Marketing Theory and Applications - ama.org · Cross-Cultural Differences in the Development of Trust in Relational Service Exchange: An Empirical Analysis in the Banking Context

2008 AMA Winter Educators’ Conference

Marketing Theory

and Applications

Editors

Tom Brown, Oklahoma State UniversityZeynep Gurhan-Canli, Koç University

Track Chairs

Sankar Sen, Baruch College, CUNYThomas Kramer, Baruch College, CUNYTom Brown, Oklahoma State UniversityCees B.M. van Riel, Erasmus UniversityGuido Berens, Erasmus UniversityJagdip Singh, Case Western Reserve UniversityRavi Sohi, University of NebraskaAysegul Ozsomer, Koç UniversityAhmet Kirca, Michigan State UniversityVikas Mittal, University of PittsburghMichael Walsh, West Virginia UniversityRajesh Chandy, University of MinnesotaJaideep Prabhu, Imperial College of LondonAric Rindfleisch, University of Wisconsin – MadisonKersi Antia, University of Wisconsin – MadisonJoshua Wiener, Oklahoma State UniversityDebra Scammon, University of UtahDeirdre Guion, St. Joseph’s UniversityMichael Kamins, University of Southern California

Volume 19

311 S. Wacker Dr. • Chicago, IL 60606

Copyright © 2008, American Marketing Association

Printed in the United States of America

Publications Director: Francesca V. CooleyProject Coordinator: Daphanee LewisCover Design: Jeanne NemcekTypesetter: Marie Steinhoff, Southeast Missouri State UniversityISSN: 1054-0806ISBN: 0-87757-329-8

All rights reserved. No part of the material protected by this copyrightnotice may be reproduced or utilized in any form or by any means,including photocopying and recording, or by any information storageor retrieval system, without the written permission of the AmericanMarketing Association.

iii

TABLE OF CONTENTS

TABLE OF CONTENTS iii

PREFACE AND ACKNOWLEDGMENTS xiii

BEST PAPERS AWARDS xiv

LIST OF REVIEWERS xv

CONSUMER PSYCHOLOGY AND BEHAVIOR

The Use of Visual Metaphors in Ads: Incongruity, the Aha Effect and AffectPraggyan Mohanty 1

Measures of Emotional Response to Television AdvertisingAnca Cristina Micu 10

The Effects of Syntactic Structure Variation on Consumers’ Memory for PrintAdvertisements

Hieu P. Nguyen, James M. Munch 12

SERVICES MARKETING AND MANAGEMENT

Changes Strategies and Ambiguous Roles: Managing Frontline Performance andPsychological Well-Being in Fast Moving Service Organizations

Jun Ye, Detelina Marinova, Jagdip Singh 14

Evaluating the Employee’s Perspective of Customer DelightDonald C. Barnes 16

The Influence of Service Climate and Personality on Nurse Customer OrientationMakoto Matsuo 18

BRANDING AND MARKETING COMMUNICATIONS

Can Companies Profit from Building a Virtual Brand Community?David M. Woisetschläger, Vivian Hartleb 26

Does Marketing Communication Suffer When Service Customers Cry Foul?Gavin L. Fox, J. Joseph Cronin 28

Hierarchy (or Anarchy) of Effects? Exploring Banner Advertising Effects andAttention

Kendall Goodrich 30

The Effect of Sponsorship on Brand Image: A Longitudinal Analysis in the AirlineIndustry

Manuel Michaelis, David M. Woisetschläger 32

GLOBAL MARKETING

Cross-Cultural Differences in the Development of Trust in Relational ServiceExchange: An Empirical Analysis in the Banking Context in China and Germany

Jan H. Schumann, Florian V. Wangenheim, Zhilin Yang 34

iv

Consumer Responses to Service Failures: A Cross-Cultural Comparison AcrossTwo Service Sectors

Venkatapparao Mummalaneni 36

Offshoring Customer Service: Communicating Across the Cultural DivideAnne Stringfellow 38


Brand Efficiency and Brand Relevance: Introducing and Linking Both ConceptsMaik Hammerschmidt, Tobias Donnevert, Hans H. Bauer 48

When Companies Go Too Far . . . and Get it Right: Exploring Low-Fit BrandExtension Success

Jill Sadler, Matthew Thomson 58

Completeness as a Product Positioning StrategyTimucin Ozcan, Daniel Sheinin 60

The Importance of the Retail Marketing Instruments and of Their PerceivedConceptual Coherence in Building a Strong Retail Brand

Bernhard Swoboda, Thomas Foscht, Hanna Schramm-Klein 62


Nature of Currency: Effects on Decision MakingAshwani Mongo, Ritesh Saini 64

The Role of Reciprocity, Sympathy, and Generosity in Predicting Tipping BehaviorGary Daniel Futrell 75

To Have or to Hold: The Influence of Payment Method on Consumer SatisfactionMichael T. Krush, Aubrey R. Fowler III 77

SALES AND RELATIONSHIP MARKETING

Mediated Effects of Self-Monitoring on Personal Selling: A Meta-AnalysisGeorge R. Franke, Jeong-Eun Park 79

The Motivation-Spillover-Principle: How Leaders Motivate Salespeople to AdoptSales Technology

Florian Kraus, Tino Kessler-Thönes, Jan Wieseke 80

Top Management Involvement with Key Accounts: The Concept, its Dimensions,and Strategic Outcomes

Rodrigo Guesalaga 82

MARKETING MANAGEMENT, STRATEGY, AND CHANNELS

Exploitation, Exploration, and Ambidexterity from a Marketing Perspective: TheImpact of Business Strategies

Silke Mühlmeier, Felicitas M. Morhart, Torsten Tomczak 84

The Role of Strategic Flexibility on Knowledge Exploitation and ExplorationKevin Zhou, Fang Wu 86

v

Beyond the Dyad: Initiating Interaction with Non-Adjacent Supply-Chain PartnersNiklas Myhr, Chapman University, Orange 87

New Product Pre-Announcements: Incumbent Responses to Competitive SignalsMayoor Mohan, Kevin E. Voss 89


Influence of Individual’s Personality Dimensions, Exposure to Advertising andPersonal Relevance on Perception of Brand Personality

Srividya Raghavan, M.S. Balaji 91

Enhancement of Brand Personality: The Use of Different Information Sources forDifferent Dimensions

Natalia Maehle, Cele Otnes, Magne Supphellen 100

Promotion Matching: The Role of Promotion Type, Self-Construal, and BrandConnection on Purchase Intentions

Karen Page Winterich, Vikas Mittal, Vanitha Swaminathan 102


Building Customer’s Trust in the Supplier Firm: The Impact and Limits ofSalesperson’s Continuity

Rodrigo Guesalaga 104

Customer-Centric Culture in Sales Organizations: On its Measurement and Impacton Salesperson Performance

Nikolaos G. Panagopoulos 106

Flattery’s Curse: How Customer-Oriented Selling Affects Short-Term CustomerSatisfaction and Decision to Buy

Alexander Haas, Peter Kenning 108


A Comparison of Television Celebrity Use in the United States and LebanonMorris Kalliny, Abdul-Rahman Beydoun, Anshu Saran, Lance Gentry 109

Celebrity Endorsement, Brand Equity, and Brand CredibilityAmanda Spry, Ravi Pappu, T. Bettina Cornwell 110

Communicating Corporate Brands Through Narratives: The Moderating Effect ofPersuasion Motives

Daniel Wentzel, Torsten Tomczak, Andreas Herrmann 112

The Neglected Power of Mood: How it Moderates Brand Placement’s EffectivenessHans H. Bauer, Melchior D. Bryant, Marcus M. Neumann 114


Flow Experience: New Approaches for Conceptualization and Modeling of aMultifaceted Construct

Jan Drengner, Pia Furchheim, Manuela Sachse 116

Exploring Perceptions of Consumer Experience OnlineNeil Frederick Hair, Deborah Colton 118

vi

Consumer Control and the Psychology of DVR UseMatthew S. Eastin, Terry Daugherty, Galit Marmor-Lavie, Sonny Rosenthal 119

Measuring Consumer Rituals: A Marketing ApplicationLarry Neale, Richard Mizerski, Alvin Y.C. Lee 121

GLOBAL MARKETING STRATEGY

Sources of Global E-Tail Advantage: Relationships among Orientations, Resources,and Performance

Deborah Colton, Martin S. Roth, William O. Bearden 122

Understanding the Drivers of the Subsidiary Innovative Capacity in Host-MarketsEsra F. Gencturk, Destan Kandemir 124

Toward an Enriched Organizational Ecology Framework of Foreign Market Entry:The Explanation for Retailers’ International Expansion

Chuanyi Tang, Eric J. Arnould 137

Reciprocal Effects of Commitment in Exporter-Importer RelationshipsFarid Ahmed 139

MARKETING RESEARCH, TECHNONOGY, AND INNOVATION

Innovate, Don’t Alienate: How Componential Factors and Constraint EnhanceCreativity in New Product Ideation

Tanawat Hirunyawipada 141

In Control or Out of Control? Active Managerial Involvement and its Impact onEscalation of Commitment

William Boulding, Abhijit Guha, Richard Staelin 142

The Effect of New Product Portfolio Management on Firm Performance: TheModerating Role of Managers’ Ambiguity Intolerance and Cognitive Style

Serdar S. Durmusoglu, Regina C. McNally 144


A Conceptual Framework of Technology Orientation: A Resource-Based PerspectiveOlivia F. Lee, Matthew L. Meuter 146

The Effects of Strategic Orientation in Emerging Economies: A Dynamic CapabilityPerspective

Caroline Bingxin Li, Kevin Zheng Zhou 148

What Happens When Firms Combine Customer and Competitor Orientations WhilePursuing Learning or Performance Goals Related to Competitors?

Xueming Luo, Mark Peterson, James M. Munch 149

Opening the Black Box of Competitive Intelligence: Uncovering AntecedentFactors and a Performance Enhancer

Adam Rapp, Raj Agnihotri, Kevin Trainor, Michael Ahearne 151


Cheers! A Means-End Chain Analysis of College Students’ Bar-Choice MotivationsRobert E. McDonald, Tillmann Wagner, Michael S. Minor 153

vii

Psychological, Product-Related and Situational Influences in Purchasing IntimateApparel

Yelena Tsarenko, Felix Mavondo 154

The Effects of Consumer Animosity and Consumption Situation on the PurchaseDecision of Nationally Iconic Products: A Means-End Chain Model

Joseph P. Little 161

The Sport Spectator Involvement Model: A Conceptual Framework and First EmpiricalResults for Understanding the Team Involvement of German Sport Spectators

Klaus-Peter Wiedmann, Nadine Hennigs, Frank Bachmann 163


Advertising Focus and Brand Association Strength: What Is the Effect of Cross-Target Interference?

Daniel W. Baack, Brian D. Till 166

Dual-Processing and Message Sidedness: The Effects of Involvement in Two-SidedAdvertising

Martin Eisend 168

Uncovering Undercover Marketing: The Role of Consumers’ Agent, Persuasion, andTopic Knowledge

Guang-Xin Xie, David, M. Boush 170


Rocky Sales-Marketing Interface: An Exploratory Examination of Symptoms, Causes,Effects, and Remedies

Avinash Malshe, Ravipreet S. Sohi 178

The Contingent Value of Sales Participation in Marketing Decision-Making onOrganizational Performance

Gabriele Troilo, Luigi M. De Luca, Paolo Guenzi 179


A Conceptual Model of Antecedents and Consequences of Consumer Uncertainty:Model Development and Critical Assessment

Edward Shiu, Louise Hassan, Gianfranco Walsh 181

Clarity and Customer ValueJoe Urbany, William Bearden, Raymond Kordupleski 183

The Influence of a Tentative Preference on Information SelectionKurt Carlson, Abhijit Guha 193

Two-Stage Lotteries and the Value of Unresolved Uncertainty in Multiple ContactsIdo Erev, Ernan Haruvy 195


Company Designers Versus Users Through the Customer Lens: Who Holds MoreCreative Potential?

Martin Schreier, Christoph Fuchs 197

viii

Examining the Influence of Social Capital on Corporate ReputationKlaus-Peter Wiedmann, Nadine Hennigs, Barbara Gassmann 199

Mission Fulfillment and the Internal Audience: Psychological Job OutcomesTaewon Suh, Mark B. Houston, Steven Barney, Ik-Whan G.Kwon 202

Do B2B Brands Make a Difference? An Empirical Investigation of the PerformanceImplications of Branding in B2B Environments

Christian Homburg, Martin Klarmann, Jens Schmitt 204

MARKETING AND SOCIETY

Avoidance, Falsification, and Pragmatic Caution: The Impact of Technology Comforton Consumers’ Online Privacy Coping Strategies

Pushkala Raman, Kartik Pashupati 206

Higher Expectations for Higher Satisfaction: The Effect of Attribute Types on Post-Trial Customer Satisfaction

Camelia C. Micu, Robin A. Coulter 208

How Do Individual-Level and Country Characteristics Affect the Intention to QuitSmoking?

Gianfranco Walsh, Louise May Hassan, Edward Shiu 216

The Potential Effects of Corrective Advertising on Consumer Beliefs Mandated byU.S. vs. Philip Morris USA, Inc. (2006)

Andrea Heintz Tangari, Brook Plack, Scot Burton, Craig Andrews 229

GLOBAL MARKETING

Does Perceived Product Quality Impact Country-of-Origin Image? An EmpiricalInvestigation

Qin Sun 231

Effects of Consumer Ethnocentrism and Animosity on Foreign Product Evaluationand Purchase Intentions: Moderating Role of Cultural Similarity and EconomicCompetitiveness

Jun Ma, Wei (Andy) Hao, Laura Lin 233

“Made in . . .” – Does it Also Apply to Services? An Empirical Assessment of theCountry-of-Origin Effect in Service Settings

Dieter Ahlert, Christof Backhaus, Johannes Berentzen, Markus Blut,

Manuel Michaelis 235


Atmosphere Measurement and MythsMorten Heide, Kjell Gronhaug 237

Return on Servicescape Investments: Does Remodeling Make a Difference?Dwayne D. Gremler, Elisabeth Brüggen, Bram Foubert 252

The Structure and Function of Language in the Strategic Rituals of ServiceProviders

Cele C. Otnes, Behice Ece Ilhan, Atul Kulkarni 254

ix

Communication of Price Increases: How Can Negative Consumer Reactions BeReduced?

David M. Woisetschläger, Heiner Evanschitzky, Hartmut H. Holzmüller 264

MARKETING RESEARCH, TECHNOLOGY, AND INNOVATION

Media Competition for National Advertising in the Era of New Media, 1997–2006John Dimmick, Osei Appiah, Matthew S. Eastin 266

Exploring the Market Structure of Online AuctionsStephen France, J. Douglas Carroll 268

Exploring Consumer Motivations for Creating User-Generated ContentLaura F. Bright, Terry Daugherty, Matthew S. Eastin 270

An Innovative Data Collection Method from the Internet for Marketing Research:Computer Spiders

Yanbin Tu 272


A Measurement Scale for Customer Co-Production (CCP) of ServicesMohammad Ali Zolfagharian 273

Customer Co-Design Competence: Model Development and Influence on CustomerSatisfaction with Mass Customization Offerings

Melanie Müller, Ralf Reichwald, Frank Piller 275

Who Should Do the Work: The Different Roles of Customer Participation and ItsImpacts on Satisfaction

Lan Xia, Xiucheng Fan 284

Cooperative vs. Rebellious Customers: How They Evaluate Consumer PenaltyYoung “Sally” Kim 292


What Role Does Customer Relationship Management Play in Marketing Strategy?Martin Reimann, Oliver Schilke, Jacquelyn Thomas 294

Are Newspaper Subscribers Closely Connected or Just Bound by Contract?An Empirical Analysis of Moderating Effects in the Four-Stage Loyalty Model

Patrick Lentz, David M. Woisetschläger, Heiner Evanschitzky 296

The Mediating Effect of Customer Satisfaction Between Switching Costs andCustomer Loyalty

Markus Blut, David M. Woisetschläger 298


A Social Marketing Model of Religious OrganizationsRobert M. Schindler 300

“What About Us? I Mean, Hey, We’re Growing, We’re Here”: Women Livingwith HIV/AIDS

Yelena Tsarenko 302

x

The Moderating Role of Managerial Ethical Predispositions in Dyadic MarketingExchange

Kelly D. Martin, Jean L. Johnson 309


Preannouncing Pioneering vs. Follower Products: What Should the Message Be?Christian Homburg, Torsten Bornemann, Dirk Totzek 311

How Does Distance Matter? The Importance of Pre-Launch Stakeholders inSpatial Diffusion of New Medicines

David Dekker, Paul H. Driessen 313

The Effect of Prior Experience and Product Type on Changes in Perceived Risk foran Innovation: The Role of Feelings

Arjun Chaudhuri, Camelia C. Micu 315

The Role of Consumers’ Need for Uniqueness in the Innovation Adoption ProcessMachiel J. Reinders, Ruud T. Frambach, Mirella H.P. Kleijnen 322


An Empirical Investigation of the Influence of Corporate Social Capital onConsumers’ Company and Product Perceptions

Klaus-Peter Wiedmann, Nadine Hennigs, Barbara Gassmann 324

Assessing Loyalty Program Effects in a Competitive Setting: Impact of MarketSaturation, Market Share, and Category Expandability

Yuping Liu, Rong Yang 326

Using Segment Attractiveness to Improve Segment Selection in the Credit CardBusiness

Ranjeet Khaira 328

GLOBAL MARKETING

A Model to Include the Potential Impact of Governmental Public and ForeignPolicies on Global Brand Equity

Peter A. Kaufman, Frederick W. Langrehr 336

National Culture and Adoption of Technology-Based New ProductsAbhijit Patwardhan, Douglas W. Vorhies 342

Standardization with Respect to What? A Conceptualization of AdvertisingStandardization Measurement in Past Research

Fernando Fastoso, Jeryl Whitelock 344

Loving a Brand: Concept and CultureNoël Albert, Deanne Brocato, Dwight Merunka, Pierre Valette-Florence 346


Stress-Induced Word-of-Mouth Seeking: A Missing Dimension in Word-of-Mouth

Research

Selina Cui, Haksin Chan 348

xi

The Consumer Strikes Back: Assessing Antecedents to Boycott ParticipationHans H. Bauer, Carmen-Maria Albrecht, Melchior D. Bryant, Tobias E. Haber 350

Selective Consumer Communications: The Role of Relationship StrengthYu Hu 352

Understanding Attitude-Behavior Gap in Green Buying as a Social DilemmaShruti Gupta, Denise T. Ogden 354


Drivers of Service Recovery Performance: Perceived Organizational Support,Learning and Psychological Job Outcomes

Christiana Raquel Lages, Nigel F. Piercy 356

Failure and Recovery in Internet Purchasing: Their Impact on Consumer Trust andLoyalty to the Company’s Site and to the Online Shopping Environment

Daniel von der Heyde Fenandes, Cristiane Pizzutti dos Santos 358

Measuring the Perceived Importance That Consumers Place on the Benefit of FreeTechnical Support: A New Scale

Lee Hibbett, Keith Absher, Joshua Shackman, Darin White 369

Spectator Rage: Examining the Dark-Side of Fan BehaviorScott A. Jones, Stephen J. Grove, Gregory M. Pickett 371


An Argument for Co-Opetition’s Pro-Consumer PostureMichael A. Levin, Sabrina Sattler 380

Functional Food Products: Issues and Implications for Society and FirmsShelly Freyn, Sreedhar Madhavaram 389

Who Is on Your Side? Agency Representation and Consumer Welfare in RealEstate Markets

Kristin Rotte, Murali Chandrashekaran 390

MARKETING, MANAGEMENT, STRATEGY, AND CHANNELS

Marketing in Second Life: If You Build It, Will They Come?Lakshmi Goel, Sonja Prokopec 399

Proposing and Conceptualizing a Service-Dominant Strategic OrientationIngo O. Karpen, Liliana L. Bove, Alexander P. Josiassen 406

Performance Implications of Developing Customer-Linking CapabilitiesKevin Trainor, Adam Rapp, Raj Agnihotri, Niels Schillewaert 415


When Can We Measure Willingness to Pay Directly? An Empirical Study on theRole of Consumers’ Involvement in the Direct Elicitation of Reservation Prices

Klaus Matthias Miller, Harley Krohmer, Reto Hofstetter 417

xii

An Investigation of the Relationship Between the Number of Response Categoriesand Scale Sensitivity

Richard J. Fox, Sunil H. Contractor 419

The Puzzle of “About Six” Indicators in Latent VariablesRobert A. Ping 421

The Vital Few and the Useful Many Using Support Vector Machines to IdentifyFuture Best Customers

Markus Wuebben, Florian Wangenheim 432

AUTHOR INDEX 443

xiii

Preface and Acknowledgments

The theme of the 2008 AMA Winter Educators’ Conference is “Marketing the Organization and Its Products andServices.” There has been an explosion of research activity in marketing and other disciplines related to what individualsknow about organizations and how they respond to this information. By studying topics such as corporate branding,consumer boycotting, consumer–company identification, corporate communication, cause-related marketing, corporatesocial responsibility, corporate brand personality, reputation/image, and corporate associations, researchers havebrought more of the focus of scholarly attention in marketing to the organization or corporation in addition to thetraditional focus on product- or service-level marketing. Many of the conference sessions reflect the state-of-the-artthinking on these topics.

Many people have been involved in the development and implementation of the conference. Most important, weappreciate the efforts of the fine scholars who have contributed the fruits of their intellectual curiosity and passion. Therewould be no conference without their stimulating input, whether in the form of papers, presentations, panel discussions,or other contributions.

The track chairs really made things happen. We challenged them to substantially increase participation in specialsessions – and they exceeded our expectations, delivering well-rounded sets of special sessions and competitive papersessions. The track chairs included are as follows:

Consumer Psychology and Behavior Sankar Sen, Baruch College, CUNYThomas Kramer, Baruch College, CUNY

Branding and Marketing Communications Tom Brown, Oklahoma State UniversityGuido Berens, Erasmus University RotterdamCees van Riel, Erasmus University Rotterdam

Sales and Relationship Marketing Jagdip Singh, Case Western Reserve UniversityRavi Sohi, University of Nebraska – Lincoln

Global Marketing Ahmet Kirca, Michigan State UniversityAysegul Ozsomer, Koç University

Services Marketing and Management Vikas Mittal, Rice UniversityMichael Walsh, West Virginia University

Marketing Research, Technology, and Innovation Rajesh Chandy, University of MinnesotaJaideep Prabhu, Imperial College London

Marketing Management, Strategy, and Channels Aric Rindfleisch, University of Wisconsin – MadisonKersi Antia, University of Wisconsin – Madison

Marketing and Society Josh Wiener, Oklahoma State UniversityDebra Scammon, University of UtahDeirdre Guion, Saint Joseph’s University

Special Interest Groups Michael Kamins, Stony Brook University, SUNY

Reviewers also gave of their time and effort to evaluate the hundreds of papers and sessions submitted to theconference. Thanks to all who were willing and able to help (listed on page xv). We also thank the members of our “BlueRibbon” Award Selection Committee who had the unenviable task of selecting a single paper to receive the “best ofconference” award.

Because we were new to running a conference using an online conference management system, Rick Peacor withAllAcademic.com had to display extreme patience as he guided us through the process. We appreciate his efforts.

Finally, we thank the American Marketing Association staff and volunteers who have gone above and beyond tohelp us throughout this endeavor. Lynn Brown (program director), Daphanee Campbell, Francesca Cooley, CherDoherty, and Pat Goodrich all did a terrific job keeping us pointed in the right direction. As usual, the SIG members andleaders came through with a great set of SIG special sessions. We also thank the members of the AMA Academic Councilwho offered this opportunity and helped in a variety of ways. Pam Ellen (Georgia State University), the current presidentof Academic Council, was always quick with the right answer when we needed it, which was often. Thanks, Pam.

Tom Brown Zeynep Gurhan-Canli

Oklahoma State University Koç University

xiv

Best PaperAwards

Best of Conference Award“Change Strategies and Ambiguous Roles: ManagingFrontline Performance and Psychological Well-Beingin Fast Moving Service Organizations”

Jun Ye, University of OregonDetelina Marinova, University of Missouri –

ColumbiaJagdip Singh, Case Western Reserve University

Best Paper Awards by Track

Global Marketing“Sources of Global E-Tail Advantage: RelationshipsAmong Firm Orientations, Resources, and Performance”

Deborah Colton, Rochester Institute ofTechnology

Martin Roth, University of South CarolinaWilliam Bearden, University of South Carolina

Consumer Psychology and Behavior“Flow Experience: New Approaches for Conceptual-ization and Modeling of a Multifaceted Construct”

Jan Drengner, University of TechnologyPia Furchheim, University of TechnologyManuela Sachse, University of Technology

Services Marketing and Management“Change Strategies and Ambiguous Roles: ManagingFrontline Performance and Psychological Well-Beingin Fast Moving Service Organizations”

Jun Ye, University of OregonDetelina Marinova, University of Missouri –

ColumbiaJagdip Singh, Case Western Reserve University

Marketing Management, Strategy, and Chan-nels“The Role of Strategic Flexibility on Knowledge Exploi-tation and Exploration”

Kevin Zhou, University of Hong KongFang Wu, University of Texas at Dallas

Marketing and Society“The Potential Effects of Corrective Advertising on Con-sumer Beliefs Mandated by U.S. vs. Philip Morris”

Andrea Heintz Tangari, University of ArkansasBrooke Plack Adams, University of ArkansasScot Burton, University of ArkansasJ. Craig Andrews, Marquette University

Branding and Marketing Communication“Mission Fulfillment and the Internal Audience: Psycho-logical Job Outcomes”

Taewon Suh, Texas State University – San MarcosMark B. Houston, Texas Christian UniversityStephen M. Barney, SSM Health CareIk-Whan G. Kwon, Saint Louis University

Marketing Research, Technology, and Inno-vation“The Vital Few and the Useful Many: Using SupportVector Machines to Identify Future Best Customers”

Markus Wuebben, Technical University MunichFlorian Wangenheim, Technische Universitaet

Muenchen

Sales and Relationship Marketing“What Role Does Customer Relationship ManagementPlay in Marketing Strategy?”

Martin Reimann, Stanford UniversityOliver Schilke, Stanford UniversityJakki Thomas, Northwestern University

xv

2008 AMA Winter Educators’ Conference

List of Reviewers

AKhaled Aboulnasr, Florida Gulf

Coast UniversityUlla R. Ahlfors, University of

JyvaskylaKirk Damon Aiken, Eastern

Washington UniversityLerzan Aksoy, Koc UniversityMuhammad Aljukhadar, HEC

MontrealArthur Allaway, University of

AlabamaWilfred Amaldoss, Duke

UniversityClinton Amos, University of North

TexasLaurel Anderson, Arizona State

UniversityTrina L. Andras, Drexel UniversityJonlee Andrews, Indiana

UniversitySyed Tariq Anwar, West Texas

A&M UniversityEvmorfia Argyriou, Aston

UniversityArmin Arnold, Technical

University of MunichInigo Arroniz, University of

CincinnatiJoep Arts, Vrije Universiteit

AmsterdamBerk Ataman, RSM Erasmus

UniversityJoann L. Atkin, Western Michigan

UniversityKwaku Atuahene-Gima, China

Europe International BusinessSchool

BDaniel W. Baack, Ball State

UniversityBarry J. Babin, University of

Southern MississippiFrank Bachmann, Leibniz

University of HannoverCem Bahadir, Emory UniversitySoumava Bandyopadhyay, Lamar

UniversityRanjan Banerjee, University of

Minnesota

Yongchuan Bao, California StateUniversity, Fullerton

Victor A. Barger, University ofWisconsin – Madison

Donald Clay Barnes, MississippiState University

Jos Bartels, LEI Wageningen URRajeev Batra, University of

MichiganEnrique P. Becerra, Texas State

University, San MarcosJoe Bellizzi, Arizona State

UniversityNeil Thomas Bendle, University of

MinnesotaKatrina J. Bens, Pennsylvania State

UniversityJoseph Ben-Ur, University of

Houston, VictoriaDoris Berger, FH KremsDavid Berkowitz, University of

Alabama, HuntsvilleThomas Bertsch, James Madison

UniversityPelin Bicen, Texas Tech UniversityGabriel Biehal, University of

Maryland, College ParkCharles Blankson, University of

North TexasVera Blazevic, Maastricht

UniversityLauren Block, Baruch College,

CUNYJames Boles, Georgia State

UniversityPaula Bone, West Virginia

UniversitySterling A. Bone, Brigham Young

UniversityAberdeen Leila Borders, University

of New OrleansDoug Bowman, Emory UniversityMairead Brady, Trinity College

DublinS. Adam Brasel, Boston CollegeEileen Bridges, Kent State

UniversityNenad Brkic, University of

SarajevoPeggy S. Brønn, Norwegian School

of Management

Terence A. Brown, Penn StateHarrisburg

Jim Burroughs, University ofVirginia

Scot Burton, University ofArkansas

Paul Busch, Texas A&MUniversity

CJoe Cannon, Colorado State

UniversityBrad D. Carlson, Texas Tech

UniversityLes Carlson, Clemson UniversityFrançois A. Carrillat, HEC

MontréalBrian R. Chabowski, University of

TulsaChiu-chi Angela Chang,

Shippensburg UniversityWei-Lun Chang, Tamkang

UniversityCristian Chelariu, Suffolk

UniversityAlka Citrin, Georgia Institute of

TechnologyBruce Clark, Northeastern

UniversityLucette B. Comer, Purdue

UniversityKatsikeas Constantine, Leeds

UniversityDelonia O. Cooley, Texas Southern

UniversityDeborah Cowles, Virginia

Commonwealth UniversityAmy E. Cox, Converse CollegeAnthony D. Cox, Indiana

UniversityDena Cox, Indiana UniversityDavid Crête, HEC MontréalPeggy Cunningham, Queen’s

School of Business

DScott G. Dacko, University of

WarwickRobert Dahlstrom, University of

Kentucky

xvi

Rajiv P. Dant, University of SouthFlorida

Brad Davis, Wilfrid LaurierUniversity

Federico de Gregorio, Universityof Alabama

Tom DeCarlo, University ofAlabama, Birmingham

Jade Sturdy DeKinder, Universityof Texas at Austin

Benedict Dellaert, ErasmusUniversity Rotterdam

Susan Kay DelVecchio, EastCarolina University

Debra M. Desrochers, Universityof Notre Dame

Irene J. Dickey, University ofDayton

Bart Dietz, RSM ErasmusUniversity

Andrea L. Dixon, University ofCincinnati

Michael J. Dorsch, ClemsonUniversity

Paul H. Driessen, RadboudUniversity Nijmegen

Cornelia Droge, Michigan StateUniversity

Shuili Du, Simmons College

EAndreas Eisingerich, University of

Southern CaliforniaJoel Evans, Hofstra UniversityKenneth Evans, University of

Oklahoma

FFernando Fastoso, University of

Bradford School of ManagementDaniel von der Heyde Fernandes,

Universidade Federal do RioGrande do Sul

Karen Flaherty, Oklahoma StateUniversity

Theresa B. Flaherty, JamesMadison University

Renee Florsheim, LoyolaMarymount University

John B. Ford, Old DominionUniversity

Gavin L. Fox, Florida StateUniversity

Ruud T. Frambach, VU UniversityAmsterdam

Stephen Lee France, RutgersUniversity

Gary Frankwick, Oklahoma StateUniversity

Frank Franzak, VirginiaCommonwealth University

Jonathan Freeman, University ofWarwick

Nancy Furlow, MarymountUniversity

Gary Daniel Futrell, Florida StateUniversity

GBashar S. Gammoh, University of

ToledoShankar Ganesan, University of

ArizonaNitika Garg, University of

MississippiJudith Anne Garretson Folse,

Louisiana State UniversityJule Gassenheimer, Rollins CollegeBarbara Gassmann, Leibniz

University of HannoverEsra F. Gencturk, Koc UniversityJames Gentry, University of

NebraskaDominik Georgi, University of

Basel, SwitzerlandPervez Nasim Ghauri, Manchester

Business SchoolDavid I. Gilliland, Colorado State

UniversityJacob Goldenberg, Hebrew

University of JerusalemPeter Golder, New York UniversityRonald E. Goldsmith, Florida State

UniversityAnuj P. Gosain, Indian Institute of

Management, BangaloreElad Granot, Cleveland State

UniversityAlbert N. Greco, Fordham

UniversityAndreas Grein, Baruch College,

CUNYRajdeep Grewal, Pennsylvania

State UniversityChristopher J. Groening,

University of PittsburghBarbara L. Gross, California State

University, NorthridgeRajesh Gulati, St. Cloud State

University

Greg Gundlach, University ofNorth Florida

Michael Guolla, University ofOttawa

Reetika Gupta, Lehigh UniversityShruti Gupta, Pennsylvania State

University, Abington

HAlexander Haas, University of

BernMignon van Halderen, Rotterdam

School of ManagementJay Handelman, Queen’s

UniversityAndrew Wei Hao, Kent State

UniversityNukhet Harmancioglu, Bilkent

UniversityJames E. Harris, Saint Norbert

CollegeAngela Hausman, Xavier

UniversityHongwei He, University of East

AngliaBill Hedgcock, University of

MinnesotaBob Heiser, University of Southern

MaineDavid H. Henard, North Carolina

State UniversityTy Henderson, University of Texas

at AustinNadine Hennigs, Leibniz

University of HannoverMonica D. Hernandez, University

of Texas, Pan AmericanKelly Hewett, University of South

CarolinaBas Hillebrand, Nijmegen School

of ManagementStacey B. Hills, Utah State

UniversityTanawat Hirunyawipada, Central

Michigan UniversityJens Hogreve, University of

PaderbornHartmut H. Holzmuller, Dortmund

UniversitätEarl Honeycutt, Elon UniversityMark B. Houston, Texas Christian

UniversityJohn Hulland, University of

Pittsburgh

xvii

Gary K. Hunter, Case WesternReserve University

Gary L. Hunter, Illinois StateUniversity

Michael D. Hutt, Arizona StateUniversity

Michael Hyman, New MexicoState University

ICaglar Irmak, Baruch College,

CUNYChiharu Ishida, Illinois State

University

JAnupam Jaju, George Mason

UniversitySandy Jap, Emory UniversityRama Jayanti, Cleveland State

UniversityMark Johlke, Bradley UniversityAllison Johnson, Queen’s

UniversityDevon S. Johnson, Northeastern

UniversityChris Joiner, George Mason

UniversityScott Jones, Clemson UniversityAlexander P. Josiassen,

Unuiversity of Melbourne

KManish Kacker, Tulane UniversityAndrew M. Kaikati, University of

MinnesotaDestan Kandemir, Bilkent

UniversityMoon Young Kang, University of

Wisconsin – MadisonIngo O. Karpen, University of

MelbourneVishal Kashyap, Xavier UniversityEvangelia Katsikea, Athens

University of Economics andBusiness

Jeremy Kees, Villanova UniversityCraig Kelley, California State

University, SacramentoStephen K. Kim, Iowa State

UniversityGurprit Singh Kindra, University

of OttawaAhmet H. Kirca, Michigan State

University

Jill Klein, INSEADAjay K. Kohli, Emory UniversityLaura J. Kornish, University of

Colorado, BoulderJohn C. Kozup, Villanova

UniversityMichael Krush, University of

Nebraska, LincolnAlok Kumar, University of

Wisconsin

LRuss Laczniack, University of IowaGene Laczniak, Marquette

UniversityCristiana Raquel Lages, Boston

CollegeDesmond Lam, University of

MacauSon K. Lam, University of HoustonJay Lambe, Seattle UniversityTimothy D. Landry, University of

OklahomaJeff Langenderfer, Meredith

CollegeThomas W. Lanis, East Central

UniversityNick Lee, Aston UniversityJames H. Leigh, Texas A&M

UniversityPatrick Lentz, University of

DortmundHillary Leonard, University of

Rhode IslandDawn Lerman, Fordham UniversityMichael Levin, Texas Tech

UniversityBeichen Liang, University of

Illinois at ChicagoYam Bahadur Limbu, New Mexico

State UniversityJoan Lindsey-Mullikin, Babson

CollegeAnthony Lowrie, Minnesota State

University, MoorheadRyan J. Luchs, Univsersity of

PittsburghDavid Luna, Baruch College,

CUNY

MGerrard M. Macintosh, North

Dakota State UniversityAvinash Malshe, University of St.

Thomas

Elliot Maltz, Willamette UniversitySudha Mani, University of Texas at

ArlingtonKen Manning, Colorado State

UniversityLawrence J. Marks, Kent State

UniversityKelly Martin, Colorado State

UniversityMarlys Mason, Oklahoma State

UniversityShashi M. Matta, Ohio State

UniversityPaul Matthyssens, University of

AntwerpDeborah Brown McCabe, Arizona

State UniversityMichael McCarthy, Miami

UniversityPeter McClure, University of

Massachusetts, BostonBob McDonald, Texas Tech

UniversityJoyce A. McGriff, Southern

Polytechnic State UniversityRegina C. McNally, Michigan

State UniversityTimothy Meyer, University of

Wisconsin, Green BayStefan Michel, Thunderbird,

School of Global ManagementCamelia Codruta Micu, Fairfield

UniveristyGeorge Milne, University of

Massachusetts, AmherstSoonhong (Hong) Min, University

of OklahomaSaurabh Mishra, McGill UniversityVincent W. Mitchell, Cass

Business SchoolAnthony Miyazaki, Florida

International UniversitySabine Moeller, WHU – Otto

Beisheim School of ManagementBruce Money, Brigham Young

UniversityErik Mooi, Aston UniversityMark Moon, University of

TennesseeNeil A. Morgan, Indiana UniversitySayantani Mukherjee, California

State University, Long BeachVenkatapparao Mummalaneni,

Virginia State UniversityPatrick Murphy, University of

Notre Dame

xviii

Laurent Muzellec, Dublin CityUniversity

Niklas Myhr, Chapman University

NCheryl Nakata, University of

Illinois at ChicagoStern Neill, University of

Washington, TacomaNoelle Nelson, University of

MinnesotaEdwin Nijssen, Eindhoven

University of TechnologyRakesh Niraj, University of

Southern CaliforniaEdward Lorenzo Nowlin,

University of Nebraska – Lincoln

OClaude Obadia, Advancia-NegociaMatthew O’Brien, Bradley

UniversityMatthew S. O’Hern, University of

Wisconsin, MadisonShintaro Okazaki, Universidad

Autónoma de MadridGlenn Omura, Michigan State

UniversityTimucin Ozcan, University of

Rhode Island

PNikolaos G. Panagopoulos, Athens

University of Economics &Business

Photis M. Panayides, CIIMBusiness School

Jeong-Eun Park, Ewha WomansUniversity

Ji Kyung Park, University ofMinnesota

Chirag Patel, ESC Rennes Schoolof Business

Shalini Pathak, IIT BombayLaura Peracchio, University of

Wisconsin, MilwaukeeA. Morys Perry, University of

Michigan, FlintRoss Petty, Babson CollegePaulo Prado, Federal University of

ParanaSonja Prokopec, ESSEC Business

School, FranceStefano Puntoni, Erasmus

University

QTianjiao Qiu, California State

University, Long Beach

RPradeep Rau, George Washington

UniversityMartin Reimann, Stanford

UniversitySubom Rhee, Santa Clara

UniversityJosep Rialp, Universitat Autònoma

de BarcelonaJef I. Richards, University of TexasLiesl Riddle, George Washington

UniversityDebra Ringold, Willamette

UniversityRichard K. Robinson, Marquette

UniversityAksel Ivar Rokkan, SNF-Research

in Economics and BusinessAdministration

Jose Antonio Rosa, University ofIllinois at Chicago

Mark S. Rosenbaum, NorthernIllinois University

William Taylor Ross, Jr.,Pennsylvania State University

Martin Roth, University of SouthCarolina

Subroto Roy, University of NewHaven

SMaria Sääksjärvi, Swedish School

of Economics and BusinessAdministration

Laura Salciuviene, University ofManchester

Saeed Samiee, University of TulsaKare Sandvik, Buskerud University

CollegeM. Kim Saxton, Indiana UniversitySerdar Sayman, Koç UniversityNancy Panos Schmitt, Westminster

CollegeJonathan E. Schroeder, University

of ExeterClifford Schultz, Arizona State

UniversityJan H. Schumann, Technical

University of Munich

Charles Schwepker, University ofCentral Missouri

Sanjit Sengupta, San FranciscoState University

Hamed M. Shamma, GeorgeWashington University

Kirby Shannahan, University ofNew Brunswick

Arun Sharma, University of MiamiMary Jane Sheffet, University of

Northern IowaAstrid Siebels, Leibniz University

of HannoverJames Simpson, University of

Alabama, HuntsvilleRamendra Singh, IIM AhmedabadEugene Sivadas, University of

WashingtonRebecca J. Slotegraaf, Indiana

UniversityKaren H. Smith, Texas State

University, San MarcosHarmeen Kaur Soch, Guru Nanak

Dev University, AmritsarAlina Sorescu, Texas A&M

UniversityArina Soukhoroukova, University

of PassauDeborah F. Spake, University of

South AlabamaJelena Spanjol, University of

Illinois at ChicagoMartin Spann, University of PassauMark Spriggs, University of St.

ThomasDavid E. Sprott, Washington State

UniversityElizabeth Stammerjohan, Jackson

State UniversityJulie Stanton, Saint Joseph’s

UniversityMark Staton, University of

WashingtonDavid Stewart, University of

California, RiversideKaren Stilley, University of

PittsburghNicola E. Stokburger-Sauer,

University of MannheimAnne Stringfellow, ThunderbirdRodney L. Stump, Towson

UniversityBharat L. Sud, University of

Western OntarioUrsula Y. Sullivan, University of

Illinois at Urbana-Champaign

xix

Tracy A. Suter, Oklahoma StateUniversity

Scott D. Swain, Boston UniversityVanitha Swaminathan, University

of PittsburghJack Swasy, American UniversityMichael J. Swenson, Brigham

Young University

TMarios Theodosiou, University of

CyprusSweta Chaturvedi Thota, James

Madison UniversityCarlos Javier Torelli, University of

MinnesotaCarlos Torelli, University of

Illinois at Urbana-ChampainJanell D. Townsend, Oakland

University

VJohan Van Rekom, Rotterdam

School of ManagementAndres Rodriguez Veloso,

Mackenzie; FEA-USPChander Velu, Cambridge

UniversityRaj Venkatesan, University of

VirginiaLeslie Vincent, University of

KentuckyMadhu Viswanathan, University of

MinnesotaKevin E. Voss, Oklahoma State

University

WTillmann Wagner, Texas Tech

UniversityGianfranco Walsh, University of

Koblenz-Landau, GermanyGuangping Wang, Pennsylvania

State UniversityFlorian Wangenheim, Technische

Universitaet MuenchenJudith Washburn, University of

TampaKenneth H. Wathne, Norwegian

School of ManagementRebecca Wells, University of

DaytonSusan Whelan, Waterford Institute

of TechnologyRick Whiteley, Calabash

Educational SoftwareJerome D. Williams, University of

Texas at AustinPhil Wilson, Midwestern State

UniversityKaren P. Winterich, Texas A&M

UniversityNancy Wong, Georgia Institute of

TechnologyVeronica Wong, Aston UniversityNatalie Wood, Saint Joseph’s

UniversityBob T. Wu, Bowling Green State

UniversityFang Wu, University of Texas at

DallasNancy V. Wuenderlich, Technical

University of MunichStefan Wuyts, Tilburg University

XLan Xia, Bentley CollegeGuang-Xin Xie, University of

Oregon

YManjit Yadav, Texas A&M

UniversityChun Ming Yang, National Sun

Yat-Sen UniversityAttila Yaprak, Wayne State

UniversityJun Ye, University of OregonSengun Yeniyurt, Rutgers

UniversityTuba Pinar Yildirim, University of

PittsburghBoonghee Yoo, Hofstra UniversityEric Yorkston, Texas Christian

University

ZAlex R. Zablah, Oklahoma State

UniversityLixuan Zhang, College of

CharlestonGuangzhi Zhao, University of

KansasYuhuang Zheng, Fordham

UniversityMarcus Steven Zimmer, Technical

University Munich, GermanyShaoming Zou, University of

Missouri, Columbia

American Marketing Association / Winter 2008 1

THE USE OF VISUAL METAPHORS IN ADS: INCONGRUITY,

THE AHA EFFECT AND AFFECT

Praggyan Mohanty, University of Missouri, Columbia

ABSTRACT

Visual metaphors, used extensively in ads, make ananalogical comparison between two terms. Thehomospatial existence of two distinct terms makes thevisual metaphor incongruous. The incongruity in visualmetaphors renders the advertisement incomprehensible atfirst glance. This provokes the viewer to resolve theconflict via metaphorical comprehension. The initial in-comprehension followed by comprehension of the ad iscalled the “aha experience” that yields positive affect. Ourstudy attempts to throw light on (a) the nature of relation-ship between incongruity of visual metaphors and affect(b) the mechanism by which incongruity in visual meta-phors leads to positive affect.

INTRODUCTION

A recent ad for Microsoft Office shows a femalestanding in an office setting with the head of a dinosaur.The ad copy reads as “Microsoft Office has evolved haveyou?” This is an example of a visual metaphor, a type ofrhetorical figure1 that is used extensively in advertisingand communication (Phillips 2003). Visual metaphorspresent two ideas or terms in relationship to one anothersuch that one (i.e., the source term, the dinosaur in theMicrosoft Office ad) is used to organize or conceptualizethe other (i.e., the target term, the female in the MicrosoftOffice ad) (Kittay 1987; Lakoff and Johnson 1980). Avisual metaphor makes an analogical comparison be-tween two terms, by stating that one term (e.g., the female)is figuratively like the other term (e.g., the dinosaur), eventhough the two are literally different (Stern 1990). Thevisual of a dinosaur’s head on a female’s body is incon-gruous, it does not make sense at a literal level. Thisincongruity in the visual attracts attention and one isdrawn toward reconciling the incongruity. Only at afigurative level, can one interpret that the ad is suggestingthat the female is “outdated, anachronous” and needs toupgrade to newer versions of Microsoft Office.

The incongruity in the visual makes the ad stimulat-ing (McQuarrie and Mick 1996; McQuarrie and Mick1999; Peracchio and Meyers-Levy 1994), elicits somelevel of exploratory behavior (Berlyne 1960) and subse-quent resolution of incongruity with meaningful per-ceived comprehension yields what is termed in semioticsliterature as “pleasure of the text” (Barthes 1986). “Plea-sure of the text” is conceptualized as the positive affectthat an individual experiences due to the person’s subse-

quent perceived comprehension of an initially incompre-hensible text.2

The phenomenon of incongruity-driven-pleasure orpositive affect experienced by the audience (McQuarrieand Mick 2003; Peracchio and Meyers-Levy 1994) is apotent one both for the advertiser and the academic.Research suggests that pleasure experienced from the adleads to positive attitude-toward-the-ad (McQuarrie andMick 1999; Mick 1992) that in turn gets transferred aspositive attitude-toward-the-brand (MacKenzie and Lutz1989; MacKenzie et al. 1986). Also, cognitive effort toreconcile the incongruity and the generation of positiveaffect due to successful comprehension of the visualmetaphor is an intriguing interplay of cognitions andemotions that should pique the interest of a theoretician.We embark on a journey to address two unansweredquestions in advertising and marketing literature (i) Whatis the nature of relationship between incongruity in visualmetaphors and affect? (ii) What is the mechanism bywhich incongruity in visual metaphors leads to positiveaffect?

What is apparent from prior studies is that incongru-ity plays an important role in “pleasure of the text” orpositive affect. But there are boundary conditions forincongruity or deviation in visual metaphor to yield aquantum of pleasure. If deviation falls below a thresholdlevel or is above an upper limit, pleasure may not beexperienced. Therefore, if incongruity is too low then thevisual metaphor becomes a degenerate literal statementand is too dull for attention (Tversky 1977). On the otherhand, if the incongruity is too high, the metaphor becomesobscure and uninterpretable thus thwarting ad evaluations(Tourangeau and Strernberg 1981). Further, there is amediatory phenomenon of incomprehension at initialperusal of the metaphor followed by subsequent compre-hension, an “arousal – release” sequence (McQuarrie andMick 2003) that seems to play a role in pleasure of the text.But there is lack of empirical validations of these theoreti-cal propositions.

First, the relationship of incongruity in visual meta-phors and positive affect has not been examined empiri-cally in the marketing literature. Research has shown thatrhetorical figures versus literal expressions have a signifi-cant positive effect on attitude-toward-the-ad. McQuarrieand Mick (1999) and Mothersbaugh et al. (2002) foundthe “figures effect”; rhetorical figures were perceived tobe more artful/ clever or incongruous than literal expres-

2 American Marketing Association / Winter 2008

sions. In a subsequent paper, McQuarrie and Mick (1999)studied the effect of four types of visual rhetorical figureson attitude-toward-an-ad. They found that the visual fig-ures had a significant positive effect on attitude-toward-an-ad versus the control (ads without the figurative execu-tion). But we are still uninformed about the effect ofdifferent levels of incongruity in rhetorical figures (visualmetaphors in our study) on ad evaluations. Second, theprocess by which incongruity in visual metaphors leads topositive affect though has attracted theoretical discus-sions; it still has not been empirically examined. There isindeed a need to understand the mechanism by whichincongruity in visual metaphors leads to affect and espe-cially positive affect. Third, in marketing literature, thedegree of incongruity or deviation in rhetorical figures hasbeen measured by scales such as “artful, clever/plain,matter-of-fact,” (McQuarrie and Mick 1992; McQuarrieand Mick 1999) and “meaning openness” (Mothersbaughet al. 2002). In one study, McQuarrie and Mick (1996)used “artful, clever/plain, matter-of-fact” scale as a mea-sure of deviation and complexity in figures but in theirlater study they used the same scale as a measure per-ceived figurativeness of the ad. In these studies (McQuarrieand Mick 1996; McQuarrie and Mick 1999; Mothersbaughet al. 2002), whether incongruity in rhetorical figuresplayed a role in positive attitude-toward-the-ad and otherad evaluations remains an open question.

In marketing academia, the seminal work by Scott(1994a, 1994b) exhorting a need for theory in visualrhetoric propelled some research on visual metaphors, butthe phenomenon of visual metaphors still eludes us. In ourstudy we make an effort to add to the body of research onadvertising and communication by empirically studyingthe phenomenon of incongruity in visual metaphors andaffect. Here I am presenting the conceptual part of ourpaper. In this conceptual paper, first I deliberate on theconcept of incongruity in visual metaphors, after which Iexpand on a process model for the effect of incongruity invisual metaphors on affect and “pleasure of the text” andfinish with the future research and conclusion section.

INCONGRUITY IN VISUAL METAPHORS

Extant research suggests that rhetorical figures canvary on the degree of deviation from expectation orincongruity. Metaphors belong to a class of rhetoricalfigures called tropes3 that are purported to be the mostdeviant of figures (McQuarrie and Mick 1996;Mothersbaugh et al. 2002). Tropes are deviations fromordinary semantic expectations because they are createdby substituting or transferring one meaning with another.They also subsume deficiency of regularity and meaninguncertainty. That is tropes do not explicitly communicatethe entire meaning and have to be resolved in order tofathom their meaning.

With an objective to acquire a thorough understand-ing of incongruity in visual metaphors we studied twotheoretical paradigms. One is the schema incongruitytheory (Mandler 1982) other is the theorizing by Berlyn(1971, 1960) on collative variables.

Schema theory and its extension as schema incongru-ity theory is very popular and “overly applied” in contextof incongruity (Fiske and Taylor 1991; Mandler 1982;Meyers-Levy and Tybout 1989). In social cognition para-digm, schema is a cognitive structure that represents theknowledge and associations about a concept or idea(Fiske and Taylor 1991). It is a framework we use tounderstand, organize, and evaluate the incessant stream ofstimuli. Schemas are therefore representations of experi-ence that guide action, perception, and thought (Mandler1982, p. 3). Schema incongruity occurs when a stimuli orevent does not correspond to or match with the schematicrepresentation occasioned by the stimuli or event at thatpoint. The extent of correspondence between the actualstimuli that is experienced versus the expected or instan-tiated schematic representation is the basis of the degreeof schematic incongruity.

Berlyn’s perspective of incongruity on the other handis couched in the cognitive psychology paradigm. Incon-gruity belongs to the group of collative properties such asnovelty, suprisingness, complexity, etc. (Berlyne 1971;Berlyne 1960). Collative properties in a stimulus engen-der some level of conflict among incompatible responsetendencies of one kind or another. According to Berlyn(1960), incongruity exists when a stimulus induces anexpectation which is disappointed by accompanying stimu-lus. It means that while some elements in a stimuli occa-sion certain responses and expectations, some other ac-companying elements may induce responses that con-flicts with or contradict the expectation set by the previousset. Incongruity is caused by cohabitation of elements,certain elements that cause certain expectation and certainother elements that counter these. For example, in a visualmetaphor of a female’s body with the head of a dinosaur,the female’s body generates an expectation of a female’shead, while the dinosaur’s head evokes the expectation ofa dinosaur’s body. Both these conflicting reactions causea perception of incongruity.

INCONGRUITY IN A VISUAL METAPHOR

We define, incongruity in a visual metaphor as thedegree to which there is the lack of structural and semanticcorrespondence between the source and target elements inthe visual metaphor in the presented context and the pre-existing knowledge structures associated with source andtarget elements in the given context. This definition of theincongruity is informed by both Berlyn’s conceptualizationof incongruity (Berlyne 1971; Berlyne 1960) and the


schema incongruity theory (Fiske and Taylor 1991;Mandler 1982). Therefore, the incongruity in visual meta-phors entails the interplay of the source and target ele-ments ensconced in an individual’s existing structural andsemantic schemas.

One of the explanations for incongruity in visualmetaphors is that the source and target terms are fromdifferent domains and therefore when one tries to map theproperties of the source onto the target it causes theproblem of domain incongruence (Ortony 1979). Thefeatures and attributes that belong to the source domainare semantically and structurally remote from the targetdomain. Therefore, agility and speed of a Puma (animal)(the source term) is not identical to the agility and speedof a shoe (Puma brand) (the target term). Hence, thefeatures or attributes of the source cannot be directlycompared or transferred to the target because the at-tributes are associated with different domains. One canbring about a comparison only by metaphorically alteringthe attributes of the source to be meaningfully compatiblewith the source.

Carroll (1994, p. 198) posits that the consequence oftwo phenomenon simultaneously occupying the samespace or homospatiality, is incongruity. Further, the twophenomena are incompossible. He is of the opinion that itis the combination of homospatiality and incompossiblitythat gives rise to the “apparent falsity” in metaphors. Incontext of metaphors, Burke (1954, p. 90) mentions the“perspective by incongruity,” a perspective that is gainedby constant juxtaposition of incongruous terms that re-veals “unsuspected connectives” and appeals to the audi-ence by exemplifying relationship that conventional formsof communication had ignored. Koestler (1964, p. 35)speaks about metaphors as a part of a creative process ofperceiving a situation or idea in two self-consistent buthabitually incompatible frames of reference.

A PROCESS MODEL FOR AFFECT AND

PLEASURE OF THE TEXT

Incongruity in a visual metaphor renders the ad,incomprehensible at first glance. This causes arousal andstimulates the audience, who makes an attempt to recon-cile the uncertainty. Reconciliation of incongruity-in-duced-uncertainty is via metaphoric comprehension. Theinitial state of incomprehension followed by the subse-quent comprehension of the visual metaphor is called the“aha experience.” The “aha experience” leads to “plea-sure of the text” or positive affect. But comprehension ofvisual metaphor is not a given. Research suggests thatcomprehension of visual metaphor is a complex cognitiveprocess brought about by cross-domain mapping. Asincongruity increases beyond a level, it becomes difficultto apply cross-domain mapping and comprehension fails.

In such a scenario, incongruity in visual metaphor canlead to negative affect.

A PROCESS MODEL: EFFECT OF

INCONGRUITY IN VISUAL

METAPHORS ON AFFECT

There has been considerable theorizing on “collativemotivation” or motivation dependent on collative proper-ties such as novelty, surprisingness, complexity, andincongruity. The basic assumption here is that collativeproperties effect arousal or drive irrespective of context.This arousal would vary depending on the frequency ofexposure and different levels of the collative property.Incongruity, which is a collative property, would cause anarousal that stems from the conflict and uncertainty evokedby visual stimulus.

Incongruity is a property of metaphors and there istheorizing to suggest metaphors elicit tension. Nilsen(1986) proposed that metaphorical process elicits linguis-tic, pragmatic or hermeneutic tension. In context of visualmetaphors, linguistic tension is an outcome of relation-ship of graphic elements in an ad that is unconventional inrelation to graphic conventions. Pragmatic tension iscaused when the depiction of objects violates their usualform in reality. And hermeneutic tension is evoked whenthe ad’s depiction of the abstract qualities of the productposes a certain level of challenge one’s belief system(Kaplan 1992). The tension or arousal should increase asincongruity in a visual metaphor increases.

There are two routes that account for the effect ofincongruity of visual metaphors on affective evaluations(Figure 1). One is the perceptual route and other is thecomprehension route. The perceptual route can be under-stood in terms of arousal boost concept and comprehen-sion route can be understood in terms of arousal lagconcept introduced by Berlyn (1971, 1960) in context ofcollative variables. Berlyn (1960) purported that botharousal-increasing and arousal-mitigation systems can belead to positive hedonic value. If one is exposed to a visualmetaphor for a very brief period so that semantic levelprocessing is inhibited, the incongruity experienced wouldbe at a perceptual level. If one could increase incongruityfrom low to high, beyond a threshold level, the hedonicvalue will increase and reach a peak when the arousalpotential is at a moderate level. After which there is adecline in hedonic value and at high level of arousalpotential, the hedonic value may become negative. There-fore, arousal-increasing system or arousal boost has anon-monotonic relationship with hedonic value.

Conflict and uncertainty are aversive states and anorganism tries to reduce or alleviate an aversive state andget back to a more comfortable state. States of high


uncertainty and conflict are states of disequilibrium andare unstable. In fact there is research to suggest thatincongruity evokes exploratory behavior so as to make upfor the lack of information and reduce uncertainty. Sub-jects spend longer time perusing a stimulus high onincongruity versus low on incongruity (Berlyne 1963).The reduction of response uncertainty to below a thresh-old value is termed as conflict resolution. The removal orattenuation of the source of discomfort is supposed to berewarding. This is the arousal jag or arousal-mitigatingphase that also contributes to the positive hedonic valuegenerated from a stimulus. In context of visual metaphor,the conflict resolution is brought about by the process ofcomprehension. Comprehension quenches the stimula-tion evoked by the incongruity in the visual metaphor.

The two routes can be reconciled to understand theoverall effect of incongruity in an ad on advertisingevaluations. As is evident from Table 1, low incongruity

evokes mild stimulation. Because incongruity is low,comprehension of the visual metaphor will require littleeffort and hence the affect experienced post comprehen-sion will also be mildly positive. Moderate level of incon-gruity in visual metaphors will elicit most pleasant level ofarousal or stimulation. One would expect that in case ofmoderate level of incongruity, comprehension can beachieved with some effort. In effect conflict resolution incase of moderately incongruous visual metaphors postcomprehension will lead to moderate positive affect. Incase of high incongruity, the hedonic value due to arousalwill be negative. The comprehension route can lead to twopossible outcomes: one in which perceived comprehen-sion takes place and other in which perceived comprehen-sion does not take place. If the metaphor cannot becomprehended, it will lead to moderate to high negativeaffect due to frustration and disappointment. What oneneeds to appreciate in case of a highly incongruousmetaphor, in which the terms of the metaphor cannot be

FIGURE 1

A Conceptual Process Model for the Effect of Incongruity of Visual Metaphors on Affect

Incongruity in visual

metaphors

Comprehension Affect

Perceptual route

Comprehension route

TABLE 1

Relationship of Incongruity in a Visual Metaphor and Affect

Perceptual Route Comprehension Route

Low Incongruity Mild Positive Affect Mild Positive Affect

Moderate Incongruity Moderate Positive Affect Moderate Positive Affect

High Incongruity Mild to Moderate Negative Affect EitherMild to Moderate Positive AffectorModerate to High Negative Affect


easily reconciled, is that even if comprehension takesplace, an individual will be uncertain about the level ofcomprehension and hence the positive affect experiencedwill be compromised. Hence high incongruity in visualmetaphors may lead to only mild to moderate positiveaffect. Overall, by reconciling the effect of the two routes,one can deduce that moderate level of incongruity shouldlead to more favorable evaluations than either high or lowlevels of incongruity.

H1: Incongruity in visual metaphors will have a non-linear relationship with affect. Moderate level ofincongruity versus low or high level of incongruity ina visual metaphor will lead to higher positive affect.

The Aha Experience and Pleasure of the Text

“Aha experience” is a transition from a state ofincomprehension to that of comprehension of the stimu-lus. “Aha experience” represents the change from a stateof incomprehension to a state of comprehension. Rhetori-cal figures are indirect form of communication. Hence thecomprehension of rhetorical figures unlike literal com-prehension is not clear at a glance. Metaphors, a type oftrope are supposed to be the most deviant of rhetoricalfigures. Metaphors are undercoded and lack informationand hence the reader gets motivated toward elaboration ofthe text so as to achieve closure. The cohabitation of twoterms in a visual metaphor that are incompossible andinconsistent, renders the message incomprehensible to theaudience at first instance. Hence the reader has to make aneffort toward understanding the visual metaphor. Thesubsequent resolution of meaning of the metaphor elicits“pleasure of the text.”

The “aha experience” is marked by a pre-experiencecharacterized by some factors connected with threat,discomfort, uncertainty, or arousal and a post-experiencecharacterized by some factors that signify safety, read-justment, clarification, or release. In physiological sensea rewarding event such as comprehension after someeffort, should produce a change in arousal level. Becausean organism is active and disturbed before it receives thereward. The onset of a reward produces a change to

quiescence and tranquility (Berlyne 1960). Therefore,“aha experience” would be the point at which there is achange in hedonic value. It can be understood as a pointof inflection on a curve.

Because “aha experience” takes place only when theinitial incomprehension is resolved, hence comprehen-sion should occur for “aha experience” to take place(please refer to Figure 2). “Aha experience” can be thoughtof as the “click” of comprehension. So “aha experience”would depend on both the initial degree of incomprehen-sion and the subsequent degree of comprehension of thevisual metaphor. As incongruity increases from low tohigh, the initial perceived incomprehension should in-crease. Simultaneously as the incongruity increases, thelikelihood for perceived comprehension to occur aftersome effort-after-comprehension should decrease. Mod-erate level of incongruity will yield the most favorablelevel of “aha experience.” Therefore, when incongruity islow, the incomprehension would be mild and comprehen-sion should not be challenging and hence “aha experi-ence” should be of low intensity. Moderate incongruitywill lead to higher level of perceived incomprehension onfirst exposure to the visual metaphor. Comprehensionshould be possible after some effort and hence a readershould experience higher level of “aha experience.” Butwhen the metaphor is highly incongruous, it becomes verydifficult to accomplish cross-domain mapping and recon-cile the meaning of the metaphor. In case of highlyincongruous visual metaphors, even if one were able toaccommodate or assimilate the visual metaphor, one maynot be certain of one’s comprehension. Hence, “ahaexperience” should reduce in case of high or extremeincongruity in visual metaphors.

H2: Incongruity in a visual metaphor will have a non-linear relationship with the “aha experience”

The “aha experience” is one explanation for the“pleasure of the text.” Pleasure of the text is the positiveaffect that a reader experiences due to the reader’s per-ceived successful comprehension of an initially incom-prehensible text. Metaphors and other rhetorical figuresare deviations from expectation. The initial incongruity

FIGURE 2

A Process Model for Pleasure of the Text

Incongruity in Visual Metaphor

Aha Experience Pleasure of the Text

+

H2 H4


FIGURE 3

Visual Metaphors Manipulated to Different Levels of Incongruity


leads to incomprehension, which is unsettling and ensu-ing comprehension and resolution of incongruity is re-warding. The “aha experience” contributes to the “plea-sure of text.” Greater is the intensity of “aha experience”or click of comprehension, the greater should be the“pleasure of text” experienced. Pleasure relates to “safety,readjustment, clarification, or release” (Berlyne 1960,p. 258). It is the reward from getting over an aversivesituation. It is the thrill one gets from solving a puzzle(Peracchio and Meyers-Levy 1994).

H3: The “aha experience” will mediate the relationshipbetween incongruity of visual metaphor and pleasureexperienced from a visual metaphor.

H4: The “aha experience” will have a linear relationshipwith pleasure experienced from a visual metaphor.

FUTURE RESEARCH AND CONCLUSION

To empirically study our research questions first, twosets of ads using visual metaphors were created with thehelp of a professional ad executive. In the choice of theproducts for the ads, two points were borne in mind. One,the products should be ones that most subjects should befamiliar with, so that knowledge factors do not confoundthe study variables. Two, the products should be such thatwe can come up with different source terms that can beused to transfer a particular attribute to the product. Aftermuch cogitation, we converged on the idea of a plus-sizeline of shirts called Big John shirts and an energy drinkcalled Z energy drink. For Big John shirts (please refer toFigure 3 on the previous page), the desired attribute to bemapped onto the shirt was “bigness.” We used “a giant,”“a whale,” “an elephant,” “a bus,” and “a mountain” assource terms for the target “Big John shirts.” In the Zenergy drink ads, the desired attribute was “electrifyingenergy.” To transfer the association of electrifying en-ergy, we used “a lightening,” “a plug point,” “electrons inan atomic model,” and “an electricity danger sign” assource terms for the target, i.e., Z energy drink. Therefore,we created five ads for Big John shirts and four ads for Zenergy drinks.

A pre-test was conducted with the objective to cali-brate the ad stimuli as low, moderate, and high incongru-ity visual metaphors. Due to constraints on availability ofsubjects, we used a within-subjects experimental design.Incongruity is reflected in perceived deviation from ex-pectation and hence a within-subjects design could posethe problem of carry-over effects. Hence, we used a Latin-square design so that carry over effect of sequence of thead exposure could be factored out. Subjects were ran-domly assigned to a within-subjects 4X4 Latin-squaredesign, in case of Z energy drink and a within-subjects5X5 Latin-square design, in case of Big John shirts.

Incongruity was measured by the “Very unexpected –Very expected” and “Very irrelevant – Very relevant”(Heckler and Childers 1992), 11-point semantic differen-tial scales. Also we had items such as “Very similar toother ads in this product category – Very different fromother ads in this product category,” “very novel – not at allnovel,” “simple – complex”; all rated on 11-point seman-tic scales for manipulation check. Further, we measuredthe extent to which the visual aroused uncertainty andconflict (Berlyne 1960). The pretest showed significantmean differences on incongruity and comprehensionmeasures between the desired pairs of stimuli.

After the pre-test, we have collected the data for thefirst study and analyses are currently underway. The studyobjective is to test the effects of low, moderate, and highlevels of incongruity on ad evaluations. The key depen-dent variables are the “pleasure-arousal” scale byMehrabian and Russell (1974); pleasantness scale (Berlyne1963); attitude-toward-the-ad; attitude-toward-the-brandand an ad hoc measure for pleasure of the text. Wemeasured the “pleasure of the text” with an 11-pointLikert scale anchored with “strongly agree – stronglydisagree.” The items in the scale are “I liked the processof going through the ad. The process of going through thead gave me a pleasant feeling / I enjoyed the process ofgoing through the ad.” After the main study, we proposeto conduct a computer-based study, with a new set of adsto investigate the “aha experience.”

Visual metaphors are an extensively used phenom-enon in advertising and brand communication (Phillips2003). They are especially useful in transferring abstractproperties to a brand. This paper proposes to make a fewtheoretical and practitioner related contributions in thearea of advertising. First, we are making an attempt toempirically examine the relationship of incongruity invisual metaphors and affect. Second, we would like tofathom the process by which incongruity in visual meta-phors leads to affect. Of particular interest to us is themechanism that leads to positive affect and thereforepositive attitude-toward-an-ad and brand. Our researchunderscores the role of comprehension as a mediatorymechanism to positive ad evaluations. These are ques-tions and issues that are still unanswered in the currentmarketing literature.

Notably, though the study focuses on visual meta-phors, the findings from this study can be applied to visualrhetorics. Incongruity is a characteristic of rhetoricalfigures in general. Rhetorical figures are called artfuldeviations (Corbett and Connors 1999; McQuarrie andMick 1996). Hence, the understanding drawn from thephenomenon of incongruity in visual metaphors and af-fect can be applied to other rhetorical figures. McQuarrieand Mick (1996) placed the various rhetorical figures


along a gradient of incongruity and complexity. Accord-ing to their framework, metaphors are most deviant offigures. Hence, one must bear their framework in mind,

when applying the results of our study to other rhetoricalfigures. One must do so metaphorically!

ENDNOTES

1 Etymology of “rhetoric” is “words” or “speech.” Arhetorical figure also called figures of speech is ageneric term for an artful deviation from ordinarymode of speaking or writing.

2 Text is a term derived from Semiotics that refers to amessage or assemblage of signs that is physicallyindependent of the sender or the receiver. It followsconventions associated with the genre and is relayed

in a particular medium of communication.3 Figures of speech or rhetorical figures can be divided

into two broad classes called tropes and schemes. Ascheme such as rhyme, alliteration, etc. is a deviationrelated to arrangement and order. Schemes manifestthrough regularity of some sort that leads toovercoding. A trope such as metaphors, irony, etc. isa deviation related to semantics or meaning. Tropesmanifest in some sort of irregularity that leads toundercoding.

REFERENCES

Barthes, R. (1986), The Rhetoric of the Image. Oxford,England: Blackwell.

Berlyne, D. (1963), “Complexity and Incongruity Vari-ables as Determinants of Exploratory Choice andEvaluative Ratings,” Canadian Journal of Psychol-ogy. 17 (3), 274–90.

Berlyne, D.E. (1960), Conflict, Arousal, and Curiosity.New York, Toronto, London: McGraw-Hill BookCompany.

Berlyne, Daniel E. (1971), Aesthetics and Psychobiology.New York: Appleton-Century-Crofts.

Burke, Kenneth (1954), Permanence and Change. NewYork: The Bobbs-Merrill Company, Inc.

Carroll, Noel (1994), Visual Metaphor. Dordrecht: Kluwer.Corbett, Edward P.J. and Robert J. Connors (1999),

Classical Rhetoric for the Modern Student. NewYork, Oxford: Oxford University Press.

Fiske, Susan T. and Shelley E. Taylor (1991), SocialCognition, 2nd ed. New York, England: McGraw-HillBook Company.

Heckler, Susan E. and Terry L. Childers (1992), “TheRole of Expectancy and Relevancy in Memory forVerbal and Visual Information: What Is Incongruen-cy?” Journal of Consumer Research, 18 (March),475–92.

Kaplan, Stuart J. (1992), “A Conceptual Analysis of Formand Content in Visual Metaphors,” Communication,13, 197–209.

Kittay, E.F. (1987), Metaphor: It’s Cognitive Force andLinguistic Structure. New York: Oxford UniversityPress.

Koestler, Arthur (1964), The Act of Creation. New York:Macmillan.

Lakoff, G. and M. Johnson (1980), Metaphors We Live

By. Chicago: University of Chicago Press.MacKenzie, Scott B., Richard J. Lutz, and George E.

Belch (1986), “The Role of Attitude Toward the Adas a Mediator of Advertising Effectiveness: A Test ofCompeting Explanations,” Journal of Marketing Re-search (JMR), 23 (2), 130–43.

____________ and ____________ (1989), “An Empiri-cal Examination of the Structural Antecedents ofAttitude Toward the Ad in an Advertising PretestingContext,” Journal of Marketing, 53 (2).

Mandler, G. (1982), The Structure of Value: Accountingfor Taste. Hillsdale, NJ: Lawrence Erlbaum Associ-ates.

McQuarrie, Edward F. (1992), “On Resonance: A CriticalPluralistic Inquiry into Advertising Rhetoric,” Jour-nal of Consumer Research, 19 (2), 180–97.

____________ and David Glen Mick (1996), “Figures ofRhetoric in Advertising Language,” Journal of Con-sumer Research, 22 (4), 424.

____________ (1999), “Visual Rhetoric in Advertising:Text-Interpretive, Experimental and Reader-ResponseAnalyses,” Journal of Consumer Research, 26 (1),37.

____________ and D.G. Mick (2003), The Contributionof Semiotic and Rhetorical Perspectives to the Expla-nation of Visual Persuasion in Advertising. Mahwah,NJ: London: Lawrence Erlbaum Associates.

Mehrabian, Albert and James Russell (1974), An Ap-proach to Environmental Psychology. Cambridge,MA: The MIT press.

Meyers-Levy, Joan and Alice M. Tybout (1989), “SchemaCongruity as a Basis for Product Evaluation,” Jour-nal of Consumer Research, 16 (1), 39.

Mick, D.G. (1992), “Levels of Subjective Comprehen-sion in Advertising Processing and Their Relations toAd Perceptions, Attitudes, and Memory,” Journal of


Consumer Research, 18 (March), 411–24.Mothersbaugh, David L., Bruce A. Huhmann, and George

R. Franke (2002), “Combinatory and Separative Ef-fects of Rhetorical Figures on Consumers’ Effort andFocus in Ad Processing,” Journal of Consumer Re-search, 28 (4), 589-602.

Nilsen, Don L.F. (1986), “The Nature of Ground inFarfetched Metahpors,” Metaphor & Symbolic Activ-ity, 1 (2), 127–38.

Ortony, A. (1979), “Beyond Literal Similarity,” Psycho-logical Review, 86, 161–80.

Peracchio, L.A. and J. Meyers-Levy (1994), “How Am-biguous Cropped Objects in Ad Photos Can AffectProduct Evaluations,” Journal of Consumer Research,21, 190–204.

Phillips, Barbara J. (2003), Understanding Visual Meta-phor in Advertising. Mahwah, NJ: Lawrence Erlbaum

Associates.Scott, L.M. (1994a), “Images in Advertising: The Need

for a Theory of Visual Rhetoric,” Journal of Con-sumer Research, 21, 252–73.

____________ (1994b), “The Bridge from Text to Mind:Adapting Reader-Response Theory to Consumer Re-search,” Journal of Consumer Research, 21, 461–80.

Stern, B.B. (1990), “Feminist Literary Criticism andDeconstruction of Ads: A Postmodern View of Ad-vertising and Consumer Responses,” Journal of Con-sumer Research, 19, 556–66.

Tourangeau, Roger and Robert J. Strernberg (1981),“Aptness in Metaphor,” Cognitive Psychology, 13,27–55.

Tversky, A. (1977), “Features of Similarity,” Psychologi-cal Review, 84, 327–52.

For further information contact:Praggyan Mohanty

MarketingCollege of Business

433 Cornell HallUniversity of Missouri – Columbia

Columbia, MO 65211Phone: 573.882.4716

E-Mail: [email protected]


MEASURES OF EMOTIONAL RESPONSE TO

TELEVISION ADVERTISING

Anca Cristina Micu, Sacred Heart University, Fairfield

SUMMARY

Emotions, Feelings, and Advertising Research Mea-

sures

According to the dictionary definition, emotion is amental state that arises spontaneously rather than throughconscious effort and is often accompanied by physiologi-cal changes, while a feeling is an affective state of con-sciousness resulting from emotions. Taking a closer lookat the consumer’s mind, neuroscientists and psycholo-gists label that first emotional reaction as either “mentalmodels” (Johnson-Laird 1981; LeDoux 1996) or “situa-tion models” (Zwaan and Radvansky 1998; Zwaan 1999),terms used to describe the structures of meanings thatexist within our brains. According to Wyer andRadvansky’s (1999) theory of social comprehension,people spontaneously construct a mental simulation inorder to comprehend an event that involved a particularindividual at a particular time and space. Such mentalsimulations are called event models and they impact affect(i.e., feelings) and cognition as well as judgments anddecision making. Mental models in our brains are con-stantly updated with information we either learn or makeup. We revisit existing meanings within our brains when-ever certain triggers are present and there is no establishedgeneral pattern for this constant maintenance of our knowl-

edge structures (Zaltman 2003). Therefore, people do notreact to the environment in a linear fashion (e.g., attention-interest-desire-action, think-feel-do). Each of our actionsmay be determined by a different set or pattern of thoughtsand feelings, every time. In addition, the first reaction toany stimulus is an emotion, not necessarily one of whichwe are consciously aware, followed by a mesh of thoughtsand feelings (LeDoux 1996).

Since our interest here is advertising research, wefind that the new knowledge about the human brainrelating to the primacy of emotions and the importance ofthe unconscious is pointing to an outdated paradigmbecause we are learning that consumers do not makedecisions in a rational or linear fashion. Hence, in additionto traditional measures of consumer thought there is aneed for information about consumer’s emotional andsubconscious reactions. We differentiate between sponta-neous and prompted thoughts and point to the pitfall ofusing verbal measures alone when researching responsesto advertising.

Next, we present the findings from an exploratorystudy where five advertising research companies agreedto employ their individual methods for measuring emo-tional responses to the same television commercials.

FIGURE 1

Measurement Methods and the Types of Reactions They Measure


The Study

We sampled five advertising research companies inan effort to span methodologies from the three types ofmeasures. The companies agreed to employ their indi-vidual methods to measure emotional responses to thesame four beer television commercials. We included threephysiological measures (SC, HRT, and facial EMG),symbolic measures (ZMET), and one of each category ofself-report measures (verbal, visual, and moment-to-mo-ment). We set out to explore differences among emotionalreactions recorded using physiological and self-report(quantitative) methods and then relate our findings to theinterpretations of advertisement meanings provided bythe symbolic (qualitative) measure.

While broadly comparing the three measure types,the study’s exploratory findings warrant further investi-gation of the specific measures under each type in order todelineate the finer differences among them. For example,it is likely not all physiological measures may measure theexact same kind or level of emotional reaction. First, welooked at the trend lines of the physiological, visual, andmoment-to-moment self report normalized responses tothe four commercials. Second, we looked at the qualita-tive interpretations and representations of ad meaning thatresulted from the ZMET technique.

Discussion

While self-report measures especially verbal self-report have been used commercially for a long time in

advertising research, physiological and symbolic mea-sures of advertising effectiveness are not employed on alarge scale. It is possible the industry’s left-brain bias isinhibiting the intuitive and creative side of advertising andbrand communication. Self-report measures shed light onthe conscious interpretation of reactions to the commer-cials (i.e., what I think I felt during/after seeing thecommercial). Self report measures based on subjectivefeelings may not always be able to capture lower-orderemotions in an accurate way, although these lower-orderemotions may have a substantial influence on consumerdecisions. Whilst self-report measures are more likely toevaluate commercials on a pre-determined scale of de-scriptors, symbolic measures leave the door open forpainting a picture of ad response from the variety of itemsautomatically triggered in the brain by the exposure.Symbolic measures point to specific items from the innerworkings of the brain that were triggered by the commer-cial. Physiological measures provide the confirmation ofthe initial emotional arousal and of the emotional peaksthat are then reflected in both symbol generation anddescriptor evaluations. The use of the various methodswill probably differ depending on the information wanted,from looking to sketch a personality of a brand as pre-sented in the commercial when consistent brand commu-nication is desired, to identifying those frames of a com-mercial that definitely should not be cut when editing, tofocusing on utilizing ad time most effectively to generatepreferred patterns of emotional reactions. References areavailable upon request.

For further information contact:Anca Cristina Micu

Welch College of BusinessSacred Heart University

5151 Park AvenueFairfield, CT 06825

Phone: 203.339.1286E-Mail: [email protected]


THE EFFECTS OF SYNTACTIC STRUCTURE VARIATION ON

CONSUMERS’ MEMORY FOR PRINT ADVERTISEMENTS

Hieu P. Nguyen, California State University, Long BeachJames M. Munch, Write State University, Dayton

SUMMARY

Researchers have used a psycholinguistic frameworkto study the effects that linguistic factors of ad copies haveon consumers’ memory for and comprehension of the ads.However, absent in the literature is research that looks intothe effects of varying the advertising claim structure(claim-evidence versus evidence-claim) on consumers’memory for the ad. It appears that in print advertising ahighly popular format of ad claims is the claim-evidencestructure (e.g., “NO MORE LONG HOURS OF READ-ING – the QUICK-LINK Pen® Elite allows you to imme-diately enter text and number into any applications: PC,PDA, or Smartphone”). However, it remains unclearwhether this is the optimal format in terms of audience’smemory for the ad.

The effects of syntactic structure on readers’ memoryand comprehension of texts have mostly been examinedin education research. Researchers (e.g., Irwin 1980)found that an order reflecting the temporal order of theevents (left-branching: Because X, Y, e.g., “Because thecolonists didn’t like the laws, they didn’t help makethem.”) was related to a higher level of comprehensionthan a reverse temporal order (right-branching: X, be-cause Y, e.g., “The colonists didn’t like the laws, becausethey didn’t help make them.”). A common theoreticalaccount for this effect is that in the left-branching format,the order in which the clauses are presented matches upwith the general cause-before-effect ordering that mostpeople prefer. However, Irwin and Pulver (1984) alsofound that for younger individuals (schoolchildren in 3rd

and 5th grades) the left-branching structure (Because X,Y) was more difficult to comprehend than the right-branching structure (X, because Y). The authors sug-gested that this might be because when children readabout a situation, they are more likely to ask themselveswhy it occurred than to ask themselves what it may havecaused. Perhaps they expect the new information to beexplained in more depth because it is temporarily thediscourse topic in their working memory.

Overall, these studies suggest that for individualswith limited working memory capacity (e.g., young chil-dren), sentences with a right-branching structure result inbetter memory and comprehension than sentences with aleft-branching structure. However, for people with nor-mal working memory capacity, sentences with a left-

branching structure (normal causal order) seem to berelated to better comprehension and memory than sen-tences with a right-branching structure (reverse causalorder), presumably because they tended to expect aneffect to follow the event in working memory (Irwin andPulver 1984).

In print advertising, the equivalent of a right-branch-ing structure is the highly popular claim-evidence format(e.g., “ZIPCONNECTTM IS CONVENIENT – the spooled,retractable cable in ZipConnectTM modules extends up to18 inches.”). In this example, an implicit connective(hyphen) replaces the explicit connective (“because”). Ifthe ad claim is to be reformatted as a left-branchingstructure, it will read as follows: “The spooled, retractablecable in ZipConnectTM modules extends up to 18 inches –ZIPCONNECTTM IS CONVENIENT,” a reversal of thepopular claim-evidence format. In accordance with find-ings in psycholinguistic research, we predict that adclaims with a left-branching structure will result in betterrecall and recognition than ad claims with a right-branch-ing structure (H1). In addition to this main effect, it islikely that motivation to process the ad (ad involvement)moderates the effect of claim structure variation onmemory. The Elaborative Likelihood Model (Petty andCacioppo 1986) suggests that high involvement with theproduct, message, or purchase decision tends to lead tostrong attention focused on central, product-related fea-tures and factual information, which results in consciousthoughts about product attributes and usage outcomes. Onthe other hand, low involvement leads to limited attentionfocused on peripheral, nonproduct features and feelings,which results in low or nonconscious information pro-cessing. In line with the ELM model, we predict that claimstructure variation will be more strongly related to memoryfor consumers low in involvement than for those high ininvolvement (H2). Both hypotheses (H1 and H2) weresupported in Study 1. We found that under low involve-ment, participants in the evidence-claim (left-branching)condition recalled more ad claim information than partici-pants in the claim-evidence (right-branching) condition.They also had a higher recognition rate than participantsin the claim-evidence condition. However, in the highinvolvement condition, this advantage of left-branchingstructure disappeared: the difference in terms of recall andrecognition between left- and right-branching partici-pants was insignificant, supporting H2.


Results from Study 1 suggest that contrary to thecommon practice in the advertising industry (claim-evi-dence format), advertising claims with left-branchingstructures (evidence-claim) result in better memory forthe ad than those with right-branching structures. How-ever, this advantage was eliminated when consumerswere highly motivated to process the ad information. Wewanted to confirm that this was caused by high involve-ment and not product unfamiliarity (because in Study 1,subjects were relatively unfamiliar with the stimulusproduct, which was a scanner/note taker). In Study 2 wecreated a high involvement condition with a familiar

product (a Global Positioning System device or GPS). Wefound that under high involvement, claim structure varia-tion had no effect on people’s memory for the ad andproduct familiarity did not have any significant effects onsubjects’ memory. Taken together, these results show thatwhen the motivation to process the information is high,familiarity with the product is not a moderating factor ofthe effects of claim structure variation on memory. Theadvantageous memory effect of a left-branching structureover a right-branching one was eliminated by high in-volvement rather than participants’ unfamiliarity with thestimulus. References are available upon request.

For further information contact:Hieu P. Nguyen

Department of MarketingCalifornia State University, Long Beach

1250 Bellflower BoulevardLong Beach, CA 90840Phone: 562.985.7132

Fax: 562.985.1588E-Mail: [email protected]


CHANGES STRATEGIES AND AMBIGUOUS ROLES: MANAGING

FRONTLINE PERFORMANCE AND PSYCHOLOGICAL

WELL-BEING IN FAST MOVING SERVICE

ORGANIZATIONS

Jun Ye, University of Oregon, EugeneDetelina Marinova, University of Missouri, Columbia

Jagdip Singh, Case Western Reserve University, Cleveland

SUMMARY

To survive and remain competitive, service organiza-tions must respond with increasing speed and agility tofast-moving markets due to continuous changes in serviceeconomies. Developing effective change strategies insuch an environment is fundamentally disruptive. To alarge degree, effectiveness of change strategies in fastmoving service organizations rests on understanding howfrontline employees perceive and respond to change, andhow their performance and well-being is affected asresult. This is because a frontline employee is the last andcrucial link to customers in the implementation of top-down change strategies.

Change strategies are especially challenging whenthey involve shifting between competing strategic em-phases, such as from a dominant emphasis on cost con-tainment to an overarching emphasis on revenue enhance-ment or vice-versa (Rust, Moorman, and Dickson 2002).Managing such change necessitates disruption of estab-lished organizational practices and frontline roles. Acrossa range of service industries, charting a course betweenrevenue and cost emphasis is a critical implementationchallenge for market driven organizations.

Juxtaposing the crucial role of the frontline employee(FLE) in change implementation with shifts between cost

and revenue strategic emphasis, we designed this study toexamine: (1) the impact of cost- and revenue-dominantstrategies on FLE’s change perception, (2) the disruptiverole of FLE’s change perception in fostering role ambigu-ity, and (3) its implications for FLE performance andpsychological well-being. Drawing from role stress theory,we isolate the negative effect of change from its functionaleffects by including mediation mechanisms governed byrole ambiguity (see Figure 1).

The conceptual framework was empirically testedusing survey data from 843 RNs and LPNs in five hospi-tals in the Northeastern United States. After measurementpurification and confirmation, the model was tested usingstructural equation modeling. In estimating the hypoth-esized coefficients, we modeled common method biasfollowing procedures outlined by Lindell and Whitney(2001) and Podsakoff et al. (2003). Specially, we esti-mated a common method factor in which each manifestitem was hypothesized to have an equal loading on themethod factor. We also included the estimated commonmethod factor in each of the structural equations.

The empirical results support all of our hypotheses.Specially, our findings indicate that a cost emphasisstrategy induces a FLE’s change perception, while arevenue enhancement strategy curbs FLE’s change per-ception. FLE’s change perception significantly increases

Revenue Enhancement

FLE Change Perceptions

FLE Role Ambiguity

FLE Service Performance

FLE Psychological

Wellbeing

Cost Containment

+

_

+

_

_

_

+

FIGURE 1

Conceptual Framework


his/her role ambiguity, which negatively affects FLE’sservice performance and psychological well-being. Wealso found out that FLE’s perception of change has apositive direct effect on his/her service performance afterthe negative effect of role ambiguity is controlled.

Overall, we have reasonable evidence to concludethat an organization’s pursuit of market driven strategiesactivates change processes which include both positiveand negative pathways that are substantively distinct and

differentially maintained. These pathways also act simul-taneously to counter each other and are consequential forFLE performance and psychological well-being. The posi-tive, functional mechanism is captured by the direct effectof change perceptions on service performance, and thenegative, dysfunctional mechanism is captured by themediated effect of role ambiguity. Isolating the positiveand negative mechanisms holds the potential to point toricher insights and relevant managerial action.

For further information contact:Jun Ye

University of Oregon1208 University of OregonEugene, OR 97403–1208

Phone: 541.346.2839Fax: 541.346.3341



EVALUATING THE EMPLOYEE’S PERSPECTIVE OF

CUSTOMER DELIGHT

Donald C. Barnes, Mississippi State University, Starkville

SUMMARY

Recently there has been a push to move customers toa more extreme state of satisfaction, referred to as delight.The reason for this is the realization that merely satisfiedcustomers may not be especially loyal (Reichheld 1996).Thus, to ensure loyalty firms should aim to delight.Although the relationship between delight and loyaltyseems intuitive, when evaluated across different studies,the results remain ambiguous. Arnold et al. (2005) foundcustomers now expect to be delighted, implying compa-nies must delight their customers in order to meet cus-tomer expectations and ensure satisfaction. Whereas,other researchers suggest caution in accepting the value ofdelighting the customer (Rust and Oliver 2000).

What seems apparent from previous research is thatthe field is not currently sure how delight should fit into afirm’s strategy. Further exacerbating this situation is thefact that major gaps exist regarding key aspects of delightthat might clarify the costs/benefits of delighting thecustomer. For example, literature does not exist thatcritically evaluates how employees view the concept ofdelight. This seems surprising as previous literature hashighlighted the fact that the front-level employee is inter-mingled in the implementation of a successful serviceencounter (i.e., Bitner, Booms, and Mohr 1994). In thehopes of filling this gap, the following research evaluatesthe employee’s perspective of delight.

Research Method

Data were collected using the critical incident tech-nique (CIT) (Flanagan 1950) which has a rich history inservices marketing literature (i.e., Bitner et al. 1994). Theconcept of delight is still relatively young and no researchhas evaluated delight from the employee’s perspective,thus CIT appears to be an ideal method. The sampleconsisted of 124 non-student service workers recruited by31 students. Respondents were asked to complete a stan-dardized form consisting of four open-ended questionsregarding a recent experience in which they felt they haddelighted the customer.

Results

Employee Generated Examples of Delightful Inci-dents. The first question on the survey addressed in-

stances in which employees felt they had delighted thecustomer. From the respondent recollections five catego-ries emerged as reasons for the delightful encounter:accurate service performance (51%), employee effort(27%), complimentary offering (11%), benevolence (8%),and service failure recovery (2%).

Categories of Employee Affect When ProvidingDelight. Research has evaluated how employee moodeffects service performance evaluation (Kelley andHoffman 1997), but does not exist that evaluates howemployees feel when they provide higher levels of ser-vice. From this research, four categories surface: delightcontagion whereby a ripple effect of positive emotionsflow from the customer to the employee (57%); a sense ofaccomplishment for the employee (33%); indifferencetoward positive or negative outcomes (6%); and excite-ment over the possibility of future benefits (4%).

Potential Problems with Providing Delight to theCustomer. Leading scholars in the services field havewarned businesses of the potential problems associatedwith attempting to delight the customer. These includeincreased expectations (Rust and Oliver 2000), decreas-ing returns to the service firm (Ngobo 1999), and highcosts (Berman 2005). Although similar concerns werevoiced by the employees in this sample, the largest cat-egory represented the belief that no problems existed withdelighting the customer (39%). The remaining categoriesincluded employee concerns of the possibility: of elevatedexpectations for the customer (23%); too much effortbeing required (20%); of encouraging negative customeractions (12%); that the only way to delight was to engagein negative employee actions (3%); of a strained relation-ship between the employee and the customer (2%).

Reward Employee Received for Delighting the Cus-tomer. The goal of the last question was to discover howemployees are rewarded after they provide delight. Thequestion was purposely ambiguous so that employeescould state either monetary or non-monetary rewards. Insupport of previous management findings whereby intan-gible rewards are especially meaningful, a non-monetaryreward was the most cited reward (39%). Future benefits(22%), and explicit monetary reward (20%) accounted fora majority of the remaining incidents. While praise (8%),none (7%), and improved relationships (5%) completedthe categories.


Conclusion

In recent years, reasonable arguments have beenmade by leading scholars against providing delight for anumber of reasons which include: costs and decreasingmarginal returns. However, there continues to be a grow-ing contingent of scholars and practitioners who see valuein providing extreme forms of satisfaction to the cus-tomer.

From examining delight from the employee perspec-tive, four themes have emerged when considering delightas a service strategy. First, a disconnect may exist betweenwhat employees and consumers perceive as delight. Thefact that 51 percent of the incidents in this research can becharacterized as “regular service performance” illustratesthis fact. Thus, firms need to train and educate theiremployees as to what constitutes delight. Second, em-ployees connect their self-concept to the services theyprovide and the environment in which they function. Thisresults in a desire to provide delight to the consumerthereby maximizing the employee’s “appearance.” Third,

employees have some concerns regarding the potentialnegative reactions from providing delight. This is relatedto the second theme, in that employees do not want to risklowering their “appearance” in the servicescape. Fourth,is a phenomenon whereby a two-way delight contagionexists between the employee and the customer. Previ-ously shown as a one-way relationship, this finding seemsto explain why certain employees can continually main-tain high levels of service. This phenomenon also repre-sents a factor which is difficult to quantify or model.Because of this difficulty, it is not clear if previous modelshave included this variable.

With increasing competition, it has become increas-ingly important to provide elevated levels of satisfactionto encourage consumer loyalty. However, recent researchhas challenged the benefits of setting delight as a strategicgoal. In the current research, there have been severalbenefits to this strategy that may not have been elucidatedin previous research models, and thus must be accountedfor before delighting as a strategy is eliminated.

REFERENCES

Arnold, Mark J., Kristy E. Reynolds, Nicole Ponder, andJason E. Lueg (2005), “Customer Delight in a RetailContext: Investigating Delightful and Terrible Shop-ping Experiences,” Journal of Business Research, 58(8), 1132–45.

Berman, Barry (2005), “How to Delight Your Custom-ers,” California Management Review, 48 (1), 129–51.

Bitner, Mary Jo, Bernard H. Booms, and Lois A. Mohr(1994), “Critical Service Encounters: TheEmployees’s Viewpoint,” Journal of Marketing, 58,95–106.

Kelley, Scott W. and Douglas K. Hoffman (1997), “AnInvestigation of Positive Affect, Prosocial Behav-iors, and Service Quality,” Journal of Retailing, 73(3), 407–27.

Ngobo, Paul-Valentin (1999), “Decreasing Returns inCustomer Loyalty: Does it Really Matter to Delightthe Customers?” Advances in Consumer Research,26 (1), 469–76.

Reichheld, Frederick F. (1996), “Learning from Cus-tomer Defections,” in Harvard Business Review. 74th

ed.Rust, Roland T. and Richard L. Oliver (2000), “Should

We Delight the Customer?” Journal of the Academyof Marketing Science, 28 (1), 86.

For further information contact:Donald Clay Barnes

Department of Marketing and Quantitative AnalysisMississippi State University

P.O. Box 9588Mississippi State, MS 39762

Phone: 662.325.6725Fax: 662.325.7012



THE INFLUENCE OF SERVICE CLIMATE AND PERSONALITY ON

NURSE CUSTOMER ORIENTATION

Makoto Matsuo, Otaru University of Commerce

ABSTRACT

The purpose of this study is to examine the anteced-ents of nurse’s customer orientation in terms of serviceclimate and their personality traits, using data from nursesof a university hospital. The results indicate that (1)conscientiousness, agreeability, stability, openness to ex-perience is positively related to customer orientation, (2)customer orientation partially mediates the relationshipbetween personality traits and performance, and (3) per-sonality traits accounted more variance for customerorientation than service climate. The author discusses theresults and their implications for service marketing re-searchers and managers.

INTRODUCTION

Considerable number of researches indicates thatmarket oriented firms tend to achieve superior perfor-mance (e.g., Jaworski and Kohli 1993; Im and Workman2004; Kirca et al. 2005; Narver and Slater 1990). Becausethe implementation of the marketing concept is accom-plished through individual service employees (Donavanet al. 2004; Saxe and Weitz 1982), it is critical for firms toenhance customer orientation (CO) of individual serviceemployees in facilitating the firm level market orienta-tion. Thus, clarifying the antecedents of CO is viewed ascritical for building theories on customer oriented man-agement.

Antecedents of individual CO have been examinedfrom various viewpoints including personality traits(Brown, Mowen, Donavan, and Licata 2002; Licta,Mowen, Harris, and Brown 2004), goal orientations (Har-ris et al. 2005), adaptive selling behavior (Franke and Park2006), incentives (Widmier 2002), coworker support(Susskind et al. 2003), service climate (Kelley 1992), andwork environment (Boles et al. 2001). These factors canbe divided into personal factors (e.g., goal orientations,personality traits, and adaptive selling behavior) andsituational factors (e.g., coworker support, service cli-mate, and work environment). Although research consis-tently suggests that behavior is a function of the interac-tion between personal characteristics and situational in-fluences (Kenrick and Funder 1988), little is known aboutthe competing effect of both factors on individual levelCO. In order to investigate the competing effect of per-sonal and situational factors on CO, the present researchfocuses on the role of personality traits and service cli-mate.

A primary goal of this study is to clarify the relativestrengths of personality traits and service climate on COof service workers using data from nurses of a universityhospital. By examining the competing effects of person-ality traits and service climate on CO, we can obtaintheoretical implication on the antecedents of CO. Resultsalso may provide service managers with practical impli-cations on whether they should give high priority torecruiting appropriate employees or changing serviceclimates in the workplace.

Second purpose of this study is to examine the medi-ating role of CO in the relationship between personality,service climate, and performance. There are three possi-bilities on the mediating role of CO (Brown et al. 2002).First is the no mediation model in which personality andservice climate influence performance without mediationby CO. Second is the full mediation model in whichpersonality and service climate affect performance onlythrough CO. Third model is the partial mediation model inwhich personality and service climate have not only directpath but also indirect path through CO on performance.

The article is organized along the following lines.First, literature on CO, service climate, and personalitytraits are reviewed. The research models and hypothesesare then presented. Next, research methods are presented.Finally, data from a survey is examined using structuralequation modeling, followed by a discussion and sugges-tions for future research.

BACKGROUND AND HYPOTHESES

CO and Performance

Saxe and Weitz (1982) was the first study to measureCO at the individual level. They viewed customer-ori-ented selling as the practice of the marketing concept atthe level of the individual salesperson and customer, anddeveloped a 24-item scale, or SOCO scale (The sellingorientation – customer orientation scale) to measure theextent to which a salesperson seeks to increase long-termcustomer satisfaction. CO is associated with high concernfor others, low pressure selling, and need satisfaction/problem solution selling approaches (Saxe and Weitz1982).

On the other hand, Brown et al. (2002) defined CO asan “employee’s tendency or predisposition to meet cus-tomer needs in an on-the-job context,” and developed a


12-item scale with two dimensions: the needs dimensionand enjoyment dimension. It should be noted that Brownet al. (2002) treat CO as a surface-level personality trait,while Saxe and Weitz (1982) conceptualizes CO as abehavioral concept. However, the distinction betweendispositional CO and behavioral CO is not significant,because many personality traits are measured by means ofitems stressing behavioral tendencies (Donavan et al.2004).

Previous empirical studies have reported that COpositively influenced job satisfaction, organizational com-mitment, organizational citizenship behaviors, and per-formance (Boles et al. 2001; Brown et al. 2002; Donavanet al. 2004; Franke and Park 2006; Harris, Mowen, andBrown 2005). Therefore, the following hypothesis isproposed.

H1: Customer orientation is positively related to salesper-sons’ performance.

Personality Traits and CO

The important of personality in services marketingcan be seen in terms of the hierarchical model of person-ality (Mowen and Spears 1999). Brown et al. (2002)distinguished basic personality, or underlying predisposi-tions of individuals arising from genetics and their earlylearning history, and surface traits, or dispositions to actwithin specific situational contexts. In this view, CO istreated as a surface personality trait influenced by basicpersonality.

In research on the relationship between personalitytraits and employees’ performance, the Big-Five factormodel of personality (Digman 1990) has been adopted.The dimensions of the Big-Five factor model are com-monly labeled: (1) extraversion (or introversion), repre-senting the degree to which a person is outgoing or shy, (2)emotional stability, which represents the anxiousness,hostility, and steadiness, (3) agreeability (or agreeable-ness), or general warmth of feelings toward others such astrusting, cooperative, and good-naturedness, (4) consci-entiousness, or the degree of orderliness, organization,and precision; and (5) openness to experience (or creativ-ity), which captures the person’s degree of imagination ororiginality (Brown et al. 2002; Frei and McDaniel 1998).

Table 1 summarizes the past studies examining theeffect of basic personality traits on CO. Although theresults vary among studies, agreeability seems to be acommon personality trait for customer oriented employ-ees across studies. This may be because agreeability ischaracterized by altruism, nurturance, and caring (Bakkeret al. 2006). There are similarities between agreeabilityand CO in terms of high concern for others (Saxe andWeitz 1982). Therefore, the following hypotheses areproposed.

H2: Agreeability is positively associated with customerorientation.

Because the effects of personality traits on CO varyamong studies except for agreeability, it is difficult topredict how the other personality traits (Intraversion,

Frei and

McDaniel Hurley Brown et al. Licata et al

(1998) (1998) (2002) (2003)

Industry Meta-Analysis Food Food Bank Food Healthcare

Personality Traits

Extraversion * *

Emotional Stability * * *

Agreeability * * * * * *

Conscientiousness *

Openness

Note: *means the significant positive relationship with CO.

TABLE 1

Previous Studies on the Relationship Between Personality Traits and CO


emotional stability, conscientiousness, and openness toexperience) influence CO. Thus, the hypotheses regard-ing other personality traits are not proposed here.

Service Climate and CO

Service climate is generally defined as employeeperceptions of the practices, procedures, and behaviorsthat get rewarded, supported, and expected with regard tocustomer service and customer service quality (Schneider,White, and Paul 1998). Service climate is a specificapplication of organizational climate (Dietz, Pugh, andWiley 2004), and it is built in the light of organizationalpractices focused on customer service (Lytle et al. 1988;Salanova et al. 2005).

Past empirical research repeatedly has shown thepositive effect of employees’ service climate perceptionon customer perceptions of service quality, customersatisfaction, and customer loyalty (Dietz et al. 2004;Johnson 1996; Jong et al. 2004; Salanova et al. 2005;Schneider et al. 1998; Schneider and White 2004). Thissort of research has come to be called “linkage research”(Wiley 1996). Several studies reported that organiza-tional citizenship behaviors (OCB), job satisfaction, andorganizational commitment mediated between serviceclimate and customer satisfaction (Paulin et al. 2006;Schneider et al. 2005).

The present research focuses on the relationshipbetween service climate and employees’ CO. Since previ-ous work (Boles et al. 2001; Kelley 1992) reported thepositive relationship between service climate or workenvironments and CO, the following hypothesis is pro-posed.

H3: Service climate positively influences customer ori-entation.

Mediating Models

Considering a hierarchical personality model (Mowenand Spears 1999), it can be predicted that CO will mediatethe relationship between personality traits, service cli-mate, and performance. In this study, the following hy-potheses are proposed.

H4: Customer orientation mediates the relationship be-tween personality traits and performance.

H5: Customer orientation mediates the relationship be-tween personality traits and performance.

There are three possible models regarding serviceclimate, personality, customer satisfaction, and perfor-

mance: no mediation model, full mediation model, andpartial mediation model shown in Figure 1. Model 1, nomediation model, assumes that there are only direct ef-fects of personality traits and service climate on perfor-mance. Model 2, full mediation model, assumes that theeffects of personality traits and service climate on perfor-mance are mediated through customer orientation.Model 3, partial mediation model, allows both direct andindirect effects (mediated through customer orientation)of the personality traits and service climate on perfor-mance. Based on the hierarchical personality model, it islikely that full mediation or partial mediation model fit thedata better than no mediation model.

METHODS

Data was collected from nurses of a university hospi-tal located in Northern Japan. Questionnaires were dis-tributed to all 639 nurses by nurse managers. They wereassured that their individual answers would be held inconfidence. To maximize privacy and minimize bias,respondents completed surveys in sealed envelopes thatwere gathered and returned to a research assistant.

Four hundred thirty-nine completed surveys werereturned for a response rate of 68.7 percent. Of these, 45were unusable because of missing data, which left 396cases for analysis (final response rate is 61.9%). Theaverage nurse experience in my sample was 13.42 yearsand the average tenure at the hospital was 5.55 years.Furthermore, 98.2 percent were women, 100 percent werefull-time employees.

Same scales were used to measure personality traits,customer orientation, and service climate. A factor analy-sis with oblique rotation of the 20 items indicated a five-factor solution (introversion, conscientiousness, instabil-ity, openness, and agreeability). Internal consistencies(Cronbach’s alphas) for the introversion, conscientious-ness, instability, openness, and agreeability were .84, .80,.82, .89, and .81, respectively. As is the Study 1, indexscores (i.e., mean across items) for each construct wereused as single-item indicators in structural equation mod-els.

Factor analysis with oblique rotation revealed a three-factor solution. Factors were named “need to deliver,”“need to read customer’ needs,” and “need for personalrelationship.” Factor structure is almost identical to that ofStudy 2 except for “need to build relationship,” whichrefers employees’ desire to contact and connect customerson a personal level Internal consistencies (Cronbach’salphas) for the need to read customer’s needs (four items),the need to deliver superior service (seven items), and theneed for personal relationship (two items) were .83, .88,


FIGURE 1

Research Models Tested

Model 1. No Mediation Model

Model 2: Full Mediation Model

Model 3: Partial Mediation Model

Introversion

Conscientiousness

Instability

Openness

Agreeability

Service Climate

Customer Orientation

PerformanceRatings

Introversion

Conscientiousness

Instability

Openness

Agreeability

Service Climate


PerformanceRatings

Introversion

Consc ientiousness

Instabili ty

Openness

Agreeabi lity

Service Climate


PerformanceRatings

Introversion

Consc ientiousness

Instabili ty

Openness

Agreeabi lity

Service Climate


PerformanceRatings

Introversion

Conscientiousness

Instability

Openness

Agreeability

Service Climate


PerformanceRatings

Introversion

Conscientiousness

Instability

Openness

Agreeability

Service Climate


PerformanceRatings

and .68, respectively. The need for personal relationshipdimension was eliminated because its Cronbach’s alphaswere below .70. Index scores for each dimension wereused as observable variables for customer orientation instructural equation models.

Internal consistencies (Cronbach’s alphas) forSchneider et al.’s (1998) scale and Jong et al.’s (2004)scales were .84 and .86, respectively. Index scores foreach construct were used as observable variables forservice climate in structural equation models.


Because nursing service is not selling activity, theirperformance was measured by the following three items;“nursing technique,” “nursing assessment,” and “com-munication with patients” using the seven-point scale (1=“among the worst in the company” and 7 = “among thebest in the company”). Performance was based on self-rating. Internal consistencies (Cronbach’s alphas) for thescale were .90. The three items were used as observablevariables for performance in structural equation models.

Questionnaires were presented to the participants inJapanese. Scales originally in English were translated intoJapanese using the back-translation procedure.

RESULTS

Table 2 provides descriptive statistics and pair wisecorrelations between variables. In the following analyses,H1–H3 are tested after examining H4 and H5, because itis necessary to decide the model which is fitted to the data.

To test the mediating role of CO, the fitness of thethree models were examined. Table 3 indicates that partialmediation model (χ² = 128.7, d.f. = 36, GFI = .95, AGFI = .89, CFI = .94, RMSEA = .08) provides better fit for thedata than no mediation model (χ² = 249.32, d.f. = 42,GFI = .90, AGFI = .82, CFI = .87, RMSEA = .11) and fullmediation model (χ² = 173.7, d.f. = 42, GFI = .93, AGFI =.87, CFI = .91, RMSEA = .08). Thus, H4 and H5 aresupported.

The standardized path coefficients of partial media-tion model suggest that the direct paths from some person-ality traits to performance are significant while the directpath from service climate to performance is not signifi-

cant. This indicates that CO fully mediates the serviceclimate – performance relationship, while CO partiallymediates the personality – performance relationship.

To test hypotheses 1–3, structural paths of partialmediation model were examined. Hypothesis 1 predicteda positive influence of CO on performance, and thishypothesis was supported (standardized path coefficient =.36; t = 5.37, p < .01). Hypothesis 2 predicted a positiveinfluence of agreeability on CO was supported (standard-ized path coefficient = .23; t = 4.93, p < .01). Hypothesis 3predicted a positive influence of service climate on cus-tomer orientation, and this hypothesis was supported(standardized path coefficient = .17; t = 3.25, p < .01).

To examine research question that pertains to thecompeting effect of personality traits and service climateon performance, variance accounted for CO by bothfactors was calculated. Personality accounted 16.4 per-cent of variance, while climate accounted only 2.8 per-cent. This means that personality traits explain morevariance for CO than service climate.

DISCUSSION

The primary goal of this study was to investigate theantecedents of CO by examining the competing effects ofservice climate and personality traits on CO using datafrom data nurses of a university hospital. The presentstudy found that (1) conscientiousness, agreeability, sta-bility, openness to experience is positively related tocustomer orientation, (2) customer orientation partiallymediates the relationship between personality traits andperformance, and (3) personality traits accounted morevariance for customer orientation than service climate.

Variable Number of Coefficient Standard

Items Alpha Mean Deviation (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

(1) Introversion 3 .84 3.90 1.37 1.00

(2) Conscientiousness 4 .80 4.38 1.09 -.02 1.00

(3) Instability 5 .82 3.54 1.09 .16 ** -.06 1.00

(4) Openness 5 .89 3.50 1.13 -.23 ** .22 ** -.02 1.00

(5) Agreeability 3 .81 4.92 .97 -.13 ** .25 ** -.12 ** .27 ** 1.00

(6) CO1(needs) 6 .83 4.39 .93 -.13 .30 ** -.15 ** .37 ** .38 ** 1.00

(7) CO2(deliver) 7 .88 4.66 .95 -.10 ** .22 ** -.15 ** .18 ** .38 ** .63 **1.00

(8) SC1 7 .84 3.37 .61 -.02 .06 -.07 .09 .22 ** .21 ** .22 ** 1.00

(9) SC2 6 .60 3.65 .58 -.03 .00 -.09 .07 .22 ** .21 ** .24 ** .65 **1.00

(10) PERI(technique) 1 N.A. 4.14 1.20 -.14 ** .25 ** -.07 .29 ** .14 ** .41 ** .15 ** .08 .06 1.00

(11) PER1(assessment) 1 N.A. 3.96 1.19 -.18 ** .32 ** -.13 ** .33 ** .15 ** .41 ** .18 ** .12 * .07 .80 ** 1.00

(12) PER2(communication) 1 N.A. 3.34 1.14 -.24 ** .26 ** -.14 ** .38 ** .27 ** .50 ** .23 ** .07 .07 .73 ** .71 ** 1.00

TABLE 2

Descriptive Statistics and Correlations


The results indicate how customer oriented behavior isdetermined by personal and situational factors in serviceorganizations. The theoretical implications of the resultscan be summarized as follows.

First, the importance of agreeability in enhancing COcorresponds to previous studies (Brown et al. 2002; Freiand McDaniel 1998; Hurley 1998; Licata et al. 2003).This suggests that agreeability characterized by kind,sympathetic, and tender hearted with others is a corepersonality trait for customer oriented service employeesregardless of service type and national culture. As highcustomer orientation is closely associated with high con-cern for others (Saxe and Weitz 1982), agreeability isrequired for customer oriented service providers.

Second, nurse’s CO is supported by various person-ality traits. This suggests that nurses are required complextasks involving medical assessment, medical treatment,and communication of patients, and they have heavierresponsibility in their tasks compared with other serviceworkers in banks or restaurants. We could say that themore complex the task is, the more personality traits areassociated with CO.

Third, CO is influenced more strongly by personaldisposition (personality traits) than situational factors(service climate). Personality accounted personality ac-counted 16.4 percent of variance, while climate accountedonly 2.8 percent. This suggests that dispositional factor ismore important determinant of customer orientation than

TABLE 3

Results of Structural Equations Analyses

Model 1 Model 2 Model 3

(No Medicaeion) (Full Medication) (Partial Medication)

Standardized Standardized Standardized

Path Path Path

Path Coefficient t-Value Coefficient t-Value Coefficient t-Value

Strudtural Paths

Introversion → performance -.12 -.266 * -.11 -2.55 *

Conscientiousness → performance .24 5.09 ** .22 4.69 **

Instability → performance -.05 -.113 -.03 -.86

Opensess → performance .16 3.39 ** .17 3.40 **

Agreeability → performance -.03 .72 -.06 -1.37

Service climate → performance .01 2.57 -.01 -.13

Customer orientation → performance .26 .24 .51 8.73 ** .36 5.37 **

Introversion → Customer orientation -.03 -.64 -.01 -.40

Conscientiousness → Customer orientation .21 4.45 ** .18 4.02 **

Instability → Customer orientation -.10 -2.30 * -.09 -2.14 *

Openness → Customer orientation .27 5.50 ** .26 5.44 **

Agreeability → Customer orientation .24 5.08 ** .23 4.93 **

Service → climate .18 3.40 ** .17 3.25

Measurement Paths

CO1 (needs) 1.38 Fixed .92 Fixed .97 Fixed

CO2 (deliver) .45 2.92 ** .66 11.36 ** .64 9.77 **

SC1 1.08 Fixed .76 Fixed .73 Fixed

SC2 .59 .32 .85 3.80 ** .85 3.63 **

PER1 .88 Fixed .89 Fixed .89 Fixed

PER2 .87 2.65 ** .88 23.10 ** .88 23.44 **

PER3 .08 19.75 ** .82 20.85 ** .82 20.99 **

Model Fit Statistics

χχχχχ2 249.32 173.70 128.70

d.f. 42.00 42.00 36.00

GFI .90 .93 .96

AGFI .82 .87 .89

CFI .87 .91 .94

RMSEA .11 .08 .08

* p < .05 ** p < . 01

Notes: n = 396. GFI - goodness of fit index. AGFI - adjusted goodness fit index. CFI - comparative fit index. RFSEA - root mean square error of

approximation.


situational factors. The finding indicates that CO shouldbe treated as surface personality traits influenced by basicpersonality traits, and supports the validity of the hierar-chical model of personality proposed by Mowen andSpears (1999).

Finally, CO partially mediates the relationship be-tween personality traits and performance. That is, person-ality traits have not only indirect impact on performancethrough CO but also direct impact on performance. Theresults correspond to Brown et al. (2002), which sup-ported the partial mediation model allowing the directimpact of personality traits on performance ratings. Thismay be because performance was measured by process(or behavior) based scales not output based scales such assales volume or profits.

There are several practical implications derived fromthe present study. First, recruitment of employees withappropriate personality traits is extremely important infacilitating CO of service workers. In selecting nurses,manager should especially pay attentions to candidate’sagreeability, open to experience (creativeness), and con-scientiousness.

Second, nurse managers should understand that ser-vice climate has limited power in enhancing CO of service

workers compared with the influence of personality traitsof workers. However, it is possible that service climateenhances performance mediated through other factorssuch as job satisfaction, organizational commitment, andmotivation of workers.

The limitations of this study should be acknowl-edged. First, this study may be limited by the sample.Additional studies might investigate a more diverseworkforce and different industries to ensure the findingsof this study. Second, this study did not investigate theeffect between service climate, personality traits, and COon perceived service quality and customer satisfaction.Linkage research between intra-organizational factorsand customer evaluation is needed to clarify the role of COin future research. Finally, it is necessary to examine theboundary conditions in which service climate and person-ality traits are relevant for excellent performance or ser-vice quality as suggested by Dietz et al. (2004). Thecharacteristic of service task should be specified andincorporated into the research model.

Despite these limitations, the empirical evidencereported here will stimulate research on service climateand customer orientation. Continued research on themechanism should advance our understanding of a cus-tomer-oriented organization.

REFERENCES

Bakker, Arnold B., Karen I. Van Der Zee, Kerry A.Lewig, and Maureen F. Dollard (2006), “The Rela-tionship Between the Big Five Personality Factorsand Burnout: A Study among Volunteer Counse-lors,” Journal of Social Psychology, 146 (1), 31–50.

Boles, James, Barry J. Babin, Thomas G. Brashear, andCharles Brooks (2001), “An Examination of theRelationships Between Retail Work Environments,Salesperson Selling Orientation-Customer Orienta-tion and Job Performance,” Journal of MarketingTheory and Practice, 9 (3), 1–13.

Brown, Tom J., John C Mowen, Todd Donavan, and JaneW. Licata (2002), “The Customer Orientation ofService Workers: Personality Trait Effects on Self-and Supervisor Performance Ratings,” Journal ofMarketing Research, 39 (February), 110–19.

Dietz, Joerg, S. Douglas Pugh, and Jack W. Wiley (2004),“Service Climate Effects On Customer Attitudes: AnExamination of Boundary Conditions,” Academy ofManagement Journal, 47(1), 81–92.

Digman, John M. (1990), “Personality Structure: Emer-gence of the Five-Factor Model,” Annual Review ofPsychology, 41, 417–40.

Donavan, Todd T., Tom J. Brown, and John C. Mowen

(2004), “Internal Benefits of Service-Worker Cus-tomer Orientation: Job Satisfaction, Commitment,and Organizational Citizenship Behaviors,” Journalof Marketing, 68 (January), 128–46.

Franke, George R. and Jeong-Eun Park (2006), “Sales-Person Adaptive Selling Behavior and CustomerOrientation: A Meta-Analysis,” Journal of Market-ing Research, 43 (4), 693–702.

Frei, Richard L. and Michael A. McDaniel (1998), “Va-lidity of Customer Service Measures in PersonnelSelection: A Review of Criterion and Construct Evi-dence,” Human Performance, 11 (1), 1–27.

Harris, Eric G., John C. Mowen, and Tom J. Brown(2005), “Re-Examining Salesperson Goal Orienta-tions: Personality Influencers, Customer Orienta-tion, and Work Satisfaction,” Journal of the Academyof Marketing Science, 33 (1), 19–35.

Hurley, Robert F. (1998), “Customer Service Behavior inRetail Settings: A Study of the Effect of ServiceProvider Personality,” Journal of the Academy ofMarketing Science, 26 (2), 115–27.

Im, Subin and John P. Workman, Jr. (2004), “MarketOrientation, Creativity, and New Product Perfor-mance in High-Technology Firms,” Journal of Mar-keting, 68 (April), 114–32.

Jaworski, Bernard J. and Ajay K. Kohli (1993), “Market


Orientation: Antecedents and Consequences,” Jour-nal of Marketing, 57 (July), 53–70.

Jong, Ad de, Ko de Ruyter, and Jos Lemmink (2004),“Antecedents and Consequences of the Service Cli-mate in Boundary-Spanning Self-Managing ServiceTeams,” Journal of Marketing, 68 (April), 18–35.

Johnson, Jeff W. (1996), “Linking Employee Perceptionsof Service Climate to Customer Satisfaction,” Per-sonnel Psychology, 49, 831–51.

Kelley, Scott W. (1992), “Developing Customer Orienta-tion among Service Employees,” Journal of the Acad-emy of Marketing Science, 20 (1), 27–36.

Kirca Ahmet H., Satish Jayachandran, and William O.Bearden (2005), “Market Orientation: A Meta-Ana-lytic Review and Assessment of Its Antecedents andImpact on Performance,” Journal of Marketing, 69(April), 24–41.

Licata, Jane W., John C. Mowen, Eric G. Harris, and TomJ. Brown (2003), “On the Trait Antecedents andOutcomes of Service Worker Job Resourcefulness:A Hierarchical Model Approach,” Journal of theAcademy of Marketing Science, 31 (3), 256–71.

Lytle, S. Richard, Peter W. How, and Michael P. Mokwa(1998), “SERV*OR: A Managerial Measure of Or-ganizational Service-Orientation,” Journal of Retail-ing, 74 (4), 455–89.

Mowen, John C. and Nancy Spears (1999), “Understand-ing Compulsive Buying among College Students: AHierarchical Approach,” Journal of Consumer Psy-chology, 8 (4), 407–30.

Narver, John C. and Stanley F. Slater (1990), “The Effectof a Market Orientation on Business Profitability,”Journal of Marketing, 54 (October), 20–35.

Paulin, Michele, Ronald J. Ferguson, and Jasmin Bergeron(2006), “Service Climate and Organizational Com-mitment: The Importance of Customer Linkages,”Journal of Business Research, 59, 906–15.

Salanova, Marisa, Sonia Agut, and Jose Maria Peiro(2005), “Linking Organizational Resources and WorkEngagement to Employee Performance and Cus-tomer Loyalty: The Mediation of Service Climate,”Journal of Applied Psychology, 90 (6), 1217–27.

Saucier, Gerard (1994), “Mini-Markers: A Brief Versionof Goldberg’s Unipolar Big-Five Markers,” Journalof Personality Assessment, 63 (3), 506–16.

Saxe, Robert and Barton A. Weitz (1982), “The SOCOScale: A Measure of the Customer Orientation ofSalespeople,” Journal of Marketing Research, 19(August), 343–51.

Schneider, Benjamin, Jill K. Wheeler, and Jonathan F.Cox (1992), “A Passion for Service: Using ContentAnalysis to Explicate Service Limate Themes,” Jour-nal of Applied Psychology, 77 (5), 705–16.

____________, Susan S. White, and Michelle C. Paul(1998), “Linking Service Climate and Customer Per-ceptions of Service Quality: Test of a Causal Model,”Journal of Applied Psychology, 83 (2), 150–63.

____________, Mark G. Ehrhart, David M. Mayer, Jes-sica L. Saltz, and Kathryn Niles-Jolly (2005), “Un-derstanding Organization-Customer Links in Ser-vice Settings,” Academy of Management Journal, 48(6), 1017–32.

Susskind, Alex M., K. Michele Kacmar, and Carl P.Borchgrevink (2003), “Customer Service Providers’Attitudes Relating to Customer Service and Cus-tomer Satisfaction in the Customer–Server Ex-change,” Journal of Applied Psychology, 88 (1),179–87.

Wiley, Jack W. (1996), “Linking Survey Results to Cus-tomer Satisfaction and Business Performance,” inOrganizational Surveys: Tools for Assessment andChange, Allen I. Kraut, ed. San Francisco: Jossey-Bass, 330–59.

For further information contact:Makoto Matsuo

Otaru University of CommerceJapan

Phone: 81.134.27.5353E-Mail: [email protected]


CAN COMPANIES PROFIT FROM BUILDING A VIRTUAL

BRAND COMMUNITY?

David M. Woisetschläger, University of Dortmund, GermanyVivian Hartleb, University of Münster, Germany

SUMMARY

The majority of brand community literature dealswith the exploration of the nature of brand communitiesand the measurement of community effects. However,existing literature on how to implement and to managecompany-run brand communities is rare. From a manage-rial point of view, companies are confronted with the riseof the popularity of (virtual) brand communities, so thatboth the management of a brand community implementedby the consumers and the implementation of a brandcommunity by the company itself gets more and moreimportant. It seems, up to now, the impact of brandcommunities is underestimated in practice. Companiesneglect the chance of getting closer to their consumersand, as a consequence, increase customer loyalty. In thepresent paper, we (a) conceptualize drivers of loyalty witha football brand community initiated by a company and(b) analyze the influence of loyalty on the brand image ofthat company. We empirically test our model with adataset of 1,025 members of that particular virtual brandcommunity.

Theoretical Background

Based on Social Identity Theory (SIT), identificationwith the online community results in the intention to staya loyal member of the community. The consequence ofself-categorization to a particular virtual brand commu-nity is a positive distinction of the community’s values,norms and behaviors toward other communities andthereby an increase of group members’ self esteem (Turner1987). Moreover, the link between loyalty and the brandimage of a community presenter is likely to be stronger inthe case of high social identification. People that indicatea high level of identification with the community will tendto evaluate the community presenter as a member of thein-group, and, as a result, rate the brand image morefavorably. A second important driver of brand communityloyalty is satisfaction with the community. Satisfactionresults as a consequence of a comparison between expect-ancy and experience of functions offered and attributesthat are connected with the virtual community. Based onthe theory of cognitive dissonance (Festinger 1957), sat-isfied community members will strive for cognitive con-sonance by their intention to continue their membership,and not risk possible dissonance by switching to anothercommunity. The third important element that contributesto community loyalty is the interaction between the mem-

bers of the community itself. The foundation for thecontinuity of a virtual community is the interaction andparticipation behavior. According to SIT, interaction withother members leads to a strengthening of in-group con-sciousness. Therefore, we propose that interaction inten-sity has a positive effect on the loyalty intention ofcommunity members. Loyalty to the brand community issupposed to influence the brand image of the presenter ofthe virtual brand community. The brand community pre-senter – similar to advertising – is directed to respondentsin a situation where they pay low attention to the stimulus(e.g., because of concentrating on the community func-tions). Therefore, it has to be repeated several times inorder to attract the attention of a respondent’s mind (Baker1999). Following balance theory (Heider 1946, 1958), wepropose that loyalty to the brand community will lead toa positive evaluation of the presenter’s brand image.People that identify the presenter to be associated with thecommunity will evaluate the brand more positively. Wepropose this direction of causality, assuming that theattitude of the community members toward their commu-nity is relatively stronger than their attitude toward thepresenter in the first place.

Methodology

We tested our hypotheses in an online study using aweb survey design. At the time when the survey wasstarted, a total number of 8,361 users were registered.Twenty-one point seven (78.3) percent of the users arefemale (male). Registered members are on average 26.32years old (Std. Dev. = 11.42) and members since 4.66months (SD = 2.94). A total of 1,025 members partici-pated in the survey, equaling a response rate of 12.26percent. Twenty-four point one (75.9) percent of theparticipants are female (male). The respondents are onaverage 26.03 years old (Std. Dev. = 11.84) and memberssince 4.72 months (SD = 3.24). According to the descrip-tive statistics above, the sample is comparable to thepopulation of all registered members. We measured atti-tude-related variables with multi-item scales. Brand im-age is measured by items that are regularly used in theliterature (e.g., Keller 1993; Mitchell 1986) Brand com-munity loyalty is operationalized as membership continu-ance intention and measured using a scale fromAlgesheimer, Dholakia, und Herrmann (2005). Interac-tion intensity, identification and satisfaction with thevirtual community are measured by scales adapted fromLöwenfeld (2006).


Results

We used structural equation modeling to test thehypothesized direct and moderating effects. Results indi-cate that identification, satisfaction, and interaction inten-sity explains most of the variance of community loyalty.Moreover, a positive influence of loyalty intention on thebrand image of the community sponsor can be confirmed.A chi-square difference test was conducted for possiblemoderating effects, comparing a restricted and a non-restricted model. Results show a stronger link of member-ship continuance intention on brand image for respon-dents with high identification.

Implications

Marketing management has to keep in mind to con-tribute positively to satisfaction, interaction intensity andidentification when new features for the community aredeveloped. In our empirical example, a look at the factormeans of the three drivers to community loyalty showsroom for improvement. Second, our results demonstratethat loyal members of the community tend to evaluate tocommunity presenter’s brand image more favorably. Thisfact should provide an incentive to marketing managers toinvest in a virtual community. References are availableupon request.

For further information contact:David M. Woisetschläger

Service ManagementUniversity of Dortmund

D-44221 DortmundGermany

Phone: +49.231.755.4611Fax: +49.231.755.3271



DOES MARKETING COMMUNICATION SUFFER WHEN SERVICE

CUSTOMERS CRY FOUL?

Gavin L. Fox, Valdosta State UniversityJ. Joseph Cronin, Florida State University

SUMMARY

Introduction and Background

Complaints can be key sources of information andmanagers in service industries need to understand howcomplaints impact profitability. From a competitive per-spective, there is no currently suggested means for com-paring complaint and complaint handling impacts acrossfirms. Limitations expressed by Nyer and Gopinath (2005)indicate that it is difficult to examine real-life dissatisfac-tion due to the difficulty in gathering together enoughfirms with dissatisfied customers that will share customerinformation with the researcher. Consumers frequentlycomplain to third parties when dissatisfied with respon-siveness at the firm level. This research utilizes easilyaccessible third-party complaint information from theBetter Business Bureau to overcome this hurdle andprovide a source of strategic marketing information.

Advertising represents a primary mean through whichcompanies disperse marketing communications and buildnew customer bases. Some firms benefit more from mar-keting messages than others and a great deal of literaturesuggests that this is primarily due to quality and brandingdifferences (Keller 2001). Consumers typically are lesssatisfied with what they perceive to be of lower qualityand complain more when they are less satisfied (Voorheesand Brady 2005), especially to third parties (Singh andPandya 1991). Lower quality firms are thus expected todraw more third party complaints, be worse at respondingto third party complaints, and ultimately have lowersatisfaction. Lower quality firms are also suggested tospend more on advertising than higher quality firms (Zhao2000). Transposing quality and complaining thereforesuggests that third party complaints should positivelyrelate to advertising expense. Hence, we suggest thefollowing hypothesis:

H1: There is a positive association between the number of

complaints to third parties about a firm and advertis-ing expenditures.

Responsiveness exists to negate, or at least minimize,the fallout from product failures (Smith and Bolton 1998)and has been shown to positively impact satisfaction(Fornell and Wernerfelt 1987) and quality perceptions(Smith and Bolton 1998) following a failure. Firms withhigher perceived quality require fewer advertising dollars

in comparison to firms of lower perceived quality(Kihlstrom and Riordan 1984). Transposing quality andresponsiveness in this case suggests that good responsive-ness should lead to lower advertising expense. We there-fore propose the following hypothesis:

H2: There is a negative association between complaint

responsiveness and advertising expenditures.

Advertising intensity (Advertising/Sales) is a mea-sure of how effective advertising is at generating sales(Farris and Buzzell 1979). In other words, it is an indicatorof how well advertising dollars are being spent. Highvalues represent relatively low effectiveness, while lowvalues represent relatively high effectiveness. Complaintssuggest lower quality, which in turn should lead to de-creased sales and negative word-of-mouth (Curren andFolkes 1987). More advertising is necessary then to com-bat the negative word-of-mouth (Kihlstrom and Riordan1984). Low quality also damages customer loyalty, whichcreates customer churn and the subsequent need to replacelost customers (Jones and Sasser 1995). Hence, higherquality firms should exhibit lower advertising intensity.Therefore, we suggest the following hypothesis:

H3: There is a positive association between the number of

complaints about a firm to third parties and advertis-ing intensity.

Advertising intensity is the ratio of sales to advertis-ing expense. Hence, any change in advertising expensewill change the intensity ratio. The previous discussion onresponsiveness suggests that advertising expense decreasesas a function of responsiveness. Responsiveness alsohelps retain customers and thus maintain or even increasesales (Smith and Bolton 1998). We therefore suggest thefollowing hypothesis:

H4: There is a negative association between complaint

responsiveness and advertising intensity.

Methods

Advertising expense, total assets, and sales for eachcompany were generated from COMPUSTAT. Numberof complaints and number of responses to complaintswere generated from the Better Business Bureau data-base. The final sample represented the firms that could bematched between the two systems during the 2003–2005


year period. Values were standardized by industry toremove industry specific variance that could confound theresults. Firms that could not be matched exactly betweenthe databases were excluded from further analysis. Thisresulted in a final usable sample of 237 observations.

Results and Discussion

The objective of this research was to understand ifand how complaints to third parties and responsiveness to

complaints to third parties impacts service firm advertis-ing efforts. The results suggest that complaints, but notresponsiveness, at the third party stage is positively re-lated to advertising expense and advertising intensity.Complaints exhibited exceptional explanatory powerabove and beyond firm size, especially when determiningadvertising intensity. As a result, firms should focus onpreventing failures (Choong 2001) or at least keepingfailure fallout within the firm and away from third partiesto alleviate advertising strain.

For further information contact:Gavin L. Fox

Valdosta State UniversityValdosta GA 31698–0065



HIERARCHY (OR ANARCHY) OF EFFECTS? EXPLORING BANNER

ADVERTISING EFFECTS AND ATTENTION

Kendall Goodrich, Florida Atlantic University, Boca Raton

SUMMARY

This paper contrasts dual attitude theory (Wilsonet al. 2000) with “cognitive” and “implicit” hierarchy ofeffects models by examining effects of ad attention onmultiple advertising outcomes. The study also addressesa research gap regarding optimal Web advertising (Dahlen,Rasch, and Rosengren 2003) and potential interrelation-ships between explicit and implicit processes (Cohen andReed 2006; Stijn et al. 2005; Wilson et al. 2000).

The impact of implicit processes is a main themecoming from consumer research in the past 15 years(Johar, Maheswaran, and Peracchio 2006). Despite sub-stantial research into mere exposure effects on attitude(e.g., Zajonc 1980), issues remain about extent to whichsubconscious processes impact other advertising effects(Stijn et al. 2005). Pioneering research into dual attitudes(Wilson et al 2000) suggests that people can have twodifferent attitudes, “implicit” (people unaware of basis forevaluation, automatically activated) and “explicit.” Theendorsed attitude depends on cognitive resources avail-able for potential explicit attitude retrieval and “override”of the implicit attitude (Wilson et al. 2000).

Based on dual attitude theory (Wilson et al. 2000) andother prior research, attention was expected to be posi-tively related to aided recall (e.g., Finlay, Marmurek, andMorton 2005; Pieters, Warlop, and Wedel 2002), butnegatively related to attitude (e.g., mere exposure; Zajonc1980; Heath, Brandt, and Nairn 2006; Herr and Page2004). Furthermore, recall (Gronhaug, Kvitastein, andGronmo 1991; Haley and Baldinger 1991; Dahlen andLange 2005) and brand attitude (e.g., Fishbein and Ajzen1980; Brown and Stayman 1992; Biehal, Stephens, Curlo1992) were expected to have positive effects on purchaseintention. However, implicit “mere exposure” effects onattitude were predicted to be “overridden” for purchaseintention by a “strong brand” (e.g., top-ranked shaver;Euromonitor 2006; Dahlen and Lange 2005; Homer 2006)and the intention question (Fitzsimons and Morwitz 1996).

Methodology, Results, and Conclusion

Attention was manipulated using different banner adstimuli, after which an online questionnaire measuredaided recall, brand attitude and purchase intention. Atten-tion was directly measured online via attention tracking,which provides a valid and reliable measure of visuo-

spatial attention (Scheier 2006). Brand attitude was mea-sured with multi-item scales (Janiszewski 1988), pur-chase intention with a single item scale (Morrison 1979)and aided recall with a dichotomous variable (Unnava andBurnkrandt 1991).

Hypotheses were tested using regression, logisticregression, and ANOVA adaptations of mediation analy-sis (Baron and Kenny 1986; Campbell and Keller 2003;Grewal et al. 1998; Hastak and Olson 1989; Kumar andKrishnan 2004). The results indicate that attention has asignificant positive effect on aided recall (Wald statistic =8.23, df = 1, p < .01) and a negative effect on attitude F(1,882) = 4.53, p < .05), supporting predictions, but anunexpected positive direct effect (F(1, 882) = 8.67, p <.01) on purchase intention. Furthermore aided recall un-expectedly did notpredict purchase intention, (F(1, 882) =.365, p > .10) and attitude is negatively related to purchaseintention (t (1, 881) = 14.88, p < .001), opposite frompredictions. When attention was split into two dichoto-mous groups, attitude was negatively related to intentionfor both the high attention group (F(1, 525) = 150.09, p <.001) and the low attention group (F(1, 353) = 75.48, p <.001), suggesting “override” of even more highly-at-tended ads and little “carry-through” of implicit attitudi-nal effects onto purchase intention.

The results of this paper suggest that (1) there is aninverse relationship between attention and attitudes, sup-porting prior mere exposure research, but that (2) atten-tion is positively related to aided recall and purchaseintention, supporting more “cognitive” effects. Accord-ing to dual attitude theory (Wilson et al. 2000), explicitfactors (e.g., strong brand, positive valence, purchaseintention question’s “cognitive trigger”) can “override”implicit ad-related effects on purchase intention. How-ever, effects of both lower-attended and more highly-attended online ads appeared to be “overridden.” Arethere better competing explanations for these results? Apure cognitive approach (e.g., attention-recognition-in-tention; Stigler 1961; Telser 1964) was not supported,since there was no significant recall-to-intention relation-ship. Likewise, a cognitive hierarchy of effects model(attention-recognition-attitude-intention; ELM; Petty,Cacioppo, and Schumann 1983; anchoring and adjust-ment; Lynch, Marmorstein, and Weigold 1988; MPAA;Cohen and Reed 2006) was not supported because aidedrecall was unrelated to either attitude or intention, andthere was a negative relationship between attention and


attitude (consistent with mere exposure theory). A pureaffective model (e.g., low attention-implicit attitudes-intention), with consumer liking preferences (e.g., from“mere exposure”) affecting purchase intention, was notsupported because there was a negative relationship be-tween attitudes and purchase intention. Thus, none ofthese prior models were fully supported by the results.

One conclusion from the data is that a clear hierarchyof advertising effects is elusive. Some researchers argue

that there is little support for any advertising effectshierarchy, and that advertising effects should be studied ina space, with affect, cognition, and experience as threedimensions (Vakratsas and Ambler 1999). It is possible,for example, that purchase intention (e.g., goals, experi-ence) had a reverse-causal effect on attention (Glasmanand Albarracín 2006). Although neuromarketing is stillstruggling with reliability and validity issues (Plassmannet al. 2007), it may ultimately hold the key to accuratelymapping advertising effects.

For further information contact:Kendall Goodrich

College of BusinessFlorida Atlantic University

777 Glades RoadBoca Raton, FL 33431Phone: 561.703.2537



THE EFFECT OF SPONSORSHIP ON BRAND IMAGE: A

LONGITUDINAL ANALYSIS IN THE AIRLINE INDUSTRY

Manuel Michaelis, University of Münster, GermanyDavid M. Woisetschläger, University of Dortmund, Germany

SUMMARY

The present paper contributes to the literature bydeveloping a dynamical perspective of sponsorship ef-fects on brand image based on learning-theory and byinvestigating the proposed effects based on a panel surveyconducted before and after the 2006 FIFA World Cup™.The empirical study focuses on the change of brand imagebetween the two measurement points and assesses whetherthe “fit” between the event image of the FIFA WorldCup™ and the brand image of the sponsor plays a criticalrole in the process of image transfer.

Theoretical Background and Hypotheses Develop-

ment

In existing research, which often focuses on theassessment of sponsorship effects at one point in time,explanations of effects are often based on learning- andconsistency-theories (Cornwell, Weeks, and Roy 2005;Dean 2002; Olson and Thjømøe 2003). In a similarmanner to advertising, sponsorship often targets consum-ers in a situation, wherein they pay relatively low attentionto the stimulus (e.g., because of concentration on thesports event itself). Therefore, repetition of the stimulus ismandatory in order to attract the respondent’s attention(Baker 1999). Therefore, we propose that the recall of asponsor results in a more favorable evaluation of thesponsor’s brand image (H

1). More precisely, the degree of

change in brand image is expected to be stronger for thoseindividuals who learn that the brand is linked to sponsor-ship of the event in the second survey, in comparison tothose individuals who recall the brand as a sponsor in bothsurveys (H

2). In the latter group, respondents have already

learned that the brand is connected to the event before thefirst measurement point. However, whether a change inbrand image occurs, and the direction in which it changes,also depends on the image of the sponsored object (i.e.,event image). Many authors claim that a “fit” between thesponsored event and the sponsoring brand is essential inorder to realize an image transfer (e.g., Cornwell, Weeks,and Roy 2005; Dean 2002; Haste 1980). However, learn-ing is less likely if individuals perceive the sponsor and thesponsored event as congruent. Following balance theory(Heider 1946, 1958), a positive – and therefore incongru-ent – evaluation of the event relative to the brand in t = 1leads to a more favorable evaluation of the brand image int = 2. People who identify a sponsor’s association with an

event and have a relatively negative opinion about thesponsoring brand in comparison to the sponsored eventwill evaluate the brand more positively over time in orderto restore congruence (H

3a). A poorer evaluation of brand

image should occur if the brand image is evaluated asstronger than the event image in t = 1, using the sameexplanation as above (H

3b).

Methodology

The airline brand Emirates, the official airline spon-sor of the 2006 FIFA World Cup™, is the object ofanalysis in this study. Data for the empirical study derivefrom a series of two online studies using a web surveydesign. A total of 433 respondents answered the firstsurvey in June 2006, 254 of them also participated in thesecond survey in August 2006, equaling a response rate of21.7 percent (58.7 %). Individuals were not notified thatthey would be contacted a second time. About 60 percentof the 254 respondents were male, with an average age of29.2 (s = 8.3).

Results

Conducting ANOVAS for the two measuring pointstests the effect of sponsorship recall on brand image. Thedifference between those consumers who recalled thesponsor and those who did not before (recall

t=1: m = 3.1,

s = 1.00; no recall t=1

: m = 3.8, s = 1.38; Brown-Forsythe =17.3, p < .01) and after the event (recall

t=2: m = 3.1, s =

1.05; no recall t=1

: m = 4.0, s = 1.43; Brown-Forsythe =24.8, p < .01) is significant (H

1). Moreover, the relations

between recall and brand image are even stronger after theevent (η²

t=1 = 5.5 %; η²

t=2 = 10.4 %). Furthermore, brand

image is more stable in the group that recalls the sponsortwice (Δbrand image = .02, ns.), in comparison to thegroup that recalls the sponsor only in the second survey(Δbrand image = .43, p < .001) giving support for H

2.

Additionally, the group that evaluates Emirates as rela-tively worse compared to the FIFA World CupTM (i.e.,group with positive distance, b = 0.39, p < .001) in the firstplace evaluates the brand image of Emirates significantlybetter over time, giving support for hypothesis H

3a. The

event image also shows a marginally significant influenceon the change of brand image in the fit-group (b = 0.3,p < .1). However, the change in brand image over time isnot significant in this group (Δbrand image = .09, ns.).Furthermore, the findings lead to a rejection of H

3b be-


cause both the change of brand image (Δ brand image =.09, ns.) and the regression of distance (b = -.29, ns.) on thechange of brand image are not significant in the groupwith a negative difference.

Discussion

These findings provide the basis for several implica-tions for practice and theory. The use of longitudinal datais adequate to analyze sponsorship effects. Linking recall

and change of brand image over time is important for areliable assessment of marketing communications. Spon-sorship recall impacts brand image, which is in line withthe literature. Moreover, the change of brand image overtime depends on the relative evaluation of event imageversus brand image. This result is in accordance withbalance theory and attribution-contrast-theory, but con-tradicts other empirical findings (e.g., Speed and Thomp-son 2000; Dean 2002). References are available uponrequest.

ACKNOWLEDGMENT

The authors greatly acknowledge the financial support ofthe German Ministry for Education and Research

(BMBF – FKZ 01 HQ 0523) for this research andthank Heiner Evanschitzky for helpful comments onan earlier draft.

For further information contact:David M. WoisetschlägerUniversity of Dortmund

Otto-Hahn-Str. 644221 Dortmund

Phone +49.231.755.4611Fax: +49.231.755.3271



CROSS-CULTURAL DIFFERENCES IN THE DEVELOPMENT OF TRUST

IN RELATIONAL SERVICE EXCHANGE: AN EMPIRICAL ANALYSIS IN

THE BANKING CONTEXT IN CHINA AND GERMANY

Jan H. Schumann, Technische Universität München, GermanyFlorian V. Wangenheim, Technische Universität of München, Germany

Zhilin Yang, City University of Hong Kong, Hong Kong

SUMMARY

Achieving mutual trusting relations is an importantgoal for customer relationship management in servicesmarketing as it has been shown to positively affect rel-evant customer behavior such as customer loyalty orcommitment in different service contexts. Today moreand more service firms operate on a global level andprovide their services to customers in different cultures.This would provide a challenge for these firms if thecentral antecedents of trust were different across cultures.Evidence for such differences exists. However, thesestudies are either theoretical contributions, based on quali-tative data or concentrate only on the effects of singletrust-building processes, e.g., the importance of personalrelations. Furthermore, these studies mainly apply sec-ondary data on cultural dimensions on the country level,which do not allow analyzing the actual influence of thesingle cultural dimensions on the development of trust.Hence, up to now, services marketing research cannotsufficiently answer the question of whether people indifferent cultures differ in their development of trust.

Therefore, the purpose of this paper is to provide anempirical analysis of differences in the trust-buildingprocesses across cultures. In order to be able to deducevalid propositions for services marketing, we will focuson a specific target group in two different countries(Germany and China) and take an individual-level ap-proach to assessing differences in cultural values. Wederive and test six research hypotheses on differences inthe trust-building processes. Managerial implications arederived and directions for further research are proposed.

Theoretical Basis

Based on Doney et al. (1997), Mayer et al. (1995) andMcKnight et al. (2002) we propose six different cognitivetrust-building processes that have been shown to influ-ence the development of trust, i.e., the capability process,the transference process, the intentionality process, theprediction process, the calculative process as well as themoral evaluation process. While prior research has re-peatedly shown that trust is well explained by these sixprocesses, we assume in accordance with Doney, Cannon,and Mullen that the strength of the effect of each of these

cognitive trust-building processes is moderated by cul-tural values. Cross-cultural value differences should there-fore account for differences in the development of trust.

Method

Survey data from 652 business students in China andGermany were collected, because major cultural differ-ences could be expected between both countries. Bankingservices were chosen as a setting as they are high credenceservices and relatively comparable across countries. Thesurveys were conducted in German and Chinese. Themeasurement instruments were adapted from the relevanttrust literature and showed satisfying reliability scores. Ina first step, differences in the cultural values of bothcultural groups were compared based on individual cul-tural values of both groups (Yoo, Donthu, and Lenartowicz2001). Different from the usually applied Hofstede scores,the Chinese and German students only differed withregard to their uncertainty avoidance, where the Chinesestudents scored significantly higher than their Germancounterparts. Based on these results and building on thework by Doney, Cannon, and Mullen (1998) hypotheseswere derived on differences in the cognitive trust-build-ing processes of Chinese and German business students.The Chinese were expected to build trust more based ona capability, transference, intentionality, and predictionprocess, while the Germans were believed to build trustmore via the calculation and moral evaluation process.

Results

The hypotheses were tested with a regression modelthat included the main effect terms for the trust-buildingprocesses as well as the interaction effects between cul-ture and the respective trust-building processes. The re-sults basically confirmed the hypotheses. All differencesin the interaction effects were in the expected direction.However, significant differences could only be found forthe transference, the prediction, the calculation as well asfor the moral evaluation process.

Discussion

The results are noteworthy and relevant for market-ing research and practice in at least three ways. First, we


were able to support our hypotheses that trust is being builtbased on different cognitive processes in different cul-tures and that these differences can be explained bydifferences in individual cultural values. Second, ourresults once more show that secondary data do not reflectthe value system of a given target group very well. Thefact that both groups differ from the secondary data oncultural values in both countries supports the experienceof cross-cultural marketing research that secondary datadoes not necessarily reflect the cultural values of a spe-

cific target group. Finally, we believe these results can betransferred to other marketing and organizational con-texts. Although the data were collected from businessstudents in the context of banking services, they are on thelevel of very fundamental cognitive processes and cul-tural values. Therefore, they should not be restricted to aspecific context and provide useful guidelines as to howto build and develop trust within and across internationalorganizations. References are available upon request.

For further information contact:Jan H. Schumann

Department of Services and Technology MarketingTechnische Universität München

Arcisstraße 2180333 München

GermanyPhone: 0049.89.289.28415

Fax: 0049.89.289.28480E-Mail: [email protected]


CONSUMER RESPONSES TO SERVICE FAILURES: A CROSS-

CULTURAL COMPARISON ACROSS TWO SERVICE SECTORS

Venkatapparao Mummalaneni, Virginia State University, Petersburg

SUMMARY

Customer satisfaction with marketers has been de-clining over the years and service failures are common-place, in spite of the increased focus on consumer satisfac-tion (Lovelock and Wirtz 2004). Customer responsechoices to service failures range from no action to directcomplaints to the marketers or even third parties (Singh1990). At times, customers have gone to the internet withtheir complaints and received worldwide attention(Zeithaml et al. 2006, p. 216). An intriguing researchquestion arising out of the service failure-customer com-plaining behavior domain concerns the nature of appro-priate customer response. Should the customer complainactively and publicly, or, take no action whatsoever? Evenmore interestingly, which of these behaviors should beconsidered “ideal” from the perspective of customer-business organization relationships? From the point ofview of the service provider, “Ideal” consumer behaviorsare proposed in the present research, as the best, wishedfor responses of consumers in case of service failures.

Results from the TARP (Technical Assistance Re-search Program) studies indicate that only a small percent-age of unhappy customers choose to complain (Goodmanand Newman 2003) and the available evidence suggeststhat demographic and psychological characteristics ofcustomers play a role in determining their response (Singh1990; McAlister and Erffmeyer 2003; Lovelock and Wirtz2004; Zeithaml et al. 2006, p. 218). The influence ofcultural differences on customer complaining behaviorhas been examined in recent research (Liu et al. 2001;Mattila 1999; Mattila and Patterson 2004). The purpose ofthis study is the systematic examination of consumerresponses to service failures across cultural contexts.

On the basis of a review of consumer complainingbehavior (CCB) literature including the recent work ex-amining the influence of cultural forces on personalcharacteristics and the service expectations of customers(Mattila 1999; Ferrer et al. 2000), hypotheses linkingconsumer’s aggressiveness and individualism as well ascountry/culture differences and the type of service tovoice and exit responses of consumers have been pre-sented in this paper.

A survey of consumers in two distinct cultures, namely,U.S., and Singapore, was conducted for testing thesehypotheses. A survey instrument has been developed on

the basis of prior literature and pre-tested. Apart fromitems measuring cultural and personal characteristics, thequestionnaire presented two scenarios involving servicelapses in the context of retail service and healthcare andasked the respondent to indicate how an “ideal consumer”would respond in each scenario. Following Singh’s (1991)approach, the selection of the service settings was purpo-sive. The retail and medical care scenarios are roughlyequivalent to the settings employed by Singh (1991). Datacollection (n = 110 for the U.S. and 201 for Singapore, outof which 110 and 193 responses were usable) was done byuniversity students as part of their coursework.

Cronbach’s alpha was computed to assess the reli-ability of each variable. The values of the coefficientsrange from 0.72 to 0.77. Variable scores were computedby summing across the items measuring each variable.Prior to the summing, responses to any negatively wordeditems were reversed. The summated scores were used inthe statistical analyses undertaken for the purpose ofhypothesis testing. Multivariate Analysis of Variance wasconsidered the appropriate statistical technique. Giventhat the study employed both a between-subjects factor,namely country and a within-subjects factor (service type)with repeated measurements from the same subjects, adoubly multivariate repeated measures model (Hair et al.1998) was employed for testing the hypotheses developedearlier. Univariate tests were run for the purpose ofgaining additional insights into consumer responses. Allstatistical analyses were conducted using SPSSX 13.0.

The results from the statistical analyses provide par-tial support for our hypotheses. Collectively, the indi-vidual/personality variables considered here, fail to showconsistent effects on consumer responses. A majority ofhypotheses focusing on these effects have been rejected.The structural factors included in this study – country andservice type variables, exhibit more consistent effects.The hypotheses concerning the effects of the countryvariable on voice and exit responses find strong supportfrom our data. As predicted, the level of voice response islower and the level of exit response is higher among theSingaporean consumers. Singaporean consumers are morelikely to exit the relationship than their American counter-parts and the responses are at a higher level for the medicalcare scenario for both sets of consumers. American con-sumers are much more likely to exercise the exit optionwhen service failure occurs in a medical care setting thanthey are in a retail setting.


While the hypotheses developed here received sub-stantial support from the empirical study, future researchis needed to confirm these results and to further explore

the notion of an “Ideal” consumer from dual perspec-tives – those of the service providers and consumers.References are available upon request.

For further information contact:Venkatapparao Mummalaneni

Department of Management and MarketingVirginia State University

P.O. Box 9209Petersburg, VA 23806Phone: 804.524.5808



OFFSHORING CUSTOMER SERVICE: COMMUNICATING ACROSS

THE CULTURAL DIVIDE

Anne Stringfellow, Thunderbird School of Global Management, Glendale

ABSTRACT

This paper examines how offshore customer serviceaffects communication between service representativesand their customers, resulting in perceptions of inferiorcustomer service. The aim is to provide a framework toallow managers’ to weigh the pros and cons of offshorecustomer service and to provide guidance in selectingoffshore service locations.

INTRODUCTION

Competitive pressures are increasingly causing firmsto cut operating costs by sending customer support workoffshore. Unfortunately offshoring the customer servicefunction is more complex than offshoring manufacturing.Since customers interact directly with service representa-tives at the foreign site, cultural differences are more of anissue than would be the case with products manufacturedoffshore (Mattila 1999). One unintended consequence ofoffshoring services is customer perceptions of reducedservice quality. Poor customer service is a problem forthree main reasons. First, it leads to customer defections(Smith and Bolton 1998). Second, customers tell othersabout their poor experience (Anderson 1998; Mahajan,Muller, and Kerin 1984), leading to a further loss ofcustomers. Finally, customers experiencing poor servicetend to reduce their amount of business with the firm(Bayon et al. 2004).

A recent survey of over 9,000 customers in ninecountries found that over three quarters of English-speak-ing customers in the USA, U.K., Canada, and Australia,prefer to obtain customer service by phone, as opposed tovia e-mail, chat-room, or regular mail (Stone andLiyanearachchi 2006). The results also showed that whenthese customers called in to contact centers, their mostcommon frustration was with accents that made it difficultfor them to understand the service representatives. An-other customer survey reported that 63 percent of thosepolled had switched their business to other firms at leastpartly because of poor customer service (Genesys 2007).This echoes a previous finding that 65 percent of Ameri-can consumers reportedly would change their buyingbehavior if they even earned that a company was using anoffshore contact center (McCafferty 2004). This is consis-tent with a previous finding that tolerance for poor serviceis lower when the customer believes that the firm hascontrol over the cause (Bitner 1990). The message fromthis research by practitioners is clear and consistent:

Customers want to obtain service by phone, they don’twant to have to communicate with agents who are difficultto understand, and, if their communication needs aren’tmet, many of them are prepared to take their businesselsewhere.

According to Call Center magazine, “The chief issueimpacting off-shoring is poor cultural affinity betweenoffshore agents and American customers, which riskscustomer satisfaction and retention problems” (Read 2004,p. 30). Firms’ recent actions suggest that managers areconcerned about these negative outcomes. For example,in 2003, Dell moved its corporate customer support func-tion from Bangalore, India back to the United States(Castro 2003), Lehman Brothers moved back a twenty-person help desk to the U.S. (Stone 2004). Continuing thistrend, Lloyd’s Bank recently announced its intention tomove its customer contact center back to the U.K. (By theNumbers 2007). Despite this recognition in the practitio-ner community of the serious repercussions of offshorecustomer service on customer attrition, there is surpris-ingly little academic research on contact centers in gen-eral, and on offshore contact centers in particular. Anexception to this is a recent paper (Thelen, Magnini, andThelen 2006) which views offshoring through an ethno-centrism lens. What has not yet been investigated is theprecise mechanism by which offshoring adversely affectscustomer service. Understanding this mechanism wouldallow managers to minimize negative effects on revenueswhile retaining the cost advantages inherent in low-wagecountries.

This research seeks to address this important issue byinvestigating the effect of offshore contact center serviceencounters on customer outcomes. Since not all custom-ers withdraw their business after contact with an accentedservice representative, we seek to uncover the factors thataffect customer sensitivity to foreign accents, and thefactors that affect the linkage between exposure to aforeign accent and customer attrition. A better under-standing of these relationships would not only contributeto marketing theory, it would also enable managers tomake informed decisions about the cost-benefit relation-ships of serving particular groups of customers fromoffshore contact centers.

In the next section, we review extant literature in themarketing and services fields. We then draw on findingsfrom the linguistics and cross-cultural studies literaturesto develop a conceptual model that links aspects of the


provision of customer service by foreign contact centerrepresentatives to the quality of customer service. Thepaper concludes with a discussion of the theoretical andmanagerial implications of testing these propositions.

CONCEPTUAL MODEL OVERVIEW

The conceptual framework developed in this re-search is presented in Figure 1. Providing customer ser-vice from an offshore location affects three dimensions ofservice provider-customer communication. The dimen-sions are speech intelligibility, identity incongruence, andcommunication style incongruence. This degradation ofcommunication quality, in turn, affects customer percep-tions of three dimensions of service quality. The dimen-sions of service quality are outcome, efficiency, andinteraction. Customer attributes (consumer ethnocentrism/antagonism and demographics) and service factors (com-plexity and urgency) moderate the relationships betweencommunication and service quality.

CUSTOMER SERVICE OUTCOMES

Researchers have identified a number of desirablecustomer service outcomes. For example, the SERVQUALmeasure of service quality consists of five dimensions,reliability, responsiveness, assurance, empathy, and tan-gibles (Parasuraman, Zeithaml, and Berry 1988). How-ever, loadings have not been consistent in research usingthis scale (e.g., Cronin and Taylor 1994, p. 128) and adifferent set of factors emerged in a study undertaken in avoice-to-voice service context (Burgers et al. 2000), sug-gesting that the dimensions may differ for call centers.The dimensions we investigate in this research are thequality of the outcome, service efficiency, and the qualityof the interaction experience. Each dimension is brieflydescribed below.

Service Outcome

Customers utilize contact centers to make purchases,to obtain information, or to solve problems. For manyfirms, customers do not use contact centers unless some-thing goes wrong. For example, a customer might call inbecause they are unable to get a recently-purchased soft-ware package to run on their computer. Typically the mostimportant driver of customer satisfaction with a contactcenter service encounter is the service outcome (Feinbergand Xu 2004).

Service Efficiency

Faster service is usually evaluated more highly bycustomers (Mattila and Enz 2002). A recent study foundthat American customers were more willing to pay extrafor expedited delivery than were Singapore customersChen, Ng, and Rao 2004), supporting the conventional

wisdom that “time is money” for the American consumer.Clearly efficiency is important.

Quality of Interaction

The interaction between service provider and cus-tomer is a further important determinant of the customer’ssatisfaction with service encounters in general (Solomonet al. 1985). As Reich (1991) described service workers:“They must also have a pleasant demeanor. . . . They mustbe pleasant and helpful, even to the most obnoxious ofpatrons. . . . Above all they must make others feel happyand at ease” (p. 176). Consistent with these beliefs, socialinteraction skills are a fundamental component for contactcenter workers (Batt, Hunter, and Wilk 2002).

COMMUNICATION IN OFFSHORE SERVICE

ENCOUNTERS

The primary task of a contact center representative iscommunication with customers. According to communi-cation theorists, interpersonal communication consists of(1) the meaning of the message itself, referred to as thepropositional content and (2) the more abstract meaningconveyed by the manner in which the message is deliv-ered, referred to as the pragmatic message (Soldow andThomas 1984; Thomas 1992). How a message is commu-nicated may be just as important as what is communicated(Mehrabian 1971; Peterson, Cannito, and Brown 1995).This research explores how offshoring affects both thesevital dimensions of communication.

In a customer service encounter with an offshoreservice representative, the reception of the propositionalcontent of the message is threatened by a lack of intelli-gibility due to the use of foreign-accented English. Con-tact center operators recognize this problem and haveinstituted accent modification training systems to addressit (e.g., Stitt 2002). The reception of the pragmatic contentis threatened by the foreign context of the message. Theforeign context includes the identification of the servicerepresentative as being of a different nationality and therepresentative’s use of an unfamiliar communication style.Each of these is discussed below.

Speech Intelligibility

Practitioner surveys have found that about 50 percentof the respondents reporting problems with offshore ser-vice reported difficulty understanding the agent’s accent(Kontzer 2004) and that 12 percent of those who wouldchange their purchasing behavior as a result of offshoreservice attributed this intention to poor intelligibility offoreign customer service representatives (Anton and Set-ting 2004). Next we consider how three dimensions ofnon-native speech, accent, speaking rate, and lack offluency, affect service quality.

40A

merican M

arketing Association / W

inter 2008

Quality of Outcome

Efficiency

Quality of Interaction

Offshore Communication

Customer Ethnocentrism Customer Animosity

Service Urgency

Perception of Service Outcome

Speech Intelligibility

Identity Incongruence

Communication Style Incongruence

P6

Customer Demographics

P7

Service Complexity

P8

P10

Service Urgency

P11

P1

P3

P5

P2

P4

FIGURE 1

A Model of Customer Perceptions of Offshore Contact Centers


Intelligibility and Service Outcome. The fact that therepresentative is not a native English speaker affectsinformation transmission in both directions. First, thecustomer may not understand the service representative,and, second, the service representative may not under-stand the customer. This lack of understanding is particu-larly serious in a telephone conversation, since the tele-phone channel relies only on the voice modality, with nopossibility for the use of visual cues. Research has linkedaccent strength with decreased comprehensibility (Burda2003), which suggests that the stronger the agent’s accent,the less likely is the customer to receive a satisfactoryoutcome to his or her service encounter.

Another Component of Speech Intelligibility IsSpeaking Rate. In general, the faster the speaking rate, thelower the listeners’ comprehension (Llurda 2000). Re-search has also shown that Americans find non-nativeEnglish speakers more difficult to comprehend as theirspeaking rate increases (Anderson-Hsieh and Koehler1988; Munro and Derwing 1998). This discussion leads tothe following research propositions:

P1a: There is a negative relationship between a servicerepresentative’s strength of accent and quality ofservice outcome.

P1b: There is a negative relationship between a servicerepresentative’s speaking rate and the quality of theservice outcome.

Intelligibility and Efficiency. Accented speech takeslonger to process than unaccented speech (Munro andDerwing 1995). In addition, its lower comprehensibilitymay cause a need for message repetition, thus slowingdown the rate of information transmission. This is sup-ported by evidence that calls to Indian contact centers takean average of 20 percent longer than those to U.K. contactcenters (Wilson and Nair 2005). We therefore proposethat accented speech decreases service efficiency, bothdue to accent strength and speaking rate.

P2a: There is a negative relationship between a servicerepresentative’s strength of accent and service effi-ciency.

P2b: There is a negative relationship between a servicerepresentative’s speaking rate and service efficiency.

Intelligibility and Quality of Interaction. Researchhas shown that speakers of accented English are viewed asless competent than standard speakers (Giles 1972; Ryan,Giles, and Sebastian 1982) and of lower status (Brennanand Brennan 1981). Interaction with accented speakers isalso associated with irritation on the part of the listener(Fayer and Krasinski 1987). In addition, fluent speakersare regarded as more credible (Erickson et al. 1978),

suggesting another perception problem for second-lan-guage service representatives. The foregoing discussionsuggests the following propositions:

P3a: There is a negative relationship between a servicerepresentative’s strength of accent and perceivedquality of the service interaction.

P3b: There is a negative relationship between a servicerepresentative’s language disfluency and perceivedquality of the service interaction.

Identity Incongruence

The similarity attraction paradigm (Bryne 1971) andsocial identity theory (Ashforth and Mael 1989; Tajfeland Turner 1979) suggest that people prefer to interactwith others with whom they share a common identity.Out-groups are generally perceived as negative (Hewstone,Jaspers, and Laljee 1982) and in-group members may beliked, despite their perceived negative attributes, simplybecause of the perceived common identity (Dion 1973).Supporting this in a marketing context, research has foundthat customer perceptions of salesperson similarity inappearance, lifestyle, and socioeconomic status, play animportant role in salesperson effectiveness (Crosby, Evans,and Cowles 1990). In a voice contact situation, the onlyclue available about the other person is their speechpattern.

Communications researchers have observed nega-tive emotional reactions to foreign-accented speech. En-glish speakers perceived to be of Korean, Arabic, andChinese origin aroused negative feelings in listeners(Llunda 2000), as did speakers of Japanese origin (Cargileand Giles 1996). The reported levels of displeasure wereunaffected by the strength of the speaker’s foreign accent,nor by the disfluency of his speech. A further study,involving Chinese accented participants (Bresnahan et al.2002), found the effect to occur even with speakers whorated high on intelligibility. Not all accents are believed tocause these negative emotional effects. Cargile and Giles(1996) also reported on a prior study (Brennan and Brennan1981) in which speakers with Mexican Spanish accentsaroused no negative feelings. For example, people withBritish, other European, and Australian accents are be-lieved not to be discriminated against (Ramirez 2003).

P4: The identity incongruence of the offshore servicerepresentative affects the perceived quality of theservice interaction.

Communication Styles Incongruence

In addition to language issues, the topic of culturalaffinity has recently also been cited as an essential com-


ponent of customer satisfaction with offshore contactcenters (Hill 2004; Read 2004). In service encounters ingeneral, the greater the mutual understanding between theservice provider and the customer, the higher the cus-tomer satisfaction with the encounter (Mohr and Bitner1991). A number of issues that affect mutual understand-ing are discussed below.

Display of Emotions. As in any other social interac-tion, emotions play an important role in service encoun-ters. Research in paralinguistics has shown that listenersare able to detect the emotions expressed in a messagesolely by the tone of voice (Argyle et al. 1970), suggestingthat emotions are on display even in voice-only serviceencounters. Display rules (Rafaeli and Sutton 1989) specifyappropriate emotions for service representatives to dis-play during service encounters. For example, flight atten-dants are expected to appear cheerful and friendly, whilefuneral directors appear somber (Ashforth and Humphrey1993). However, display rules depend on cultural norms.For example, the antecedents of customer satisfactionwith a service encounter in Japan differ considerably fromthose in the United States (Winsted 1997, 1999), withformality being much more important in Japan than in theUnited States. This suggests that incongruity in emotiondisplay norms between representative and customer willtend to reduce the perceived quality of the service interac-tion.

P5a: Service encounters with representatives with incon-gruent emotional display norms are negatively asso-ciated with the quality of service interaction.

Communication Directness. American communica-tion styles are regarded as blunt, direct, and self-assertive(Pornpitakpan 2000). Other countries, such as Japan,prize harmony and use an indirect communication stylethat avoids confrontation. In the West, the communicatorbears the responsibility for making the information clearto the listener. In Eastern societies, the listener assumesthe responsibility of understanding what he/she is told(Nisbett 2003). These differences may cause Americansto feel, for example, that Asians are hiding informationfrom them.

P5b: Service encounters with representatives with incon-gruent communication directness are negativelyassociated with the quality of service interaction.

CUSTOMER AND SITUATIONAL FACTORS

THAT AFFECT CUSTOMER SERVICE

Factors may moderate some of the proposed relation-ships are broadly categorized as customer factors or assituational factors. We consider their effects below.

Customer Factors

Consumer Ethnocentrism and Animosity. A surveyof consumer attitudes to offshore customer service foundthat 48 percent of those who would change their purchas-ing behavior attributed this intention to feelings of nation-alism (Anton and Setting 2004). Closely related to nation-alism are the constructs of consumer ethnocentrism andconsumer antagonism. Consumer ethnocentrism encom-passes consumer beliefs about the appropriateness ormorality of purchasing foreign-made products (Shimpand Sharma 1987), leading to reduced preferences forforeign versus domestic products (Papadopoulos andHeslop 1993; Hong and Wyer 1989). Research has foundthat ethnocentrism carries over to communication prefer-ence, with people exhibiting strong ethnic identity exhib-iting a preference for American English, while peoplewith weak ethnic identity are more accepting of foreignaccents (Bresnahan 2002). Extending this to the offshoreservice context results in the following propositions.

P6a: Ethnocentrism amplifies the negative effect of iden-tity incongruence on quality of service interaction.

Consumer animosity is anger related to previous orongoing political, military, economic, or diplomatic eventsin a particular foreign country (Klein, Ettenson, andMorris 1998). Customer animosity was related to Chineseconsumers’ preference for South Korean cars over Japa-nese cars (Klein 2002). This result illustrates that animos-ity is country specific. For example, television and moviesoften portray the “bad guy” as being from particularforeign countries (Lippi-Green 1997). In a contact centercontext, callers routed to a less-developed country mayresent having to deal with someone who, on the basis ofaccent, is judged to be evil, socially backward, or fillinga job that rightful belongs at home.

P6b: Consumer animosity amplifies the negative effectof identity incongruence on quality of service inter-action.

Customer Demographics. Two demographic factorsthat may moderate the relationship between offshorecommunication and service quality are customer age andgeographic region.

Communication research has found that adults (sixtyand older) have significantly greater difficulty in under-standing accented speech than younger age groups (Burdaet al. 2003). It may therefore be expected that olderAmericans will be less likely to experience a high qualityservice outcome when interacting with an offshore con-tact center.


P7a: The relationship between service providers’ accentstrength and quality of service outcome is moder-ated by customer age.

Research in linguistics has found that listeners’ fa-miliarity with accented speech also influences its intelli-gibility (Gass and Varonis 1984; Wingstedt and Schulman1987), suggesting that customers living in ethnicallyhomogeneous areas, where they are not normally exposedto a variety of accents, might have more difficulty under-standing foreign accents. In a contact center context, thereis anecdotal evidence that “people on the U.S. east andwest coasts don’t mind the sound of a foreign voice, butother parts of the country are less tolerant,” (Knowledgeat Wharton 2002). Thus, the following proposition:

P7b: The relationship between service providers’ accentstrength and quality of service outcome is moder-ated by the customer’s geographic region.

Situational Factors

We consider the impact of service complexity andservice urgency on the relationship between offshorecommunications and service quality.

Service Complexity. The complexity of a service isthe number and intricacy of the steps required to performit (Shostack 1987). Customer satisfaction with contactcenter service is known to vary by industry, with 83percent of callers to government contact centers reporting“top box” satisfaction, as compared with only 30 percentfor computer software contact centers (Feinberg et al.2002). Since calls regarding computer software may bemore complex than questions calls to government offices,this suggests a negative relationship between call com-plexity and customer satisfaction.

P8: Service complexity amplifies the negative effect ofpoor speech intelligibility on quality of outcome.

Service Urgency. Just as remote service encountersvary in complexity, so too do they vary in urgency. Forexample, a customer who has just had their credit cardstolen has an urgent need to freeze the account, and mayperceive any delay due to poor understanding as longerthan it really is, suggesting the following research propo-sition:

P9: Service urgency amplifies the negative effect of poorspeech intelligibility on service efficiency.

Individuals vary in their self-disclosure behavioraccording to whether their interaction partners are fromwithin or outside their own culture (Kim 1991), suggest-ing that it is easier to establish rapport with a person whois perceived to be similar to oneself. In urgent situations,

there may be more desire to work with a service represen-tative who is perceived as being similar to oneself, andmore emotional distress if faced with incongruent com-munication styles.

P10: Service urgency amplifies the negative effect ofidentity incongruence on quality of service interac-tion.

P11: Service urgency amplifies the negative effect ofcommunication style incongruence on quality ofservice interaction.

DISCUSSSION

To summarize, the objective of this research was toinvestigate the drivers of customer resistance to offshorecustomer service, with the ultimate aim of enabling firmsto make informed choices about offshore customer ser-vice. In the process, we also hoped to advance theory inthe field of cross-cultural communications. The concep-tual framework presented is but the first step in thisprocess, providing as it does a framework around whichto organize future empirical research.

A Research Agenda

Besides the need for empirical verification of thepropositions developed in this paper, there are severalavenues for further research. One important area is toinvestigate the acceptability of specific non-native En-glish accents to customers in English-speaking countries.Anecdotal evidence exists suggests, for example, thatFilipino accents may be more acceptable to Americanthan East Indian accents (Read 2004); however, what isneeded is a systematic investigation of accent acceptabil-ity. This will allow managers to balance customer satis-faction against costs when evaluating offshore locations.

Another issue of vital importance to practitioners isthe relative role of speech intelligibility versus accentidentification in creating customer dissatisfaction withoffshore customer service. One potential difficulty inteasing apart these effects is that a negative attitude towardthe speaker of a particular variety of English tends todecrease intelligibility, despite the listener’s familiaritywith the accent (Eisenstein and Verdi 1985). The answerto the intelligibility versus identity question has clearimplications for the lucrative accent modification indus-try. While anecdotal evidence exists for the ability ofaccent neutralization to improve intelligibility (Whitneyand Wilson 2007), it appears unlikely that this training isable to completely disguise a customer servicerepresentative’s national identity. If accent identity playsan important role in customer dissatisfaction, then accentmodification may be shown to be less effective for repre-sentatives interacting with highly ethnocentric customers.


A related issue is the issue of cultural compatibility.As proposed in our model, compatible communicationstyles may be particularly important in complex and/orurgent service encounters, where emotional issues cometo the fore. The concept of interaction distance(Stringfellow, Teagarden, and Nie 2007), which wasconceptualized in the context of globally-distributed ser-vice work, may prove useful here. The cultural dimensionof interaction distance is grounded in the cultural clustersthat emerged from Hofstede’s seminal work (1980) andhave been recently built upon in the GLOBE study (Houseet al. 2004). If there is a need for emotional rapport, thenlocating an offshore contact center in a similar culturalcluster may provide an advantage.

One important aspect of national culture is the extentto which countries are oriented toward customer service.In the aggregate, United States customers have higheroverall service quality expectations than many other na-tional groups (Donthu and Yoo 1998; Sultan and Simpson2000). In less-developed countries, such as India, wheredemand for goods and services has typically exceededsupply, there has been less incentive to provide high-quality customer service. A recent survey ranked India48th out of 104 countries on degree of customer orienta-tion, while the U.S. ranked third, after Japan and Egypt(Porter et al. 2004). The customer orientation of a countrycould serve as a valuable proxy for the availability ofworkers with a service mindset, and may therefore aid indeciding on offshore locations.

The United States possesses a strong national identity(Keillor, Erffmeyer, and Babkus 1996), as opposed toJapan, Hong Kong, and Mexico, which possess interme-diate national identity, or Sweden, which possesses aweak national identity (Keillor and Hult 1999). Thissuggests that resistance to offshore contact centers may beless of an issue in countries other than the United States.

There is evidence that dislike of foreign accents in contactcenter is highest in English-speaking countries (Stone andLiyanearachchi 2006); however, this result may simplyreflect the fact that most offshored calls are conducted inEnglish.

Consumer ethnocentrism varies with demographics,with ethnocentric United States consumers more likely tobe older, Caucasian, female, employed in blue-collaroccupations (Han and Terpstra 1988), and to belong toless educated, lower income groups (Lee, Hong, and Lee2003; Sharma et al. 1995; Wall and Heslop 1986). Con-sumer animosity also varies with age and gender (Klein,Ettenson, and Morris 1998). If ethnocentrism and/or ani-mosity has a detrimental effect on customer satisfaction,it may, therefore, be feasible to identify the customersegments most likely to object to offshore customer ser-vice and to selectively route them to onshore servicerepresentatives.

CONCLUSION

In this paper we present a conceptual model that linksattributes of offshore contact center service to customersatisfaction. Motivating this effort is an observed growthin new offshore customer contact centers, while at thesame time, a growing number of firms are bringing theircustomer service back onshore. This suggests that thereare countervailing forces at play – the savings in operatingcosts do not always outweigh the costs in customerattrition. By combining the results of research conductedby contact center practitioners with insights from thecommunications literature, this research provides the ba-sis for further theory-based empirical investigation. Assuch, it provides an entry point to a new research domainthat holds great potential, both for academic researchersand for customer service practitioners.

REFERENCES

Anderson, Eugene W. (1998), “Customer Satisfactionand Word of Mouth,” Journal of Service Research, 1,5–17.

Anderson-Hsieh, J. and K. Koehler (1988), “The Effect ofForeign Accent and Speaking Rate on Native SpeakerComprehension,” Language Learning, 38, 561–613.

Anton, Jon and Teresa Setting (2004), The AmericanConsumer Reacts to the Call Center Experience andthe Offshoring of Service Calls. Purdue University:Center for Customer-Driven Quality.

Argyle, Michael, Veronica Salter, H. Nicholson, M. Wil-liams, and P. Burgess (1970), “The Communicationof Inferior and Superior Attitudes by Verbal and

Nonverbal Signals,” British Journal of Social andClinical Psychology, 9 (3), 222–31.

Ashforth, Blake E. and Ronald H. Humphrey (1993),“Emotional Labor in Service Roles: The Influence ofIdentity,” Academy of Management Review, 18 (1),88–115.

Batt, Rosemary, Larry Hunter, and Steffanie Wilk (2002),“The Quality of Jobs and Mobility Opportunities forContact Center Workers,” in The Labor Market forLow Wage Workers, Eileen Appelbaum, AnnetteBernhardt, and Richard Murnane, eds. New York:Russell Sage Foundation.

Bayon, Thomas, Florian van Wangenheim, F. OlafBackofen, and Michael Fockel (2004), “Capacity-Driven Up-and Downgrading of Service Customers:


Short- and Long-Term Effects on Customer Profit-ability,” Marketing Science Institute Conference onCollaborative Research, Yale University, (Decem-ber 11).

Bitner, Mary Jo (1990), “Evaluating Service Encounters:The Effect of Physical Surroundings and EmployeeResponses,” Journal of Marketing, 54 (April), 69–82.

Brennan, Eileen M. and J.S. Brennan (1981), “AccentScaling: Language Attitudes and Reactions TowardMexican American English Speech,” Language andSpeech, 24, 207–21.

Bresnahan, Mary J., Rie Ohashi, Reiko Nebashi, Wen Y.Liu, and S. Morinaga Shearman (2002), “Attitudinaland Affective Response Toward Accented English,”Language and Communication, 22, 171–85.

Brown, B.L., Howard Giles, and J.N. Thakerar (1985),“Speaker Evaluations as a Function of Speech Rate,Accent, and Context,” Language and Communica-tion, 5 (3), 207–22.

Burda, Angela, J.A. Scherz, C.F. Hageman, and H.T.Edwards (2003), “Age and Understanding Speakerswith Spanish or Taiwanese Accents,” PerceptualMotor Skills, 97 (1), 11–20.

Burgers, A., K. de Ruyter, C. Keen, and S. Streukens(2000), “Customer Expectation Dimensions of Voice-to-Voice Encounters: A Scale Development Study,”International Journal of Service Industry Manage-ment, 11 (2), 142–61.

Byrne, Donn (1971), The Attraction Paradigm. NewYork: Academic Press.

Cargile, Aaron C. and Howard Giles (1996), “LanguageAttitudes Toward Varieties of English: An Ameri-can-Japanese Context,” Speech Communication Con-ference. San Diego.

Castro, A. (2003), “Amid Complaints, Dell Closes Over-seas Contact Centers,” USA Today, (November 24).

Chen, Haipeng, Sok Ling Ng, and Akshay R. Rao (2005),“Cultural Differences in Consumer. Impatience,”Journal of Marketing Research, 42 (3), 291–301.

Cronin, J. and S.A. Taylor (1994), “SERVPERF versusSERVQUAL. Reconciling Performance-Based andPerceptions-Minus-Expectations Measurements ofService Quality,” Journal of Marketing, 58 (Janu-ary), 125–31.

Crosby, L.A., K.R. Evans, and D. Cowles (1990), “Rela-tionship Quality in Services Selling; An Interper-sonal Influence Perspective,” Journal of Marketing,54 (3), 68–91.

Dawson, K. (2004), “Customers: Not as Happy as YouThink,” Call Center Magazine, (September), 16–22.

Dion, K.L. (1973), “Cohesiveness as a Determinant ofIngroup-Outgroup Bias,” Journal of Personality andSocial Psychology, 28, 163–71.

Donthu, Naveen and Boonghee Yoo (1998), “CulturalInfluences on Service Quality Expectations,” Jour-nal of Service Research, 1 (2), 178–86.

Doucet, Lorna (2004), “Service Provider Hostility andService Quality,” Academy of Management Journal,47 (5), 761–71.

Eisenstein, M. and G. Verdi (1985), “The Intelligibility ofSocial Dialects for Working Class Adult Learners ofEnglish,” Language Learning, 35, 287–98.

Erickson, B., E.A. Lind, B.C. Johnson, and W.M. O’Barr(1978), “Speech Style and Impression Formation ina Court Setting: The Effects of ‘Powerful’ and ‘Pow-erless’ Speech,” Journal of Experimental Social Psy-chology, 14, 266–79.

Fayer, J.M. and E. Krasinski (1987), “Native and Non-Native Judgements of Intelligibility and Irritation,”Language Learning, 37, 313–26.

Feinberg, Richard A., Leigh Hokama, Rajesh Kadam, andIkSuk Kim (2002), “Operational Determinants ofCaller Satisfaction in the Banking/Financial ServicesContact Center,” International Journal of Bank Mar-keting, 20 (4), 174–80.

Feinberg, Richard A. and Wenti Xu (2004), “It’s theSolution, Stupid,” CRMProject.com, 5, (October).

Gass, S.M. and E.M. Varonis (1984), “The Effect ofFamiliarity on the Comprehensibility of NonnativeSpeech,” Language Learning, 34, 65–89.

Giles, Howard (1972), “Evaluation of Personality Con-tent from Accented Speech as a Function of Listen-ers’ Social Attitudes.” Perceptual and Motor Skills,34, 168–70.

Han, C. and V. Terpstra (1988), “Country-of-Origin Ef-fects for Uni-National and Bi-National Products,”Journal of International Business Studies, (Sum-mer), 135–56.

Hewstone, Miles, J. Jaspars, and M. Lalljee (1982), “So-cial Representations, Social Attribution, and SocialIdentity: The Intergroup Images of ‘Public’ and ‘Com-prehensive’ Schoolboys,” European Journal of So-cial Psychology, 12, 241–69.

Hill, K. (2004), “Skirting Contact-Center Offshoring Pit-falls,” Contact Center Today, (April 22).

Hofstede, Geert (1980), Culture’s Consequences. BeverlyHills: Sage.

House, Robert J., Paul J. Hanges, Mansour Javidan, PeterW. Dorfman, and Vipin Gupta (2004), Culture, Lead-ership, and Organizations: The GLOBE Study of 62Societies. Thousand Oaks, CA: Sage.

Kettinger, William J., Choong C. Lee, and Sunro Lee(1995), “Global Measures of Information ServiceQuality: A Cross-National Study,” Decision Sci-ences, 26 (5), (September/October), 569–88.

Kim, Young Yun (1991), “Intercultural CommunicationCompetence,” in Cross-Cultural Interpersonal Com-munication, Stella Ting-Toomey and FelipeKorzenny, eds. Newbury Park, CA: Sage, 259–75.

Kimble, C.E. and S.D. Seidel (1991), “Vocal Signs ofConfidence,” Journal of Nonverbal Behavior, 15(Summer), 99–105.

Klein, Jill (2002), “Us Versus Them or Us Versus Every-


one? Delineating Consumer Aversion to ForeignGoods,” Journal of International Business Studies,33 (2), 345–63.

Klein, Jill, Richard Ettenson, and Marlene D. Morris(1998), “The Animosity Model of Foreign ProductPurchase: An Empirical Test in the People’s Repub-lic of China,” Journal of Marketing, 62 (January),89–100.

Knowledge at Wharton (2002), “The Case for and AgainstShifting Back-Office Operations Overseas,” [http://knowledge.wharton.upenn.edu/100902_ss1.html].

Kontzer, Tony (2004), “Behind the Numbers: SuperiorCustomer Service Proves Elusive,” Information Week,(November 15), 90–91.

Lippi-Green, R. (1997) English with an Accent: Ideologyand Discrimination in the United States. New York:Routledge.

Llurda, Enric (2000), “Effects of Intelligibility and Speak-ing Rates on Judgments of Non-Native SpeakersPersonalities,” International Review of Applied Lin-guistics in Language Teaching, 38, (3/4), 289–99.

Mahajan, Vijay, Eitan Muller, and Roger A. Kerin (1984),“Introduction Strategy for New Products with Posi-tive and Negative Word-of-Mouth,” ManagementScience, 30, 1389–1404.

Malhotra, N.K., F.M. Ugaldo, J. Agarwal, and I. Baalbaki(1994), “International Services Marketing: A Com-parative Analysis of the Dimensions of Service Qual-ity Between Developed and Developing Countries,”International Marketing Review, 11 (2), 5–15.

Mattila, Anna S. (1999), “The Role of Culture in theService Evaluation Process,” Journal of Service Re-search, 1 (3), 250–61.

____________ and Cathy A. Enz (2002), “The Role ofEmotions in Service Encounters,” Journal of ServiceResearch, 4 (4), 268–77.

McCafferty, Joseph (2004), “Hello There,” CFO Maga-zine, (November), 17.

Mohr, Lois A. and Mary Jo Bitner (1991), “MutualUnderstanding Between Customers and Employeesin Service Encounters,” Advances in Consumer Re-search, 18, 611–17.

Munro, M.J. and T. Derwing (1995), “Processing Time,Accent, and Comprehensibility in the Perception ofNative and Foreign-Accented Speech,” Languageand Speech, 38, 289–306.

____________ and ____________ (1998), “The Effectsof Speaking Rate on Listener Evaluation of Nativeand Foreign-accented Speech,” Language Learning,48 (2), 159–82.

Nisbett, R.E. (2003), The Geography of Thought: HowAsians and Westerners Think Differently. . . and Why.New York: Free Press.

Parasuraman, A., Valarie A. Zeithaml, and Leonard L.Berry (1988), “SERVQUAL: A Multiple Item Scalefor Measuring Customer Perceptions of Service Qual-ity,” Journal of Retailing, 64 (1), 14–40.

Peterson, Robert, Michael P. Cannito, and Steven P.Brown (1995), “An Exploratory Investigation ofVoice Characteristics and Selling Effectiveness,”Journal of Personal Selling and Sales Management,XV(1), 1–15.

Pornpitakpan, C. (2000), “Trade in Thailand: A Three-Way Cultural Comparison,” Business Horizons,(March/April).

Porter, Michael E., Klaus Schwab, and Xavier Sala-i-Martin, Lopez-Claros (2004), The Global Competi-tiveness Report 2004–2005. Geneva: World Eco-nomic Forum.

Ramirez, Y. (2003), “Studies Show Accents Create Ob-stacles for Job-Seeking Foreigners,” Daily Universe,Brigham Young University, 11/21.

Read, Brendan (2004), “How to Offshore Customer Opsthe Right Way,” Customer Contact, (April) 20.

Reich, Robert B. (1991) The Work of Nations: PreparingOurselves for 21st Century Capitalism. New York:Alfred A. Knopf.

Ryan, Ellen B., Howard Giles, and R.J. Sebastian (1982),“An Integrative Perspective for the Study of Atti-tudes Toward Language Variation,” in Attitudes To-ward Language Variation, E.B. Ryan, H. Giles, eds.London: Edward Arnold, 1–19.

Shimp, Terrance A. and Subhash Sharma (1987), “Con-sumer Ethnocentrism: Construction and Validationof the CETSCALE,” Journal of Marketing Research,24 (3) 280–89.

Shostack, G. Lynn (1987), “Service Positioning ThroughStructural Change,” Journal of Marketing, 51 (1),34–43.

Smith, Amy K. and Ruth N. Bolton (1998), “An Experi-mental Investigation of Customer Reactions to Ser-vice Failures: Paradox or Peril?” Journal of ServiceResearch, 1 (1), 65–81.

Soldow, G.F. and G.P. Thomas (1984), “Relational Com-munication: Form Versus Content in the Sales Inter-action,” Journal of Marketing, 48 (1), 84–93.

Solomon, Michael R., Caroline Surprenant, John Czespiel,and Evelyn Gutman (1985), “A Role Theory Per-spective of Dyadic Interactions: The Service En-counter,” Journal of Marketing, 49, 99–111.

Stitt, Gregg (2002), Diverted to Delhi. Film Australia. 55minutes.

Strauss, B. and P. Wang (1999), “‘Culture Shocks’ inInter-Cultural Service Encounters?” Journal of Ser-vices Marketing, 13 (3/4), 329–46.

Stringfellow, Anne, Mary Teagarden, and Winter Nie(2007), “Invisible Costs in Offshoring ServicesWork,” Journal of Operations Management, forth-coming.

Sultan, Fareena, Merlin C. Simpson, Jr. (2000), “Interna-tional Service Variants: Airline Passenger Expecta-tions and Perceptions of Service Quality,” Journal ofServices Marketing, 14 (3), 188–216.

Tajfel, H. and J.C. Turner (1979), “The Social Identity


Theory of Group Behavior,” in The Social Psychol-ogy of Intergroup Relations, William G. Austin andStephen Worchel, eds. Monterey, CA: Brooks-Cole,33–47.

Thelen, Shawn, Vincent P. Magnini, and Tanya K. Thelen(2006), “Identifying Elements of Offshore ServiceEthnocentrism,” Working Paper, Hofstra University.

Thomas, G.P. (1992), “The Influence of Processing Con-versational Information on Inference, ArgumentElaboration, and Memory,” Journal of ConsumerResearch, 19 (2), 83–92.

Tsalikis, J., O.W. DeShields, Jr., and M.S. LaTour (1991),“The Role of Accent on the Credibility and Effective-ness of the Salesperson,” Journal of Personal Selling& Sales Management, 11 (Winter), 31–41.

Whitney, Kellye and Amy Wilson (2007), “Operational

Efficiency: Lessons from Global Enterprises,” ChiefLearning Officer, 6 (6), 46–49.

Wingstedt, M. and R. Schulman (1987), ‘”Comprehen-sion of Foreign Accents,” in Phonologica 1984:Proceedings of the Fifth International PhonologyMeeting, Eisenstadt, W. Dressler, ed. 25–28 June1984, Cambridge: Cambridge University Press, 339–44.

Winsted, Kathryn Frazer (1997), “The Service Experi-ence in Two Cultures: A Behavioral Perspective,”Journal of Retailing, 73 (3), 337–60.

____________ (1999), “Evaluating Service Encounters:A Cross-Cultural and Cross-Industry Exploration,”Journal of Marketing Theory and Practice, 7 (2),(Spring), 106–23.

For further information contact:Anne Stringfellow

Thunderbird School of Global Management15249 N. 59th AvenueGlendale, AZ 85306Phone: 602.978.7452



BRAND EFFICIENCY AND BRAND RELEVANCE: INTRODUCING

AND LINKING BOTH CONCEPTS

Maik Hammerschmidt, University of Mannheim, GermanyTobias Donnevert, University of Mannheim, Germany

Hans H. Bauer, University of Mannheim, Germany

ABSTRACT

Brand managers are under increased pressure toillustrate the performance of their multimillion dollarexpenditures. This paper provides three contributions inthe context of brand management. First, we introduce theconcept of brand efficiency as a measure for the perfor-mance of the brand management process. Second, wedevelop a measure for the industry-specific brand rel-evance (influence of branding on purchase decisions).Third, we link these concepts and examine if brandmanagement efficiency in an industry is influenced bybrand relevance.

INTRODUCTION

Several authors stress that most of existing brandequity approaches are developed ad hoc and are used in anisolated way (Ambler and Barwise 1998; Chumpitaz,Kerstens, and Paparoidamis 2006; Luo and Donthu 2006).They note that a variety of “stand alone” brand equityconcepts exists, providing only fragmented insights ratherthan a holistic perspective on brand performance. Thus,the authors call for a stronger integration of the differentoutcome variables. Moreover, most approaches do notrelate brand outcomes to brand investments which wereemployed to create the outcomes. The first contribution ofour paper is to address these issues by introducing theconcept of brand efficiency as a broader, more integratedmeasure of brand performance. To the best of our knowl-edge, no study has examined the efficiency of brandmanagement so far or has provided a methodologicallyand theoretically sound measure for the efficiency of thebrand management process. The advantages of the brandefficiency concept are that we consider both the input andthe output side simultaneously (resulting in a measure forthe “return on brand investment”) and conceptualize bothsides multi-dimensionally. By measuring efficiency as aratio of multiple outputs to multiple inputs we embed themultitude of brand equity measures in an economic frame-work. This allows identifying overspendings and there-fore helps to avoid misallocations of brand managementresources. In doing so, we answer Rust and colleagues’(2004, p. 83) call for new methodologies to measuremarketing productivity, in which they clearly identify akey gap in the literature when they state that “few methodscurrently exist for comprehensively modeling the chain of

marketing productivity all the way from tactical actions tofinancial impact or firm value.”

We propose that firms are not equally successful inachieving high efficiency of their brand managementprocess because they do not align their brand investmentsto the influence of brands on consumers buying decisions(brand relevance). The long-term success of a branddepends on the sustainable influence brands have towardthe buying decision in a specific product category. If theinfluence of the brand on the buying decisions of consum-ers is low, a brand management that invests heavily tocreate a strong brand (in terms of customer-based out-come) is assumably inefficient as the brand investmentswill not lead to a high financial outcome. As a secondcontribution we therefore develop a multi-item measurefor the brand relevance in an industry or a product marketin terms of the “brand driveness” of purchase decisions inthat industry (Riesenbeck and Perrey 2006). To our knowl-edge, this is the first academic study addressing this issue.

After introducing the two concepts of brand effi-ciency and brand relevance our third contribution is to linkthese two concepts to investigate whether different levelsof brand relevance across industries can explain differ-ences in brand management efficiency. We argue thatcreating high brand efficiency depends on the alignmentof brand investments with the level of brand relevance. Totest this hypothesis we conduct two studies using brandrelated data from five industries or product categories,respectively (automotive, desktop computers, casual cloth-ing, financial services, banks). Our findings answer thequestion under what circumstances the focus of marketingshould be on brand building or on other decisions criteriasuch as price or customer service. In other words, weanalyze when the focus of marketing investments shouldbe on maximizing brand equity and when it should be oncreating customer equity in the respective industry.

CONCEPTUAL FRAMEWORK

Brand Efficiency

We use the concept of brand efficiency as a measureof brand performance. Brand efficiency or brand manage-ment efficiency is to be understood as a ratio of multiplebrand outputs to multiple brand inputs. It reflects whether


brand management achieves the best transformation ofdeployed brand investments (e.g., advertising spendings)into brand outcomes (consumer based and financial brandsuccess). Due to the multitude of branding instruments(inputs) and brand-related success measures (outputs),brand efficiency has to be conceptualized multi-dimen-sionally (Staat and Hammerschmidt 2005). Then a simul-taneous analysis of different investments and outcomefactors is possible. The target is to gain insights into howbrand management can transform deployed inputs (adver-tising, distribution, quality management) into outputsaligned to the steps of the brand management process. Inorder to conceptualize this brand management process werefer to the brand value chain proposed by Keller andLehmann (2006).

Keller and Lehmann (2006) provide a conceptualframework for the creation of brand value and proposethat brand equity is built and should therefore also bemeasured along the “brand value chain” (Figure 1). In thefirst step of the chain brand investments are consideredwhich directly affect both the cognitive and affectivedimension of customer based brand equity. These twodimensions are captured appropriately by brand aware-ness and brand image respectively as recent studies show(Keller 1993; Kapferer 2004). According to empiricalresults they are mainly driven by investments in com-munication, distribution, and product quality (Johnson,Herrmann, and Huber 2006; Yoo, Donthu, and Lee 2000;Rossiter and Percy 1997).

In the second step, brand image and brand awarenesslead to financial outcomes (financial brand success).Here, we distinguish between product market perfor-mance represented by brand revenue or brand profitabil-ity and stock market performance represented by stockprice, P/E-ratio or market capitalization. This follows thelogic that from the marketing perspective consumers arethe major constituency driving brand revenue, while share-holders constitute the central stakeholder from a financialperspective driving stock price. Recently, brand manag-ers are seen as accountable not only for the success of theindividual brand but for creating shareholder value inorder to strengthen marketing’s credibility (Madden, Fehle,and Fournier 2006). Instead of capturing stock marketperformance by the company’s market capitalization asdone in several publications (e.g., Rust, Lemon, andZeithaml 2004) we believe that from a financial marketperspective the stock price relative to the earnings (Price/Earnings-ratio) should be used as an outcome of brandinvestments. Both the company earnings (E

i) and the

willingness of investors to pay for it (measured by the P/E-ratio) are the drivers of market capitalization (MV

i).

PMVi = Ei x ––

E

Furthermore research findings in Finance show thatthe P/E-ratio (willingness to pay) is positively influencedby the liquidity and breadth of a stock. According toGrullan, Kanatas, and Weston (2004) brand investmentsare a key driver of liquidity and breadth of a stock becausestocks of strong brands are heavily traded. Additionally,investors view a strong brand as an indicator for thecompany’s ability to create and to ensure future cashflows (Madden, Fehle, and Fournier 2006). As a result, theP/E-ratio is a more comprehensive indicator for the stockmarket performance. Figure 1 summarizes our conceptualframework.

This paper investigates how efficient brand invest-ments are transformed into customer-based outcomes(step 1) and, subsequently, how efficient this customer-based brand impact is transformed into financial out-comes (step 2). We propose that the efficiency of thetransformation of customer-based outcomes into “hard”financial results (step 2) highly depends on the degree onwhich the decision criterion “brand” influences purchasedecisions (i.e., the level of brand relevance).

Brand Relevance

Building strong brands is not a promising strategy forany industry. This is because brands are not equallyimportant to purchasing decisions in every market. Brandrelevance is defined as the degree to which the brand playsa key role in consumers’ choice process for a product in agiven product category. The stronger the role of the brandagainst other purchasing decision criteria, such as price,customer service, or product quality, the more relevant thebrand appears (Perrey et al. 2003).

Brand relevance is an often-used phrase, but it gener-ally has not been well defined or explained. In literature,definitions exist that differ from our understanding ofbrand relevance. Aaker (2004) regards a specific brand asrelevant if three conditions are met: (1) a product orservice category or subcategory – defined by some com-bination of attributes, applications, user groups, or otherdistinguishing characteristics – exists or emerges. (2)There is a perceived need or desire on the part of acustomer segment for the category or subcategory. (3)The brand is part of the evoked set of brands that a segmentconsiders as being material to the product category orsubcategory. Brand relevance involves two stages of thecustomer-brand interaction: First, when the customerselects a product category or subcategory perceived to berelevant to the problem or opportunity (e.g., he or she maydecide to buy a luxury sports sedan rather than a compactor an SUV). Second, the customer determines whichbrands to consider (in this case the choice might includeAudi, BMW, Lexus, and Cadillac). A specific brand’srelevance depends on both stages. Although preference( )

i


based on a differentiated offering and a positive usageexperience can help to enhance a brand’s relevance, if theneed or category association is missing, the brand lacksrelevance, and no differentiation, attitude, or relationshipwill help. In other words, according to Aaker (2004) abrand (e.g., Cadillac) is relevant if it is associated with aproduct category or subcategory (e.g., SUV). In order tobe relevant, a brand should at least be recalled without aid.

Kapferer and Laurent (1992) present another ap-proach called “sensibilité aux marques” (brand sensitiv-ity). They define brand sensitivity as the influence of abrand on the buying decision of a specific consumer.Brand sensitivity exists if the consumer takes the brandinto consideration for his buying decision as a matter ofprinciple.

Obviously, existing approaches relate brand relevanceto specific brands (Aaker 2004) or to specific individuals(Kapferer and Laurent 1992). Our understanding of brandrelevance is related to the relevance of the decision crite-rion “brand” in general to an average consumer in amarket (Perrey et al. 2003). It is the average brand sensi-tivity in an industry. Here, we follow Kotler’s (2003,p. 442) definition of a brand as a “name, sign, symbol ordesign which identifies the goods and services of oneseller and differentiates them from those of competitors.”Hence, with the term brand we consider consumers’associations and feelings related to a brand name (Aaker1991).

To our knowledge, McKinsey & Company con-ducted the only empirical studies on brand relevance sofar (Perrey et al. 2003; Riesenbeck and Perrey 2006). A

crucial shortcoming of these studies is that brand rel-evance is operationalized as a single-item, global measureonly not capturing the different facets of the construct.Therefore, we introduce a multi-item measure of brandrelevance. Based on a review of existing literature(Riesenbeck and Perrey 2006; Perrey et al. 2003; Kapfererand Laurent 1992) we generated five items that captureour understanding of brand relevance. We reveal theinfluence of the buying decision criterion “brand” in anindustry by asking consumers if in industry X (1) thebrand plays an important role compared to other decisioncriteria (e.g., price); (2) the brand is a very importantdecision criterion; (3) it is important for them to buybranded products; (4) they would buy a branded producteven if they would have to incur extra efforts; (5) the brandis very important for the purchase decision. Using thismeasurement model researches can classify existing prod-uct categories or industries according to their level ofbrand relevance (e.g., industries with high, medium, orlow brand relevance).

Linking Brand Efficiency and Brand Relevance

We suggest that the efficiency of the first step trans-formation (creating superior awareness and image throughbranding instruments) depends mainly on factors that areunder control of brand management e.g., managerial skillsand competencies (Murthi,Srinivasan, and Kalyanaram1996). Recent literature proposes an increasing standard-ization and homogenization of brand strategies and poli-cies across firms (cervino and Cubillo 2004; Teece, Pisano,and Shuen 1997). At the same time the number of mergersand acquisitions as means to capture marketing resourceshas grown significantly (capron and Hulland 1999). In

FIGURE 1



line with these developments it can be assumed that eveninvisible brandind assets become more and more mobi-lized making an easy and efficient transfer of brandingskills between companies possible (Itami and Roehl 1991).This leads us to conclude that in the first step, highefficiency can be achieved in every industry irrespectiveof brand relevance. Thus, we propose H

1: In the first step

of the brand value chain brand efficiency does not differsignificantly between product markets with high vs. me-dium or low brand relevance. In contrast, for efficientlytransforming consumer-based outputs into financial out-comes in the second step of the brand value chain, brandrelevance is a crucial prerequisite. As explained above,brand relevance can be seen as a market characteristicwhich is exogenously given and cannot be controlled bybrand management, at least not within a short and mediumterm horizon (Smith 1992). Presumably, only if brandshave a significant impact on the buying decision ofconsumers (i.e., the level of brand relevance is high), highexpenditures to build up strong brands will translate intohigh financial success. Thus, we formulate H

2: In the

second step of the brand value chain, brand efficiency issignificantly higher in product markets with high brandrelevance than in product markets with medium or lowbrand relevance.

RESEARCH SETTING AND METHODOLOGY

Research Setting

To test our hypotheses we follow a three-stage ap-proach: First, we measure brand relevance in differentproduct categories and rank them according to their levelof brand relevance. Second, we choose three industrieswith significantly different levels of brand relevance(high, medium, and low) and measure the efficiency ofbrands in each industry using Data Envelopment Analysis(DEA). Third, we compare the average efficiency be-tween the industries for step 1 (H

1) and step 2 (H

2) of the

brand value chain to analyze differences between productcategories with different levels of brand relevance.

Stage 2 and 3 are conducted twice. In study 1 we use“mono-brand” manufacturers that are publicly traded.Therefore, as the financial output for step 2 we use a stockmarket related metric (price/earnings ratio). To test for therobustness and generalizability of the findings we con-duct a second study now including brands that don’t havean own stock price either because the manufacturer is notpublicly traded at all or is a “multi-brand” publicly tradedfirm (i.e., only the parent brand is listed). As stock marketrelated metrics become inappropriate in those cases weuse earnings before interest, taxes, depreciation, and am-ortization (EBITDA) to capture the profitability of thesebrands. We believe that examining one sample withoutstock metrics is important as several studies emphasize

that stock returns may be driven simply by the fact thatinvestors and analysts observe high brand investments.This may lead to the fact that they expect higher apprecia-tion potential although the long run sales or profitabilityremains unaffected, resulting in a pure “investor responseeffect” (Joshi and Hanssens 2004; Mizik and Jacobson2005). Joshi and Hanssens (2004) show, that high adver-tising has a direct effect on valuation, i.e., an effectindependent of its indirect effect via revenue and profitresponse. Frieder and Subrahmanyam (2005) find thatinvestors favor stocks with strong brand names, eventhough these powerful brands did not generate superiorreturns. Thus, even if sales response to branding activitiesis demonstrably weak, investors are willing to pay apremium for aggressive brand management initiatives.Therefore, one could argue that the high brand efficiencyscores and the close relationship between brand relevanceand brand efficiency occurred due to the use of a stockmarket performance metric instead of a conventionalprofit measure like EBITDA.

Measurement of Brand Relevance

In order to get a brand relevance ranking of productcategories, we randomly selected 26 business-to-con-sumer (B2C) product categories from the Consumer PriceIndex (CPI) market basket. The five statements intro-duced above were transferred to an online questionnaireusing a seven-point Likert-scale ranging from “absolutelydo not agree” to “absolutely agree.” To ensure that re-spondents only answered questions about product catego-ries they are familiar with, we first asked them for theirpurchase experience in the 26 categories within the last 12months. Among these “familiar” product categories sevenwere randomly selected and respondents answered thefive relevant questions per category and several questionsregarding socio-economic characteristics. Brand relevanceis measured as an index of the five items for which theaverage values of all respondents within an industry weretaken.

Measurement of Brand Efficiency

To capture the two steps of the brand value chain weuse a two-step Data Envelopment Analysis (DEA) model.DEA is a nonparametric tool that can deal with multipleinputs and outputs when measuring inefficiency. It esti-mates an efficient frontier by maximizing the weightedoutput/input ratio of each brand, thus producing a singlemeasure of overall efficiency (Charnes, Cooper, andRhodes 1978). Efficient brands (best practices) are thosefor which no other brand or linear combination of brandscan generate as much as or more of the output given theinput levels. Because the weights for input and outputvariables of a brand are computed in order to maximize theratio and then compared to similar ratios of best-perform-


ing units, the measured value is also referred to as relativeefficiency (Seiford 1996). The efficiency results given byDEA will provide firms valuable information for assess-ing the adequacy of their spending.

According to the brand value chain the influence ofthe resources consumed by brand management instru-ments on financial performance is indirect, utilizing psy-chographic outputs as intermediate factors to generatefinancial outputs (Keh and Chu 2003). Thus, we recast thebrand value chain as a chain of two DEA models. In thefirst step DEA model we examine the influence of brandinvestments on the psychographic variables awarenessand image covering the cognitive and affective dimensionof customer-based brand equity. Subsequently, in thesecond step DEA model it is investigated whether psycho-graphic variables are translated successfully into “hard”economic facts. Such a multi-stage model allows insightsinto the sources of overall brand (in)efficiency. Not de-composing the overall efficiency score would maskwhether inefficiency arises from internal, operationalaspects (conversion of resources in superior awarenessand image) or from market-related aspects (capitalizingon awareness and image). We argue that the success in thefirst step mainly depends on the skills of the brandmanagers (H

1) while success in the second step depends

on brand relevance as a key market characteristic (H2).

DATA AND SAMPLE

Data and Sample for the Brand Relevance Ranking

Three hundred fifty respondents answered the onlinequestionnaire. As every respondent rated seven industrieswe obtained more than 2,500 evaluations in total, i.e., 100evaluations per industry. The sample’s socio-economiccharacteristics are representative for the Central Europepopulation. We conducted exploratory and confirmatoryfactor analysis indicating excellent fit measures(Cronbach’s alpha: 0.95, Variance Extracted: 83%, χ2/df:2.95, RMSEA: 0.03, NNFI: 0.99, CFI: 0.99; RFI: 0.99;SRMR: 0.015). As a result we obtained the brand rel-evance ranking of the 26 industries shown in Table 1.

Data and Sample for the Measurement of Brand Effi-

ciency

In order to examine the differences in brand effi-ciency between industries and the moderating role ofbrand relevance we first selected the following productcategories (brand relevance score in parentheses): Study 1:automotive (high brand relevance: 5.1), desktop comput-ers (medium: 3.8) and financial services (low: 2.5); Study 2:automotive (high: 5.1), casual clothing (medium: 4.4) andbanks (low: 2.2). The differences in brand relevance arehighly significant at p < 0.01.

TABLE 1

Brand Relevance Ranking of Industries

Rank Industry Brand Relevance Rank Industry Brand Relevance

1 Consumer Electronics 5.16 14 Desktop Computers 3.80

2 Automotive 5.10 15 Soft Drinks 3.73

3 Cigarettes 5.08 16 Convenience Food 3.68

4 Sports Shoes 5.04 17 Cellular Phone Networks 3.52

5 Cosmetics 5.00 18 Shower Gel 3.43

6 Business Clothing 4.91 19 Sunglasses 3.35

7 Beer 4.90 70 Fixed Line Networks 2.91

8 Laptop Computers 4.79 21 Furniture 2.74

9 Television 6.63 22 Mineral Water 2.61

10 Casual Clothing 4.48 23 Financial Services 2.52

11 Sparkling Wines 4.46 24 Banks 2.23

12 Power Tools 4.01 25 Electricity 2.07

13 Insurances 3.90 26 Toilet Paper 1.60


Second, to analyze brand efficiency we chose 16brands per industry in Study 1 and 20 brands per industryin Study 2. For each industry the brands cover at least 60percent of the market volume. Thus, no major brand ismissing in our data set. Note that the number of brands waslimited as data for all steps and variables of the brand valuechain had to be available. To get the input and output datawe collected secondary data for the period 2005. Commu-nication investments were taken from Nielsen MediaResearch including expenditures for print (newspaper,magazines), broadcast (television, radio) and outdoor(expenditures in more than 300 outdoor plant operatormarkets). To control for lagged and carryover effects ofadvertising we used a function of previous period (2004)and current period (2005) expenditures as the communi-cation input (Charnes et al. 1997). As most studies onadvertising response modeling found that 90 percent of alladvertising effects dissipate after 15 months at latest (seethe review in Vakratsas and Ambler 1999) this time spanseems adequate. For quality costs we used costs of goodssold as this metric represents all resources that go into theproduct and thus determine its quality (e.g., labor, mate-rial). Distribution costs refer to the costs for making theproduct available in a great number of stores in order tooffer the brand where and when consumers want it andthus reducing the time consumers must spend searching,providing convenience in purchasing, and making it easierto get services related to the product. Hence, distributioncosts encompass costs for outlets, sales force, and trademarketing (Smith 1992; Yoo, Donthu, and Lee 2000).Both quality and distribution costs were taken fromCOMPUSTAT. Image (as an index of 13 indicators likeinnovativeness, design, and prestige evaluated on a 7-point scale) and awareness (measured as aided recall) areobtained from a cross-industry brand survey. It is one ofthe most comprehensive and widely circulated surveysbased on over 10,000 consumer interviews that are repre-sentative for 50.3 million people. Data on brand specificrevenue, EBITDA, and P/E ratio were obtained fromCOMPUSTAT database. Note that DEA estimates theefficiency without a priori information on tradeoffs amonginputs and outputs (Chen and Agha 2004; Luo and Donthu2006). Thus, this method is advantageous for our study aswe have no prior knowledge about which part of the brandexpenditures produces which part of the outputs.

Regarding the sample size of DEA studies necessaryfor meaningful results, the literature commonly suggeststhat the amount of observed units (in our case brands) hasto be larger than double the amount of the product of thenumber of inputs and number of outputs. This test isregarded as valid for assessing the appropriateness ofdatasets for DEA (Dyson et al. 2001; Vassiloglou andGiokas 1990). For both steps of the model this conditionis fulfilled. To check for potential outliers which is crucialdue to high error sensitivity of DEA results, we conductedsuper-efficiency analysis. Brands with abnormal super-

efficiency scores extremely push out the frontier leadingto biased efficiency evaluations. As all brands’ super-efficiency scores are below the suggested screen level of1.2 (Banker and Chang 2006) there is no need for remov-ing brands from the dataset. In summary, the DEA resultscan be expected to be robust and without systematic errors(Doyle and Green 1995).

RESULTS

Study 1

Comparison of First Step Results Between the In-dustries (Test of H1). On the first step of the brand valuechain (see Table 2) the fraction of efficient brands is verysimilar (automobile and financial services: 50%; comput-ers: 40%). Using the nonparametric Mann-Whitney rankstatistic we find that the differences in first step efficiencyscores between the three industries are non-significant.This test avoids the hazard of making assumptions aboutthe distribution of DEA efficiency scores. In contrast toparametric techniques, non-parametric rank procedurescan be used for any sample size; studies have shown thatthe asymptotic normal approximation used here onlyrequires n

1, n

2 > 10 (Brockett and Golany 1996). The

results confirm H1 implying that in the first step of the

brand value chain operational management capabilitiesdrive branding success. As expected, in average the brand-ing competence – which can be learned internally orexternally acquired in any industry – is quite similaracross the industries.

Even in the industry with lowest brand relevancethere is little room for improvements in the first step. Theaverage efficiency score in the financial industry indicatesthat the mean brand could have reduced spending by 13percent (1 – 0.87) while holding the level of outputs(image, awareness) constant.

Comparison of Second Step Results Between theIndustries (Test of H2). The results for the second stepmodel are totally different. The Mann-Whitney test indi-cates that both the difference in average efficiency be-tween automotive and computers (0.94 – 0.72 = 0.22) andthe difference between computers and financial services(0.72 – 0.63 = 0.09) are significant. Moreover, whilewithin the automobile industry (high brand relevance) 40percent of brands are efficient, this fraction declines to 25percent in the financial services industry (low brandrelevance). These results confirm H

2 showing that the

efficiency of the transformations in the second step of thebrand value chain is influenced by brand relevance be-cause the initial situation for all branches was comparable:For example the 2005 brand advertising expenditures inthe automotive industry (1.39 billion euros) and in thefinancial service industry (1.22 billion euros) are veryclose (Nielsen Media Research 2007). They also reachnearly the same efficiency scores in step 1, but as the brand


is not the key driver of buying decisions in markets withlow brand relevance well-known brands with a positiveimage do not translate into economic performance. As theefficiency score for the financial service industry (0.63)indicates, for the mean brand almost 40 percent of thecurrent values of awareness and image are not capitalized,i.e., are wasted. In contrast, for automotives investmentsin awareness and image are nearly fully reflected in thebottom line.

Comparison of Overall Results Between the Indus-tries. The overall efficiency score shows the efficiencythat is reached by the brands in the entire brand manage-ment process; it is calculated by multiplying the efficiencyscores of both steps (Golany and Storbeck 1999). Theoverall efficiency is also significantly higher for theautomotive brands (0.86) than in the other industries (0.67and 0.53). As the results show, the high overall ineffi-ciency in the computer and financial services industry ispredominantly caused by inefficiencies in the second stepof the brand management process. Obviously, there existtwo possibilities to increase the overall efficiency: (1)through creating higher revenue and stock value by re-aligning brand resources to other more brand drivenproduct markets; (2) through reducing wasted brand ex-penditures (thereby improving brand efficiency) and re-assigning them to the actual drivers of buying decisionsuch as price or service quality. These results confirm oursuggestion that low brand investments should be em-ployed in industries with low brand relevance. Thus, it isnot adequate that the brand advertising expenditures ofthe financial service industry are that close to the automo-tive ones.

Comparison of Results for the Inefficient BrandsBetween the Industries. Finally, an analysis of the ineffi-cient brands is relevant for the deduction of managerialimplications as well. The overall efficiency of inefficientbrands is considerably high in the automotive industry(0.81). Especially in the second step, the efficiency valuesof the inefficient brands are close to 1. This indicates thatit would be easy for all brands in the sample to reach the

efficient frontier. The derived values indicate a lead of theautomotive industry, even within the inefficient units.This provides further support for H

2. Our model can

confirm the assumption that brand relevance is a goodindicator for the optimum level of brand investment sinceit shows a high predictive validity for brand managementefficiency.

Study 2

In order to ensure the robustness of our findings werepeated the analysis for a new data set modifying severalaspects of the first study. In Study 2 we focused on non-publicly traded brands, in order to check whether ourimplications are robust when using conventional profit-ability measures instead of stock market performance.Thus, in step 2 of the DEA model we substituted the outputprice/earnings-ratio by EBITDA (earnings before inter-est, taxes, depreciation, and amortization). Again we usedthe automotive industry to represent high brand relevance.As the industry for the medium brand relevance level wenow used casual clothing manufacturers instead of desk-top computers; for low brand relevance we more specifi-cally used banks instead of financial services providers ingeneral. Moreover, in Study 2 we used data for 20 brandsfor each industry (instead of 16 in Study 1). The inputs andoutputs of the first step DEA were the same as in Study 1.Results for Study 2 are shown in the following Table andcan be interpreted analogously to Study 1.

As the results of Study 2 show, the ranking of thethree industries with respect to second step brand effi-ciency (from high to low efficiency) exactly correspondswith the brand relevance ranking of the industries (fromhigh to low brand relevance). This holds true for all threeefficiency indicators, i.e., fraction of efficient brands,average efficiency score across all brands and averageefficiency score across inefficient brands. Thus, the re-sults of both studies are highly consistent indicating thatour results are robust with respect to industry type, typesof financial output measures and number of observations(brands) included in the DEA model.

TABLE 2

DEA Results for Brand Management Efficiency in Study 1

Automotive Computer & Equipment Financial Services

(High Brand Relevance) (Medium Brand Relevance) (Low Brand Relevance)

Step 1 Step 2 Overall Step 1 Step 2 Overall Step 1 Step 2 Overall

Fraction of efficient brands 50% 40% 33.3% 40% 33% 26.6% 50% 25% 6.2%

Average efficiency score (all 0.91 0.94 0.86 0.91 0.72 0.67 0.87 0.63 0.53

Brands)

Average efficiency score 0.80 0.91 0.81 0.86 0.60 0.56 0.73 0.50 0.49

(inefficient brands)


DISCUSSION

Brand managers are accountable for the task of get-ting the most out of company resources such as commu-nication, distribution, and quality investments. Brandinvestments become increasingly threatened since theyentail a large part of the overall marketing costs. Thus, itbecomes important for brand managers to show the pro-ductivity of their multimillion dollar spending (Rust,Lemon, and Zeithaml 2004). A methodologically soundmeasure of brand performance is challenging becausefirms often target their expenditures to promote multipleoutcomes simultaneously, such as both visible sales andstock performance and invisible brand image and aware-ness.

Theoretical Contribution

First, we provide a model to measure the efficiency ofboth steps of the brand management process with effi-ciency being defined as a multiple-output to multiple-input ratio. In the first step we examined the transforma-tion of brand investments into customer-based metrics(brand awareness and brand image). The first step modelrepresents the cognitive and affective aspect of brandequity. Subsequently, in the second step model it has beeninvestigated whether customer metrics have been trans-lated successfully into “hard” financial metrics. As asecond theoretical contribution we developed a multi-item measure to assess brand relevance in an industry.This allows researchers to analyze if brands have animpact on consumers’ purchase decisions.

Managerial Contribution

By analyzing five industries differing significantly intheir level of brand relevance we revealed a significantinfluence of brand relevance on brand management effi-ciency. As the results show, firms are equally successfulin creating consumer-based outputs but differ signifi-cantly in creating brand related financial success along the

level of brand relevance in the product market they areoperating in. This is because brand relevance, in contrastto the consumer-based outputs, cannot be influenced bybrand managers in the short run. Thus, brand relevanceshould be seen as a solid metric for determining theoptimal extent and allocation of brand investments. Thefindings show that high brand investments in marketswith high brand relevance are justified while the enor-mous costs to build up well known brands in markets withlow brand relevance have to be scrutinized. In suchmarkets other criteria than brands seem to be the keydrivers of consumers buying decisions.

An understanding of brand relevance provides com-panies with a more solid basis for determining how muchto spend on communication. High communication inten-sity can only be justified if brand relevance is high. Ifbrand relevance is low, such investments should, at thevery least, be carefully scrutinized. Our findings highlightthe key issue in any marketing decision: Whether it ispricing, distribution, or brand, the marketing lever that isused must be effective enough to justify the investment.For example, in the financial service sector brand invest-ments are highly inefficient due to the low significance ofbrands in this market. Consequently, the use of othermarketing tools would be more beneficial. The low im-portance of the brand as a purchase criterion in thefinancial services sector is consistent with the discussionthat the personal relationship with customers is the keysuccess factor in that industry. Therefore, when it comesto the decision which metrics should be used for resourceallocation the customer equity concept may be moreappropriate. Our findings do question the recent discus-sion whether customer equity and brand equity shouldnecessarily be integrated (Leone et al. 2006; Ambler et al.2002). Instead, we propose that this integrated approachis only appropriate in industries with high brand rel-evance. From a financial perspective companies in indus-tries with low brand relevance should concentrate oncustomer equity only.

TABLE 3

DEA Results for Brand Management Efficiency in Study 2

Automotive Casual Clothing Bank

(High Brand Relevance) (Medium Brand Relevance) (Low Brand Relevance)

Step 1 Step 2 Overall Step 1 Step 2 Overall Step 1 Step 2 Overall

Fraction of efficient brands 55% 30% 20% 25% 25% 0% 40% 20% 15%

Average efficiency score 0.80 0.92 0.73 0.47 0.56 0.20 0.51 0.31 0.20

(all brands)

Average efficience score 0.56 0.89 0.68 0.28 0.40 0.20 0.20 0.14 0.06

(inefficient brands)


REFERENCES

Aaker, D.A. (1991), Managing Brand Equity: Capitaliz-ing on the Value of a Brand Name. New York.

____________ (2004), Brand Portfolio Strategy: Creat-ing Relevance, Differentiation, Energy, Leverage,and Clarity. New York.

Ambler, T. and P. Barwise (1998), “The Trouble withBrand Valuation,” Journal of Brand Management, 5(5), 367–77.

____________, C.B. Bhattacharya, J. Edell, K.L. Keller,K.N. Lemon, and V. Mittal (2002), “Relating Brandand Customer Perspectives on Marketing Manage-ment,” Journal of Service Research, 5, 13–25.

Banker, R.D. and H. Chang (2006), “The Super-Effi-ciency Procedure Is for Outlier Identification, Not forRanking Efficient Units,” European Journal of Op-erational Research, 175, 2, 1311–20.

Brockett, P.L. and B. Golany (1996), “Using Rank Statis-tics for Determining Programmatic Efficiency Dif-ferences in Data Envelopment Analysis,” Manage-ment Science, 42, 466–72.

Capron, L. and J. Hulland (1999), “Redeployment ofBrands, Sales Forces, and General Marketing Man-agement Expertise Following Horizontal Acquisi-tions: A Resource-Based View,” Journal of Market-ing, 63, 41–54.

Cerviño, J. and J.M. Cubillo (2004), “A Resource-BasedPerspective on Global Branding: An Analysis ofTrademark Registration,” International Journal ofManagement, 21 (4), 351–63.

Charnes, A., W.W. Cooper, and E. Rhodes (1978), “Mea-suring the Efficiency of Decision Making Units,”European Journal of Operational Research, 2, 429–44.

____________, ____________, B. Golany, D.B. Learner,F.Y. Phillips, and J.J. Rousseau (1997), “AMultiperiod Analysis of Market Segments and BrandEfficiency in the Competitive Carbonated BeverageIndustry,” in Data Envelopment Analysis: Theory,Methodology, and Application, A. Charnes, W.W.Cooper, A.Y. Lewin, and L.M. Seiford, eds. 3d ed.Boston, 145–65.

Chen, Y. and A. Agha (2004), “DEA Malmquist Produc-tivity Measure: New Insights with an Application toComputer Industry,” European Journal of Opera-tional Research, 159 (1), 239–49.

Chumpitaz, R., K. Kerstens, and N. Paparoidamis (2006),“Performance in Marketing: From Uni- to Multi-Dimensional and from Function to Frontier Con-cepts,” in Marketing Efficiency, H.H. Bauer, M. Staat,and M. Hammerschmidt, eds. Munich, 3–30.

Doyle, J.R. and R.H. Green (1995), “Cross Evaluation inDEA: Improving Discrimination among DMUs,”INFOR, 33 (3), 205–22.

Dyson, R.G., R. Allen, A.S. Camanho, V.V. Podinovski,C.S. Sarrico, and E.A. Shale (2001), “Pitfalls and

Protocols in DEA,” European Journal of Opera-tional Research, 132, 245–59.

Frieder, L. and A. Subrahmanyam (2005), “Brand Per-ceptions and the Market for Common Stock,” Jour-nal of Financial and Quantitative Analysis, 40, 65–85.

Golany, B. and J.E. Storbeck (1999), “A Data Envelop-ment Analysis of the Operational Efficiency of BankBranches,” Interfaces, 29 (3), 14–26.

Grullon, G., G. Kanatas, and J.P. Weston (2004), “Adver-tising, Breadth of Ownership, and Liquidity,” TheReview of Financial Studies, 17 (2), 439–61.

Itami, H. and T.W. Roehl (1991), Mobilizing InvisibleAssets. Cambridge.

Johnson, M.D., A. Herrmann, and F. Huber (2006), “TheEvolution of Loyalty Intentions,” Journal of Market-ing, 70, 122–32.

Joshi, A. and D.M. Hanssens (2004), “Advertising Spend-ing and Market Capitalization,” MSI Report, No. 04–110, Cambridge: Marketing Science Institute.

Kapferer, J.N. and G. Laurent (1992), “La Sensibilité AuxMarques – Marchés sans Marques, Marchés ÀMarques,” Editions d’Organisation, HEC School ofManagement, Paris.

____________ (2004), Strategic Brand Management:New Approaches to Creating and Evaluating BrandEquity, 3d ed. London.

Keh, H.T. and S. Chu (2003), “Retail Productivity andScale Economies at the Firm Level – A DEA Ap-proach,” Omega, 31, 75–82.

Keller, K.L. (1993), “Conceptualizing, Measuring, andManaging Customer-Based Brand Equity,” Journalof Marketing, 57 (1), 1–22.

____________ and D. R. Lehmann (2006), “Brands andBranding: Research Findings and Future Priorities,”Marketing Science, 25 (6), 740–59.

Kotler, P. (2003), Marketing Management: Analysis, Plan-ning, Implementation and Control, 11th ed. EnglewoodCliffs.

Leone, R.P., V.R. Rao, K.L. Keller, A.M. Luo, L.McAlister, and R. Srivastava (2006), “Linking BrandEquity to Customer Equity,” Journal of Service Re-search, 9 (2), 125–38.

Luo, X. and N. Donthu (2006), “Marketing’s Credibility:A Longitudinal Investigation of Marketing Commu-nication Productivity and Shareholder Value,” Jour-nal of Marketing, 70 (4), 70–91.

Madden, T.J., F. Fehle, and S. Fournier (2006), “BrandsMatter: An Empirical Demonstration of the Creationof Shareholder Value Through Branding,” Journal ofthe Academy of Marketing Science, 34, 224–35.

Mizik, N. and R. Jacobson (2005), “How Brand AttributesDrive Financial Performance,” MSI Report, No. 05–003, Marketing Science Institute Cambridge.

Murthi, B.P.S., K. Srinivasan, and G. Kalyanaram (1996),“Controlling for Observed and Unobserved Manage-rial Skills in Determining First-Mover Market Share


Advantage,” Journal of Marketing Research, 33 (3),329–36.

Nielsen Media Research (2007), “Brands Invest More inAdvertising,” Press Release, 10.01.2007, Hamburg.

Perrey, J., J. Schroeder, K. Backhaus, and H. Meffert(2003), “When Do Brand Investments Pay Off?”Mckinsey Marketing & Sales Practice on Branding,London, 31–46.

Riesenbeck, H. and J. Perrey (2006), Power Brands:Measuring, Making and Managing Brand Success.Wiley.

Rossiter, J.R. and L. Percy (1997), Advertising & Promo-tion Management. New York.

Rust, R.T., K.N. Lemon, and V.A. Zeithaml (2004),“Return on Marketing: Using Customer Equity toFocus Marketing Strategy,” Journal of Marketing,68 (1), 109–27.

Shocker, A.D. and B. Weitz (1988), “A Perspective onBrand Equity Principles and Issues,” in Defining,Measuring and Managing Brand Equity, L.Leuthesser, ed. Report 88–104, Marketing ScienceInstitute, Cambridge, 2–4.

Seiford, L. (1996), “Data Envelopment Analysis: TheEvolution of the State of the Art,” Journal of Produc-tivity Analysis, 7, 99– 137.

Smith, D.C. (1992), “Brand Extension and AdvertisingEfficiency: What Can and Cannot Be Expected,”Journal of Advertising Research, 32 (6), 11–20.

Staat, M. and M. Hammerschmidt (2005), “Product Per-formance Evaluation – A Super-Efficiency Model,”International Journal of Business Performance Man-agement, 7 (3), 304–19.

Teece, D.J., G. Pisano, and A. Shuen (1997), “ProductPerformance Evaluation – A Super-EfficiencyModel,” Strategic Management Journal, 18 (7), 509–33.

Vakratsas, D. and T. Ambler (1999), “How AdvertisingWorks: What Do We Really Know?” Journal ofMarketing, 63 (1), 26–43.

Vassiloglou, M. and D. Giokas (1990), “A Study of theRelative Efficiency of Bank Branches: An Applica-tion of Data Envelopment Analysis,” The Journal ofthe Operational Research Society, 41, 591–97.

Vavricka, J. and B. Trailer (1997), “Implementing aFormal Selling Process and Performance Measuresin a Sales Organization,” in Operational Perfor-mance Measurement – Increasing Total Productiv-ity, W. Kaydos, ed. Boca Raton, 1–21.

Yoo, B., N. Donthu, and S. Lee (2000), “An Examinationof Selected Marketing Mix Elements and BrandEquity,” Journal of the Academy of Marketing Sci-ence, 28, 195–211.

____________ and ____________ (2001), “Developingand Validating a Multidimensional Consumer-BasedBrand Equity Scale,” Journal of Business Research,52, 1–14.

For further information contact:Maik Hammerschmidt

University of MannheimL 5, 1

68131 Mannheim, GermanyPhone: +49.0621.181.1569

Fax: +49.0621.181.1571E-Mail: [email protected]

Tobias DonnevertUniversity of Mannheim

L 5, 168131 Mannheim, GermanyPhone: +49.0621.181.1566


Hans H. BauerUniversity of Mannheim

L 5, 168131 Mannheim, GermanyPhone: +49.0621.181.1563



WHEN COMPANIES GO TOO FAR . . . AND GET IT RIGHT:

EXPLORING LOW-FIT BRAND EXTENSION SUCCESS

Jill Sadler, Queen’s School of Business, KingstonMatthew Thomson, Queen’s School of Business, Kingston

SUMMARY

In the last ten years, consumers have witnessed theintroduction of many brand extensions including Panasonicbicycles, Teflon apparel, Ferrari laptops, and Virgin bridalboutiques. Many of these efforts have thrived, a result thatis surprising given that they mesh poorly with or evencontradict the notion of fit with parent brand. Klink andSmith (2001) touched on this issue in their review ofthreats to the external validity of existing brand extensionresearch. They noted that research has identified fit as oneof the most powerful predictors of brand extension suc-cess yet examples such as those provided earlier appear tocontradict this phenomenon. Seemingly incongruent brandextensions highlight that fit may be more flexible than isreflected in current research or that there are additionalfactors are at play.

The purpose of this study is to explore the conditionsunder which a low-fit brand extension can succeed. Spe-cifically, we examine what effect consumers’ strong rela-tionships with an existing brand have on the success oflow fitting extensions. Generally, a person with a strongand positive relationship with a partner will be moreaccommodating and tolerant (Rusbult et al. 1991), ratealternatives less favorably (Johnson and Rusbult 1990),and perceive their partner more positively (Murray,Holmes, and Griffin 2003). In this study, we propose thatstrong relationships will predict positive evaluations oflow fit extensions due to perceptions of (1) innovativenessand (2) fit. That is, as the strength of consumer-brandrelationship increases, consumers are more likely to per-ceive the brand extension as innovative and as a notterrible fit with the parent brand, explaining their im-proved perceptions of the extension as a whole.

We administered a survey to a sample of 200 adultswho were recruited online using Study Response in ex-change for being entered into a random draw of six prizesof $50. Eighty-four respondents were able to identify

what they thought were successful low fit extensions.Examples of low fit extensions provided by respondentswere Hanes perfume, Adidas cologne, Honda TV, EldersISP, and Vera Wang beds. Several covariates were alsoassessed to address alternative explanations and includedmood, attitude toward the parent brand, breadth of offer-ing of the parent brand, and manufacturing difficulty ofboth the parent brand and extension. Only Mood andManufacturing Difficulty predicted consumers’ evalua-tions.

Results generally support our thinking and suggestthat consumers who have a strong relationship with aparticular brand evaluate low fit extensions positivelybecause they view such extensions as more innovativeand better fitting than consumers with weaker relation-ships. That is, even though the respondents in our studyrated the extensions as low fitting (M = 3.5/7), fit still hada positive effect on evaluations of extensions. Whileresearch on biased partner perceptions suggests individu-als in a strong relationship tend to view their partner in apositive light (Murray, Holmes, and Griffin 2003), ourresults likewise point to some combination of forgivenessand rationalizing by consumers.

Further, based on our data, it is apparent that if acompany launches a low-fitting brand extension, theconsumer with a strong brand relationship will pick outand make evaluations based on characteristics of theextension that are superior, unique, or innovative. Theseunique and innovative characteristics, whether real orimagined, would allow the consumer to re-affirm theirrelationship with the brand.

Our findings confirm the importance of buildingstrong relationships between brands and consumers. Strongrelationships provide protection from branding mistakesthat could otherwise damage brand equity and ensuregreater flexibility for their business in the future. Refer-ences are available upon request.


For further information contact:Jill Sadler

Queen’s School of BusinessQueen’s University143 Union Street

Kingston, Ontario, K7L 3N6Canada



COMPLETENESS AS A PRODUCT POSITIONING STRATEGY

Timucin Ozcan, University of Rhode Island, KingstonDaniel Sheinin, University of Rhode Island, Kingston

SUMMARY

Marketers often try to differentiate their products byintroducing additional product capabilities. In some prod-uct categories, introducing all possible product features ina single product may be positioned as “complete product”by the marketer to increase perceived compatibility of theproduct with consumers’ needs and wants. Such completepositioning has been introduced in a range of products,including toothpaste (e.g., Colgate Total – complete 12hour protection), multivitamins (e.g., Centrum – from Ato Zinc), online retailers (e.g., eBay – whatever you’relooking for you can get it on eBay), personal care products(e.g., Braun – 360° Complete Shaver), and softwareprograms (e.g., McAfee Total Protection). A crucial re-search question, therefore, centers on the impact of per-ceived completeness of product features on consumerjudgment and choice.

The objective of this study is to understand theinfluence of completeness as a positioning strategy onproduct beliefs, evaluations and purchase intentions. Wepropose that complete positioned products have morepositive product beliefs, better evaluations, and higherlevels of purchase intentions compared to the same prod-uct without such positioning. It is also proposed thatinformation load as a result of decision complexity andproduct complexity, and price will moderate the impact ofcompleteness. Specifically, it is proposed that higherlevels of information load and price will have a positiveeffect on preference for complete products.


Merriam Webster’s collegiate dictionary (1995) de-fines completeness as “having all necessary parts, ele-ments, or steps.” Complete products, by definition, shouldinclude all attributes, additional features and options thatare offered in that particular product category with itscurrent technology. In a choice context where consumerneeds to choose between a specialized versus completepositioned product, the concept of compensation mayexplain the potential outcome. Within the choice context,compensatory decision strategy implies that a superiorfeature of a product can compensate for poor value on it’sanother feature, while noncompensatory strategy impliesthat a superior performance of one attribute cannot com-pensate for a poor value on another. The central researchquestion is, therefore, whether consumers use compensa-

tory or noncompensatory decision strategies when com-paring complete and specialized alternatives.

Zero-risk bias, in other terms certainty effect, sug-gests that individuals favor small benefits which aredefinite to larger benefits that are indefinite. For example,individuals are more likely to assign a higher value ondecreasing the possibility of a negative outcome from fivepercent to zero percent than decreasing the possibility ofanother, possibly more severe, risk from 30 percent to 20percent (see Baron 1994 and Gowda 1999 for a discus-sion). In terms of possessing attributes, complete posi-tioning would imply reducing the potential risk to zerowith the choice since it would suggest that the product hasall potential attributes. While specialized positioning maysuggest that the product is excelled in one or more at-tributes, it does not suggest a definitive benefit in terms ofhaving all attributes. Hence, zero-risk bias should operatein favor of complete positioned alternative over special-ized alternative. Another cognitive process that couldinfluence the evaluation between complete positionedand specialized product is availability heuristic. The avail-ability heuristic states that “people assess the frequency ofa class or the probability of an event by the ease withwhich instances or occurrences can be brought to mind”(Tversky and Kahneman 1974, p. 1127). Complete posi-tioning should evoke availability heuristic by stating allpossible attributes that alternative should offer. There-fore, availability heuristic should favor complete posi-tioned alternative especially when choice context in-volves both specialized and complete positioned alterna-tives.

While we propose that complete positioned productswill have more positive product beliefs, better evalua-tions, and higher levels of purchase intentions, we alsoposit that there are some moderating variables that influ-ence this relationship. First, information load that is causedby decision complexity (number of products in the choiceset) and product complexity should moderate the impactof completeness. Another variable that is proposed toimpact the effect of completeness is price of the focalproduct. When price of complete positioned product islower than specialized alternatives, consumers are ex-pected to be more skeptic about the effectiveness of theproduct. Therefore, we propose that when price of thecomplete positioned product is higher than other alterna-tives, product beliefs, evaluations, and purchase inten-tions for complete positioned product increase.


Experiment and Discussion

The participants, 133 undergraduate students (48.1%females) in a northeastern university, were recruited toparticipate in a study that investigates consumer-decisionmaking by evaluating hypothetical choice alternatives.They were given extra credit for participating in thisexperiment. A 2 (completeness) x 2 (decision complexity)x 2 (price) between-subjects design was employed toinvestigate research hypotheses. Product complexity wasthe within subjects-factor.

Results indicate that complete products have higherpreferences compared to specialized products in both

high and low complex product groups. Even at the indi-vidual attribute beliefs, our results demonstrate that com-plete positioned products are more likely to be evaluatedmore positively. Our results also indicate that price is notthe main moderator of the completeness-product prefer-ences relationship. Our results show that rather than price,information overload with decision and product com-plexities moderates complete product preferences. Spe-cifically, our results show that completeness as a position-ing strategy significantly elevates product evaluations,beliefs, and purchase intentions when product is highlycomplex and there are many alternatives in the choice set.References are available upon request.

For further information contact:Timucin Ozcan

University of Rhode IslandKingston, RI 02881–0802



THE IMPORTANCE OF THE RETAIL MARKETING INSTRUMENTS

AND OF THEIR PERCEIVED CONCEPTUAL COHERENCE IN

BUILDING A STRONG RETAIL BRAND

Bernhard Swoboda, University of Trier, GermanyThomas Foscht, California State University, East Bay

Hanna Schramm-Klein, Saarland University, Germany

SUMMARY

With the growing realization that brands are one of afirm’s most valuable resources (Srivastava, Fahey, andChristensen 2001), branding has emerged as a top man-agement priority in the past decade. Establishing strong,conceptually coherent and consistent/durable brands tendsto be one of the main goals that managers strive to achieve(Aaker 1996; Kay 2006). Given its highly competitivenature, branding can be especially important in the retail-ing industry in influencing customer perceptions, and inmotivating store choice and loyalty (Ailawadi and Keller2004, p. 331; Hartman and Spiro 2005). Thus, the rise ofthe retailer as a brand is one of the most important trendsin retailing (Grewal, Levy, and Lehmann 2004).

This article addresses how retailer attributes and theirperceived conceptual coherence affect customer-basedretail brand equity when considering retailers as brands. Itinvestigates the antecedents of a strong retail brand.Unlike other investigations, a model is applied to fiveretail sectors so that some generally applicable and alsosome sector-specific conclusions can be drawn. We applya broad catalogue of retailer attributes in order to detectvarying relevance depending on the sector concerned. Instore image research, retailer attributes were usually in-vestigated in isolation only, without looking at whetherthese attributes matched one another. In other areas,however, this aspect plays a major role, which is why wehave integrated it into our study and thus comply with therequest made by Marks (1976), who pointed out that theindividual attributes should not only be regarded in isola-tion because they interact with one another in the mind ofthe consumer. Human cognition and perception is anintegrated process, which means that perception is not anisolated process of specific senses, but an integratedassimilation of stimuli (Gerrig and Zimbardo 2004). Per-ception psychology research, particularly with regard todissonance theory (Festinger 1957), speaks of conver-gence and divergence, which refer to the customer’scognition being matched or contradicted. In our contextthis means that marketing instruments either fit one an-other or do not. The match of integrated instruments in thecognition of the customer can be labeled perceived con-ceptual coherence (Murphy and Medin 1985). So theretailer attributes and their conceptual coherence, are

processed internally in the consumer’s mind. The short-term perception is the ground on which the consumer-based retail brand equity (the long-term attitudes towardthe retailer) is formed. This then influences actual shop-ping behavior (Steenkamp and Wedel 1991; Hanna andWozniak 2001; Burnkrant and Page 1982).

The empirical study is based on 3,026 face-to-faceinterviews (conducting quota sampling by sex and age)spread over the grocery, textiles, DIY, electronic, andfurniture retailing sectors. Structural equation modeling isused to illustrate the impact of central dimensions of theperception of retailer attributes and their perceived con-ceptual coherence on customer-based retail brand equity.

First of all, a model with excelled overall fit measuresis presented which analyses the impact of the retailerattributes and their perceived conceptual coherence on theretail brand equity examining the total sample, i.e., all fiveretailing sectors simultaneously in order to reveal effectsthat occur across the retail industry. Intersectorally theconsiderable relevance of all five single dimensions andthe perceived conceptual coherence of the retailer at-tributes is revealed. Perceived conceptual coherence isshown to be important on the one hand, and one the otherhand in a comparison of the individual perception dimen-sions, service is the dimension that stands out.

The next step was to compare their influence withinthe five specific sectors forming a strong retail brand.AMOS, based on a multiple-group analysis, was used toreview the sectors individually. The overall fit measuresindicate that this model can be applied to each single retailsector and substantiate the proposition that the fiveintersectoral dimensions of the retailer attributes and theircoherence can be applied both intersectorally and specifi-cally to an individual retail sector. The individual dimen-sions show different levels of influence in the individualretail sector. Nevertheless, the perceived conceptual co-herence of the retailer attributes has very high and con-stant influence compared to the individual dimensions.When comparing only the individual dimensions, theservice dimension is also highly relevant in each specificretail sector. In four out of five sectors analyzed, it is thedimension with the strongest influence on retail brandequity.


The study demonstrates the impact of retailer at-tributes in building a strong retail brand in general, as wellas highlighting some differences that are very muchspecific to the individual retail sector concerned. Theresults sensitize the reader to the fact that building apositive retail brand must take account of special featuresspecific to the retail sector in question. In general, wepoint out the importance of the perceived conceptual

coherence of the retailer attributes. As a retailer it isimportant to follow one distinct strategy that defines eachsingle retail marketing instrument with the aim that theconsumer’s perception of the retailer is consistent. Theconsumer must obtain a clear image of the retailer thatremains stable over the course of time. References areavailable upon request.

For further information contact:Bernhard Swoboda

Institute for Marketing and RetailingTrier University

Universitätsring 15D-54286-Trier

GermanyPhone: 0049.651.3050



NATURE OF CURRENCY: EFFECTS ON DECISION MAKING

Ashwani Mongo, University of Texas, San AntonioRitesh Saini, George Mason University

ABSTRACT

In this paper, we explore how the nature of currencyused in a decision making situation can influence a deci-sion maker’s susceptibility to using heuristics. Prior re-search reveals that when information such as magnitudeof risk, prior investments, or time delay is processed byindividuals in the context of time versus money, there arequantitative differences in how this information is utilizedto arrive at judgments and decisions. We contribute to thisstream of research by demonstrating that a qualitativelydifferent form of decision making – making decisionsbased on quick and easy heuristics rather than an analysisof the available information – gains prominence when oneworks with time rather than money. We also offer evi-dence for the heuristic process via response time mea-sures, and by showing how this difference between timeand money can be eliminated.

INTRODUCTION

Consider a mom trying to find the right university forher promising high-school child. She could search forinformation about universities by spending time (e.g.,scouring the Internet), money (e.g., buying relevant maga-zines), or both (e.g., traveling to different universities).More searches will possibly increase the payoffs (i.e.,finding suitable universities) but will also increase costs(i.e., time or money expenditures). Therefore, she mightmake an a priori decision regarding how much to search.For instance, she might decide on the number of univer-sities to visit so that she can accordingly budget for therequired travel time (e.g., by applying for leave fromwork) and money (e.g., by saving on other expenses).

As illustrated above, consumers frequently searchthe environment in order to resolve uncertainties (Moorthy,Ratchford, and Talukdar 1997; Punj and Staelin 1983).Literature on search suggests that, given a potential pay-off, the extent to which people search is dictated by thecosts of search (Stigler 1961). In this paper, we explorehow the willingness to search could be driven by heuris-tics that are completely unrelated to search costs andpayoffs. For instance, the mom trying to decide on thenumber of universities to visit could rely on heuristicsevoked by prior experience (“I always check 3 optionsbefore making a decision; let me visit 3”) or by the context(“Out of my three friends, one visited 2 universities, onevisited 5, and one visited 8; let me take the middle optionand visit 5”). We find that the propensity to use heuristics

is higher when search involves spending time rather thanmoney. This finding provides new insights into howdecision making in time is different than that in money. Italso adds to our understanding of the search process byshowing how it differs with the currency in which it isconducted. Finally, our results are of relevance to con-sumers trying to budget time and money, and to marketerstrying to stay in the consideration set of consumers search-ing for products and services.

CONCEPTUAL BACKGROUND

A search decision is akin to solving an optimization-under-constraints problem in which costs are traded againstpayoffs (Stigler 1961). In line with this, experimentaleconomists have demonstrated that, given no change inpayoffs, search behavior increases with a decrease in therelevant search costs (Kogut 1992; Schotter and Braunstein1981; see Davis and Holt 1993 for a review). Consumerresearchers, however, have long argued that decisions canbe based not only on a careful analysis of relevant infor-mation but also on simple thumb rules, or heuristics(Payne, Bettman, and Johnson 1993). Could search deci-sions also be driven by heuristics that are actually unre-lated to search costs and payoffs? Furthermore, couldheuristics have a differential effect depending on whethersearch is conducted by spending time or by spendingmoney? Examining this time-money difference in thecontext of search is interesting for three reasons.

First, apart from rare exceptions involving time (Rosenand Olshavsky 1987; Smith, Venkatraman, and Dholakia1999), search experiments rely overwhelmingly on thecurrency of money. This is true for experimental econom-ics research (Cox and Oaxaca 1989; Kogut 1992; Schotterand Braunstein 1981) as well as experimental consumerresearch (Diehl 2005; Srivastava and Lurie 2001; Urbany1986; Zwick et al. 2003). The use of money is appropriatebecause it enables easy quantification of search costs asresearchers focus on the phenomena of interest. However,introspection reveals that in conventional markets (e.g.,physical stores) as well as electronic ones (e.g., websites),time is very frequently spent in order to search. This is alsoevident from field research that assumes time as a primarycurrency when studying search behavior in real-life set-tings. Search costs are measured by directly assessingrespondents’ own valuation of the time required to search(Moorthy et al. 1997; Srinivasan and Ratchford 1991) orindirectly assessing their opportunity costs of time fromother indicators (Punj and Staelin 1983).


Second, as discussed above, the theoretical predic-tion of search costs affecting search behavior has beensupported in many experiments. However, few field stud-ies (Moorthy et al. 1997) show this effect. In other fieldresearch, the effect of search cost has been found to beeither marginally significant (Punj and Staelin 1983) orcompletely non-significant (Srinivasan and Ratchford1991). We recognize that this discrepancy between ex-perimental and field results may be driven by the numer-ous other differences between the two settings. However,these results do underscore the importance of isolating theeffect of currency on search decisions.

Finally, time and money are inherently interestingbecause despite being economically equivalent consump-tion resources (Becker 1965), they differ in terms of theirrisk aversion (Leclerc, Schmitt, and Dube 1995; Okadaand Hoch 2004), consideration of sunk costs (Soman2001), and discounting (Soman 1998; Zauberman andLynch 2005). The crux of these papers is that whenconsumers are faced with a decision problem, they ana-lyze the relevant information in a different manner if theinformation pertains to time than to money. What thisstream does not tell us, however, is whether a qualitativelydifferent form of decision making – making decisionsbased on heuristics – gains prominence when one workswith one currency rather than the other. This question canbe well studied in the context of search not only becausesearch decisions are an integral part of the consumptionexperience but also because unlike purchase situations, inwhich money is the dominant currency, search situationsdo frequently involve time as well.

THEORETICAL DEVELOPMENT

Both theory (Stigler 1961) and monetary experi-ments (Kogut 1992; Schotter and Braunstein 1981) sug-gest analytic decision making in which people carefullyconsider the trade-off between costs and payoffs. Al-though this decision making is consistent with the econo-mists’ view of a rational choice theory, the information-processing approach (Bettman 1979) suggests that such atheory is incomplete; decision makers often do not pro-cess information methodically because they are limited intheir capacity to do so (Simon 1955). Prior work suggeststhat when the currency is that of time rather than money,people are indeed less able to process information.

Soman (2001) shows that individuals do not accountfor past investments of time the way they account formoney because they find it harder to do accounting fortime. He argues that people face difficulties in accountingfor time because they do not routinely transact in time theway they do with money. Although people in someprofessions (e.g., lawyers) do keenly monitor their timeexpenditures, most other people are not trained to do so.

Soman (2001) does find that time-money differencesdisappear when the accounting of time is facilitated byvarious means: providing a wage rate, increasing salienceof opportunity costs, and educating about economic ap-proaches to time. Inability to account for time could alsoarise because unlike money which is unambiguous – adollar is a dollar in all circumstances – time cannot beprecisely assessed because its use may vary from onesituation to another; the value of a temporal expense is notconcrete but open to interpretation (Okada and Hoch2004). That time is hard to account for is also evident fromthe fact that people are not adept at judging extendedexperiences based on their duration. Apart from somesituations, such as when individuals can easily comparedurations (Ariely and Loewenstein 2000), people show aduration neglect (Varey and Kahneman 1992). Thesefindings suggest that it is indeed harder to work with timethan with money.

If people are limited in their capacity to scrutinize theinformation provided, they are less likely to be able toidentify and process the relevant and important pieces(Alba and Hutchinson 1987) and less likely to analyze theinformation in order to arrive at a judgment (Ratneshwarand Chaiken 1991). What they might do instead, is to relyon simple thumb rules that they find easier to work with(Payne, Bettman, and Johnson 1993). This idea of relyingon cognitive short-cuts instead of a careful analysis ofrelevant information is also integral to many dual-processmodels (see Chaiken and Trope 1999 for a review). Forinstance, in the heuristic-systematic model (Chen andChaiken 1999), people make judgments based on simplecues or thumb rules if they are limited in their cognitiveability or capacity to scrutinize judgment-relevant infor-mation. Taken together, the above discussion suggeststhat when individuals work with temporal search costs,their search decisions will be driven by heuristics becausetime is not easy to work with. When individuals work withmonetary search costs, however, they are less likely to relyon unrelated heuristics because they are quite adept ataccounting for expenditures of money.

We next present four non-sequential-search experi-ments. In such settings, the search decision is taken priorto doing any search (Burdett and Judd 1983). For ex-ample, one can decide on the number of universities tovisit before actually starting travel. Similarly, a consumertrying to buy a book at the lowest price could make an apriori decision about the total number of bookstores tovisit in order to check prices. Experiment 1 shows thatpeople use self-generated heuristics more often when thecurrency is time rather than money. Experiment 2 showsthat when a heuristic strategy is made available – theoption to choose the middle of three options – people usethat strategy more for time than for money. Experiment 3shows a similar effect for the anchoring heuristic. Experi-


ment 4 offers evidence for the heuristic process viaresponse time measures, and by showing how this time-money difference can be eliminated.

EXPERIMENT 1

Overview

This experiment was a preliminary test of our predic-tion. Participants considered searching for a moving com-pany. If one chose to search more (i.e., invite more movingcompanies for estimates), the likelihood of a desirablepayoff (i.e., finding the cheapest moving company) in-creased but so did the search costs (i.e., cost of inviting themoving companies).

Design

A between-subjects design was used in which cur-rency of search (time vs. money) was manipulated. Ourmain dependent variable was the extent to which a heuris-tic strategy was mentioned in the verbal protocol that therespondents wrote.

Procedure

Forty-three undergraduate students participated inthis experiment in exchange for partial course credit. Inthe following paragraph, underlining highlights the twoconditions, to which the participants were randomly as-signed. As is clear from the scenario, we considered $5 tobe roughly equivalent to 30 minutes. This wage rate isbased on our pretests and is close to recent work that hasemployed a wage rate of $12.50 per hour (Okada andHoch 2004). However, it needs to be noted that an exactequivalence between the two is not critical to our experi-ment.

Imagine that you are moving to a different locationand have decided to hire a moving company. Becauseyou have never tried movers before, you ask a knowl-edgeable friend for help. He picks up the YellowPages and tells you the names of 50 moving compa-nies that provide a good level of service. He adds thatthe prices could vary a lot – anywhere from $500 to$1000 for the amount of stuff you have. He thereforesuggests that you randomly choose some of these 50moving companies, get them to your apartment sothat they can give you their price estimates, and thenpick the one that is the cheapest.

The problem that you now face is to decide how manyyou should get home for an estimate. To get thecheapest rate, the best thing to do would be to let eachof 50 companies visit your apartment and give you anestimate so that you may pick the cheapest one.However, there is a cost involved in terms of the

amount of money/time you spend on this activity.Each moving company will charge $5/take 30 min-utes to inspect your stuff and provide an estimate.

Out of the 50 moving companies that you are consid-ering, how many are you going to get over for priceestimates? (Please provide one number, not a range.)

I will ask ______ moving companies to come overand provide their price estimates.

After participants indicated the willingness to search(i.e., no. of companies), they turned to the next page andtook two minutes to explain the thought process that led totheir answer.

RESULTS AND DISCUSSION

We did not have any prediction regarding the willing-ness to search. The results indicated that there was nostatistical difference between time and money (F(1,41) =.6; p > .44) but, numerically, the willingness to search washigher for money (M = 6.7) than for time (M = 5.8). Toassess the variable of interest, we analyzed the cognitiveresponses – the comments that the respondents wroteabout their thought process. We asked two independentjudges who were blind to the experimental conditions tocode participants’ responses as either 0 (non-heuristic-strategy) or 1 (heuristic-strategy). Through examples anddiscussions, it was explained to the judges that heuristicprocessing is reflected if participants rely on their ownthumb rules in order to arrive at their search decision. Thejudges coded some responses as 0 (“If each companycharged $5, I’d be willing to spend $50 to get estimates,which is 10 companies” or “I thought about how muchtime I wanted to spend), and other responses as 1 (“Ask 3companies, so you have a high, middle, and low range.”“Pick 1 out of every 10, hence 5 companies in all”).

Out of the 43 responses, the two judges agreed on 41(r = .83; Cohen’s = .81). They then discussed the twodissimilar responses and arrived at a consensus. We usedthe mutually-agreed list of responses to analyze 22 partici-pants in the time condition and 21 participants in themoney condition. The number of respondents who indi-cated having used a heuristic strategy was higher for timethan for money; 9 (40.9%) for time and 3 (14.3%) formoney (z = -1.95; p < .03).

These results provide preliminary evidence for thehigher use of heuristics in time (vs. money). However,because we assessed self-generated heuristics, we couldnot verify the causal mechanism. We therefore manipu-late the availability of context-based heuristics – menu-dependence and anchoring – to better examine the differ-ential use of heuristics.


Menu dependence (Huber, Payne, and Puto 1982;Simonson 1989) occurs if preferences for options areirrationally influenced by the menu of options in thechoice set. One type of menu dependence, the compro-mise effect, runs counter to the notion of similarity. Giventwo options B and C, the inclusion of an option that is moresimilar to B (say, A) ought to reduce the preference for B(relative to C) because some of those who earlier chose Bcan now choose the adjacent option A. However, Simonson(1989) demonstrates that including A increases the pref-erence for B. Because B is the middle option in the {A, B,C} set, it is an easy-to-choose compromise option. In thecontext of search, one can envision two options regardingthe willingness to search (B = low search and C = highsearch). The inclusion of a third option (A = lowestsearch) would make B the compromise option. Given ourpredictions, the compromise effect is more likely to emergefor time than for money. If accounting for temporal searchcosts is hard, choosing the compromise option B willrepresent an easy-to-use thumb rule to decide whichsearch option to pick. Therefore, the preference for B(relative to C) will increase when A is introduced. Thiseffect will be relatively weaker for money because thesearch decision will be driven by analysis of search costsrather than a reliance on thumb rules. We test this idea inexperiment 2.

Anchoring (Tverksy and Kahneman 1974) is thetendency to rely heavily, or anchor, on one piece ofinformation when making decisions. For instance, thewillingness to pay for a product can be influenced by anarbitrary anchor such as the last two digits of one’s SocialSecurity number (Simonson and Drolet 2004). Whenconsumers decide on how much to search, their decisionscould similarly be anchored by numbers that they are firstexposed to. If the currency is time (vs. money), people aremore likely to rely on the anchoring heuristic because,given the difficulty in accounting for search costs, theanchor provides an easy way to arrive at a search decision.We test this in experiment 3. In experiment 4, we offerevidence for the heuristic process via response time mea-sures, and by incorporating a prime for accounting.

EXPERIMENT 2

Overview

The experimental stimuli were similar to that ofexperiment 1. The main difference was that rather thanhaving to decide on the number of moving companies thatthey would invite, participants had to choose one searchoption (i.e., number of companies to invite) from a menuof options. Our prediction was that a change in the menuof search options (two vs. three options) would lead to astronger change in the relative share of the focal option ifthe currency of search was time rather than money. The

focal option was the lower-search option in the binary setbut the compromise (i.e., between lowest and highestsearch options) in the trinary set.

Design

A between-subjects design was used in which bothcurrency of search (time vs. money) and menu of searchoptions ({3,5} or {1,3,5}) were manipulated. The trinaryset includes one option (i.e., invite 1 company) that ismore similar to the search option of 3 companies than to5 companies. Therefore, in the absence of a compromiseeffect, the 1-company option would more adversely affectthe share of the 3-companies option rather than the 5-companies option. However, if people rely on the com-promise effect, the share of the 3-companies option willactually increase relative to the 5-companies option(Simonson 1989). This change in the relative share of the3-companies option (vs. the 5-companies option) was thedependent variable.

Procedure

In exchange for partial course credit, two hundredand eleven undergraduates participated in this experi-ment. After reading the main scenario, they read thefollowing in the binary-set:

As you think about the number of moving companiesyou should get over for estimates, you considergetting either 3 or 5 moving companies. So you needto make a decision now. How many of the 50 movingcompanies will you get over for estimates? (Pleasemark only one of the following two options.)

I will ask 3 companies to come over for priceestimates.


In the trinary set, participants read the following afterhaving read the main scenario:

As you think about the number of moving companiesyou should get over for estimates, you considergetting 1, 3, or 5 moving companies. So you need tomake a decision now. How many of the 50 movingcompanies will you get over for estimates? (Pleasemark only one of the following three options.)

I will ask 1 company to come over for a priceestimate.




Results

We employed a binary logistic regression model inwhich we analyzed the number of people choosing the 3-companies option or the 5-companies. The dependentvariable was the relative share. The currency of searchand the menu of search options were the between-subjectsindependent variables. Figure 3 depicts the percentagerelative share ((No. choosing 3 companies / No. choosingeither 3 or 5 companies) * 100).1

There was no main effect of either currency (β = .22;Wald = .52; p > .47) or the menu of search options (β =-.47; Wald = 2.43; p > .11) but the predicted interactionwas significant. The menu-of-options x currency interac-tion, (β = -1.29; Wald = 4.62; p < .05) confirmed that thecompromise effect emerged differently in the moneyversus time conditions. Specifically, when the currencywas time, the focal 3-companies option was chosen by 28out of 52 (53.9%) participants in the binary condition and39 out of 50 (78.0%) in the trinary condition. In line withthe compromise effect, this increase was significant (z =-2.25; p = .01); the share of the 3-companies optionrelative to the 5-companies option increased when the 1-company option was introduced. In contrast, there was nostatistical change in relative share when the currency wasmoney (z = .43; p > .67). The 3-companies option waschosen by 34 out of 53 (64.2%) participants in the binarycondition and by 30 out of 50 (60.0%) in the trinary

condition. When offered the easy solution of relying on acompromise heuristic, participants relied on that optionmore in time (vs. money). We now examine time-moneydifferences in the anchoring heuristic.

EXPERIMENT 3

Overview

Participants considered the situation of searching forinformation before purchasing a used car. Although itwould be desirable to find out all information about eachused car being considered, searching for such informationusually involves certain costs, as is evident by the pres-ence of companies that sell vehicle history reports for aprice (e.g., Carfax®). The scenario clearly represented therelationship between search costs and search behavior. Ifone chose to search more (i.e., view more historicalrecords), the likelihood of a desirable outcome (i.e.,finding a used car of acceptable quality) increased but sodid the search costs (i.e., cost of obtaining the records).The classic two-step procedure was used to anchor par-ticipants (Tverksy and Kahneman 1974). First, they wereasked whether the number of records they view would beless than or equal to a certain number, or more than thatnumber. Second, they were asked to provide the exactnumber of records they would view. Our prediction wasthat the change in the anchor value (2 vs. 40) would morestrongly influence the willingness to search (i.e., exactnumber of records to view) when search is conducted intime rather than money.

M o n e yT im e

C u r r e n c y o f S e a r c h

1 0 0

9 0

8 0

7 0

6 0

5 0

4 0

3 0

2 0

1 0

0

Re

lat

ive

Sh

are

(%

Ch

oo

sin

g 3

R

1 , 3 , o r 5

3 , o r 5

M e n u o f S e a r c h O p t i o n s ( N o . o f

C o m p a n i e s )

FIGURE 1

Experiment 2: Effect of Search Currency and Menu of Options on Relative Share of One

Specific Option to Search for a Cheap Moving Company


Design

A between-subjects design was used in which cur-rency of search (time vs. money) and anchor value (lowvs. high) were manipulated. The dependent variable wasthe willingness to search – the number of historical recordsthat the participants wanted to view.

Procedure

One hundred and forty-one undergraduate studentsparticipated in this experiment in exchange for partialcourse credit. In the following paragraph, underlining isused to highlight the four conditions to which the partici-pants were randomly assigned.

You’re planning to buy a used car within the next fewdays. Your friends recommend that you consult areputed used-car website which offers extensive in-formation that is hard to find anywhere else. On thiswebsite, you cannot only view the listings of the usedcars but can also view detailed historical records(accidents, water damage, number of previous own-ers, etc.).

You go to the used-car website, put in your searchcriteria (price range, brand name, etc.) and find thatthere are 80 cars that meet these criteria. You arenow thinking about viewing their historical records.However, the website informs you that it takes $1 / 5minutes to view the details of one historical record.Therefore, the more records you view, the moremoney / time you spend on this website. You now haveto decide how many of these 80 historical records youare going to view.

Will you view up to 2 / 40 records or will you viewmore than 2 / 40 records? (Mark one option) (a) I willview 2 / 40 records or less (b) I will view more than2 / 40 records.

Exactly how many records will you view? (Write onenumber, not a range.)

I will view _______ records.


Two participants provided inconsistent responses tothe two questions indicating that they had not read thequestions well. For example, one participant marked “Iwill view 2 records or less” in response to the first questionand then marked 5 records in response to the secondquestion. Therefore, they were excluded. The followingresults are based on a sample size of 139 rather than 141.We used a between-subjects design in which willingness

to search was the dependent variable and currency ofsearch and anchor value were the between-subjects inde-pendent variables. Figure 2 depicts the results of theAnalysis of Variance.

The currency x anchor interaction was significant(F(1, 135) = 3.5; p = .06) confirming that the willingness-to-search difference between the low and high anchorconditions was greater for time (M = 14.6) than for money(M = 3.4). That is, when the anchor value increased (from2 to 40), participants in the time (vs. money) conditionswere more sensitive to this increase. In fact, the effect ofanchoring was significant only for time, not for money.The willingness to search in the time condition was higher(F(1, 135) = 13.5; p < .001) when anchor value was high(M = 23.7) rather than low (M = 9.1), but the willingnessto search in the money condition was statistically the same(F(1, 135) = .57; p > .44) irrespective of whether theanchor value was high (M = 19.2) or low (M = 15.8).Consistent with our earlier experiments, we demonstratethat people are more susceptible to relying on heuristicswhen the currency is time (vs. money).

EXPERIMENT 4

Overview

The experimental stimuli were similar to that ofexperiment 3 with a few exceptions. One, we used highersearch costs ($3/15 minutes instead of $1/5 minutes) tocheck whether our results replicate. Two, we manipulatedan accounting prime to be either absent or present. Givenour theorizing about heuristic strategies being driven bythe fact that people do not usually account for time (vs.money), an accounting prime should get the time partici-pants closer to the money participants. As an additionalcheck, we also measured response times to see if theaccounting prime does indeed make the time participantsspend more time in doing relevant calculations, ratherthan simply relying on the anchor. To that effect, wemeasured the response time for the willingness-to-searchquestion (i.e., specific number of records to view) – thetime it took from the moment this question was posed tothe moment it was answered. To enable measurement ofresponse time, we conducted this study using MediaLabsoftware.

Design

A between-subjects design was used in which cur-rency of search (time vs. money), anchor value (low vs.high), and accounting prime (absent vs. present) weremanipulated. The dependent variables were the willing-ness to search (the number of historical records that theparticipants wanted to view) and the associated responsetime.


Procedure

One hundred and forty-five undergraduate studentsparticipated in exchange for partial course credit. If par-ticipants were assigned to the time (money) condition,they either saw the scenario, as in experiment 3, or wrotean essay on the value of time (money) as follows:

Please take a couple of minutes to write a short essayon THE VALUE OF TIME/MONEY. In this essay,please explain how time / money is a resource thatone should spend carefully.

When accounting prime was manipulated to bepresent, participants saw the above instructions, typed intheir essays, and pressed “Continue.” The computer thendisplayed the used-car scenario. When accounting primewas absent, participants were directly presented with theused-car scenario. At the bottom of the scenario, theanchor was manipulated by asking participants to chooseone of two options; they marked whether they would viewless than or equal to, or more than, a certain number ofrecords (2 vs. 40). Then, on the next screen, participantswere asked to type in the exact number of records theywould view.


As in experiment 3, some participants were excludedbecause their responses revealed that they had not read the

questions. The results are based on a sample of 143 ratherthan 145.

Willingness to Search. When the accounting prime isabsent, we expect to replicate the results of experiment 3such that a change in the anchor value (2 vs. 40) shouldmore strongly impact willingness to search when thecurrency is time (vs. money). When the accounting primeis present, we expect this difference between time andmoney to diminish. To test our predictions, we used abetween-subjects design in which willingness to searchwas the dependent variable and currency of search, an-chor value, and accounting prime were the between-subjects independent variables. Figure 3 depicts the re-sults of the Analysis of Variance.

As is evident from Figure 3 and the currency x anchorx accounting-prime interaction (F(1, 135) = 2.8; p < .1),the accounting prime changed the pattern. When account-ing prime was absent, the currency x anchor interaction(F(1, 135) = 3.4; p = .06) was similar to that of experi-ment 3; the willingness-to-search difference between thetwo anchor conditions was greater for time (M = 23.6)than for money (M = 8.4). The anchoring effect wassignificant only for time, not for money; the willingness tosearch in the time condition was higher (F(1, 135) = 16.3;p < .001) when anchor value was high (M = 29.4) ratherthan low (M = 5.9), but the willingness to search in themoney condition was statistically the same (F(1, 135) =2.1; p > .15) irrespective of whether the anchor value was

NOTE. Numbers are means (standard deviations). Error bars show 95 percent confidence intervals.

M o n e yT i m e


3 5

3 0

2 5

2 0

1 5

1 0

5

0

Willi

ng

ne

ss

t

o S

ea

rc

h (

Nu

mb

er

o

H i g h

L o w

A n c h o r V a l u e

FIGURE 2

Experiment 3: Effect of Search Currency and Anchor Value on Willingness to

Search for Information about Used Cars


high (M = 11.8) or low (M = 3.4). When accounting primewas present, however, these time-money differences van-ished. The currency x anchor interaction was not signifi-cant (F(1, 135) = .3; p > .58); the willingness-to-searchdifference between the two anchor conditions was similarfor time (M = 5.9) and money (M = 10.3).

Response Time. Based on our theorizing about heu-ristic processing in time, we expect an interaction ofaccounting prime with search currency. When partici-pants are asked to indicate the specific number of recordsthey are going to view, those in the money condition arelikely to calculate their response based on a relatively

precise calculation regarding the search costs and thesearch payoffs. They are likely to do this accountingirrespective of whether the prime is absent or present.Consequently, the accounting prime is not expected tohave a significant influence on the time taken to respond.Conversely, however, those in the time condition are moreinclined to rely on a heuristic. So if the accounting primeis absent, they are likely to come up with a quick answerbased on the anchor they have already formed. But if theaccounting prime is present, they are likely to spend timeon a more precise calculation based on search costs.Consequently, the accounting prime is expected to sig-nificantly increase the time taken to respond. To test our

FIGURE 3

Experiment 4: Effect of Search Currency, Anchor Value, and Accounting Prime on

Willingness to Search for Information about Used Cars

NOTE. Numbers are means (standard deviations). Error bars show 95 percent confidence intervals.


M o n e yT i m e

Wi

llin

gn

es

s t

o S

ea

rc

h (N

um

be

r o

f R

4 5

4 0

3 5

3 0

2 5

2 0

1 5

1 0

5

0

H i g h

L o w



M o n e yT i m e

Wi

lli

ng

ne

ss

t

o S

ea

rc

h (N

um

be

r o

f R

ec

or

ds

V

iew

ed

)

4 5

4 0

3 5

3 0

2 5

2 0

1 5

1 0

5

0

H i g h

L o w


ACCOUNTING PRIME = ABSENT

ACCOUNTING PRIME = PRESENT

5.9 (3.2)

29.4 (29.2)

3.4 (2.6)

11.8 (18.2)

11.9 (14.1)

17.8 (18.2)

5.3 (11.3)

15.7 (24.1)

11.9 (14.1)


predictions, we used a between-subjects design in whichcurrency of search, anchor value, and accounting primewere the between-subjects independent variables. Thedependent variable used was log-transformed willingnessto search. This transformation was done in order to adjustfor extreme response times. However, the results reportedbelow are substantively identical to the results we ob-tained without doing any transformation. To ease inter-pretation, the means reported below are the actual re-sponse times (in milliseconds).

Although we used the complete model to use the fullinformation, there is no reason to expect anchor value tohave any impact on the response time; the currency xanchor x accounting-prime interaction was not significant(F(1, 135) = .7; p > .39). The only term statisticallysignificant in the model was the critical currency x ac-counting-prime interaction (F(1, 135) = 8.2; p < .01).Specifically, for money, the response time was statisti-cally the same (F(1, 135) = 1.1; p > .29) irrespective ofwhether the accounting prime was absent (M = 13065) orpresent (M = 11204). But, for time, the response time wasgreater (F(1, 135) = 8.9; p < .01) when the accountingprime was present (M = 14912) than when it was absent(M = 9144). These results for response time indicate thatpriming accounting does indeed make time participantsspend more time on accounting. This is consistent with thewillingness-to-search results that indicate that the ac-counting prime makes time participants rely less on heu-ristics

Viewing the currency x accounting-prime interactionfrom a different perspective provided new insights intothe heuristic process for time. Specifically, when theaccounting-prime was absent, the response time was

greater for money than for time (F(1, 135) = 2.9; p < .1).Conversely, when the accounting-prime was present, theresponse time was greater for time than for money (F(1,135) = 5.5; p < .05). These results provide convergingevidence for the process we propose. When the account-ing prime is absent, those in the money condition arerelatively slow because they answer the question bythinking carefully about the provided information regard-ing search costs and payoffs; those in the time conditionare relatively fast because they simply rely on the anchorthat they have already formed. When the accountingprime is present, those in the money condition are notaffected because they continue to do the analysis that theywould have done anyway. However, the prime makesthose in the time condition do a careful analysis as well.And because it is not easy to work with search costs intime, it takes longer to reach an answer relative to the timeit takes for search costs in money.

GENERAL DISCUSSION

The finding that people are more prone to usingheuristics in time (vs. money) provides a new view oftime-money differences. It also helps us grasp the searchprocess better because it suggests a significant role forsearch currency. Managers need to be aware of the searchcurrency of their consumers. For instance, when trying tosell a house, one could either pay a real-estate agent orspend one’s own time looking for the highest bidder.Whether people use money or time has implications forother consumers (i.e., home buyers) as well as businesses(i.e., real-estate agents). Therefore, businesses need toadapt to the currency of search used by consumers, or tryto manage it, in order to succeed in terms of higher profitsand more satisfied consumers.

ENDNOTE

1 Given our focus on the compromise effect (Simonson1989), we examined the share of the 3-companiesoption relative to the 5-companies option. Becausethe 1-company option exists only in the trinary con-

dition, its results cannot be compared to the binarycondition. However, for the reader who might inter-ested in the number of people who chose the 1-company option in the trinary conditions, it was 2 fortime and 4 for money.

REFERENCES

Alba, Joseph A. and J. Wesley Hutchinson (1987), “Di-mensions of Consumer Expertise,” Journal of Con-sumer Research, 13 (March), 411–54.

Ariely, Dan and George Loewenstein (2000), “WhenDoes Duration Matter in Judgment and DecisionMaking?” Journal of Experimental Psychology:General, 129 (December), 508–23.

Becker, Gary S. (1965), “A Theory of the Allocation ofTime,” Economic Journal, 75 (September), 493–517.

Bettman, James R. (1979), An Information ProcessingTheory of Consumer Choice. Reading, MA: Addison-Wesley Publishing Co.

Burdett, Kenneth and Kenneth Judd (1983), “EquilibriumPrice Dispersion,” Econometrica, 51 (July), 955–69.

Chaiken, Shelly and Yaacov Trope (1999), Dual-Process


Theories in Social and Cognitive Psychology. NewYork: Guilford.

Chen, Serena and Shelly Chaiken (1999), “The Heuristic-Systematic Model in its Broader Context,” in Dual-Process Theories in Social and Cognitive Psychol-ogy, Shelly Chaiken and Yaacov Trope, eds. NewYork: Guilford.

Cox, James C. and Ronald L. Oaxaca (1989), “LaboratoryExperiments with a Finite Horizon Job-SearchModel,” Journal of Risk and Uncertainty, 2 (Septem-ber), 301–29.

Davis, Douglas D. and Charles A. Holt (1993), “Deci-sions and Games,” in Experimental Economics,Princeton: Princeton University Press, 67–124.

Diehl, Kristin (2005), “When Two Rights Make a Wrong:Searching Too Much in Ordered Environments,”Journal of Marketing Research, 42 (August), 313–22.

Fredrickson, B.L. and Daniel Kahneman (1993), “Dura-tion Neglect in Retrospective Evaluations of Affec-tive Episodes,” Journal of Personality and SocialPsychology, 65 (July), 45–55.

Huber, Joel, John W. Payne, and Christopher Puto (1982),“Adding Asymmetrically Dominated Alternatives:Violations of Regularity and the Similarity Hypoth-esis,” Journal of Consumer Research, 9 (June), 90–98.

Kogut, Carl A. (1992), “Recall in Consumer Search,”Journal of Economic Behavior and Organization, 17(January), 141–51.

Leclerc, France, Bernd H. Schmitt, and Laurette Dube(1995), “Waiting Time and Decision Making: IsTime Like Money?” Journal of Consumer Research,22 (June), 110–19.

Moorthy, Sridhar, Brian T. Ratchford, and DebabrataTalukdar (1997), “Consumer Information Search Re-visited: Theory and Empirical Analysis,” Journal ofConsumer Research, 23 (March), 263–77.

Okada, Erica M. and Stephen J. Hoch (2004), “SpendingTime versus Spending Money,” Journal of Con-sumer Research, 31 (September), 313–23.

Payne, John, James Bettman, and Eric Johnson (1993),The Adaptive Decision Maker. New York: Cam-bridge University Press.

Punj, Girish N. and Richard Staelin (1983), “A Model ofConsumer Information Search Behavior for NewAutomobiles,” Journal of Consumer Research, 9(March), 366–80.

Ratneshwar, S. and Shelly Chaiken (1991), “Comprehen-sion’s Role in Persuasion: The Case of its ModeratingEffect on the Persuasive Impact of Source Cues,”Journal of Consumer Research, 18 (June), 52–62.

Rosen, Dennis L. and Richard W. Olshavsky (1987), “TheDual Role of Informational Social Influence: Impli-cations for Marketing Management,” Journal of Busi-ness Research, 15 (April), 123–44.

Schotter, Andrew and Yale M. Braunstein, (1981), “Eco-

nomic Search: An Experimental Study,” EconomicInquiry, 19 (January), 1–25.

Simon, Herbert A. (1955), “A Behavioral Model of Ratio-nal Choice,” Quarterly Journal of Economics, 69(February), 99–118.

Simonson, Itamar (1989), “Choice Based on Reasons:The Case of Attraction and Compromise Effects,”Journal of Consumer Research, 16 (September),158–74.

____________ and Aimee Drolet (2004), “AnchoringEffects on Consumers’ Willingness-to-Pay and Will-ingness-to-Accept,” Journal of Consumer Research,31 (December), 681–90.

Smith, Gerald E., Meera P. Venkatraman, and Ruby RoyDholakia (1999), “Diagnosing the Search Cost Ef-fect: Waiting Time and the Moderating Impact ofPrior Category Knowledge,” Journal of EconomicPsychology, 20 (June), 285–314.

Soman, Dilip (1998), “The Illusion of Delayed Incen-tives: Evaluating Future Effort-Money Transactions,”Journal of Marketing Research, 35 (November),427–37.

____________ (2001), “The Mental Accounting of SunkTime Costs: Why Time Is Not like Money,” Journalof Behavioral Decision Making, 14 (July), 169–185.

Srinivasan, Narasimhan and Brian T. Ratchford (1991),“An Empirical Test of a Model of External Search forAutomobiles,” Journal of Consumer Research, 18(September), 233–42.

Srivastava, Joydeep and Nicholas Lurie (2001), “A Con-sumer Perspective on Price-Matching Refund Poli-cies: Effect on Price Perceptions and Search Behav-ior,” Journal of Consumer Research, 28 (Septem-ber), 296–307.

Stigler, George J. (1961), “The Economics of Informa-tion,” Journal of Political Economy, 69 (June), 213–25.

Tversky, Amos and Daniel Kahneman (1974), “JudgmentUnder Uncertainty: Heuristics and Biases,” Science,27, 185 (September), 1124–31.

Urbany, Joel E. (1986), “An Experimental Examinationof the Economics of Information,” Journal of Con-sumer Research, 13 (September), 257–71.

Varey, Carol and Daniel Kahneman (1992), “ExperiencesExtended Across Time: Evaluations of Moments andEpisodes,” Journal of Behavioral Decision Making,5 (July/September), 169–85.

Zauberman, Gal and John G. Lynch, Jr. (2005), “Re-source Slack and Propensity to Discount DelayedInvestments of Time versus Money,” Journal ofExperimental Psychology: General, 134 (February),23–37.

Zwick, Rami, Amnon Rapoport, Alison King Chung Lo,and A.V. Muthukrishnan (2003), “Consumer Se-quential Search: Not Enough or Too Much,” Market-ing Science, 22 (Fall), 503–19.


For further information contact:Ashwani Mongo

University of Texas, San AntonioSan Antonio, TX 78249–0631

Ritesh SainiGeorge Mason University

Phone: 703.993.1796E-Mail: [email protected], VA 22030–4444


THE ROLE OF RECIPROCITY, SYMPATHY, AND GENEROSITY IN

PREDICTING TIPPING BEHAVIOR

Gary Daniel Futrell, Florida State University, Tallahassee

SUMMARY

Though its origin is unknown, the practice of tippinghas become commonplace in many service sectors –hairdressers, cab drivers, ushers, waiters, and waitressesare among the most commonly tipped service workers. Itis estimated that in the United States, $16–$27 billion isgiven away in tips each year (Azar 2004b; Conlin, Lynn,and O’Donoghue 2003; Lynn and McCall 2000a). Tip-ping is an important part of the exchange process, yet littlehas been written about this subject in marketing literature(Koku 2005).

The primary objective of this study is to create afoundation for a predictive tipping behavior model thatfuture researchers and managers can use to identify moti-vating factors that determine tipping behavior. The pro-posed model suggests that obligation (a negative affectiveconstruct often accompanied by feelings of indebtednessthat create great discomfort), gratitude (an enjoyable statethat has been shown to be a better predictor of futurecompliance than obligation (Goei and Boster 2005;Watkins et al. 2006)), and sympathy (a person’s aware-ness of the feelings of others or the capacity to respond tothe concerns of others (Escalas and Stern 2003)), havedirect effects on tipping behavior. Past theoretical studiessuggest that the effects of gratitude and obligation (theprimary agents of reciprocity) are moderated by differ-ences in personal variables, such as generosity, that deteror enhance social behavior motivated by reciprocity.

Three hundred and ten online surveys were collectedfrom undergraduate students at a large southeast Ameri-can university. The 45-item survey included establishedLikert-type scales for measuring gratitude (Kolyesnikova2006), obligation, and sympathy (Escalas and Stern 2003).A semantic differential scale was developed by the authorto measure generosity. Tipping behavior, the primarydependant variable, was measured by asking respondentsto indicate how much they would tip given a $30 food billat a casual dining restaurant.

Of the 310 surveys, 160 were completed by femalestudents (51.6%). The large majority of respondents de-

scribed themselves as White/Caucasian (74.8%); othersdescribed themselves as Black/African American (9.4%),Hispanic/Latino (9%), Asian (3.5%), or “other” (3.2%).Participant ages ranged from 18 to 49, with a mean andmode of 20.

The mean tip amount in response to the tippingscenario was $5.44, or 18.13 percent of the $30 food bill.Among the various ethnic groups, Whites had the highesttipping rate (18.5%) and Blacks the lowest (16.3%).Significance tests show no significant difference betweenmale and female respondents.

The psychometric properties of five variables with atotal of 13 items were tested simultaneously in one modelconsistent with Anderson and Gerbing’s (1988) two-stepapproach for structural equation modeling (SEM). SRMR,RMSEA, and TLI were .036, .030, and .998 respectively;with a χ2 of 71.9 and 56 degrees of freedom (p < .074).Together, these results indicate a good fit to the data andthe results of the structural test partially support theproposed model. Hypothesis 1 (gratitude has a direct andpositive influence on tipping behavior), and Hypothesis 2(obligation has a direct and positive influence on tippingbehavior) were supported, but Hypothesis 3 (sympathyhas a direct and positive influence on tipping behavior)was not. Additionally, the moderating effects of generos-ity (Hypothesis 4) were not supported, that is, no differ-ence was found between highly generous individuals, andthose who are not.

The results of this study suggest that consumers aremotivated to tip by both gratitude and obligation, therebyproviding additional support for applying the norm ofreciprocity to explain this unique phenomenon in whichconsumers voluntarily pay extra for services rendered. Asan expression of gratitude, tipping allows patrons toreward service providers for their efforts while at the sametime presenting opportunities for altruistic behavior. Thereciprocal nature of altruistic acts between patrons andservice providers may explain (at least in part) the per-petuation of tipping as a social norm. References areavailable upon request.


For further information contact:Gary Daniel Futrell

The College of BusinessFlorida State University

821 Academic WayP.O. Box 3061110

Tallahassee, FL 32306–1110Phone: 850.644.3869



TO HAVE OR TO HOLD: THE INFLUENCE OF PAYMENT METHOD ON

CONSUMER SATISFACTION

Michael T. Krush, University of Nebraska – LincolnAubrey R. Fowler III, University of Nebraska – Lincoln

SUMMARY

The acquisition of goods is not always accomplishedthrough the outright purchase of those goods. In factmarketers utilize a range of payment methods which allowconsumers to gain possession of but not necessarily own-ership of the product. Today, consumers may rent or leasedurable products such as cars (Howard 2005), excavationequipment (Lovelock and Gummesson 2004) yachts(Wetschler 2006), and specialty and fashion goods likehandbags (Jackson 2005). Despite the substantial growthin the rental and leasing of consumer goods, a relativelyunexplored area lies in understanding whether alternativepayment methods such as renting and leasing influenceconsumer satisfaction, especially in comparison to moretraditional payment methods that result in ownership.Hence, the goal of this paper is to address the gap in ourunderstanding of consumer satisfaction and its relation-ship to payment methods. By doing so, our theoreticalcontribution results from revealing the levels of satisfac-tion that emanate from various payment methods and byexpanding upon a rather limited set of research on rentingand leasing (Trocchia and Beatty 2003). In addition toexamining the link between satisfaction and the leasing/rental versus ownership dichotomy, we argue that certainconsumer characteristics (i.e., the consumer’s level ofmaterialism and psychological ownership) may influencethe payment method-satisfaction relationship.

Constructs Examined

We conceptualize payment method in two ways, asownership, (i.e., whether one gains legal title to theproperty through outright purchase or financing)(Obenberger and Brown 1976b) or as access (i.e., a legalarrangement of rental or leasing payments in which theconsumer agrees to give back the good at some pre-ordained time) (Patton 1990; Rifkin 2000). To focus ourinvestigation, we concentrate on the consumer market,rather than the business segment markets and we narrowour examination to a specific class of consumer products,shopping goods (Murphy and Enis 1986). We utilize theshopping goods construct (Murphy and Enis 1986) as it isbased on two key pricing elements (risk and effort). As ourcentral focus is on a form of a pricing tactic (i.e., accessversus ownership terms) we find a price-based typologyaligns with the nature of our study. We also follow thelogic of Oliver (1993), Spreng et al. (1996) and Gardial

et al. (1994) and conceive of a new construct, acquisitionsatisfaction that serves as an antecedent to overall satis-faction. This marketer-generated aspect of satisfaction isdefined as the consumer’s appraisal of the pricing termsand structure through which they attained the product.

Hypotheses

Next, we build a number of hypotheses that relatepayment method with acquisition satisfaction. Previousresearch builds a strong argument that links consumerswho utilize an access payment method with greater levelsof acquisition satisfaction than those who use an owner-ship payment method. The literature suggests the struc-ture of access payment options provides consumers witha range of benefits that potentially yield greater acquisi-tion satisfaction than those who own. These benefitsinclude a decreased sense of responsibility and mainte-nance (Trocchia and Beatty 2003), lower levels of con-cern regarding product obsolescence, and decreased lev-els of risk of living with a bad purchase decision for anextended amount of time (Berry and Maricle 1973).

Next, we examine the influence of two consumercharacteristics (i.e., materialism and psychological own-ership) on the payment-method/acquisition-satisfactionlink. The multi-faceted nature of materialism leads us topresent contrasting hypotheses. First, we suggest that oneof materialism’s central elements is control (Belk 1985;Burroughs et al. 2002; Kilbourne et al. 2005). We believethat the access pricing option requires the consumer togive up more control over the product than the ownershipoption does. When leasing, the consumer must return theitem after a pre-determined period of time and mustadhere to the terms of the leasing agreement. Under highlevels of materialism, we would expect that satisfactionfor consumers who use ownership payment terms will begreater than the satisfaction felt by consumers who useaccess payment terms. On the other hand, certain accessoptions may provide greater satisfaction to materialisticconsumers, as it allows them to utilize their consumptionbudget in a broader fashion. Materialists are often dissat-isfied due to their circumstances in life (Richins andDawson 1992). Renting or leasing may allow theconsumer’s budget to go further, may provide access to agreater number and variety of objects and may aid theconsumer to maintain consumption with a perceived peergroup regardless of financial constraints associated with


ownership methods (Durgee and O’Connor 1995). Ourcompeting hypothesis suggests that under high levels ofmaterialism, satisfaction for consumers who use accesspayment terms will be greater than the satisfaction felt byconsumers who use ownership payment terms.

We also integrate a second customer characteristic,psychological ownership, into our examination. Our goalis to delineate between the notion of property ownershipand the notion of psychological ownership. Psychologicalownership is “the state in which individuals feel as thoughthe target of ownership (material or immaterial in nature)or a piece of it is ‘theirs’ (i.e., a feeling of “being psycho-logically tied to an object”)” (Pierce et al. 2001, p. 299).We submit that one’s feelings of psychological ownershipwill influence the payment-method-acquisition-satisfac-tion relationship. First, psychological ownerships gener-ate a greater sense of responsibility and stewardshiptoward the item (Pierce et al. 2003) which may reduce the

freedom provided by the access payment agreement andinfluence acquisition satisfaction. Second, psychologicalownership may increase the level of integration into one’sself concept and the mere ownership effect (Beggan1992). Because access pricing terms explicitly limits theterm of possession, one may lose a part of the self at theend of lease term. Thus, acquisition satisfaction may beimpacted. Therefore we posit that for those exhibitinghigh levels of psychological ownership, the individualswho use ownership payment methods will have greateracquisition satisfaction than those who use access pay-ment methods.

Finally, we suggest the payment-method/acquisi-tion-satisfaction relationship has both theoretical andmanagerial implications. Future studies may examinevarying consumer and business markets, product classes,and other pricing strategies. References are availableupon request.

For further information contact:Michael T. Krush

University of Nebraska – LincolnCBA 320

P.O. Box 880492Lincoln, NE 68588

Phone: 402.742.3886Fax: 402.472.9777



MEDIATED EFFECTS OF SELF-MONITORING ON PERSONAL

SELLING: A META-ANALYSIS

George R. Franke, University of Alabama, TuscaloosaJeong-Eun Park, Ewha Womans University, Korea

SUMMARY

Meta-analyses of research on personal selling haveprovided valuable empirical generalizations, but the em-phasis has been on estimating relationships between twovariables at a time. Combining meta-analytic procedureswith structural equation modeling allows researchers totest multiple simultaneous pair wise relationships, reveal-ing the processes mediating the effect of one personalselling variable on another (Brown and Peterson 1993;Franke and Park 2006). This study focuses on the medi-ated effects of self-monitoring (SM), a personality traitthat involves the extent to which people monitor re-sponses to their expressive behavior and adjust their self-presentation to fit the social setting (Snyder 1974). Highself-monitors modify their public appearances to suitdifferent situations and “seem to perform particularly wellin occupations that call for flexibility and adaptiveness indealings with diverse constituencies,” such as sales(Gangestad and Snyder 2000, p. 533). A previous meta-analysis reports correlations between self-monitoring andother variables for workers in general (Day et al. 2002),but not for salespeople in particular.

A plausible mediator of the effects of SM is adaptiveselling behavior (ASB), the practice of altering salesbehaviors across interactions with prospects to suit thesituation (Spiro and Weitz 1990; Weitz 1981). Integratingmeta-analytic findings for SM with existing results forASB and other variables (Franke and Park 2006) suggestsa causal process for the influence of SM on sales forceperformance and satisfaction. Updating and expandingpast literature reviews also provides a broad basis forempirical generalizations about relationships among themodel’s variables.

The methods in this study are similar to those ofFranke and Park (2006). The structural analysis is basedon 36 meta-analytically derived correlations betweengender, experience, SM, ASB, customer orientation (CO),three measures of performance, and job satisfaction. Thecorrelations are taken from studies of native English-speaking sales forces, dated 1979 or later. Correlationsused are from 158 journal articles, 78 dissertations, and 11conference proceedings and working papers, reportingresults from 224 different samples containing a total of48,728 salespeople.

The results show that the direct and total effects ofSM on ASB are significant and substantial. SM has anindirect effect on CO mediated by ASB, but no additionaldirect effect. SM has a positive direct effect on self-ratedperformance, but only indirect effects on manager-ratedand objective performance. SM has a strong negativedirect effect on job satisfaction. This path is partiallyoffset by positive effects mediated through ASB and othervariables, though contrary to Day et al. (2002) the totaleffect is significantly negative. SM tendencies are signifi-cantly higher for salesmen than saleswomen. Sellingexperience is unrelated to SM. The other structural rela-tionships are largely but not completely consistent withthe results of Franke and Park’s (2006) preferred model.Where there are differences, many of the new results arearguably more plausible. For example, Franke and Park(2006) find ASB influences CO but not the reverse,whereas the current study reveals significant effects inboth directions. Salespeople who are oriented to solvingcustomers’ problems and stimulating long-term satisfac-tion are likely to treat each customer as an individualrather than minimizing adaptation across customers. There-fore, an effect of CO on ASB, not just ASB on CO, isreasonable. References are available upon request.

For further information contact:George R. Franke

Department of Management & MarketingUniversity of Alabama

105 Alston HallTuscaloosa, AL 35487–0225

Phone: 205.348.9435Fax: 205.348.6695



THE MOTIVATION-SPILLOVER-PRINCIPLE: HOW LEADERS

MOTIVATE SALESPEOPLE TO ADOPT SALES TECHNOLOGY

Florian Kraus, University of Marburg, GermanyTino Kessler-Thönes, University of Marburg, Germany

Jan Wieseke, University of Mannheim, Germany

SUMMARY

Motivation of salespeople is important because theywork at the boundaries of organizations where oversightand supervision is difficult. Since “customer contact em-ployees are the first and only representatives of a servicefirm” (Hartline, Maxham, and McKee 2000, p. 35) salesorganizations are constantly faced with the challenge offinding ways to improve salesforce performance. Previ-ous studies especially underline the importance of salesmanagers’ behavior for the sales employees’ behavior(e.g., Wotruba and Rochford 1995; Hartline and Ferrell1996). As Ilies, Judge, and Wagner (2006, p. 1) put it:“leader behaviors result in follower heightened motiva-tion to attain designated outcome(s) which then leads toperformance.” Despite the plausibility of these argumentsempirical research investigating the processes how salesmanagers influence the motivation of subordinate em-ployees is conspicuous by its absence.

A second issue of importance in contemporary salesmanagement besides motivation spillover from leaders tofollowers are the antecedents of the usage of a salestechnology. According to Mathieu, Ahearne, and Taylor(2007) “information technology is transforming the orga-nizational landscape […]. Nowhere is this truer than insales environments, where technology tools are changingthe way in which salespeople operate. Consequently,researchers and practitioners alike are beginning to exam-ine closely the issues surrounding the implementation oftechnology in sales environments and its effectiveness”(p. 528).

Regarding the need for multilevel research in both thearea of motivation as well as sales technology adoption,the goal of this paper is to develop and test a multilevelframework that combines the transfer of motivation (fromsales managers and salespeople) and its effects on salestechnology usage in sales organizations. At the heart ofthis framework we propose a motivation spillover effect.That is, we argue that the different dimensions of motiva-tion transfer from leaders to followers. Specifically wepropose a spillover of valences, instrumentalities and

expectancies (concerning sales technology usage), thethree components of Vrooms’ (1964) widely acceptedmotivation theory.

We believe that during sales technology implementa-tion motivation of the sales force is crucial for success.Therefore, we see the sales technology context as an idealapplication field for the motivation-spillover-principle.

A key asset of our research is that we are able to testthe motivation spillover and its consequences on salestechnology adoption on the basis of a large-scale multi-level data set, which links data from each of these twolevels (sales managers and salespeople). Data sets werematched via code numbers. We utilize hierarchical linearmodeling to adequately account for the nested data struc-ture of this data set.

We found that there are several spillover effects ofsales managers’ motivation components on salespeople’smotivation. Specifically, aspects of sales managers’ in-strumentalities and expectancies affect their counterpartson salespeople-level positively. Moreover, a higher levelof salesforce motivation is positively related to an in-creased sales technology usage. In contrast to our initialreasoning, sales managers’ valence does not affect thevalences of their subordinates. A reason for this findingcould be that valences of sales manager and salespeopleare varying. In this respect, Churchill, Ford, and Walker(1979) argue that “personal characteristics do influencethe kinds of rewards salespeople find most motivating”(p. 48).

Our results underline the importance of utilizing salesmanagers as multipliers during an introduction process ofnew sales technologies in large-scale organizations. Mo-tivating salespeople is a major task of sales managers,leading to an ongoing need to understand how salespeopleare likely to respond to organizational directives. Againstthis background, our findings provide an insight in thisblack box and offer guidance for managers on how totrigger the usage of new sales technologies.


For further information contact:Florian Kraus

Department of Business Administration, Marketing, and Retail ManagementUniversity of Marburg

Universitätstr. 2435037 Marburg

GermanyPhone: +49.6421.2823764

Fax: +49.6421.2826598E-Mail: [email protected]


TOP MANAGEMENT INVOLVEMENT WITH KEY ACCOUNTS: THE

CONCEPT, ITS DIMENSIONS, AND STRATEGIC OUTCOMES

Rodrigo Guesalaga, Pontificia Universidad Catlica de Chile, Chile

SUMMARY

The business community recognizes that top execu-tives need to be committed to the management of thefirm’s most strategic customers. The president and chiefexecutive officer of the Strategic Account ManagementAssociation (SAMA) has observed that: “Top manage-ment sponsorship and involvement is the most criticalindicator of success” (Napolitano 1997, p. 5). The aca-demic literature also recognizes that top managementinvolvement is a critical factor in key account manage-ment (KAM) success. Millman and Wilson (1999) con-sider commitment from senior management as a precon-dition that facilitates the implementation of KAM pro-cesses. Furthermore, Homburg, Workman, and Jensen(2002) find that companies with a KAM approach charac-terized by high-top management involvement tend tohave higher profitability, while Workman, Homburg, andJensen (2003) show that top management involvement(TMI) in KAM is positively related to KAM effective-ness.

True, top management involvement can be crucial fordeveloping new business, setting long-term agendas, andsteering the firm through sticky situations. However, itcan also be detrimental to key account management (KAM)efforts when, for example, top executives make decisionsregarding a customer without having enough knowledgeabout the customer’s way of doing business, competitivesituation or environment (Harro and Blinde 2005), orwhen senior managers try to overtake the account duringnegotiations while diminishing the key account manag-ers’ role as a primary contact (Koerner 2001).

In spite of the potential value and risks of top manage-ment involvement, the literature offers little guidance.Specifically, it fails to illuminate what top managementinvolvement is – i.e., the breadth and nature of the con-struct and the specific ways in which senior managersshould be involved with key accounts. Secondly, there isa dearth of empirical evidence as to how top managementinvolvement affects key strategic outcomes such as per-formance in sales and profits, and the quality of therelationship with customers. Finally, there is no identifi-cation of the conditions under which top managementinvolvement might be more beneficial or detrimental tothe supplier company.

This research contributes in several ways to theunderstanding of this issue. First, it proposes aconceptualization of top management involvement (TMI)

in KAM. Fifteen in-depth interviews with executivesinvolved in key account management were conducted tounderstand why and how top executives participate in thebusiness and the relationship with strategic customers,and how top management involvement can affect theperformance with customers. Based on these interviewsand a review of the business press on-key account man-agement, the following four dimensions of top manage-ment involvement were identified:

• TMI in business interaction refers to the extent towhich top managers sell, negotiate, or participate inother business activities directly with customers.This interaction can be either strategic or tactical.Strategic business interaction occurs when strategicissues are discussed with the customer, e.g., thedevelopment of new business opportunities, or theformation of a strategic alliance. Tactical businessinteraction takes place when tactical issues are dis-cussed with the customer, such as marketing mixactions with the current product line, or problemswith the service provided, among others.

• TMI in social interaction is the extent to which thesupplier’s top managers have social (nonbusinessrelated) contact with the buyer’s personnel.

• TMI in decision-making is the extent to which topmanagers make decisions with respect to key ac-counts. Strategic decision-making covers issues suchas the expansion of the current business with a cus-tomer to new products/services or markets, or changesin the organizational structure to attend a customer,among other aspects. Tactical decision-making in-cludes marketing mix-mainly product, price, distri-bution, and promotion-decisions for current productsor services, payment conditions, the frequency ofcontact with a customer, etc.

• TMI in KAM alignment is the extent to which topmanagers have an active role in aligning the goals andprocesses of the different functional areas toward acustomer.

The first two dimensions reflect top managers’ directinvolvement with key customers (i.e., directly interactingwith them), while the third and fourth dimensions revealtop managers’ indirect involvement with key customers(i.e., without directly interacting with them, but partici-pating through internal processes).


Second, the author analyses how each of these di-mensions of TMI affects the following KAM outcomes:KAM performance refers to the supplier firm’s economicachievements with key customers in terms of dollar sales,profitability, and share of customers’ purchases (in thesame product- or service-category). Relationship qualityis defined as how well the whole relationship fulfillsexpectations; it is a higher-order construct that encom-passes satisfaction, trust, and commitment. The resultsshow that TMI in social interaction has a positive effect onboth KAM performance and relationship quality, whiletactical business interaction has a negative impact on bothoutcomes. It is also found that TMI in strategic decision-making has a positive effect on KAM performance, andTMI in aligning the goals of those involved in managingthe customer has a positive effect on relationship quality.

In addition to the main effects, two boundary condi-tions were examined. First, it is analyzed whether theexistence of a high team spirit among those involved inmanaging or serving a key account reduces the impor-tance of TMI in internal processes (decision-making andKAM alignment) or not, and whether the existence of ahigh competitive intensity in the supplier’s industry in-creases the relevance of top managers’ direct involvementwith key accounts (through business and social interac-tion). The results suggest that the existence of a high teamspirit among those who are involved with customersmakes top management involvement in internal processes(decision-making and alignment) less critical, and thatsuppliers participating in highly competitive environ-ments should have top executives actively interactingwith customers in business-related issues. To test thismodel, 261 key account managers (as key informants)were surveyed from several industries, from United Statesand Chile (roughly 50% of the sample from each country).The sample frame was built mainly from the following

sources: Institute for the Study of Business Markets(ISBM) company members, ANDA (Chilean Associationof National Advertisers), Strategic Account ManagementAssociation’s peer directory, and Emory UniversityAlumni database. The model was tested empirically usingseemingly unrelated regression analysis; this approachimproves the efficiency of the estimation by incorporatingthe possible correlation of the disturbance terms in bothregression equations (Kmenta 1997). The following con-trol variables were included in the analysis as predictors ofKAM performance and relationship quality: executivesponsorship (i.e., whether there was a formal assignmentof the top manager to the account or not), sales team size,age of the buyer-supplier relationship, and size of thesupplier company. This research has several managerialimplications for supplier firms. Based on the findings,senior executives are more likely to contribute to KAM bysocially interacting with customers, and definitely avoid-ing participation in business meetings with customers onmatters that are tactical in nature. However, supplier firmsmay benefit from having senior executives interact withcustomers on strategic business issues when the companyfaces a highly competitive environment. In addition, theresults suggest that top managers have an important inter-nal role in KAM in terms of supporting decision-makingon strategic issues and aligning the customer team, espe-cially when there is a lack of a strong team spirit amongthose involved in managing or serving the account.

Overall, both senior executives and key accountmanagers need to recognize that top management in-volvement in KAM is not always beneficial to the suppliercompany. The key, then, is to determine when top manag-ers should actively participate in KAM, and how. Thisresearch provides some valuable insights to help manag-ers in that task. References are available upon request.

For further information contact:Rodrigo Guesalaga

Pontificia Universidad Catlica de ChileAv. Vicuæa Mackenna 4860, Macul

Santiago, ChilePhone: 56.2.354.7982


[email protected]


EXPLOITATION, EXPLORATION, AND AMBIDEXTERITY FROM

A MARKETING PERSPECTIVE: THE IMPACT OF

BUSINESS STRATEGIES

Silke Mühlmeier, University of St.Gallen, SwitzerlandFelicitas M. Morhart, University of St.Gallen, Switzerland

Torsten Tomczak, University of St.Gallen, Switzerland

SUMMARY

Innovation has become the hallmark of today’s busi-ness. To stay innovative, firms need to refine and improvetheir existing competencies and/or develop new knowl-edge and skills. In this regard, the duality of explorationand exploitation (March 1991) has emerged as a valuableconcept in strategic management research (e.g., Danneels2002; Vera and Crossan 2004). However, though sometentative advances have been made to transfer the notionof exploration/exploitation into marketing strategy re-search (Atuahene-Gima 2005; Kyriakopoulos andMoorman 2004), this framework is far from being estab-lished in this discipline. We aimed to contribute to thisnascent research effort by examining the impact of differ-ent business strategies on exploitative, exploratory, andambidextrous learning processes, thereby answering arecent call for research on this issue by several scholars(e.g., Atuahene-Gima 2005; Smith and Tushman 2005).In doing so, we resort to the state-of-the-art paradigm ofbusiness strategies by Treacy and Wiersema (1995) (i.e.,operational excellence, customer intimacy, and productleadership), and build on the literature on the resource-based view of the firm, the knowledge-based view of thefirm and organizational learning to conduct our line ofargument.

Four types of organizational learning are examinedas dependent variables in this study: (1) exploration, (2)exploitation-extension, (3) exploitation-refinement, and(4) ambidexterity. We conceive of a business strategy asa unifying frame of reference that emphasizes the align-ment of a firm’s resources in terms of knowledge, skills/competences, and routines to a specific value proposition.As such, a business strategy provides norms for dealingwith organization-wide intelligence which in turn set thestage for the type of organizational learning processes thatset in. We assume that operationally excellent firms – withtheir value proposition lying in delivering products andservices at lowest total cost for the customer – foster aknowledge orientation in terms of reliance on currentintelligence, i.e., the articulation, dissemination, and mul-tiplication of existing knowledge. Hence, we hypothesizethat a business strategy of operational excellence has apositive impact on exploitation-refinement, but neither onexploitation-extension, nor on exploration, nor on ambi-

dexterity. In contrast, we assume that product leaders –with their value proposition lying in providing customerswith cutting-edge products and services – facilitate aknowledge orientation in terms of knowledge generation,be it through completion and combination of existingknowledge or through creation of completely new knowl-edge. Therefore, we hypothesize that a business strategyof product leadership has a positive impact on exploita-tion-refinement, exploitation-extension, exploration, andambidexterity. Further, we assume that customer intimatecompanies – with their value proposition lying in provid-ing patrons with custom-made solutions – also foster aknowledge orientation of combining the use and exten-sion of existing knowledge with the creation of newknowledge. Thus, we hypothesize that a business strategyof customer intimacy also has a positive impact on exploi-tation-refinement, exploitation-extension, exploration, andambidexterity. Finally, according to the logic of “nocompany can be good at everything” put forth by manyscholars (e.g., Kotler 1999; Porter 1985; Thornhill andWhite 2007; Treacy and Wiersema 1995), we hypothesizethat no combination of the three business strategies willlead to higher levels in any of the four learning types(exploitation-refinement, exploitation-extension, explo-ration, and ambidexterity) than each of the business strat-egies in its purity.

Data for this study were collected from 377 individu-als out of a commercial sampling list of 2500 CEOs andchief marketing executives in Swiss services and manu-facturing firms. Whenever possible, established measuresof the constructs were used and adapted to this study’scontext. Exploratory and confirmatory factor analyseswere used to ensure that study measures exhibited accept-able psychometric properties. Hierarchical moderatedregression analyses were used to test our hypotheses.Results show support for all hypotheses stated.

This study is the first to empirically examine businessstrategies in their function as antecedents to exploitativeand exploratory learning processes and to identify themost valuable approach in a marketing strategy context.Our findings suggest managers to either concentrate on astrategy of product leadership or customer intimacy inorder to make most of their organization-wide learningpotential. References are available upon request.


For further information contact:Silke Mühlmeier

Institute of Marketing and RetailingUniversity of St. Gallen

Dufour Street N° 40aSt. Gallen, CH–9000

SwitzerlandPhone: +41(0)71.224.2841

Fax: +41(0)71.224.2857E-Mail: [email protected]


THE ROLE OF STRATEGIC FLEXIBILITY ON KNOWLEDGE

EXPLOITATION AND EXPLORATION

Kevin Zhou, University of Hong Kong, Hong KongFang Wu, University of Texas at Dallas, Richardson

SUMMARY

Do the core capabilities of firms lead to their corerigidities? In today’s highly competitive market environ-ments, developing core capabilities represents a criticalmeans for firms to survive and achieve sustainable com-petitive advantage. However, in rapidly changing envi-ronments, core capabilities may become “core rigidities”in that firms tend to heavily exploit their core capabilitiesbut forego the exploration of new technologies and knowl-edge for future success.

To address this challenge, we develop a model toresolve the above conflict with an attempt to contribute tothe literature in three ways:

• First, we focus on two critical organizational capa-bilities for NPD, namely marketing and technologycapabilities, and examine how they affect knowledgeexploitation and exploration. In doing so, we directlyaddress whether core capabilities can lead to exploi-tation but drive out exploration.

• Second, we propose that strategic flexibility repre-sents one important factor that may help firms over-come the rigidity dilemma. Our study fills this re-search gap by examining the facilitating role ofstrategic flexibility in managing the relationship be-tween an organization’s core capabilities and itsknowledge exploitation/exploration.

• Third, we examine how knowledge exploitation andexploration differentially affect firm financial per-

formance. We propose that though both might en-hance firm performance, exploitation will have astronger performance impact in the short run, whichexplains why it tends to crowd out exploration.

The conceptual model is empirically tested using across-section of 192 high-tech firms in China. To elimi-nate common method bias, we obtained different infor-mation from different sources. Specifically, senior man-agers provided information about the strategic level, suchas capabilities and strategic flexibility; middle managersprovided data about the operational level, such as knowl-edge exploitation and exploration, as well as two controlvariables (i.e., market growth and technological turbu-lence). Information on firm profitability (i.e., return onassets [ROA]), firm age, size, ownership, industry, andlocation came from archival data provided by the researchfirm.

After measurement purification and confirmation,the overall conceptual model is tested using hierarchicalregression analysis. The results indicate that even thoughmarketing and technology capabilities can foster knowl-edge exploitation, marketing capability does not improveknowledge exploration and high levels of technologycapability actually dampen knowledge exploration. How-ever, strategic flexibility can solve this rigidity dilemmaby strengthening the positive effects of technology capa-bility on knowledge exploration. Moreover, knowledgeexploitation and exploration differentially affect firmfinancial performance. References are available uponrequest.

For Further Information Contact:Fang Wu

School of ManagementUniversity of Texas at Dallas

Richardson, TX 75083Phone: 972.883.4740



BEYOND THE DYAD: INITIATING INTERACTION WITH

NON-ADJACENT SUPPLY-CHAIN PARTNERS

Niklas Myhr, Chapman University, Orange

SUMMARY

As managers increasingly recognize that they cannotgo it alone they embrace the notion that companies onlyrepresent interdependent parts of larger business net-works (Håkansson and Snehota 1989). Within such net-works, companies strive to establish interactive partner-ships with adjacent supply-chain members, i.e., theirimmediate customers and suppliers, and research to dateoffers managers plentiful insights into the management ofthese types of dyadic relationships. Beyond dyadic rela-tionships, there has long been a recognition of the fact thatdyads both affect and are affected by interactions with thesurrounding environment of the dyad (Achrol, Reve et al.1983; Anderson, Håkansson et al. 1994). Indeed, there areresearchers that have begun the process of coming to gripswith a business network perspective that looks beyonddyadic interactions in supply chains and marketing chan-nels (Anderson, Håkansson et al. 1994; Antia and Frazier2001; Rindfleisch and Moorman 2001; Wathne and Heide2004; Wuyts, Stremersch et al. 2004).

There is still a shortage, however, of research andguidance with respect to if, when and how companiesshould engage in activities with the specific aim to forminteractive relationships with firms beyond their adjacentsupply-chain partners (Van den Bulte and Wuyts 2007).For example, are there situations when a company wouldbenefit from interacting with the suppliers of their suppli-ers or with the customers of their customers? This paperexamines the choice of whether or not a focal companyshould initiate direct interaction with non-adjacent sup-ply-chain partners. Specifically, the objectives of thispaper are (1) to identify theoretical arguments supportingeach alternative and (2) to discuss potential criteria thatcan help managers in choosing the best alternative giventheir particular circumstances.

Interaction in Business Networks

Interaction is the degree to which two parties engagein a two-way communication process and interaction inbusiness networks has been analyzed through the use ofdifferent theoretical foundations. A common theme ofthese works is that dyads such as specific buyer-supplierrelationships cannot be analyzed without also taking thesurrounding business network into account. In short, thedyadic perspective is increasingly being characterized asmyopic and potentially misleading.

While fully embracing the social network perspec-tive may seem a daunting challenge to researchers fromboth a theory-building and data collection standpoint,significant insights may be gained by merely extendingthe dyadic perspective to small networks like triads, i.e.,social networks involving three actors (Van den Bulte andWuyts 2007). Therefore, for the purposes of this paper, wewill restrict our attention to triads involving a focal com-pany, an adjacent supply-chain partner of the focal com-pany, and a non-adjacent supply-chain partner that cur-rently is involved in a direct exchange relationship with anadjacent supply-chain partner of the focal company. Thatis, we are looking into the prospect of a focal companyinitiating interaction with a company with “two degrees ofseparation.”

In extending the dyadic interaction perspective to atriadic one, we also build upon and extend trust-basedtheoretical arguments commonly used in dyadic research.We propose that trust plays a critical role in determiningthe best approach of the focal company to interact withnon-adjacent supply-chain partners. The higher the de-gree of trust in the adjacent supply-chain partner both interms of competence and goodwill trust, the more likely itis that the focal company can benefit from an indirectinteraction approach with non-adjacent supply-chain part-ners. The dilemma is that too much trust in adjacentsupply-chain partners also makes the focal company morevulnerable to exploitation. Some companies prefer directinteraction with non-adjacent supply-chain partners as away for them to be better able to monitor and controladjacent supply-chain partners and not become too de-pendent upon trust in the willingness of the adjacentsupply-chain partner to be on their side.

Thus, it seems as if trust alone is not a sufficientdeterminant of interaction mode and managers wouldbenefit from more precise guidance as to how they strate-gically can manage interactions in their business net-works. Using characteristics of social networks (networkdensity, betweenness centrality, and tie strength), wesuggest conditions under which the potential for exploita-tion is more apparent. Consequently, such conditions canalso influence the relative effectiveness of the indirectinteraction approach with non-adjacent supply-chain part-ners. References are available upon request.


For further information contact:Niklas Myhr

Argyros School of Business and EconomicsChapman UniversityOne University Drive

Orange, CA 92866Phone: 714.532.6092



NEW PRODUCT PRE-ANNOUNCEMENTS: INCUMBENT RESPONSES

TO COMPETITIVE SIGNALS

Mayoor Mohan, Oklahoma State University, StillwaterKevin E. Voss, Oklahoma State University, Stillwater

SUMMARY

The globalizing world is teeming with firms all hop-ing to make it big with their next new product launch.Everyday bears witness to the launch of countless newproducts and services. One phenomenon is the tendencyof firms to announce new products in advance of theiractual initial date of sale. Accordingly, investigation ofnew product pre-announcements (NPPAs) is an importantarea of research.

Wind and Mahajan (1987) discuss pre-announce-ments as part of broader pre-launch activities that a firmconducts to increase marketing hype for a new product.We adopt the NPPA definition provided by Eliashbergand Robertson (1988), “a formal, deliberate communica-tion before a firm actually undertakes a particular market-ing action such as a price change, a new advertisingcampaign, or a product line change.”

NPPAs are market signals. How these signals areinterpreted is largely dependent on contingent factors,based on which their might be a competitive response.Hultink and Langerak (2002) define competitive responsesas the set of decisions made by a firm in response to anobserved new product launch. Empirical research sug-gests that new product launch decisions usually invokecompetitive responses that vary in terms of strength andspeed. Note that competitive responses can be multi-dimensional in nature (Kuester, Homburg, and Robertson1999). We look at two dimensions in particular: (1) speedof response and (2) aggressiveness of response.

Speed of response is the time it takes a competitor toreact to a particular competitive signal sent by anotherfirm. Speed of response to an NPPA appears to be deter-mined by a number of factors, which include theinnovativeness of the product (Gatignon, Robertson, andFein 1997) and perceived hostility (Robertson, Eliashberg,and Rymon 1995). This paper focuses on perceived hos-tility because it is the one factor that is a function of theNPPA itself.

A competitive response may be aggressive to varyingdegrees. An aggressive response is one that tries to elicita sense of threat within the original signal sender (cf., Heiland Robertson 1991; Heil and Walters 1993). When anaction is perceived as hostile, some kind of competitive

response from incumbent firms is likely (Robertson,Eliashberg, and Rymon 1995). Research has found per-ceived hostility and signal credibility help determine theaggressiveness of the response.

We argue that the presence or absence of an ally willhelp determine the perceived credibility of the NPPAsignal. We also argue that the stated lead time (short orlong) to new product introduction will affect the per-ceived hostility of the NPPA signal. This results in aninteraction effect such that the likelihood of response willbe higher when both an ally and a short lead are present inthe NPPA.

Signal credibility determines whether or not an in-cumbent in the marketplace is likely to react to an NPPAsignal (Chen and Miller 1994). The credibility of a signalis two dimensional in the sense that it depends on thesender’s reputation, and the irreversibility of the signal. Itcan be hypothesized that credible signals will inducefaster responses. Situations could arise in which somefirms may have difficulty sending a credible signal throughan NPPA. One possible solution to this conundrum isusing a reputable ally. Since a credible NPPA signalshould affect both the strength and speed of response bycompetitors, we offer the following propositions:

P1A

– The presence of a well-known, reputable ally inan NPPA will lead to a more aggressive response bythe signal receiver.

P1B

– The presence of a well-known, reputable ally inan NPPA will lead to a speedier response by thesignal receiver.

Similarly, signal hostility has been suggested to havean influence on speed and aggressiveness of response.Information about lead-time to product introduction orpatent status information can influence the perceivedhostility within an NPPA. If the NPPA indicates a shortlead-time to introduction coupled with information that apatent is pending, this should be perceived as more hostileby competitors than such information is not included. Weoffer the following:

P2A

– An NPPA with a shorter lead-time and a patentpending status will be associated with a more aggres-sive response by the signal receiver.


P2B

– An NPPA with a shorter lead-time and a patentpending status will be associated with an increasedspeed of response by the signal receiver.

A signal must be credible before it is perceived ashostile. We propose that when perceived signal credibilityis low, the patent status of the new product or the lead-timementioned in the NPPA is irrelevant. Essentially, thesignal is not believable. Based on this argument, weshould see a low level of response likelihood on bothspeed of response and strength of response when a signalis judged to be non-credible. We suggest the followinginteractions:

P3A

– An NPPA with shorter lead-time and patentpending status will lead to a more aggressive re-sponse when an ally is present, but will have no effectotherwise.

P3B

– An NPPA with shorter lead-time and patentpending status will lead to an increased speed ofresponse when an ally is present, but will have noeffect otherwise.

References are available upon request.

For further information contact:Mayoor Mohan

Department of MarketingWilliam S. Spears School of Business

Oklahoma State UniversityStillwater, OK 74078Phone: 405.762.1429



INFLUENCE OF INDIVIDUAL’S PERSONALITY DIMENSIONS,

EXPOSURE TO ADVERTISING AND PERSONAL RELEVANCE

ON PERCEPTION OF BRAND PERSONALITY

Srividya Raghavan, ICFAI Institute for Management Teachers (IIMT), IndiaM.S. Balaji, ICFAI Institute for Management Teachers (IIMT), India

ABSTRACT

The objective of this study was to examine the rela-tionship between individual’s personality (measured bythe big-five inventory – BFI) and perception of brandpersonality (measured using the brand personality scale –BPS). An experimental study was designed after a reviewof conceptual and empirical studies in the area of brandpersonality and human personality. The results of thestudy indicate significant relationship between individual’spersonality dimensions and perception of brand personal-ity. In addition, the results indicate that advertising mightbe a stronger factor that influences brand personalityperceptions. The two results take together indicates thatwhile individual differences can affect brand personalityperceptions, good advertising may be able to overrideindividual personality differences.

Keywords: Brand personality, personality, personalrelevance, marketing communication, and advertising.

INTRODUCTION

“Beauty lies in the eyes of the beholder” states an oldadage. Does the perception of Brand personality, then,also depend on the perceiver’s personality?

Intuitively, one might theorize that individual per-sonality differences affect perception of other personali-ties, including those of brands. While differences ingeneral personality dimensions have indicated differ-ences in job or task performances (Barrick and Mount1991), educational achievement (Smith 1967) the affectof differences in individual personality on Brand Percep-tions has been largely conjectural and contextual. Thisstudy proposes to address the lack of scientific evidencein this quarter.

LITERATURE REVIEW

Individual Personality has been extensively studiedand extracted by psychologists, beginning as early in thehistory of psychology, as Sigmund Freud who determinedthe three part id-ego-superego personality, progressing tothe trait theorists such as Gordon Allport, Cattel, andMcAdams, who determined dominant and consistent traitsthat describe an individual’s personality (Kassarjian 1971).

Among a plethora of personality trait theories, a popularand more general demarcation of personalities is madepossible by the “Big Five” trait theory which includes thetraits of Extroversion, Agreeableness, Conscientiousness,Neuroticism, (Emotional Stability) and Openness(Goldberg 1971). In the context of marketing, “Personal-ity” has been one of the more inclusive constructs bor-rowed from psychologists. A number of consumer relatedactivities such as purchase behavior, media choice, seg-mentation, choice, perceived risk, opinion leadership, andattitude change have been related to Personality (Kassarjian1971). While a number of studies emphasize the linkbetween personality and product or brand choice andstress on the actual decision making process, very fewstudies dwell in the realm how individual personalitiesprocess brand personality information and create brandperceptions. Even as far back as the 1940’s it was sug-gested that personalities might affect consumer decisionmaking and it was suggested that various personalitieswould be predisposed to specific styles, colors, brands, orstores (Brody and Cunningham 1968). It has even beenshown that personalities drive information processingwhich can in turn influence their perception of the brand.In an attempt to show that variances in consumer’s per-sonality affect their brand choice, Fry (1971) studiedpersonality disposition and brand choice in the context ofcigarette brand choice. Other studies have demonstratedthe relationship among a few personality traits, such asextroversion and openness and hedonic value, brandaffect, and loyalty. Matzler, Bidmon, and Grabner-Kräuter(2006), provide evidence that individual personality dif-ferences can account for differences in the values expec-tation of the consumer and can also impact formation ofbrand affect and loyalty in the individual consumer.

Academic interest in the marketing domain, how-ever, appears to have moved away from individual per-sonality to defining and measuring personality of prod-ucts and brands themselves (Aaker 1997). This is possiblya reason for the limited availability of empirical evidenceconnecting generic Individual personality dimensionsand brand personality dimensions, which are of morerecent origin. Literature shows the importance of specificpersonality traits such as need for cognition (Haugtvedt,Petty, and Cacioppo 1992), self confidence (Fry 1971);conscious and unconscious needs (Brody and Cunningham1968) etc., on brand choice or preference. These studies


have not focused on more general personality dispositionsas described in the trait theory of personality such as the“Big Five” personality dimensions, the 16PF or the MMPI.The studies have dwelt, instead on contextual personalityvariables which have direct implication for the researcher’sinterest.

Brand personality has been widely accepted as a partof the branding theory and is often considered as theclosest variable to the consumer during the purchasedecision making process (Rajagopal 2006). A positivebrand personality is thought to increase the levels of trust,preference, usage, loyalty, encourage active processing ofthe information and provide the basis of differentiation(Aaker 1996; Fournier 1998). In addition, a strong, posi-tive and favorable brand personality leads to favorable,unique, strong and congruent brand associations thusenhancing brand equity. Aaker (1997) compared thebrand personality with human personality and identifieda stable set of five personality dimensions that are thoughtto underlie the construct. The Aaker’s brand personalityscale (BPS) measures the extent to which the brandpossesses any of these personality traits. The brand per-sonality dimensions include: Sincerity; Excitement; Com-petence; Sophistication; and Ruggedness.

Besides, previous research suggests that the greaterthe congruity between the brand personality and thehuman characteristics that describes an individual’s ac-tual or ideal self, the greater the preference for the brand(Kim, Lee, and Ulgado 2005). Human Personality andBrand Personality can be thought of as counterparts in thehuman and brand world. Since markets involve interac-tion of human and brands, it may be conceivable that thetwo personalities influence each other. While variousstudies have noted the possibility of conceiving brands aspersonalities and the influence of brand personality onconsumer’s responses toward the brand (Freling andForbes 2005), few studies have sought to understand howhuman personality variations may affect perception of thebrand (Matzler, Bidmon, and Grabner-Krauter 2006).While Brand personality itself is derived as a parallel tohuman personality, the perception of the brand personal-ity as communicated through marketing channels may beinfluenced by human personality variations in the con-sumers themselves. As pointed out by Freling and Forbes(2005), “creation of personality is a ‘joint venture’ be-tween brand’s management and the consumers.”

The perceptions of brand personality are shaped andinfluenced by both direct (user imagery; McCracken1989) and indirect (product-related attributes, productcategory associations, brand name, symbol, advertisingstyle, pricing, and distribution channel; Batra, Lehmann,and Singh 1993) contact that consumer has with the brand.Brand communication helps consumers to conjure upbrand personality in the consumers’ mind that facilitates

better recall and positioning of the brand. Consequently,this affects the equity of the brand. The advertisementcannot by itself transfer the brand personality until theviewer makes the connection between the personalities ofthe celebrity/person in the advertisement and/or the per-sonality derived from the meaning of the advertised prod-uct. Thus consumers conceptualize and humanize brandsto give them human-like personalities.

It has been demonstrated that individuals are moresusceptible and receptive to messages that match theirpersonalities (Moon 2002) and hence the communicationinteracting with the personality type could also causedifferences in perception of Brand personality.

An individual must process information in somemanner in order to form a perception or attitude about theobject in contention. Celsi and Olson (1988), point outthat “felt involvement is a motivational state that affectsthe extent and focus of consumers’ attention and compre-hension processes, and thus the specific meanings that areproduced.” Involvement with a product category wouldoccur when the product category is of some relevance tothe recipient of the information. Hence, while consideringhow an individual processes information, involvementdue to personal relevance has been considered in severalstudies and it has been shown that involvement can playa moderating role in processing information (Celsi andOlson 1988). In experimental conditions, involvement isoften manipulated by inducing personal relevance tomotivate involvement which can influence processing.

This study, therefore, attempts to link individualpersonality dimensions, specifically the “Big Five” to theperception of brand personality dimensions. In addition,the study also examines the role of advertising and per-sonal relevance of the target category in moderating theperception of brand personality by people with differ-ences in various personality dimensions. This study pro-vides evidence that individual consumer’s personalitydifferences among the various Big Five dimensions canaffect their perception of brand personality with or with-out the presentation of a common marketing communica-tion.

EXPERIMENT

Research Problem

This research is based on the premise that individualpersonality dimensions may influence brand personalityperception since individual’s process information differ-ently. Similarly, personal relevance and involvement be-come motivational factors that influence processing ofbrand related information and exposure to advertisingprovides individuals with an opportunity to process infor-mation. This study was, therefore, designed to study the


Individual

Perception ofBrand

PersonalityDimensions

IndividualPersonalityDimensions

PersonalRelevance &

MarketingCommunication

effects of individual personality dimensions, personalrelevance of the issue under contention and exposure tomarketing communication on the perception of brandpersonality. Furthermore, considering that both brandpersonality and individual personality are multi-dimen-sional, the effect of each individual personality dimensionon the perception of each dimension of brand personalitywas sought, since there is evidence from literature thatcertain brand personality traits are preferred by certainindividual personalities due to perception of the self.

It is expected that individual personality dimensionswould have main effects in perception of some of thedimensions of the brand personality. The exposure toadvertising is expected to have significant main effectssince it directly influences motivation as well as nature ofprocessing. Personal relevance may not have significantmain effects in perception of brand personality, but mod-erating effects are expected in the form of interactioneffects with both personality dimension as well as expo-sure to advertising. A significant three-way interaction isalso expected. However, this may not be consistent for allindividual personality dimensions.

Hypothesis. Based on the aforesaid, the followinghypotheses evolve

H1: Individual’s personality dimensions cause variationsin perception of Brand personality dimensions.

H2: Exposure to advertising causes variations in percep-tion of brand’s personality dimensions.

H3: Individual’s with high personal relevance perceivethe brand’s personality differently from those whohave less personal relevance.

H4: Individual’s personality dimensions with higher per-sonal relevance perceive brand’s personalities differ-ently from those with lower personal relevance

H5: Individual’s with varied personalities exposed toadvertising perceive brand’s personality differentlyfrom those who are not exposed to advertising.

H6: Individuals with varied personal relevance who areexposed to advertising perceive brand’s personalitydifferently from those who are not exposed to adver-tising.

H7: Individual personalities with varied personal rel-evance levels who are exposed to advertising, per-ceive brand’s personality differently from those whoare not exposed to advertising.

The following sections explain the method, proce-dure and analysis of the study designed to answer theaforementioned questions.

FIGURE 1



Participants and Procedure. Using the standardizedinstrument for gauging the “Big Five” dimensions ofhuman personality (BFI), the data was collected from 270students enrolled into the MBA program. Un-cued recallwas elicited for corporate advertisements and the respon-dents were asked to recollect information about the adver-tisements recalled (information regarding name of com-pany, punch line, and concept). This initial data wascollected in order to identify groups for the five individualpersonality dimensions as well as to effect the personalrelevance manipulation.

A second phase of data collection on the variousbrand personality dimensions was done after two weeksof the initial data collection from two groups of respon-dents, half of who were shown print corporate advertise-ment of two Information Technology companies and theother half who were not shown any advertisement. It wasensured that half of those in both these groups were eitherinterested in IT companies for future placements andcareer choices and fell on one end of each of the person-ality dimension.

Method and Measures. Following the initial datacollection, the respondents were sorted in groups based ontheir personal interest in information technology industry.This was determined based on their personal interest ofchoosing IT company for future career prospects as wellas immediate placements. As final year MBA studentswho were looking forward to facing selection boardsduring campus placements, the group of respondents, hadcorporations, at the top of their minds and were consider-ing these corporations as brands that have immediatepersonal relevance. With II companies paying the largestpay packets and booming as an industry in India, it was feltthat the focus on IT companies for the study makesintuitive sense.

After two weeks, responses on the brand personalityscale (BPS: Aaker 1997) for both the information technol-ogy (IT) companies [Apple Inc. and Hindustan Comput-ers Limited (HCL)] was collected from one group ofrespondents. The second group of students was exposedto print advertisement of both the information technologycompanies and responses on the brand personality scalewere collected for both the companies, one after the other.The groups had an equal mix of those who had showninterest in the information technology industry and thosewho did not. It was then ensured, respondent wise, that thetwo groups had equal number of members who fell oneach end of each of the personality dimensions. Thus, a2 × 2× 2 factorial design was enabled.

In addition to the 44 items that measures the person-ality dimensions, the data collection instrument also in-cluded questions that captured the interest of the respon-

dents in the information technology industry. The averagescores of the set of items that captured each of thepersonality dimensions: extroversion, agreeableness, con-scientiousness, openness and neuroticism, were used togroup respondents. Those who had an average score ofless than a cut of (see Table 2 for averages and cut offs) onthe 5-point multi-item scale (agree strongly – disagreestrongly: items such as is talkative, is reserved, is full ofenergy etc.) were assigned to the “low on the dimension”condition and those who got a score of more than the cut-off point were assigned to “high on the dimension”condition.

Personal Interest Manipulation. Considering thatthe sample used was a set of final semester MBA studentswho had indicated their pre-placement interests, the ex-perimenters chose to use this information to manipulatepersonal interest conditions. The conjecture was that thestudents would have prior brand images of specific targetcompanies in the industries of interest. Given that infor-mation technology is a popular industry due to its promi-nence in the Indian context, its high remuneration struc-ture and high level of awareness of the industry, theinformation technology industry and two of the informa-tion technology companies were chosen for the study.Those respondents who ranked the IT industry as theirfirst, second or third choice as their target industry forfuture career prospects, were assigned to the high per-sonal interest condition and those that gave it a rank offour or more were assigned to the low personal interestcondition.

Advertisement Exposure. The pretest conducted dur-ing the initial contact contained questions that probed theawareness of corporate advertising of the respondents.Two companies that did not feature in the list of compa-nies for which students had seen corporate advertisingwere chosen as target companies whose brand personalityperception was captured on the brand personality ques-tionnaire. The companies chosen were Apple Inc. andHindustan Computers limited, both of which were com-panies that had 100 percent awareness among the respon-dents. The two companies were also selected since bothwere multinational but had American and Indian originrespectively. The respondents from the pretest were di-vided into two groups with almost equal composition ofthose in the two extroversion conditions and the twopersonal interest conditions. One group was then asked torespond to the brand personality questionnaire when onlythe name of the companies was given and the secondgroup was asked to respond to the same brand personalityquestionnaire after exposure to a print advertisement forthe company that they had not seen before. Thereby, thetwo conditions for exposure to advertising were manipu-lated.


Analysis

The dependent variables used were the five brandpersonality dimensions. The average scores from themulti-item 5-point scale for each dimension were used todenote the perception of brand personality along thosedimensions (Sincerity, Excitement, Competence, Sophis-tication, and Ruggedness). The independent variableswere each of the five dimensions of individual personal-ity, and the personal interest and exposure to advertising.Five multivariate analyses of covariance were done, foreach dimension of individual personality groupings usingSPSS 13.0 to determine the impact of the independentvariables on the personality dimensions. In order to elimi-nate chances of confounding results caused by gendereffects, this variable was controlled by using it as acovariate. The analysis for the two companies was doneindependently and the results are reported as such.

Results

Two of the manipulations, individual’s personalitydimension and personal interest were pre-determined.Equal number of those with personal relevance and with-out, belonging to higher or lower of each of the individualpersonality dimensions were assigned to “exposed to ad”and “not exposed to ad” conditions. Each of the conditionshad equal number of participants.

Manipulation check for personality dimensions wasagain done by running a one-way ANOVA for the groupsunder each dimension. Table 2 presents the results whichindicate that the groups were indeed significantly differ-ent from each other on each of the dimensions.

For the exposure to advertising manipulation, thepretest had provided information from recall questionsthat the respondents did not remember the corporateadvertisements for the two companies chosen in thisstudy. During the experimentation when groups wereeither exposed or not exposed to advertising, specificquestions related to their knowledge and memory of theads presented elicited answers that indicated that 100percent of the participants were aware of the companies,but had not been exposed to the advertising used in theexperiment. In the “not exposed to advertising” group, therespondents were shown the ads after brand personalityscores were elicited in order to be sure that they had notbeen exposed to the advertisements used in the study.

Analysis of Data. MANCOVA was used to analyzethe data from each of the 2×2×2 (individual personalitydimension × exposure to advertising × personal relevance)designs. SPSS 13.0 was used for the analysis. Five designswere obtained based on the five individual personalitydimensions. The results are presented on the basis of theindividual personality dimensions as the other indepen-

dent variables remain the same in each of the designs. Theresults of the analysis are presented for one of the design.

The results of the design using the groups based onextroversion scores, yielded results that indicated thatextroversion and advertising exposure have an overallsignificant main effect on the dependent variables forApple Computers. The gender covariate also shows sig-nificant main effects. There is a marginal significant maineffect of the advertising exposure and extroversion inter-action. Considering each dependent variable individu-ally, advertising exposure seems to have an effect on mostof the dimensions of brand personality excepting thedimension of “ruggedness.” Whether a person scores lowon the extroverted dimension or not seems to indicate theperception of the “sophistication” dimension of brandpersonality, while females and males perceive the “sincer-ity” dimension differently. Surprisingly, advertising ex-posure and extroversion show an interaction effect for theperception of the “ruggedness” personality dimension(Table 3a).

A similar result with very minor variances is noted inthe perception of HCL’s brand personality. The resultsindicate that advertising exposure, extroversion and gen-der have significant main effects, while a three waymarginal interaction effect is noted for the three factors.

As seen in the case of Apple Computers, advertisingexposure effects perception of four of the brand dimen-sions. Extroversion seems to indicate that it affects onlytwo dimensions, but the overall main effect is highlysignificant. Gender effects perception of sincerity as wellas sophistication. A three-way interaction is noted for theexcitement dimension at 5 percent significance level(Table 3b).

Similar analysis was done on other dimensions ofhuman personalities. The results indicate the following:

1. There are consistent and significant main effects foradvertising exposure.

2. Advertising exposure affects perception more thanone brand personality dimension

3. Each dimension of the individual’s personality hassignificant main effects of perception of brands per-sonality.

4. Each dimension of individual’s personality affectsdifferent dimensions of brand’s personality

5. The effect of individual’s personality dimensions onbrand personality dimensions is not consistent acrossbrands. This may be due to other factors outside the


TABLE 3(a)

Extroversion Design: Significant Main Effects and Means

APPLE Computers Inc.

Per Per rel* Ad Ad exp*

Dependent Extro- Personal Adv. Rel * ad exp* Extro*

Variable Version Relevance Exposure Gender Extro exp extro per rel

Multivariate

Test results P = .001 – P =.000 P =.005 – – P = .070 –

Between Subject Effects and Means

Sincerity – – P = .007 P = .000 – – – –(3.194;3.374)

Excitement – – P = .014 – – – – –(4.177; 4.346)

Competence – – P = .000 – – – – –(3.792; 4.237)

Sophistication P = .000 – P = .000 – – – – –(3.488; 3.811) (3.502; 3.797)

Ruggedness – – – – – – P = .008 –

TABLE 3(b)

Extroversion Design: Significant Main Effects and Means

HCL

Per Per rel* Ad Ad exp*

Dependent Extro- Personal Adv. Rel * ad exp* Extro*

Variable Version Relevance Exposure Gender Extro exp extro per rel

Multivariate

Test results P = .000 – P=.000 P =.007 – – – P =.085

Between Subject Effects and Means

Sincerity P = .000 – P = .007 P = .007 – – – –(3.265; 3.504) (3.577; 3.192)

Excitement – – P = .017 – – – – P = .036(3.042; 3.271)

Competence – – P = .000 – – – – –(3.364; 3.953)

Sophistication – – P = .001 P = .022 – – – –(2.353; 2.647)

Ruggedness P = .005 – – – – P= .087 – –(2.747; 3.005)


purview of the study such as prior knowledge andfamiliarity with the brands.

6. The existence of personal relevance of the categoryhas not shown consistent and significant effects onperception of brand’s personality.

7. The gender covariate appears to have a consistentmain effect on the perception of brand personality.However, it affects only one dimension, namelysincerity.

8. Interaction effects are minimal and hence inclusiveby way of results that the factors used in the studyconsistently interact in producing variations in brandpersonality perceptions.

8. While the effects of the factors considered in thestudy are more or less consistent across both brandsused in the study, some personality dimensions mayaffect some brands differently as evidenced in thecase of “Openness to experience” and “Neuroticism”designs.

9. It is quite apparent that interaction effects among thefactors are unpredictable. This indicates that theresults of the study cannot be used as evidence toprove that individuals with certain personality traits,process advertising according to whether the productcategory is relevant to them in perceiving brandpersonality dimensions.

DISCUSSION

Instead of focusing on the creation of brand person-ality and its impact on consumers, this study examines theimpact of human personality in the very conception andunderstanding of brand personality. In addition, this re-search takes into consideration the role of marketingcommunication (advertising) in shaping this perception.As rightly pointed out by Matzler, Bidmon, and Grabner-Krauter (2006) “a walk down any street in any townswiftly reveals that people are different. The days whenmarketers could treat them as being the same are longgone.” Their research concluded that human personalitydimensions such as extroversion and openness to experi-ence (human personality dimensions) influences theconsumer’s level of loyalty toward a brand. They furtheropined that their individual responses to the hedonisticproperties of the brand and propensity for affective re-sponses may cause these variations in loyalty. While thisstudy establishes the differences in brand perception bydifferent personalities, a link between brand personalitiesand human personalities was not divulged.

This study was designed to examine the influence ofthe five dimensions of human personality on the

consumer’s perception of brand personality as proposedby Aaker (1997). The study indicates significant effectsfor the factors – exposure to advertisement, and specificindividual personality dimensions – while, the results areinconclusive for the factor – personal relevance. Thisindicates that each of the individual’s personality dimen-sions does, in fact, affect the perception of brand’s person-ality. Moreover, each dimension significantly effects theperception of different set of brand’s dimensions. Onemay note that there are more significant effects for theHCL than for Apple Computers. This could be due to thefact that the Indian students who participated in the studyare less familiar with Apple computers than with HCL.This is however conjectural as the study does not indicateas such in the results. Further, it was found that identifica-tion of prior knowledge, attitude etc. about the brand wasout of the scope of the study and there was fear ofcontaminating the data by getting the participants to thinkabout these brands before the experiment was conducted.

With further research, the specific connections be-tween brands and individual personalities could be iden-tified and relationships between brands and consumerscould be built on the basis of consumer personality pro-files and brand personality characteristics that would bestinfluence them. Implications for marketers would be, forinstance, if they found that their typical target scored highon extroversion, then, depending on the prior knowledgeand attitude toward the brand, they could point out thelevel of sophistication, in a brand such as apple andSincerity and ruggedness for a brand like HCL. The studyalso indicates consistent and significant impact of adver-tising on the personality of brands. Although outside thescope of this study, it may be noted that the content of theadvertisement can play a moderating role in causingvariations in perception of the brand’s personality. How-ever, this study indicates that advertising, in fact, doesimpact perception of brands personalities and in tandemwith the information on the individual personality dimen-sion, marketing communication could be made moreeffective in delivering brand’s personality information tothe consumer.

Further, this study has been inconclusive and insig-nificant in most cases with respect to the personal rel-evance factor. This could mean that personal relevance,by itself may only affect the depth of processing ofinformation as suggested by literature (Petty and Cacioppo1983; Celsi and Olson 1988), but not affect the perceptionof brand’s personality by impacting the nature of process-ing. The lack of consistent and significant interactiveeffects suggests that the influence of individual personal-ity dimensions and advertising exposure could be inde-pendent and they may not moderate each other. Hence, themarketer must look at each of these factors and assess heirinfluence separately. For instance, a study of theindividual’s personality dimensions and the brands per-


sonality dimensions may be done to assess a positioningplank, whereas creating and disseminating advertisingmay be done to render this information to the consumer.Each of these would determine how the consumer per-ceives the brand’s personality, albeit not in tandem.

LIMITATIONS AND SCOPE FOR

FUTURE RESEARCH

The focus of this study was only to examine, whetheran individual’s unique personality, marked by differencesin the dimensions would affect the perception of anotherpersonality, in the case, that of brands. The study includedmarketing communications and personal relevance in theform of motivation since literature suggests that thesefactors might significantly impact how brands are per-ceived. Therefore, the study did not look into aspects ofattitude and prior knowledge.

These additional measures in the study could haveadded value in assessing the nature of the results. How-ever, the researchers kept these variables out of the studyin order to keep measurements and data collection instru-ments simple, as well as not contaminate the data byallowing the respondents to dwell on the nature of thebrands used in the study.

This limitation however, creates scope for futurework that could examine the reasons for the results ob-tained in the study. One such important question is whetherthe prior knowledge/attitude toward a brand would causevariations in perception of the brand’s personality byindividuals with differing personality profiles.

Another question that might arise from this study iswhether processing of information is a function ofindividual’s personality, advertising exposure and per-sonal relevance. If the depth and nature of processing areconsidered as dependent variables instead of brand per-sonality dimensions, one might be able to examine the roleof individual personality dimensions.

Yet another variation of the study is possible in a non-factorial design condition, where the independent vari-able used in the study could have been the various dimen-sions of individual personality with the five brand person-ality dimensions as the dependent variables. This mighthave provided direct information on the manner in whichthe various personality dimensions interact. However,given the nature of the design and scope of work envi-sioned for this study, this was not possible. Hence thisstudy focused on the factors including the advertisingexposure and personal relevance which provided addi-tional insights into how these factors interact.

REFERENCES

Aaker, Jenifer (1997), “Dimensions of Brand Personal-ity,” Journal of Marketing Research, 34 (2), 347–56

Barrick, Murray R. and Michael K. Mount (1991), “TheBig Five Personality Dimensions and Job Perfor-mance: A Meta-Analysis,” Personnel Psychology,44 (1), 1–26.

Batra, Rajiv, D. Lehmann, and D. Singh (1993), “TheBrand Personality Component of Brand Goodwill:Some Antecedents and Consequences,” in BrandEquity and Advertising: Advertising’s Role in Build-ing Strong Brands, Aaker and Biel, eds. Hillsdale:Lawrence Erlbaum, 83–95.

Brody, Robert P. and Scott M. Cunningham (1968),“Personality Variables and the Consumer DecisionProcess,” Journal of Marketing Research, 5 (1), 50–57

Celsi, Richard L. and Jerry C. Olson (1988), “The Role ofInvolvement in Attention and Comprehension Pro-cesses,” The Journal of Consumer Research, 15 (2),210–24

Freling, Traci H. and Lukas P. Forbes (2005), “An Exami-nation of Brand Personality Through Methodologi-cal Triangulation,” Journal of Brand Management,13 (2).

Fry, Joseph N. (1971), “Personality Variables and Ciga-rette Brand Choice,” Journal of Marketing Research,8 (3), 298–304

Goldberg, L.R. (1981), “Language and Individual Differ-ences: The Search for Universals in Personality Lexi-cons,” in Review of Personality and Social Psychol-ogy, L. Wheeler, ed. Beverly Hills, CA: Sage, 2, 141–65.

Haugtvedt, Curtis P, Richard E. Petty, and John T. Cacioppo(1992), “Need for Cognition and Advertising,” Jour-nal of Consumer Psychology, 1 (3), 239–60.

John, O.P. and S. Srivastava (1999), “The Big Five TraitTaxonomy: History, Measurement, and TheoreticalPerspectives,” in Handbook of Personality: Theoryand Research, 2nd ed., L.A. Pervin and O.P. John,eds. New York: Guilford, 102–38.

Kasarjian, Harrold H. (1971), “Personality and ConsumerBehavior: A Review,” Journal of Marketing Re-search, 8 (4), 409–18

Kim, Hae Ryong, Moonkyu Lee, and Francis M. Ulgado(2005), “Brand Personality, Self-Congruity and theConsumer-Brand Relationship,” Asia Pacific Ad-vances in Consumer Research, 6, 111–17

Matzler Kurt, Bidmon Sonja, and Grabner-Krauter Sonja(2006), “Individual Determinants of Brand Affect:The Role of the Personality Traits of Extroversion


and Openness to Experience,” Journal of Productand Brand Management, 15 (7), 427–34.

McCracken, Grant (1989), “Who Is the Celebrity En-dorser? Cultural Foundations of the EndorsementProcess,” Journal of Consumer Research, 16 (3),310–21.

Moon, Youngme (2002), “Personalization and Personal-

ity: Some Effects of Customizing Message StyleBased on Consumer Personality,” Journal of Con-sumer Psychology, 12 (4), 313–26.

Smith, G.M. (1967), “Usefulness of Peer Ratings ofPersonality in Educational Research,” Educationaland Psychological Measurement, 27, 967–84.

For further information contact:Srividya RaghavanResearch Scholar

ICFAI Institute for Management Teachers (IIMT)Hyderabad, India

M.S. BalajiResearch Scholar

ICFAI Institute for Management Teachers (IIMT)Hyderabad, India

Phone: 91.402.343.04.69.94E-Mail: [email protected]


ENHANCEMENT OF BRAND PERSONALITY: THE USE OF DIFFERENT

INFORMATION SOURCES FOR DIFFERENT DIMENSIONS

Natalia Maehle, Norwegian School of Economics and Business Administration, NorwayCele Otnes, University of Illinois at Urbana-Champaign

Magne Supphellen, Norwegian School of Economics and Business Administration, Norway

SUMMARY

A brand, like a person, may be characterized as“modern,” “old-fashioned,” “lively,” or “exotic.” Thesecharacterizational aspects of the brand fall under what iscalled the brand personality (Plummer 1985). Much re-search has emphasized the importance of brand personal-ity (Johar, Sengupta, and Aaker 2005). For example,brand personality can serve a symbolic or self-expressivefunction (e.g., Belk 1988; Johar and Sirgy 1991; Levy1959).

However, the primary focus of previous studies hasbeen either on understanding the effects of brand person-ality or on measurement issues related to the construct(e.g., Aaker 1997; Azoulay and Kapferer 2003; Goversand Schoormans 2005). Fewer studies explore the natureof brand personality and identify its sources. Neverthe-less, a number of authors (e.g., Batra, Lehmann, and Singh1993) express their concern regarding the lack of theresearch on this topic. It is important to understand howconsumers form their perceptions of brand personalityand to understand how different sources contribute tobrand personality.

Study 1

We assume many sources can contribute to formingthe personality of a brand. These include not only thetraditional ones like typical brand user or endorser(McCracken 1989) but also indirect sources like price,product-related attributes, advertising style, distributionchannel, and symbols used in all phases of brand commu-nication, sales promotions, and media advertising (Batra,Lehmann, and Singh 1993). Brand personality itself is acomplicated concept that encompasses five different di-mensions: Sincerity, Excitement, Competence, Sophisti-cation, and Ruggedness (Aaker 1997). These dimensionsdiffer in their nature. According to the “match-up” hy-pothesis (Kahle and Homer 1985), there should be amatch between the stimuli (or source of brand personalityin our case) and the desired message about the brand themarketer wants to deliver through these stimuli. Thisassumption means marketers cannot use the same sourcesto convey different messages about brand personality. Inorder to make the message about brand personality effec-tive, it is necessary to find the sources that match this

message. Thus, we hypothesize that different informationsources are relevant for forming different dimensions ofbrand personality.

A survey of ten brands was conducted to test thishypothesis. The results support our assumption. More-over, we found systematic differences regarding whichsources are relevant for building each dimension. Themost important sources of Competence and Sincerity arecompany-level sources such as company’s moral values,the CEO and company employees. In contrast, Sophisti-cation and Ruggedness seem formed to a large extent bybrand symbols such as endorsers, typical brand users,brand name, and brand logos. Excitement seems to beformed by a blend of company-level sources (employees),symbols (endorser), and advertising style.

Study 2

A related issue not previously explored in the litera-ture is the enhancement of brand personality. Brandimage/personality enhancement is a serious challenge tobrand managers (Keller 1998). When brand personalitystagnates or declines, the result could be lower brandpreference and reduced brand equity. However, there isstill lack of research on this phenomenon. A number ofstudies explore adapting brands to changing consumertaste (e.g., Van Rekom, Jacobs, and Verlegh 2006) but amore general approach based on understanding the di-verse nature of brand personality is lacking.

Thus, after identifying the sources of different brandpersonality dimensions, we sought to determine whetherthese sources could be effective for the purpose of brandpersonality enhancement. In this study we chose to focuson two personality dimensions that are different in nature:Competence and Sophistication. We hypothesize that thesources found relevant in Study 1 can be effectively usedfor brand personality enhancement. For example, com-pany-level sources (company’s employees, company’smanaging director, and product attributes) are effectivefor brand personality enhancement on the Competencedimension, and brand symbols (celebrity endorser, typi-cal brand user, and brand logo) are effective for brandpersonality enhancement on the Sophistication dimen-sion.


The experiment was designed to assess these as-sumptions. The results show that the sources that werefound relevant for particular brand personality dimensionin Study 1 are also, for the most parts, effective for brandpersonality enhancement in Study 2.

Discussion, Conclusions, and Further Research

We investigate two important issues related to brandpersonality. First, we identify the sources of brand person-ality. Second, we explore whether these sources can beeffective for the purpose of brand personality enhance-ment. The results of these two studies have both theoreti-cal and managerial implications. They contribute to un-derstanding the nature of brand personality and to theliterature on brand personality enhancement. The results

can also be used by managers for choosing strategies forbrand personality development.

There are a number of possibilities for further re-search. The nature of brand personality is not yet fullyexplored. We isolate which sources are responsible forforming different dimensions, but it is still necessary tounderstand why some sources are more relevant thanothers. It would also be interesting to compare the effectsof relevant and less relevant sources on brand personalityenhancement. To our mind, less relevant sources shouldbe less effective for the enhancement of brand personality.One more direction of extending the current research is totry to enhance the rest of brand personality dimensions(e.g., Sincerity, Excitement, and Ruggedness).

REFERENCES

Aaker, Jennifer (1997), “Dimensions of Brand Personal-ity,” Journal of Marketing Research, 34 (3), 347–56.

Azoulay, Audrey and Jean-Noël Kapferer (2003), “DoBrand Personality Scales Really Measure Brand Per-sonality?” Journal of Brand Management, 11(2),143–55.

Batra, Rajeev, Donald R. Lehmann, and Dipinder Singh(1993), “The Brand Personality Component of BrandGoodwill: Some Antecedents and Consequences” inBrand Equity and Advertising, David A. Aaker andAlexander Biel, eds. Hillsdale, NJ: Lawrence ErlbaumAssociates, 83–96.

Belk, Russell W. (1988), “Possessions and the ExtendedSelf,” Journal of Consumer Research, 15, 139–68.

Govers, P.C.M. and J.P.L. Schoormans (2005), “ProductPersonality and Its Influence on Consumer Prefer-ence,” Journal of Consumer Marketing, 22 (4), 189–97.

Johar, J.S. and M. Joseph Sirgy (1991), “Value-Expres-sive versus Utilitarian Advertising Appeals: Whenand Why to Use Which Appeal,” Journal of Advertis-ing, 20 (3), 23–33.

Johar, Gita V., Jaideep Sengupta, and Jennifer L. Aaker(2005), “Two Roads to Updating Brand PersonalityImpressions: Trait versus Evaluative Inferencing,”Journal of Marketing Research, 42 (4), 458–69.

Kahle, Lynn R. and Pamela M. Homer (1985), “PhysicalAttractiveness of the Celebrity Endorser: A SocialAdaptation Perspective,” Journal of Consumer Re-search, 11 (4), 954–61.

Keller, Kevin Lane (1998), Strategic Brand Manage-ment: Building, Measuring, and Managing BrandEquity. New Jersey: Prentice Hall.

Levy, Sidney J. (1959), “Symbols for Sales,” HarvardBusiness Review, 37 (4), 117–24.

McCracken, Grant (1989), “Who Is the Celebrity En-dorser? Cultural Foundations of the EndorsementProcess,” Journal of Consumer Research, 16 (3),310–21.

Plummer, Joseph T. (1985), “How Personality Makes aDifference,” Journal of Advertising Research, 24 (6),27–31.

Van Rekom, Johan, Gabriele Jacobs, and Peeter W.J.Verlegh (2006), “Measuring and Managing the Es-sence of a Brand Personality,” Marketing Letters, 17,181–92.

For further information contact:Natalia Maehle

Norwegian School of Economics and Business AdministrationBreiviksveien 40

5045 BergenNorway

Phone: +47.55959672Fax: +47.55959430



PROMOTION MATCHING: THE ROLE OF PROMOTION TYPE,

SELF-CONSTRUAL, AND BRAND CONNECTION ON

PURCHASE INTENTIONS

Karen Page Winterich, Texas A&M University, College StationVikas Mittal, Rice University, Houston

Vanitha Swaminathan, University of Pittsburgh, Pittsburgh

SUMMARY

With promotions taking from 25 to 50 percent ofcompanies’ marketing budgets for consumer productsand packaged goods (Ailawadi et al. 2006; Raghubir,Inman, and Grande 2004), consumers are showered withpromotions each day. Yet, the effectiveness of manypromotions in building sales is not clear. We examine theeffectiveness of special promotions based on the contextin which consumers evaluate promotions. Self-construal,defined as an individual’s sense of self in relation toothers, has been found to influence consumer responses toadvertisements and brand information (Aaker and Lee2001; Agrawal and Maheswaran 2005). We propose thatthe effect of special promotions may be moderated by self-construal.

Sales promotions may lower brand evaluations(Dodson, Tybout, and Sternthal 1978), but this finding isnot consistent (Davis, Inman, and McAlister 1992). Re-search indicates that promotions lead to increased sales,but these sales may be short-lived as they are largely aresult of brand switching and stockpiling (Gupta 1988).On the contrary, promotions may increase brand equity byincreasing brand knowledge (Palazón-Vidal and Delgado-Ballester 2005). One promotion that has increased sales isthat of employee pricing. After promoting automobilesusing the “Employee Pricing for Everyone” tagline, Gen-eral Motors’ sales increased 41 percent for the month ofJune (Munoz 2005). We term these types of promotions“inclusive” promotions. Under what conditions will suchinclusive promotions be more effective? Given the ten-dency of individuals characterized by an interdependentself-construal to focus on ingroups such as family andfriends (Markus and Kitayama 1991), inclusive promo-tions may be evaluated differentially based on one’sinterdependence. We argue that interdependent consum-ers will have higher purchase intentions for an inclu-sively-framed discount than that of those with a lowinterdependent self-construal while interdependence willhave no effect on purchase intentions for a regular dis-count.

Consumers may also be characterized by an indepen-dent construal of self. The independent self-construal ischaracterized by one’s focus on individual thoughts and

feelings (Markus and Kitayama 1991). Recommendingthat companies remember customers’ birthdays and offerthem incentives, Harrington (2006) states, “Successfulbusinesses take every opportunity to offer unique, per-sonalized products and services, and using the retailpromotions calendar can help you do just that.” Wepropose that an independent self-construal will enhancethe effect of exclusively-framed promotions (i.e., birth-day discount, unique customer) on purchase intentionswhile an interdependent self-construal will enhance theeffect of inclusively-framed promotions on purchase in-tentions.

In the first study (N = 247 students), participants wereexposed to either an inclusively-framed (employee) pro-motion or regular promotion for a young apparel retailer.They then indicated their purchase intentions as well astheir chronic interdependence on a commonly used scale.Results indicated that the interaction of promotion typeand interdependence is significant (F(1, 240) = 3.92;p < .05). Comparing cell means, purchase intentions forthose in the employee promotion condition were signifi-cantly greater for those with high interdependence thanfor those with low interdependence (M

High = 4.69 vs.

MLow

= 3.94; t = 2.52, p < .05). In contrast, for those in theregular promotion condition, purchase intentions werenot significantly different for those with high interdepen-dence and those with low interdependence (M

High = 4.17

vs. MLow

= 4.22; t = 0.20, ns).

In the second study, participants (N = 235 adultconsumer panelists) are exposed to one of four conditionsin a 2 (Promotion: Employee vs. Birthday) X 2 (Self-construal: Interdependent vs. Independent (manipulated))between-subjects design. Participants’ self-brand con-nection was also measured using nine items adapted fromEscalas and Bettman (2005) to examine the three-wayinteraction of promotion type, self-construal, and self-brand connection. A manipulation check was conductedto ensure that the promotions were viewed as inclusively-or exclusively-framed, as intended. Importantly, the analy-sis reveals that interaction of promotion type, self-construal,and self-brand connection is significant (F(1, 234) = 4.21;p < .05), controlling for pre-purchase intentions. Specifi-cally, for those in the employee promotion condition withlow self-brand connection, purchase intentions were sig-


nificantly greater for those in the interdependent primethan those in the independent prime (M

Inter = 5.34 vs.

MIndep

= 4.87; t = 1.99, p < .05). In contrast, for those in thebirthday promotion condition with low self-brand con-nection, purchase intentions were significantly greater forthose in the independent prime than those in the interde-pendent prime (M

Indep = 5.62 vs. M

Inter = 4.96; t = 2.77, p <

.05). For those with high self-brand connection, there wasno difference in purchase intentions, regardless of promo-tion type or self-construal (p’s > .40). These findingsindicate that when promotions are framed to match a

consumer’s self-construal, purchase intentions may in-crease. However, this effect is limited to those consumerswith lower self-brand connection. While marketing man-agers are constantly offering promotions to increase salesand market leadership, this research examines how spe-cial promotions (i.e., inclusively- and exclusively-framed)can have positive impacts on the brand via purchaseintentions when appropriately matched with consumers’self-construal. Importantly, these effects appear to belimited to those consumers with lower connections to thebrand.

REFERENCES

Aaker, Jennifer L. and Angela Y. Lee (2001), “‘I’ SeekPleasures and ‘We’ Avoid Pains: The Role of Self-Regulatory Goals in Information Processing and Per-suasion,” Journal of Consumer Research, 28 (June),33–49.

Agrawal, Nidhi and Durairaj Maheswaran (2005), “TheEffects of Self-Construal and Commitment on Per-suasion,”Journal of Consumer Research, 31 (March),841–49.

Ailawadi, Kusum L., Bari A. Harlam, Jacques Cesar, andDavid Trounce (2006), “Promotion Profitability for aRetailer: The Role of Promotion, Brand, Category,and Store Characteristics,” Journal of MarketingResearch, 43 (November), 518–35.

Davis, Scott, J. Jeffrey Inman, and Leigh McAlister (1992),“Promotion Has a Negative Effect on Brand Evalua-tions – Or Does It? Additional Disconfirming Evi-dence,” Journal of Marketing Research, 39 (Febru-ary), 143–48.

Dodson, Joe A., Alice M. Tybout, and Brian Sternthal(1978), “The Impact of Deals and Deal Retraction onBrand Switching,” Journal of Marketing Research,15 (February), 72–81.

Escalas, Jennifer and James Bettman (2005), “Self-construal, Reference Groups, and Brand Meaning,”Journal of Consumer Research, 32 (December), 378–89.

Gupta, Sunil (1988), “Impact of Sales Promotions onWhen, What, and How Much to Buy,” Journal ofMarketing Research, 35 (November), 342–55.

Harrington, Wendy (2006), “Making the Calendar Workfor Your Business,” Missouri Small Business Devel-opment Centers, (December 8).

Markus, Hazel and Shinobu Kitayama (1991), “Cultureand the Self: Implications for Cognition, Emotion,and Motivation,” Psychological Review, 98 (April),224–53.

Munoz, Olivia (2005), “GM to Extend its Employee-Pricing Discount,” ForbesAutos.com, (August 25).

Palazón-Vidal, Mariola and Elena Delgado-Ballester(2005), “Sales Promotions Effects on Consumer-Based Brand Equity,” International Journal of Mar-ket Research, 47 (2), 179–204.

Raghubir, Priya, J. Jeffrey Inman, and Hans Grande(2004), “The Three Faces of Consumer Promotions,”California Management Review, 46 (Summer), 23–42.

For further information contact:Karen Page WinterichMays Business SchoolTexas A&M University

220G Wehner, 4112 TAMUCollege Station, TX 77843-4112

Phone: 979.845.5452Fax: 979.862.2811


Vikas MittalJones Graduate School of Management

249 McNair HallRice University

Houston TX 77005Phone: 713.348.6234


Vanitha Swaminathan344 Mervis Hall

University of PittsburghPittsburgh, PA 15260Phone: 412.648.1579



BUILDING CUSTOMER’S TRUST IN THE SUPPLIER FIRM: THE

IMPACT AND LIMITS OF SALESPERSON’S CONTINUITY

Rodrigo Guesalaga, Pontificia Universidad Católica de Chile, Chile

SUMMARY

Do supplier firms benefit from having their relation-ship managers look after the same customers for anextensive time period? Previous research suggests that, tobuild long-term relationships with customers, supplierfirms must ensure continuity between the relationshipmanager and the buying company (Capon 2001; Weitzand Bradford 1999). Moreover, sales managers tend toavoid the reallocation of salespeople to different custom-ers because they risk losing business; salespeople see thisoption as stressful and unattractive when they are com-fortable with their existing customers; and customers mayperceive the assignment of a new sales representative asdisruptive, particularly when in-depth customer knowl-edge is needed for effectiveness (Zoltners and Lorimer2000). Bendapudi and Leone (2002) argue that customersmay have potential negative responses when the keycontact employee is unable to continue serving them,especially when he or she plays a critical role in determin-ing satisfaction with the supplier, and when the customerhas developed special bonds with the sales representative.

However, maintaining continuity of salespeople withtheir customers is not always possible, because new jobopportunities may arise for the relationship manager, orthe supplier company may need to realign sales territoriesto increase sales productivity or adopt a new sales strategy(Zoltners and Lorimer 2000). Furthermore, it is not clearwhether continuity of the salesperson with a customerreaches a point where its benefits start to wash out, or theconditions under which salesperson continuity becomesless effective as a mean to build long-term relationshipswith customers.

This gap in the literature motivates this research,which looks at how salesperson’s continuity with custom-ers can affect customers’ trust in the supplier firm. Spe-cifically, it is examined whether there is an inverted-Urelationship between salesperson’s continuity (measuredas the number of years that the salesperson has been incharge of the same customer) and customer’s trust in thesupplier firm (in terms of these two variables, and whetherthe linear effect between continuity and trust is moderatedby the size of the customer team, the dedication of uniqueresources to the customer (in terms of product specifica-tions, procedures, and marketing activities), and the lengthof the relationship between the supplier and the buyerfirms (i.e., relationship age).

To test this model, 261 key account managers (as keyinformants) were surveyed from several industries, fromUnited States and Chile (roughly 50% of the sample fromeach country). The sample frame was built mainly fromthe following sources: Institute for the Study of BusinessMarkets (ISBM) company members, ANDA (ChileanAssociation of National Advertisers), Strategic AccountManagement Association’s peer directory, and EmoryUniversity Alumni database. The model was tested em-pirically using regression analysis. The following controlvariables were included in the analysis as predictors ofcustomer’s trust: access to resources (i.e., the extent towhich the relationship manager receives support from thefollowing functions: Sales, Customer Service, Marketing,and Production), esprit de corps among those involved inserving or working with the customer, and top manage-ment involvement in decision making.

The results show that, although salesperson’s conti-nuity is positively associated with customer’s trust in thesupplier firm, there is a limit after which this effect washesout. After a few years (in this sample, after eight years),securing the continuity of salespeople with the samecustomers does not contribute significantly to buildingcustomer’s trust in the supplier company. Second, thefindings provide evidence that, under certain conditionsrelated to sales organization, idiosyncratic investments inthe customer, and the long-term orientation of the buyer-supplier relationship, salesperson’s continuity becomesless relevant as a determinant of customer’s trust in thesupplier. Specifically, I find that the importance ofsalesperson’s continuity declines – and eventually van-ishes – when the supplier has a large cross-functional teaminvolved in servicing or working with the customer, whenthe supplier dedicates a large amount of unique resourcesto the customer (in the form of unique product specifica-tions, procedures, and marketing activities), and when therelationship between the supplier and the buyer has ex-isted for a long time.

This study provides useful insights for supplier firmsand, in particular, for sales managers who have the great-est responsibility for deciding how long salespeople shouldlook after the same customer. This is especially relevant intimes when the competition is intense and global, andconsequently there are many circumstances where turn-over or transfer of salespeople may occur. All referencesin this abstract are available upon request.


For further information contact:Rodrigo Guesalaga

Pontificia Universidad Católica de ChileAv. Vicuña Mackenna 4860, Macul

Santiago, ChilePhone: 56.2.354.7982

E-Mail: [email protected][email protected]


CUSTOMER-CENTRIC CULTURE IN SALES ORGANIZATIONS: ON ITS

MEASUREMENT AND IMPACT ON SALESPERSON PERFORMANCE

Nikolaos G. Panagopoulos, Athens University of Economics & Business, Greece

SUMMARY

Over the last few years, there has been a growingrecognition among researchers and practitioners that thedevelopment of customer centricity in organizations con-stitutes the foundation of marketing (Sheth, Sisodia, andSharma 2000) and represents a source of competitiveadvantage (Day 2003; Gulati 2007; Shah et al. 2006).Interestingly, this recognition comes with the realizationthat very few studies have been conducted in the area thatcan guide companies in their transition to being customer-centric (e.g., Sheth, Sisodia, and Sharma 2000). To date,published studies are mostly conceptual and are discuss-ing customer-centricity at the organizational level, therebypaying less attention to organizational sub-processes,such as the sales process (Leigh and Marshall 2001).Against this backdrop, the aim of the present study isthreefold. First, we build and test a multiple-layer modelof customer-centric sales culture with the hope to provideinsights on the measurement issues surrounding the de-velopment of customer-centricity in sales organizations.This multi-layer conceptualization is grounded on thework of Homburg and Pflesser (2000) and echoes the callof Leigh and Marshall (2001) for future work on the area.Importantly, this approach allows for an in-depth analysisand understanding of the different layers comprisingcustomer-centric sales culture. Second, we examine thefactor structure of the customer centric culture construct.Third, we empirically examine the impact of customer-centric sales culture on salesperson performance.

Theory and Hypotheses

Culture was defined and conceptualized on the basisof Homburg and Pflesser’s work (2000). Thisconceptualization identifies four different layers of cul-ture: “values,” “norms,” “artifacts,” and “behaviors.”Briefly, culture exists simultaneously on three levels: onthe surface are behaviors and artifacts, underneath arti-facts lie norms, and at the core are basic shared values.Culture is, therefore, generated through values, but mani-fested and expressed sequentially through norms, arti-facts, and behaviors. Drawing on the above discussion,the present study approaches customer-centric sales cul-ture as a hierarchical model comprising four interrelatedlayers (Homburg and Pflesser 2000): (a) shared basicvalues supporting a customer-centric culture, (b) norms

for customer-centricity, (c) artifacts of customer-centric-ity, and (d) customer-centric behaviors.

Consistent with Leigh and Marshall’s recommenda-tions, the identification of values, norms, artifacts andbehaviors surrounding and supporting a customer-centricsales culture was based on Chally Group’s (2006) re-search report. In addition, to ensure content validity we (a)searched for related materials in major marketing jour-nals, textbooks, and Fortune 100 company web sites, and(b) discussed definitions and items with eight managersand salespeople. This process resulted in the identifica-tion of six dimensions tapping the customer centric salesculture construct: (1) Cross-functional coordination, (2)Adoption of a relationship/partnership business model,(3) Formalization of customer needs analysis processes,(4) Continuous improvement stressing customer satisfac-tion, (5) Repositioning the salesperson as business con-sultant, (6) Educating customers.

Based on prior work, we generated five researchhypotheses (e.g., Gebhardt, Carpenter, and Sherry 2006;Homburg and Pflesser 2000). In particular, according tothe multi-layer model of culture, values influence norms,which influence artifacts and customer-oriented behav-iors, which, in turn, influence behavioral and outcomesalesperson performance.

Research Methods

Data were collected from salespeople working in theGreek pharmaceutical industry; salesperson names andaddresses were provided by a mailing-list broker special-ized in the prescription-based pharmaceutical industry.We randomly selected 800 salespersons from the list.Salespeople received questionnaires via mail. A total of323 usable questionnaires were returned, representing aresponse rate of 40 percent.

Analysis

Measurement scales (i.e., customer-centric sales cul-ture, salesperson performance) were subjected to confir-matory factor analysis, which indicated that measurespossess adequate psychometric properties. Structural equa-tion modeling was then employed to asses the hypoth-esized model relationships. The fit statistics indicated that


the conceptual model fits the data well (RMSEA = .09;NNFI = .92; CFI = .93).

Discussion and Implications

Our study answers the calls for research in the realmsof customer-centric marketing (Sheth, Sisodia, and Sharma2000) and customer-centric sales culture (Leigh andMarshall 2001). Specifically, the present study contrib-utes to the sales management literature in three importantways: (1) we conceptualize a multi-layer model of cus-tomer-centric sales culture, (2) we empirically examine itsfactor structure, and (3) we test the nomological validityof the model by testing its effects on salesperson perfor-mance. Overall, our study provides evidence of discrimi-nant validity for the different layers of customer-centricsales culture and identifies six important dimensions ofcustomer-centrism that can be further explored in futureresearch.

On the basis of our research results, we identifyspecific actions that both senior sales executives and salesmanagers must implement for customer centricity to oc-cur within sales organizations. First, our study tells man-

agers that the establishment of customer-centric valuesand associated norms should be a top priority in modernsales organizations. This is so because developing anappropriate system of cultural values enhances norms andbehaviors supporting a customer-centric orientation,thereby increasing salesperson performance. Second, theeffects of customer-centric culture primarily occur throughvalues, norms, and behaviors. Artifacts (i.e., arrange-ments, stories, rituals, and language) do not appear to havea significant effect on salesperson performance. Thisresult is interesting and it partially reflects that, at least forthe sample companies, artifacts do not constitute impor-tant performance levers. This finding may also signal thatin contrast with employees that work inside the company(see Homburg and Pflesser 2000), artifacts in the form ofarrangements, stories and language, do not have an impor-tant effect in the enhancement of performance for bound-ary-spanners (i.e., salespeople) who are spending most oftheir time in the field. Rather, it is the set of basic valuesand expectations (norms) that will affect their behaviorand hence their performance. Finally, sales executives canuse our scale to measure customer-centricity among mem-bers of their sales forces with the intent to guide manage-rial actions. References are available upon request.

For further information contact:Nikolaos G. Panagopoulos

Athens University of Economics & Business47 Evelpidon & 33 Leykados

Athens, 113 62Greece

Phone: +302108203853Fax: +302108203851



FLATTERY’S CURSE: HOW CUSTOMER-ORIENTED SELLING

AFFECTS SHORT-TERM CUSTOMER SATISFACTION

AND DECISION TO BUY

Alexander Haas, University of Bern, GermanyPeter Kenning, Zeppelin University, Germany

SUMMARY

More than 80 years ago, Strong (1925) emphasizedthat personal selling strategies should be directed towardsecuring customer satisfaction as well as making the sale.Nearly 60 years later, Saxe and Weitz (1982) suggestedcustomer-oriented selling (COS) to be a key aspect ofsecuring customer satisfaction (Saxe and Weitz 1982).Since then, a substantial amount of research has beendirected toward COS (for an overview, see Schwepker2003). This work has focused on long-term effects of COSand has largely overlooked the short-term effects of COS.Accordingly, the question of how COS affects customers’decision to buy (CDTB) at the end of a given salesencounter remains still open. This is surprising given thatmaking the sale is an important objective of salespeople.

Against this background the aims of this study arethreefold. First, we will investigate the mechanism of howCOS affects customers’ decisions to buy. Building onprevious work, we consider direct and indirect effects,mediated by customer satisfaction with sales interaction(CSSI). In addition, these effects are investigated on thelevel of the dimensions of COS. Second, we will suggestand test a three-dimensional model of COS adding socio-emotional behaviors to the two dimensions of the con-struct as measured with the customer orientation andselling orientation sub-scales of the SOCO scale (Saxe

and Weitz 1982). Third, relative effects of the dimensionsof COS will be examined.

In order to achieve these three aims, we collected datain multiple retail outlets whose managers had agreed toparticipate in the research project and had granted thepermission to interview customers within the outlets in amajor German city. Research assistants were instructed tointerview consumers who had engaged in sales interac-tions right after the sales interactions. This process yieldedusable data about 203 sales interactions. Twenty-onepoint two percent of the sales interactions related toapparel, 78.8 percent to electrical appliances and tookplace in stores and retail outlets of different sizes. As to thesample demographics, 50.2 percent of the respondentswere males, 49.8 females. Their ages ranged from 17 to 82with a mean of 43.4 years.

Our empirical results support the dimensions’ indi-rect effects on CDTB, mediated by CSSI. The COS-Dimension “building a positive interaction climate” wasfound to have the strongest impact on CSSI and a directnegative effect on CDTB. Also, results supported theexpected inverted u-shaped relationship between “avoid-ing pressure to close” and CDTB. Finally, COS was foundtoo only have a little impact on CDTB. References avail-able upon request.

For further information contact:Alexander Haas

Marketing DepartmentInstitute of Marketing and Management

University of BernEngehaldenstrasse 4

CH-3012 BernGermany

Phone: +41(0)31.631.4541Fax: +41(0)31.631.8032

Mail: [email protected]


A COMPARISON OF TELEVISION CELEBRITY USE IN THE

UNITED STATES AND LEBANON

Morris Kalliny, University of Missouri – RollaAbdul-Rahman Beydoun, University of International Florida

Anshu Saran, University of Texas of the Permian BasinLance Gentry, University of Missouri – Rolla

SUMMARY

Jagdish and Kamakura (1995) argued that celebrityendorsement has become a prevalent form of advertisingin the United States. Approximately 20 percent of alltelevision commercials feature a famous person, andapproximately 10 percent of the dollars spent on televi-sion advertising are used in celebrity endorsement adver-tisements (Advertising Age 1987; Sherman 1985).

There are several reasons for the extensive use ofcelebrities in advertising. Research findings show thatcelebrities make advertisements believable (Kamins et al.1989), enhance message recall (Friedman and Friedman1979), aid in the recognition of brand names (Petty,Cacioppo, and Schumann 1983), create a positive attitudetoward the brand (Kamins et al. 1989), and create adistinct personality for the endorsed brand (McCracken1989). Because it is believed that celebrity endorsementsare likely to generate a greater likelihood of customerschoosing the endorsed brand (Heath, McCarthy, andMothersbaugh 1994; Kahle and Homer 1985), businessare willing to pay the high price to obtain it.

The purpose of this study is to compare the use ofcelebrity endorsement between the United Sates and Leba-non in terms of two fundamental cultural dimensions: (1)low versus high context, and (2) individualism versuscollectivism. This study will investigate differences andsimilarities regarding celebrity characteristics in the U.S.and Lebanon.

Celebrity and Celebrity Endorsements in the Arab

World

Although celebrity fascination takes place in theArab world, it is likely that the way people react to

celebrity is different from the way people do in the U.S.due to religious and cultural factors. In spite of thesecultural and religious values; however, Arab fascinationwith celebrity has recently increased due to more expo-sure to the West and particularly the United States. Forexample, BBC News (2005) reported that “a Saudi star ofa reality TV show caused such commotion among fans ina Riyadh mall he was reportedly arrested for sparking an“indecent scene.” Saudi fans tried to hug and kiss the starwhich is against religious and cultural values of thekingdom.

Research Questions

RQ1: Which one of the two countries (U.S. or Lebanon)uses more celebrity in television advertising?

RQ2: What are the types and characteristics of celebritiesappearing in Lebanese and U.S. television adver-tising?

RQ3: What are the product types for which celebrities arecommonly employed as endorsers in Lebanese andU.S. television advertising?

Research Hypotheses

H1: Lebanese Celebrity will be more likely to conveyless information compared to their U.S. counter-parts due to the high context nature of the Lebaneseculture (was not supported).

H2: Lebanese celebrity will be more likely to employcollectivism appeals compared to their U.S. coun-terparts due to the collectivistic nature of the Leba-nese celebrity (was not supported).

For further information contact:Morris Kalliny

University of Missouri – RollaRolla, MO 65401



CELEBRITY ENDORSEMENT, BRAND EQUITY, AND

BRAND CREDIBILITY

Amanda Spry, University of Queensland, AustraliaRavi Pappu, University of Queensland, Australia

T. Bettina Cornwell, University of Queensland, Australia

SUMMARY

Celebrity endorsement is considered a valuable pro-motional tool by marketers worldwide. Companies investheavily in celebrities, with one-in-four advertisementsusing celebrity endorsement (Market Watch 2006, p. 29).

The term, brand equity, refers to the incrementalvalue added by a brand name to a product (Farquhar 1989,p. 7). Firms’ major marketing activities are generallyaimed at building, managing and exploiting this importantmarketing metric (Yoo and Donthu 2001, p. 1). Brandequity has emerged as a priority area in the marketingdiscipline, leading to calls for research identifying strate-gies for building brand equity.

Several researchers have emphasized the importanceof brand associations in building brand equity. Celebrityendorsement represents one tool marketers can use todevelop and maintain appropriate brand associations(McCracken 1989, p. 12). Despite suggestions from re-searchers (e.g., Till 1998) that celebrity endorsementcould lead to improved brand equity, extant research doesnot offer much guidance on this issue. Specifically, exist-ing research does not provide any empirical evidence forthis relationship.

The principal objective of the present research is toprovide a conceptual framework for explaining the impactof the credibility of the celebrity endorser used on theconsumer-based equity of the endorsed brand. The medi-ating role of the credibility of the endorsed brand on the“celebrity credibility-brand equity” relationship is alsoexamined. The moderating role of the type of brandingemployed on the ‘celebrity credibility-brand equity’ rela-tionship is also examined. Three propositions are ad-vanced for the proposed relationships. Guidelines are alsoprovided for empirical testing.

We conceptualize the credibility of the celebrityendorser based on the source model theory which con-tends that endorser credibility is based on his/her attrac-tiveness, expertise and trustworthiness (e.g., Biswas,Biswas, and Das 2006, p. 18; Erdogan 1999, p. 297;Ohanian 1990, p. 41). Consumer-based brand equity isconceptualized based on Accurst (1991, p. 16) brandequity framework, as a multidimensional construct com-prising of brand awareness, brand associations, perceived

quality and brand loyalty, which is consistent with recentwork in this area (e.g., Pappu, Quester, and Cooksey2006; Yoo and Donthu 2001; Washburn and Plank 2002).Brand credibility has been conceptualized as a two-di-mensional construct comprising trustworthiness and ex-pertise (Erdem and Swait 2004, p. 192). Brand credibilityrefers to how effectively information is conveyed by thebrand signal and how truthful and dependable that infor-mation is considered to be (Erdem and Swait 1998,p. 137).

The relationship between credibility of a celebrityendorser and brand equity is explained using the associa-tive network memory model (e.g., Anderson 1973) fromcognitive psychology and the persuasion knowledge model(Friestad and Wright 1984). Previous research has usedassociative learning principles to articulate the underlyingprocess of celebrity endorsement (e.g., Till 1998). Weargue that by using endorsement, a celebrity could belinked to a brand within a consumer’s memory, henceproviding a favorable association conducive to the cre-ation of brand equity.

The mediating role of brand credibility on the cred-ibility-brand equity relationship is explained using thebrand signaling theory from information economics. Weargue that the credibility of a celebrity endorser will addcredibility to the signal of the endorsed brand. Subse-quently, brand credibility can build brand equity by favor-ably influencing consumer attribute perceptions, infor-mation costs and perceived risk.

It is proposed that the impact of celebrity credibilityon brand equity would differ depending on whether acelebrity is endorsing a parent brand or a sub-brand. Asub-brand is a type of brand extension, whereby a newbrand is combined with a parent brand (Keller 2003,p. 577). Using associative learning principles, it is arguedthat a celebrity endorser would be more effective for asub-brand because a sub-brand would have a less-estab-lished associative network in consumer memory.

The proposed framework could be tested using anexperimental design. This experiment could follow a 3 x 2between-subjects factorial design. That is, source cred-ibility could be varied at three levels (control group/lowcredibility endorsement/high credibility endorsement).Type of branding could be manipulated at two levels


(parent brand/sub-brand). Valid and reliable measures areavailable from past research to measure the variablesincluded in our conceptual framework. The experimentalmanipulations could be implemented using print adver-tisements featuring celebrities of varying credibility for abrand and its sub-brand selected through pre-testing. Datafor this research could be collected using a mall-interceptapproach which would yield a sample of real consumers.To strengthen the validity of this research, the proposedexperiment could be replicated across two brands withina chosen product category.

The present research has significant implications formarketing theory and practice. The framework proposedin this research is one of the first to conceptualize a

relationship between celebrity credibility and consumer-based brand equity. The present research also contributesby identifying the mediating role of brand credibility andthe moderating role of type of branding on the “celebritycredibility and consumer-based brand equity” relation-ship. This research also demonstrates another applicationof associative learning theory. In terms of practical impli-cations, this research will be useful for managers in theirselection of celebrities. Further, the present research pro-vides managers with compelling justification for the useof celebrity endorsement as an effective promotionaltechnique which can genuinely create value for a firmsuch as improvements in intangible assets like brandequity. References are available upon request.

For further information contact:Ravi Pappu

UQ Business SchoolUniversity of Queensland

Brisbane QLD 4072Australia

Phone: +61.7.3365.6567Fax: +61.7.3365.6988



COMMUNICATING CORPORATE BRANDS THROUGH NARRATIVES:

THE MODERATING EFFECT OF PERSUASION MOTIVES

Daniel Wentzel, University of St. Gallen, SwitzerlandTorsten Tomczak, University of St. Gallen, Switzerland

Andreas Herrmann, University of St. Gallen, Switzerland

SUMMARY

In recent years, researchers have paid increasingamounts of attention to the management of corporateassociations and corporate brands (e.g., Brown et al.2006; Berens, van Riel, and van Bruggen 2005; Brownand Dacin 1997). A strong corporate brand is a strategicasset because it can positively affect how different stake-holders evaluate and respond to the organization and tothe products and services it offers. As such, managing andcommunicating a corporate brand constitutes an impor-tant strategic task and, consequently, more research isneeded that shows how this can be achieved successfully(Dacin and Brown 2002). The purpose of this research isto investigate if a corporate brand and its underlyingvalues can be communicated effectively through the useof narratives or stories (van Riel 2000; Dowling 2006).

Narratives and the Persuasion Knowledge Model

Narratives or stories are defined as “knowledge struc-tures that consist of a sequence of thematically and tem-porally related events” (Adaval and Wyer 1998, p. 208).A great deal of studies have shown that information thatis presented in a narrative form can lead to positiveconsequences in terms of memory, affect, and persuasion(e.g., Escalas 2004; Adaval and Wyer 1998; Polyorat,Alden, and Kim 2007; Pennington and Hastie 1988).Narratives exert their persuasive impact through a cogni-tive process called “transportation” (Escalas 2004, 2007;Green and Brock 2000). Transportation denotes the extentto which individuals immerse themselves into the contentof the narrative and get “lost” in it. Transportation en-hances persuasion by eliciting strong affective reactionsand reducing the amount of negative cognitive responses(Escalas 2007; West, Huber, and Min 2004). These stud-ies would suggest that narratives are an effective way tocommunicate the values of a corporate brand to consum-ers.

We, however, posit that under some circumstancesnarratives cause consumers to infer a persuasion motiveand to resist the message of the narrative. This perspectiveis consistent with the Persuasion Knowledge Model(Friestad and Wright 1994). For instance, Campbell (1995)demonstrated that attention-getting advertising tactics

can cause viewers to question the motives of the advertiserand to draw inferences of manipulative intent. Theseinferences, in turn, exerted a negative impact on theevaluations of the advertiser. We posit that similar pro-cesses may be at work in the context of narratives. Whenconsumers do not infer a persuasion motive, narrativesmay be a very effective means for communicating thevalues of a corporate brand. Consumers are then drawninto the narrative and can form a stronger connectionbetween their own selves and the brand that is depicted inthe narrative. When, however, an ulterior motive for usingnarratives is made accessible, consumers may infer thatthe company is trying to manipulate them by unfairmeans. They may think that narratives are powerful toolsof persuasion and that marketers are using narratives forprecisely that reason. As a result of these inferences,consumers may discount the narrative relative to situa-tions when an ulterior motive is not as evident (Campbelland Kirmani 2000; Campbell 1995). Hence, narrativesmay lose some of their persuasive power if they areassociated with a persuasion motive.

Methods, Results, and Discussion

To test our hypotheses, we conducted two experi-ments with adult German and Swiss consumers. In bothexperiments, we asked participants to evaluate a corpo-rate brand that was either represented in a narrative or anexpository, factual format. Furthermore, the extent towhich a persuasion motive was accessible was also ma-nipulated. In the first experiment, we manipulated theaccessibility of a persuasion motive by suggesting that thecompany may have had an ulterior motive for communi-cating its corporate brand. In the second experiment, wetriggered a persuasion motive by embedding the brandvalues into a narrative theme that was not entirely cred-ible. The effect was similar in both cases. When ulteriormotives were easy to discern, participants inferred thatthey were being unfairly persuaded and, as a result, thepositive impact of the narrative was attenuated. Partici-pants partially discounted the brand and its values andconsidered them to be less relevant and credible. Interest-ingly, the presence or absence of a persuasion motive didnot affect participants’ responses when the brand wasrepresented in an expository fashion. These results holdimportant implications for managers that decide to use


narratives to communicate their corporate brands. Whennarratives are executed poorly and ignite a persuasionmotive, they are likely to be met with resistance and may

eventually backlash against the brand. References areavailable upon request.

For further information contact:Daniel Wentzel

Institute of Marketing and RetailingUniversity of St. GallenDufour Street No. 40a

St. Gallen, 9000Switzerland

Phone: +41(0)71.224.71.62Fax: +41(0)71.224.28.57



THE NEGLECTED POWER OF MOOD: HOW IT MODERATES BRAND

PLACEMENT’S EFFECTIVENESS

Hans H. Bauer, Mannheim University, GermanyMelchior D. Bryant, Mannheim University, Germany

Marcus M. Neumann, Mannheim University, Germany

SUMMARY

Brand placements take root as a successful advertis-ing approach reflected by an expected increase of globalsales of 30 percent to US$4.4 bn. in 2007 (PQ Media2007). Following d’Astous and Chartier (2000), brandplacements contain all intentional placements of brandedproducts in motion pictures for promotional purposes.Thus, this approach becomes more and more importantfor the advertising industry as a visible flash of the brandduring the current program can be guaranteed even thoughviewers increasingly avoid commercials (Harvey 2001).

While the effectiveness of brand placements in termsof brand attitudes and purchase intentions is alreadyshown in studies (e.g., Law and Braun 2000; Russel 2002)the role of the viewers’ mood has not been explored.Thereby, mood is defined as “feeling states that aresubjectively perceived by individuals” (Gardner 1985,p. 282). Further, motion pictures are able to evoke posi-tive or negative mood states in the audience which in turninfluence the effectiveness of traditional commercials(e.g., Goldberg and Gorn 1987; Holbrook and Batra1987). Karrh, McKee, and Pardun (2003) revealed in anexploratory study with marketing professionals the beliefthat induced mood is likely to influence the viewers’reactions toward brand placements. However, no empiri-cal study has examined whether mood moderates theknown antecedents of brand placement’s success.

To address this research gap we apply a two-stepprocedure. At first, we are developing a core model ofbrand placement’s effectiveness including antecedentsbased on findings of existing literature. We are usingstructural equation modeling for assessing the signifi-cance and the effect size of each antecedent. Then, we aretesting every path in the core model for being moderatedby movie-induced mood.

Altogether, our core model consists of six hypoth-eses: We assume that the attitude toward advertising ingeneral determines the attitude toward the brand place-ment in the movie which in turn influences the attitudetoward the placed brand. Further, we hypothesize that theimage congruency between the placed brand and themovie affects the attitude toward the brand placement inthe movie, whereas the image congruency between the

placed brand and the character has an effect on the attitudetoward the placed brand. In addition, the attractiveness ofthe character affects also the attitude of the placed brand.Finally, we postulate that the attitude of the placed brandinfluences the respective purchase intention.

Based on our core model, we discuss movie-inducedmood as a moderating variable of the hypothesized rela-tionships. Basically, we assume that every path on thecore model is positively moderated by the viewer’s per-ceived mood.

We used data of 237 study participants (average ageof 26.3 years, 32.1 percent female, 67.9 percent male) inan online setting to assess our assumptions. At first, testpersons were asked to choose one of two movies (“Hitch”with a placement of a car brand and “Ocean’s Twelve”with a placement of a mobile phone brand). Afterwards,test persons were exposed to the official movie trailer andto the scene that included the brand placement. At last, thequestionnaire was shown to the participants. The com-puted results confirmed all hypotheses.

The study shows for the core model that the use ofbrand placements influences desirable dispositions of thepurchase behavior. The following recommendations mightbe given. Brands should be placed in a positive environ-ment. To increase the acceptability of the message, it isessential to integrate brands exclusively in movies withcorresponding images. If the positioning is too inappro-priate the recipient will perceive it as an attempt ofpromotional influence, which in turn will lower the ad-vantage of the placement. This conscious perception canlead to an intense negative attitude toward the advertisedbrand. Likewise, it is important to pay attention to a highdegree of congruity between the image of the characterand the advertised brand. The attractiveness of the charac-ter also constitutes an important factor to activate identi-fication processes of recipients and, therefore, increasethe effects of brand placements through imitating pro-cesses.

Reviewing the core model for moderating effectsreveals that the paths are influenced by the viewers’ moodstate. In the positive mood condition effect sizes in thecore model are strengthened while relationship effectsdecrease in the less positive mood condition and, there-


fore, lower brand placement’s success. This finding is ofutmost importance as brands should only be placed inentertainment programs, which can assure positive moodeffects. Especially the integration of brands into liveprograms like sport events or shows are critical due tounpredictable mood causing contents, if the favored char-acters are not performing as expected. In this case, thesuccess of brand placements cannot be controlled by

exogenous variables like context congruity or character’sattractiveness.

Resumptive, our research helps managers to under-stand the effectiveness of brand placements and givesthem a guideline to successfully implement branded prod-ucts in motion pictures.

REFERENCES

d’Astous, Alain and Francis Chartier (2000), “A Study ofFactors Affecting Consumer Evaluations and Memoryof Product Placements in Movies,” Journal of Cur-rent Issues and Research in Advertising, 22 (2), 31–40.

Gardner, Meryl P. (1985), “Mood States and ConsumerBehavior: A Critical Review,” Journal of ConsumerResearch, 12 (3), 281–99.

Goldberg, Marvin and Gerald J. Gorn (1987), “Happy andSad TV Programs: How They Affect Reactions toCommercials,” Journal of Consumer Research, 14(3), 387–403.

Harvey, Bill (2001), “Measuring the Effects of Sponsor-ship,” Journal of Advertising Research, 41 (1), 59–65.

Holbrook, Morris B. and Rajeev Batra (1987), “Assessingthe Role of Emotions as Mediators of ConsumerResponses to Advertising,” Journal of Consumer

Research, 14 (3), 404–20.Karrh, James A., Kathy B. McKee, and Carol J. Pardun

(2003), “Practitioners’ Evolving Views on ProductPlacement Effectiveness,” Journal of AdvertisingResearch, 43 (2), 138–49.

Law, Sharmistha and Kathryn A. Braun (2000), “I’ll HaveWhat She’s Having: Gauging the Impact of ProductPlacements on Viewers,” Psychology and Market-ing, 17 (12), 1059–75.

PQ Media (2007), “PQ Media Market Analysis FindsGlobal Product Placement Spending Grew 37% in2006,” press release, (March 14), (accessed June 28,2007), [available at http://www.pqmedia.com/about-press-20070314-gppf.html].

Russell, Cristel A. (2002), “Investigating the Effective-ness of Product Placements in Television Shows: TheRole of Modality and Plot Connection Congruenceon Brand Memory and Attitude,” Journal of Con-sumer Research, 29 (3), 306–18.

For further information contact:Melchior D. Bryant

Mannheim UniversityL 5, 1

Mannheim 68161Germany

Phone: +49.621.181.1570Fax: +49.621.181.1571



FLOW EXPERIENCE: NEW APPROACHES FOR

CONCEPTUALIZATION AND MODELING OF A

MULTIFACETED CONSTRUCT

Jan Drengner, Chemnitz University of Technology, GermanyPia Furchheim, Chemnitz University of Technology, Germany

Manuela Sachse, Chemnitz University of Technology, Germany

SUMMARY

The flow experience has gained increasingly moreattention in marketing research. So far, it has mainly beenused to explain the impact of commercial websites, soft-ware use or the analysis of marketing events. There havebeen several attempts to develop a detailed definition,conceptualization, and operationalization of the construct.However, these findings often encounter problems interms of content validity. One problem regards a blurredconceptualization of flow. Either only some of the facetsare considered or further constructs describing flow areadded. A second problem can be identified in respect todifferent views of the distinction between reflective andformative facets of the flow construct, its antecedents andconsequences. A third problem area concerns the model-ing of this multifaceted construct. Based on literaturereview and three empirical studies, the authors developeda revised and valid conceptualization of flow and exam-ined several approaches of modeling multifaceted con-structs. For that, the framework of Bagozzi and Heatherton(1994) was applied.

Empirical Studies

To clarify the conceptual framework and to examinethe developed scale, two pilot studies were conducted

with spectators of the FIFA World Cup Germany 2006.The data for the main study was collected with the help ofan online questionnaire using participants of the Mas-sively Multiplayer Role-Playing Game “World ofWarcraft.” The final sample of 618 respondents was usedto compare the different approaches of modeling multi-faceted constructs and their adaptation to the flow con-struct. The parameters of the models were estimated withthe help of CFA.

Conceptualization of Flow

Reviewing the literature, the mainly used facets weretaken into account for this conceptualization in order torepresent the construct as a whole. To distinguish forma-tive and reflective facets (see Table 1), the authors adaptedappropriate decision rules by Jarvis, MacKenzie, andPodsakoff (2003).

Some facets neither build manifestations of the con-struct (reflective view) nor the elimination of these facetsnecessarily alters the conceptual domain of the flowconstruct (formative view). Hence, it seems to be reason-able to separate these facets from the construct as anteced-ents. The classification of the facets Control and Concen-tration can be termed both as reflective facet and anteced-ent. Therefore, both views were considered. With respect

TABLE 1

Reflective Facet Formative Facet Antecedent Consequence

• action-awareness merging – • clear goals • autotellic experience(enjoyment, positivejubjective experience)

• loss of self-consciousness • unambiguous andseamless feedback

• transformation of time • challenge -skill balancebalance

• concentration on task at hand • convenience on task at hand

• control


to their perceived control, the results of the empiricalstudies showed no distinction between respondents whoexperienced flow and those who did not. Hence, this facetshould only be conceptualized as an antecedent of flow.

In conclusion, the following definition can be de-rived. Flow is a holistic experience, explaining the state inwhich the acting individual is absent-minded, loses allsense of time while being highly concentrated and havingthe impression of its consciousness and action merging.

TABLE 2

Approach Description

I Partial Aggregation with The facets are organized hierarchically as indicators of an underlyingHierarchical Organization factor. The indicators of each facet are summed up as indices and load asof Facets indicators on a higher order single-factor (flow).

IIa Total Disaggregation The facets are formed as correlated latent variables. Each item is used toFirst-Order operationalize its respective facet.

IIb Total Disaggregation The facets are organized as first-order factors. The second-order factor canSecond-Order be thought as an abstract representation of the overall flow construct.

Approaches Representing Flow as a Multifaceted Con-

struct

Comparing the different fit indices (χ2, RMSEA,SRMR, NNFI, CFI, PNFI, CVI), it can be concluded thatthe analyzed approaches that do not make any distinctionbetween the various facets of flow have to be rejected.Moreover, it can be constituted that all approaches reflect-ing the variety of a multifaceted construct in one oranother way (see Table 2) can be supported.

In summary, the following practical implications canbe deduced: Both total disaggregation models (II) can beused if the researcher wants to focus on the structure of theflow construct. If the aim is to integrate the flow constructinto a higher-order structure, one can choose between twoapproaches. Focusing on the impact of the respectivefacets on other key constructs, the first order approach

(IIa) can be suggested. The partial aggregation model (I)would be useful to analyze the relationship between theflow experience as a global construct and other keyconstructs. Hence, due to its lower complexity, the partialaggregation model seems to be the most practicable alter-native to the total disaggregation second order model.References are available upon request.

For further information please contact:Jan Drengner

Department of MarketingChemnitz University of Technology

Thueringer Weg 709107 Chemnitz

GermanyPhone: +49.371.531.341.58

Fax: +49.371.531.261.39E-Mail: [email protected]


EXPLORING PERCEPTIONS OF CONSUMER EXPERIENCE ONLINE

Neil Frederick Hair, Rochester Institute of Technology, RochesterDeborah Colton, Rochester Institute of Technology, Rochester

SUMMARY

This paper explores consumers’ perceptions of theironline experience beyond website interface issues. UsingPersonal Construct Theory (Kelly 1955) this researchidentified twelve themes including; aesthetics, customerorientation, customization, dependability, entertainment,functionality, identification, informative, innovative, price,product/service offer, and social. These findings buildupon those reported in existing literatures on Web sitequality. Specifically, this study captured identificationand social aspects of online experience as importantissues.

Increasingly, customers are using the internet forhedonistic rather than utilitarian purposes. The majorityof customer experience research conducted to date hasfocused on the web site interface or specifically on theconsumer’s feelings or state of mind. The Web site qualityliterature predominantly considers the interface betweenthe consumer and the Web site. The main consideration ofthe customer experience literature has been the customerhimself. In order to explore this gap in the literature thispaper considers the perceptions of consumers’ experi-ences online.

This study draws from George Kelly’s Personal Con-struct Theory (1955). Originally designed as a psycho-logical instrument for exploring patients’ perceptions ofinterpersonal relationships the approach was adopted by

marketers looking to explore consumers’ perceptions ofmarket offerings (Guttman 1982). The methodology isqualitative exploring consumers’ perceptions of productsor services through a process of comparisons betweenobjects. Initially participants are asked to select threewebsites and note, “how two are immediately similar butdifferent from a third in terms of your online experience.”The answers to this triadic sorting are noted as a pair of“constructs” which then forms the basis of a rating scale.Web sites are then rated against the two “construct poles”and the process resumes with a new triadic sort. Oncesaturation is achieved, the results are collated and codeddrawing out patterns of similarities and differences result-ing in the derivation of key themes.

The sites used in this study were taken from an openpoll of graduate and undergraduate students who com-prised the sample frame for this study. Seventy-sevenInternet Marketing students were asked to identify mostfrequently used sites. The results were collated and the tenmost frequently cited chosen. The ten sites used were;Amazon.com, Dell.com, Netflix.com, Walmart.com,BestBuy.com, eBay.com, Google.com, ITunes.com, andFacebook.com.

Overall the study revealed twelve key themes and601 pairs of constructs denoting the samples perceptions’of their experience with Web sites (1202 construct poles).References are available upon request.

For further information contact:Neil Frederick Hair

E. Philip Saunders College of BusinessRochester Institute of Technology

108 Lomb Memorial DriveRochester, NY 14623Phone: 585.475.6322


Deborah ColtonE. Philip Saunders College of Business

Rochester Institute of Technology108 Lomb Memorial Drive

Rochester, NY 14623Phone: 585.475.4939



CONSUMER CONTROL AND THE PSYCHOLOGY OF DVR USE

Matthew S. Eastin, University of Texas at AustinTerry Daugherty, University of Texas at Austin

Galit Marmor-Lavie, University of Texas at AustinSonny Rosenthal, University of Texas at Austin

SUMMARY

Media fragmentation, consumer interactivity, andpersonalization are the product of technology advance-ments. Furthermore, these advancements lead to oneoutcome – the empowerment of the consumer. Recently,Andrew Heyward, former president of CBS news, con-sumers create their own media environment and “watchwhat they want when they want” (Page 2006, p. 18).Designating this phenomenon as the “the control revolu-tion,” Shapiro (1999) claims that technology has broughtwith it a reduction of institutional control resulting in anincrease of individual control. Recently, Cover (2006)defined the control revolution as a “push-and-pull ofauthor versus (intended) audience control over the text”(p. 153). Control represents a personality trait that linksuse with gratifications sought from media. Certainly, therelationship between psychological personality traits andmedia use is frequently noted in the literature (Finn 1997;Hall 2005; and so on). We adopt the “desire of control”construct suggested by Burger and Cooper (1979) tobetter understand the use of DVRs. Building on Eastin’swork (2005), the role of self-efficacy in DVR use is alsoexamined. Finally, enjoyment and social interaction asoutcomes of DVR use will be examined.

As shown in Figure 1, DVR self-efficacy, priorexperience, and desirability of control all act directly onDVR use. DVR self-efficacy and desirability of controlalso act indirectly on perceived outcomes from use. Fi-nally, DVR use directly influences perceived enjoymentand social expectations.

Methods

Participants were part of a large opt-in online re-search panel recruited over a seven-day period throughEMI Market Research. A total of 3,000 emails were sent.The survey was closed once 325 completed surveys werecompleted (day 7). The current research was only inter-ested in DVR users, thus, only those who indicated beingDVR consumers were included in analysis (N = 106). Ofthose participants, 63 were male and 43 were female.Ages ranged from 21 to 84 (M = 43.50, SD = 15.05).Eighty-five percent were Caucasian, 4 percent were Afri-can American, 4 percent were Asian, 6 percent wereLatino, and 2 percent indicated other.

DVR Use. Guided by Ferguson and Perse’s (2004) 14DVR functions, participants were asked to indicate therelative frequency of use on a Likert-type scale where 1 =Never and 7 = Always (M = 4.28, SD = 1.09). Prior DVRExperience. Prior DVR experience used a single item toestimate the length of time (months) each participant hasowned a (M = 27.49, SD = 24.75). Desirability of control.Participants were asked to indicate on a Likert scale theiragreement with eight statements where 1 = StronglyDisagree (SD) and 7 = Strongly Agree (SA) (M = 5.34,SD = .89) (i.e., “When I see a problem, I prefer to dosomething about it rather than sit by and let it continue”and “I enjoy having control over my own destiny”; α =.86)(Althaus and Tewksbury 2000; Burger and Cooper 1979).DVR Enjoyment. Participants were asked to indicate on aLikert scale their agreement with four statements where1 = SD and 7 = SA (M = 4.93, SD = 1.33). The statementsreflected the likelihood they would “enjoy the showmore,” “be more entertained,” “have more fun,” and “bemore relaxed” when using a DVR (α = .94) (LaRose andEastin 2004). DVR Social Engagement. Participants wereasked to indicate on a Likert scale their agreement withfive statements where 1 = SD and 7 = SA (M = 4.60, SD =1.23). The statements reflected the likelihood they would“spend more time with others,” “watch shows to talkabout with others,” “stay current with programs otherswatch,” “talk about television programming,” and “watchtelevision with others” when using a DVR (α = .91)(Eastin 2005). DVR Self-Efficacy. Participants were askedto indicate on a Likert scale their agreement with fourstatements where 1 = SD and 7 = SA (M = 5.46, SD = 1.50).From Eastin’s (2002), the statements reflected attitudes of“my interaction with the digital video recorder is clear andunderstandable,” “interacting with the digital video re-corder does not require a lot of mental effort,” “I find thedigital video recorder is easy to use,” and “I find it easy toget the digital video recorder to do what I want it to do”(α = .96).

Results

Path analytic results (Figure 1) indicate the datarepresent a relatively good fit of the model proposed,2(3) = 19.07; p < 0.05; CFI = 0.90; RMSEA = 0.22; GFI =0.90.


Conclusions

Through the integration of desirability of control andthe social cognitive construct of self-efficacy, this paperadvances current understanding of how and why peopleare using DVRs. Further, these constructs, as well asactual DVR use, provide insight and predictive power forthe outcomes of enjoyment and social interaction. More-over, as a central component to this study, desirability ofcontrol also displayed positive relationships with DVRconsumption and the expected outcomes. This is hardly

FIGURE 1

Proposed and Tested Path Analysis for DVR Use

.25

DVR Self-Efficacy

Prior Experience

Desirability of Control

DVR Use R2= .15

DVR Enjoyment

R2 = .44

DVR Social Interaction

R2 = .41

.11

.50

.06.16

.11 .44

.31.30

surprising considering the desirability of control con-struct represents the aspiration of a person to be in control(behaviorally speaking) (Burger 1984; Burger 1992). Inconclusion, the current paper sides with the idea thatconsumers increasingly holding more control over whatand when they view mediated content. The current paperoffers a new direction in this area, suggesting desirabilityof control and self-efficacy as two cognitive constructsthat influence behavior. References are available uponrequest.

For further information contact:Matthew S. Eastin

Department of AdvertisingThe University of Texas

CMA 7.143Austin, TX

Phone: 512.471.3429Fax: 512-471-7018



MEASURING CONSUMER RITUALS: A MARKETING APPLICATION

Larry Neale, The University of Western Australia, AustraliaRichard Mizerski, The University of Western Australia, Australia

Alvin Y.C. Lee, The University of Western Australia, Australia

SUMMARY

Some brands seem to garner uncommon levels ofloyalty from their customers. These brands can weathereconomic downturns, long-term competitive disadvan-tage and continual performance failures to emerge with acore of dedicated, committed, and loyal consumers. Goodexamples of this phenomenon come from sports. Somesports teams have fans who proudly proclaim their loyaltyas well as financially support their team through atten-dance, yet live their entire life without witnessing theirteam win a championship. Why would they do this, whenswitching brands is possible?

One of the features of the sporting industry is theritualized way in which it is consumed across the world.Rituals are an important part of society, and are a frequenttopic of investigation among sociologists and anthropolo-gists. Fans of every sport use rituals and superstitions tohelp them enjoy the spectacle, socialize with other like-minded fans, and reduce some of the anxiety of watchingtheir team play. Yet marketers generally do not includeconsumption rituals in their marketing plans. One pos-sible explanation is that there is currently no way toobjectively quantify and measure rituals in a consumptioncontext.

This study uses a sample of 651 attendees at anAustralian Football League game to explore ritual behav-ior, define the game-day rituals observed, and design ascale to measure sports fan ritual. Sample items wererefined through a four-step process, (1) observation of

fans at professional football games, (2) reviewing litera-ture involving sports fans, (3) interviews with sport mar-keting practitioners, and (4) consultations with sport mar-keting academics. A sixteen-item scale was designed forspectators of Australian football during game day. Thesescale items are behaviorally based and include bodypainting, wearing a lucky charm, and singing the teamsong.

A two-dimensional construct was proposed. Thedimensions were labeled Social Rituals and PersonalRituals. Social rituals are performed in groups, or with thepurpose of involving others in the ritual such as singingthe team song. Personal rituals are carried out individuallyand include praying for team success. Exploratory factoranalysis supported the two-dimensional construct, whichwas confirmed using Structural Equation Modeling.

For academic researchers, the findings are importantto establish the role of ritual in consumption and loyalty,while opening future research opportunities in other prod-uct categories. For sports marketers, the results indicatethe importance of developing and facilitating consump-tion rituals tied to game day attendance, with a view togenerating uncommon loyalty. While the results from thescale validation procedure are encouraging, further scaledevelopment and refinement is suggested as there appearto be other latent dimensions affecting ritualistic con-sumption. This study serves as a starting point to measureritualistic consumption. References are available uponrequest.

For further information contact:Larry G. Neale

UWA Business School – M261The University of Western Australia35 Stirling Hwy., Crawley WA 6009

AustraliaPhone: +61.8.6488.7154



SOURCES OF GLOBAL E-TAIL ADVANTAGE: RELATIONSHIPS

AMONG ORIENTATIONS, RESOURCES, AND PERFORMANCE

Deborah Colton, Rochester Institute of Technology, RochesterMartin S. Roth, University of South Carolina, Columbia

William O. Bearden, University of South Carolina, Columbia

SUMMARY

Retail e-commerce continues to grow as an importantchannel of distribution in the global business environ-ment. Annual U.S. e-tail sales revenue growth of 22.2percent between 2004 and 2005 strongly outpaced thetotal retail sales growth of 6.3 percent during the sameperiod (U.S. Census Bureau 2007). Similar and oftengreater growth rates are common in other markets glo-bally. Despite the growth of e-tailing, many e-tailersstruggle to understand how to create customer value andgain competitive advantage (Larson 2001).

Firms that achieve superior performance typicallypossess and leverage important resources such as assets,knowledge, and processes that allow them to efficientlyand effectively develop and implement successful strate-gies. These strategies form the basis of competitive ad-vantage, and the more resistant the resources are to dupli-cation or imitation, the more sustainable the advantageand superior performance. This resource-based view ofthe firm (Wernerfelt 1984; Barney 1991) provides atheoretical framework through which to assess how e-tailers can achieve competitive advantage and superiorperformance. As described by Day (1994) and Hult andKetchen (2001), firms often utilize critical capabilitiesthat independently and collectively contribute to the cre-ation of unique resources. Thus, competitive advantageand subsequent performance are contingent on firmsmanaging capabilities to create valuable and inimitableresources.

Following Hult and Ketchen’s approach (2001), weidentify certain capabilities (market orientation, entrepre-neurial orientation, and foreign market orientation) thatcan contribute to e-tailers’ abilities to create strong brandsand supplier relations. Firm orientations have been linkedto performance (Lumpkin and Dess 1996). Similar toNoble, Sinha, and Kumar (2002), we hypothesize that theimpact of these orientations on performance are not direct,but rather are mediated by the resources they help create.This study investigates the ability of important e-tail firmresources to mediate the effects of market orientation,entrepreneurial orientation, and foreign market orienta-tion on global revenue growth and firm performance.

Two key sources of customer value creation andcompetitive advantage in e-tailing are brand strength and

supplier relations. While price is a key driver of e-com-merce transactions, retailer name and reputation are im-portant criteria in e-tailer selection (Shop.org 2004). Thestrength of consumers’ associations of a brand (Keller1993; Park, Jaworski, and MacInnis 1986), create one ofthe most valuable assets a firm can own (Aaker 1991),including those in the retailing industry (Ailawadi andKeller 2004). Studies indicate that online brand loyaltyand market share are positively correlated (Danaher,Wilson, and Davis 2003) and strong brands do betteronline than weaker brands (Degeratu, Rangaswamy, andWu 2000). Creating and leveraging a strong brand can bea strategic asset and competitive advantage for onlineretailers. Effective branding can help create experientialvalue that may otherwise be lacking online (Mathwick,Malhotra, and Rigdon 2001).

While the store’s brand name and reputation maydraw customers, product availability and prompt deliveryare also requisite conditions for consummating onlinetransactions (Korper and Ellis 2001; Lancioni, Smith, andOliva 2000). Accurate fulfillment and reliability are keyfactors of consumer e-tail satisfaction and quality percep-tions (Wolfinbarger and Gilly 2003). By managing theirrelations with key suppliers, e-tailers can assure theircustomers quality product selection and provision. Online,brand strength and supplier relations represent organiza-tional resources that can lead to competitive advantagebased on increased site traffic and superior service whichsubsequently lead to superior performance (Day 1994).Such a focus on strategic resources is commensurate withthe resource-based view of the firm (RBV), which positsthat unique assets, such as strong brands and supplierrelations, are difficult for competitors to replicate and candifferentiate a firm (Barney 1991).

Using survey data from a cross-national sample ofmarketing and e-commerce decision-makers, the modelwas tested using structural equation methods. Key infor-mants were marketing or e-commerce decision makersresponsible for the operations of online retailing andmarketing activities. Using Steenkamp and Baumgartner’s(1998) prescribed hierarchical confirmatory factor analy-sis procedure; the multi-item measures were evaluated forcross-national measurement invariance to examine thestrength and generalizability of scales between the U.S.and non-U.S. respondents.


The overall objective of this research was to enhanceunderstanding of the relationships among a series ofhypothesized antecedents of e-retailer firm performanceand to investigate the potential for firm resources tomediate the effects of important firm orientations onrevenue growth and performance relative to objectives(cf., Avlontis and Karayanni 2000). The results providepartial support for the ability of the resource-based vari-ables (i.e., brand strength, supplier relations) to mediatethe effects of the antecedent orientation variables on thetwo performance outcomes. Thus, this research showshow certain unique resources important to e-tailing areestablished and how different types of capabilities andresources relate to one another.

The findings support the resource-based view of thefirm that competitive advantages can be accrued through

the development and deployment of unique and inimi-table assets (Wernerfelt 1984; Barney 1991). Further, ourstudy reinforces the importance of key capabilities, spe-cifically market and foreign knowledge orientations, onresources and ultimately performance (Day 1994; Hultand Ketchen 2001). The results suggest that the impact oforientations on performance was mediated by the re-sources that the orientations help create. That orientationeffects on performance are indirect (see also Noble, Sinha,and Kumar 2002) contributes to our knowledge of theinterrelationships between capabilities, resources, andperformance. Resources such as brand strength and sup-plier relation are thus critical mediators of the orienta-tion – performance relationship. References are availableupon request.

For further information contact:Deborah Colton

Marketing and International BusinessE. Philip Saunders College of Business

Rochester Institute of Technology108 Lomb Memorial Drive

Rochester, NY 14623Phone: 585.475.4939



UNDERSTANDING THE DRIVERS OF THE SUBSIDIARY INNOVATIVE

CAPACITY IN HOST-MARKETS

Esra F. Gencturk, Koc University, TurkeyDestan Kandemir, Bilkent University, Turkey

ABSTRACT

A common theme in innovation research is the notionthat innovation leads to superior performance because itprovides successful innovators with a competitive edge.However, a major void exists in our current understandingbecause of the assumed yet empirically untested competi-tive benefits of innovative capacity. This research exam-ines key attributes of organizational culture, firm struc-ture, and micro- and macro-host country environment asdrivers of a foreign subsidiary’s innovative capacity,which in turn, provides the subsidiary with a competitiveedge and more satisfactory performance. Results from asurvey with 137 informants at 103 multinational corpora-tions (MNC) provide support for the proposed conceptualmodel.

Keywords: Innovative capacity, competitive equity,headquarters-subsidiary relationships, multinational cor-porations.

INTRODUCTION

In today’s increasingly dynamic and competitiveglobal markets, many executives are aware that theircompanies need to innovate and to improve how theymanage innovations. In Booz Allen’s (2004) survey ofsenior executives, nearly half of the respondents said theywere dissatisfied with their companies’ innovation per-formance. The practice of keeping the R&D functionclose to the center of firm’s operations, producing aproduct domestically first and then introducing the sameproduct with minor modifications into other countries nolonger guarantees success. Today, many companies inindustries ranging from fast-moving consumer goods tohigh-tech consumer durables are introducing new prod-ucts simultaneously in many countries. Furthermore, firms’innovative behavior has become increasingly more dis-persed as evidenced by growth and geographical diversityof the R&D investments. For example, Booz Allen’s(2004) European innovation survey estimates that Euro-pean companies will increase their R&D expenditures inEastern Europe and Asia from 6 percent to 31 percent by2007. As Doh et al. (2005) argue, this trend is makinglocation of innovation and its management increasinglyimportant to MNCs. So, the critical managerial question ishow to support and leverage the ability of foreign subsid-iaries to adopt or implement new ideas or products suc-

cessfully, namely how to manage a foreign subsidiary’sinnovative capacity.

The managerial necessity and importance of innova-tion has provided the impetus for studies investigating itsantecedents and performance consequences (e.g., Ambosand Schlegelmilch 2007; Bonner, Ruekert, and Walker2002; Damanpour 1991; Deshpandé, Farley and Webster1993; Deshpandé and Webster 1989; Hurley and Hult1998; Sinkula, Baker, and Noordewier 1997; Slater andNarver 1995). The result is a substantial amount of re-search on innovation across multiple disciplines in bothdomestic and multinational settings. The common de-nominator of this diverse research is the belief that inno-vation leads to superior performance because it providessuccessful innovators with a sustainable competitive edge.The presumed competitive effects of innovation are at-tributed to information asymmetry, high risks, and highcosts that characterize innovative behavior, and createentry and mobility barriers to competitors. If innovationinvolves a deliberate but risky and costly attempt to createa new product with commercial value then the companymust establish a strong competitive advantage to differen-tiate itself before it can enjoy outstanding performance.That is, innovation can result in superior and sustainablesuccess if it first creates an advantage relative to competi-tors. However, in none of the existing empirical studieshave the competitive effects of innovation been directlytested. Hence, despite considerable research accumulatedto-date, a major void in our existing knowledge concernsthe competitive benefits that give innovation its strategicvalue. While this traditionally assumed but empiricallyuntested relationship frames the central research problemof the study, we also argue that to be successful at productinnovations, firms must know when and how to supportinnovative capacity and when to scale back. That is, theattributes of innovative behavior and its importance as asource of competitive advantage also make discretion andcare paramount in managing innovations. As a result, inthis study, we also examine the drivers of innovationaddition to its consequences.

To this end, we integrate and extend the resource-based view of the firm (Barney 1991; Wernerfelt 1984)and the Day and Wensley’s (1988) positional advantageframework to identify the necessary organizational cul-ture, firm structure, and environmental drivers that stimu-late innovative behavior. In so doing, we provide a more


comprehensive framework for understanding both thedrivers and the immediate as well as subsequent conse-quences of innovative capacity. Furthermore, the empiri-cal test of this framework addresses the frequently ex-pressed need for testing the generalizability of these well-established theories and frameworks in the context ofMNCs and their subsidiaries,

CONCEPTUAL DEVELOPMENT

The conceptual framework proposed in this studyidentifies foreign subsidiary’s innovative capacity as acentral construct of interest and examines its contextualdrivers as well as its impact on organizational success.The theoretical justification for the relationships hypoth-esized and subsequently tested is discussed below.

Innovative Capacity

Zaltman, Duncan, and Holbek (1973, p. 2) defineinnovation “as an idea, practice or material artifact per-ceived as new by the relevant unit of adoption.” Firmshave considered innovation as a means of responding toenvironmental changes and uncertainties to increase orsustain their effectiveness and competitiveness(Damanpour and Evan 1984; Thompson 1967). Whileinnovation is associated with success, stability is relatedto failure (Daft 1982). Thus, the purpose of innovation isto contribute to the performance of the firm throughcreating a change in its economic or social environment.

Building on Cohen and Levinthal’s (1990) absorp-tive capacity, Hurley and Hult (1998) consider innovativecapacity as an important building block of innovationconsisting of the abilities to evaluate, assimilate, andapply knowledge from the external environment. Thus,developing and maintaining innovative capacity is criticalto a firm’s competitiveness and success because firmswith higher levels of innovative capacity are assumed tobe better and faster at exploiting the valuable develop-ments in the market. Based on this reasoning, innovativecapacity is generally viewed as an organizational outcomeand has been measured by the number and rate of newproduct introductions (Damanpour and Gopalakrishnan2001; Hurley and Hult 1998; Persaud 2005).

It has been shown that developing and maintaininginnovative capacity requires a number of facilitators at avariety of levels: organizational culture (e.g., Deshpandé,Farley, and Webster 1993; Hurley and Hult 1998; Sinkula,Baker, and Noordewier 1997; Slater and Narver 1995),firm structure (e.g., Ambos and Schlegelmilch 2007;Bonner, Ruekert, and Walker 2002; Damanpour 1991;Deshpandé and Webster 1989), and environmental forces(e.g., Cooper and Brentani 1984; Cooper and Kleinschmidt1993). Based on this literature, our conceptual modelspecifically focuses on a centralization of subsidiary de-

cision making, home office managers´ orientation towardinternational marketing, host-market attractiveness, andhost country environmental similarity as key drivers of asubsidiary’s innovative capacity in the foreign market.

Drivers of Innovative Capacity

Centralization refers to the concentration of decision-making authority in the hands of top management (Aikenand Hage 1968). In an MNC context, centralization wouldimply the extent of foreign subsidiary’s decision-makingautonomy in general, and with respect to its local market-ing activities, in particular (Venaik, Midgley, and Devinney2005). Theoretically, it has been argued that subsidiarieswith low levels of decision-making authority will be notbe motivated to take initiatives in introducing new prod-ucts. That is, centralization makes it difficult for localexperimentation to occur because home country manag-ers, who decide and approve on new product innovations,are often isolated from the subsidiary managers, who arein a position to know what is needed and to deliver theseinnovations. Similarly, MNCs favoring a centralized struc-ture will be careful not to over complicate their operationsand vary their product offerings because variety canundermine efficiency, span of control, and authority ofhome-office managers. Additionally, the established waysof doing business reinforced by centralization encouragestatus quo and hinder the adoption of novel, tailored, andflexible approaches to serving foreign markets. Thesetheoretical arguments have been empirically supported inthe extant literature, where a negative association be-tween centralization and innovation has generally beenobserved in the organizational innovation literature(Damanpour 1991). Similarly, in an MNC context, em-pirical support has been found for the positive effects ofsubsidiary autonomy on the level of innovations(Birkinshaw, Hood, and Jonsson 1999; Ghoshal andBartlett 1988; Venaik, Midgley, and Devinney 2005).Based on this research, we argue that because centraliza-tion will necessitate gaining headquarter’s approval ofnew product initiatives, it may delay speed to market,making the launch less pronounced or even immaterial ifand when it is actually realized. Therefore,

Hypothesis 1: The higher the centralization of for-eign subsidiary’s decision-making, the lower will beits innovative capacity.

Managerial orientation toward international market-ing has generally been viewed as an important componentof organizational culture because it representsheadquarter’s subjective evaluations of international mar-keting activities which result in deeply embedded atti-tudes and perceptions (Zou and Cavusgil 2002). We knowof no study that has investigated the relationship betweeninternational marketing orientation and innovative capac-ity either theoretically or empirically. Accordingly, we


borrow from the market orientation literature, whichattributes a significant role to market orientation in in-creasing a firm’s innovation performance (Deshpandé,Farley, and Webster 1993; Gatignon and Xuereb 1994;Han, Kim, and Srivastava 1998). Drawing on these stud-ies, we expect a subsidiary to have a more innovativecapacity in an MNC with a strong international marketingorientation because home office managers will be morecommitted to and allocate greater resources to interna-tional marketing activities. Unless a subsidiary gets clearsignals from headquarter managers about the importanceof being responsive to local opportunities, the subsidiaryis not likely to invest in developing its innovative capac-ity. Therefore,

Hypothesis 2: The more favorable the home-officemanagerial orientation toward international market-ing, the higher will be the foreign subsidiary’s inno-vative capacity.

Host-market attractiveness refers to the favorabilityof economic conditions of a particular segment/industryin a foreign country. Several new product scholars haveidentified different market characteristics (e.g., marketsize, growth rate, and customer need level) as importantfactors for new product success (Cooper 1980; Cooperand Kleinschmidt 2007; Globe, Levy, and Schwartz 1973;Montoya-Weiss and Calantone 1994). For example, aStanford study conducted by Maidique and Zirger (1985)shows that 14 percent of the time, managers considermarket characteristics as critical for new product success.This literature also considers market attractiveness as akey dimension in new product screening and evaluation(Cooper and Brentani 1984; Cooper and Kleinschmidt1993). Additionally, several studies have provided em-pirical support for the positive effect of market attractive-ness on a firm’s resource allocations to new products(Abell and Hammond 1979; Luck and Ferrell 1985).Apart from new product literature, research on MNCs hasidentified economic factors as drivers of innovation mea-sured by foreign R&D investments (Doh et al. 2005).Building on the implications of these two research streams,we postulate that existence of many potential and profit-able customers, who are open to try new products inattractive host markets, will encourage the subsidiary toengage in a greater number and a higher rate of newproduct introductions to exploit such attractive marketopportunities. Hence;

Hypothesis 3: The greater the perceived attractive-ness of the host-market, the higher will be the foreignsubsidiary’s innovative capacity.

Host country similarity, which refers to the similaritybetween the home- and host-country macro-environmen-tal characteristics (i.e., socio-cultural), is a prominent

explanatory variable of subsidiary behavior in both inter-national business and international marketing literatures.This perceived similarity between the socio-culturalcharacteristics (i.e., language, culture, and business prac-tices) of countries has been termed psychic distance(Johanson and Widersheim-Paul 1975). Previous researchon standardization-adaptation (Jain 1989; Szymanski,Bharadwaj, and Varadarajan 1993) as well as diffusion ofinnovations (Ganesh, Kumar, and Subramaniam 1997;Hagerstrand 1967; Rogers 1983) have shown that firmsoperating in markets that are socio-culturally similar orwith shorter psychic distance are likely to adopt similarproducts (Levitt 1983; Sorenson and Weichmann 1975)because of ready transferability of knowhow betweensuch markets. The perceived host country similarity wouldallow home country managers with accumulated experi-ence to easily transfer their knowledge to host marketswith socio-economic characteristics, and hence bettergauge the strengths of certain product features and decidewhich customer problems will be more advantageous tosolve. On the other hand, since home country managershave relatively little established knowledge or past expe-rience to draw upon, culturally dissimilar countries willprevent exploitation of existing knowledge base. Instead,they will demand experimentation, speculative invest-ments, and possible failure; all of which entail high costand risk. As a result, we expect host market dissimilarityto temper down the innovative behavior. Thus;

Hypothesis 4: The greater the perceived host countrysimilarity between home- and host-country, the higherwill be the foreign subsidiary’s innovative capacityin a foreign market.

Consequences of Innovative Capacity

The link between innovative capacity and firm per-formance has been well-documented in the marketingliterature across various definitions and operationalizationsof organizational success (Hult and Ketchen 2001; Noble,Sinha, and Kumar 2002; Siguaw, Simpson, and Enz2006). For example, Ittner and Larcker (1997) observe asignificant positive relationship between innovation andreturn on assets and growth. Likewise, Han, Kim, andSrivastava (1998) report that technical and administrativeinnovations are positively related to organizational per-formance measured in terms of growth, profitability, andreturn on assets. Based on this extensive literature, twocritical manifestations of organizational success are usedin this paper.

First, called competitive equity, is adapted fromSchatzel and Calantone (2006), and refers to relativecompetitive position of the company and its productswithin the served market. In this context, we are explicitlyacknowledging and testing the competitive advantages


that may stem from innovative capacity by consideringthe relative reputation of the subsidiary as well as therelative quality of and loyalty for its products. The secondmanifestation of organizational success considered in thispaper captures the satisfaction of home office managerswith the operational performance of its foreign subsidiarymeasured in terms of sales, sales growth, market share,and profits. The theoretical rationale behind consideringinnovative capacity as a determinant of competitive eq-uity and satisfaction with operational performance is thatnew products may enable the subsidiary to cope with thechanges in the foreign market, satisfy customer needsbetter than the existing goods by continuously offeringgreater benefits, and prevent competitors from capturingthe subsidiary’s customers with their own products(Szymanski, Bharadwaj, and Varadarajan 1993). Giventhis theoretically grounded empirical support in the litera-ture, we expect the subsidiary’s innovative capacity to bepositively linked to its competitive equity and the satisfac-tion of corporate management with the subsidiary´s op-erational results. Thus;

Hypothesis 5: The higher the foreign subsidiary’sinnovative capacity;

(a) the greater will be its competitive equity.

(b) the greater will be the home office manag-ers’ satisfaction with its operational results.

In addition to postulating a direct relationship be-tween subsidiary’s innovative capacity and the two com-ponents of organizational success, we expect a sequentialordering between these two components, whereby com-petitive equity would contribute to home office managers’satisfaction with the subsidiary’s operational performance.Competitive equity capturing the relative reputational andquality position of the subsidiary as well as its productsshould have a positive impact on the subsidiary’s opera-tional performance in terms of sales, market share, andprofits. For example, the importance of product quality inachieving satisfactory operational results has been high-lighted by several marketing researchers (Day 1994; Huntand Morgan 1995; Song and Parry 1997). Specifically,customers’ evaluations of product quality and reputationhave been shown to contribute to their purchase decisions(Spreng, MacKenzie, and Olshavsky 1996), positivelyeffecting sales and market share (Tatikonda and Montoya-Weiss 2001). Hartman and Teece (1990) also find marketshare to be positively influenced by brand reputation.Thus;

Hypothesis 6: The higher the foreign subsidiary’scompetitive equity, the greater will be the homeoffice managers’ satisfaction with its operationalresults.

RESEARCH METHOD

Sample and Data Collection

Consistent with the recommendations of past studies(e.g., Myers and Harvey 2001; Zou and Cavusgil 2002),our unit of analysis was at the product-market level of afirm’s operations in a single product or product line soldin a specific foreign country. The data collection wasaccomplished in three stages. In the first stage, the presi-dents/CEOs of 350 manufacturing selected from a statedirectory of firms with international operations werecontacted through mail to request their participation in thestudy as well as to obtain the names of managers respon-sible for the respective firms’ international marketingoperations. Service firms as well as those in primaryindustries (i.e., metals) were excluded because of ob-served differences in their international expansion andmarketing strategies (e.g., Erramilli 1991) as well as ourtheoretical interest in adoption of product (vs. service)innovations. Of the 229 firms that responded to the intro-ductory letter, 163 firms that agreed to participate in thisstudy identified 252 informants representing 227 divi-sions.

In the second stage, 27 informants from 18 differentfirms were used to pretest the format and content of thesurvey instrument. No difficulties were encountered andhence no major changes were made in the measures or thequestionnaire used. In the third stage of data collection,questionnaires along with a cover letter were mailed to theremaining 225 informants at the 132 divisions across the103 firms. To facilitate development of valid measuresand control for extraneous sources of variation, respon-dents were asked to self-select a product unit and oneforeign market served by this unit with which they weremost familiar with, and answer all questions with respectto the chosen foreign market. After a postcard reminderand two follow-up letters, completed useable question-naires were received from 137 informants at 103 firms, fora response rate of 61 percent.

Non-response bias was assessed by comparing earlyand late respondents on key variables. To assess non-response bias, we divided our data into two groups basedon the date on which we received the completed surveys.Based on a comparison of the averages of annual salesvolume, years of exporting, and the number of employees,t-tests between the mean responses of the early and the laterespondents indicated no statistically significant differ-ences at the .05 level (Armstrong and Overton 1977).Additionally, using secondary data on employees andannual sales volume available in the sampling frame, wecompared the 103 responding firms with the 76 nonpar-ticipant firms as well as with 42 nonresponding firms. Thet-tests revealed no significant differences between re-


spondents and the two groups of nonresponding firms.Accordingly, non-response bias was not considered to bea significant problem in the data.

The survey respondents held upper-managementpositions (President/CEO, Director, Vice-President, AreaManager/General Manager), had on average 13.02 yearsof experience working with the current firm and 6.23years of experience in making decisions about the firm’smarketing activities in the foreign country they had se-lected. These characteristics suggest that the respondentsare likely to be familiar with their firm’s foreign opera-tions, knowledgeable about the questions asked in thesurvey, and competent to serve as a key informant for theparticular foreign venture of their respective firms.

Measures

Several considerations guided the development ofthe scales to operationalize each of the constructs dis-cussed in our conceptual model (Churchill 1979; Nunnally1978). Based on the constitutive definition of innovativecapacity and its antecedents as adopted in this study, aninitial list of items was generated through an exhaustivereview of the literature. Then, in view of the extensivelyreported concerns about the adequacy of single indicators,multiple item scales were used to measure each constructof interest. Finally, the order of the scales was randomizedand some of the items were negatively stated in order tominimize position and response set bias. Table 1 providesa complete list of scale items for all the measures used inthis study.

The centralization scale assessed the degree to which(1) any changes in the international marketing activities ofthe foreign subsidiary must first be cleared with thecorporate management; (2) corporate management ap-proval is required for all subsidiary marketing activities;and (3) home-office managers of the subsidiary have thefinal say in marketing decisions (Davidson 1984). Themanagerial orientation toward international marketingscale was adapted from the measures used by Cavusgiland Nevin (1981) and Cavusgil (1984) and assessed thefavorability of home-management’s subjective evalua-tions of international marketing with a 4-item scale. Bothscales were measured by a 6-point Likert scale rangingfrom “Definitely Agree” to “Definitely Disagree.”

The host-market attractiveness scale, representingthe perceived potential for the host market to contribute toa business’ foreign marketing objectives, was adaptedfrom the measures developed by Burke (1984). The mea-sure of host-market attractiveness involved a 5-item 7-point Likert scale ranging from “Low” to “High.” Thehost country similarity scale assessed the degree to whichthe foreign country in question was perceived as beingdifferent from the home country. The measurement of

environmental similarity was accomplished by develop-ing a 6-item 6-point Likert scale ranging from “VerySimilar” to “Very Different.”

The operationalization of subsidiary’s innovativecapacity was intended to measure the number as well asthe rate of new product introductions in the host-marketserved. Contrary to past research, in this study, we as-sessed innovative capacity not only relative to home-country but also relative to competitors in the host-coun-try. While the relative to home country measure includeda 3-item 6-point Likert scale ranging from “Much Lowerthan U.S.” to “Much Higher than U.S.,” the relative tohost-country competitors, measure consisted of a 3-item6-point Likert scale ranging from “Much Lower thanMajor Competitors” to “Much Higher than Major Com-petitors.”

To explore the relationship between the subsidiary´sinnovative capacity in a foreign market and its organiza-tional success in that market, we used two outcomes: (1)competitive equity and (2) satisfaction with operationalresults. Competitive equity assessed the degree to whichthe relative positional advantages such as brand loyaltyand reputation were attained by a subsidiary in the host-market served. It was measured by a 4-item 6-point scaleranging from “Definitely Agree” to “Definitely Disagree.”The second assessment of performance measured theextent to which home-market managers were satisfiedwith the operational results (e.g., sales, sales growth,market share, and profits) of the foreign subsidiary. Thesatisfaction scale was measured by a 5-item 6-point scaleranging from “Extremely Dissatisfied” to “ExtremelySatisfied.”

Control Variables. To account for the effects ofextraneous variables, a business’ international experienceand international involvement were included as controlvariables. We used the number of majority owned foreignsales subsidiaries to represent international experience.We measured international involvement by the number ofits employees residing in the foreign country as a percent-age of its total (i.e., worldwide) workforce.

ANALYSIS AND RESULTS

The Measurement Model

We evaluated the psychometric properties of ourmeasures using a confirmatory factor analysis (CFA) thatcombined each factor measured by reflective scales(Bagozzi, Yi, and Philips 1991; Gerbing and Anderson1988). The CFA was fitted using the maximum likelihoodestimation procedure with the raw data as input in EQS 6.1(Bentler 1995). After we dropped some items that had lowfactor loadings or high cross loadings, the confirmatorymodel fit the data satisfactorily.


TABLE 1

Results of the CFA

Standardized

Scale Items Loading t-valuea

Centralization

AVE = 49.7%; HSV = 11%; CR = .70When an unusual marketing problem is encountered, it is handled without checking .80 8.81

with the home-office managers of this product unit. rc

When a marketing decision is to be made for which rules and procedures do not exist, .74 8.17the decision is made without referring to the home-office managers of thisproduct unit. rc

When a marketing problem is encountered, the home-office managers of this product .55 6.11unit are contacted for an answer.

When a marketing decision is to be made, the recommendations of home officemanagers of this product unit are followed.*

Home office managers of this product unit have the final say in marketing decisions.*

Managerial Orientation Toward International Marketing

AVE = 44.2%; HSV = 11%; CR = .70Consider domestic marketing activities to be more important than international .67 7.19

marketing activities.rc

Actively explore international opportunities. .61 6.55Frequently travel abroad. .71 7.57Have no intention of increasing the unit’s international marketing activities.*

Host-Market Attractiveness

AVE = 67.6%; HSV = 6%; CR = .90Prospects for future profits. .57 6.99Average industry gross margin. .99 14.33Average industry pretax profits. .85 11.49Short-term market growth rate.*

Long-term market growth rate.*

Perceived Host Country Similarity

AVE = 68.1%; HSV = 5%; CR = .90Cultural values. .83 11.03Distribution infrastructure. .70 8.86Norms of business conduct. .93 12.98

Communication infrastructure.*Predominant language.*Predominant religion.*

Subsidiary’s Innovative Capacity Relative to Host-Country Competitors

AVE = 90.3%; HSV = 22%; CR = .90Number of new product introductions. .93 13.32Rate of new product introductions. .97 14.43Number of product changes in your unit.*


TABLE 1 (CONTINUED)

Results of the CFA

Standardized

Scale Items Loading t-valuea

Subsidiary’s Innovative Capacity Relative to Home-Country

AVE = 76.9%; HSV = 7%; CR =.90

Number of new product introductions. .98 9.69Rate of new product introductions. .76 7.92

Number of product changes in your unit.*

Competitive Equity

AVE = 74%; HSV = 24%; CR =.90Our products have a better overall reputation. .81 11.09We have higher quality products. .81 11.07We have a better company reputation. .95 13.94There is higher brand loyalty for our products.*

Satisfaction with Operational Results

AVE = 72.9%; HSV = 24%; CR =.90Sales of your product unit. .88 12.41Sales growth rates of your product unit. .81 11.04Market share of your product unit. .87 12.16

Profit margins of your product unit.*

Operating profits of your product unit.*

Model Fit Statistics: χ2 = 315.15 (df = 181, p < .05)NNFI = .89CFI = .92IFI = .92RMSEA = .07

aThe t-values from the unstandardized solution; *Deleted items; rc Reverse coded.Notes: AVE = Average variance extracted; HSV = Highest shared variance with other constructs; CR = Compositereliability.

We assessed the convergent and discriminant valid-ity of the focal constructs by estimating an 8-factor con-firmatory measurement model. Each measurement itemloaded only on its latent construct. The model showed achi-square (χ2) value of 315.50 (df = 181) having a p valueless than .05, indicating significance of the χ2 test. Therehas been considerable discussion in the Structural Equa-tion Modeling (SEM) literature concerning the validity ofthe χ2 test as an index of model fit, especially when thesample size is small as in our study. As a result, a numberof adjunct fit indices have been proposed to reflect theimprovement in fit of a specified model, in which all

parameters are fixed to zero. Accordingly, the modelprovides a satisfactory fit to the data (Bentler-Bonettnonnormed fit index [NNFI] = .89; comparative fit index[CFI] = .92; Bollen’s fit index [IFI] = .92; root meansquare error of approximation [RMSEA] = .07), indicat-ing the unidimensionality of the measures (Anderson andGerbing 1988) (Table 1). Furthermore, all the factorloadings were statistically significant (p < .01), and thecomposite reliabilities of all constructs exceeded the thresh-old value of .70 (Nunnally 1978). Thus, the measuresdemonstrate adequate convergent validity and reliability.


Discriminant validity was examined by calculatingthe shared variance between all possible pairs of con-structs verifying that they were lower than the averagevariance extracted for the individual constructs (Fornelland Larcker 1981). The highest level of shared variancebetween any pair of constructs was 24 percent. Theaverage variances extracted ranged between 90.3 and44.2 percent. These results showed that the average vari-ance extracted by the measure of each factor was largerthan the squared correlation of that factor’s measure withall measures of other factors in the model (see Table 1).Given these values, we concluded that all the factors in themeasurement model possess strong discriminant validity.In light of this evaluation, we are able to conclude that allfactors in the measurement model possessed both conver-gent and discriminant validity, and that the CFA model fitthe data adequately. Table 1 presents key results of theCFA.

Hypothesis Testing

The hypothesized model was estimated by usingSEM, with the EQS 6.1 program. Although the chi-squaretest was statistically significant (χ2

(218)= 357.86, p < .05),

the Bentler-Bonett nonnormed fit index [NNFI = .89]; thecomparative fit index [CFI = .91]; the Bollen’s fit index[IFI = .91]; the root mean square error of approximation[RMSEA = .07] indicated a good fit with the hypothesizedstructural model (see Figure 1).

Centralization was found to be negatively associatedwith the foreign subsidiary’s innovative capacity relativeto host-country competitors (β = -.29; p ≤ .05). However,the relationship between centralization and the subsidiaryinnovative capacity relative to home country was positivebut not significant (β = .16; p ≥ .10). Managerial orienta-tion toward international marketing was found to havepositive effects on subsidiary’s innovative capacity rela-tive to host-country competitors (β = .44; p ≤ .01) andrelative to home country (β = .24; p ≤ .10). Similarly, themore attractive the host market the higher the subsidiary’sinnovative capacity relative to host country competitors(β = .28; p ≤ .01) and relative to its home country (β = .15;p ≤ .10). In terms of host-country environmental similar-ity, the higher degrees of perceived similarity has apositive and significant effect on the subsidiary’s innova-tive capacity relative to host country competitors (β = .27;p ≤ .01). In contrast, subsidiary’s perceived environmen-tal similarity had a negative effect on the subsidiary’sinnovative capacity relative to home country (β = -1.78;p ≤ .10).

In examining the consequences of innovative capac-ity, we found that the subsidiary’s innovative capacityrelative to host country competitors had positive effectson both its competitive equity (β = .42; p ≤ .01) and home

country managers’ satisfaction with its operational results(β = .27; p ≤ .01). On the other hand, the subsidiary´sinnovative capacity relative to home country did notappear to be significantly related to brand advantage (β =-.01; p ≥ .10). However, we found the subsidiary´s inno-vative capacity relative to home country to be signifi-cantly related to its satisfaction (β = .21; p ≤ .05). Further-more, competitive equity had a positive effect on homeoffices managers’ satisfaction with subsidiary’s opera-tional results (β = .40; p ≤ .01).

The level of home office managers’ internationalexperience and international involvement were includedas control variables. The results indicated that only inter-national experience was significantly, negatively relatedto subsidiary’s innovative capacity relative to host-countycompetitors (β = -.21; t-value = -1.95; p .10). In addition,neither of the control variables did have significant effectson subsidiary’s innovative capacity relative to home coun-try. Furthermore, we found that only international in-volvement was significantly, negatively related to com-petitive equity (β = -.15; t-value = -1.77; p .10). Finally,neither international experience nor international involve-ment did appear to be significantly related to home officemanagers’ satisfaction with operational results.

DISCUSSION

Much of the existing research had assumed the com-petitive benefits of innovation considering it an antidoteto being indistinguishable from competitors, and thus acure for operating with paper thin profit margins. In thisstudy, we provide an explicit and more precise test of thisassumption than previously offered in the literature. Inparticular, we propose, test, and find empirical support forcompetitive equity as an important outcome of innovativecapacity in addition to the traditional operational metricssuch as sales and profits considered in the extant research.We also provide a slightly different but in our opinionmuch more relevant measurement of operational perfor-mance by considering the frequently used metrics in termsof home market manager’s satisfaction with the opera-tional performance of its foreign subsidiary.

Furthermore, our study represents one of the firstattempts we know of operationalizing subsidiary innova-tive capacity not only in reference to that of the homecountry but also in relation to the competitors in a givenhost-market. The results support the importance of thisdistinction since all the hypothesized drivers have a morepronounced relationship with the subsidiary’s innovativecapacity relative to its competitors in the host market. Ourstudy has focused on a representative range of factorsdriving a subsidiary’s innovative capacity, with particularattention paid to organizational culture, firm structure andmicro as well as macro host country environmental forces.

132A

merican M


inter 2008

FIGURE 1

Results of Hypothesis Testing

***p d” .01; **p d” .05; *p d” .1.Notes: t-values are in parentheses


Consistent with the findings of previous research, theresults support the general view that both internal homemarket and external host country factors play an importantrole in determining the innovative capacity of foreignsubsidiaries. Yet, our study clearly highlights the fact thatthe role played by these drivers is especially critical whenthe MNC wants and or needs to support innovative capac-ity of its subsidiary relative to its host-country competi-tors. Under these circumstances, a decentralized structurealong with managers’ favorable attitudes toward interna-tional activities create an internal climate conducive toenhancing the number and rate of new product introduc-tions by the subsidiary to a level beyond that of itscompetitors. Similarly, subsidiaries operating in attrac-tive foreign markets as well as in countries perceived to besimilar to their home market are likely to enjoy a greaterinnovation capacity than their competitors.

Additionally, and more critically, while a subsidiary’sinnovative capacity relative to its competitors contributessignificantly to the subsidiary’s competitive standing ascaptured by its competitive equity, its relative to homeinnovative capacity has no competitive effect whatsoever.That is, a key finding of our study is that the historicallypresumed but untested competitive benefits of innovativecapacity will be realized only when the foreign subsidiaryachieves an innovative capacity that is superior to itscompetitors in the host-market. Apparently, foreign sub-sidiaries with a higher innovative capacity than that oftheir home market in terms of either the number or the rateof new product launches do not necessarily enjoy thepresumed competitive advantages of innovation in themarkets they are serving.

Limitations and Directions for Future Research

This study’s contributions as well as limitations alsosuggest directions for future research on this importanttopic.

First, the findings of this study unequivocally showthat future research should consider more than the homecountry operations as a reference point and include com-petitors in the host market as well in assessing the innova-tive capacity of foreign subsidiaries.

Second, this study also illustrates the relevance andimportance of assessing competitive benefits of innova-tive capacity directly and explicitly. While our attempt totest this presumed relationship focuses on competitiveequity capturing the relative reputation of the subsidiaryas well as the relative quality of and loyalty for itsproducts, future research may build upon and extend boththe conceptualization and operationalization of this con-struct.

Additionally, in extending the implication of diverseresearch on innovations and innovative capacity into therealm of foreign subsidiaries, our study both reinforcesthe results of previous studies and provides additionalempirical evidence in support of the importance of bothinternal organizational and external host country climatein fostering innovative capacity. While the drivers ofinnovative capacity considered in this study are represen-tative of the key and empirically supported considerationsobserved in the extant literature, both the depth and breathof explanatory variables can be expanded in future re-search. For example, since host countries benefit from theexternal spillovers from the innovative capacity of theforeign subsidiaries operating within their national bor-ders, host government policy and national innovationefforts are becoming relevant factors to consider in futureresearch on this topic.

Finally, while we used a multi-industry sample toincrease observed variance and to strengthen thegeneralizability of the findings, future studies shouldexpand the scope of the proposed model beyond manufac-turing firms and consider adoption of service innovationsin foreign markets.

REFERENCES

Abell, Derek F. and John S. Hammond (1979), StrategicMarket Planning. Englewood Cliffs, NJ: Prentice-Hall.

Aiken, Michael and Jerald Hage (1968), “OrganizationalInterdependence and Intraorganizational Structure,”American Sociological Review, 33 (December), 912–30.

Ambos, Björn and Bodo B. Schlegelmilch (2007), “Inno-vation and Control in the Multinational Firm: AComparison of Political and Contingency Ap-proaches,” Strategic Management Journal, 28 (5),473–86.

Anderson, J.C. and D.W. Gerbing (1988), “StructuralEquation Modeling in Practice: A Review and Rec-ommended Two-Step Approach,” PsychologicalBulletin, 103 (3), 411–23.

Armstrong, J. Scott and Terry S. Overton (1977), “Esti-mating Nonresponse Bias in Mail Surveys,” Journalof Marketing Research, 14 (3), 396–402.

Bagozzi, Richard P., Youjae Yi, and Lynn W. Phillips(1991), “Assessing Construct Validity in Organiza-tional Research,” Administrative Science Quarterly,36 (September), 421–58.

Barney, Jay (1991), “Firm Resources and Sustained Com-petitive Advantage,” Journal of Management, 17 (1),99–120.


Bentler, Peter M. (1995), EQS Structural Equations Pro-gram Manual. Encino, CA: Multivariate Software,Inc.

Birkinshaw, Julian, Neil Hood, and Stefan Jonsson (1999),“Building Firm Specific Advantages in Multina-tional Corporations,” Journal of International Busi-ness Studies, 10 (4), 221–42.

Bonner, Joseph M., Robert W. Ruekert, and Orville C.Walker, Jr. (2002), “Upper Management Control ofNew Product Development Projects and Project Per-formance,” The Journal of Product Innovation andManagement, 19, 233–45.

Burke, Marian C. (1984), “Strategic Choice and Market-ing Managers: An Examination of Business-LevelMarketing Managers,” Journal of Marketing Re-search, 21 (November), 345–59.

Cavusgil, S.Tamer and John R. Nevin (1981), “InternalDeterminants of Export Marketing Behavior: AnEmpirical Investigation,” Journal of Marketing Re-search, 18 (1), 114–19.

____________ (1984), “Differences Among ExportingFirms Based on Their Degree of Internationaliza-tion,” Journal of Business Research, 12 (2), 195–208.

Churchill, Jr., Gilbert A. (1979), “A Paradigm for Devel-oping Better Measures of Marketing Constructs,”Journal of Marketing Research, 16 (1), 64–73.

Cohen, Wesley M. and Daniel A. Levinthal (1990), “Ab-sorptive Capacity: A New Perspective on Learningand Innovation,” Administrative Science Quarterly,35 (1), 128–52.

Cooper, Robert G. (1980), “Project NewProd: Factors inNew Product Success,” European Journal of Mar-keting, 14, 277–92.

____________ and U. de Brentani (1984), “Criteria forScreening New Industrial Products,” Industrial Mar-keting Management, 13, 149–56.

____________ and Elko J. Kleinschmidt (1993), “MajorNew Products: What Distinguishes the Winners inthe Chemical Industry?” The Journal of ProductInnovation and Management, 10, 90–111.

____________ and ____________ (2007), “WinningBusinesses in Product Development: The CriticalSuccess Factors,” Research Technology Manage-ment, 50 (3), 52–66.

Daft, Richard (1982), “Bureaucratic versus Nonbureau-cratic Structure and the Process of Innovation andChange,” Research in the Sociology of Organiza-tions, 1, 129–66.

Damanpour, Fariborz and William M. Evan (1984), “Or-ganizational Innovation and Performance: The Prob-lem of Organizational Lack,” Administrative ScienceQuarterly, 29, 392–409.

____________ (1991), “Organizational Innovation: AMeta Analysis of Effects of Determinants and Mod-erators,” Academy of Management Journal, 34 (3),

555–90.____________ and Shanthi Gopalakrishnan (2001),

“Theories of Organizational Structure and Innova-tion Adoption: The Role of Environmental Change,”Journal of Engineering and Technology Manage-ment, 15, 1–24.

Davidson, William H. (1984), “Administrative Orienta-tion and International Performance,” Journal of In-ternational Business Studies, 15 (Fall), 11–23.

____________ (1994), “The Capabilities of Market-Driven Organizations,” Journal of Marketing, 58 (4),37–52.

Day, George S. and Robin Wensley (1988), “AssessingAdvantage: A Framework for Diagnosing Competi-tive Superiority,” Journal of Marketing, 52 (2), 1–20.

Deshpandé, Rohit and Frederick E. Webster, Jr. (1989),“Organizational Culture and Marketing: Definingthe Research Agenda,” Journal of Marketing, 53 (1),3–15.

____________, John U. Farley, and Frederick E. Webster,Jr. (1993), “Corporate Culture, Customer Orienta-tion, and Innovativeness in Japanese Firms: A QuadradAnalysis,” Journal of Marketing, 57 (1), 23–37.

Doh, Jonathan P., Gary K. Jones, Ram Mudambi, andHildy Teegen (2005), “Foreign Research and Devel-opment and Host Country Environment: An Empiri-cal Examination of U.S. International R&D,” Man-agement International Review, 45 (Special Issue),121–54.

Erramilli, M.K. (1991), “The Experience Factor in For-eign Market Entry Behaviour of Service Firms,”Journal of International Business Studies, 22 (3),479–501.

Fornell, Claes and David F. Larcker (1981), “StructuralEquation Models with Unobservable Variables andMeasurement Error: Algebra and Statistics,” Journalof Marketing Research, 18 (3), 382–88.

Ganesh, Jaishankar, V. Kumar, and Velavan Subramaniam(1997), “Learning Effect in Multinational Diffusionof Consumer Durables: An Exploratory Investiga-tion,” Journal of the Academy of Marketing Science,25 (3), 214–28.

Gerbing, David W. and James C. Anderson (1988), “AnUpdated Paradigm for Scale Development Incorpo-rating Unidimensionality and its Assessment,” Jour-nal of Marketing Research, 25 (2), 186–92.

Gatignon, Hubert and Jean-Marc Xuereb (1994), “Strate-gic Orientation of the Firm and New Product Perfor-mance,” Journal of Marketing Research, 34 (Febru-ary), 77–90.

Ghoshal, Sumantra and Christopher A. Bartlett (1988),“Creation, Adoption, and Diffusion of Innovationsby Subsidiaries by Multinational Corporations,” Jour-nal of International Business Studies, 19 (3), 365–88.

Globe, Samuel, Girard W. Levy, and Charles M. Schwartz(1973), “Key Factors and Events in the Innovation


Process,” Research Management, 16 (July), 8–15.Hagerstrand, T. (1967), Innovation Diffusion as a Spatial

Process. Chicago: University Press.Han, Jin K., Namwoon Kim, and Rajendra K. Srivastava

(1998), “Market Orientation and Organizational Per-formance: Is Innovation a Missing Link?” Journal ofMarketing, 62 (4), 30–45.

Hartman, Raymond S. and David J. Teece (1990), “Prod-uct Emulation Strategies in the Presence of Reputa-tion Effects and Network Externalities: Some Evi-dence from the Minicomputer Industry,” Economicsof Innovation and New Technology, 1, 157–82.

Hult, G. Tomas M. and David J. Ketchen (2001), “DoesMarket Orientation Matter? A Test of RelationshipBetween Positional Advantage and Performance,”Strategic Management Journal, 22 (9), 899–906.

Hunt, Shelby D. and Robert M. Morgan (1995), “TheComparative Advantage Theory of Competition,”Journal of Marketing, 59 (April), 1–15.

Hurley, Robert F. and G. Tomas M. Hult (1998), “Innova-tion, Market Orientation, and Organizational Learn-ing: An Integration and Empirical Examination,”Journal of Marketing, 62 (3), 42–54.

Ittner, Christopher D. and David F. Larcker (1997), “Prod-uct Development Cycle Time and OrganizationalPerformance,” Journal of Marketing Research, 34(1), 13–23.

Jain, Subhash C. (1989), “Standardization of Interna-tional Marketing Strategy: Some Research Hypoth-eses,” Journal of Marketing, 53 (January), 70–79.

Johanson, Jan and Finn Wiedersheim-Paul (1975), “TheInternationalization of the Firm: Four Swedish CaseStudies,” Journal of Management Studies, (Octo-ber), 305–22.

Levitt, T. (1983), “The Globalization of Markets,” HarvardBusiness Review, (May/June), 92–102.

Luck, David J. and O.C. Ferrell (1985), Marketing Strat-egy and Plans. New Jersey: Prentice Hall.

Maidique, M.A. and B.J. Zirger (1985), “The New Prod-uct Learning Cycle,” Research Report Series, Inno-vation & Entrepreneurship Institute, School of Busi-ness Administration, University of Miami, CoralGables, FL, (February), 85–101.

Montoya-Weiss, Mitzi M. and Roger Calantone (1994),“Determinants of New Product Performance: A Re-view and Meta-Analysis,” The Journal of ProductInnovation and Management, 11, 397–417.

Myers, Matthew B. and Michael Harvey (2001), “TheValue of Pricing Control in Export Channels: AGovernance Perspective,” Journal of InternationalMarketing, 9 (4), 1–29.

Noble, Charles H., Rajiv K. Sinha ve Ajith Kumar (2002),“Market Orientation and Alternative Strategic Orien-tations: A Longitudinal Assessment of PerformanceImplications,” Journal of Marketing, 66 (October),25–39.

Nunnally, Jum C. (1978), Psychometric Theory. NewYork: McGraw-Hill.

Persaud, Ajax (2005), “Enhancing Synergistic Innova-tive Capability in Multinational Corporations: AnEmpirical Investigation,” The Journal of ProductInnovation and Management, 22, 412–29.

Rogers, E.M. (1983), Diffusion of Innovations. New York:Free Press.

Schatzel, Kim and Roger Calantone (2006), “CreatingMarket Anticipation: An Exploratory Examinationof the Effect of Preannouncement Behavior on a NewProduct’s Launch,” Journal of the Academy of Mar-keting Science, 34 (3), 357–66.

Siguaw, Judy A., Penny M. Simpson, and Cathy A. Enz(2006), “Conceptualizing Innovation Orientation: AFramework for Study and Integration of InnovationResearch,” The Journal of Product Innovation andManagement, 23, 556–74.

Sinkula, James M., William E. Baker, and ThomasNoordewier (1997), “A Framework for Market-BasedOrganizational Learning: Linking Values, Knowl-edge, and Behavior,” Journal of the Academy ofMarketing Science, 25 (4), 305–18.

Slater, Stanley F. and John C. Narver (1995), “MarketOrientation and the Learning Organization,” Journalof Marketing, 59 (3), 63–74.

Song, X. Michael and Mark E. Parry (1997), “The Deter-minant of Japanese New Product Successes,” Jour-nal of Marketing Research, 34 (February), 64–76.

Sorenson, R. and U.E. Weichmann (1975), “How Multi-nationals View Marketing Standardization,” HarvardBusiness Review, (May/June), 38–167.

Spreng, R.A., S.C. MacKenzie, R.W. Olshavsky (1996),“A Reexamination of the Determinants of ConsumerSatisfaction,” Journal of Marketing, 60 (3), 15–32.

Szymanski, David M., Sundar G. Bharadwaj, and P.Rajan Varadarajan (1993), “An Analysis of the Mar-ket Share – Profitability Relationship,” Journal ofMarketing, 57 (3), 1–18.

Tatikonda, Mohan V. and Mitzi M. Montoya-Weiss (2001),“Integrating Operations and Marketing Perspectivesof Product Innovation: The Influence of Organiza-tional Process Factors and Capabilities on Develop-ment Performance,” Management Science, 47 (1),151–72.

Thompson, J.D. (1967), Organizations in Action. NewYork: McGraw Hill.

Venaik, Sunil, David F. Midgley, and Timothy M.Devinney (2005), “Dual Paths to Performance: TheImpact of Global Pressure on MNC Subsidiary Con-duct and Performance,” Journal of InternationalBusiness Studies, 36, 655–75.

Wernerfelt, Birger (1984), “A Resource-Based View ofthe Firm,” Strategic Management Journal, 5 (2),171–80.

Zaltman, Gerald, Robert Duncan, and Jonny Holbek


(1973), Innovations and Organizations. New York:John Wiley and Sons.

Zou, S. and Tamer S. Cavusgil (2002), “The GMS: A

Broad Conceptualization of Global Marketing Strat-egy and its Effect on Firm Performance,” Journal ofMarketing, 66 (4), 40–56.

For further information contact:Esra F. Gencturk

Department of MarketingKoç University

Rumelifeneri Yolu 34450 Sar1yerTurkey

Phone: +90.212.338.1000E-Mail: [email protected]


TOWARD AN ENRICHED ORGANIZATIONAL ECOLOGY

FRAMEWORK OF FOREIGN MARKET ENTRY:

THE EXPLANATION FOR RETAILERS’

INTERNATIONAL EXPANSION

Chuanyi Tang, University of Arizona, TucsonEric J. Arnould, University of Arizona, Tucson

SUMMARY

Foreign market entry is an important issue in interna-tional marketing. In recent years, the dominant researchstream based on transaction cost analysis or efficiencyreceived criticism; organizational ecology theory has beenproposed as an alternative explanation. Organization ecol-ogy theory complements the traditional research perspec-tive by conducting population level analysis and empha-sizing the effects of resource constrains, inter-organiza-tional competition, and temporal disequilibrium on orga-nizational viability. However, organization ecology theoryitself has some limitations. This paper intends to addressthese limitations and establish an enriched organizationalecology framework for organizations’ foreign marketentry. To develop the theoretical framework, interna-tional retailing as an under-researched field is employedas the research context, and four main dimensions oforganizational ecology are examined.

Carrying Capacity and Demand Dynamics

According to organizational ecology theory, carry-ing capacity influences the survival of organizations(Hannan and Freeman 1989). However, most researchfocuses only on the effects of carrying capacity size, whiledynamic change in carrying capacity is seldom consid-ered. We propose that the number of foreign retail entriesinto a foreign market is positively relate to not only thehost country market demand (size of carrying capacity)but also the growth rate of the host country market demand(growth rate of carrying capacity).

Density Dependence and Population Dynamics

Most ecological studies on foreign market entryfocus on the analysis of population density or organiza-tions’ prior entry and exit in a host market. However,density dependence theory has two major problems. Oneis that it simply counts the total number of firms anddoesn’t take into account firm characteristics and popula-tion composition. In fact, these factors reflect marketstructure and opportunities, and thereby influence poten-tial newcomers’ entry decision. Following resource parti-tioning theory (Carroll 1985), we propose that generalist

retailers (e.g., Wal-Mart and Carrefour) are less likely toenter (or survive) highly concentrated markets than spe-cialist retailers (e.g., Footlocker and Body Shop), sincethere are fewer survival opportunities for generalists thanspecialists.

Another weakness of density dependence theory liesin its unrealistic assumption: all organizations exert equiva-lent impact on other organization’ mortality rates. Weargue that this unrealistic assumption should be relaxedand let imitative isomorphism provide the boundary of thetheory instead. According to the imitative isomorphismmechanism, the densities of different sub-sets of thewhole population have different effects on other retailers’entry decision, and the bundled effects of density depen-dence can be disaggregated into characteristics of the sub-sets: aggregated number based, trait based, and outcomebased. Therefore, we propose that the number of foreignretail entries into a host country not only has a U-shapedrelationship with the density of the retail outlet in the hostcountry, but also has stronger U-shaped relationships withthe number of prior entries and exits made by all the otherforeign retailers, retailers with the same retail format, andretailers from the same home country, etc. Overall, theimitative isomorphism mechanisms and density depen-dence theory complement each other based on the com-mon ground of legitimacy seeking. After addressing theirboundary problems, the demarcation between these twotheories disappears.

Organization Inertia, Environment Uncertainty, and

Their Moderating Effects

International experience and firm size are good prox-ies of firm inertia in international market. Since selectionprocesses favor organizations with inert structures, weexpect that experienced and large organizations tend tomake foreign entry decision consistent with their ownexperience and care less about the signals of populationdensity. Since organizations reduce risks through mim-icking other firms in a highly uncertain environment (e.g.,DiMaggio and Powell 1983), we expect that the higher theenvironment uncertainty, the more likely organizationsconduct legitimacy seeking and thereby are more affectedby the signals of different population densities.


This paper provides an important alternative expla-nation for retailers’ foreign market entry and contributesto the existing literature in two aspects. Firstly, we providenew insights into organization ecology theory. The unre-alistic assumption of density dependence is relaxed byincorporating the characteristics of individual organiza-tion and organization salience into population analysis.Furthermore, the confounding effects of density depen-dence are decomposed, and the underlying mechanismsand patterns of density dependence are identified. Thisdisaggregating helps us to understand the inconsistent

results of density dependence studies and answers Zucker’s(1989) call for more credible proxy for cognitive legiti-macy. We propose aggregated number based, trait based,and outcome-based densities as proximal proxies of le-gitimacy in the context of foreign market entry. Second,by exploring the boundaries and limitations of organiza-tional ecology and imitative isomorphism, this paperidentifies the potential opportunity for integrating thesetwo theories and argues that these two theories are comple-mentary. Managerial implications are provided. Refer-ences are available upon request.

For further information contact:Chuanyi Tang

The University of ArizonaP.O. Box 210033

Tucson, AZ 85721–0033Phone: 520.621.3346



RECIPROCAL EFFECTS OF COMMITMENT IN EXPORTER-

IMPORTER RELATIONSHIPS

Farid Ahmed, University of Western Sydney, Australia

SUMMARY

Successful relationships with foreign partners offervarious advantages: greater operational efficiencies; re-duced risks (commercial or political); the ability to entermarkets that may otherwise be closed to foreign firms;exploiting national and/or differential advantages of part-ners; reducing the capital investment by sharing somevalue-adding activities with foreign firms; gaining speedin getting products to markets; and establishing long-termrelationships with a global network of suppliers, import-ers/distributors and other intermediaries. Thus, formingand maintaining a successful relationship should improvebusiness performance in an international market. How-ever, research on the impact of dyadic aspect of exporter-importer relationships on export performance is scarce.Given the unit of analysis in exporter-importer relation-ships is the relationship, rather than either firm, studyingthe interaction from the perspective of both partners isnecessary for a full understanding of the impact of rela-tional variables on export performance.

In this study, reciprocal effects of relationship com-mitment on exporter-importer relationship performanceare tested with data from 125 pairs of Australian-Thaiexporter-importer relationships. Drawing on earlier workin domestic distribution channels, and inter-organiza-tional relationship theories, this study develops and testsa dyadic model that hypothesizes links between reciprocaleffects of commitment in exporter-importer relationshipsand export performance.

Relationship performance is modeled as a function ofeach party’s commitment toward the other partner. Eachparty’s commitment toward the other partner is driven byreciprocal cycle of each partner’s perception of the other’scommitment. The cycle from exporter’s commitmentimporter’s perception of exporter’s commitment import-er’s commitment exporter’s perception of importer’s com-mitment, describes a positive reinforcement that increasesthe level of commitment by both partners over time.Perceptions of each other’s commitment are influencedby perceptions of how dependent each party is on theother, and perceptions of the level of relationship specificinvestment being made by the other party.

To test the hypotheses, we conducted a survey ofmatched pairs of Australian exporter and Thai importers.The sampling frame for the cross-sectional study com-

prised the Australian Trade Commission database ofexporters to Thailand. The database yielded 500 compa-nies. The final sample comprised responses from 125Australian exporters and their corresponding import part-ners in Thailand. Scales were sourced from the literature,and in some cases modified for the context and based onthe results of qualitative interviews. For scale validation,first we conducted exploratory factor analysis to assessthe underlying factor structure of the items. Next, weconducted confirmatory factor analysis so as to assess thevalidity of the measures. The fit indices indicate that themodels fit the data extremely well. We estimated thehypothesized structural model using LISREL VIII. Thefindings support the overall model, highlighting the recip-rocal effects of commitment and the link between com-mitment and exporter-importer relationship performance.

Although relationship commitment has often beenviewed as the central factor to the ongoing success ofbuyer-seller relationships, there is limited understandingof the reciprocal role of commitment in an internationalexchange relationship. This study adds to the body ofknowledge on the effect of partner’s mutually reinforcingcycle of commitment on international relationship (i.e.,exporter-importer relationship) performance. While mosthypotheses were supported, our results should be inter-preted in light of some limitations inherent in this re-search. The study was conducted within the context ofAustralian exporter – Thai importer relationships. It wouldbe valuable to replicate or refute the results of this studyin other cross-border relational contexts. Our study exam-ines the relationship at one point of time. Longitudinalstudy in future would provide a more in-depth under-standing of the development of relationship commitment.In future studies, large sample from various relationshipscan be explored.

We contribute to international marketing research inseveral ways. First, by focusing on the relationship ofexporter-importer exchanges, the study provides substan-tive support for previous findings in inter-firm relationsliterature and fresh insights about the determinants ofrelationship performance in cross-cultural exchange rela-tions. Second, this is one of the very few studies that tookthe perspectives of both sides of the dyad into account inidentifying the determinants of relationship performance.Third, the study highlights the significance of reciprocaleffects of commitment on both partners within the dyad ina cross-cultural relationship. Finally, the study reinforces


the importance of commitment in building successfulcross-border inter-organizational relationships. A full

working paper including references is available uponrequest.

For further information contact:Farid Ahmed

School of MarketingUniversity of Western Sydney

Penrith South DC, Locked Bag 1797, NSW 1797Australia

Phone: +61.2.96859686Fax: +61.2.96859612



INNOVATE, DON’T ALIENATE: HOW COMPONENTIAL FACTORS AND

CONSTRAINT ENHANCE CREATIVITY IN NEW PRODUCT IDEATION

Tanawat Hirunyawipada, Central Michigan University, Mt. Pleasant

SUMMARY

The development of new product idea (or new prod-uct ideation) is the transformation of a raw idea into arobust concept with consideration of fit and feasibility oftechnologies, customer benefits, and market opportunity.Although the ideation process often involves ambiguousprocesses, it is the most critical activity in new productdevelopment.

As a creativity task, the ideation process is tradition-ally considered heuristic rather than algorithmic in nature.The idea generation is heuristic in the sense that it is theprocess whereby individuals generate variation of ideas insomewhat mysterious ways. In contrast, several studieshave attempted to address the weakness of heuristicprocess by attempting to replace this process with analgorithmic approach. However, making the ideation pro-cess to be completely heuristic (i.e., anything goes) needsnot result in breakthroughs while the rigid, algorithmicprocedure often leads to just conventional outcomes.

Rooted in cognitive psychology, this study proposesthat ideation activities in new product development shouldbe pursued as the constrained stochastic behavior. Anideation task not only needs good componential factorsbut also requires constraint to frame the task by precludingunwieldy ideas while promoting high variability of ideas.

Focusing on the inputs and attempting to strike a balancebetween algorithmic and heuristic ideation process mayprovide the mechanisms to manage the psychologicalperceptions with an aim to stimulate and orchestrate theideation staff’s cognitive efforts to generate the creativeidea.

To achieve this goal, new product idea creativity isconsidered as the ideas that could turn out to be productsthat are novel to and useful for customers, and appropriateto firms’ existing production systems. All these dimen-sions should be used simultaneously to appraise the extentto which new product ideas are considered creative. Inaddition, this study asserts that componential factorsinclude two factors: (1) specialization representing ideacreators’ depth of new product development knowledge,experience, and skills in a product domain, and (2) diverseexpertise representing the breadth of ideation team’sknowledge, experience, and skills concerning the samedomain of new product development. These factors areessential and collectively can enhance creativity in thedevelopment of new product ideas. Finally, goal con-straint is defined and incorporated in the development ofnew product idea framework. This constraint encapsu-lates the overall criteria and stylistic principle for a par-ticular product domain and reflects the frame of referencefor new product idea development. Reference availableupon request.

For further information contact:Tanawat Hirunyawipada

Central Michigan UniversitySmith 105

Mt. Pleasant, MI 48859Phone: 989.774.2827



IN CONTROL OR OUT OF CONTROL? ACTIVE MANAGERIAL

INVOLVEMENT AND ITS IMPACT ON ESCALATION

OF COMMITMENT

William Boulding, Duke University, DurhamAbhijit Guha, Duke University, Durham

Richard Staelin, Duke University, Durham

SUMMARY

Introduction

Most marketing decisions are made in environmentswhere the final outcome is uncertain and not fully underthe firm’s control, e.g., the launch of a new product isimpacted not only by the decisions of marketing manag-ers, but also by competitor reactions, which are not underthe direct control of the firm. Yet many managers feel thatthey can use managerial skill to minimize the role ofexogenous factors (March and Shapira 1987). This ten-dency to overestimate the impact of managerial skill istermed as “managerial conceit.”

In this paper we examine the causes and conse-quences of managerial conceit. We do this examination inthe context of escalation of commitment (henceforth,escalation) to a failed new product launch. Extant escala-tion studies conceptualize managers as either (1) passiveobservers, or (2) passive investors, who view project dataand then may assign some funds to the project. The keypoint here is that, in both cases, there is little scope for theexercise of skill. These conceptualizations are in sharpcontrast to practice. If a project is viewed as going badly,a manager will get involved and use managerial skill tosuggest, for example, a revised pricing policy, with theintent of increasing the expected payoff associated withthe project. We ask the following: (1) Will the manageroverstretch the impact of his/her involvement, an involve-ment that is – prima facie – linked to managerial skill, andhence exhibit managerial conceit? (2) Will involvementin a decision related to a specific component of a projectlead the manager to be more likely to recommend persist-ing with the overall project?

Experiments/Results

Langer (1975) posits that skill-based interventionstypically lead to better outcomes. However, in somecontexts, interventions do not impact outcomes. Becausedistinguishing between contexts is a difficult task, indi-viduals use the heuristic that potential for skill-basedintervention (also called skill-cues) always lead to supe-rior outcomes. Hence even in contexts wherein interven-tions do not impact outcomes, individuals with some

element of skill-based involvement exhibit “illusion ofcontrol” (the managerial analogue of which is managerialconceit) and believe they can achieve better outcomesthan what is normatively possible. By modeling ourexperimental contexts on Langer (1975), where manage-rial skill does not impact project outcomes, differences inmanagerial behavior due to skill-cues can wholly beattributed to managerial conceit.

We run two studies, using 225 manager-participants(MBAs) with an average of six years of managerialexperience. Participants are told that they are part of acommittee at Quality Valves Company (QVC) that ap-proves new product launches. In period one, the partici-pants approve the launch of a new automotive valve –RXT 1. In period two, participants are asked to imaginethemselves two years in the future. The updated perfor-mance report provides information about RXT 1’s (lack-luster) post-launch performance. The participant is toldthe firm is planning to hire a stellar salesman to improvethe current situation. A revised report, which explicitlytakes into account the hiring of this stellar salesman,includes a distribution of NPVs indicating that the meanNPV from future cash flows is lower than the certain cash-out value associated with the available alternative ofdisposing the RXT 1 business. The normative decision isto terminate the project.

Participants are randomly split into a control groupand an experimental group. A separate manipulation(skill-cue) is provided to the experimental group, incorpo-rating potential for experimental group participants to beinvolved in the salesman hiring process. Pretests/analysesestablish that there is no normative difference in salesmanvaluation across the two conditions. Nevertheless, theinvolvement manipulation creates illusory differencesbetween groups, such that experimental group partici-pants rate the to-be-hired salesman as having a highervaluation. Consistent with our hypotheses, the results ofboth studies indicate that (1) a greater proportion of theexperimental group participants exhibit escalation, and(2) increased salesman valuation in the experimentalgroup explains the higher levels of escalation seen in theexperimental group. Not only does this study illustrate thelink between managerial conceit and escalation, but alsowe use various quantitative and qualitative measures


(e.g., thought listings) to provide a rich and deep under-standing of how managerial conceit leads to increasedescalation.

Conclusion

Our research hence suggests that there are two inde-pendent reasons why managers close to a project escalate.

One reason relates to the biasing impact of positive priors(Biyalogorsky, Boulding, and Staelin 2006). The otherreason – explored in this research – relates to the biasingimpact of managerial conceit, which is triggered by thepotential for skill-linked managerial involvement. Theabove findings contribute to research on both escalationof commitment and managerial conceit. References avail-able upon request.

For further information contact:Abhijit Guha

Duke University1 Towerview Drive

Durham, NC 27708–0120Phone: 919.218.6652



THE EFFECT OF NEW PRODUCT PORTFOLIO MANAGEMENT ON

FIRM PERFORMANCE: THE MODERATING ROLE OF MANAGERS’

AMBIGUITY INTOLERANCE AND COGNITIVE STYLE

Serdar S. Durmusoglu, University of Dayton, DaytonRegina C. McNally, Michigan State University, East Lansing

SUMMARY

New Product Portfolio Management (NPPM), thedecision process of evaluating, selecting, prioritizing, andallocating resources to new product development (NPD)projects, is pivotal for firm success. NPPM is comprisedof three dimensions: strategic connection, value maximi-zation, and balance.

In strategic connection, managers evaluate NPDprojects based on the extent to which they reflect firmstrategy (Cooper, Edgett, and Kleinschmidt 2001a). Theproject selection decision therefore involves assessinghow well a project meets strategic goals and justificationof the product opportunity (Krishnan and Ulrich 2001).Strategic connection is reflected by fit and contribution,which includes marketing and technological synergy(Henard and Szymanski 2001; Montoya-Weiss andCalantone 1994), order of market entry (Henard andSzymanski 2001), and how well the target market fits thecompany’s image and vision of its preferred target market(Krishnan and Ulrich 2001).

In the value maximization dimension, managers evalu-ate NPD projects based on the financial returns they areexpected to generate, such as long-term profitability,return on investment, or net present value (Cooper, Edgett,and Kleinschmidt 2001a; Sorescu, Chandy, and Prabhu2003). Studies reveal that this criterion is the most oftenused to evaluate new product projects (Cooper, Edgett,and Kleinschmidt 2001b).

NPD projects can be evaluated for their generalcontribution to the firm’s NPD project portfolio based onthe extent to which they create a diverse product mix (Blauet al. 2004). Therefore, in the balance dimension, manag-ers evaluate NPD projects based on the extent to which themix of NPD projects balances multiple concerns, such astime-frame, innovativeness, financial returns, and re-sources (Cooper, Edgett, and Kleinschmidt 2001a). Inaddition, balance can be evaluated relative to a collectionof specific criteria, such as speed to market (Griffin 1997),product innovativeness (Ali, Kalwani, and Kovenock1993), and various financial criteria like risk and return(Blau et al. 2004).

Research suggests that performance is enhanced whenmultiple NPPM components are considered in makingdecisions rather than employing a single criterion, such asevaluating only financial returns (Cooper, Edgett, andKleinschmidt 1999). Although limited, there is a growingnumber of studies about constructing NPD project portfo-lios that support firm strategy, maximize value, and pro-vide balance. However, there is a gap in the literatureregarding the differing effects the three NPPM compo-nents have on firm performance and the contingenciesinvolved in their effect strength. This study provides aninitial investigation to fill these gaps by first operational-izing and empirically testing the effect of NPPM dimen-sions on firm performance, and second, by identifyingtwo moderators that alter the influence of NPPM dimen-sions on firm performance.

The existing validated measures of the constructswere used where appropriate and, when needed, newmeasures were developed (e.g., NPPM dimensions) fol-lowing Churchill’s (1979) recommendations. All NPPMdimensions are measured with multiple items. Theseitems were pre-tested for relevance, interpretation, andclarity with six academics familiar with NPD literature aswell as with three practitioners. Based on the comments ofthese experts, some item wordings were modified andadditional items were included in the survey. Then, datawas collected from business professionals enrolled in apart-time MBA program at a large Midwestern university.Overall, the sample consists of 94 business professionals.

Based on our analysis, we find that only value maxi-mization directly enhances firm performance. Moreover,we find that decision makers’ associative cognitive styleinteracts with the balance dimension to hinder perfor-mance. We also hypothesize a negative interaction be-tween strategic connection and ambiguity intolerance onfirm performance, but find no support for this hypothesis.Despite this result, the growing evidence that personalitytraits can be expressed differently in different situationssuggests that future research should examine other per-sonality traits potentially influencing the relationship ofNPPM dimensions and firm performance.


The major managerial implication of our study is thatexecutives should continue to emphasize value maximi-zation during NPPM decisions while keeping in mind thatmanagers’ personality traits could affect the composition

of their firm’s NPD portfolio, which would have crucialtop and bottom line consequences. References are avail-able upon request.

For further information contact:Serdar S. Durmusoglu

Management and Marketing DepartmentSchool of Business Administration

University of Dayton300 College Park

Dayton, OH 45469–2271Phone: 937.229.3540



A CONCEPTUAL FRAMEWORK OF TECHNOLOGY ORIENTATION:

A RESOURCE-BASED PERSPECTIVE

Olivia F. Lee, St. Cloud State University, St. CloudMatthew L. Meuter, California State University, Chico

SUMMARY

Today’s business landscape is one of increased com-petitiveness, global perspective, and saturated markets.By recognizing the pivotal role technology has in achiev-ing market success, organizations are focusing more at-tention on acquiring technology knowledge and manag-ing technology resources across the organization. Tech-nology plays a central role in connecting employees,customers and the organization. More and more, organi-zations see the benefits of integrating technology intomarketing functions; advantages include increasing effi-ciency (Meuter et al. 2000), extending customer interfacehours (Dabholkar 1996), managing capacity (Radas andShugan 1998), enabling customization (Meuter et al.2000) and developing product innovation (Zhou, Yim,and Tse 2005). However, unless the use of technology isstrategically incorporated into a firm, unintended nega-tive consequences can result. Technology failures canrange from careless strategic planning to inferior designsor poor implementation. Without adequate technologylearning and knowledge management, it is difficult toeffectively assimilate technologies into firms.

In order to utilize technology effectively in anyelement of the marketing mix and service design, under-standing a firm’s technology orientation is essential. Tech-nology orientation (TECHOR) is defined as a continuumof technology integrated practices that either directly orindirectly enhance value creation. TECHOR exists alonga continuum because firms vary in the extent to which theyengage in technology strategies and practices. In the sameindustry, firms may perform differently because of differ-ences in their technology resources, competence, and

capabilities to explore or exploit technology opportunities(Teece, Pisano, and Shuen 1997).

To develop a multi-dimensional TECHOR construct,we conducted semi-structured brain storming sessionsand focus groups over a two-year period with more than300 health care employees in two leading hospitals inNortheastern Ohio. Based on our literature review andfield work, we operationalize TECHOR as a firm’s abilityto effectively (1) allocate technology resources, (2) de-velop technology competencies, and (3) sense and re-spond to technology opportunities. The antecedents ofTECHOR are comprised of two external forces (technol-ogy policy and industry characteristics), and three factorsof internal dynamics (top management, interdepartmentalgoals and organizational factors). We posit that TECHORlevels influence key employee outcomes (job satisfaction,job performance, and technology learning) and customeroutcomes (perceived quality, loyalty, and technologylearning). In addition, two environmental factors (marketturbulence and technology turbulence) influence the rela-tionship between TECHOR and key organizational out-comes such as firm performance, service quality, andcompetitive advantage.

This paper marks our initial efforts to understand thebenefits of technology learning and knowledge manage-ment in relation to marketing activities. By developing aTECHOR construct, firms will have the knowledge tobetter manage their technology initiatives, more effec-tively determine an appropriate level of technologicalintegration within their marketing operations, and de-velop stronger technological capabilities leading to sus-tainable competitive advantage.

REFERENCES

Dabholkar, Pratibha (1996), “Consumer Evaluations ofNew Technology-Based Self-Service Options: AnInvestigation of Alternative Models of Service Qual-ity,” International Journal of Research in Marketing,13, 29–51.

Meuter, Matthew L., Amy L. Ostrom, Robert I. Roundtree,and Mary Jo Bitner (2000), “Self-Service Technolo-gies: Understanding Customer Satisfaction with Tech-nology-Based Service Encounters,” Journal of Mar-

keting, 64 (July), 50–64.Radas, Sonja and Steven M. Shugan (1998), “Managing

Service Demand: Shifting and Bundling,” Journal ofService Research, 1 (1), 47–64.

Teece, David J., Gary Pisano, and Amu Shuen (1997),“Dynamic Capabilities and Strategic Management,”Strategic Management Review, 18 (7), 509–33.

Zhou, Kevin Zheng, Chi Kin (Bennett) Yim, and David K.Tse (2005), “The Effects of Strategic Orientations onTechnology- and Market-Based Breakthrough Inno-vations,” Journal of Marketing, 69 (April), 42–60.


For further information contact:Olivia F. Lee

Department of Marketing and Business LawG.R. Herberger College of Business

St. Cloud State UniversitySt. Cloud, MN 56301–4498

Phone: 320.308.4256Fax: 320.308.4061


Matthew L. MeuterDepartment of Finance and Marketing

College of BusinessCalifornia State University

Chico, CA 95929–0051Phone: 530.898.5880



THE EFFECTS OF STRATEGIC ORIENTATION IN EMERGING

ECONOMIES: A DYNAMIC CAPABILITY PERSPECTIVE

Caroline Bingxin Li, The University of Hong Kong, Hong KongKevin Zheng Zhou, The University of Hong Kong, Hong Kong

SUMMARY

Strategic orientation defines firms’ relative emphasison ways to acquire, allocate, and utilize resources; how tocreate organizational capabilities; and how to integratethe resources and capabilities into a cohesive whole toattain competitive advantages, according to the resource-based view (RBV) (Day 1994; Gatignon and Xuereb1997). Although the RBV as a theoretical framework ishighly influential for explaining how to achieve competi-tive advantage, it does not explain adequately how firmsachieve competitive advantage in the context of fastchanging environments (Eisenhardt and Martin 2000).Researchers extend the RBV further to the dynamic capa-bility perspective, stressing the critical role of capacitiesfor renewing, reconfiguring, and adapting existing firm-specific resources to achieve congruence with the chang-ing environment (Teece, Pisano, and Shuen 1997). How-ever, despite growing interest in the dynamic capabilityperspective, most studies are theoretical and conceptualand call for more empirical research to examine andvalidate this perspective (Lavie 2006). Extant literatureon strategic orientation also fails to integrate the dynamiccapability perspective.

To address these research gaps, this article examinesthe role of strategic orientation in the emerging economyof China. Emerging economies offer a rich context forstudies of dynamic capabilities because their unprec-edented changes and high uncertainties raise serious stra-tegic problems for firms (Hoskisson, Eden, Lau, andWright 2000). In particular, adaptive capability repre-sents a key element of dynamic capability through whichstrategic orientations may affect performance. This study

includes three types of strategic orientations (customer,competitor, and orientation) and examines how they in-fluence adaptive capability and consequently firm perfor-mance. And we propose that market dynamics (i.e., de-mand uncertainty and competitive intensity) may moder-ate the effects of strategic orientations on adaptive capa-bility.

We conducted a large scale survey in China andobtained a total of 380 usable responses. Our resultsempirically demonstrate that adaptive capability – a keyelement of dynamic capabilities – enhances firm perfor-mance in emerging economies. This finding supports thedynamic capability perspective by showing its positiverole in uncertain environments. Second, this study identi-fies strategic orientation as an important driver of adaptivecapability and thereby fills the research gap pertaining tohow to build dynamic capability. Third, as the interactiontest shows, the effects of strategic orientations on adaptivecapability depend on the dynamics of market environ-ments. When market demand is increasingly uncertain,the role of customer orientation declines, whereas tech-nology orientation becomes more prominent. With moreintense competition, both competitive and technologyorientations help build adaptive capability. Therefore, inorder to succeed in this dynamic market, firms shouldactively build their adaptive capability to integrate, build,and reconfigure existing competencies that provide strat-egy – environment congruence and thereby superior per-formance. Furthermore, to build adaptive capability, firmscan undertake a customer or technology orientation, butthey must understand the boundary conditions of thesestrategic orientations. References are available upon re-quest.

For further information contact:Caroline Bingxin LiSchool of Business

Room 730Meng Wah Complex

The University of Hong KongPokfulam Road

Hong KongPhone: 852.22415164

Fax: 852.28585614E-Mail: [email protected]


WHAT HAPPENS WHEN FIRMS COMBINE CUSTOMER AND

COMPETITOR ORIENTATIONS WHILE PURSUING LEARNING

OR PERFORMANCE GOALS RELATED TO COMPETITORS?

Xueming Luo, University of Texas at Arlington, ArlingtonMark Peterson, University of Wyoming, Laramie

James M. Munch, Wright State University, Dayton

SUMMARY

Two questions about goal setting for firms motivatethis study. First, what is the nature of the synergy a firmcan gain from combining a customer orientation with acompetitor orientation when it pursues organizationalgoal-setting focused upon learning? Second, what is thenature of the synergy of combining these two orientationswhen goal setting is focused upon performance? Thepurpose of this research is to better understand the pos-sible three-way interaction of these elements, as well asthe two-way interaction possible when combining a cus-tomer orientation with a competitor orientation. In thisway, scholars will be able to gain deeper understandingfor the outcomes of important strategic stances that thefirm can take in the competitive marketplace.

The Two-Way Interaction of Competitor and Cus-

tomer Orientations

In brief, the interaction between competitor and cus-tomer orientation is critical to the understanding of theinterdependent nature of the performance advantages ofeach orientation. Without a simultaneous employment ofboth, a firm is not likely to benefit fully from eachorientation’s advantages and at the same time to avoid theassociated limitations.

H1: Both orientations have synergistic benefits; competi-tor and customer orientationwill enhance each other’spositive influence on firm financial performance.

The Three-Way Interaction of External Orientations

with Organizational Goal-Setting

Goal setting is conceptualized as consisting of twodistinct types: (1) learning goal-setting, and (2) perfor-mance goal-setting (Seijts and Latham 2001; Seijts et al.2004). We predict the contingent effects of goal-settingtypes will likely differ.

Positive Moderating Effects of Learning Goal-Set-ting. Goal-setting related to learning from competitors islikely to augment the benefits of both orientations on firmprofitability. First, this is because if organizations arestrongly committed to and spend a lot of effort in learning

from their rivals, acquiring and exploiting knowledge ofcompetitors would bring “learning advantages” to thefirm (Grant 1996; Luo, Slotegraaf, and Pan 2006). Theseknowledge-based effects would enable the competitor-and customer oriented firm to satisfy customer needsmore effectively. Logically, the best partners from whichto learn would likely be one’s strong competitors due tothe distinct and complementary knowledge such competi-tors wield in the same industry as one’s firm (Lado, Boyd,and Hanlon 1997). In addition, setting learning-basedcompetitive goals would amplify the benefits of superiororganizational capabilities of satisfying customer needs atlower costs than the rivals.

H2: The synergistic benefits of both orientations will bestronger in organizations emphasizing learning goal-setting, than in those not emphasizing learning goalsetting.

Negative Moderating Effects of Performance Goal-

Setting. We predict that performance goal-setting is likelyto weaken the benefits of both orientations on firm prof-itability. First, when firms adopt performance-based com-petitive goals, they are more likely to focus on competi-tion as a win-lose rivalry. Second, firms with performancegoals focused on competitors may engage in more “pricewars” and desperately push for higher market share thana customer-focused firm (Armstrong and Collopy 1996).Similarly, a push for higher market share may not alwaysbe productive, because Armstrong and Collopy (1996,p. 191) conclude that “many studies have found a nega-tive relationship between market share and profits.”

H3: The synergistic benefits of both orientations will beweaker in organizations emphasizing performancegoal-setting than in those not emphasizing perfor-mance goal setting.

In this study, the authors use a multi-source approachfor data collection resulting in a final sample of 251business executives as well as the publicly available profitreports for these executives’ firms. Such a multi-sourceapproach allows for a better understanding of the interac-tions that may occur when firms combine strategic orien-tations, such as a customer orientation and a competitororientation, while pursuing learning or performance goals


related to competitors. Hierarchical regression was usedto assess the hypotheses of the study.

Each of the three hypotheses of the study was sup-ported. First, the results suggest that combining a cus-tomer orientation with a competitor orientation tends tobring synergistic gains in firm performance. Second,

results suggest that such positive synergistic gains areamplified when the organization is intrinsically motivatedthrough goal setting related to learning from competitors.Third, results also suggest that the same synergistic gainsare reduced when the organization is extrinsically moti-vated through goal setting related to performance relativeto competitors.

REFERENCES

Armstrong, J. Scott and Fred Collopy (1996), “Competi-tor Orientation: Effects of Objectives and Informa-tion on managerial Decisions and Profitability,” Jour-nal of Marketing Research, 33 (May), 188–99.

Grant, R.M. (1996), “Prospering in Dynamically-Com-petitive Environment: Organizational Capability asKnowledge Integration,” Organization Science, 7 (4),375–87.

Lado, Augustin A., Nancy G. Boyd, and Susan C. Hanlon(1997), “Competition, Cooperation, and the Searchfor economic Rents: A Syncretic Model,” Academyof Management Review, 22 (1), 110–41.

Luo, Xueming, Rebecca Slotegraaf, and Xing Pan (2006),“Cross-Functional Coopetition: The SimultaneousRole of Cooperation and Competition within Firms,”Journal of Marketing, 70 (2), 67–80.

Seijts, Gerard and Gary Latham (2001), “The Effect ofDistal Learning, Outcome, and Proximal Goals on aModerately Complex Task,” Journal of Organiza-tional Behavior, 22, 291–302

____________, ____________, Kevin Tasa, and Bran-don Latham (2004), “Goal Setting and Goal Orienta-tion: An Integration of Two Different Yet RelatedLiteratures,” Academy of Management Journal, 47(2), 227–39.

For further information contact:Mark Peterson

Department of Management and MarketingCollege of Business

University of WyomingDept 3275, 1000 E. University Ave.

Laramie, WY 82071Phone: 307.766.3124



OPENING THE BLACK BOX OF COMPETITIVE INTELLIGENCE:

UNCOVERING ANTECEDENT FACTORS AND

A PERFORMANCE ENHANCER

Adam Rapp, Kent State University, KentRaj Agnihotri, Kent State University, KentKevin Trainor, Kent State University, Kent

Michael Ahearne, University of Houston, Houston

SUMMARY

Over the years, a number of theoretical and empiricalresearch studies have postulated conceptual models ofcompetitive intelligence (hereafter, CI) and offered vari-ous perspectives on CI practices. Despite diversity interms of approach and method, one common theme isvisible among previous studies. The CI literature agreeson the importance of the role that firm personnel play inthe CI process. Among a firm’s employees, the sales forceis recognized as the single best internal source of informa-tion concerning what is occurring in the market withcustomers and competitors (Lambert, Marmorstein, andSharma 1990). This recognition relies on the fact thatsalespeople are the boundary spanners between the firmand the outside world and have a vital role in bringinginformation to the firm (Rapp, Ahearne, Mathieu, andSchillewaert 2006). This is especially true regarding com-petitors’ information due to the fact that the salespeople’srealm is contiguous to the competitors’ domain.

Although previous studies enrich our understandingof CI, several dimensions remain unexplored and needattention. For example, the significance of CI is confinedto the organizational level. To this point, the importanceof CI for a salesperson’s performance has not been dis-cussed. Addressing this limitation, this research studyexamines the impact of competitive intelligence (CI)within a sales context. More specifically, the purpose ofthis research is twofold: (1) we examine the antecedentfactors that influence a salesperson’s CI level, and (2) weexplore ways through which CI may affect a salesperson’sperformance.

Parallel to the salesperson cooperation model pro-posed by Yilmaz and Hunt (2001), we characterize theantecedent factors of CI into three categories: (i) indi-vidual, (ii) organizational, and (iii) relational. We believethat these three antecedents represent separate mecha-nisms by which salespeople will increase their CI level.Salesperson effort represents the individual’s actions,technology resources represent an organizational capabil-ity which facilitates the salesperson’s intelligence level,and the relationship quality represents the customer half

of the dyad, or person providing the information. Addi-tionally, based on contingency theory, we propose that therelationship between salesperson CI and performancewill be moderated by experience. Proposing a contin-gency framework of salesperson effectiveness, Weitz(1981) suggested that the relationship between salesper-son behavior and performance is contingent upon ormoderated by characteristics of the salesperson, the envi-ronment, and/or the work situation. In a similar vein,several researchers have demonstrated empirically thatthe relationship between performance and individual char-acteristics varies across sales circumstances (Franke andPark 2006; Weitz 1981).

Our data comes from a sample of salespeople andmanagers in the pharmaceutical field, along with archivalmeasures of salesperson performance. To examine therelationships empirically, we employed structural equa-tion modeling. Model fit for the measurement model(while controlling for common method variance) wasacceptable (χ2 = 295.4(94), p < .01; CFI = .95; RMSEA =.057) as was the linear effect model (χ2 (100) = 222.5, p <.01; CFI = .974; RMSEA = .04). Our results establish thepositive influence of salespeople’s effort on their CI level(H1; β = .08, p < .05); the positive impact of relationshipquality on salesperson CI (H3; β = .60, p < .01); thepositive relationship between salesperson CI and perfor-mance (H4; β = .101, p < .05). Notably, however, thelinear effects of technology resources and CI (H2; β =.028,ns), was not significant. Also, our interaction effect modelfit the data quite well (χ2 (99) = 218.61, p < .001; CFI = .97;RMSEA = .04) and was a significant improvement overthe linear effects model (Δχ2 (1) = 3.89, p < .01). We foundthat the interaction between experience and CI was posi-tive and significant (H5; β =.09, p < .05).

The findings of this study have their greatest meaningfor salespeople. They need to focus on CI for their ownbenefit. In an era of rapidly, and sometimes dramatically,changing market environment, CI helps them recognizethreats and opportunities posed by competitors. Also, afirm should ensure that its salesforce is sensitive about theexternal environment and possess critical competitiveinformation. The strong relationship found stemming from


a salesperson CI suggests this to be a valuable investment.A salesperson that has high level of CI exhibits behaviorsthat impact the customer and ultimately organizational

performance. We hope that our research provides a foun-dation on which future research efforts can build toaddress this issue. References are available upon request.

For further information contact:Adam Rapp

Department of MarketingCollege of Business Administration

Kent State UniversityP.O. Box 5190

Kent, OH 44242Phone: 330.672.1230



CHEERS! A MEANS-END CHAIN ANALYSIS OF COLLEGE STUDENTS’

BAR-CHOICE MOTIVATIONS

Robert E. McDonald, Texas Tech University, LubbockTillmann Wagner, Texas Tech University, Lubbock

Michael S. Minor, University of Texas – Pan Am, Edinburg

SUMMARY

Drinking alcohol is a major activity among collegestudents (Levine 1980). CBS News recently reported thatAmerican college students spend $4.2 billion each year onalcoholic beverages. Most of the research on this subjecthas focused on students’ motivations for excessive drink-ing and the associated public health concerns (e.g., Hingsonet al. 2005). Little research has investigated the context inwhich students drink alcohol (Clapp et al. 2006; Paviset al. 1997), specifically, why students choose to drink inparticular bars.

The present study uses a means-end-chain methodol-ogy to explore students’ motivation for selecting bars inwhich they drink. Depth interviews with 36 college stu-dents revealed the attributes of a bar that are most impor-tant to them, as well as the consequences and values thatresult from these attributes (Gutman and Alden 1985;Olson and Reynolds 2001). The interviews were con-ducted by a trained, male interviewer who was the sameage as many of the students who participated in the study.The interviews were conducted just before students en-tered a bar in the entertainment district near the collegecampus. Using Laddermap 5.4 software, responses wereanalyzed to develop implication matrices and ultimately,the hierarchical value map. When asked to describe fourattributes of an ideal bar, the students reported the follow-ing: patrons of the opposite sex, music, atmosphere,roominess, facility quality, prices, location, and servicequality. The two attributes mentioned most often amongthe students were drink prices and music. Nearly all of thestudents listed these attributes.

Students reported that they anticipated the followingconsequences of their desired bar attributes: drinking anddriving concern, accessibility, save money, buy moredrinks, get drunk, approach members of the opposite sex,

fast service, financial security, comfort, socialization,sexual encounters, fun, and a positive mood. The mostcommon anticipated consequence was socialization fol-lowed closely by having fun, being in a positive mood andsaving money. Surprisingly, getting drunk was mentionedby about only one third of the respondents. Interestingly,the consequences led to the same three values in mostcases: happiness, quality of life, and relationships.

Overall, the empirical evidence of this study suggeststhat college students’ bar choices are largely driven byhedonic motivations relating to having fun and pursuingpositive mood states, which are frequently facilitated bymeans of alcohol consumption and socialization (Shimand Maggs 2005). In doing so, social interaction tends tobe regarded from a short-term perspective where studentsseek sexual encounters as well as a long-term perspectiveassociated with improved and lasting relationships. Barsattempting to cater to this audience would do best toemphasize reasonable prices of beverages and admission,a good bar atmosphere including pleasant music, roomyfacilities, and good service quality.

In selecting bars with these characteristics, studentsseek opportunities for socialization with friends and po-tential sexual partners, comfort, a positive mood and fun,on a limited budget. They expect that these consequenceswill lead to relationships, improved quality of life andhappiness. While drinking is certainly a facilitating activ-ity to many students and inebriation a goal for some, theoverall emphasis is on the social and sexual human inter-action that they expect to enjoy at and after the barrespectively. This is consistent with the findings of Orfordet al. (2004) that students are motivated to drink by socialfactors. Emphasis on social interaction should be para-mount in the design and operation of these establishments.References are available upon request.

For further information contact:Bob McDonald

Rawls College of Business, MS 42101Texas Tech University

Lubbock, TX 79409–2101Phone: 806.742.1175



PSYCHOLOGICAL, PRODUCT-RELATED AND SITUATIONAL

INFLUENCES IN PURCHASING INTIMATE APPAREL

Yelena Tsarenko, Monash University, AustraliaFelix Mavondo, Monash University, Australia

ABSTRACT

Three significantly different profiles of involvementin the purchasing of intimate apparel were identified –enthusiasts who are highly involved in purchasing a bra,pragmatists with relatively moderate interest, and realistswith very limited interest. Traditional involvement theoryis extended by jointly examining consumer personality,product and situational factors.

INTRODUCTION

The fundamental insights of the literature examiningconsumer behavior in the fashion clothing market drawson consumer involvement theory. This theory postulatesthat consumers engage in a range of information process-ing activities stimulated by their level of involvementaccording to the relevance of the purchase. As Martin(1998) suggests, the consumer involvement measure re-flects a specific consumer response to the product. Thismeasure of involvement is essentially determined by howconsumers learn, acquire, and process information whenmaking purchase decisions.

A significant body of research literature exists thatdefines and discusses the application of involvementtheory in various product categories, advertising andpurchase decisions (Day 1970; Krugman 1965; Bloch1980). The question which always remains of interest ishow involvement profiles differ from each other. Whileconsumer demographic characteristics are traditionallyused as descriptors of profiles, there is some agreement infindings among researchers that demographic factors arenot always strong differentiators across fashion involve-ment profiles (O’Cass 2004; Moye and Kincade 2003).Based on previous studies, Zaichkowsky (1985) identi-fied three alternative factors, all of which contribute tovarious levels of consumer involvement: individual per-sonal characteristics, product characteristics, and situ-ational factors.

THE BRA AS A PRODUCT CONCEPT

Very little research literature concerns bra purchas-ing specifically, with most focusing on clothing in gen-eral. To the best of our knowledge only one study dealingexclusively with this product is available in the marketingliterature. Hart and Dewsnap (2001) conducted “An ex-ploratory study of the consumer decision process for

intimate apparel,” which outlined a number of challengesand issues associated with the decision-making process.The authors concluded that the bra shopper is a highlyinvolved consumer who perceives a high level of riskassociated with the purchase and is motivated by a rangeof factors such as the desire to be brand loyal, which hasthe added benefit of reducing the extent of the searchprocess. However, commensurate with the exploratorynature of that study, the methodology was only based onfocus groups, open discussion and a blind test experiment.

A number of factors may account for the lack ofresearch in marketing-related literature on this topic.Firstly, bras or intimate apparel can be considered as asubset of the apparel constellation that researchers feel isa sensitive and intrusive topic, and hence difficult toresearch. Secondly, the nature of the product is highlygender oriented, and its function as a garment is equivo-cal. It is a necessary commodity which traditionally formeda discrete part of women’s clothing, yet over recentdecades its fashion meaning has changed significantly.While it continues to serve its traditional functionality,modern technologies have brought variety in styles, de-signs and fabrics. Aided by technological advances inmanufacturing, the bra is increasingly presented as aproduct of high fashion that has moved from privateboudoirs to the public arena. In some social contexts, thebra becomes a distinguishable part of attire that indexesthe increasing dominance of visual culture. This raises thequestion of whether the bra represents support and com-fort, or allure and sex appeal – or both?

However, while the bra is often marketed as highfashion, fun, or a sensually satisfying product, actuallyshopping for a bra appears to be considered by many as “anecessary evil.” For example, anecdotal reflections onconsumer attitudes include the notion that “shopping fora sports bra may rank right up there with a visit to thedentist” (De Moeller 2004, p. 4), or that bra shoppinginvolves “being measured, poked, prodded and fitted bya matronly, but reassuring elderly woman” (Hall 2003,p. 45).

To further complicate this matter, choice has ex-panded dramatically in the last ten years, with a largenumber of new labels appearing, many of them fromfashion designers or celebrities such as “Elle MacPherson,Oroton, Kookai, and Kylie” (Hall 2003, p. 44). While it isacknowledged that the bra potentially may contribute to


enhanced personal confidence and positive self image(Koff and Benavage 1998), it is also potentially a sourceof contradictory messages. All these factors contribute toconsumer confusion and create some unexpected barriersin the process of bra purchasing.

CONSUMER INVOLVEMENT

The concept of involvement has been the topic ofmany consumer research studies that attempt to under-stand and explain consumers’ behavior (Laurent andKapferer 1985). Consumer involvement can be broadlydefined as “the degree of psychological identification,affective and emotional ties the consumer has with astimulus or stimuli” (Martin 1998, p. 8). However, anadequate measure of consumer involvement is not easilydetermined. For example, O’Cass (2004) has identifiedthe challenge of understanding purchase and consump-tion patterns, and consumer experiences in relation tofashion clothing. In addition, Laaksonen (1994, p. 22 ) hasnoted the elemental nature of the concept of involvement,which cannot be analyzed without taking into consider-ation its contextual properties. Hence, simply measuringconsumer involvement is not enough to explain variousfacets of consumers’ behavior. According to Laaksonenboth social psychological and consumer behavior re-searchers consider involvement as a property of product-related cognitive structure, determined either in terms ofan attitude or knowledge structure in relation to theproduct (1994, p. 28). Based on this view, in order to betterdelineate involvement profiles, we took into consider-ation consumers’ personal characteristics, the productrelated characteristics of the bra, and situational factors inbra purchasing.

SELF-CONCEPT

The findings from the exploratory study conductedby Hart and Dewsnap suggests that bras are related to self-image or self-concept (2001, p. 115). This finding isconsistent with earlier research by Gould and Stern (1989),which states that “Stereotypical biological distinctionsare important in consumption spheres, especially thosepertinent to socially visible products which express sexualidentity: most notably, fashion apparel items” (p. 130).Malhotra (1988, p. 6) states that “consumers prefer, intendto buy or use brands/products which are more congruentwith their self concepts.” The self-concept scale devel-oped by Malhotra consists of three dimensions – ideal self,actual self and social self – with the semantic differentialemployed as the measurement technique. However, manystudies are based on the actual self-concept since theothers (ideal and social) produce equivocal results(Malhotra 1988). We hypothesize that:

H1: Women highly involved in the purchase of intimateapparel are more likely than less involved women to

consider themselves (a) colorful, (b) vain, (c) indul-gent, and (d) excitable.

PRODUCT-RELATED CHARACTERISTICS

Product-related attributes reflect style, design, andbrand factors. Consumers vary in their approach to pur-chasing this product. Therefore, given the diverse motiva-tions – from ideational to practical – involved in brapurchase decisions, female consumers may place highutility on branded products or select products with spe-cific styles and designs. Some women actively seek tomaximize their sex appeal while another group is con-cerned with the functionality of the garment in the supportand comfort it provides. In other respects, celebrity en-dorsements have resulted in the constant appearance ofnew brands that may encourage female shoppers to ex-plore several options on the market. Also, situations mayarise when customers need to shop around to match theirintimate apparel to a specific style or color of attire.

We hypothesize that:

H2: Women highly involved in the purchase of intimateapparel are more likely than other women to considerthe importance of (a) bra style and design; (b) brabrands.

SITUATIONAL CHARACTERISTICS

Factors associated with specific retail situations mayinduce or enhance satisfaction with the purchase experi-ence or lead one to consider shopping as a chore. Asignificant amount of research from various disciplinesfocuses on the impact of the environment upon humanbehavior, such as the extent to which the environmentshapes approach or avoidance behavior. If the environ-ment is considered relatively pleasing and friendly anindividual is likely to exhibit approach behavior (Greenlandand McGoldrick 2003). In retailing literature the mainemphasis is placed on the physical attributes of the stores.Gould and Stern (1989) found “fashion conscious womenwill tend to look for appearance-related reinforcement inshopping activity. The presence of mirrors, flatteringlighting, and sales people trained to provide positiveappearances are reinforcers” (p. 142) of satisfying pur-chasing environments. Advice is very important to brabuyers as, in comparison to a t-shirt or dress, a bra is ahighly complex apparel item with an enormous range ofstyles, fabrics, brands, and sizes that cannot be allowed tocompromise its underlying function. We therefore hy-pothesize that:

H3: Women who are highly involved in purchase ofintimate apparel are more likely than other women toconsider the importance of situational factors like (a)store environment (b) ambience of the lingerie de-


partment/section (c) availability of professional ad-vice.

METHOD

Sample and Data Collection

The population of interest in this study was Austra-lian females over the age of 18 years. A seven-pagequestionnaire was designed to capture a range of issuesassociated with the purchase and use of a bra. Questionsfocused primarily on five aspects: (a) fashion conscious-ness; (b) attitude toward bra shopping; (c) product andsituational characteristics; (d) self-concept; and (e) demo-graphic characteristics. Six hundred questionnaires weredistributed to shoppers who were approached at a varietyof retail outlets: a prestigious department store; discountdepartment stores; and an independent boutique lingeriestore. Those who agreed to participate in this study weregiven the questionnaire with a reply-paid envelope to takehome and fill out. In total 221 usable questionnaires werereceived, which corresponds to a 36 percent response rate.The data was entered into SPSS, noting the date thequestionnaire was received. This allowed for testing non-response bias through comparing early against late re-spondents. Comparisons using t-tests indicated there wereno differences between early and late respondents sug-gesting non-response bias was not a serious problem(Armstrong and Overton 1977).

DEVELOPMENT AND VALIDATION OF

MEASURES

Involvement Profiles

The items for the involvement scale were developedfrom Laurent and Kapferer (1985) and O’Cass (2004).Four subscales were extracted through exploratory factoranalysis (EFA). These were interpreted as (a) Hedonicvalue (Cronbach Alpha = 0.885); (b) Competent shopper(α = 0.796); (c) Interest in product (0.746); (d) Importanceattached to bra as an extension of one’s identity (α =0.701). The psychometric properties for this scale wereacceptable. These four factors were used for K-meanscluster analysis of the sample. Several cluster solutionswere attempted; specifically, two-, three-, and four- clus-ter solutions. These were compared for their efficacy andthe resulting distribution of respondents to the clustercells. The three-cluster solution appeared the best repre-sentation across all measures used. In order to allowdeeper examination of three clusters, other techniquessuch as profiling or discriminant analysis was used to seewhat internal variables account for the clustering. Thethree clusters were then subject to further detailed profil-ing. Results in Table 1 show the profiles of the threeclusters. Cluster 1 was labeled Enthusiasts as it was

characterized by high scores for hedonic value, highscores for interest in bra and very high scores on the braas an extension of one’s identity. Enthusiasts had signifi-cantly higher hedonic value scores than both Pragmatistsand Realists. Cluster 2 was labeled Pragmatists and wascharacterized by moderate hedonism, very low compe-tence in bra purchase, low interest in bra and little consid-eration of the bra as an extension of the individual. The lastcluster was labeled Realists. This group was characterizedas having no hedonic association with the bra, very limitedinterest in purchasing the bra, and no consideration of thebra product as a projected identity of the individual.However, respondents in this group considered them-selves as very competent bra shoppers. For competence inpurchasing bras, the Realist buyers were surprisinglysimilar to Enthusiasts and significantly more competentthan Pragmatists (p < .001). One major distinguishingfeature of the groups is revealed in the context of the braas an extension of the individual. The Enthusiasts scoredsignificantly higher (p < .001) than both Pragmatist andRealist buyers.

Personal Characteristics

The 15 bipolar items of the Self-concept scale werederived from Malhotra (1981) without any modifications.

Product Related Characteristics

This section of the questionnaire was developed froman extensive fashion clothing literature review (Deeter-Schmelz et al. 2000; Hart and Dewsnap 2001; O’Cass2000; Greenland and McGoldrick 2003) and adapted forthis study.

Situational Factors

This was measured in the following subsectionswhich were related to: store environment, ambiance, andimportance of advice.

In addition, we assessed convergent and discriminantvalidity of dimensions of situational and product relatedcharacteristics, using AMOS 6.0. All constructs (seeTable 2) had acceptable level of reliability and averagevariance extracted (AVE) greater than the 0.5 cut-offpoint.

RESULTS AND DISCUSSION

In this section we discuss the differences betweeninvolvement profiles based on demographic characteris-tics. We then examine the relationships between degree ofinvolvement and self-concept, product, and situationalcharacteristics.


One-way ANOVA was employed to examine thedifference across involvement segments on consumerpersonality, product and situational factors.

Involvement Profiles and Self-Concept

The self-concept paradigm was applied to investigatedifferences in aspects of personality across involvementlevels. The results presented in Table 3 demonstrate thatthe three groups defined by degree of involvement in brapurchase are not different across most of the self-concepttraits. However, significant differences were found insome cases; for example, for ruggedness (p < .001), whichresulted from comparisons of the highly involved (Enthu-siasts) and the minimally involved (Realists) female shop-pers. This finding was not previously hypothesized. Thefindings for colorfulness (p < .05) and vanity (p < .001)support H1a and H1b, and indulgence (p < .001) lendssupport to H1c. In addition, the minimally involved group(Realists) has a much higher score than Enthusiasts for

perceiving themselves as unexcitable (p < .05) leading tosupport of H1d

From a marketing perspective the most sharply de-fined profile in this analysis is that of the Enthusiast brabuyer who could be characterized as rugged, excitable,indulgent, colorful and vain. These characteristics couldbe used as a basis for segmentation, targeting and adver-tising. These findings are largely consistent with those ofGoldsmith, Moore, and Beaudoin (1999) who examinedcustomer degrees of innovativeness in fashion clothes.

Involvement Profiles and Product-Related Charac-

teristics

The results in Table 3 show that there are no signifi-cant differences across the segments defined by degree ofinvolvement in terms of bra style and design, leading torejection of H2a. This suggests that all bra purchasersconsider these characteristics to be critical (as evidenced

TABLE 1

Profile of Respondents

Variable Enthusiasts Pragmatists Realist F-ratio Different Sets

Hedonic 4.95 2.89 2.68 131.98*** E>P & R

Competency 4.84 2.94 5.08 154.17*** E & R>P

Interest in bras 5.47 4.40 3.84 58.80*** E & P>R

Functionality 5.85 5.81 5.37 4.90** E & P>R

Projected identity 4.77 2.97 2.87 67.10*** E > P & R

Age 2.16 2.44 2.77 9.29*** R > E

** p < .01, ***p < .001

TABLE 2

Correlation Matrix of Measures

Internal Consistency 1 2 3 4 5

1 Bra style and design 0.82 0.74

2 Brand attachment 0.85 .298** 0.73

3 Store environment 0.84 .427** .280** 0.76

4 Ambiance of lingerie section 0.65 .518** .409** .374** 0.57

5 Importance of advice 0.59 .240** .218** .223** .355** 0.65

Diagonal (in italics) shows the square root of the average extracted for each construct.** p < .01


by the high mean values). Thus, one can argue that allfemales place high expectations on the functional at-tributes of the bra irrespective of their fashion orientationor degree of bra involvement. However, as we anticipated,significant differences emerge in the importance attachedto brand names across groups. The highly involved groupconsiders brands and brand names very important whenpurchasing a bra (p < .001). This finding supports H2b.This result highlights the increasing importance of brand-ing for products such as intimate apparel, since purchas-

ing is associated with high levels of risk. Attachment tobrands reflects the increasing use of celebrities in braadvertising while the importance of design and styleperhaps reflects an emphasis on functionality, comfort,and simplicity.

Involvement Profiles and Situational Factors

Store attributes are essential situational factors indelivering value to the customer. These factors include the

TABLE 3

One-Way ANOVA Test of Differences Across Consumer Personality, Product, and

Situational Characteristics by Degree of Involvement

Highly Moderately Minimally

Involved Involved Involved

Enthusiast Pragmatist Realist Group

(E) (P) (R) F-ratio Differences

N = 82 N = 64 N = 75

Consumer Personality Characteristics (self-concept)

H1a: Colorful 5.56 4.97 5.28 4.203* E > R

H1b: Vain 3.70 3.33 2.91 6.917*** E > R

H1c: Indulgent 4.50 3.67 3.75 7.717*** E > P and R

H1d: Unexcitable 3.68 4.23 4.63 7.099*** R > E

Rugged 4.62 4.58 3.97 7.043*** E and P > R

Uncomfortable 5.20 5.06 5.32 0.681 NS

Dominating 3.70 3.83 4.09 2.245 NS

Pleasant 2.32 2.25 2.28 0.056 NS

Contemporary 3.43 3.25 3.49 0.709 NS

Organized 2.80 2.36 2.61 1.43 NS

Rational 3.80 3.31 3.33 1.858 NS

Youthful 4.17 4.13 4.72 2.972 NS

Formal 4.30 4.67 4.75 2.21 NS

Orthodox 5.10 5.16 4.88 0.959 NS

Complex 3.72 3.80 4.23 2.111 NS

Product-Related Characteristics

H2a: Bra style and design 5.81 5.70 5.45 2.366 NS

H2b: Attachment to brands 4.45 3.68 3.44 13.760*** E > P and R

Situational Characteristics

H3a: Store environment 5.26 5.20 4.86 2.349 NS

H3b: Ambiance of bra

department/section 4.69 4.08 3.79 14.065*** E > P and R

H3c: Importance of advice 4.38 4.23 3.53 8.935*** E > R


general store environment, availability of professionaladvice from service personnel and the ambience of thelingerie department. There are no significant differencesacross the groups for general store environment, leadingto rejection of H3a. This result perhaps suggests that mostoutlets provide adequate atmospherics so that this is not adistinguishing feature across outlets. However, the im-portance attached to ambience in the lingerie section/department differs significantly across groups (p < .001)with Enthusiasts considering ambience much more im-portant than both the Pragmatists and Realists. This lendsstrong support to H3b. Finally, there is a significantdifference between Enthusiasts and Realists (p < .001) intheir appreciation of professional advice provided by thestore. In general, H3c is supported. However, given theself-perception of the Realists as highly competent in brapurchasing, it might suggest that they perceive them-selves as knowledgeable and independent customers whodo not require any assistance in choosing the garment. Theresults suggest that despite high competence in bra pur-chase, the Enthusiasts still need and seek advice fromsales staff. It could be speculated that perhaps this appar-ent “need” for advice is not practical as much as an“aesthetic” need to fulfil their sense of a satisfying shop-ping experience.

The results discussed so far indicate that although thebra has become increasingly a fashion accessory, it is stillconsidered as highly personal and intimate apparel. There-fore, product characteristics and situational factors suchas store attributes play a significant role for customersdespite different involvement levels. A comparison of thethree groups shows that Enthusiasts are a potentiallyexciting and dynamic market segment since members ofthis group love shopping for bras, they are more indulgentthan other groups, they consider themselves more color-ful, and the majority of respondents in this segment areyounger females. This group attaches a high level ofimportance to branded products, they are willing to spendmoney for the right brands, and will consciously look forbra brands that match aspects of their self-concept.

CONCLUSION

While many studies focus on differentiation of vari-ous segments of female shoppers, this is the first studywhich attempts to segment women in relation to purchaseof intimate apparel. The segments that emerged in thisstudy clearly indicate the scope for specifically tailoredretail services and offers that address the various degreesof involvement.

From the academic perspective, this study signifi-cantly contributes to the literature on consumer behaviorin this specific product context. The overall goal was notonly to attempt to examine the existence of various in-volvement profiles, but also to expand the theory ofinvolvement through incorporating personal as well prod-uct and situational characteristics to gain deeper insightsinto various customer segments as it was theoreticallypostulated in Zaichkowsky (1985). In this study we usedthe paradigm of self-concept which was developed byMalhotra (1988). While the findings suggest that theprimary distinguishing factor of those who are highlyinvolved is not an overall self-concept, the individualaspects of self-concept (colorful, indulgent, etc.) that doreflect significant differences seem to form a coherent setof attributes that could be used for segmentation, targetingand promotion. In future empirical studies, a differentscale for self-concept from psychology would be recom-mended to compare the differences.

The results of this study and the associated literaturereview indicate that the drivers of interest in bra purchasesare only understood anecdotally. Managers of retail out-lets for bras can take heart that the principal drivers ofpurchase are similar to those associated with fashionpurchases in general. However, in probing the store-related situational factors, the results suggest managersshould pay significant attention to the selection, trainingand employment of perceptive and customer friendly staffwho are able to provide appropriate assistance in brapurchases. This type of interaction is perceived as sensi-tive and may be seen as an intrusion by some customers.Managers should invest in designing, locating and fur-nishing change rooms to provide privacy and a pleasantenvironment for women to fit and assess bras for pur-chase. There is a constant need to monitor the ambience ofthe lingerie department to ensure an appealing environ-ment since this may be a source of customer satisfactionand competitive advantage.

Another noteworthy finding is that females valuequality, style, and material regardless of their specificinterest in this intimate product. In addition, female con-sumers welcomed this study and demonstrated significantand genuine interest in this topic. The researcher receivednumerous emails, phone calls, and notes at the end of thequestionnaire in which females shared their concerns andideas. Evidence of this active interest was also confirmedby the response rate which indicates that women arewilling to engage in conversation concerning this poten-tially sensitive product.


REFERENCES

Aaker, J., S. Fournier, and S.A. Brasel (2004), “WhenGood Brands Do Bad,” Journal of Consumer Re-search, 31 (June), 1–16.

Armstrong, J.S. and T. Overton (1977), “Estimating Non-Response Bias in Mail Surveys,” Journal of Market-ing Research, 14 (3), 396–440.

Bloch, P.H. (1980), “An Exploration into the Scaling ofConsumers’ Involvements with a Product Class,”Advances in Consumer Research, Association forConsumer Research, 61–65.

Day, G.S. (1970), Buyer Attitudes and Brand ChoiceBehavior. New York: Free Press.

De Moeller, O. (2004), “How to Choose a Sports Bra,”The West Australian, (June 9), [http://www.thewest.com.au/].

Deeter-Schmelz, D.R., J.N. Moore, and D.J. Goebel (2000),“Prestige Clothing Shopping by Consumers: A Con-firmatory Assessment and Refinement of thePRECON Scale with Managerial Implications,” Jour-nal of Marketing Theory and Practice, 8 (4), 43–58.

Gould, S.J. and B.B. Stern (1989), “Gender Schema andFashion Consciousness,” Psychology and Market-ing, 6 (2), 129 –45.

Greenland, S.J. and P.J. McGoldrick (2003), “Measuringthe Atmospheric Impact on Customers,” (Workingpaper).

Hall, G. (2003), “Function versus Fashion,” ApparelMagazine, (April), 44.

Hart, C. and B. Dewsnap (2001), “An Exploratory Studyof the Consumer Decision Process for Intimate Ap-parel,” Journal of Fashion Marketing and Manage-ment, 5 (2), 108 –19.

Koff, E. and A. Benavage (1998), “Breast Size Perception

and Satisfaction, Body Image, and PsychologicalFunctioning in Caucasian and Asian American Col-lege Women,” Sex Roles, 38 (7/8), 655–73.

Krugman, H. (1965), “The Impact of Television Advertis-ing: Learning Without Involvement,” Public Opin-ion Quarterly, 29 (Fall), 349–56.

Laaksonen, P. (1994), Consumer Involvement: Conceptsand Research. London and New York: Routledge.

Laurent, G. and J.N. Kapferer (1985), “Consumer In-volvement Profiles: A New Practical Approach toConsumer Involvement,” Journal of AdvertisingResearch, 25 (6), 48–56.

Malhotra, N.K. (1988), “A Methodology for MeasuringConsumer Preferences in Developing Countries,”International Marketing Review, 5 (Autumn), 52.

Martin, C.L. (1998), “Relationship Marketing: A High-Involvement Product Attribute Approach,” Journalof Product and Brand Management, 7 (1), 6–26.

Moye, L.N. and D.H. Kincade (2003), “Shopping Orien-tation Segments: Exploring Differences in Store Pa-tronage and Attitudes Toward Retail Store Environ-ments among Female Apparel Shoppers,” Interna-tional Journal of Consumer Studies, 27 (1), 58–71.

O’Cass, A. (2000), “An Assessment of Consumers Prod-uct, Purchase Decision, Advertising and Consump-tion Involvement in Fashion Clothing,” Journal ofEconomic Psychology, 21 (5), 545–76.

____________ (2004), “Fashion Clothing Consumption:Antecedents and Consequences of Fashion ClothingInvolvement,” European Journal of Marketing, 38(7), 869–82.

Zaichkowsky, J.L. (1985), “Measuring the InvolvementConstruct,” Journal of Consumer Research, 12 (3),341–52.

For further information contact:Yelena Tsarenko

Department of MarketingMonash University

S 7.18, Chisholm Tower26 Sir John Monash Drive

P.O. Box 197Caulfield East, VIC 3145

AustraliaPhone: +61.3.9903.2354



THE EFFECTS OF CONSUMER ANIMOSITY AND CONSUMPTION

SITUATION ON THE PURCHASE DECISION OF NATIONALLY

ICONIC PRODUCTS: A MEANS-END CHAIN MODEL

Joseph P. Little, Saint Louis University, St. Louis

SUMMARY

Previous consumer animosity research has focusedon how animosity toward a country-of-origin affectspurchasing behavior. A problem arises in the fact thatmany consumers are not aware of a product’s country-of-origin (Samiee, Shimp, and Sharma 2005). In fact, Samieeet al. (2005) found that consumer’s Brand Origin Recog-nition Accuracy (BORA) is only 35 percent. The goal ofthe current analysis is to contribute to the consumeranimosity research stream by developing a conceptualmodel of how consumer animosity affects the purchasedecision of nationally iconic products, or those productsthat are highly visible as originating from a certain coun-try, in different consumption situations.

Previous literature has defined animosity as angerdirected at a particular country due to political, economic,diplomatic, or military events (Klein, Ettenson, and Mor-ris 1998). Also, recently religious differences have beensuggested as causing consumer animosity (Riefler andDiamantopoulos 2007), as experienced in the case ofArab countries’ anger over a Danish newspaper publish-ing caricatures of Mohammed. Whatever the cause forfeelings of animosity, previous literature demonstratesthat a company’s sales can be damaged by consumers’animosity toward the firm’s home country.

Klein et al. (1998) were the first to link consumeranimosity to the willingness of the consumer to purchaseproducts from the country for which the animosity isfocused. Using the background of the Nanjing massacre(300,000 Chinese died), Klein et al. (1998) showed thatanimosity had a negative impact on Chinese consumers’purchasing of Japanese products. However, the researchalso suggested that feelings of animosity did not affect thequality perceptions of the Japanese products. Feelings ofanimosity affect purchase decisions separately from con-sumer product quality evaluations.

Based on Samiee et al. (2005) findings the questionarises: If consumer animosity negatively affects the pur-chase decision of a foreign product, but not the actualproduct quality judgments, and consumers struggle toaccurately name a products country-of-origin, then what,if any, products are actually affected by consumer ani-mosity? The logical answer would be to focus attention onproducts that are highly visible as originating from a

certain country. The current analysis will refer to theseproducts as nationally iconic products.

Furthermore, people consume differently based onthe consumption situation. Research has shown that peopleconsume and behave differently when the act of consump-tion takes place in a private situation compared to a publicsituation (Bearden and Etzel 1982). Fishbein’s theory ofreasoned action (1975) examines both cognitive and af-fective components of attitude formation. Fishbein em-phasizes the effects of social norms and subjective moti-vations of consumers to adhere to the norms of referencegroups. If a consumer chooses to purchase a foreignproduct, they will have to deal with the potential criticismor other social consequences of the consumer’s referencegroup.

The goal of developing a means-end chain model thatincludes the consumption situation is to build a frame-work to examine how animosity affects purchase deci-sions when others are present during consumption (pub-lic), and when the consumer will be alone (private). In aprivate-use situation, consumption of a product allowsthat product to contribute and reflect our own identity(Belk 1998). Consumers seek, confirm, and ascertain asense of being through what they have (Sartre 1943). In apublic consumption situation, not only are consumersseeking to find a product that reflects their identity, theyare also motivated to comply with the expectations ofothers (Lutz 1991).

In conclusion, the consumption situation, the recog-nition of a product as originating from a foreign country,perceived consequences, and level of animosity are alltaken into account in order to give value to the consumer’sfeelings of animosity. Once a consumer’s animosity hasbeen weighted, compared to other values and consider-ations, a purchase decision is made. After the purchasedecision is made, a consumer will face the consequencesof that decision.

Based on the literature review and the developmentof the means-end chain model, the following hypothesesare presented:

H1: Consumer animosity will more often affect the pur-chase decision of nationally iconic products thanother products from the same country-of-origin.


H2: Consumer animosity will have a greater negativeaffect on the purchase decision of nationally iconicproducts than on other products from the same coun-try-of-origin.

H3: A public consumption situation will cause consumersto give greater weight to their feelings of animositythan in a private consumption situation when makinga purchase decision of a nationally iconic product.

H4: Prevention focused individuals will give greaterweight to their feelings of animosity than promotion

focused individuals when making a purchase deci-sion of a nationally iconic product.

H5: A public consumption situation, compared to a pri-vate consumption situation, will cause preventionfocused consumers to increase their weighting ofanimosity more than promotion focused consumerswhen making a purchase decision of a nationallyiconic product. References are available upon re-quest.

For further information contact:Joseph P. Little

Saint Louis University3674 Lindell Blvd.

St. Louis, MO 63108–3397Phone: 314.531.0991



THE SPORT SPECTATOR INVOLVEMENT MODEL: A CONCEPTUAL

FRAMEWORK AND FIRST EMPIRICAL RESULTS FOR

UNDERSTANDING THE TEAM INVOLVEMENT

OF GERMAN SPORT SPECTATORS

Klaus-Peter Wiedmann, Leibniz University of Hannover, GermanyNadine Hennigs, Leibniz University of Hannover, Germany

Frank Bachmann, Leibniz University of Hannover, Germany

SUMMARY

Professional team sport events are getting more andmore important in the planning schedules of the everydaylife; the fact of being a fan and supporter of one particularprofessional sport team becomes generally accepted in thesociety. To explore the multidimensional sport spectatorinvolvement construct, the present study replicated andextended the Team Sport Involvement (TSI) model andthe Sport Interest Inventory (SII) (Funk, Ridinger,Moorman 2004), a psychometrically sound survey instru-ment and valuable tool for both future researchers andpractitioners to examine unique motivational factors amongsport consumers. Originally developed and tested onconsumers of women’s sport, we adapted the TSI modelon German professional men’s soccer and the FIFAWorld Cup 2006 to examine individual spectators’ differ-ences referring to the interest, social, environment andentertainment dimension of sport spectator involvement.

Conceptual Model: Determinants of Sport Spectator

Involvement

The primary goal of this paper was to establish amultidimensional sport spectator involvement frameworkand explore a related factor structure, which specifies thedimensions of individual interest, motivation, and arousalrelated to an individual’s involvement with a competitivesport team or event. Focusing on its potential motivationalorigins and with reference to the TSI model, we suggestthat the sport spectator involvement can be described byfour strongly correlated but not identical motivationaldimensions representing emotional and cognitive fea-tures: (1) interest, environment, experience, and socialvalue encompassing the facets interest in team, interest insport, interest in players, and sport knowledge (interestdimension); (2) role model, socialization, bonding withfriends and family, and escape (social value dimension);(3) aesthetics, drama, excitement, and entertainment value(entertainment dimension); (4) wholesome environment,community pride and vicarious achievement (environ-ment dimension). Figure 1 shows the proposed conceptualmodel to investigate the strongly correlated but not iden-

tical dimensions and origins of sport spectator involve-ment.

Methodology

To measure the underlying dimensions of sport spec-tator involvement against the background of our multidi-mensional framework, we did both, using already existingand tested measures and generating further items resultingfrom exploratory interviews. As a result, the question-naire included the items of the SII scale (Funk, Ridinger,Moorman 2004), supplemented by additional items con-cerning the FIFA World Cup 2006 in Germany. Theseadditional items were added to examine the impact of thisspecial sport event on the sport spectator involvement.The items were rated on a seven-point Likert scale (1 =strongly disagree, 7 = strongly agree). The data for thecurrent study were collected using the following proce-dures: According to the spectator behavior and involve-ment literatures, we used a four-step procedure to developand validate a conceptual model for examining the profes-sional team sport involvement. The sample used in thisstudy was defined as male or female respondents, aged 18years and older. A total of 424 interviews were conductedin winter 2006/2007.

Main Results

The main contribution of our research was to repli-cate and extend the TSI model in order to enhance currentunderstanding of consumer motives related to sport ingeneral and soccer in particular. Our research shows thatthe TSI model and its extension to German soccer fans andthe FIFA World Cup 2006 leads to reliable and validresults demonstrating that male and female spectatorsupport for a soccer team differs based upon interest,social, environment, and entertainment motivational di-mensions and individual involvement origins. The SIIfactors that were originally developed for women’s pro-fessional sport were rated favorably by all respondents inour men’s soccer research context with aesthetics, excite-ment and sport knowledge as the most influential factorsfor both, male and female sport spectators. In summary, a

164A

merican M


inter 2008


comparison of male and female fans and spectator in termsof demographics showed that they are very similar what isa clear indication of the SII’s cross-gender generalizability(i.e., that the TSI dimensions are as relevant to men as itis to women). While the interest dimension was of special

importance for male respondents, the other three dimen-sions (social, environment, and entertainment) were per-ceived to be more important by female respondents.References are available upon request.

For further information contact:Klaus-Peter Wiedmann

Institute of Marketing and ManagementLeibniz University of Hannover

Koenigsworther Platz 130167 Hannover

GermanyPhone: ++49.511.762.4862

Fax: ++49.511.762.3142E-Mail: [email protected]


ADVERTISING FOCUS AND BRAND ASSOCIATION STRENGTH: WHAT

IS THE EFFECT OF CROSS-TARGET INTERFERENCE?

Daniel W. Baack, Ball State University, MuncieBrian D. Till, Saint Louis University, Saint Louis

SUMMARY

The advertising literature is consistent in its call forfocused advertising messages (Aaker 1996; Adamson2006; Farquhar 1989; Ogilvy 1983). To quote Ogilvy(1983) “Every advertisement should be thought of as acontribution to the brand image. It follows that youradvertising should consistently project the same image,year after year” (p. 14). One of advertisings’ main goals isthe construction and reinforcement of brand equity (Keller2005). Focused advertising, therefore, has important brandmanagement implications and should influence the build-ing of brand meaning. There have been recent calls in theliterature for research on the transfer of meaning to brands(Keller 2001, 2003). This paper begins to meet this call inthe literature through an exploration of the efficacy offocused advertisement in building brand associations.

This study focuses on consumer-based brand equityusing an associative network perspective, and focuses onbrand associations as an indicator of the accessibility ofthe association for the consumer. Keller’s (1993) perspec-tive on brand associations is rooted in associative networktheory which assumes that “declarative knowledge isrepresented as a network of concept nodes connected bylinks that are strengthened each time two events co-occur”(Van Osselaer and Janiszewski 2001, p. 12). Associativenetworks are constructed of two key elements: (1) nodes(e.g., Coca-Cola, red, American) and (2) the links be-tween these nodes (Henderson et al. 1998).

Advertising is often attempting to modify the asso-ciative network around the advertised brand. Advertisingeffectiveness, however, can be hindered by cross-targetinterference, i.e., the negative effect of multiple targetassociations on the ability to add a new association orstrengthen an existing association in memory. Thesemultiple target associations diffuse node activation acrossall of the associations (Anderson 1983). In the case offocused advertising messages, advertisements attemptingto add or strengthen just one association will suffer fromless interference than advertisements attempting tostrengthen or add multiple associations.

This study also compares the relative sensitivity oftwo measures of brand association strength: (1) freeassociation tasks (Keller 1993) and (2) response latencytasks (Hennessey et al. 2005). To the best of our knowl-

edge, no study has compared them in a brand associationcontext. The response latency task has potentially impor-tant benefits that may make it a more effective measure ofbrand association strength for the following reasons: (1)the implicit nature of the method may make it less suscep-tible to biases (Fazio et al. 1992) and (2) it is a moresophisticated and potentially more sensitive measure.

The hypotheses tested were:

H1: Focused advertising will result in a stronger linkbetween the targeted association and brand than willa non-focused advertising.

H2: The response latency task will be a more sensitivemeasure of strength of association than the freeassociation task.

The study was a 2 X 2 between subjects factorialdesign. The two factors were focus of advertising (one vs.five association messages) and measure of strength ofassociation (free association vs. response latency). Thedependent variable is strength of association. A total of164 students, 26 using the response latency method, 138using the free association method participated. A graphicartist created ten color advertisements for the brand Lay’sPotato Chips. The association “classic” was the targetassociation to be strengthened. Two groups of subjectswere shown five advertisements. For Group One, the fiveadvertisements focused on the same single association“classic.” For Group Two five different associations werepresented. After, subjects completed a memory-clearingtask, the strength of the association for “classic” wasmeasured using either the response latency or the freeassociation method.

For response latency, focused advertising had a stron-ger post-exposure association than did non-focused ad-vertising (focused mean difference = 346.09ms, non-focused mean difference = 119.73ms). A one-tailedplanned contrast analysis found this difference to besignificant (t

(24) = -1.833, p < 0.05). For free association,

there was also a difference between the focused and non-focused conditions (focused mean difference = 0.2260;non-focused mean difference = 0.1077). A one-tailedplanned contrast analysis found this difference to bestatistically significant (t

(24) = 4.013.833, p < 0.01). To

measure the difference between the two strength of asso-


ciation measures, eta-squared scores were computed. Thedifference between the scores was not enough to claimany advantage for either method (response latency =0.123, free association = 106).

This study finds that the more focused an advertisingcampaign, the greater the campaign’s ability to add a newassociation. This result was true for both the responselatency and free association task measures. This study

provides preliminary evidence that the interference effectinfluences the effectiveness of advertising. For managersthis means that while maintaining the discipline necessaryto maintain advertising focus may be difficult this imagemay then lead to a stronger brand identity (Aaker 1996).The study also compared the response latency and freeassociation strength of association measures. The resultsof the eta-squared analysis found no statistical advantagefor either measure. References are available upon request.

For further information contact:Daniel W. Baack

Ball State UniversityMuncie, IN 47306–3435

Phone: 765.285.3435Fax: 314.285.4315



DUAL-PROCESSING AND MESSAGE SIDEDNESS: THE EFFECTS OF

INVOLVEMENT IN TWO-SIDED ADVERTISING

Martin Eisend, Freie Universität Berlin, Germany

SUMMARY

Theories that explain two-sided persuasion are basedon the assumption that recipients carefully scrutinize andevaluate positive and negative information. Also mostempirical studies investigating two-sided persuasion arebased on experimental designs where participants processtwo-sided messages in a rather elaborate manner. This,however, may not resemble how consumers usually pro-cess two-sided messages. This study examines whetherand how the underlying mechanisms of two-sided persua-sion for low and high involvement consumers differ.

The theoretical argument is based on dual-processingmodels which provide an explanation for the differentways of message processing under low and high involve-ment. Recipients differing in involvement evaluate anincoming message in different profoundness. They godifferent ways in forming their attitudes toward the mes-sage and are influenced by different aspects of the mes-sage. Whereas the effect of message sidedness is due toargument quality for high involved recipients, other fac-tors such as the number of arguments play a more impor-tant role for low involved recipients. When recipientsevaluate the message based on arguments, one-sidedmessages with only positive information should be evalu-ated more favorably than two-sided messages with bothnegative and positive information. An increasing propor-tion of negative information should enhance unfavorableevaluations of the message. However, when involvementis low, consumers are more likely to neglect the valence ofthe information and they rely on the mere amount ofinformation which impacts message evaluation. Messageevaluation is considered a mediator variable for the effecton attitude toward the brand.

The hypotheses were tested by way of an experimentwith a 2 (high vs. low involvement) by 4 (proportionnegative/all information in the message: 0/7 vs. 1/7 vs. 0/9 vs. 3/9) between-subjects factorial design. Two hundredtwenty undergraduates at a German University volun-teered to participate in the experiment. An advertisementfor a SONY notebook was chosen for the experiment. Theadvertisement includes a table comparing characteristicsof several product attributes of the SONY notebook with

characteristics of an average notebook of the same pricecategory. The advertisements differed in the number ofattributes (seven vs. nine) and the proportion of attributeswhere the SONY notebook was worse than the averagenotebook (0/7 vs. 1/7 vs. 0/9 vs. 3/9). Involvement wasmanipulated according to recommendations in the litera-ture. Attitudes toward the ad and attitude toward the brandare the dependent variables.

The results of the ANOVA show a significant maineffect of proportion and an interaction effect of proportionand involvement on attitude toward the ad. For lowinvolved participants, the effect of one-sided vs. two-sided messages with the same amount of information doesnot differ for both messages. However, the increasingamount of information (from seven to nine attributes)enhances attitudes toward the ad. For high involvedparticipants, attitude differs between one-sided and two-sided messages covering nine attributes while the differ-ence for the message covering seven attributes is notsignificant, though the results show the expected direc-tion. Furthermore, the overall effect of the amount ofinformation is not significant for high involved consum-ers. A path analysis supports the mediating role of attitudetoward the ad on attitude toward the brand for either theproportion of negative information for high involvedparticipants or the effect of the number of attributes forlow involved participants.

Message sidedness has an effect on ad evaluation asdescribed in previous research. However, the underlyingmechanism of this effect differs for low and high involvedconsumers. High involved consumers are influenced bythe negativity of the information provided by the message,i.e., the proportion of negative to positive attributes. Lowinvolved consumers are influenced by the mere amount ofinformation. Those variables impact message evaluationswhich in turn impacts brand evaluation. While the resultsdo not contradict the overall effect of message sidednessas described in previous studies, they provide furtherinsights on how two-sided messages impact low and highinvolved consumers in different ways. They also supportthe contribution provided by dual-process-models fortwo-sided persuasion research that has not been consid-ered so far. References are available upon request.


For further information contact:Martin Eisend

Marketing DepartmentFreie Universität BerlinOtto-von-Simson-Str. 19

14195 BerlinGermany

Phone: +49.30.838.544.60Fax: +49.30.838.545.57



UNCOVERING UNDERCOVER MARKETING: THE ROLE OF

CONSUMERS’ AGENT, PERSUASION, AND

TOPIC KNOWLEDGE

Guang-Xin Xie, University of Oregon, EugeneDavid, M. Boush, University of Oregon, Eugene

ABSTRACT

The present research identifies three aspects of con-sumer knowledge in coping with marketers’ use of surrep-titious interpersonal influence, and provides evidence forthe distinctive role of each in four experiments.

INTRODUCTION

With the diminishing effectiveness of traditionalmarketing techniques, undercover marketing (also called“stealth marketing”) has emerged as a popular alternativemethod to persuade consumers surreptitiously. Under-cover marketers often present a new product or service bycleverly creating and spreading “buzz” in an obtuse orsurreptitious manner (Kaikati and Kaikati 2004). A mar-keting company could, for example, pay an actor to use acertain product visibly and convincingly in locationswhere target consumers congregate. The actor may alsotalk about their product with people they befriend in somelocations, even handing out samples when it is economi-cally feasible.

Undercover marketing aims to penetrate consumers’defense mechanisms against marketplace persuasion. It isanchored on the premise that consumers trust genuinepersonal experience and unintended advice from othersmore than they trust marketers. However, consumers mayhave knowledge about tactics, acquired from firsthandexperience in everyday life and secondhand experiencefrom social interactions, media, and organizations. Suchpragmatic knowledge acts as a critical moderator duringmarketplace persuasion episodes (Boush, Friestad, andRose 1994). Whether because of personal experience ormedia coverage (e.g., Vranica 2005), this knowledge mayextend to undercover marketing tactics. The present re-search is one of the first empirical studies to investigatehow consumers cope with undercover marketers’ use ofinterpersonal influence. Using the Persuasion KnowledgeModel (Friestad and Wright 1994), four experimentsexamine the effects of consumers’ agent, persuasion andtopic knowledge on their judgments about this marketingtactic. This study aims to answer: (1) when consumersinfer ulterior sales motives of an undercover marketer, (2)what happens when persuasion tactics are accessible, (3)how consumers respond when they detect or are informedabout an undercover marketing attempt, and (4) how

product quality and purchase satisfaction influence theresponse to an undercover marketing attempt.

THEORY AND HYPOTHESES

It has been well documented that consumer knowl-edge is fundamental in the formation of valid perceptionsand judgments (Ratchford 2001). Consumer knowledgeis a complex and multidimensional construct that hasprofound impacts on consumers’ attitude, judgment, andbehavior (Alba and Hutchinson 1987; Brucks 1985). Thechallenge is to decompose the general dimensions andreconstruct the context-specific knowledge in a particularfield of interest.

The Persuasion Knowledge Model (Friestad andWright 1994) postulates that consumers’ three knowl-edge structures interact to shape and determine the out-comes of persuasion attempts: (1) Agent knowledge –beliefs about the traits, competencies, and goals of thepersuasion agent such as an advertiser or a salesperson;(2) Topic knowledge – beliefs about the topic of themessage such as a product, service, social cause, or acandidate; and (3) Persuasion knowledge – beliefs aboutan influence agent’s motives and tactics. Persuasion lit-erature has provided abundant evidence of the effect ofagent knowledge and topic knowledge, but less is knownabout the role of persuasion knowledge and the interac-tions among these knowledge structures.

Functionally, persuasion knowledge may influenceattitude formation about an influence agent through acognitive alarm mechanism. Persuasion knowledge alertsconsumers to detect, control and resist a variety of persua-sive attempts. In particular, Kirmani and Campbell (2004)argued that consumers actively attempt to manage amarketing persuasion interaction to achieve his or herown goals, rather than being a mere recipient or resister ofinfluence. They demonstrated that consumers choosefrom a large repertoire of strategies (i.e., seeker strategiesand sentry strategies), depending upon their goals andrelationships with the influence agent.

Previous PKM research defines accessibility as howreadily a construct is coded in terms of a given category(i.e., Campbell and Kirmani 2000; Kirmani and Campbell2004; Williams, Fitzsimons, and Block 2004). Campbell


and Kirmani (2000) manipulated the accessibility of ulte-rior sales motives by changing the timing of remarks in aninterpersonal sales interaction. A flattering remark from asalesperson before the consumer’s decision to buy ajacket was proved to activate participants’ persuasionknowledge about their sales motives. When the remarkswere made after the purchase decision, in comparison,consumers were less likely to infer the ulterior salesmotives.

Accessing persuasion knowledge can be thought ofas raising psychological “suspicion” in the context ofundercover marketing. The basic notion of suspicion isthat an influence agent’s remarks, either from a salesper-son or a stranger, will be judged by the target’s inferenceabout their motives (e.g., Fein 1996; Vonk 1998). If aconsumer thinks an influence agent’s positive remarks ontheir choices are intended to persuade them to purchase,such remarks are considered purposeful and insincere(Campbell and Kirmani 2000). For example, a typicaladult consumer understands that the social role of asalesperson is to make sales (i.e., agent knowledge), andtherefore, he or she is more likely to infer that a salesper-son has ulterior persuasion motives and is less sincere intheir positive remarks on consumers’ choices (i.e., per-suasion knowledge).

In an undercover sales context, however, an influ-ence agent by definition appears to be much less associ-ated with sales motives and acts as a stranger or a friend.As a result, the timing of positive remarks (before or afterpurchase) will not change a target’s inference about anulterior sales motive. Accordingly, we hypothesize that:

H1a: Consumers are more likely to infer to an ulteriorsales motive when an influence agent is an identi-fied salesperson than an unidentified undercovermarketer.

H1b: Timing of positive remarks before (after) the pur-chase will increase (decrease) accessibility of per-suasion knowledge when the influence agent isidentified, but will not increase (decrease) accessi-bility of persuasion knowledge when the influenceagent is unidentified.

Disclosure of Undercover Marketing

The Federal Trade Commission’s (FTC) “Endorse-ment Guidelines” concludes that “the failure to disclosethe relationship between the marketer and the consumerwould be deceptive unless the relationship was otherwiseclear from the context.” However, it remains unclear howconsumers respond to disclosure of undercover market-ing attempts. In general, consumers tend to respond nega-tively toward potentially misleading persuasive attempts,

as suggested in the advertising deception literature (e.g.,Olson and Dover 1978; Johar 1995; Darke and Ritchie2007). Disclosure, either from marketers (e.g., compara-tive advertising claims) or a third-party (e.g., media andother users), can be utilized to correct for invalid infer-ences. For example, Johar and Simmons (2000) found thatdisclosure mediates message elaboration on brand qual-ity, given particular goals and cognitive capacity of infor-mation processing. In regular interpersonal sales interac-tions, an influence agent’s association with a company isself-evident and consumers’ agent knowledge leads to ahigher level of skepticism toward salespersons’ positiveremarks.

In the undercover marketing context, however, a layperson’s agent knowledge assumes that an influenceagent has no association with a marketing purpose. Whenthe ulterior sales motive is uncovered, consumers aremore likely to feel that they are “set up” and thereforeperceive the persuasion attempt as deceptive. More spe-cifically, even if undercover promotional agents disclosethat true motive, their previous interactions could beconsidered purposeful and therefore unacceptable. Whendisclosed by a third party, undercover marketers could beconsidered even more deceptive and their behaviors unac-ceptable because they intentionally hide their motives.Given this, we hypothesize that:

H2a: Consumers perceive influence agents as more de-ceptive when persuasion knowledge is accessible.

H2b: Consumers perceive influence agents as more de-ceptive and undercover persuasive attempts as lessacceptable when agents’ sales motives are dis-closed by a third party than by the agents them-selves.

Topic Knowledge and Goal-Oriented Consumers

The PKM defines consumers’ topic knowledge asbeliefs about the topic of the message such as a product,service, social cause, or a candidate (Friestad and Wright1994). Topic knowledge could play a multiple functionalrole in the formation of perception and judgment, includ-ing specific source-related attributes (Crowley and Hoyer1994; Tormala and Petty 2004), memory (Johar,Maheswaran, and Peracchio 2006), schema (Wheeler,Petty, and Bizer 2005), previous attitudes (Cohen andReed 2006), etc. In particular, Andrews, Burton, andNetemeyer (2000) found that disclosures are particularlyeffective versus the control in reducing misperceptionwhen combined with the general ads claim for consumerswith a high level of nutrition knowledge.

The same logic can be applied in the context ofundercover marketing. The premise is that when consum-


ers are not aware of undercover marketers’ sale motivesand tactics, their attitude should rely more on topic knowl-edge about the product (e.g., quality and satisfaction) andagent knowledge (e.g., relationship and trust). However,as Kirmani and Campbell (2004) pointed out, consumertargets can be active goal-directed individuals and theirpersuasion knowledge may interact with topic and agentknowledge. Along this line, for example, Xie, Boush, andBoerstler (2007) found that consumers’ attitudes towarddeceptive advertisements were more negative when theexpected consequences of being misled were more seri-ous. In coping with undercover marketing, consumersmay judge the tactic partially based on the topic knowl-edge of potential benefits, so that the perceived deceptive-ness and their willingness to buy become two related butseparate constructs. If the product quality is high, theymay be more willing to buy the product, despite that theyconsider the marketing tactic deceptive. Moreover, whenthey are satisfied with the purchase, they may be lessresistant toward undercover marketing because they ben-efit from the persuasion episode. Thus, we hypothesizethat:

H3: Consumers consider undercover marketing tacticsmore acceptable when they are satisfied with thepurchase.

H4: Consumers are more willing to buy the demonstratedproduct if the quality is good, regardless of thepotential deceptiveness of the undercover marketingtactic.

STUDY 1

The purpose of Study 1 was to test the hypotheses thatagent knowledge could affect perceived deceptiveness.The experiment employed a 2 (Influence Agent: Salesper-son vs. Undercover marketer) X 2 (Positive RemarksTiming: before vs. after purchase decision) between-subject design. A total of ninety undergraduate marketingstudents were randomly assigned across four experimen-tal conditions.

Procedure

This experiment modified the written scenario ofCampbell and Kirmani’s (2000) Study 1, and manipulatedthe influence agents (salesperson vs. stranger couple) aswell as the flattering remarks timing (before vs. afterpurchase decision). All participants read a twelve-line(about 230 words) retail store shopping scenario on com-puter screens from a target’s perspective (you) in a com-puter lab. The scenario described a consumer looking fora jacket in a department store, who engaged in an interac-tion with either a salesperson or a stranger couple in thatstore. Participants were encouraged to imagine they werethe consumer. After reading the scenario, they were asked

questions about their perceptions of the salesperson or thecouple.

Influence agent role was manipulated by changingthe person whom a consumer talks with, either a salesper-son who is working, or a stranger couple shopping in thatstore. As in Campbell and Kirmani’s (2000) Study 1,ingratiation was made either before or after the purchasedecision.

Result

Inference to the Ulterior Sales Motive. Participantswere asked the question “Pat (or the couple) said thejacket looked great because she/he (they) was (were)trying to make a sale” on a Likert scale (1 = completelydisagree, to 7 = completely agree). As predicted, a 2 X 2ANOVA shows significant main effects of two indepen-dent variables on accessibility of sales motives. Partici-pants were much more likely to infer the ulterior salesmotives toward a salesperson (M = 5.18, SD = 1.36, n =50) than a stranger couple (M = 2.68, SD = 1.54, n = 40),F (1, 86) = 75.505, p < .001. Timing of the flatteringremarks was also significant: when the remarks weremade before the purchase decision (M = 4.35, SD = 1.94,n = 40), participants were much more likely to infer theulterior sales motives than when the remarks were madeafter the purchase decision (M = 3.84, SD = 1.85, n = 50).

Further one-way ANOVA tests show that positiveremarks before the purchase decision (M = 5.70, SD =1.13, n = 20) increased accessibility of sales motivescompared to after the purchase decision (M = 4.83, SD =1.38, n = 30) when the influence agent was salesperson, F(1, 48) = 5.52, p < .05. When the influence agent was astranger couple, timing of flattering remarks did notsignificantly change accessibility of sales motives, F (1,38) = 1.81, p > .05.

Participants in this experiment were more likely toinfer the ulterior sales motive when an influence agentwas an identified salesperson rather than a stranger couple.H1a was supported. This is consistent with most under-cover marketers’ assumption that usually consumers donot consider a “natural” conversation as a sales attempt. Inline with this, timing of the flattering remarks before thepurchase decision increased the likelihood to detect theulterior sales motive when the influence agent was anidentified salesperson, but it did not make a differencewhen the influence agent was an unidentified strangercouple. H1b was supported. The finding is consistent withthe notion that during an encounter with ordinary people(who could be undercover marketers), consumers relymore on their rudimentary agent knowledge to infer thesales motive. Persuasion knowledge tends to remain inac-cessible when an influence agent appears to have noselling mission.


The result suggests that consumers’ agent knowledgeplays an important role to infer the ulterior motive of aninfluence agent. However, this experiment did not di-rectly examine participant’s perception of undercovermarketers. The scenario only identifies the couple asstrangers, and it is not clear to the participants whether thestranger couple was actually carrying out a sales mission.

STUDY 2

The purpose of Study 2 was to test the hypothesis thataccessibility of persuasion tactics and disclosure typecould affect consumers’ perception and judgment of anundercover marketing attempt. The experiment employeda 2 (Accessibility of Undercover Marketing Tactics: prim-ing vs. no priming) X 3 (Disclosure: no disclosure vs.marketers’ self disclosure vs. third-party disclosure) be-tween-subject design. A total of one hundred and forty-eight students from undergraduate marketing classes wererandomly assigned across six conditions.

Procedure

This experiment modified the design and the writtenscenario of Study 1. Half of the participants first read a156-word short article on computer screens about under-cover marketing and Sony Ericson’s 2002 undercovermarketing campaign. As suggested by the accessibility-diagnosticity model (Feldman and Lynch 1988), anyfactor that increases the accessibility of an input shouldalso increase the likelihood with which that input will beused. Accordingly, an exposure or experience of under-cover persuasion will make the tactics-relevant memorymore ready to process, and therefore more likely toinfluence the subsequent judgment of the persuasionattempts.

All participants then read the retail store shoppingscenario as in Study 1. The scenario described the sameinteraction with a stranger couple when they made flatter-ing remarks before the purchase decision. Participantsthen were asked two questions on the same web page that“While I read the scenario, I was thinking whether thecouple are paid actors” (1 = completely disagree, to 7 =completely agree). Next, on the second page, participantsin the control condition were not informed about thecouple’s motives and only answered questions about theirperception about the couples. In the other two conditions,participants read one more sentence that says either thecouple themselves disclosed (saying they were paid topromote the jacket) or his friend Pat (who is working inthat store) disclosed that the couple were paid actors topromote the jacket. After reading the scenario, partici-pants were asked questions about their perceptions andjudgments of the couple.

Result

Perceived Deceptiveness was measured as an aver-age of three seven-point items about the influence agents:sincere/insincere, honest/dishonest, and deceptive/notdeceptive. Factor analysis using an oblique (promax)rotation extracted one factor solution and shows that thescale was unidimensional. The scale reliability is satisfac-tory, too: Cronbach’s alpha = 0.83. Acceptability wasmeasured as an average of two seven-point items aboutthe influence agents’ behaviors: reasonable/unreason-able, acceptable/unacceptable. Factor analysis using anoblique (promax) rotation extracted one factor solutionand shows that the scale was unidimensional. The scalereliability is satisfactory, Cronbach’s alpha = 0.81. Afterreading the scenario, participants were asked on the samepage that “While I read the scenario, I was thinkingwhether the couple are paid actors” (1 = completelydisagree, to 7 = completely agree). Participants in theprimed condition (M= 4.48, SD = 2.03, n =73) weresignificantly more skeptical about the couple’s motivethan those in the non-primed condition (M =3.79, SD =2.24, n = 75), t (146) = 3.68, p < .001. The manipulationwas successful.

A 2 X 3 ANOVA shows significant main effects ofaccessibility of persuasion tactics and disclosure on per-ceived deceptiveness. Participants rated the couple moredeceptive in the primed condition ((M = 4.00, SD = 1.70,n = 75) than the non-primed condition (M = 3.32, SD =1.79, n = 73), F (1, 142) = 7.54, p < 0.01. Perceiveddeceptiveness was significantly different across threedisclosure conditions (Mnondisclosure = 4.29, SD = 1.36,n = 50; MMarketer = 3.51, SDMarketer=1.49, n = 48;Mthird-party = 3.18, SDthird-party = 1.71, n = 50), F(2,142) = 7.12, p < 0.01.Mutiple comparisons (LSD)suggest that participants rated the couple more deceptivein either marketer or third-party disclosure conditionscompared with the non-disclosure condition; however,marketer disclosure was not significantly different fromthird-party disclosure at p < .05. Regarding acceptability,neither the accessibility nor the disclosure type made anysignificant difference across conditions. The marketers’self-disclosure did not make their behaviors more or lessacceptable than if a third-party disclosure did. No interac-tions were significant at p < .05.

Study 2 examined the effects of accessibility ofpersuasion tactics and disclosure type on the perceiveddeceptiveness and acceptability of an undercover market-ing tactic. The results supported the notion that when apersuasion tactic is more accessible, consumers are morelikely to consider an influence agent deceptive. H2a wassupported. Meanwhile, undercover marketers’ behaviorwas not considered acceptable no matter whether market-


ers “debriefed” at the end of the persuasion episode or athird party informed consumers the truth. In this experi-ment, the participants appeared to be unforgiving of theundercover marketers’ potentially misleading behaviors.The undercover marketer’s confession did not appear tohelp restore their credibility. H2b was not supported. Theresult of this experiment pinpointed the role of persuasionknowledge in the formation of perception and judgmentof undercover marketing. However, neither Study 1 nor 2controlled for the extent to which the conversation withthe couple would influence consumers’ decisions. Thenext study addressed this concern, and explored the effectof topic knowledge.

STUDY 3

The purpose of Study 3 was to test the hypotheses thatpurchase satisfaction and disclosure would affect con-sumers’ judgment of an undercover marketing attempt.We speculated that acceptability also depends upon howsatisfied a consumer feel about the purchase. This experi-ment employed a 2 (Satisfaction: satisfied vs. not satis-fied) X 3 (Disclosure: no disclosure vs. marketers’ selfdisclosure vs. third-party disclosure) between-subjectdesign. A total of one hundred and twenty-six under-graduate marketing students were randomly assignedacross six experimental conditions.

Procedure

This experiment modified the written scenario inStudy 2. All participants read a twelve-line retail storeshopping scenario without being primed any undercovermarketing tactic. The scenario described the same interac-tion with a stranger couple when they made flatteringremarks before the purchase decision only. At the end ofthe scenario, half of the participants were told that “A fewdays later a close friend says the wool-silk blend jacketlooks good on you”; and the other half were told that “Afew days later a close friend says the wool-silk blendjacket does not look good on you.” Next, all participantswere asked “Based on the scenario, I’m satisfied with thejacket I buy: (1: completely disagree, to 7: completelyagree) and “The conversation with the couple influencesmy decision about the jacket (1: completely disagree, to 7:completely agree).” The reason to add the last questionwas to control for the extent to which the conversationwith the couple would influence consumers’ decisions.As in Study 2, three disclosure types were equally appliedacross six conditions.

Measures were the same as those in Study 2 exceptthe manipulation check of how satisfied consumers arewith the purchase. The ANCOVA result shows that par-ticipants in the positive feedback condition (M= 5.17,SD = 1.52, n = 63) were significantly more satisfied withthe purchase than those in the negative feedback condition

(M = 3.39, SD = 1.76, n = 61), F (1,121) = 35.39, p < .001,controlling for the effect of how much influence thecouple had on the decision (which is not statisticallysignificant, F (1,121) = 2.18, p > .05). The manipulationwas successful. Also, the dependent measure scale (Ac-ceptability) was one-dimensional and Cronbach’s alpha =0.81.

Result

The 2 X 3 ANCOVA suggests that participants ratedthe couple’s behavior (M = 4.58, SD = 1.46, n = 63) moreacceptable in the satisfied condition than the dissatisfiedcondition (M = 4.05, SD = 1.41, n = 61), F (1, 117) = 5.39,p < 0.05. Participants in three disclosure conditions (M =5.034, SD = 1.01, n = 44) perceived the couple signifi-cantly different (MMarketer = 4.00, SDMarketer = 1.60,n = 38; Mthird-party = 3.86, SDthird-party = 1.45, n = 42),F (2,117) = 10.02, p < 0.001. Multiple comparisons(Turkey’s LSD) suggest that the couples’ behaviors wereless acceptable in either marketer or third-party disclosureconditions compared with the no disclosure condition;however, marketer’s self-disclosure was not significantlydifferent from third-party disclosure at p < .05. No interac-tions were found significant at p < .05.

The result suggests that participants were more ac-cepting of undercover marketing if they were satisfiedwith the product. H3 was supported. Participants in thisexperiment would not accept undercover marketers’ per-suasive attempts merely because they eventually informedconsumers of their sales motives. As in Study 2, themarketers’ self-disclosure (versus a third-party disclo-sure) did not make their behavior more or less acceptable.

STUDY 4

The purpose of Study 4 was to test the hypothesis thatproduct quality would affect consumers’ judgment of anundercover marketing attempt and intention to purchase.We speculated that consumers are more willing to buy thedemonstrated products if the product quality is good,despite perceived deceptiveness toward undercover mar-keting tactics. The experiment employed a 2 (ProductQuality: high vs. low) X 2 (Influence Agent: UndercoverMarketer vs. Tourists) between-subject design. A total ofeighty-two students from undergraduate marketing classeswere randomly assigned across four conditions.

Procedure

This experiment used a new scenario reflecting SonyEricson’s 2002 undercover marketing campaign. All par-ticipants read a thirteen-line written scenario on computerscreens. The scenario described a similar interaction witha couple posing as tourists when they asked people to takepictures of them in a downtown area during a tourism


season. Participants were told that besides taking pictures,there was an interaction between them (you as the partici-pant) and the couple about features of the new fancycamera. Participants then were asked two questions on thesame page that “The couple is promoting the sales of thecamera (1: completely disagree, to 7: complete agree).”Next, on the second page, half the participants were toldthat the quality of the camera was high and the other halfwere told the quality was low. Then all participants an-swered questions about their perception and intention tobuy the camera.

The influence agent role was manipulated by thesentence at the end of the first webpage in the scenario.Half the participants read that after the interaction, “Imag-ine that after shopping at Nordstrom, you see that coupleshowing the camera to others in another street”; and theother half of the participants did not read that sentence. Atthe second page, half were told that “A week later you readabout the camera from a credible source. The overallrating is pretty high.” And the other half were told that “Aweek later you read about the camera from a crediblesource. The overall rating is pretty high.” Next, all par-ticipants were asked “Overall, the camera is good: (1:completely disagree, to 7: completely agree).”

Result

ANOVA shows that both manipulations were suc-cessful. Participants in the experimental condition (M =5.44, SD = 1.10, n = 41) rated the camera quality signifi-cantly higher than those in the control condition (M =3.71, SD = 1.68, n = 41), F (1, 80) = 30.63, p < .001.Participants in the undercover marketer condition (M =5.70, SD = 1.64, n = 43) rated that the couple has asignificantly higher sale motive than those in the touristcondition (M = 3.92, SD = 2.15, n = 39), F (1, 80) = 17.90,p < .001. Perceived deceptiveness measure was the sameas in early experiments, unidimensional with highCronbach’s alpha = 0.88. Intention to buy was measuredby “Based on the scenario, I’d like to buy the camera: (1:completely disagree, to 7: completely agree).”

As expected, ANOVA result shows participants inthe undercover marketer condition (M = 3.56, SD = 1.31,n = 43) rated that the couple were more deceptive thanthose in the tourist condition (M = 5.01, SD = 1.43, n = 39),F (1, 78) = 23.57, p < .001. However, no interaction wassignificant at p < .05. In comparison, participants’ inten-tion to buy in the high quality condition (M = 3.42, SD =.23, n = 41) was significantly higher than those in the lowquality condition (M = 2.19, SD = .23, n = 41), F (1, 78) =13.96, p <.001. As expected, the influence agent role wasnot significant, p > .05. No interaction was found signifi-cant at p < .05.

The result suggests that although participants consid-ered the undercover marketing tactic deceptive, they weremore likely to buy the product if the quality is high. H4was supported. However, it is noticeable that the mean ofpurchase intention was below the scale midpoint (4),which indicates that the overall purchase intention in anundercover marketing situation was low.

GENERAL DISCUSSION

Combined, the four experiments suggest that a con-sumer is not usually ready to resist undercover marketing.Compared with a salesperson, an undercover marketer isless likely to be detected if behaving properly. However,consistent with the PKM predictions, consumer persua-sion knowledge of an influence agent’s motives andtactics has a significant impact on perceived sincerity andacceptability of an undercover marketing attempt. Whenconsumers infer an ulterior sales motive or readily accessthe information about undercover marketing tactics, theyare more likely to detect and resist undercover marketers.Once uncovered, marketers cannot restore the perceivedsincerity by fully informing consumers afterward. Fur-ther, when product quality is higher and they are satisfiedwith the purchase, they may be less resistant towardundercover marketing because they benefit from the per-suasion episode.

There are several marketing implications from thisstudy. When marketers consider going undercover, therecould be a lot of risks: whether their target is vulnerableenough without much persuasion knowledge, whether aspecific tactic is well executed to avoid inference to thesales motives, whether the product quality is good enough,and whether a prospective buyer would be satisfied withthe purchased afterwards. Besides, undercover marketersface ethical challenges. Based on the finding of this study,there seems that their tactics will be perceived as decep-tive and not acceptable, at least in the sample population.

The current research is subject to the usual limitationsof lab experiments based on written scenarios. First, anexperimental study requires us to trade off the richness ofreal-world persuasion with discovering a single piece ofisolated effect by manipulation. An interpersonal persua-sion attempt is often an iterative dynamic process. Asimplified written scenario may tell us what could happen,rather than what will happen. Second, this study aims toincorporate several broad theoretical constructs: knowl-edge, trust, acceptability and satisfaction. The next step isto refine these constructs and better present them in anexperiment setting. Third, a continuous improvement ofthe scenario approach to this complicated process couldalways be a goal of follow-up studies.


Despite the limitations, the present research is one ofthe first empirical studies to address basic questions aboutconsumers’ responses to undercover marketing attempts.

This is a small but important step forward in understand-ing the effect of rapidly changing undercover marketingtechniques.

REFERENCES

Alba, Joseph W. and J. Wesley Hutchinson (1987), “Di-mensions of Consumer Expertise,” Journal of Con-sumer Research, 13 (March), 411–54.

Andrews, J. Craig, Scot Burton, and Richard G. Netemeyer(2000), “Are Some Comparative Nutrition ClaimsMisleading? The Role of Nutrition Knowledge, AdClaim Type and Disclosure Conditions,” Journal ofAdvertising, XXIX (3).

Armstrong, Gary M., Metin N. Gurol, and Frederick A.Russ (1979), “Detecting and Correcting DeceptiveAdvertising,” Journal of Consumer Research, 6 (De-cember), 237–46.

Boush, David M., Marian Friestad, and Gregory M. Rose(1994), “Adolescent Skepticism Toward TV Adver-tising and Knowledge of Advertiser Tactics,” Jour-nal of Consumer Research, 21 (June), 165–74.

Brucks, Merrie (1985), “The Effects of Product ClassKnowledge on Information Search Behavior,” Jour-nal of Consumer Research, 12 (June), 1–16.

Campbell, Margaret C. and Amna Kirmani (2000), “Con-sumers’ Use of Persuasion Knowledge: The Effectsof Accessibility and Cognitive Capacity on Percep-tions of an Influence Agent,” Journal of ConsumerResearch, 27 (June), 69–83.

Cohen, Joel B. and Americus Reed II (2006), “A MultiplePathway Anchoring and Adjustment (MPAA) Modelof Attitude Generation and Recruitment,” Journal ofConsumer Research, 33 (1), 1–15.

Crowley, Ayn E. and Wayne D. Hoyer (1994), “AnIntegrative Framework for Understanding Two-SidedPersuasion,” Journal of Consumer Research, 20(March), 561–74.

Darke, Peter R. and Robin J.B. Ritchie (2007), “TheDefensive Consumer: Advertising Deception, De-fensive Processing, and Distrust,” Journal of Mar-keting Research, XLIV (February), 114–27.

Fein, Steven (1996), “Effects of Suspicion on AttributionalThinking and the Correspondence Bias,” Journal ofPersonality and Social Psychology, 70 (June), 1164–84.

Feldman, Jack M. and John G. Lynch, Jr. (1988), “Self-Generated Validity and Other Effects of Measure-ment on Belief, Attitude, Intention, and Behavior,”Journal of Applied Psychology, 73 (3), 421–35.

Friestad, Marian and Peter Wright (1994), “The Persua-sion Knowledge Model: How People Cope withPersuasion Attempts,” Journal of Consumer Re-

search, 21 (June), 1–31.Gordon, Randall A. (1996), “Impact of Ingratiation on

Judgments and Evaluations: A Meta-Analytic Inves-tigation,” Journal of Personality and Social Psychol-ogy, 71 (July), 54–70.

Johar, Gita Venkataramani (1995), “Consumer Involve-ment and Deception from Implied AdvertisingClaims,” Journal of Marketing Research, XXXII(August), 267–79.

____________ and Carolyn J. Simmons (2000), “TheUse of Concurrent Disclosures to Correct InvalidInferences,” Journal of Consumer Research, 26(March), 307–22.

____________, Durairaj Maheswaran, and Laura A.Peracchio (2006), “Mapping the Frontiers: Theoreti-cal Advances in Consumer Research on Memory,Affect, and Persuasion,” Journal of Consumer Re-search, 33 (June), 139–49.

Kaikati, Andrew M. and Jack G. Kaikati (2004), “StealthMarketing: How to Reach Consumers Surrepti-tiously,” California Management Review, 46 (4), 6–24.

Kirmani, Amna and Margaret C. Campbell (2004), “GoalSeeker and Persuasion Sentry: How Consumer Tar-gets Respond to Interpersonal Marketing Persua-sion,” Journal of Consumer Research, 31 (Decem-ber), 573–82.

Olson, Jerry C. and Philip A. Dover (1978), “CognitiveEffects of Deceptive Advertising,” Journal of Mar-keting Research, XV (February), 29–38.

Ratchford, Brian T. (2001), “The Economics of Con-sumer Knowledge,” Journal of Consumer Research,27 (March), 397–411.

Tormala, Zakary L. and Richard E. Petty (2004), “SourceCredibility and Attitude Certainty: A MetacogntiveAnalysis of Resistance to Persuasion,” Journal ofConsumer Psychology, 14 (4), 427–42.

Vonk, Roos (1998), “The Slime Effect: Suspicion andDislike of Likable Behavior toward Superiors,” Jour-nal of Personality and Social Psychology, 74 (4),849–64.

Vranica, Suzanne (2005), “Getting Buzz Marketers toFess Up,” in Wall Street Journal, 245 (28), (February9).

Wheeler, S. Christian, Richard E. Petty, and George Y.Bizer (2005), “Self-Schema Matching and AttitudeChange: Situational and Dispositional Determinantsof Message Elaboration,” Journal of Consumer Re-search, 31 (March), 787–97.


Williams, Patti, Gavan J. Fitzsimons, and Lauren G.Block (2004), “When Consumers Do Not Recognize‘Benign’ Intention Questions as Persuasion At-tempts,” Journal of Consumer Research, 31 (Decem-ber), 540–50.

Xie, G., David, M. Boush, and Courtney N. Boerstler

(2007), “Consumer Marketplace Intelligence andVulnerability: Advertising Deception, DeceptionKnowledge, and Expectancies,” Paper presented atthe annual conference of Society of Consumer Psy-chology, Las Vegas, NV.

For further information contact:Guang-Xin Xie

Department of MarketingUniversity of Oregon

1208 University of OregonEugene, OR 97403–1208

Phone: 541.346.1438Fax: 541.346.3341



ROCKY SALES-MARKETING INTERFACE: AN EXPLORATORY

EXAMINATION OF SYMPTOMS, CAUSES, EFFECTS,

AND REMEDIES

Avinash Malshe, University of St. Thomas, St. PaulRavipreet S. Sohi, University of Nebraska – Lincoln

SUMMARY

In spite of the acknowledged importance of examin-ing the sales-marketing interface, barring a few notableexceptions, research in this area is scant. Further, scholarswho have studied this area have noted that the interfacebetween these two functions remains rocky – specifically,Dewsnap and Jobber (2000, p. 109) note, “the relationship[between these two functions] is characterized by a lack ofcohesion, poor coordination, conflict, non-cooperation,distrust, dissatisfaction, and mutual negative stereotyp-ing.”

Though we understand that this interface could berocky, it is possible that the two functions may not feud inthe open all the time. It is plausible that there might besome brewing troubles between sales and marketing thathave not (yet) reached feudal proportions. These brewingtroubles may hinder organizational activities such asmarket sensing, strategy creation, and strategy executionby hampering free flow of information and collectiveresponsiveness. Research examining the sales-marketinginterface till date has not identified the nature of suchtroubles and the symptoms that might manifest thesebrewing troubles that if not identified and looked into mayreach feudal proportions in future.

This study is the first step in the exploration of thisarea. Using qualitative methodology and interview datacollected from 38 sales and marketing managers fromdifferent industries, this study identifies six symptomsthat may manifest brewing troubles between sales andmarketing. The main theme of this paper – identificationand explication of six symptoms is an important contribu-tion of this study. These symptoms bring to the surface thelatent tensions (Gaski 1984; Kruglanski 1970; Pondy1967) between these two functions. In doing so, this studyextends literature on sales-marketing interface and opensup a new research avenue. This study also pushes theenvelop further by identifying (a) the root causes/under-

lying troubles that these symptoms may manifest, (b) howthese symptoms may flag for marketers how strategyimplementation may be affected, and (c) what remedialactions marketers may take in such situations. By provid-ing a comprehensive picture of the causes – symptoms –effects – and remedies, this study also enriches our under-standing of the sales marketing interface. Further, a lookat many of the remedies indicates that they emphasize theneed for greater interfunctional communication (Li andCalantone 1998; Ruekert and Walker 1987) and collabo-ration (Atuahene-Gima 2007). The study thus indicatesthat these symptoms may serve as “triggers” withinorganizations bringing forth the need to enhance theseactivities in order to increase interfunctional integration.Extant research has not examined the effect of sales-marketing interface on marketing strategy implementa-tion. By highlighting how marketers may use certainsymptoms to predict the fate of future implementation ofmarketing strategies, this study also helps us take the firststep toward situating the literature on sales-marketinginterface in the context of strategy implementation.

Managers may find the study results useful. Specifi-cally, identification of six symptoms may help marketingmanagers spot the brewing troubles in the sales marketinginterface and take proactive measures to address thosetroubles. Further, explication of how the underlyingtroubles may affect strategy implementation may providethem with understanding of how they may address eitherthe root problems or work to contain its effects on strategyimplementation processes. Last, the overall findings ofthis study may also provide them with a comprehensivepicture linking the problems between sale and marketingto the symptoms, effects, and remedies.

Future research can look at many different areas suchas (a) identifying a broader repertoire of symptoms and (b)examining the interaction between different causes andtheir symptomatic manifestation.

For further information contact:Avinash Malshe

Department of MarketingUniversity of St. Thomas

2115 Summit Ave.St. Paul, MN 55105

Phone: 651.962.4287 ♦ E-Mail: [email protected]


THE CONTINGENT VALUE OF SALES PARTICIPATION

IN MARKETING DECISION-MAKING ON

ORGANIZATIONAL PERFORMANCE

Gabriele Troilo, Bocconi University, ItalyLuigi M. De Luca, Aston Business School, United Kingdom

Paolo Guenzi, Bocconi University, Italy

SUMMARY

Aim of the Study

In market oriented companies the Sales unit partici-pates in making most of marketing decisions. The com-monly held belief says that participation is always benefi-cial to the firm because it increases comprehensiveness ofdecisions and enhances consensus and commitment totheir implementation. However, organization theory, strat-egy making theory, and also marketing contingency theorysuggest that benefits are often counteracted by detrimen-tal effects. These effects have been traditionally over-looked by marketing scholars. In this study we explorefactors that can enhance the effectiveness associated tohigher participation of the Sales unit in marketing deci-sion-making along with factors that can accentuate thelimited efficiency produced by increased participation. Inparticular, we investigate the moderating role of organiza-tional context – namely, organizational culture, values,and structure – on the relation between Sales participationin marketing-decision making and market performance.

Definitions and Hypotheses

Organizational culture is viewed in terms of customeroriented culture, defined as the set of beliefs that puts thecustomer’s interest first, while not excluding those of allother stakeholders, in order to develop a long-term prof-itable enterprise. We expect that customer oriented cul-ture enhances the effects of Sales participation in market-ing decision-making on organizational market perfor-mance.

We conceptualize values in terms of distributivejustice defined as the perception of fairness of the alloca-tion of. We argue that an allocation of rewards betweenMarketing and Sales which is not informed by the prin-ciples of distributive justice negatively impacts the effectsof Sales participation in marketing decision-making onorganizational performance.

Organizational structure is viewed in terms of func-tional specialization, namely, the clear division of labor,roles and responsibilities among different functional units.We argue that specialization prevents jurisdictional ambi-

guity and conflicts which make implementation of deci-sions suboptimal and increase inefficiencies in participa-tive decision-making.

H1: The joint occurrence of Sales participation in market-ing decision-making and customer oriented culturehas a positive effect on organizational market perfor-mance.

H2: The joint occurrence of Sales participation in market-ing decision-making and distributive injustice has anegative effect on organizational market performance.

H3: The joint occurrence of Sales participation in market-ing decision-making and functional specializationhas a positive effect on organizational market perfor-mance.

Methods and Results

A survey was conducted among 870 managers at-tending different courses for Marketing and/or Salesexecutives at a major business school in Europe the datacollection resulted in 326 usable questionnaires, achiev-ing a response rate of 37.5 percent. Informants belongedto Marketing (53.1%) or Sales (46.9%). Thirty-one pointsix percent of the companies sell pure goods, 17.6 percentpure services, 51.8 percent a combination of goods andservices. We measured Sales participation in nine market-ing decisions, ranging from new product development totactical pricing and Promotions. The other study con-structs were measured with 5-point Likert scale itemsdrawn from the literature. We used CFA to assess theproperties of the measurement model and OLS regressionfor hypothesis testing. We controlled firm size, long termorientation, Marketing-Sales collaboration, influence ofother functional units, and superior customer value.

Although no specific hypothesis is advanced on itseffect on market performance, Sales participation in mar-keting decision making seems to have a marginally sig-nificant negative effect on market performance, suggest-ing the actual presence of some “negative side” of partici-pation itself. The interaction between Sales participationin marketing decision-making and customer oriented cul-ture is not significant. The interaction between Sales


participation in marketing decision-making and distribu-tive injustice has a negative and significant effect onorganizational performance. Finally, the interaction be-tween Sales participation in marketing decision-makingand functional specialization has a positive and signifi-cant effect on organizational performance. Thus H2 andH3 are supported, H1 is not.

Discussion and Managerial Implications

This study contributes to the marketing literature inseveral respects. First, it enriches the contingency view ofmarketing organization by suggesting that organizationalcontext plays and important role in influencing the effectsof participative marketing decision-making. Second, wegive emphasis to organizational factors, namely, distribu-tive justice, and functional specialization, which havebeen under-investigated in prior marketing research. Third,consistent with recent research on MarketinG-Sales inter-face, we add new evidence to the role that the interactionsof these two departments have on organizational perfor-mance. Fourth, our results confirm that marketing activi-ties are cross-functionally dispersed and that this disper-sion does not imply that the influence of the two units

within the firm is being reduced. Last, we advance pastresearch on Sales participation in marketing decision-making, going beyond a focus on single decisions or ongeneric participation to decision-making.

Our study also has several managerial implications.Our study informs managers that any attempt to increaseSales participation in marketing decision-making shouldbe pursued after an assessment of the organizationalcontext where it will take place. An additional implicationfor managers regards the design and management of therelationship between the two functions. Our findingsconfirm that both Marketing and Sales actively participatein marketing decision-making, with Marketing having apredominant influence on new product development, ex-pansions in new geographical markets, and advertisingcampaigns, whereas Sales predominates in distribution,tactical pricing and promotion decisions. Managers re-sponsible for the organization of the marketing functionshould carefully design integration mechanisms that al-low the effective collaboration of the two units and theestablishment of a cooperative relational atmosphere.References available upon request.

For further information contact:Gabriele Troilo

Bocconi University and SDA Bocconi Graduate School of ManagementVia Bocconi, 8

Milano, Italy, 20136Phone: +39.02.5836.6518



A CONCEPTUAL MODEL OF ANTECEDENTS AND CONSEQUENCES

OF CONSUMER UNCERTAINTY: MODEL DEVELOPMENT

AND CRITICAL ASSESSMENT

Edward Shiu, University of Strathclyde, United KingdomLouise Hassan, University of St. Andrews, United Kingdom

Gianfranco Walsh, University of Koblenz – Landau, Germany

SUMMARY

Introduction

Consumers have been found to experience uncer-tainty in decision-making in terms of information, avail-ability and choice. A fuller understanding of consumeruncertainty is necessary for manufacturers and serviceproviders to overcome this dysfunctional state of mindwhich deters consumers from engaging in mutually ben-eficial consumption experiences. Urbany et al. (1989)examine uncertainty and its relationship with informationsearch and postulate a two-dimensional representation ofconsumer uncertainty. They concluded with a call forfurther research to explore an additional dimension ofevaluation uncertainty as well as identifying antecedentsto uncertainty. To the best of our knowledge, this call hasyet to be answered.

Our paper makes four contributions to the literature.First, we take up Urbany et al.’s (1989) call to develop andoperationalize a new dimension of evaluation uncer-tainty. Second, we introduce a new conceptual modelenveloping the multidimensional uncertainty constructswithin a consumer decision framework, taking into ac-count of antecedents and consequences. Third, we pro-pose a set of antecedents (complexity, ambiguity, conflict,and credibility) of uncertainty. Although some of thesefactors have been separately examined in the literature,this paper brings together all four of these antecedentswithin a single framework underpinning consumer uncer-tainty. Fourth, Urbany et al. (1989) examined the effect ofuncertainty on information search, we extend their workby simultaneously considering the impact of uncertaintyon intention to purchase.

Model Overview and Hypotheses

Consumer decision making can be conceptualized asa sequential process comprising stages from informationacquisition to high order mental processing through tojudgment and choice. The consumer seeks and acquiresinformation and processes the information by interpret-ing, evaluating, and synthesizing it with their existingknowledge. Based on this processing, an assessment orjudgment is formed regarding the relative merits of alter-

natives and a consumption choice is made. McGuire(1972) posits that the results of the processing at eachstage have a direct impact on subsequent stages. Hence,the quality of the early stage of information acquisition iscrucial to allow effective evaluation and synthesis of theinformation pertinent to decision making. According toJacoby et al. (1994, p. 292), “the amount, content, andsequence of information input are expected to impact onthe formation, development, and change of such higher-order processes as beliefs, attitudes, evaluations, images,impressions, and intention.” Thus factors exerting a stronginfluence over consumer’s ability to acquire and processinformation are central to understanding the consumer’smental state of uncertainty. Factors within the first stageof information acquisition which have been identified inprevious research to have a direct impact on high ordermental states such as uncertainty are ambiguity, complex-ity, conflict, and credibility. Specifically, we propose

H1a: Ambiguity positively impacts knowledge uncer-tainty.

H1b: Ambiguity positively impacts choice uncertainty.

H1c: Ambiguity positively impacts evaluation uncer-tainty.

H2a: Complexity positively impacts knowledge uncer-tainty.

H2b: Complexity positively impacts choice uncertainty.

H2c: Complexity positively impacts evaluation uncer-tainty.

H3a: Conflict does not impact knowledge uncertainty.

H3b: Conflict positively impacts choice uncertainty.

H3c: Conflict positively impacts evaluation uncertainty.

H4a: Credibility negatively impacts knowledge uncer-tainty.

H4b: Credibility negatively impacts choice uncertainty.


H4c: Credibility negatively impacts evaluation uncer-tainty.

Faced with multiple options and incomplete informa-tion, consumers will either proceed with the decision-making process and make a purchase decision or searchfor information to reduce their uncertainty to a tolerablelevel (Mitchell et al. 2005; Mitchell and Papavassiliou1997). Uncertainties regarding the outcomes of purchasesare perceived as risks by the consumer and likely to resultin the consumer conducting information search with aview to risk reduction (Srinivasan and Ratchford 1991),however, the level of information search prompted byuncertainty is likely to vary depending on the type ofuncertainty perceived. In terms of purchase decisions,Greenleaf and Lehmann (1995) also proposed that con-sumers would delay decision making because of uncer-tainties and perceived risks.

We draw on Expected Utility Theory (Savage 1954;Schoemaker 1982) to develop our consequences-relatedresearch propositions. Broadly speaking, Expected Util-ity Theory suggests that consumers are motivated tomaximize their expected utility. The theory predicts thatconsumers make either risk-averse or risk-neutral choicesdepending upon the magnitude of the stakes relative totheir total wealth. When facing uncertainty, consumers

tend to evaluate the consequences both of choosing onealternative and of foregoing the other (e.g., Shafir andTversky 1992). Expected utility theory predicts that theconsumer should maximize subjective expected utility byselecting the alternative with the highest value. In order toidentify that option, uncertainty must be minimized. Weargue that utility can be maximized and uncertainty mini-mized by either increasing or decreasing informationsearch. If the consumer perceives a necessity to engage infurther information search to cope with uncertainty, apostponement of the purchase decision is likely. Wetherefore expect a high level of uncertainty to lead toinformation search and low intention to purchase. Hence

H5a: Knowledge uncertainty positively impacts con-sumer search behavior.

H5b: Knowledge uncertainty negatively impacts theconsumer’s purchase intention.

H6a: Knowledge uncertainty positively impacts con-sumer search behavior.

H6b: Knowledge uncertainty negatively impacts theconsumer’s purchase intention.

References are available upon request.

For further information contact:Edward Shiu

University of Strathclyde173 Cathedral Street

Glasgow G4 0RQUnited Kingdom

Phone: +44.141.548.4318Fax: +44.141.552.2802



CLARITY AND CUSTOMER VALUE

Joe Urbany, University of Notre DameWilliam Bearden, University of South Carolina

Raymond Kordupleski, Customer Value Management, LLC

ABSTRACT

Humans are driven to seek certainty and to avoidambiguity. These objectives are often difficult to achievein the marketplace, where there are many sources ofuncertainty, particularly in markets with complex goodsand services like health care, banking, and financialservices. Here, we suggest that, in those basic humanvalues, there exists an opportunity for firms to createunique, difficult-to-copy customer value by reducing un-certainty in consumption experiences. We define uncer-tainty broadly, and define clarity as a commodity that afirm produces in helping consumers resolve uncertainty.We consider why clarity may be an important contributorto customer value, yet undervalued by firms, who mayunderestimate the returns from creating clarity and focuslargely on the incremental costs. Data illustrating thepotential importance of clarity in consumer evaluationsare presented, and research directions discussed.

INTRODUCTION

Human beings are driven to seek certainty in judg-ment and decision-making (Einhorn and Hogarth 1985).In both consumer and organizational contexts, more am-biguous inputs are often outweighed in decision-makingby cues that can be judged with greater certainty (Deighton1984; Hoch and Ha 1986; Montgomery, Moore, andUrbany 2005). People tend to avoid ambiguity in deci-sion-making (Cyert and March 1992; Adams, Day, andDougherty 1998), instead seeking more certain and easilyaccessible information (Culnan 1983; Day and Wensley1988; O’Reilly 1982).

The drive for certainty often leads to frustration forconsumers in a marketplace that can be confusing andunfriendly, particularly in industries such as financialservices, insurance, and healthcare. Rankin (2004) notessubstantially rising consumer complaints about bankingservices and suggests that in considering investments,“. . . decision-making is confusing and stressful (and)customers are often later surprised by the penalties in-curred . . .” (see also Scott 2004). Two-thirds of commer-cial insurance buyers seek “better information on pricesand terms being transacted” in the insurance markets(Bradford 2005). The LA Times cites a Towers-Perrinsurvey in stating that 80 percent of respondents ques-tioned about healthcare desire more information aboutprices (Kaiser Network Daily Reports 2005), while

Markard and Holt (2003) report that consumers express asignificant concern for additional information in energyprovider decisions.

It is the clash of humans’ desire for certainty withmarketplace practices that provides the starting point forthis paper. Our objective is to consider the impact for firmsof seeking to reduce uncertainty for consumers. We sug-gest that “clarity” is a commodity that a firm may producethat represents value for customers in the reduction ofuncertainty. Although the marketing discipline (modelingthe foundational discipline psychology) has invested agreat deal of research emphasis into the study of howconsumers behave in the face of uncertainty, there hasbeen little research to explore the effects of reducinguncertainty. Here, we first consider various definitions ofuncertainty around the stages of the consumption process.Further, we consider the underlying theory for why con-sumers may value uncertainty-reduction, followed by anempirical exploration of one firm’s efforts to reducecustomer uncertainty on several dimensions.

DEFINING UNCERTAINTY AND CLARITY

While uncertainty is generally understood to reflectthe difficulty of prediction or doubt about current or futurecircumstances, it has a more precise meaning in consumerand choice theory. As related to consumer experiences,uncertainty has many sources. Immediately below, weconsider three general categories of uncertainty, followedby a discussion of how consumers may experience thesetypes of uncertainty in different stages of the consumptionprocess.

Variance

The concept of uncertainty in variance is fundamen-tal to search theory (Stigler 1961; Moorthy, Ratchford,and Talukdar 1997; see also Izquierdo and Luis R.Izquierdo 2007). The greater the expected variance inalternatives that the consumer must search across, thegreater the uncertainty the consumer experiences in search.The variance concept is common to information theory,information integration theory, and other theories of hu-man judgments. The intuition behind uncertainty in thiscontext is that the larger the array of outcomes, the morechallenging it will be to find one in an acceptable set thatfits personal preference.


Uncertain or Incomplete Knowledge

There is also uncertainty caused by incomplete knowl-edge of a phenomenon. This may relate to missing at-tribute information, or potential errors in estimating thatinformation, uncertainty about what attributes to includeand the consumer’s general confidence about what theyknow. In the search literature, the most relevant conceptwithin this category is the consumer’s subjective or per-ceived knowledge (e.g., Brucks 1985).

Uncertain Preference

There is a distinctive type of uncertainty that centersaround evaluation and choice, which Wang, Venkatesh,and Chatterjee (2007) label preference uncertainty, butwhich is also referred to as choice uncertainty (Urbany,Dickson, and Wilkie 1989). In making a choice, a deci-

sion-maker must integrate information (complete or not)and deal with some degree of ambiguity as to whichchoice option they prefer. Evidence suggests that the moresimilar the options are (cf., Häubl, Dellaert, and Usta2007; Lanzetta 1963) and the greater the degree of at-tribute conflict (superiority on some attributes but inferi-ority on others; Fischer, Luce, and Jia 2000), the greaterthe preference uncertainty exists. Further, we can extendthis to suggest that uncertainty about personal values thataffect decision weights, as well as a lack of knowledge ofappropriate decision heuristics to use, may also producepreference uncertainty.

Table 1, patterned after the stages in the consumptionprocess articulated by LaSalle and Britton (2003), illus-trates the different types of uncertainty that may be expe-rienced in different stages of consumption. It additionallyprovides examples of what some firms have done to

TABLE 1

Type of Uncertainty by Stage of Consumption Process

Stage of

Consumption

Process Type of Uncertainty What Firms May Do to Reduce Uncertainty

Search • Variance: perceive a wide • Progressive Insurance: Provides quotes on insurancerange in available options policies from competitive firms.

• Incomplete information

Evaluation/Choice • Incomplete information about • Exempla Healthcare: publishes quality metrics (Meyersattributes/brands. 2005)

• Values/weights • Expert systems help consumers evaluate alternatives• Decision hueristics (how to (Haubl JCR)

choose)• Perceived risk

Integration/ • Incomplete information: • Apple – simplicity in product design for usageConsumption process, instructions, how • Cigna consumer intranet shares healthcare info (e.g.,

to use the product, ongoing drugs) and customizing capability (Woehr 2006)billing and terms • Healthcare firms simplifying billing, providing online

account management (Hammer 2006)• FedEx – online tracking system (Song 2003)• Kemp Little (law practice that has a website allowing a

client to see stage of project and current billing)• Ketchum PR – web site to show all stages of project

execution• We Energies – for every service call, phoning ahead to

give an estimate of arrival time and phoning to follow-up and check on satisfaction

Extension • Incomplete information • Southwest Airlines’ simplified loyalty program RapidRewards. One point for each leg flown; 16 points (8round trips) get a free flight.


address uncertainty. Beginning at the top of the table, thesearch process has been extensively modeled in theinformation economics literature and, as noted, the con-cept of uncertainty is largely a function of perceivedvariance in the distribution over which the consumersearches (Kohn and Shavell 1974; Ratchford 1982;Weitzman 1979). Progressive Insurance has addressedthese situations by providing rate information for bothitself and competitors on its web site, clarifying both theprice distribution range and particular firms’ positions onthat distribution (cf., Moorthy et al. 1997).

In the evaluation/choice stage, uncertainty is a func-tion of the completeness of information available abouteach option (cf., Ross and Creyer 1992), as well as theattribute weights used in the assessment. In addition,consumers may lack knowledge about how to chooseeven if they are reasonably confident about what theyknow (Urbany et al. 1989). Expert systems are beingoffered online in many product or service categories thatseek to reduce the uncertainty experienced in this stage(Urban et al. 2000; Urban 2006).

The stage integration/consumption considers thatconsumer’s taking possession of the product or serviceand the time and process over which it is consumed.Several different types of uncertainty may be experiencedover this stage, surrounding product use, ongoing billing,and service interactions with the firm. Apple and othersprovide examples of firms that seek to simplify andprovide information for consumers regarding product useand service to enhance customer value.

Finally, extension refers to a firm’s efforts to buildrelationships with consumers. While there may be severaldimensions, it is most commonly captured today in cus-tomer relationship management programs and loyaltyprograms, which are often complex and frustrating (Stauss,Schmidt, and Schoeler 2005). Southwest Airlines is anexception, having created the simplest loyalty program inthe industry, with 16 individual routes or legs flown(regardless of miles) qualifying the flyer for a free flight.

CONSUMER VALUE POTENTIALLY

OBTAINED FROM CLARITY

To this point, we have identified potential sources ofuncertainty at several points in the consumer’s consump-tion experience. Yet, we have yet to explore why consum-ers may value clarity. Growing evidence and discussionare leading to the conclusion that the openness and clarityof information shared are key drivers of customer satis-faction and loyalty, particularly in web-based environ-ments (Kim and Kim 2006; Urban et al. 2000). Below, webriefly explore the driving forces behind the potentialvalue that consumers may place on clarity.

Economic Value

The denominator of the consumer’s implicit calcula-tion in judging value-for-the-money is “give,” or theconsumer’s sense of the value of what she/he pays (broadlydefined) for the goods or services obtained. To the extentthat the firm provides greater clarity or reduces uncer-tainty surrounding the terms and pricing of the product orservice as needed, it may enhance the consumer’s sense ofoverall value received in the purchase. A more confidentassessment of value might be obtained as well by creatinggreater certainty around the numerator (i.e., the “get,” orbenefits received from the product or service).

Cognitive Value

Cognitive returns to clarity may best be captured inunderstanding and control, both core human values. Pro-viding people the reasons behind requests or actionsenhances understanding and appears to have a significantimpact on their responses to those efforts to enhanceclarity. Taylor and Bower (2005), for example, find thatcompliance with instructions in product usage signifi-cantly increased in both lab and field settings when theinstructions added a simple phrase explaining the reason-ing behind usage instructions for a pesticide product (i.e.,“you should wear gloves” vs. “you should wear gloves toprevent skin irritation or staining”). Such an effect ofunderstanding (or the absence thereof) may explain con-sumers’ tendency to discount a brand on attribues forwhich there is missing information (Ross and Creyer1992), and may also be reflected in Kwong and Soman’s(2006) recent findings that spending of earned loyaltypoints is strongly associated with the cognitive ease withwhich points can be translated into cash. The importanceof perceived control (cf., Langer 1975; Parker 1993), isreflected in the results of Hui and Zhou (1996), whoconclude that the familiar finding of increased satisfactionwhen given information on waiting time in service en-counters is better explained by the fact that it gave theconsumer a greater sense of control during the waitingtime (as opposed, for example, to improving accuracy oftime perception). Finally, it is possible that significanteffort in seeking to provide clarity may be seen by con-sumers as a signal of the firm’s quality, provided it reflectsa distinctive effort relative to competitors (e.g., which isassumed to be a default inference in the case of advertisingspending (Kirmani and Wright 1989; Kirmani 1990) andthat the signal is bonded (Ippolito 1990; Kirmani and Rao2000).

Relational, Affective Value

There may be another “layer” to the effects of clarity.To the extent that a firm has undertaken an effort toprovide new insight or clarity uniquely to customers, it


will often reflect a response to having listened to customerneeds. That is, greater openness in a relationship maylikely lead to a customer’s sense that the firm has a deeperunderstanding of her/his perspective, and greater per-ceived care and compassion about customers. There maybe two underlying dimensions to the value here, eachmoving beyond the more functional consequences aboveof economic value certainty, control, understanding, andperceived quality. First is the consumer’s sense that thefirm is listening and reponding uniquely to the consumer’sneeds; a sense of empathy and appreciation that the firmcares. Second, though, is the firm’s willingness to be opento customers about issues and concerns, some of whichmay not always be flattering. There is a substantial litera-ture in psychology regarding “self-disclosure,” suggest-ing that even among strangers in a laboratory setting,greater disclosure of personal information produces greaterliking (Collins and Miller 1994). It is in part this relationaldimension of value that Eggert and Helm (2002) argueexplains the positive impact of relationship transparencyon customers satisfaction and value in business marketsthat they observe. The following section explores whetherthis and other potential sources of value emerge in firms’efforts to create clarity.

EMPIRICAL OBSERVATIONS

The basis for the empirical observations below is anelectric and gas utility that has sought to explore in severalways how clarity affects the firm’s ongoing measures ofcustomer satisfaction and value. We consider two dimen-sions of clarity examined by the firm and then brieflypresent a follow-up study that examined the potentialunderlying drivers of the effects observed.

Billing Clarity

A critical touchpoint for consumers in the utilityindustry is billing. Does clarity in billing contribute toconsumer’s perceptions of a firm’s value? One position isthat billing is simply a supplement to the core serviceprovided, and clarity in billing is just one dimension of it.In this view, the predictive value of billing clarity shouldbe dominated by other, more fundamental predictors ofvalue.

Data from two quarterly satisfaction assessmentsfrom a demographically representative sample in themidwest for the same electric utility firm provides a basisfor exploring whether clarity in billing influences con-sumer perceptions of value. The study involved a broadrange of various dimensions of satisfaction (e.g., quality,service, image, billing, price, billing), and then an aggre-gate assessment asking for overall value for the money. Asthere is a common scale, there is a significant degree ofmulticollinearity among the predictors. A factor analysisof the forty-one items identified a first factor that appears

to be scale-related, as all items loaded highly on it. Thefour remaining factors with eigenvalues above 1.0 wereeasily interpreted as reflecting satisfaction with: (1) ser-vice (7 items; e.g., knowledgeable, professional, easy todo business with, responsive, concern and caring, under-standing your needs, being knowledgeable, communicat-ing with customers; α = .96), (2) the quality of powerdelivery (6 items; e.g., reliability, safety of power deliv-ery, no spikes, no interrruptions, supply power whenneeded, respond promptly to outages, α = .93), (3) clarityof billing (3 items: easy to read, easy to understand, easyto determine exact amount owed; α = .89), and (4) price(3 items: reasonable rates, firm works to control costs,firm cares about keeping prices low; α = .95).

Scales were created by averaging the items in eachpredictor set, and then regressing them on overall value(i.e., satisfaction with whether what you get from thecompany is “worth what you pay for it”). The results inTable 2 demonstrate that, consistent with the concerns ofconsumers expressed over time, price dominates the ex-planation of value. Importantly, though, holding and otherfactors constant, billing clarity is a significant predictor ofvalue, providing as much explanatory power as Service.While it may in fact be a surprise that a dimension as minoras billing clarity contributes significantly to customerperceived value relative to more fundamental drivers ofquality, service, and price, this phenomenon is not lost inthis and other industries. Particularly given a shift towardpatients paying directly, a number of firms in the healthcareindustry have begun to provide online access to accounts,to eliminate paper statements, to simplify billing state-ments, and to consolidate bills into single statements. OneCEO notes that “hospital bills are the . . . greatest PRproblem for the nation’s acute care facilities” (Hammer2006, p. 121). In the qualitative study described below, weexplore what are the deeper drivers that make billingclarity important to consumers. Prior to that, however, weconsider another area in which the firm under study madestrides in producing clarity for consumers.

Calling Ahead on Service Calls

Consumers were expressing significant dissatisfac-tion for this firm in what might best be described as a“hostile” pricing environment, as several rate increaseshad been pushed through over a period of five years.During this time, the utility undertook a pilot program tomake follow-up calls to customers who had experiencedoutages. It was discovered in this program that consumerswho received a call explaining an outage, later rated thecompany nearly half-a scale point (about 7 percent) higheron overall customer satisfaction. Remarkably, this effectimproved satisfaction with the company’s pricing evenmore (increasing two different dimensions 13 to 16 per-cent). Intrigued, the company examined calling consum-ers ahead of service calls and following-up to assure


satisfaction after the fact for those receiving new service,and found a similar 15 to 18 percent increase in latersatisfaction ratings.

It appears that these telephone calls had weight onsatisfaction disproportionate to their size/effort. We wouldsuggest that such calls serve to reduce customers’ uncer-tainty about service delivery both pre-service and post-service. Why might it have such a dramatic effect onoverall satisfaction?

Qualitative Follow-Up Study

Subjects. A follow-up study was undertaken with 60adult Executive MBAs course at a major Midwesternuniversity, averaging 36.5 years of age and 15-year workexperience. In the interest of focusing on the potentialimportance of the different dimensions of clarity, wepresent subjects with scenarios designed to reflect firms’efforts to reduce consumer uncertainty in specific stagesof the consumption process. The idea is to obtain aconservative estimate of the importance of clarity and,more significantly, why clarity might be important toconsumers.

Method. In this exploratory study, we examined theimpact of clarity – broadly defined – at different stages ofthe consumer’s consumption experience (Table 1). Table 3presents four different contexts around which short sce-narios were built, and sample sizes. Each subject waspresented with and evaluated two of these scenarios.Scenarios were counterbalanced.

Measurement

Importance and Laddering. Each subject was askedto read through the scenario, to envision themselves in theposition of the customer described in the scenario, andthen to read the scenario a second time. After completinga second reading, subjects were asked the followingquestion:

“Think about the scenario as if you were the con-sumer in it. If all else were equal, why might ___(clarity dimension in the scenario) _________ beimportant to you in evaluating these two differentbrands?”

Laddering. The particular subject pool in this studyhad already been exposed to training on laddering, aresearch method designed to understand how customersrelate attributes to deeper personal values (cf., Reynoldsand Gutman 1988; Wansik 2002). The firm’s approach toenhancing clarity (e.g., the serviceman called ahead) isessentially the “attribute” under study. Laddering meth-odology then seeks to uncover deeper evaluation driversby drilling down into the importance of that attribute. Sothe questions followed:

Why might “pre-arrival phone call” be important toyou in evaluating these two different brands? (Theanswer will tend to be a functional consequence)

What do you get from the [functional consequence]you mentioned? Why is that important to you? (Theanswer will tend to be a psychosocial consequence)

TABLE 2

Regressions of Quality, Service, Clarity of Billing, and Price on Overall Value

(A) (B) (C) (D) (E)

Full Drop Drop Drop Drop

Predictors Model Quality Service Billing Price

Quality .08a -- .12a .10a .21a

Service .11a .13a -- .13a .30a

Billing Clarity .09b .12a .11a -- .31a

Price .68a .68a .71a .71a --

R2 .74 .71 .73 .73 .49

Dependent variable: Overall satisfaction with the service value; i.e., “Considering both quality and price, rateyour satisfaction with what you get from the company being worth what you pay for it.”

a p < .01b p < .05


Finally, why is the [psychosocial consequence] youmentioned important to you? (the answer will gener-ally be a personal value).

Final Importance Judgment. Once subjects hadcompleted the laddering phase of their survey response,they were given an opportunity to provide a formal ratingto indicate the importance of the clarity dimension theywere evaluating: “After thinking through the importance

and the laddering exercise, how important do you believe(the clarity type LABEL . . . e.g., “a pre-arrival phonecall”) is to you in your evaluation of these two firms?Subjects responded on a scale of one (very unimportant)to 10 (very important).

The objective of the qualitative study is to make theclarity issue salient and to determine whether or not clarityevokes particular meaning. The laddering technique en-

TABLE 3

Qualitative Study Scenarios

CONTEXT SCENARIO N

1. Provide You have recently moved to a new town and need to purchase a lawnmower. Based 20competitive upon Consumer Reports, you decide upon a particular lawnmower brand and model,prices (retailer) and then identify two stores that carry this brand. The stores are both lawn and garden

stores and are equally distant from your house. You first go to store J, where youSEARCH find the lawnmower you want available and the service seems acceptable.

You tell the salesperson that you’ve got one more store to shop before deciding. Heasks you “which store? Our policy is to provide customers with the current prices ofall stores in the area carrying the same brands.” He pulls out a notebook and allowsyou to look at the prices of the lawnmower at other stores in the area.

2. Explanation of You have recently moved to a new town and are choosing between two insurance 29higher pricing companies for automobile insurance. The two companies – Firm A and Firm B – are(insurance) very similar on rates, coverage features, location, and rated service responsiveness.

For your car, the rates from each firm seem higher than for other similar makes andEVALUATION/ higher than you expected.CHOICE

The two firms differed very little except that Firm A gave you a brochure thatcontained a detailed and very clear explanation of why its rates were higher forcertain kinds of cars than others. The explanation seemed satisfactory to you.

3. Clear billing You have recently moved to a new town and need to set up new mobile phone 26(mobile phone) service. In choosing between two mobile telephone services, you get brochures from

each that indicate very little difference in service between the two. Rates are fairlyINTEGRATION/ similar, each has a web site and customer service operations available 24 hours aCONSUMPTION day.

Within the materials you receive are sample bills. Firm X, has an exceptionally clearand even interesting billing statement, giving you clear access to usage and costinformation, as well as tracking of monthly billing amounts over time. Firm Y’sbilling statement was less clear, even confusing at points.

4. Service: call- You have recently moved to a new town. In your new house, you discover a pipe 24ahead (plumber) leaking. Using just the yellow pages, you call Adams’ Plumbing, explain the emergency,

and await the plumber’s arrival. About 90 minutes later, you get a call from theplumber Bob Adams plumber, saying that he will be there within 15 minutes. He

INTEGRATION/ arrives about 10 minutes later. Mr. Adams fixes the leaky pipe. The next day, heCONSUMPTION calls back to see if the problem has been fixed and to ask if you are satisfied.


courages respondents to identify consequences and val-ues at succeedingly higher levels of abstraction. At thesame time, the process does not impose values where theydo not exist. If there are no plausible deeper values,laddering methods will reveal this as subjects will strugglewith answering questions.

The goal of the qualitative study is to obtain a generalsense of whether consumers believe these dimensions areimportant in their assessments elicited by the scenariosand potential explanations as to why different dimensionsof clarity might be important to consumers. As notedearlier, our potential explanations include both standardeconomic explanations (e.g., more certain financial re-turns, time savings), as well as deeper values that have lessan economic interpretation, such as empathy, understand-ing, self-esteem/respect, control, and empathy. In addi-tion, potential inferences about firm “intelligence” andquality due to clarity might fit in between economic andnon-economic categorizations.

RESULTS

In presenting the pilot study results, we initiallyconsider the first level of the ladders, i.e., subjects’ firstresponse regarding why the clarity dimension was impor-tant. Once these functional consequences of each claritytype are identified, the higher order values laddering upfrom those functional consequences with a high fre-quency of mention can be focused upon. The results firstindicate that subjects were able to enumerate a variety offunctional consequences regarding transparency and thatthere were logical patterns in laddering up to higher ordervalues. Scenario 1 produced three predominant functionalconsequences, mixing economic (e.g., saves me time),with more inferential (e.g., seller is honest, is aware of themarket) responses. In the other pricing-related Scenarios2 and 3, a sense of certainty about value and knowing whatone is paying for predominated. Interestingly, the insur-ance pricing explanation Scenario 2 also prompted somerespondents to discuss inferences about the sellers’ hon-esty and empathy for the customer. Scenario 4 (callingahead on a service call) also produced a mixture offunctional consequences, with the most frequent focusingon the relational issue of seller care and concern. Lookingacross the rows, the functional consequences mentionedmost frequently were “knowing what I’m paying for”(i.e., economic), and “seller is truthful/honest” (i.e., infer-ential, relational).

Figure A presents the key features of clarity repre-sented in each scenario as attributes at the lowest level.The nine functional consequences that emerged from thesurveys (i.e., the pentagons at the second level) contain amixture of types of consequences. The higher-order fac-tors – represented in the hexagons and circles in Figure A –reflect consequences and values that emerged from asking

subjects to elaborate on why the lower-order conse-quences they mentioned were important to them.

The most frequent functional consequences for eachscenario are indicated in Figure A by the percentages nearthe bottom of the map. The figure illustrates that the twoscenarios involving price explanations (#2 insurance) andclear billing (#3 mobile phone) produced clear economicinterpretations, leading to judgments of financial value. Incontrast, the phenomenon of a service person callingahead (scenario #4), rather than evoking a strictly timemanagement value, instead produced beliefs that ulti-mately reflected respect and appreciation. Similarly, thenotion that a retailer would provide competitors’ pricesevoked foremost a sense of time savings, and valuecreated via time left over for other valuable sources,including family.

Rated Importance of Clarity. Following each sce-nario, subjects were asked to rate the importance of clarityin future choice on a 10-point scale. We do find relativelyhigh mean importance ratings of 8.54, 8.04, and 8.04 forthe insurance, mobile phone, and plumber scenarios,respectively. Yet, there was some discrimination, as re-spondents found the retailer providing competitive pricesto be less compelling and important (mean = 6.38). Onemight speculate that the lower importance of Scenario 1,providing competitive prices, is due to the unusual natureof this competitive behavior, although again this is pre-cisely the practice that Progressive Insurance has imple-mented in the automobile insurance market.

CONCLUSIONS

This paper offers the observation that clarity – thefirm’s efforts to reduce consumer uncertainty in variousstages of the consumption experience – may contributesubstantially to customer value. In today’s competitiveenvironment, firms that build deep understanding of andrelationships with customers may obtain competitive ad-vantage because the resulting assets are “relatively rareand difficult for rivals to replicate” (Srivastava et al. 2001,p. 779). Firms appear to be increasingly discovering theimpact of clarity on customer satisfaction and loyalty (seeTable 1).

The area of clarity provides a rich ground for study.Deep exploration of the incremental impact of clarity onconsumer decision-making, satisfaction, and loyalty overtime is needed. Although our evidence provides initialinsight, there is work to be done in exploring the extent towhich clarity rises to the surface as a driver of customersatisfaction and loyalty across industries. There are avariety of interesting and important issues in consumerpsychology that may emerge as well. For example, clarityimplies that the firm “opens up” about issues which mayhave both positive and negative impact on consumer


FIGURE A

Primary Paths for Scenario Ladders

NOTE: On Interpretation of percentages – 46% of the subjects who saw Scenario 2 (the insurance companyproviding an explanation of its higher prices) mentioned “knowing the value they’re getting” as the key reasonwhy clarity was important to them.

Attributes

Values

In Control Peaceof Mind

Respect, Trust Honesty Family

(4) Service: Calling ahead

(3) Billing Clarity

(1) Provide competitiveprices

(2) Explainhigher pricing

Financialvalue

Security

Confident of resolution

Managing my finances

I feel valuedRespects my time

Signals Quality

( reliable)

I can planGetting value/no bad deal

Signals quality

Seller is aware of the

market

Helps memanage

time

Sense of certainty

Seller is considerate/

cares

Seller understands

needs

Seller is honest

Saves metime

Know the value I’m getting

Time is valuable

Most frequently occuring ladders

46%

43%

21%

69%

54%

19%

35%

26%

13%

35%

30%

25%

assessments. Does the positive assessment of being openand enhancing understanding ever outweigh the negativeimpact of sharing bad news? If yes, under what condi-tions? More generally, what are the conditions underwhich greater clarity will be ineffective or even backfire?In addition, the study of clarity provides a fertile groundfor exploring issues in message framing, the integration ofhuman values in communications, deeper understandingof providing insight and reasoning into service interac-tions, and the nature and effects of simplicity in productdesign.

Clarity in the relationship between firm and con-sumer may provide a rare, difficult-to-copy source ofcompetitive advantage in today’s marketplace ofcommoditized products and services. Although firms mayperceive this to be a costly area to explore for its potentialin differentiation, it in fact may create opportunities forgains in distinctive competitive position for firms willingto invest in building assets and processes that deliverunique value to consumers. There are significant contri-butions that researchers in marketing can make to under-standing these issues.

REFERENCES

Adams, Marjorie E., Georg S. Day, and Deborah Dougherty(1998), “Enhancing New Product Development Per-formance: An Organizational Learning Perspective,”

Journal of Product Innovation Management, 15,403–22.

Brucks, Merrie (1985), “The Effects of Product ClassKnowledge on Information Search Behavior,” Jour-nal of Consumer Research, 12 (June), (1), 1–16.


Collins, Nancy L. and Lynn Carol Miller (1994), “Self-Disclosure and Liking: A Meta-Analytic Review,”Psychological Bulletin, 116 (3), 457–75.

Culnan, M.J. (1983), “Environmental Scanning: The Ef-fects of Task Complexity and Source Accessibilityon Information Gathering Behavior,” Decision Sci-ences, 14, 194–206.

Cyert, Richard M. and James G. March (1992), A Behav-ioral Theory of the Firm, 2nd ed. Oxford: BlackwellPublishers.

Day, George S. and Robin Wensley (1988), “AssessingAdvantage: A Framework for Diagnosing Competi-tive Superiority,” Journal of Marketing, 52 (April),1–20.

Einhorn, Hillel J. and Robin M. Hogarth (1985), “Ambi-guity and Uncertainty in Probabilistic Inference,”Psychological Review, 92 (4), 433–61.

Eggert, Andreas and Sabrina Helmb (2002), “Exploringthe Impact of Relationship Transparency on BusinessRelationships: A Cross-Sectional Study among Pur-chasing Managers in Germany,” Industrial Market-ing Management, 32, 101–8.

Fischer, Gregory W., Mary Frances Luce, and Jianmin Jia(2000), “Attribute Conflict and Preference Uncer-tainty: Effects on Judgment Time and Error,” Man-agement Science, 46 (Jaunary), 88–103.

Hammer, David C. (2006), “Adapting Customer Serviceto Consumer-Directed Health Care,” Healthcare Fi-nancial Management, 60 (9), 118–22.

Häubl, Gerald, Benedict G.C. Dellaert, and Murat Usta(2007), “Ironic Effects of Personalized Product Rec-ommendations on Subjective Consumer DecisionOutcomes,” working paper, School of Business,University of Alberta.

Hoch, Stephen J. and Young-Won Ha (1986), “ConsumerLearning: Advertising and the Ambiguity of ProductExperience,” Journal of Consumer Research, 13(September), 221–32.

Hui, MIchael K. and Lianxi Zhou (1996), “How DoesWaiting Duration Information Influence Customers’Reactions to Waiting for Services?” Journal of Ap-plied Social Psychology, 26 (19), 1702–17.

Ippolito, Pauline M. (1990), “Bonding and NonbondingSignals of Product Quality,” Journal of Business, 63(1), 41–60.

Izquierdo, Segismundo S. and Luis R. Izquierdo (2007),Journal of Business Research, 60 (8), 858–67

Kim, Kijoo Eyong B. Kim (2006), “Suggestions to En-hance the Cyber Store Customer’s Satisfaction,” Jour-nal of American Academy of Business, 9 (March),233–40.

Kirmani, Amna and Peter Wright (1989), “Money Talks:Perceived Advertising Expense and Expected Prod-uct Quality,” Journal of Consumer Research, 16(December), 344–53.

____________ (1990), “The Effect of Perceived Adver-tising Costs on Brand Perceptions,” Journal of Con-

sumer Research, 17 (September), 160–71.____________ and Akshay R. Rao (2000), “No Pain, No

Gain: A Critical Review of the Literature on Signal-ing Unobservable Product Quality,” Journal of Mar-keting, 64 (April), 66–79.

Kohn, Meir G. and Steven Shavell (1974), “The Theory ofSearch,” Journal of Economic Theory, 9 (April), 92–123.

Kwong, Jessica Y.Y. and Dilip Soman (2006), “The Roleof Fluency in the Decision to Spend Loyalty ProgramPoints,” Working Paper, The Chinese University ofHong Kong.

Kaiser Network Daily Reports (2005), “Los AngelesTimes Examines Consumer-Driven Plans’ Lack ofInformation on Prices,” [Dec 13, 2005] ed. Vol.2007: Henry J. Kaiser Family Foundation.

Langer, E.J. (1975), “The Illusion of Control,” Journal ofPersonality and Social Psychology, 32, 311–28.

Lanzetta, John T. (1963), “Information Acquistion inDecision-Making,” in Motivation and Social Inter-action-Cognitive Determinants, O.J. Harvey, ed. NewYork: Ronald Press, 239–65.

LaSalle, Diana and Terry A. Britton (2003), Priceless:Turning Ordinary Products into Extraordinary Ex-periences. Boston, MA: Harvard Business SchoolPress.

Markard, Jochen and Edward Holt (2003), “Disclosure ofElectricity Products – Lessons from Consumer Re-search as Guidance for Energy Policy,” Energy Policy,31, 1459–74.

Meyers, Susan (2005), “The Road to Transparency,”Trustee, 58 (5), 18–22.

Montgomery, David B., Marian C. Moore, and Joel E.Urbany (2005), “Reasoning about Competitive Re-actions: Evidence from Executives,” Marketing Sci-ence, 24 (Winter), (1), 138–49.

O’Reilly, C. (1982), “Variations in Use of Decision Mak-ers’ Use of Information Sources: The Impact ofQuality versus Accessibility of Information,” Acad-emy of Management Journal, 25, 756–71.

Parker, Louise (1993), “When to Fix It and When toLeave: Relationships Among Perceived Control, Self-Efficacy, Dissent, and Exit,” Journal of AppliedPsychology, 78 (December), 949–59.

Rankin, David (2004), “Simplicity and Transparency:Returning from the Land of the Bland,” Journal ofFinancial Services Marketing, 9 (2), 172–78.

Ratchford, Brian T. (1982), “Cost-Benefit Models forExplaining Consumer Choice and Information-Seek-ing Behavior,” Management Science, 28 (February),197–212.

Reynolds, Thomas J. and Jonathan Gutman (1988),“Laddering Theory, Method, Analysis, and Interpre-tation,” Journal of Advertising Research, 28 (1), 11–34.

Ross, Jr., William T. and Elizabeth H. Creyer (1992),“Making Inferences about Missing Information: The


Effects of Existing Information,” Journal of Con-sumer Research, 19 (June), 14–25.

Scott, Jeremy (2004), “Ethics, Governance, Trust, Trans-parency, and Customer Relations,” The Geneva Pa-pers on Risk and Insurance, 29 (January), 45–51.

Song, Hongjun (2003), “E-Services at FedEx,” Commu-nications of the ACM, 46 (6), 45–46.

Srivistava, Rajendra K., Liam Fahey, and H. KurtChristensen (2001), “The Resource-Based View andMarketing: The Role of Market-Based Assets inGaining Competitive Advantage,” Journal of Man-agement, 27, 777–802.

Stauss, Bernd, Maxie Schmidt, and Andreas Schoeler(2005), “Customer Frustration in Loyalty Programs,”International Journal of Service Industry Manage-ment, 16 (3/4), 229–62.

Taylor, Valerie A. and Amanda Bower (2004), “Improv-ing Product Instruction Compliance: ‘If you Tell MeWhy, I Might Comply,’” Psychology and Marketing,21 (March), 229–45.

Urban, Glen L, Fareena Sultan, and William J. Qualls(2000), “Placing Trust at the Center of Your InternetStrategy,” Sloan Management Review, (Fall), 39–48.

____________ (2006), “Customer Advocacy: A NewParadigm for Marketing?” in Does Marketing NeedReform? Jagdish N. Sheth and Rajendra S. Sisodia,eds. Armonk, NY: M.E. Sharpe, Inc.

Urbany, Joel E. (1986), “An Experimental Examinationof the Economics of Information,” Journal of Con-sumer Research, 13 (September), (2), 257–63.

____________, Peter R. Dickson, and Rosemary Key(1990), “Actual and Perceived Consumer Vigilancein the Retail Grocery Industry,” Marketing Letters, 2(1), 15–25.

____________, ____________, and Alan G. Sawyer(2000), “Insights Into Cross- and Within-Store PriceSearch: Retailer Estimates vs. Consumer Self-Re-ports,” Journal of Retailing, 76 (Summer), (2), 243–56.

____________, ____________, and William L.Wilkie(1989), “Buyer Uncertainty and Information Search,”Journal of Consumer Research, 16 (September), (2),208–15.

Wang, Tuo, R. Venkatesh, and Rabikar Chatterjee (2007),“Reservation Price as a Range: An Incentive-Com-patible Measurement Approach,” Journal of Market-ing Research, 44 (May), 200–13.

Wansik, Brian (2003), “Using Laddering to Understandand Leverage a Brand’s Equity,” Qualitative MarketResearch, 6 (2), 111–18.

Weitzman, Martin L. (1979), “Optimal Search for theBest Alternative,” Econometrica, 47 (May), 641–54.

Woehr, Maria (2006), “Give ‘em What they Want,”Insurance and Technology, 31 (June), 45–47.

For further information contact:Joe Urbany

Department of Marketing372 Mendoza College of Business

University of Notre DameNotre Dame, IN 46556–5646



THE INFLUENCE OF A TENTATIVE PREFERENCE ON

INFORMATION SELECTION

Kurt Carlson, Duke University, DurhamAbhijit Guha, Duke University, Durham

SUMMARY

During pre-decisional choice processes, consumersare either engaged in the evaluation of information, theseeking of new information, or a combination of the two.Upon evaluating early information, most consumers spon-taneously form a tentative preference for one of theoptions (this preferred option is labeled as the leader).Though research has revealed a tendency for consumersto bias their evaluation of new information to favor aleading brand (Russo, Meloy, and Medvec 1998), we areaware of no systematic exploration of how the existenceof a leader influences consumers’ search for new informa-tion.

Given that consumers bias evaluations to supporttheir leading brand, a logical prediction is that consumerswill search for information that they expect confirms theirleader as the best option overall, i.e., the mere tentativepreference for a leader during a pre-decisional choiceprocess will compel consumers to seek information theyexpect will support this preference (i.e., preference-con-firming search). In contrast, it is possible that consumersmay opt to focus their attention on a focal option (Buehlerand McFarland 2001). Presumably, in pre-decisionalchoice contexts, the leading option would be focal. If so,then consumers would seek information about their leader(i.e., leader-focused search).

In Studies 1–3, participants examine a sequence ofattributes, each of which presents information for a pair ofproducts on that attribute. After participants have viewedall the attributes, and indicated a leader choice afterviewing the final attribute, participants are given a choiceto read one of two blogs, each of which provides informa-tion about one of the two products. When informationavailable is positive about both products, both tendencies(i.e., tendency for preference-confirming search and ten-dency for leader-focused search) will result in the selec-tion of positive information about the leader. However,when information about the products is negative, prefer-ence-confirming search predicts that consumers will seekto examine negative information about the trailing option(to confirm their hypothesis that the leader is better),whereas leader-focused search predicts that consumerswill seek to examine new information about the leader,even if such information is negative.

Study 1 (Study 2) relates to a choice between twohotels (backpacks). In each study, participants were as-signed to either a positive information condition or anegative information condition. In the positive (negative)information condition, each blog provided positive (nega-tive) news about one of the products. Blog choice datarevealed that participants chose to read the blog abouttheir preferred product more often than the blog contain-ing information about the trailing product, and the ten-dency to read the blog about the leader was unimpacted bywhether the information in the blog was expected to bepositive or negative. Thus, it seems that consumers exhib-ited leader-focused search, and not preference-confirm-ing search, when seeking new information in the presenceof a leading option.

In Study 2, we also examined blog choice in anadditional condition, wherein the two blogs were ex-pected to provide comparative statements indicating thatone, or the other, product was better. Since both blogsprovided information about both products, consumerscould not exhibit leader-focused search in this condition.We found that participants were not more likely to selectthe blog that was expected to cast their leader in a positivelight over the blog that was expected to cast their leader ina negative light, i.e., participants did not exhibit prefer-ence-confirming search.

In Study 3, we again gave participants the option ofreading two blogs, each of which provided negativeinformation about either the leading product or the trailingproduct. In this case however, a pre-test rated one of theblogs as providing less diagnostic information. We stillfound that participants preferred to read the blog that gaveinformation about their leader, even if this meant readinga blog that was less diagnostic. Thus the tendency toexhibit leader-focused search can have material conse-quences for the quality of the information search.

In Study 4, we sought to determine whether prefer-ence-confirming search (in any fashion) would emergeusing the approach employed in the studies above. Wereasoned that preference-confirming search would bemost likely to exist when individuals had strong prefer-ences, and found that this was indeed the case. Universitystudents in the positive (negative) information choiceselected between two online columns that presented posi-


tive (negative) news about either their own basketballteam or a rival school’s team. Most students in the positiveinformation condition opted to read about the leader, i.e.,their own team, whereas significantly fewer students inthe negative information condition opted to read about theleader. This behavior suggests that strong preferencesresult in preference-confirming information search.

Overall, what we found is that leader-focused search,and not preference-confirming search, drives consumers’information search in pre-decisional contexts. This find-ing has important implications for marketers seeking toinfluence the consumer’s choice process. References areavailable upon request.

For further information contact:Abhijit Guha

Duke University1 Towerview Drive

Durham, NC 27708–0120Phone: 919.218.6652



TWO-STAGE LOTTERIES AND THE VALUE OF UNRESOLVED

UNCERTAINTY IN MULTIPLE CONTACTS

Ido Erev, Technion, Haifa, IsraelErnan Haruvy, University of Texas at Dallas, Richardson

SUMMARY

Many repeated contacts between firms and consum-ers involve delayed realization of uncertainty. As a result,new decisions are often made before complete resolutionof the uncertainty associated with past decisions. Forexample, a consumer may have to decide about the pur-chase of a product before the resolution of all the uncer-tainty associated with similar products bought in the past.The current paper examines the effect of the timing ofuncertainty resolution. It focuses on simple examples inwhich the uncertainty can be resolved immediately and ata delayed stage. These examples can be abstracted as two-stage lotteries.

A particularly clear example of the situations studiedhere is provided by prize promotion campaigns. In thesecases certain actions (buying a particular product) arerewarded with a lottery ticket, and the firm determines thetiming of the resolution of this lottery. A second examplewould be frequency programs, where loyalty points areawarded at purchase time but rewards are realized later on.Two-stage prize schemes could also be applied to Internetpromotions. One could view visits to a website as a firststage in a two-stage promotion scheme. In the first stage,the consumer chooses whether or not to visit the store orwebsite. If he visits, he may find new promotions. Entryin each of these promotions gives a very low probabilityof winning a large prize.

The extant literature can be used to support threeseemingly contradictory predictions concerning the be-havioral effect of delayed uncertainty resolution. Underthe “show-me-the-money” hypothesis, immediate resolu-tion of uncertainty is preferred to delayed resolution.Under the “frequency and anticipation” hypothesis, de-layed realization is most effective. Finally, the “contin-gent weighting” hypothesis predicts that a two-stagerealization of uncertainty is likely to be most effective.

We consider situations in which part of the outcomesassociated with a particular action can be described by thetwo-stage lottery. The lottery yields a positive prize if twoindependent events occur, and it pays nothing otherwise.

The first event (E1) is realized immediately, and thesecond event (E2) occurs at a later stage. The probabilitiesof the two events are p

1 and p

2; thus, the probability of

winning the prize is pw = (p

1) (p

2). The current analysis

focuses on lotteries in repeated contacts with differentvalues of p

1 and p

2and a fixed p

w value.

The current work presents three Internet studies thatevaluate the effect of the timing of uncertainty resolution.The results demonstrate that when the probability of apositive outcome is low, delayed resolution may be moreeffective than immediate resolution of uncertainty.

Study 1 gave participants a choice between twoInternet stores selling CDs. One of the stores used a prizepromotion with attractive but infrequent prizes. The re-sults suggest that the effectiveness of the prize promotionis maximized by a two-stage resolution of the uncertainty.

Study 2, also in an Internet store choice environment,seeks to establish boundary conditions when prizes arefrequent. It suggests that the effectiveness of the two-stage procedure decreases when prizes are small andfrequent.

In Study 3, participants were asked to visit a websiteas often as possible and were given monetary prizes asincentive to do so. Unlike Studies 1 and 2, where the focuswas on store choice, Study 3 focused on frequency ofvisits to the web site. The purpose of Study 3 was toexamine whether the advantage of the two-stage promo-tion could be replicated in this different task. It showedthat two-stage promotions could significantly increase thefrequency of visits to a promoted web site.

The results of all three studies suggest that two factorscontribute to the positive effect of delayed resolution ofuncertainty: frequency and contingent weighting. Underthis interpretation, the effectiveness of the two-stageprocedure is maximized when it provides frequent butsurprising (not too frequent) rewards. This interpretationexplains the observation that in the context of lotteriesconsumers behave “as if” they prefer uncertainty anddelayed realization of uncertainty.


For further information contact:Ernan Haruvy

School of ManagementUniversity of Texas at Dallas

800 West Campbell RoadRichardson, TX 75080–3021

Phone: 972.883.4865Fax: 972.883.6727



COMPANY DESIGNERS VERSUS USERS THROUGH THE CUSTOMER

LENS: WHO HOLDS MORE CREATIVE POTENTIAL?

Martin Schreier, Vienna University of Economics and Business Administration, AustriaChristoph Fuchs, University of Vienna, Austria

SUMMARY

Companies have begun to harness the creative poten-tial among user communities by allowing individual usersto submit ideas for new product designs (e.g., Adidas,BBC, BMW, Boeing, Ducati). The most promising amongthese user designs are then taken up by the firm, reworkedwhere necessary, and finally marketed to the masses. Thuscustomers (i.e., users) switch from the role of a passiveconsumer to that of an active collaboration partner whoco-creates value with the firm (Lusch, Vargo, and O’Brien2007; Vargo and Lusch 2004). Co-creation refers to a“collaborative new product development (NPD) activitythat actively engages customers in the design and devel-opment of a new product offering” (O’Hern and Rindfleisch2007, p. 3). In its most extreme form, as for exampleapplied by the US fashion start-up “Threadless” (Ogawaand Piller 2006), co-creation constitutes a radical depar-ture from conventional wisdom and common practice inNPD, as companies begin to ask their users – instead oftheir company designers – to come up with (ideas for) newproduct designs.

Whereas scholars have long explored the “direct”effects of co-creation (the creativity or commercial attrac-tiveness of user-generated products; von Hippel 2005),the potential “indirect” effects (how consumers perceivecompanies which foster co-creation) have received hardlyany attention thus far. The few exceptions focus onconceptual and/or empirical analyses of those consumerswho have actively participated in co-creation (Berthonet al. 2007; Nambisan and Baron 2007; O’Hern andRindfleisch 2007; Pitt et al. 2006; Prahalad andRamaswamy 2004). As Pitt et al. (2006) argue, however,this is only one important stakeholder group. Those cus-tomers who merely observe, but are not directly involvedin co-creation – the “observers” – constitute a secondimportant segment worth investigating because they ac-count for the bulk of the market (whereas those whoactively participate are a minority).

In this context, it appears plausible that companieswhich focus on co-creation are generally perceived to be“closer” to their customers (i.e., more customer-oriented)than conventional companies. To NPD managers, this“closeness” might be a welcome – albeit insufficient –means of positioning the company in the market. What if,

on the other hand, consumers perceive this co-creationfocused company as poorly positioned to come up withtruly novel and useful products? After all, research hasshown that being to close to one’s customers mightsometimes hinder innovation (Christensen 1997; forcounterarguments, see von Hippel 2005). From a market-ing perspective, it is a threat to the company’s long-termsuccess if consumers attribute inferior creative abilities toits NPD processes. In this paper, we therefore aim toexplore how the “observer” consumers perceive co-cre-ation in its most extreme form (users generating ideas fornew products) versus company-creation (company de-signers generating ideas for new products) in relation to afirm’s ability to derive creative new products. It is impor-tant to note that we focus our analysis on consumers’perceptions of a firm’s creative ability – not on the“objective” innovativeness of a firm’s new products (asthis has been shown elsewhere; cf., von Hippel 2005).

Most importantly, we propose that consumers willperceive firms which focus on co-creation as superior totraditional creation-focused firms in terms of creativeabilities. We further hypothesize that this specific per-ceived corporate ability will be related to the consumers’evaluations of the firm as well as its product offerings. Inparticular, we conjecture that a firm’s ability to generatecreative new products shows a positive link to consumers’brand attachment (i.e., consumers will be more attached tocreative companies) and to perceived product creativity(i.e., the company’s products will be perceived as morecreative). Finally, we hypothesize that brand attachmentas well as perceived product creativity are positivelyrelated to behavioral intentions.

In the course of an experiment in the context of t-shirts (following the Threadless-example), we find thatconsumers perceive co-creation as superior to a company’sability to develop creative new products. This is an impor-tant insight, as we find a positive connection between thiscorporate ability association and the consumer’s brandattachment and product attitudes, both of which have animpact on favorable behavioral intentions. We thus pro-vide initial evidence that – through the lens of customerperception – companies which foster co-creation mightgain a competitive advantage over traditional firms due tothe significantly higher creative potential attributed to theformer.


We believe that these findings will be very useful toresearchers and managers interested in understanding theenduring consequences of co-creation on a firm’s strate-gic positioning in the market. At the same time, we note

that our study is based on only one product category (t-shirts) and we encourage future research to address thegeneralizability of the findings reported in this study.References are available upon request.

For further information contact:Christoph Fuchs

International MarketingInstitute of Business Administration

University of ViennaBruenner Strasse 72

A–1210 ViennaAustria

Phone: +43(1)4277.38035Fax: +43(1)4277.38034



EXAMINING THE INFLUENCE OF SOCIAL CAPITAL ON

CORPORATE REPUTATION


Barbara Gassmann, Leibniz University of Hannover, Germany

SUMMARY

Embedded in an increasing expansion and density ofeconomic and social interactions, recognizing the exist-ence and improving the quality of a company’s network ofrelations has become a critical success factor. Stronglyassociated with the strategic relevance of corporate repu-tation as critical resource in gaining competitive advan-tage and maximizing a company’s economic and non-economic (social) status, a better understanding and man-agement of a company’s social capital and the identifica-tion of those internal and external actors and relationshipswith the highest potential for network-related value ef-fects is vitally important. This paper aims at developing aconceptual framework which gives reason for adding theconstruct of social capital to the field of reputation man-agement.


Defined as a characteristic or attribute ascribed to acompany by its stakeholders, corporate reputation is atleast in part the product of the company’s network ofrelations. Closely related to the network construct is theconcept of social capital referring to the value that lieswithin and may derive from a network of relations. Under-stood as the awareness or perception about corporatesocial behavior, corporate social capital will contribute tothe development of corporate image, reputation and “so-cial legitimization” of the company in the network mem-bers’ view. Figure 1 shows our proposed conceptualframework to investigate whether and to what extentcorporate social capital with its key driver’s relational andreputational capital affects corporate reputation.

Relational capital addresses the network of relationsthat go further the company’s organizational frontierswith the agents that are part of its closer environment orindustry. Referring to the shape and extent of networklinks, corporate relational capital can be defined as acomponent of corporate social capital with special focuson the value of the network of relations that a companymaintains with its stakeholders, the different environmen-tal agents. Apart from the relational dimension in terms ofthe presence or absence of network relations to a company’sstakeholders, the quality and content of the relationships

are important to identify, manage and control. As prereq-uisite for the formation of effective and stable relationships,reputational capital refers to the emotional aspects ofrelationships (i.e., shared norms and values, interpersonalobligations and expectations, reciprocal obligation, mu-tual identification, commitment, understanding, honestyand trust) which organizations or groups of individualshave developed with each other through a history of socialinteractions. Deriving from membership in specific net-works of mutual acquaintance and recognition, socialcapital in the form of reputation – reputational capital –can be described as generating trust between organiza-tions and between organizations and individuals.

An important aspect of our model is a differentiatedmeasure of corporate reputation referring to seven inter-related characteristics – Credibility, Reliability, Respon-sibility, Trustworthiness, Loyalty, Satisfaction, and Ac-countability – which can be regarded both, goal constructsof corporate social capital and fundamental componentsof corporate reputation.

First Empirical Hints

Focusing on the link between corporate relational,reputational, and social capital, this paper has examinedthe role of social context in determining a company’sreputation. We suggest that corporate social capital isrooted in the presence and quality of a certain set ofnetwork relations in which a company is embedded. Theresults of our empirical pre-study in the financial servicessector indicated that corporate social capital and corporatereputation are perceived at and have to be analyzed in anetwork-level of analysis. We found empirical evidenceto believe that association with trustworthy and presti-gious “others” might be statusenhancing as well as repu-tation-enhancing, thus, attributing reputational capitaleffects to a single company or type of company embeddedwithin a certain network of relationships might be prob-lematic. With regard to the interplay between the rela-tional and reputational capital dimension, a lot of positiveand negative correlations exist. Following a social net-work perspective, companies of high status and reputationare attracted to other companies of high status and repu-tation. Social capital and reputation can derive from themembership in a wellknown status-enhancing network;



however, the access to such a network requires a certaindegree of social capital and reputation. In addition, com-paring different companies and different groups of stake-

holders, more or less the same social capital drivers mightlead to totally different perceptions of corporate reputa-tion. References are available upon request.


Leibniz University of HannoverInstitute of Marketing and Management


GermanyPhone: ++49.511.762.4862Fax: ++49/511/762-3142



MISSION FULFILLMENT AND THE INTERNAL AUDIENCE:

PSYCHOLOGICAL JOB OUTCOMES

Taewon Suh, Texas State University – San MarcosMark B. Houston, Texas Christian University

Steven Barney, SSM Health CareIk-Whan G.Kwon, Saint Louis University, St. Louis

SUMMARY

A stream of research in organizational psychologyfinds that congruence in core values between an employeeand an organization is a key element of “fit” that canimpact the success of an employment relationship (cf.,Kristof 1996). In addition to being an important selectioncriteria (Kristof-Brown 2000), person-organization fithas a demonstrated influence on important employeepsychological outcomes and behaviors, such as satisfac-tion (Cable and Edwards 2004), organization commit-ment (Kalliath, Bluedorn, and Strube 1999), turn overintentions (Ostroff, Shin, and Kinicki 2005), and citizen-ship behaviors (Cable and DeRue 2002). Given its criti-cality, it is important to understand how marketing actionsthat relate to a firm’s core values impact this “internalaudience” (Gilly and Wolfinbarger 1998; Walker andRuekert 1987).

In this paper, we argue that value congruence –normally seen as having primarily positive consequences –operates as a moderator that can actually heighten nega-tive employee outcomes. This proposition is examined ina key internal marketing context. We draw from esteem-based theories of identity to argue that value congruenceplays an important moderating role in the relationshipbetween internal marketing activities and employee psy-chological outcomes. Person-organization value congru-ence can intensify employee reactions (positive and nega-tive).

In short, we argue that an organization’s core valuesare communicated to employees, potential employees,and other stakeholders through a variety of mechanisms,including formal mission statements and other publicproclamations. These statements of core values serve asreferents for employees’ assessments of person-organiza-tion value congruence. The degree to which the organiza-tion proceeds to act in a manner that is consistent with thecore values articulated in these symbolic statements will,in turn, impact the desirability to an employee of support-ing and being affiliated with the organization. We labelthis value-consistent behavior mission fulfillment, de-fined as an organization consistently acting (publicly andprivately) in a manner that is congruent with, and leads tothe fulfillment of, the corporate mission statement. We

expect mission fulfillment to influence affective commit-ment (positive) and turnover intentions (negative), bothdirectly and through mediating variables (organizationalidentification and emotional exhaustion). We suggest thatvalue congruence will positively moderate the impact ofperceived mission fulfillment on psychological outcomes.If an employee’s personal values align closely with thecore organization values that are embodied in a missionstatement, failures by the organization to live up to thosevalues will have implications for employee self-definition(Bhattacharya, Rao, and Glynn1995), more strongly dam-aging the degree to which the employee identifies with thefirm and heightening emotional exhaustion as the em-ployee wrestles with reconciling reality with the ideal(Foreman and Whetten 2002). In turn, commitment isreduced and turnover intentions are amplified.

Mission statements and person-organization fit aretraditionally seen as falling within the domain of manage-ment scholars and have received little attention in market-ing. However, mission statements are similar to externaladvertising messages (Gilly and Wolfinbarger 1998) inthat they affect employees by serving both ideological(Alvesson 1998) and identity (Poole1998) functions, andthus are key tools at management’s disposal for internalmarketing (Ireland and Hitt1992).

Research Sample and Results

The research setting is a large, non-profit, health-caresystem in a major Midwest U.S. city. Surveys weredistributed at work to 12,000 service employees. Themission statement of the system has a value-related themeand was inserted in the first page of the survey package.We received 3,999 usable responses (effective responserate = 33.3%).

We utilized structural equations modeling. First, whilemission fulfillment’s direct influence on affective com-mitment is statistically significant, its impact on turnoverintentions is not. Next, mission fulfillment relates posi-tively to organizational identification and, in turn, organi-zation identification relates positively to affective com-mitment and negatively to turnover intentions. Missionfulfillment relates negatively to emotional exhaustionand, in turn, emotional exhaustion relates positively to


turnover intentions, but does not relate significantly toaffective commitment, although the sign of the path isnegative, as predicted. Emotional exhaustion also relatesnegatively to organizational identification.

Value congruence moderates (intensifies) the posi-tive impact of mission fulfillment on organizational iden-tification. In other words, mission fulfillment has a sig-nificantly stronger impact on organizational identifica-tion when value congruence is high than when it is low.

Also as hypothesized, value congruence moderates (in-tensifies) the negative effect of mission fulfillment onemotional exhaustion; that is, mission fulfillment has astronger negative impact one motional exhaustion whenvalue congruence is high than when it is low. Contrary toexpectations, value congruence does not moderate thepositive direct relationship between mission fulfillmentand affective commitment and the negative direct rela-tionship between mission fulfillment and turnover inten-tions.

For more information contact:Mark B. Houston

Eunice and James L. West Chair of American EnterpriseMJ Neeley School of Business

Texas Christian UniversityTCU Box 298530

Fort Worth, TX 76123E-Mail: [email protected]


DO B2B BRANDS MAKE A DIFFERENCE? AN EMPIRICAL

INVESTIGATION OF THE PERFORMANCE IMPLICATIONS

OF BRANDING IN B2B ENVIRONMENTS

Christian Homburg, University of Mannheim, GermanyMartin Klarmann, University of Mannheim, Germany

Jens Schmitt, University of Mannheim, Germany

SUMMARY

For most companies in B2C environments, develop-ing and maintaining strong brands is traditionally a keyelement of their marketing strategy. In comparison, com-panies targeting mainly business customers have typi-cally put much less strategic emphasis on branding. It maywell be that the lack of attention for branding in B2Benvironments is rooted in one important view of research-ers and practitioners, where organizations are seen asrational, cognitive and problem-solving decision-makers.They apply objective and economic decision criteria whenmaking purchasing decisions. From this perspective, whichwe call the “full rationality perspective on organizationalbuying,” it can well be assumed that in many cases brandsare ineffective in B2B markets.

However, the rationality of organizational buyers hasbeen increasingly questioned in recent years. More spe-cifically, there is a large stream of research emphasizingthe role of predominantly emotional phenomena such astrust and commitment in B2B buyer-seller relationships,the role of emotions in personal selling or the role ofheuristics in organizational buying processes. Althoughthese studies do not completely deny the rationality oforganizational buying processes, they show that it is muchmore limited than suggested by the full rationality view oforganizational buying outlined above. This perspectivewhich we call “the limited rationality perspective onorganizational buying” suggests that brand functions mayapply to a much larger extent to B2B buying decisionsthan suggested by the full rationality perspective.

Thus, there is no unequivocal view on whether andwhen branding will be effective in B2B environments.Interestingly, empirical knowledge on branding effec-tiveness in B2B markets is scarce. While first empiricalresults suggest that B2B brands lead to increased cus-tomer loyalty and price premiums, these studies typicallyfocus on single industries. Against this background, wedevelop a branding effectiveness framework, linkingbranding success (i.e., brand awareness and brand liking)to market-related outcomes (i.e., sales volume develop-ment and price level) and financial outcomes (i.e., acompany’s return on sales). We then draw on informationeconomics and theories from social psychology on group-

level information processing to develop hypotheses re-garding the impact of branding success on market-relatedoutcomes. According to the two distinguishable perspec-tives on organizational buying behavior, we submit twocompeting sets of hypotheses to empirical exploration: apositive impact versus no impact of branding success onmarket-related outcomes.

Furthermore, previous research has not analyzedwhich factors moderate the relationship between brand-ing success and market-related outcomes in businessmarkets. However, facing the large investments usuallyassociated with establishing a strong brand, it is importantfor marketing managers in B2B firms to know, whetherbranding will be an effective marketing instrument in theirparticular marketplace. Consequently, we hypothesizemoderating effects of product characteristics (productcomplexity, technological stability, and product impor-tance) and buyer characteristics (buying center size, buy-ing center heterogeneity, and buyer experience), againbased on information economics and theories from socialpsychology on group-level information processing.

We empirically tested our hypotheses based on asurvey among marketing and sales managers from 310B2B firms from a broad range of industries (machinebuilding, electronics, chemicals, automotive, and others)using structural equation modeling. In a first step, weestimated a model (M1) containing the main effects. Ascontrol variables only SBU size, brand coverage, andbrand share of revenues were included. In a second step,we estimated the same model (M2), but additionally thecontrol variables technical product quality and servicequality were included in the model. Results show thatbrand awareness has a positive effect on volume-relatedsuccess and – partially confirmed – on the effective pricelevel. Brand liking only has a positive effect on volume-related success and on the effective price level for M1 butnot for M2. To test the moderating hypotheses, we in-cluded latent interactions between the moderator and therespective independent variables in model M2 with allcontrol variables. Results show that product complexity,technological stability, buying center heterogeneity, andbuying center size moderate the relationship betweenbranding success and market-related outcomes.


While our findings show that there is a strong effectfrom B2B brand awareness on volume-related successand a weak link between B2B brand awareness and pricepremium, brand liking has no effect on market-relatedoutcomes. Thus, our results suggest that the effects ofB2B branding on market-related outcomes are muchweaker than implied by the limited rationality perspec-tive, mostly supporting the full rationality perspective onorganizational buying. Our results also demonstrate theimportance of including control variables in models ana-lyzing the effects of brands in business markets. It appearsthere is a danger in B2B branding research that the effectsof branding are confounded with the effect of product and

service quality. Furthermore, this study is also relevantfrom a managerial perspective. With regard to branding,increasing brand awareness is the key instrument to in-creasing sales and achieving higher prices in businessmarkets. When developing a branding strategy, firmsshould tailor their marketing activities to increase brandawareness. By contrast, a positive attitude toward thebrand does not lead to beneficial outcomes. Thus, otherthan in B2C markets, investing marketing resources in thecreation of associations with the brands that are onlyloosely connected to the actual products, are most likelynot very beneficial.

For further information contact:Christian Homburg

Marketing DepartmentUniversity of Mannheim

68131 MannheimGermany

Phone: +49.621.181.1555Fax: +49.621.181.1556



AVOIDANCE, FALSIFICATION, AND PRAGMATIC CAUTION: THE

IMPACT OF TECHNOLOGY COMFORT ON CONSUMERS’

ONLINE PRIVACY COPING STRATEGIES

Pushkala Raman, Texas Woman’s University, DentonKartik Pashupati, Southern Methodist University, Dallas

SUMMARY

Marketers have to maintain a delicate balance be-tween the need to collect data, and the need to respectconsumers’ privacy concerns. On one hand, Internet andWeb based technologies have enabled marketers to cus-tomize their product and service offerings, and build one-to-one relationships through the use of databases. On theother hand, growing consumer concerns about privacyinvasion can place severe limits on the amount of data thatmarketers can collect.

In order to formulate strategies that can overcomeconsumers’ privacy concerns, marketers must first under-stand the nature and determinants of strategies used byconsumers to protect their privacy. Some of these strate-gies have been identified by previous researchers, includ-ing falsification of information, use of multiple e-mailaddresses, avoidance of the Internet, use of a credit cardexclusively designated for online transactions, and reli-ance on privacy ratings and seals provided by third-partysources such as TRUSTe and Verisign. While the use ofdifferent coping strategies has been investigated in previ-ous research, less is known about why and when consum-ers choose a particular coping strategy. This leads to thequestion of whether consumers’ coping strategies can bepredicted by consumers’ individual characteristics.

This paper has two main goals. First, it investigatesthe coping strategies used by consumers in the face ofthreats to their privacy. Second, it tries to determine ifthere is a relationship between consumers’ individualcharacteristics (specifically, their level of comfort withtechnology) and their likelihood of adopting differenttypes of coping behaviors to protect their privacy.

Data were collected from an online survey of 2,545adult consumers who had made at least one purchase onthe Internet during the six months prior to the survey. Wepropose that consumers engage in three main privacycoping strategies, (1) internet avoidance; (2) falsificationof information online; and (3) pragmatic caution. The typeof strategy adopted by individuals is determined by indi-vidual characteristics such as perceived online transactionrisk, attitude toward technology, apprehension about tech-nology usage, and comfort with technology. The follow-

ing three hypotheses and one research question weretested:

H1: Avoidance and denial coping strategies will be:

H1a

: Used by consumers who have a higher level ofperceived online transaction risk.

H1b

: Used by consumers who have greater apprehen-sion about using technology.

H1c

: Less used by consumers with a positive attitudeto technology.

H1d

: Less used by consumers with higher levels ofSPTC.

H2: Falsification coping strategies will be:

H2a

: Used by consumers who have a higher level ofperceived online transaction risk.

H2b

: Used by consumers who have greater apprehen-sion about using technology.

H2c

: Less used by consumers with a positive attitudeto technology.

H2d

: Less used by consumers with higher levels ofSPTC.

H3: Pragmatic caution coping strategies will be:

H3a

: Used by consumers who have a lower level ofperceived online transaction risk.

H3b

: Used by consumers who have a lesser apprehen-sion about using technology.

H3c

: More used by consumers with a positive attitudeto technology.

H3d

: More used by consumers with higher levels ofSPTC.

RQ1: What is the impact of age, sex, and income on theprivacy coping strategies adopted by consumers?


Hierarchical regression analysis was used to test therelationships between individual characteristics and cop-ing strategies. The key results are summarized in Table 1.The data suggest that individuals who are more comfort-able with technology are more likely to use pragmatic

caution in coping with threats to online privacy, whereasindividuals who are less comfortable with technology aremore likely to use avoidance and falsification strategies.References are available upon request.

TABLE 1

Summary of Hypotheses and Results

Coping Strategies

Predictor Variables Avoidance Falsification Pragmatic Caution

Perceived Online H1a

supported H2a

supported H3a

not supportedTransaction Risk (OTR) (significant relationship

in the opposite direction)

Technology apprehension H1b

not significant H2b

supported H3b

supported

Attitude toward technology H1c

not supported H2c

not supported H3c

supported(significant relationship (significant relationshipin the opposite direction) in opposite direction)

Self-perceived H1d

not supported H2d

not supported H3d

supportedtechnological competence (significant relationship (significant relationship(SPTC) in the opposite direction) in the opposite direction)

Demographic Variables No effect Sex has small impact Age has small but(RQ1) (males more likely to statistically significant

falsify) impact

Age has significantimpact (youngerconsumers more likelyto falsify)

For further information contact:Pushkala Raman

School of ManagementP.O. Box 425738

Texas Woman’s UniversityDenton, TX 76204–5738



HIGHER EXPECTATIONS FOR HIGHER SATISFACTION: THE EFFECT

OF ATTRIBUTE TYPES ON POST-TRIAL CUSTOMER SATISFACTION

Camelia C. Micu, Fairfield University, FairfieldRobin A. Coulter, University of Connecticut, Storrs

ABSTRACT

In the context of a highly diagnostic product, theauthors found that creating inflated (versus realistic)expectations about search attributes result in higher cus-tomer satisfaction. Moreover, inflated expectations aboutsearch attributes result in higher satisfaction than inflatedor realistic expectations about experiential attributes.

INTRODUCTION

Does creating higher expectations about a productlead to higher customer satisfaction? One of the mostessential goals of all companies is to satisfy their custom-ers (Fornell 1992; Morgan, Anderson, and Mittal 2005;Oliver 1997). As a result, customer satisfaction has be-come one of the most widely surveyed customer issueswithin the past decade (Woodruff and Gardial 2001) andsatisfaction research is often a firm’s largest annual ex-penditure on market intelligence (Morgan et al. 2005).

Customer satisfaction can be defined as “a represen-tation of the customer’s reaction to the value received”from a product offering (Woodruff and Gardial 2001,p. 86) and arises from an appreciation of the expectancy-disconfirmation model (Olson and Dover 1979). Accord-ing to this model, satisfaction feelings are the result of theperceived disconfirmation, which is the discrepancy thatresults from comparing the product’s performance to theexpectations. Thus, high customer satisfaction or evendelight is created by exceeding customer expectations(e.g., Kotler 2000, p. 36). On the other hand, dissatisfac-tion and deflection are the result of failure to meet orexceed these expectations, which can have negative con-sequences on consumer welfare. The prediction of satis-faction using the expectancy-disconfirmation model isintuitively appealing: the higher the expectations, thebigger the disappointment. But what if the perception ofthe product performance during trial is susceptible to themagnetic force of expectations? In fact, several studiesindicate that prior expectations can shape the satisfactionwith product performance (Kopalle and Lehmann 2006;Ofir and Simonson 2007; Oliver 1997; Wong 2005).

In this paper, we explore the effects of different levelsof expectations (realistic versus inflated) created by ad-vertising claims about either search or experiential at-tributes on customers’ post-trial satisfaction. Search at-tribute information (e.g., information about price, ingre-

dients, calorie content) can be gained from secondhandsources such as advertising and word-of-mouth (versusproduct experience), without having to buy or try theproduct. Information about experiential attributes (e.g.,taste, speed, softness) is effectively provided by producttrial (Wright and Lynch 1995) and can be only verified by(limited) use of product.

We extend previous research on consumers’ expec-tations and satisfaction, as well as on advertising andproduct trial, in several ways. First, we focus on highlydiagnostic products (i.e., products for which trial offerstangible, credible evidence of the product and its at-tributes) and argue that advertising can impact consum-ers’ post-trial satisfaction with highly diagnostic prod-ucts. Whereas past research has indicated that advertisingusually has an impact on subsequent trial responses forlow diagnosticity products, for which trial is ambiguous,we conjecture that consumers’ post-trial satisfaction canbe affected by advertising for highly diagnostic productsas well. This will depend on the type of pre-trial advertis-ing claims to which consumers are exposed (see also Micuand Coulter 2007). Second, whereas past studies providedcustomers with information about search and experientialattributes combined, we disentangle the effect of thesetwo attribute types. Examining the effect of attribute typesseparately may be important, since previous researchindicated that advertising is more effective than producttrial in communicating information about search (versusexperiential attributes).

Our findings indicate that advertising claims aboutsearch attributes that generate inflated (versus realistic)expectations lead to consumers being more satisfied withthe product experience. The effect is not observed foradvertising claims about experiential attributes. Further,inflated expectations about search attributes lead to highersatisfaction than experiential attribute information. Ourfindings have implications for managers (i.e., when theycan overstate product information to make customersmore satisfied), as well as public policy makers (i.e., whento, and when not to, regulate).

BACKGROUND LITERATURE

Consumer Expectations

Expectation has been defined in many ways. Olsonand Dover (1979) defined expectations as pre-trial beliefs


about the product. Beliefs represent the subjective prob-ability of association between the product and its at-tributes; i.e., the extent to which consumers “expect” theproduct to have specific attributes. According to theexpectation-disconfirmation framework (Olson and Do-ver 1979), expectations are used by consumers to becompared with the actual product performance, whichresults in confirmation/ disconfirmation (either positiveor negative), which in turn determines satisfaction levels.

For expectations to dominate a satisfaction decision,the processing of expectations must be more salient to thecustomer than is the processing of performance or ofcomparing expectations to performance (Oliver 1977).For example, when consumers lack the ability to confirm(or disconfirm) the claims that an advertisement makesabout certain product attributes during a trial experience(like the power of a cleaning product to kill germs), theymay rely on the high expectations raised by the ad to formpost-trial evaluations of the product (see Hoch and Ha1986; Oliver 1997). This is likely to happen if productshave ambiguous performance dimensions, such as“healthy” foods (Oliver 1997) or, similarly, when theinformation conveyed through advertising cannot be physi-cally tested (e.g., search attribute information). Whenfacing an ambiguous experience, consumers may havedifficulty recognizing the discrepancy, if any, thereforethey tend to assimilate the ambiguous stimuli in thecontextual stimuli (Hoch and Ha 1986; Schwarz and Bless1992).

On the other hand, the disconfirmation (resultingfrom comparing product’s performance to the expecta-tions) dominates the expectations’ effect when perfor-mance clearly and unambiguously refutes expectations(e.g., for unambiguous product experiences, when prod-ucts are rich in experiential attributes). Individuals whorecognize discrepancies from expectations, and are will-ing to accept them regardless of the potential damage toone’s ego, should demonstrate disconfirmation influ-ences (Oliver 1997).

Advertising and Product Trial

Several studies in the advertising-trial domain havealso indicated that the assimilation bias of expectations iscontingent on the noticeability of the discrepancy (Hochand Ha 1986; Kempf and Smith 1998; Smith 1993).Specifically, the studies show that pre-trial advertisinghas little or no effect on consumer evaluations of highlydiagnostic products (i.e., products for which trial offerstangible, credible evidence of the product and its at-tributes), because trial is perceived as more useful incontrast to pre-trial advertising in evaluating the product’sattributes (Deighton and Schindler 1988; Hoch and Ha1986; Kempf and Smith 1998). However, pre-trial adver-

tising can influence the subsequent trial of low diagnosticityproducts (i.e., products for which trial is not very helpfulin judging the quality of the product because of theambiguous nature of the experiential attribute informa-tion).

For example, Hoch and Ha (1986) found that post-trial self-assessment of polo shirts’ quality (considered tobe a low diagnostic product because the assessment ofquality during trial is open to multiple interpretations)assimilated the information from the pre-trial ad. Further,Shapiro and Spence (2002) found that the market infor-mation (i.e., Consumer Reports ratings) about the soundquality of a stereo provided during trial was assimilatedinto participants’ post-trial sound quality ratings when noevaluative criteria have been given. However, when par-ticipants were provided with evaluative criteria (i.e., theywere told to pay attention to the sound quality, how “full-bodied” the sound is, and the range of instruments that canbe heard), their post-trial sound quality ratings were notinfluenced by the market information. The authors arguethat this occurred because the presence of the evaluativecriteria clarified the nature of an otherwise ambiguousattribute (i.e., sound).

As mentioned previously, to the degree to whichperformance unambiguously refutes expectations, con-sumers’ prior expectations should lose their dominanceand assimilation will not occur (Hoch and Ha 1986;Kempf and Smith 1998; Shapiro and Spence 2002; Smith1993). This is more likely to happen for highly diagnosticproducts for which trial provides useful information aboutthe product that helps consumers decide the productquality. In fact, for several highly diagnostic products,such as a grammar checker (Kempf and Smith 1998) andpaper towel (Hoch and Ha 1986), no assimilation effectwas found because consumers were able to verify the truthof the claims during trial.

HYPOTHESIS DEVELOPMENT

Experiential Attribute Information

When consumers are exposed to advertising and thento product trial, they tend to place more weight on the trialinformation than on advertising claims when evaluatingthe product (Kempf and Smith 1998; Shapiro and Spence2002; Smith and Swinyard 1988). Shapiro and Spence(2002) found that participants placed a relatively largeamount of weight on the sensory attribute (i.e., sound)(approximately 70%) versus market information (i.e.,Consumer Reports ratings) when making their stereo-preference decision. These results indicate that consum-ers weight too much the product experiential attributesduring trial, thus lowering the positive effect that the admight have on the post-trial evaluations.


For high-diagnosticity products, the focus of ourresearch, experiential attributes provide valuable infor-mation that help consumers determine the product qualityduring trial. Thus, having additional information aboutthe same attributes from a less credible source, such asadvertising, cannot significantly advance consumers’evaluations. In fact, previous studies not finding an as-similation effect of the ad-based attitudes on post-trialevaluations of highly diagnostic products (i.e., subjects inthe trial and advertising-trial condition had similar re-sponses) put perhaps too much emphasis on the experien-tial attributes in the ad, thereby lowering the effect that anad could have had on a subsequent trial experience (seeHoch and Ha 1986; Kempf and Smith 1998).

If an advertisement emphasizes information aboutexperiential attributes, according to the literature on satis-faction and disconfirmation, positive prior expectationswill lose their dominance to the degree to which perfor-mance clearly and unambiguously can refute expecta-tions. Moreover, the expectancy-disconfirmation model(Olson and Dover 1979) indicates that inflated expecta-tions lead to less satisfaction. In the context of a highlydiagnostic product, inflated expectations about experien-tial attributes are able to be easily disconfirmed duringtrial, thus leading to lower satisfaction. In conclusion, inthe context of experiential attribute information, we ex-pect consumers exposed to claims that create inflatedversus realistic expectations to report lower post-trialsatisfaction.

H1: In the context of experiential attribute information,inflated expectations lead to lower satisfaction thanrealistic expectations.

Search Attribute Information

The literature on consumer satisfaction indicates thatprior expectations can dominate satisfaction decisions(e.g., higher expectations raised by the ad can lead tohigher satisfaction, regardless of trial information) whenproducts have ambiguous performance dimensions (Oliver1997) or, similarly, when the information conveyed throughadvertising cannot be physically tested (e.g., search at-tribute information). Indeed, several studies in the adver-tising-trial literature (see Hoch and Ha 1986; Kempf andSmith 1998) indicate that for low diagnostic products,advertising has the power of influencing the subsequentproduct trial. In those studies, the advertisements empha-sized the non-experiential (credence and search) ratherthan experiential attribute information. And because con-sumers lack the ability to disconfirm the claims that anadvertisement makes about search attributes during trialexperience, the expectations raised by the pre-trial adwere incorporated in the product’s post-trial evaluations.

In conclusion, for search attribute information, we expecthigher expectations to lead to higher post-trial satisfac-tion.

H2: In the context of search attribute information, inflatedexpectations lead to higher satisfaction than realisticexpectations.

Consumers are less skeptical of search than experien-tial attribute claims made by an advertisement (Ford,Smith, and Swasy 1990). Wright and Lynch (1995) foundthat advertising is more effective than product trial incommunicating information about search attributes. Infact, Micu and Coulter (2007) show that, in the context ofa highly diagnostic product, search attribute informationgenerates more positive post-trial evaluations than expe-riential attribute information. Thus, we posit:

H3: Inflated expectations about search attributes createhigher post-trial satisfaction than:

H3a: Inflated expectations about experiential attributes

H3b: Realistic expectations about experiential attributes.

METHOD

Stimulus Pretest and Classification of Attributes

Identification of a Highly Diagnostic Product.Twenty-four undergraduates were asked to list severalproducts relevant and familiar to them. In a second pretest,fifty-four undergraduate students were asked to rate theproducts’ diagnosticity by responding to two 7-pointitems (“Overall, if you were able to try the product, howconfident would you be in your ability to judge the qualityand performance of this product?” and “Overall, if youwere able to try the product, how easy do you think wouldbe for you to judge the quality of the product?”) (Hoch andHa 1986; Kempf and Smith 1998). The two items werecombined to obtain a product diagnosticity scale. Pensand ice cream scored highest in diagnosticity and nodifferent than each other (M = 6.05 vs. M = 5.83, p = .21).Ice cream was chosen for the experiment because ofsuitability to be tested in a group, laboratory setting.

Attribute Identification and Evaluation. We exam-ined ads for ice cream, and informally solicited students’opinions about the important attributes when buying icecream. Using this information, we developed a list ofattributes that we included in the pretest. Fifty-sevenparticipants responded to “If you were going to buy icecream, how important would the following attributes be toyou?” on a 7-point scale (1 = “Not at all important,” 7 =“Very important”) (Darley and Smith 1993).


Six attributes scored above the midpoint (M = 4.00)of the importance scale: taste, how flavorful it is, creami-ness, price, calorie content, and variety of flavors. Twoindependent judges trained in the definition of search andexperiential attributes identified price, calorie content,and variety of flavors as search attributes and taste, howflavorful it is, and creaminess as experiential attributes.The agreement level between the judges was 100 percent.

Expectation Level of Print Advertisements. Fourprint advertisements, two containing search attribute in-formation (one creating a realistic level of expectationsand one an inflated level of expectations) and two contain-ing experiential attribute information (with realistic ver-sus inflated expectations), were developed. To assure thatthe expectation level manipulation is successful, expecta-tions were first measured as expectancy values, which arethe consumers’ belief strength about the six attributesmultiplied by their belief confidence and summed acrossattributes (Marks and Kamins 1988). The second measurewas an attribute-level expectation measure (Oliver 1997)and asked participants to indicate if they expect the icecream to have each of the six attributes (1 = “Stronglydisagree,” 7 = “Strongly agree”). One expectation scorewas calculated by averaging the expectations of the sixattributes.

As expected, participants in the search-inflated ex-pectations condition, compared to participants in thesearch-realistic expectations condition expressed bothhigher expectancy values (M = 161.8 vs. M = 124.5, p <.001) and higher expectation (M = 6.53 vs. 4.52, p < .05).Participants in the experiential-inflated expectations con-dition, compared to participants in the experiential-realis-tic expectations condition expressed both higher expect-ancy values (M = 122.1 vs. M = 88.7, p < .05) and higherexpectancy (M = 4.86 vs. M = 4.20, p < .001).

Development of Ad Stimuli

The four print advertisements used in the pretest wereused in the main experiment. The ads were designed toresemble magazine advertisements in both content andlayout. A picture of the ice cream in a glass was coveringmost of the print ad space. The four advertisements wereidentical, except for the different attribute informationlisted below the picture.

Experimental Procedures

Participants.One hundred twelve-undergraduate stu-dents (31 participants in search-inflated expectations con-dition, 25 participants in the search-realistic expectationscondition, 31 in the in experiential-inflated expectationscondition, and 25 in the experiential-realistic expecta-tions condition) participated in the main study for extracredit. During data collection, the maximum number of

participants in a session was seven and participants werespaced apart from one another to control for potentialreactance bias.

Procedures. Our procedures closely follow those ofSmith and Swinyard (1983). Participants in each of thefour conditions received an ad booklet upon entering theexperimental session. The booklet contained only thetarget advertisement (Deighton 1984; Kempf and Smith1998; Marks and Kamins 1988). After examining the ad,participants were instructed to complete an online ques-tionnaire. After completing the questions about the adver-tisement, they were asked to stop and wait for furtherinstructions from the researcher. Next, participants weregiven a sample of the ice cream to test and asked tocomplete the remaining of the questionnaire.

Measurement: Manipulation Checks

Product Diagnosticity. Same as in the pre-test, prod-uct diagnosticity was measured after trial using a 7-pointscale of trial usefulness in judging product quality andease-of-judging product quality (Hoch and Ha 1986). Thetwo items were combined to give a diagnosticity score (r =.55, p < .001). The diagnosticity score was significantlyabove the scale midpoint [M = 5.83 vs. M = 4.00, t(110) =20.71, p < .001].

Experiential/Search Attributes. To test theexperientiality of product attributes, participants wereasked to indicate whether each attribute they saw in theadvertisement could be judged directly by product trial (7-point scale; 1 = “Trial did not enable me to judge thisattribute,” 7 = “Trial fully enabled me to judge thisattribute”). An ANOVA with attribute type (search versusexperiential) as the independent variable and experientialityof attributes as the dependent variable shows an attributetype effect [F(1, 101) = 494.98, p < .001], which indicatesthat participants mentioned that trial enabled them tojudge the experiential attributes (M = 6.49) more than thesearch attributes (M = 1.90).

Expectation Level. To measure whether the claimsused in the advertisements are creating the realistic versusinflated expectations, two measures were used, same as inthe pretest. For search attribute claims, the inflated-expec-tation condition, compared to the realistic-expectationcondition, generated more positive expectancy values(M = 138.5 vs. M = 112.8, p < .05) and higher expectancy(M = 5.74 vs. 5.03, p < .05). Participants in the experien-tial-inflated expectations condition, compared to partici-pants in the experiential-realistic expectations condition,expressed both higher expectancy values (M = 117.7 vs.M = 89.6, p < .01) and higher expectancy (M = 6.08 vs.M = 5.12, p < .001). Further, there was no differencebetween the search attribute-inflated expectations condi-tion (M = 6.08) and the experiential attribute-inflated


expectations condition (M = 5.74, t(60) = 1.19, p < .001).Thus, the expectation level manipulation was successful.

Measurement: Confound Checks

Ad Likeability. Ad likeability was measured with two7-point semantic-differential scales: “Unappealing/Ap-pealing” and “Ineffective/Effective” (Kempf and Smith1998). Advertisements communicating different claimsshould not be different from each other. An ad likeabilityscore was obtained by averaging of the two items (r = .80,p < .001). An 2 (attribute type: search vs. experiential) X2 (expectation level: realistic vs. inflated) ANOVA indi-cates no significant effect of attribute type [F(1, 99) =1.12, p = .29] or expectation level [F(1, 99) = 1.67, p = .20]on ad likeability.

Attribute Importance. If search and experiential at-tributes differ in their importance, any comparison in theireffectiveness will be confounded. Participants used a 7-point scale (1 = “Not at all important,” 7 = “Very impor-tant”) to answer the following question: “If you weregoing to buy ice cream, how important would the follow-ing features be to you?” (see Darley and Smith 1993). Twoattribute importance indices were constructed by averag-ing the importance ratings for the three search and threeexperiential attributes. A 2 (attribute type: search vs.experiential) X 2 (expectation level: realistic vs. inflated)MANOVA with search attribute importance and experi-ential attribute importance as dependent variables indi-cates a significant attribute type effect [Wilks’ Lambda =.93, F(2, 98) = 3.65, p < .05]. Follow-up t-tests indicatethat the experiential attributes were perceived as beingmore important than the search attributes [M = 6.23 vs.M = 5.25, t(110) = 8.19, p < .001], thus attribute impor-tance calculated across the six attributes is used as acovariate in the analyses.

Measurement: Dependent Variables

Post-Trial Satisfaction. The post-trial satisfactionwas assessed by asking subjects “Considering everything,how satisfied are you with this ice cream?” measured ona 7-point semantic differential scale with 1 = “Verydissatisfied” to 7 = “Very satisfied.”

RESULTS

To test H1 and H2, a 2 (attribute type: search vs.experiential) X 2 (expectation level: realistic vs. inflated)ANCOVA, with attribute importance as a covariate, wasconducted. The dependent variable is participants’ post-trial satisfaction with the ice cream. ANCOVA resultsindicate a significant effect of expectation level [F(1,98) = 9.59, p < .01], a non-significant attribute type effect[F(1, 98) = 1.01, p = .31] and a non-significant interaction[F(1, 98) = .29, p = .59]. The effect of the attribute

importance as a covariate was not significant [F(4, 253) =.22, p = .64]. To test the specific hypotheses, we per-formed post-hoc comparisons, which indicate no differ-ence between the inflated (M = 5.74) and the realisticexpectation conditions (M = 5.16, t(54) = 1.65, p = .11) forexperiential attribute information. Thus, H1 is not sup-ported. Further, participants in the search-inflated expec-tations condition reported more satisfaction with the icecream (M = 6.10) than the participants in the search-realistic expectations condition (M = 5.28, t(53) = 3.19,p < .01). Thus, H2 is supported (see Figure A–1).

H3a and H3b stated that the participants in the searchattribute-inflated expectations condition will report highersatisfaction than the participants in the experiential at-tribute (inflated or realistic expectations) condition.Planned comparisons indicate that creating higher expec-tations for search attributes (M = 6.10) indeed leads tohigher satisfaction than creating realistic expectations forexperiential attributes (M = 5.16, t(53) = 2.66, p < .05)(H3b), but only marginally higher satisfaction than creat-ing inflated expectations about experiential attributes(M = 5.74, t(59) = 1.77, p = .08) (H3a).

DISCUSSION AND MANAGERIAL

IMPLICATIONS

This research integrates and extends past findings inthe expectations and satisfaction literature and advertis-ing and product trial area in numerous ways. First, wefocus on a highly diagnostic product, for which researchhas generally found no effect of pre-trial advertising onpost-trial evaluations. Second, we distinguish betweenattribute type claims and suggest differential impacts ofthese claims on consumers’ post-trial satisfaction, de-pending on the claims’ generated level of expectation.This section summarizes the hypotheses and the findings,introducing explanations for the results that were notsupported. We conclude with implications for managers,consumer welfare, and public policy officials, and direc-tion for future research.

H1 predicted that generating inflated expectationsabout experiential attributes will result in consumersreporting lower satisfaction than creating realistic expec-tations about the same attributes. The results show nosignificant difference between the two conditions. Inother words, creating inflated expectations about experi-ential attributes did not hurt the product. The results aresimilar to Marks and Kamins’ (1988) findings that indi-cate that slightly exaggerated (versus realistic) claimsabout both search and experiential attributes combineddid not result in negative evaluations of the product. Thismay be due to the ambiguous nature of product experi-ence, in general. As Hoch (2002) argues, “most of theexperience carries with it a certain level of ambiguity” (p.448), which means that the inflated expectations about


experiential attributes cannot be fully disconfirmed dur-ing trial. More research is needed to find the mechanismsthat lead to consumers being able or willing to assimilatethe higher expectations into the performance evaluationand satisfaction.

As predicted in H2, in the context of highly diagnos-tic products, generating higher expectations about searchattribute information results in customers feeling moresatisfied with the product experience. Thus exaggerationof search attribute information in an advertisement pre-ceding product trial may have beneficial effects on con-sumer welfare, making consumers more satisfied with theproduct. The results extend past research findings thatindicate that higher expectations lead to higher satisfac-tion if the performance of the product is ambiguous(Oliver 1997) by showing that the effect holds true also forattribute information that is more effectively conveyed byadvertising (i.e., search attribute information) (Wrightand Lynch 1995).

H3a and H3b predicted that inflated expectationsabout search attributes result in higher satisfaction than

experiential attribute information generating either in-flated or realistic expectations. We found marginal sup-port for the former and support for the latter hypothesis.The results indicate that, in the context of highly diagnos-tic products, managers will be able to generate a highersatisfaction level among consumers if they convey search(as opposed to experiential) attribute information thatcreates inflated expectations.

Thus, how can firms maximize satisfaction? Theanswer that results from our research indicates that man-agement must instill high expectations about search asopposed to experiential attributes. But how high shouldthe expectation level be? Management must decide theappropriate level of (high) expectations when commu-nicating information about search attributes, keeping inmind that too high expectations may lead to negativedisconfirmation, which results in dissatisfaction (Marksand Kamins 1988). Thus, the management of expectationsis a key component of the marketing strategy.

The results have implications for public policy offi-cials as well (when to, and when not to, regulate). The

Figure A-1

Post-trial Satisfaction

Realistic Expectations (grey) and Inflated Expectations (black) Means for Attribute

Type Conditions

5.745.28 5.16

6.10

Search Experiential

Experimental conditions

Sa

tisf

act

ion

FIGURE 1

Post-Trial Satisfaction Realistic Expectations (grey) and Inflated Expectations (black)

Means for Attribute Type Conditions


officials have to arbitrate in the matter of exaggeratedclaims that produce inflated expectations, but they en-courage advertising that is beneficial to customers (seecomparative advertising). The results of this study helpthe public policy officials know under what conditionscreating inflated expectations may be beneficial to con-sumers. For example, stretching information about healthand nutritional benefits of certain foods within reasonableboundaries may make consumers more satisfied with theproduct experience.

Finally, several limitations need to be mentioned.First, the study was considering an attribute-based deci-sion making. Consumers make affective- or attitude-based decision in some situations. Future research mayexamine how the product expectations in those types ofdecisions affect post-trial customer satisfaction, as well asinvestigate how inflated expectations about credence at-tributes (e.g., healthiness and freshness of food) can affectsatisfaction levels.

REFERENCES

Darley, William K. and Robert E. Smith (1993), “Adver-tising Claim Objectivity: Antecedents and Effects,”Journal of Marketing, 57 (October), 100–13.

Deighton, John (1984), “The Interaction of Advertisingand Evidence,” Journal of Consumer Research, 11(December), 763–70.

____________ and Robert M. Schindler (1988), “CanAdvertising Influence Experience?” Psychology andMarketing, 5 (Summer), 103–15.

Ford, Gary T., Darlene B. Smith, and John L. Swasy(1990), “Consumer Skepticism of Advertising Claims:Testing Hypotheses from Economics of Informa-tion,” Journal of Consumer Research, 16 (4), 433–41.

Fornell, Claes (1992), “A National Customer SatisfactionBarometer: The Swedish Experience,” Journal ofMarketing, 56 (1), 6–21.

Hoch, Stephen J. and Young-Won Ha (1986), “ConsumerLearning: Advertising and the Ambiguity of ProductExperience,” Journal of Consumer Research, 13(September), 221–33.

____________ (2002), “Product Experience is Seduc-tive,” Journal of Consumer Research, 29 (Decem-ber), 448–54.

Kempf, DeAnna S. and Robert E. Smith (1998), “Con-sumer Processing of Product Trial and the Influenceof Prior Advertising: A Structural Modeling Ap-proach,” Journal of Marketing Research, 35 (Au-gust), 325–38.

Kopalle, Praveen K. and Donald R. Lehmann (2006),“Setting Quality Expectations when Entering a Mar-ket: What Should the Promise Be?” Marketing Sci-ence, 25 (January/February), 8–24.

Kotler, Philip (2000), Marketing Management: Analysis,Planning, Implementation, and Control, 10th ed. UpperSaddle River, NJ: Prentice Hall.

Marks, Lawrence J. and Michael A. Kamins (1988), “TheUse of Product Sampling and Advertising: The Ef-fects of Sequence of Exposure and Degree of Adver-tising Claim Exaggeration on Consumers’ BeliefStrength, Belief Confidence, and Attitudes,” Journal

of Marketing Research, 25 (August), 266–81.Micu, Camelia C. and Robin A. Coulter (2007), “Adver-

tising and Product Trial: The Impact of Product Typeand Attribute Information on Consumer Evaluations,”American Marketing Associations 2007 Winter Edu-cators’ Conference Proceedings, Chicago: Ameri-can Marketing Association.

Morgan, Neil A., Eugene W. Anderson, and Vikas Mittal(2005), “Understanding Firms’ Customer Satisfac-tion Information Usage,” Journal of Marketing, 69(3), 131–51.

Ofir, Chezy and Itamar Simonson (2007), “The Effect ofStating Expectations on Customer Satisfaction andShopping Experience,” Journal of Marketing Re-search, 164–74.

Oliver, Richard L. (1997), Satisfaction: A BehavioralPerspective on the Consumer. New York : McGrawHill.

Olson, Jerry C. and Philip Dover (1979), “Disconfirmationof Consumer Expectations Through Product Trial,”Journal of Applied Psychology, 64 (April), 179–89.

Schwarz, Norbert and Herbert Bless (1992), “Construct-ing Reality and Its Alternatives: An Inclusion/Exclu-sion Model of Assimilation and Contrast Effects inSocial Judgment,” The Construction of Social Judg-ments, Leonard Martin and Abraham Tesser, eds.Lawrence Erlbaum Associates, Inc.: Hillsdale, NJ,217–45.

Shapiro, Steward and Mark T. Spence (2002), “FactorsAffecting Encoding, Retrieval, and Alignment ofSensory Attributes in a Memory-Based Brand ChoiceTask,” Journal of Consumer Research, 28 (March),603–17.

Smith, Robert E. and William R. Swinyard (1983), “Atti-tude-Behavior Consistency: The Impact of ProductTrial Versus Advertising,” Journal of MarketingResearch, 20 (August), 257–67.

____________ (1993), “Integrating Information fromAdvertising and Trial: Processes and Effects on Con-sumer Response to Product Information,” Journal ofMarketing Research, 30 (May), 204–19.

Wong, Andrew K.C. (2005), “No Greater Satisfactionthan to Vindicate Expectation: How Affective Ex-


pectations Shape Consumption Experience,” 2005AMA Winter Educators’ Conference: MarketingTheory and Applications, American Marketing As-sociation, 34–35.

Woodruff and Gardial (2001), Know Your Customer:New Approaches to Understanding Customer Value

and Satisfaction. Blackwell Press.Wright, Alice Ann and John G. Lynch (1995), “Commu-

nication Effects of Advertising versus Direct Experi-ence When Both Search and Experience Attributesare Present,” Journal of Consumer Research, 21(March), 708–18.

For further contact information:Camelia C. Micu

Charles Dolan School of BusinessFairfield University

1073 North Benson RoadFairfield, CT 06824

Phone: 203.254.4000, Ext. 2894Fax: 203.254.4105



HOW DO INDIVIDUAL-LEVEL AND COUNTRY CHARACTERISTICS

AFFECT THE INTENTION TO QUIT SMOKING?

Gianfranco Walsh, University of Koblenz, GermanyLouise May Hassan, University of Stirling, United Kingdom

Edward Shiu, University of Strathclyde, United Kingdom

ABSTRACT

Whether antismoking advertising should be used forprevention largely depends on its effectiveness. However,the scarce conclusive evidence of a direct link betweenantismoking advertising and smoking cessation inten-tions has not resolved this issue. The authors develop andtest a model of individual-level drivers that affect smok-ers’ inclination to think more responsibly about smokingand country characteristics that affect intentions to quit.They test hypotheses on data collected from more than25,000 consumers from all 25 countries of the EuropeanUnion. The results suggest that individual-level driversinfluence smokers’ inclination to think responsibly andthat country context engenders intentions to quit smoking.

INTRODUCTION

Social marketers try to solve social problems bychanging long held, deep-seated beliefs and associatedbehaviors that have a detrimental effect on consumers’well-being (Kotler and Andreasen 1996). In the UnitedStates and Europe, which host approximately 50 millionand 215 million smokers, respectively, smoking repre-sents a major social and health issue with adverse effectson consumers’ well-being (Anonymous 2007; Helliker2007). The largest single cause of preventable death,smoking accounts for some 440,000 deaths in the UnitedStates and more than 500,000 deaths in the EuropeanUnion (EU) per year (Anderson 2007; ASPECT 2004;Helliker 2007).

In response to this major cause of preventable diseaseand death, the EU has instituted various tobacco controldirectives (e.g., European Council 2001, 2003) andlaunched a multimillion dollar HELP antismoking adver-tising campaign across the 25 EU member states (Euro-pean Commission 2005). The HELP campaign aims tohighlight the harmful effects of both active and passivesmoking and encourages smokers to think more responsi-bly about their habit, including the harm it can do tononsmokers, to prompt them to consider quitting. Itsprinciple method is a series of television advertisementsthat employ identical visual content with equivalentvoiceover messages in the native language for each mem-ber state.

Social marketing literature contains much discussionabout the use of antismoking advertisements and theireffectiveness, but little evidence suggests promoting an-tismoking goals can succeed in getting smokers to quit(Lewit et al. 1997; Wolburg 2004). The internationalreach of the HELP and other antismoking campaignsmake it pertinent to examine how the campaign’s effec-tiveness (in terms of instigating smoking cessation) mayvary depending on smokers’ and country characteristics.Such an examination could improve our understanding ofthe factors that drive smoking cessation and also contrib-ute to the hitherto scarce cross-cultural social marketingliterature. Extant literature on drivers of smoking cessa-tion tends to focus on individual-level drivers, such assmokers’ comprehension of and attitude toward anti-smoking messages (e.g., Andrews et al. 2004; Hassanet al. 2007), but, with a few exceptions (e.g., Reardonet al. 2006), largely neglects the role of culture in anti-smoking advertising and smoking cessation. Reardonet al. (2006, p. 116) even bemoan the “paucity of interna-tional and cross-cultural studies” in the literature onantismoking advertising.

This study makes a threefold contribution. First, wedevelop a conceptual model that considers mediating anddirect effects and assumes that individual-level behav-ioral change depends on the conscious processing of amessage in terms of comprehension and elaboration; italso examines the relationship between individual-levelindependent variables (age, message comprehension, andattitude toward the campaign) and outcome variables(elaboration and intention to quit smoking), positing thatthinking more responsibly about the behavior plays acentral mediating role. Second, we use insights fromelaboration likelihood theory to examine how individual-level characteristics affect thinking. We also draw oninstitutional theory to explore the direct effect of thecultural/institutional context in which smokers live ontheir intention to quit smoking. To the best of our knowl-edge, ours is the first study to model the impact of theinstitutional context on smokers’ intentions to quit. Third,we test the hypotheses using data from 25 different EUcountries collected immediately after the airing of thestandardized EU HELP television campaign.


CONCEPTUAL MODEL AND HYPOTHESES

Our model includes three individual-level character-istics as drivers of responsible thinking: smokers’ age,message comprehension, and attitude toward the anti-smoking advertisement. In line with Hassan et al. (2007),we posit that responsible thinking is a key component ofthe advertising persuasion model that fully mediates theeffects of age, comprehension, and attitude on intention toquit smoking. Our conceptual model further postulatesthat country characteristics directly influence smokers’intention to quit smoking. The conceptual model shown inFigure 1 delineates the drivers of responsible thinking andintention to quit smoking. Next, we discuss the rationalefor each relationship and formulate hypotheses.

Direct and Indirect Effects of Individual-Level Driv-

ers

Age. Social marketing literature considers age animportant variable in its own right (Pechmann and Reibling2006), and in the context of smoking, adolescents repre-sent a particularly vulnerable group. For example, Lewitet al.’s (1997) study examines, among other aspects, theeffect of exposure to pro-smoking and antismoking mes-sages on teens’ smoking behavior. They report that fre-quency of exposure to both pro-smoking and antismokingadvertising correlates positively with increased smoking.This latter positive link between antismoking advertising

and smoking seems somewhat counterintuitive but maypoint to a larger problem, namely, the ineffectiveness ofantismoking advertisements among younger target audi-ences. Moreover, it challenges Romer and Jamieson’s(2001) argument that antismoking advertising operates ina similar and opposite manner to that of tobacco advertis-ing.

Drawing on social learning theory, we expect youngersmokers to be more prone to pro-smoking advertisements.Social learning theory suggests that people learn throughthe process of observing and imitating role models(Bandura 1977). Flynn et al. (1994) report that teensexposed to antismoking advertisements featuring non-smoking role models are less likely to smoke later in life,though recent research suggests that smoking teens tendto underestimate the chances that they will continuesmoking, and nonsmoking teens underestimate the chancethat they will start to smoke (Schoenbaum 2005). Further-more, adolescents may be more receptive to the promo-tional practices of the tobacco industry and more prone topeer and social influences (Gilpin, Pierce, and Rosbrook1997; Pechmann and Knight 2002), which may make itmore difficult to convince them, compared with adults, ofthe harms of smoking through antismoking advertising.We therefore posit that younger smokers elaborate less onthe merits of the antismoking HELP advertisements, whichleads to our first hypothesis:

FIGURE 1

Conceptual Model

Attitude toward advertising

Intention

to change behavior

Three pillars of institutions(cultural characteristics)

Regulative

Normative/Moral

Cultural Cognitive Individualism

Good

citizenship

Government

effectiveness

Age

Individual-level drivers

H1

H2

H3

H4

H7

H8

H9

Message comprehension

Responsiblethinking

H5

H6Attitude toward advertising

Intention

to change behavior


Regulative

Normative/Moral

Cultural Cognitive Individualism

Good

citizenship

Government

effectiveness

Age


H1

H2

H3

H4

H7

H8

H9


Responsiblethinking

H5

H6Attitude toward advertising

Attitude toward advertising

Intention

to change behavior


Regulative

Normative/Moral

Cultural Cognitive IndividualismIndividualism

Good

citizenship

Good

citizenship

Government

effectiveness

Government

effectiveness

AgeAge


H1

H2

H3

H4

H7

H8

H9


Responsiblethinking

H5

H6


H1: Smokers’ age relates positively to their degree of

responsible thinking.

Comprehension is well accepted as an essential firststep in the persuasion process (Romaniuk et al. 2004;Wells 1964), and without it, a major opportunity toinfluence consumers is lost. In this study, we conceptual-ize comprehension as a unidimensional construct that weuse to assess the amount of correctly attributed meaningsderived from the advertised message. Hassan et al. (2007)emphasize the central role comprehension plays in theadvertising persuasion process in an antismoking promo-tional context and demonstrate a direct causal link be-tween comprehension and elaboration of the advertisedmessage.

Attitude. Antismoking advertising is arguably thesocial marketing communications tool most frequentlyused to influence or change consumers’ attitude towardsmoking. Therefore, we consider the central paradigm inthe marketing literature regarding the impact of consum-ers’ attitude toward the ad on their attitude toward andintention to buy the advertised products (e.g., MacKenzie,Lutz, and Belch 1986). Furthermore, attitude has beenfound to be stable over time and to have a strong impacton engagement with the persuasive appeal in terms ofthinking about the message and associated issues (Priesteret al. 2004).

Two components underlie attitude: cognition (per-ceived utility) and affect (feelings). Cognitive modelsassume rational information processing by the receiver,whereas affect models assume that attitudes developthrough emotional responses (Dabholkar 1996). In thecontext of social marketing, cognition gains importancebecause of the serious nature of the message and theintractability of the beliefs and behaviors being addressed(Borland 1997). Assessing utility is therefore vital, andattitude evaluations should seek to establish how under-standable and worthwhile the campaign is.

Responsible Thinking. The intractability of the be-haviors involved in social marketing means that a highlevel of elaboration – or continued engagement withpersuasive and credible messages that address underlyingbeliefs – also is crucial. Petty and Cacioppo’s (1986)elaboration likelihood model characterizes elaboration asthinking about the message and its meanings, as well asassessing the merits of the information and the argumentspresented. According to the model, a high level of elabo-ration would result in changes in consumer beliefs aboutthe behavior’s attributes and benefits, which then resultsin greater motivation for behavioral change. In our case,we conceptualize elaboration as responsible thinking, thatis, an inclination by smokers to think responsibly abouttheir smoking.

The marketing literature primarily discusses respon-sible behavior in relation to organizations and in thecontext of ethical behavior (e.g., Lichtenstein, Drumwright,and Braig 2004; Sen and Bhattacharya 2001). Hassanet al. (2007) shift the focus from firms to consumers andargue that consumers who are highly involved with themessage are more likely to engage in responsible think-ing. According to Schlenker et al. (1994), responsibilitymakes people accountable for their actions, whether tothemselves or to an audience, and depends on threeconditions: a person’s social or personal standards withclear implications for a particular action, identificationwith these standards (identity relevance), and control overthe action. The HELP campaign addresses these sameissues and clearly states the consequences of tobacco usefor both smokers and nonsmokers, as well as the actionsnecessary to mitigate these effects.

Drawing on these discussions, we make several de-ductions about the interrelationships of comprehension,attitude, and responsible thinking. First, comprehensionis a necessary prerequisite and hence an antecedent tomessage elaboration (i.e., responsible thinking). Second,attitude toward the ad should facilitate or hinder mentalengagement with the message and therefore also serve asan antecedent to responsible thinking. In turn, we proposethe following hypotheses regarding the links betweenindividual-level drivers and responsible thinking:

H2: Smokers’ message comprehension positively affects

their responsible thinking.

H3: Smokers’ favorable attitude toward the ad positively

affects their responsible thinking.

Intention. Finally, motivation or intention to initiatebehavioral change represents a central aim of any socialmarketing campaign. Assessments of the motivation toalter behavior, such as quitting smoking, as a consequenceof viewing the advertisement are essential for an overallunderstanding of the persuasion process. Without signifi-cant motivation, the consequential change in behavioreither does not take place or is short lived. In line withHassan et al. (2007), we consider responsible thinking akey component of the advertising persuasion model thathas a direct effect on motivation to quit smoking. There-fore,

H4: Smokers’ responsible thinking positively affects their

intentions to quit smoking.

According to Greenwald and Leavitt (1984), com-prehension of a message results in memory traces at thepropositional level. However, they further assert thatwithout elaboration, receivers do not form an associationbetween the message’s propositional content and their


existing knowledge. Elaboration thus refers to the processof relating new information in the message with existingpropositional knowledge. Consistent with previous re-search, we argue that message comprehension in itself isnot a sufficient condition to prompt behavioral changeintentions. In a social marketing context, message com-prehension without subsequent elaboration probably willnot lead to message acceptance, without which any behav-ioral intention in accordance with the campaign’s goalsremains weak (e.g., Areni 2002; McGuire 1985). In addi-tion, distinct elements in a message may attract attentionto the advertisement but detrimentally affect messagecomprehension by directing attention to the distinct ele-ments and away from the message claims (Sternthal andCraig 1973). Therefore, we predict no significant directrelationship between comprehension and intention.

Similarly, we do not expect attitude toward the cam-paign to have a significant direct influence on intention toquit smoking. Depending on the attitude object and situ-ational and consumer characteristics, the association be-tween attitude and behavioral intention may be weak(Ajzen and Fishbein 1977). For example, a consumer mayhave a positive attitude toward healthy food choices anda healthy lifestyle but not intend to buy healthy foodoptions because they are more expensive than “regular”food and the consumer cannot afford them. However,some research (Murry, Lastovicka, and Singh 1992) indi-cates that attitude toward the ad can predict behavioralintent, and Greenwald and Banaji (1995) argue that im-plicit cognition explains the lack of correspondence be-tween consumers’ expressed attitudes and their behavior.The concept of implicit cognition posits that past experi-ences affect attitudes, even if people do not remember theinfluential experience. In our context, implicit cognitionmay suggest that some smokers remember the advertise-ments and form attitudes toward the campaign but cannotrecall why they have formed positive or negative attitudesin the first place, which means the campaign will have aweak or negligible influence on their behavioral inten-tions. On balance, we predict no significant direct rela-tionship between attitude and intention. Hence,

H5: Smokers’ level of message comprehension does not

affect their intention to quit smoking.

H6: Smokers’ attitude toward the ad does not affect their

intention to quit smoking.

Furthermore, we do not hypothesize a main effect ofage on intentions to quit smoking, because it is notobvious why smoking cessation intention might be sys-tematically influenced by smokers’ age.

Direct Effects of Country-Level Drivers

Cultural characteristics likely directly influence smok-ers’ intention to quit smoking. Treating cultural character-

istics as drivers of consumers behavior aligns with previ-ous research (e.g., Aaker and Maheswaran 1997; Moneyet al. 1998), which notes that people of a country tend tobehave in accordance with that country’s cultural norms(Usunier 2000). In this section, we therefore draw oninstitutional theory to explore how national contextualcharacteristics may affect smokers’ intentions to quitsmoking.

Smokers are constituents of larger institutional sys-tems within their respective nations. Their views, atti-tudes, and behaviors not only stem from their own evalu-ative structure but also account for and reflect the conflictsand compatibilities imposed through their respective in-stitutional systems. According to Giddens (1984, p. 24),institutions are “the most enduring features of sociallife . . . giving ‘solidity’ across time and space.” Thefundamental issues, values, and fabric of a society mani-fest themselves through institutional priorities, which inturn govern citizenship behaviors. A stable and orderlysociety requires shared institutional priorities among con-stituents, reinforced by instituted social and economicreward contingencies. Thus, in some countries, smokingas an individual behavior with wider social costs probablyis considered less conducive to maintaining institutionalpriorities than in others.

Scott (2001) sets forth three distinct pillars of institu-tions: regulative, normative/moral, and cognitive-cultural.That is, an institution is enforced by law (regulative),supported by societal norms (normative/moral), and cultur-ally anchored in terms of common beliefs, values, andbehaviors (cognitive-cultural). These pillars regulatesociety’s interactions and thus likely affect smoking-related behavior. Some key characteristics of the threepillars of institutions appear in Table 1; we choose one keycomponent to represent each of them, namely, govern-ment effectiveness (regulative), good citizenship (norma-tive/moral), and national-cultural individualism (cogni-tive-cultural).

The regulative institutional pillar represents the rulesand laws of the institutional environment. Regulativeinstitutions directly relate to rule setting, monitoring, andsanctioning activities in a society, such as laws statingwhich behaviors are condoned (Scott 2001). Power, au-thority, and coercion also play important roles in theenactment of regulative institutions. Because these insti-tutions reflect rules and regulations, failing to obey themresults in legal sanctions (e.g., not complying with asmoking ban in bars), and a country’s government effec-tiveness is central to its ability to issue sanctions andmaintain the regulative system.

In turn, government effectiveness refers to the qualityof public services, civil services, and policy formulationand implementation, as well as the degree to whichservices are independent from political pressures and the


credibility of the government’s commitment to policies(Kaufmann et al. 2005). Higher levels of governmenteffectiveness engender wider public policy and welfareactivities. Because organized states offer health provi-sions, smokers in countries characterized by higher levelsof government effectiveness may be less fearful of theconsequences of smoking, and hence less inclined to quit,because they expect the government to take care of thesick. Furthermore, many smokers tend to think the gov-ernment has no business telling them whether or not tosmoke, particularly among citizens of the EU, who tend toview “interference” from the central “federal” EU Com-mission with skepticism. Because higher levels of gov-ernment effectiveness likely provide more comprehen-sive antismoking measures, smokers may perceive thattheir freedom is being threatened and determine that theywill resist, which makes them less likely to quit smokingas a consequence of government policies.

H7: Government effectiveness relates negatively to smok-

ers’ intentions to quit smoking.

The normative/moral institutional pillar refers tovalues, beliefs, norms, and assumptions that exist in theinstitutional environment and thus captures prescriptive,evaluative, and obligatory dimensions in social life andprovides structures of acceptable behavior (Scott 2001).Normative institutions encompass heuristics, rules ofthumb, operating procedures, occupational standards, and

educational curricula (Hoffman 1999). In addition, theyare based on social interactions and obligations and com-prise values and norms (Scott 2001; Wicks 2001). Theirability to influence smokers’ behavior stems from confor-mity goals, social obligations, social necessity, and sharedunderstandings of what is proper among a cultural group.In many Western countries, the popularity of and society’scondoning of smoking has declined during the past twodecades (Moore 2005; Rozin and Singh 1999), whichsuggests smokers have greater difficulty obtaining sociallegitimacy. In this context, Rozin and Singh (1999) findthat over time, moral opinions and attitudes can change, aprocess they refer to as “moralization.” Previous researchin marketing also notes the important role of one particu-lar construct in explaining organizational behavior, namelygood citizenship (e.g., Donavan et al. 2004; Piercy et al.2006). We shift the focus to country-level citizenshipbehavior of individuals.

Citizenship in modern society draws from ideas suchas political participation, social justice, humanitarianism,community involvement, voluntarism, and shared re-sponsibilities (Box 1998; Fredrickson 1997). In countriesin which people value good citizenship, we expect smok-ers to be more motivated to abide by societal norms for thegreater good of society. Hence,

H8: Good citizenship relates positively to smokers’ inten-

tions to quit smoking.

TABLE 1

Three Pillars of Institutions

Regulative Normative/Moral Cognitive-Cultural

Basis of compliance Expedience Social obligation Taken-for-grantedness,shared understanding

Basis of order Regulative rules Binding expectations Constitutive schema

Mechanisms Coercive Normative Mimetic

Logic Instrumentality Appropriateness Orthodoxy

Indicators Rules, laws, sanctions Certification, Prevalence, isomorphismaccreditation

Basis of legitimacy Legally sanctioned Morally governed Culturally supported,conceptually correct

Primary scientific Economics Sociology, Psychology Anthropology, Sociologydomain

Source: Adapted from Scott (2001)


Finally, the cognitive-cultural institutional pillar re-fers to widely shared social knowledge and cognitivecategories, such as stereotypes and schemata that repre-sent the models of “individual behavior based on subjec-tively constructed rules and meanings that dictate appro-priate thought, feeling and action” (Wicks 2001, p. 665).Cognitive-cultural institutions represent “symbols – words,signs, and gestures – as well as cultural rules and frame-works that guide understanding of the nature of reality andthe frames through which that meaning is developed”(Hoffman 1999, p. 353), which get reproduced throughmimetic processes (i.e., mimicking successful practices,implementing others). Compliance with cognitive com-ponents of the institutional environment occurs becauseof taken-for-granted traits or routines (Scott 2001), and akey component of the cognitive-cultural pillar involvesnational-cultural individualism (Allik and Realo 2004).

The construct of individualism/collectivism, gener-ally identified with Hofstede (1980), plays a prominentrole in cross-cultural research (Gudykunst and Ting-Toomey 1988; Kagitcibasi 1997; Maheswaran and Shavitt2000). This national culture dimension distinguishes be-tween individualistic and collectivistic societies, such thatpeople within individualistic societies tend to appreciateindependence and prefer a loosely knit social framework,whereas those in collectivistic societies tend to be emo-tionally integrated within a tightly knit social framework(Hofstede 1980; Triandis 1995). From an antismokingperspective, people from individualistic societies oftenare more concerned with their own personal needs, goals,and interests than are people from collectivistic societies(Triandis 1995).

Smokers in individualistic societies therefore havegreater concerns for themselves and their health, whichmay make them more prone to quit. In this context, Marinet al. (1990) examine cultural differences in attitudestoward smoking and smoking cessation and find that interms of attitudes, smokers from more collectivistic cul-tural groups (i.e., Hispanics) are more concerned aboutfamily-related consequences and effects on others (e.g.,bad smell) than are smokers from more individualisticcultural groups (i.e., non-Hispanic whites). Furthermore,drawing on Hofstede (2001), we might surmise that cul-tural differences with regard to internal locus of controlrelate to smokers’ quitting decisions, because they deter-mine whether smokers believe events in their lives (e.g.,poor health) are consequences of their own behavior.Specifically, an internal health locus of control mightrelate to better health and healthier behaviors in countriesthat score higher on individualism, compared with coun-tries high in collectivism (Piko 2004). Therefore,

H9: National-cultural individualism relates positively to

smokers’ intentions to quit smoking.

METHOD

Data Collection and Sample

Data was collected after two waves of the HELPcampaign had been aired on national and pan-Europeanchannels across the 25 EU member states in October 2005.In 23 of the member states, telephone interviews wereconducted, but in Lithuania and Portugal, personal inter-views provided the responses. The target sample size ineach country was 1,000, giving a final total sample size of25,113. In response to the 10-minute survey, 6,139 re-spondents indicated they were smokers, and of these,2,010 were aware of the campaign (5,189 respondentswho classified themselves as nonsmokers or former smok-ers also were aware of it). Of these 2,010 smokers, 56percent were men, and the average age of the respondentswas 37.8 years (SD = 14.62). The country sample sizesranged from 30 (Sweden) to 117 (U.K.), which exceed the15 responses per country recommended in the literature(Steenkamp and Geyskens 2006) for cross-cultural stud-ies of this type.

To explore the role of institutional theory, we also usesecondary data sources. Our primary data come from theEU Commission (the executive body of the EU), and ourstudy involves a reanalysis of these data.

IPSOS France, the research agency employed toconduct the interviews, developed the survey. Prior to itsimplementation, the survey instrument was sent to IPSOS’spartners in each EU nation for translation, and the Euro-pean Network on Smoking Prevention in each countryverified this translation to ensure the relevancy and con-sistency of the meanings of the phrases in the question-naire across the 20 EU national languages. A pilot test ofthe survey was undertaken in France by IPSOS withrespondents screened to ensure that they were aware of atleast one of the three campaign advertisements. A briefdescription of each advertisement was read to each re-spondent, and in response, they indicated if they had seenit. If respondents remembered at least one advertisement,they completed the remaining survey questions. Appen-dix 1 provides a list of the survey items used within themodel. The questionnaire also included measures of gen-der, age, and smoking status.

Measures

Independent Variables. Individual-level drivers.Age was measured with a single item that simply asks therespondent to report his or her age. In line with publishedresearch (e.g., Jacoby and Hoyer 1987; Thorson andSnyder 1984), we assess comprehension during post-exposure interviews that question respondents about theiridentification of the intended meanings. Message com-


prehension was assessed with eight items anchored on afive-point, “strongly agree” to “strongly disagree” re-sponse scale. To measure consumer attitude toward thecampaign, we use eight items based on a four-point “yes,definitely” to “no, not at all” response scale. We employaggregates for the message comprehension and attitudetoward the campaign measures and average the itemswithin each scale.

Independent Variables: Cultural (Institutional-Level) Characteristics. To gather country-level data, werely on three sources. We capture government effective-ness according to information from the World Bank,provided by Kaufmann, Kraay, and Zoido-Lobaton (1999).Six items from the European Social Survey (ESS 2003)use an 11-point scale spanning “extremely unimportant”to “extremely important” indicate the good citizenshipratings; we average the six items to create country scores.However, for countries with no available score (i.e.,Cyprus, Estonia, Latvia, Lithuania, and Malta), we con-sider these values missing. Finally, we assess individual-ism/collectivism using scores provided by Hofstede (2001),which are available for most counties except Cyprus,Latvia, and Lithuania. Because we lack scores for Cyprus,Latvia, and Lithuania for both good citizenship and indi-vidualism, we exclude these three countries from ouranalyses of country-level effects.

Dependent Variables. To measure responsible think-ing, we use four items that capture the extent to which thecampaign has led consumers to think responsibly abouttheir smoking (Hassan et al. 2007). The second dependentvariable, intention to quit smoking in response to thecampaign, relies on one item. The items for both depen-dent variables employ a four-point “yes, definitely” to“no, not at all” response scale, and we average the fourresponsible thinking items to form an aggregate measure.

Measurement Validation and Analysis Procedure

To test for convergent and discriminant validity, weconducted a confirmatory factor analysis of the indi-vidual-level constructs (excluding age). The fit of themodel is good, according to the usual conventions (Huand Bentler 1999), though the chi-square statistic is highlysignificant, as we expected: χ2 (184) = 1051.05, p < .001;comparative fit index (CFI) = .92; Tucker-Lewis index(TLI) = .90; and root mean square error of approximation(RMSEA) = .05. All indicators load significantly (p <.001)and substantively on their respective constructs with allstandardized factor loadings exceeding .40. These posi-tive results provide evidence of the convergent validity ofthe measures. In addition, Cronbach alpha values (andcomposite reliability) are acceptable with .74 (.60) formessage comprehension, .81 (.80) for attitude toward thecampaign, and .74 (.67) for responsible thinking. We use

Anderson and Gerbing’s (1988) procedure to test fordiscriminant validity, examine a series of nested modelsof the individual-level constructs, and compare a con-strained (correlation between each pair of constructsequals 1) with a free model. In all cases, we obtainsignificant results; therefore, the constructs of compre-hension, attitude, responsible thinking, and intention ex-hibit discriminant validity.

We employ a path model to test the hypothesizedrelationships and estimate the model using maximumlikelihood with AMOS 6.0. Gender appears in the modelas a control variable.

Results

In Table 2, we list the correlations among the modelindicators, and in Table 3, we present the results of thehypothesized model. All nine hypothesized paths aresignificant. Gender does not add to the explanation ofresponsible thinking, and message comprehension is fullymediated by responsible thinking, whereas attitude ispartially mediated. The model fits the data very well, withall indices greater than their recommended levels (χ2

(37) = 102.31, p < .01, CFI = .96, TLI = .91, RMSEA =.05).

We find that age affects responsible thinking, suchthat older respondents are more likely to engage in think-ing responsibly about their smoking as a result of thecampaign, as we propose in hypothesis 1. The findingsfurther suggest that, in agreement with hypotheses 2and 3, comprehension of the campaign message and atti-tude toward the campaign have positive effects on en-gagement with the message in terms of thinking responsi-bly about smoking. The results further show that the moresmokers think responsibly about smoking, the greatertheir motivation to quit, in support of hypothesis 4. Thesefindings are consistent with previous research (Hassanet al. 2007).

Turning to the direct relationships between compre-hension and attitude and intention to quit, we find thatmessage comprehension has no direct effect, and there-fore, comprehension is fully mediated by responsiblethinking, as we propose by hypothesis 5. However, incontrast with hypothesis 6, a positive attitude toward thecampaign has a positive direct effect on intention to quit,which may suggest that antismoking advertisements evokestronger attitudes that (compared with weak attitudes) aremore likely to instigate behavioral changes than are “nor-mal” attitudes (Petty and Krosnick 1995).

For the country-level effects, we find support for allthree hypotheses. Consistent with hypothesis 7, the effectof government effectiveness on intention to quit is nega-


tive; that is, smokers in countries whose citizens perceivetheir own government as effective are more resistant toEU-wide campaigns. In support of hypothesis 8, we findthat respondents from countries that place greater impor-tance on being a good citizen exhibit higher intentions toquit. Finally, countries with a higher level of individual-ism give rise to greater intentions to quit by respondentsin these countries, in support of hypothesis 9.

DISCUSSION

The inspiration for this research comes partly fromthe calls for more empirical research into the construct ofresponsible thinking and additional cross-cultural researchinto social marketing and consumer well-being. Socialmarketing campaigns advocate a particular behavior andare designed to influence people’s behavior in ways thatwill improve their well-being. Because smoking is amajor threat, researchers need a better understanding ofthe drivers of associated behavioral changes.

This article attempts to shed new light on the adver-tising persuasion process in an antismoking context, ad-vance the process of empirically integrating responsiblethinking as a relevant construct in an advertising persua-sion model, and test the model relationships along with themediating and direct effects within the model. Specifi-

cally, we formulate testable hypotheses of mediating andindividual- and country-level direct effects and test themodel with empirical data.

We find support for eight of our nine hypotheses inthe empirical data (see Table 3). The results of this studythus suggest that individual- and country-level character-istics have differential effects on smokers’ intention toquit. Specifically, smokers’ age, attitude toward the cam-paign, and message comprehension relate positively totheir proneness to think about the antismoking message.Moreover, responsible thinking fully mediates the rela-tionship between the independent variable comprehen-sion and behavioral intention and partly mediates therelationship between attitude and behavioral intention. Inaddition, all three country characteristics (i.e., pillars ofinstitutions) exert a direct influence on smokers’ intentionto quit smoking. We believe these findings represent animportant step forward in providing the field of socialmarketing and communication with a parsimonious ad-vertising persuasion model that explicitly considers re-sponsible thinking and culture-related drivers.

Contributions and New Insights

Our findings are consistent with consumer researchthat suggests responsible thinking is a central catalyst for

TABLE 2

Descriptive Statistics

Variable 1 2 3 4 5 6 7 8 9

1. Age 1.000

2. Comprehension .082 1.000

3. Attitude -.063 .473 1.000

4. Thinking .041 .461 .584 1.000

5. Intention .062 .271 .380 .562 1.000

6. Government effectiveness .081 .031 -.074 -.053 -.062 1.000

7. Good citizenship .033 -.116 -.066 -.019 .051 .095 1.000

8. Individualism .099 .013 -.118 -.116 -.039 .428 .095 1.000

Covariate 1.000

9. Gender .052 .064 -.028 -.004 -.026 .047 -.004 .075

Mean 37.890 1.045 3.069 2.640 2.010 1.340 6.811 64.682 1.460

SD 14.648 .816 .670 .881 1.138 .503 .428 15.469 .499

Notes: N = 1749 for correlations between the individual-level variables and between individual- and country-level variables, as well as for means and standard deviations for individual-level variables; N = 20/22 forcorrelations between country-level variables and for means and standard deviations for country-level variables.


attitudinal and behavioral change (Hassan et al. 2007) butalso extends previous work by considering age an impor-tant individual-level driver. Our contribution to theorylies in developing and testing a model that conceives ofresponsible thinking as an important outcome as well as amediator variable and that notes that age, comprehension,and attitude toward the campaign have differential im-pacts on responsible thinking. Smokers harm themselvesand endanger the health and well-being of others, sogetting smokers to reflect on their abhorrent behaviorseems a promising way of promoting smoking cessation.

Our findings also resonate with discussions of theeffects of culture on smoking-related behavior (Reardonet al. 2006) and highlight culture’s important role ininfluencing smokers’ intention to quit. We agree withReardon et al. (2006) that the importance of culture effectsin an antismoking context remains under researched andtherefore contribute to the expansion of the researchdomain by including cultural characteristics in a compre-hensive advertising persuasion model. Specifically, weshow that cultural characteristics directly affect smokers’intention to quit smoking, but government effectivenessexerts a negative influence. Thus, our study distinguishesbetween individual-level and country-level characteris-tics and attempts to measure mediated and direct effects ofthe persuasion levels of social marketing advertising.Furthermore, we believe that integrating responsible think-ing as an advertising persuasion consequence has benefits

at both conceptual and managerial levels, because ourresultant model enables social marketers to study andmeasure the persuasion effects of advertising at differentlevels of abstraction.

To measure persuasion effects without consideringpotential mediators entails losing the ability to determinehow different independent variables work. Researchersand social marketers should consider individual relation-ships from our model (lower level of abstraction) and alsolook at the overall model with its mediating effects (higherlevel of abstraction) to learn more about the drivers ofpersuasion in advertising.

Limitations and Future Research Direction

As with any research, our work is not free of limita-tions, some of which open avenues for further research.Additional work could investigate if our advertising per-suasion model is equally effective in other socially re-sponsible behavioral contexts, such as drunk driving orhealthy eating. In a similar vein, our model should betested in non-EU countries (e.g., Norway, Russia, Swit-zerland), as well as outside Europe, where smoking ratesare among the highest in the world. For example, theannual per capita cigarette consumption in Russia in-creased by 75 percent to 2,058 cigarettes from 1994 to2004 (Johnson 2006). During the same period, the annualper capita cigarette consumption in France decreased by

TABLE 3

Path Model Results

Standardized Supported

Predictor Coefficient CR p Yes/No

Individual-Level EffectsH1: Age Thinking .051 2.645 <.01 Yes

H2: Comprehension Thinking .240 10.004 <.001 Yes

H3: Attitude Thinking .471 21.072 <.001 Yes

H4: Thinking Intention .515 20.328 <.001 Yes

H5: Comprehension Intention .008 .294 n.s. Yes

H6: Attitude Intention .080 3.135 <.01 No

Country-/Institutional-Level Effects

H7: Government effectiveness Intention -.062 -2.819 <.01 Yes

H8: Good citizenship Intention .080 3.799 <.001 Yes

H9: Individualism Intention .068 3.079 <.01 Yes

Notes: n.s. = not significant at 5% level. CR = critical ratio.


42 percent to 1,556 and in the United Kingdom by 40percent to 1,448 cigarettes (Johnson 2006). Furthermore,the WHO (2002) reports that regions such as East Asiaand the Pacific have the highest smoking rates in theworld, with nearly two-thirds of men smoking.

Researchers also might consider adding independentvariables to the model’s specification. In this study, weexamine one marketing mix element, advertising, in rela-tion to consumers’ behavioral change. However, asCullwick (1975) stresses, the entire marketing mix playsa strategic role in demarketing (see also Gerstner, Hess,and Chu 1993), which suggests other antismoking mea-sures, such as increased taxes (price) or fewer consump-tion spaces and smoking bans (place), could be includedin the model as independent variables. Another extensionof the model might use the variables we address, espe-cially responsible thinking and the culture-related vari-

ables, to determine market segmentation. By taking theseconstructs one step further, researchers might identifydifferences within populations and explore the existenceof national and pan-European responsible thinking seg-ments. Because many tobacco companies have adoptedthe role of “good citizens” these days, in which contextthey provide information about the dangers of smokingand warn parents about teen smoking, it becomes evenmore pertinent to explore the degree of responsible think-ing across different groups of smokers.

We examine smokers’ perceptions of antismokingadvertisements, but the normative power of referencegroups and word-of-mouth influences suggests it alsomight be useful to investigate perceptions of nonsmokers.Finally, we test our conceptual model and the hypotheseswith cross-sectional data, so longitudinal data from amultistage survey could extend our findings even further.

REFERENCES

Aaker, Jennifer L. and Durairaj Maheswaran (1997),“The Effect of Cultural Orientation on Persuasion,”Journal of Consumer Research, 24 (December), 315–28.

Ajzen, Icek and Martin Fishbein (1977), “Attitude-Be-havior Relations: A Theoretical Analysis and Reviewof Empirical Research,” Psychological Bulletin, 84,888–918.

Allik, Juri and Anu Realo (2004), “Individualism-Collec-tivism and Social Capital,” Journal of Cross-Cul-tural Psychology, 35 (January), 29–49.

Anderson, James C. and David W. Gerbing (1988), “Struc-tural Equation Modeling in Practice: A Review andRecommended Two-Step Approach,” PsychologicalBulletin, 103 (3), 411–23.

Anderson, John Ward (2007), “France Isn’t Just BlowingSmoke,” The Wall Street Journal, (January) 30, 9.

Andrews, J. Craig, Richard G. Netemeyer, Scot Burton,D. Paul Moberg, and Ann Christiansen (2004), “Un-derstanding Adolescents’ Intentions to Smoke: AnExamination of Relationships among Social Influ-ence, Prior Trial Behavior, and Anti-Tobacco Cam-paign Advertising,” Journal of Marketing, 68 (July),110–23.

Anonymous (2007), “Nicotine Inhaler Maker Plans $500mln HK IPO-Paper,” Tobacco.org (accessed January27, 2007) [available at http://www.tobacco.org/news/230923.html].

Areni, Charles S. (2002), “The Proposition-ProbabilityModel of Argument Structure and Message Accep-tance,” Journal of Consumer Research, 29 (Septem-ber), 168–87.

ASPECT (2004), “Analysis of the Science and Policy forEuropean Control of Tobacco,” European Commis-

sion.Bandura, Albert (1977), Social Learning Theory.

Englewood Cliffs, NJ: Prentice-Hall.Borland, Ron (1997), “Tobacco Health Warnings and

Smoking-Related Cognitions and Behaviors,” Ad-diction, 92, 1427–35.

Box, Richard C. (1998), Citizen Governance: LeadingAmerican Communities into the 21st Century. Thou-sand Oaks, CA: Sage.

Cullwick, David (1975), “Positioning Demarketing Strat-egy,” Journal of Marketing, 39 (2), 51–57.

Dabholkar, Pratibha A. (1996), “Consumer Evaluationsof New Technology-Based Self-Service Options: AnInvestigation of Alternative Models of Service Qual-ity,” International Journal of Research in Marketing,13, 29–51.

Donavan, D. Todd, Tom J. Brown, and John C. Mowen(2004), “Internal Benefits of Service-Worker Cus-tomer Orientation: Job Satisfaction, Commitment,and Organizational Citizenship Behaviors,” Journalof Marketing, 68 (1), 128–46.

European Social Survey (2003), Technical Report, Lon-don: Centre for Comparative Social Surveys, CityUniversity.

European Commission (2005), “Help-for a Life WithoutTobacco,” Memo 05/68 (accessed December 21,2005) [available at http://europa.eu.int/rapid/pressReleasesAction.do?reference=MEMO/05/68&format=HTML&aged=0&language=EN&guiLanguage=fr].

European Council (2001), “Directive 2001/37/EC of theEuropean Parliament and of the Council of 5 June2001 on the Approximation of the Laws, Regula-tions, and Administrative Provisions of the MemberStates Concerning the Manufacture, Presentation andSale of Tobacco Products,” Official Journal of the


European Communities, L194/26.____________ (2003), “Directive 2003/33/EC of the

European Parliament and of the Council of 26 May2003 on the Approximation of the Laws, Regula-tions, and Administrative Provisions of the MemberStates Relating to the Advertising and Sponsorship ofTobacco Products,” Official Journal of the EuropeanCommunities, L152/16.

Flynn, Brian S., John K. Worden, Roger H. Secker-Walker, P.L. Pirie, Gary J. Badger, J.H. Carpenter,and Berta M. Geller (1994), “Mass Media and SchoolInterventions for Cigarette Smoking Prevention: Ef-fects 2 Years After Completion,” American Journalof Public Health, 84 (7), 1148–50.

Fredrickson, H. George (1997), The Spirit of PublicAdministration. San Francisco: Jossey Bass.

Gerstner, Eitan Hess, James D. Hess, and Wujn Chu(1993), “Demarketing as a Differentiation Strategy,”Marketing Letters, 4 (1), 49–57.

Giddens, Anthony (1984), The Constitution of Society.Berkeley, CA: University of California Press.

Gilpin E.A., J.P. Pierce, and B. Rosbrook (1997), “AreAdolescents Receptive to Current Sales PromotionPractices of the Tobacco Industry?” Preventive Medi-cine, 26 (1, January), 14–21.

Greenwald, Anthony G. and Clark Leavitt (1984), “Audi-ence Involvement in Advertising: Four Levels,” Jour-nal of Consumer Research, 11 (1), 581–92.

____________ and Mahzarin R. Banaji (1995), “ImplicitSocial Cognition: Attitudes, Self-Esteem, and Ste-reotypes,” Psychological Review, 102, 4–27.

Gudykunst, William B. and Stella Ting-Toomey (1988),Culture and Interpersonal Communication. NewburyPark, CA: Sage Publications.

Hassan, Louise M., Gianfranco Walsh, Edward Shiu,Gerard Hastings, and Fiona Harris (2007), “Model-ing Persuasion in Social Advertising – A Study ofResponsible Thinking in Antismoking Promotion inEight Eastern EU Member States,” Journal of Adver-tising, 36 (2), 15–31.

Helliker, Kevin (2007), “Nicotine Fix: How Drug Com-panies Influence U.S. Antismoking Policy,” The WallStreet Journal, (February 9), 12–13.

Hoffman, Andrew J. (1999), “Institutional Evolution andChange: Environmentalism and the U.S. ChemicalIndustry,” Academy of Management Journal, 42 (4),351–71.

Hofstede, Geert (1980), Culture’s Consequences: Inter-national Differences in Work-Related Values. BeverlyHills, CA: Sage Publications.

____________ (2001), Culture’s Consequences: Com-paring Values, Behaviors, Institutions, and Organi-zations Across Nations. Thousand Oaks CA: SagePublications.

Hu, Li-tze and Peter M. Bentler (1999), “Cutoff Criteriafor Fit Indices in Covariance Structural Analysis:Conventional Criteria Versus New Alternatives,”

Structural Equation Modeling, 6 (1), 1–56.Jacoby, Jacob and Wayne D. Hoyer (1987), The Compre-

hension and Miscomprehension of Print Communi-cations. New York: Advertising Educational Foun-dation.

Johnson, K. (2006), “Spain’s Smoking EstablishmentWould Rather Fight than Quit – Big Tobacco SlashesPrices, Bars Defy New Laws,” The Wall Street Jour-nal, 24 (38), 1 & 10.

Kagitcibasi, Cigdem (1997), “Individualism and Collec-tivism,” in Handbook of Cross-Cultural Psychology:Social Behavior and Applications, John W. Berry,Marshall H. Segall, and Cigdem Kagitcibasi, eds.Boston: Allyn & Bacon, 3, 1–49.

Kaufmann, Daniel, Aart Kraay, and Pablo Zoido-Lobaton(1999), Governance Matters. Policy Research Work-ing Paper 2196, Washington: The World Bank.

____________, ____________, and ____________(1999), “Aggregating Governance Indicators,” PolicyResearch Paper No. 2195, The World Bank.

____________, ____________, and Massimo Mastruzzi(2005), Governance Matters IV: Governance Indica-tors for 1996–2004. Washington, DC: World Bank.

Kotler, Philip and Alan Andreasen (1996), StrategicMarketing for Nonprofit Organizations, 5th ed. UpperSaddle River, NJ: Prentice-Hall.

Lewit, Eugene M., Andrew Hyland, Nancy Kerrebrock,and K. Michael Cummings (1997), “Price, PublicPolicy, and Smoking in Young People,” TobaccoControl, 6 (suppl 2), S17–S24.

Lichtenstein, Donald R., Minette E. Drumwright, andBridgette M. Braig (2004), “The Effect of CorporateSocial Responsibility on Customer Donations to Cor-porate-Supported Nonprofits,” Journal of Market-ing, 68 (October), 16–32.

MacKenzie, B. Scott, Richard J. Lutz, and George E.Belch (1986), “The Role of Attitude Toward the Adas a Mediator of Advertising Effectiveness: A Test ofCompeting Explanations,” Journal of MarketingResearch, 23 (May), 130–43.

Maheswaran, Durairaj and Sharon Shavitt (2000), “Issuesand New Directions in Global Consumer Psychol-ogy,” Journal of Consumer Psychology, 9 (2), 59–66.

Marin, Barbara Vanoss, Gerardo Marin, Eliseo J. Perez-Stable, Regina Otero-Sabogal, and Fabio Sabogal(1990), “Cultural Differences in Attitudes TowardSmoking: Developing Messages Using the Theory ofReasoned Action,” Journal of Applied Social Psy-chology, 20 (6), 478–93.

McGuire, William J. (1985), “Attitudes and AttitudeChange,” in Handbook of Social Psychology, Vol. II,3d ed., G. Lindzey and E. Aronson, eds. New York:Random House.

Money, R. Bruce, Mary C. Gilly, and John L. Graham(1998), “Explorations of National Culture and Word-of-Mouth Referral Behavior in the Purchase of In-


dustrial Services in the U.S. and Japan,” Journal ofMarketing, 62 (October), 76–87.

Moore, Robert S. (2005), “The Sociological Impact ofAttitude Toward Smoking: Secondary Effects of theDemarketing of Smoking,” The Journal of SocialPsychology, 145 (6), 703–18.

Murry, Jr., John P. John L. Lastovicka, and Surendra N.Singh (1992), “Feeling and Liking Responses toTelevision Programs: An Examination of Two Ex-planations for Media-context Effects,” Journal ofConsumer Research, 18 (March), 441–51.

Pechmann, Cornelia and Susan J. Knight (2002), “AnExperimental Investigation of the Joint Effects ofAdvertising and Peers on Adolescents’ Beliefs andIntentions About Cigarette Consumption,” Journalof Consumer Research, 29 (June), 5–19.

____________ and Ellen Thomas Reibling (2006), “Anti-Smoking Advertisements for Youth: An Indepen-dent Evaluation of Health, Counter-Industry, andIndustry Approaches,” American Journal of PublicHealth, 96 (5), 906–13.

Petty, Richard and John Cacioppo (1986), “The Elabora-tion Likelihood Model of Persuasion,” in Advancesin Experimental Social Psychology, L. Berkowitz,ed. New York: Academic Press, 19, 123–205.

Petty, Richard E. and Jon A. Krosnick, eds. (1995),Attitude Strength: Antecedents and Consequences.Hillsdale, NJ: Erlbaum.

Piercy, Nigel F., David W. Cravens, Nikala Lane, andDouglas W. Vorhies (2006), “Driving Organiza-tional Citizenship Behaviors and Salesperson In-Role Behavior Performance: The Role of Manage-ment Control and Perceived Organizational Sup-port,” Journal of the Academy of Marketing Science,34 (2), 244–62.

Piko, Bettina F. (2004), “Interplay between Self andCommunity: A Role for Health Psychology in East-ern Europe’s Public Health,” Journal of Health Psy-chology, 9 (1), 111–20.

Priester, Joseph R., Dhananjay Nayakankuppam, MoniqueA. Fleming, and John Godek (2004), “The A2SC2

Model: The Influence of Attitudes and AttitudeStrength on Consideration and Choice,” Journal ofConsumer Research, 30, 574–87.

Reardon, James, Chip Miller, Bram Foubert, Irena Vida,and Liza Rybina (2006), “Antismoking Messages forthe International Teenage Segment: The Effective-ness of Message Valence and Intensity Across Dif-ferent Cultures,” Journal of International Marketing,14 (3), 115–38.

Romaniuk, Jenni, Byron Sharp, Samantha Paech, andCarl Driesener (2004), “Brand and Advertising

Awareness: A Replication and Extension of a KnownEmpirical Generalisation,” Australasian MarketingJournal, 2 (3), 70–80.

Romer, Daniel and Patrick Jamieson (2001), “Advertis-ing, Smoker Imagery, and the Diffusion of SmokingBehavior,” in Smoking: Risk, Perception, & Policy,Paul Slovic, ed. Thousand Oaks, CA: Sage Publica-tions, 127–55.

Rozin, Paul and Leher Singh (1999), “The Moralizationof Cigarette Smoking in the United States,” Journalof Consumer Psychology, 8, 321–37.

Schlenker, Barry R., Thomas W. Britt, John Pennington,Rodolpho Murphy, and Kevin Doherty (1994), “TheTriangle Model of Responsibility,” PsychologicalReview, 101 (4), 632–52.

Schoenbaum, Michael (2005), “The Accuracy of Teens’Expectations of Future Smoking,” American Journalof Preventive Medicine, 28 (3), 274–80.

Scott, W. Richard (2001), Institutions and Organizations.Thousands Oaks, CA: Sage Publications.

Sen, Sankar and C.B. Bhattacharya (2001), “Does DoingGood Always Lead to Doing Better? Consumer Re-actions to Corporate Social Responsibility,” Journalof Marketing Research, 38 (2), 225–44.

Steenkamp, Jan-Benedict E.M. and Inge Geyskens (2006),“How Country Characteristics Affect the PerceivedValue of Web Sites,” Journal of Marketing, 70 (July),136–50.

Sternthal, Brian and Samuel S. Craig (1973), “Humor inAdvertising,” Journal of Marketing, 37, 12–18.

Thorson, Esther and Rita Snyder (1984), “Viewer Recallof Television Commercial Scripts,” Journal of Mar-keting Research, 21, 127–36.

Triandis, Harry C. (1995), Individualism and Collectiv-ism. Boulder, CO: Westview Press, Inc.

Usunier, Jean-Claude (2000), Marketing Across Cul-tures, 3d ed. Harlow, England: Prentice-Hall.

Wells, William D. (1964), “Recognition, Recall, andRating Scales,” Journal of Advertising Research, 4(3), 2–8.

WHO (2002), Smoking Statistics. (accessed January 22,2007) [available at http://www.wpro.who.int/media_centre/fact_sheets/fs_20020528.htm].

Wicks, David (2001), “Institutionalized Mindsets of In-vulnerability: Differentiated Institutional Fields andthe Antecedents of Organizational Crisis,” Organi-zation Studies, 22 (4), 659–92.

Wolburg, Joyce M. (2004), “The Need for New Anti-Smoking Advertising Strategies that Do Not ProvokeSmoker Defiance,” Journal of Consumer Marketing,21 (3), 173–74.


MEASUREMENT APPENDIX

Individual-Level ConstructsComprehension (adapted from Hassan et al. 2007)

It is hard to stop smoking

It is ridiculous to smoke to be like othersThere is help available to stop smokingThere is a telephone number and/or a website that one may contact about smokingSmoking is absurdSmokers and non-smokers need to respect each otherSmoking harms the health of others around youYou should not start smoking

Attitude toward the campaign (adapted from Hassan et al. 2007)

The campaign delivers a worthwhile messageThe campaign is interestingThe campaign is an incentive to look for information and / or helpThe campaign is meant for people like youThe campaign is easy to understandThe campaign is in people’s best interestThe campaign uses humor to convey the messageThe campaign talks about smoking in a new way

Responsible Thinking (adapted from Hassan et al. 2007)

The advert / advertisements made you think about your smokingThe advert / advertisements made you think that smoking is absurdThe advert / advertisements made you think about the behavior of smokers in the presence of non-smokersThe advert / advertisements made you think about the value of help in quitting smoking

Country-Level ConstructsGood Citizenship (from European Social Survey 2003)

To be a good citizen:How important to support people worse offHow important to vote in electionsHow important to always obey laws/regulationsHow important to form independent opinionHow important to be active in voluntary organizationsHow important to be active in politics

Individualism (we used scores provided by Hofstede 2001)

Government effectiveness (ratings provided by Kaufmann, Kraay, and Zoido-Lobaton 1999)

For further information contact:Gianfranco Walsh

University of KoblenzGermany



THE POTENTIAL EFFECTS OF CORRECTIVE ADVERTISING ON

CONSUMER BELIEFS MANDATED BY U.S. VS. PHILIP MORRIS USA, INC. (2006)

Andrea Heintz Tangari, University of Arkansas – FayettevilleBrook Plack, University of Arkansas – FayettevilleScot Burton, University of Arkansas – FayettevilleCraig Andrews, Marquette University, Milwaukee

SUMMARY

Recently, a United States federal court has orderedthe use of corrective advertising and promotion to aug-ment consumer knowledge about smoking by targetingpotential misperceptions in this area (U.S. vs. PhillipMorris USA, Inc. 2006). The goal of this corrective adcampaign is to mitigate inappropriate or inaccurate con-sumer beliefs and to thwart any future deceptive market-ing practices that could contribute to or encourage to-bacco use. Drawing from this court decision and previousresearch, this study addresses the following questions:

1. Will the strength of the specific beliefs identified bythe Court suggest that consumers have been misled ordeceived by the tobacco companies for some or all ofthese beliefs?

2. Do the proposed corrective ads have effects and arethese effects similar across all beliefs mandated inU.S. vs. Phillip Morris USA, Inc. (2006)?

3. What are the effects of antismoking corrective adsthat include text only versus ads that include text andgraphics on the focal belief themes related to smok-ing?

4. Will the strength of beliefs differ for smokers andnonsmokers, and will these corrective ads influenceintentions to quit smoking for current smokers?

Method

A pilot study was initially performed to developmulti-item measures of the six themes noted in the caseand of interest in predictions: (a) adverse health effects ofsmoking, (b) smoking addictiveness, (c) lack of healthbenefit from smoking light/low tar cigarettes, (d) tobaccocompanies’ manipulation of cigarette design to ensureoptimum nicotine delivery, (e) secondhand smoke ef-fects, and (f) tobacco company deceptiveness. Items forthe pilot study were generated through a review of theliterature (e.g., Andrews et al. 2004; Tangari et al. 2007).Participants’ beliefs in each of the themes are assessedusing multi-item seven point, Likert-type scales with

endpoints of “Strongly Disagree” (coded as a ‘1’) and“Strongly Agree” (coded as a ‘7’). Both factor analysesand reliability tests were used to reduce the number ofbelief theme items and develop reliable multi-item mea-sures, ranging in length from five to eight items. Allcoefficient reliability estimates for each of the measuresexceed .70.

In the main study, a 3 (corrective ad condition) X 6(belief themes) mixed experimental design was used. Thebetween-subjects factor is the corrective ad conditionwith three levels: (1) a control with no ad; (2) a correctivead containing copy only; and (3) a corrective ad thatcontained copy and two graphic visuals. The within-subjects factor consisted of the six belief themes. Partici-pants were undergraduate students at a Southern univer-sity. The sample size was 226 participants with 40 percentmales. Ages ranged from 18-36 with an average age of 23.Several measures were also included to address specificsmoker reactions, such as their desire to quit smoking.

Results

The means for the six beliefs were relatively strongacross all ad conditions. Means for the beliefs are particu-larly high for the corrective ad conditions. Even in the noad control condition, all belief means are significantlyabove the scale midpoint of ‘4’ (p < .01 for all tests). Allbelief themes were significantly different and higher inthe corrective ad conditions (p’s < .05) except for thedeceptiveness belief (p = .09). Also, there was a signifi-cant interaction (F(10,1115) = 3.07, p < .01) between thead condition and the belief themes, indicating that theeffect of the corrective ads is not consistent across allbeliefs. Specifically, the effect of the ad on the beliefregarding the health benefits of light/low tar cigarettes isstronger compared to the ad effect on other beliefs. Theaddition of the graphic visuals to the corrective adsincreased the overall strength of some (but not all) beliefthemes, compared to corrective ads without the visuals.

We also explored whether the effects of the correc-tive ads differ for the beliefs of smokers vs. nonsmokers.Beliefs regarding tobacco company deceptiveness, sec-ondhand smoke and health effects of smoking were all


significantly stronger (p < .01) for nonsmokers comparedto smokers. Finally, we also examined whether the adsaffect smokers’ intentions to quit. Results show thatcorrective ad exposure does not have a significant effecton the desire to quit smoking, although the pattern ofmeans is in the desired direction. (However, the samplesize of smokers is small (n = 55); larger samples mayprovide a stronger test on these intention variables.)

Summary

Although most of the belief themes were signifi-cantly affected by the corrective ads, many of the meanlevels for these belief themes were already high. Thisindicates that despite past misleading actions and market-ing tactics of tobacco companies, these study participantsdo not appear to have high levels of “incorrect” beliefs

about smoking and its consequences. There was a signifi-cant effect of exposure to the corrective ads, but thestrength of this effect varied across the beliefs. Specifi-cally, the light/low tar belief received the lowest level ofagreement in the control condition, and the ad exposurehad the strongest effect on this belief. Thus, the mosteffective approach may be to weight any corrective cam-paign toward the weaker beliefs, such as the light/low tarbelief theme, where the opportunity to “correct” con-sumer misperceptions appears to be the most substantial.The findings for the addition of graphics to advertise-ments offered mixed results with only some of the beliefthemes becoming significantly stronger. The findings ofthis research show that, in general, consumers’ beliefsabout smoking potentially can be affected with correctiveadvertising in a manner consistent with the objectives ofthe Court.

For further information contact:Andrea Heintz Tangari

Department of Marketing and Logistics 302Sam M. Walton College of BusinessUniversity of Arkansas – Fayetteville

Fayetteville, AR 72701–1201Phone: 479.575.4005



DOES PERCEIVED PRODUCT QUALITY IMPACT COUNTRY-OF-

ORIGIN IMAGE? AN EMPIRICAL INVESTIGATION

Qin Sun, University of North Texas, Denton

SUMMARY

Over the last 45 years, country of origin (COO) effecthas been extensively investigated in the marketing litera-ture. This continued interest in COO effect is partiallyfueled by the increasing presence of multinational corpo-rations in different countries and interconnection of glo-bal economy (Ahmed and d’Astous 2007; Mihailovich2006). Most studies in COO literature have focused on theimpact of COO on product evaluation, i.e., how countryimage can affect perceived product quality (Krishnakumar1974; Chao 1993; Insch and McBride 1998), consumerattitude (Papadopoulos et al. 1987; Smith 1993; Okechuku1994), and purchase intention (Wang and Lamb 1983;Roth and Remeo 1992; Han 1990; Baker and Michie1995; Tse et al. 1996). The researchers in this streamassume that country image is static and national stereo-type is the rule to judge the product/brand quality in aspecific product category. However, a few recent studieshave challenged this prevalent viewpoint, and have startedto look at the country image as a target variable rather thanas an explanatory variable. These studies argue that coun-try image is dynamic and can be impacted by changes inproduct image (Moss and Estrada 2002). Nonetheless, noempirical study has systematically investigated the im-pact of perceived product quality on country image. To fillthis gap, this study investigates the effect of perceivedproduct quality on the perception of COO image.

The impetus for this study comes from the fact thatpolicy makers may want to increase the country image toattract the foreign investments and to increase exports(Beverland and Lindgreen 2002). In addition,O’Shaughnessy and O’Shaughnessy (2000), Papadopou-los and Heslop (2000, 2002) and Paswan et al. (2002)propose that country is a brand. Specifically, there is anincreasing interest among practitioners and governmentagencies in how product/brand quality may influencecountry brand image. Against this backdrop, this studyintends to investigate the effect of perceived productquality on the perception of country of origin (COO). Thearguments presented in this study are theoretically an-chored in attribution theory. This paper has two majorcontributions: first, to the best of our knowledge, this is thefirst attempt in the marketing literature to empiricallyexamine the effect of product quality on COO image;second, this paper tests the contingent effects of some keyvariables such as consumer preference on the effect ofperceived product quality on COO image.

Research Method

We collected the data using self-administered sur-veys from students at one large U.S. public university. Therespondents were divided randomly into two groups inorder to avoid the speculations about the purpose of thesurvey. Respondents in one group were given the ques-tionnaire where they evaluated the image of country-of-origin based on the description of the “average” productquality, and the others in the second group were given adifferent questionnaire where they evaluated the COOimage based on the description of the “excellent” productquality. Two products categories – color TV and back-pack were selected because these are most commonlybought products by the target respondents. A total of 623students were asked to participate in the survey, but only391 completed the survey, of which 371 are usable (theresponse rate is 62.8% and the usable rate is 59.6%). Theproposed hypotheses were tested using multiple regres-sion method. The result substantiates our hypothesis thatperceived product quality positively influences the COOimage. There is a consensus that the country where theproducts with perceived excellent quality come from hashigher brand image than the country where the productswith perceived average quality come from. This findingcorroborates the arguments presented by authors such asLoo and Davies (2006). Improved perception of productquality can enhance the country image. This may in turnreinforce the perceived product quality, and then impactproduct attitudes and purchase intention.

In addition, we conclude that the product durability,social class, nationality of the respondents (especiallyU.S., China, Korea, and maybe Thailand), and consumerpreference for well-known brand have some influence onthe COO image, particularly on the COO-People dimen-sion. The results also indicate that nationality (UnitedStates, China, and Korea) could influence the COO-People image in durable product (e.g., color TV). Incomparison, for non-durable products (e.g., backpack),social class and nationality (United States, India, China,and Thailand) could moderate the effect of perceivedproduct quality on the COO-People image. Finally, theresults show that consumer preference to well-knownbrands may affect the COO image (both people andinteraction dimensions) in non-durable products (e.g.,backpack).


Discussion

The findings of this study have important managerialimplications for both international marketers and interna-tional trade policy makers. We conclude that the percep-tion of product quality is a key variable affecting thecountry image. The country’s reputation constitutes anasset which is of great value to all businesses from aparticular country and safeguarding its reputation is aparamount obligation of these businesses. In addition, the

success of Japan and South Korea suggests that countryimage changes over time. The marketing efforts of thebusinesses from a country facilitate this transition. Mar-keters from countries with unfavorable reputation need tobe aware that their marketing activities not only promotetheir products, but also their countries. On the other hand,the country governments could facilitate the businessactivities to enhance its image in the international market.References are available upon request.

For further information contact:Qin Sun

University of North Texas1167 Union Circle, COBA #236

Denton, TX 76203–1396Phone: 940.565.3338



EFFECTS OF CONSUMER ETHNOCENTRISM AND ANIMOSITY ON

FOREIGN PRODUCT EVALUATION AND PURCHASE INTENTIONS:

MODERATING ROLE OF CULTURAL SIMILARITY AND

ECONOMIC COMPETITIVENESS

Jun Ma, IPFW, Fort WayneWei (Andy) Hao, Kent State University, Kent

Laura Lin, Kent State University, Kent

SUMMARY

This study explored the moderating effect of culturalsimilarity and economic competitiveness on the relation-ships between consumer ethnocentrism (CE)/animosityand product judgment/reluctant to buy. The effects ofconsumer ethnocentrism and animosity on product judg-ment and willingness to buy have been explicitly ex-plained by the animosity model proposed by Klein et al.(1998). In their model, CE has a negative impact onproduct judgment and willingness to buy while animosityonly affects willingness to buy. Product judgment also hasa positive impact on willingness to buy. Klein and hiscolleagues collected data from adult consumers in NanJing,China, where the Nanjing Massacre occurred 70 yearsago. The effect of animosity on willingness to buy Japa-nese products is salient. The effects of animosity and CEon product judgment and willingness to buy may be heldin a different context. This study proposed that culturalsimilarity and economic competitiveness moderate therelationship between CE and product judgment/willing-ness to buy and the relationship between animosity andwillingness to buy respectively.

The level of economic competitiveness indicates thecountry’s “ability to manufacture products that require acertain level of skill and technology” (Verlegh andSteenkamp 1999). The level of competitiveness is repre-sented by production technology, product uniqueness,physical capital, managerial skills, physical infrastruc-ture, or human resources (Balabanis and Diamantopoulos2004). When the economic competitiveness of the targetcountry is high, the negative effect of animosity on will-ingness to buy may be weakened.

CE indicates that consumers have the tendency toevaluate domestic product favorably than imported prod-ucts (Shimp and Sharma 1987). CE results an overestima-tion of the overall quality of domestic goods and underes-timation of the quality of imported products (Sharma et al.1995). When cultural similarity is high between two

countries, the consumers in one country may perceiveanother country as in-group and feel closer to that country.As the consequence, it may create favorable attitudes andpurchase intentions toward products imported from cul-turally similar countries (Lantz and Loeb 1996; Sharmaet al. 1995; Watson and Wright 2000). Therefore, theeffect of CE on product judgment and willingness to buymay be weakened when the cultural similarity is high.

We collected data in Taiwan via self-reported onlinesurvey. Respondents were asked to indicate their agree-ments on the seven-point of Likert scales (1= “stronglydisagree” to 7 = “strongly agree) with statements in termsof the following constructs: (1) consumer ethnocentrism,(2) product judgment, reluctance to buy, animosity to-ward Japanese products, (3) product judgment, reluctanceto buy, and animosity toward Chinese products, (4) prod-uct judgment, reluctance to buy, and animosity towardAmerican products. Among the three target countries,China is considered the most similar to Taiwan in terms ofculture followed by Japan and the United States. Accord-ing to the World Economic Forum Rankings, Japan hasthe highest economic competitiveness index comparingto the United States and China (Balabanis andDiamantopoulos 2004).

The results of this study partially support the moder-ating effect of cultural similarity on the relationship be-tween CE and foreign product evaluation and purchaseintentions. The CE does not show a significant effect onproduct judgment and willingness to buy when the targetcountry is China. However, the effects for Japan were notdifferent from that for the United States even thoughJapan was deemed as more culturally similar to Taiwanthan the United States was. The results of this study do notsupport the moderating effect of the economic competi-tiveness on the relationship between animosity and will-ingness to buy. It seems that the animosity is negativelyassociated with willingness to buy foreign products con-sistently.


For further information contact:Wei (Andy) Hao

Marketing DepartmentCollege of Business Administration

Kent State UniversityKent, OH 44242

Phone: 330.672.1268Fax: 330.672.5006



“MADE IN . . .” – DOES IT ALSO APPLY TO SERVICES? AN EMPIRICAL

ASSESSMENT OF THE COUNTRY-OF-ORIGIN EFFECT

IN SERVICE SETTINGS

Dieter Ahlert, University of Muenster, GermanyChristof Backhaus, University of Muenster, Germany

Johannes Berentzen, University of Muenster, GermanyMarkus Blut, University of Muenster, Germany

Manuel Michaelis, University of Muenster, Germany

SUMMARY

The fact that the country of origin (COO) of a producthas an essential influence on consumers’ quality assess-ments has already been proven by Schooler (1965). Sub-sequently, most internationalization literature has exam-ined COO effects with respect to products, and onlylimited research has been conducted on such effects onservices (Ahmed, Johnson, Ling, Fang, and Hui 2002).However, due to the specific characteristics of services(Zeithaml, Parasuraman, and Berry 1985) these effectsmay differ. It is widely acknowledged that these distinc-tive service characteristics affect the perceived risk re-lated to consuming services (Zeithaml, Bitner, and Gremler2006). In the case of a service company entering a foreignmarket, the ability to establish trust, and thus to lowerperceived risk, seems to be crucial, because informationavailable about the novice is less reliable and consumerscannot rely on others’ experiences to any significantdegree (Coulter and Coulter 2003). Therefore, consumersrely on alternative quality indicators such as COO andbrand name to diminish perceived risk. Against this back-ground, the main objective of our study is to analyze (1)whether COO also affects consumer purchase intention ofservices and (2) whether negative effects in the case of anegative COO can be alleviated by providing additionalinformation to the consumer.


ment

In order to analyze the COO effect on purchaseintention, we base our research framework on the prod-uct-related study of Ulgado and Lee (1993). Their resultssupport the assumption that consumers especially useextrinsic information if no intrinsic information about aproduct is available (Olsen and Jacoby 1972). Conse-quently, consumers use the brand name and the COO asextrinsic information for the product assessment if onlythis information is provided. Moreover, previous studiesfound that additional information on product features canlead to changes in consumer preferences (Ueltschy 1998).Based on these results, it can be assumed that the COO will

have a weaker influence on the service assessment whenadditional information is provided. Thus, we hypothesizethat if price, COO, and basic service information is pro-vided to a consumer, the COO has an influence on theassessment of the service (H1). Moreover, if additionalinformation on an offer is provided to the consumer, theimpact of the COO for the assessment of the service willsignificantly diminish (H2).

Methodology

In order to assess the role of the COO effect forservices, four experiments were conducted, using limitconjoint analysis (LCA). As service settings, direct bank-ing and low-cost airlines were selected. For Studies 1and 2 (banking), Switzerland (positive COO), Germany(neutral), and Poland (negative) were used as COO in thedesign. In addition, rate of interest (2, 3, and 4%) and theperiod of investment (1, 2, and 3 years) were taken intoconsideration as significant service features. In the secondstudy, the indication of an “A-Ranking” by an indepen-dent agency was given as an additional quality cue. ForStudies 3 and 4 (low-cost airlines), Germany (positiveCOO), Spain (neutral) and Czech Republic (negative)were used as COO in the design. In addition, price (50, 75,and 100 • ) and the distance from the airport to the place ofresidence (30, 60, and 90 km) were included in theconjoint design. The age of the airplane (2 years) and theservice interval were used as additional information inStudy 4. We collected data from 208 students to controlthe sample’s homogeneity. The average age was 21.3years.


In Study 1 (banking without additional information),relative importance of the COO was 31.14 percent (inter-est rate 58.58%; period of investment 10.28%). In Study 2,COO had an impact of only 28.18 percent (rate of interest54.63%; period of investment 17.19%). Similar to theresults of Studies 1 and 2, in Study 3 the price was foundto be the major factor of influence on the choice of airlines(46.14%), followed by distance to airport (28.36%). In


contrast to the banking setting, COO achieved only 25.5percent. Study 4 showed a stronger influence of COOproviding additional information (27.98%), more than theinfluence of distance (23.84%). Again, price had thehighest relative importance, with 48.18 percent.

The results of Study 4 contradict the assumption thatthe COO significantly loses its importance for the assess-ment of service when additional information is provided.In conclusion, H1 could be confirmed in all experiments,indicating that “made in . . .” plays a significant role fordecision making also in a services context. H2 can only bepartially confirmed, i.e., is confirmed only in the case ofdirect banks. In case of low-cost airlines additional infor-mation cannot eliminate the COO effect. This could be

explained by the higher level of perceived risk (i.e., risk tolife and health in the worst case) which is associated withthe use of airlines in comparison to direct banking ser-vices. These results show that COO should also be takeninto account as information that influences the consumerdecision-making process in service settings. Thus, servicecompanies can distinguish themselves from their com-petitors in international markets by indicating a positivelyperceived COO. References are available upon request.

ENDNOTE

The authors greatly acknowledge the financial support ofthe German Ministry for Education and Research(BMBF – FKZ 01 HQ 0523).

For further information contact:Johannes Berentzen

Marketing Center MuensterUniversity of MuensterAm Stadtgraben 13–15

Muenster, 48143Germany

Phone: +49.251.83.22050Fax: +49.251.22032



ATMOSPHERE MEASUREMENT AND MYTHS

Morten Heide, University of Stavager, NorwayKjell Gronhaug, Norwegian School of Economics and Business Administration, Norway

ABSTRACT

Atmosphere is generally viewed as a key factor toattract and satisfy customers. However, the phenomenonis considered to be vague and difficult to grasp, whichrepresents a substantial challenge with regards to mea-surement, management and improvement. This paperreports an effort to capture the phenomenon of atmo-sphere as applied in the hospitality industry. The work wasdone in two phases; an inductive (collecting actual de-scriptions of the phenomenon) and deductive (refiningthese into a set of empirical indicators, first by usingjudgment-based criteria and secondly through empiricaltesting).

A field survey of hotel guests was conducted to testthe applicability of these indicators, which revealed arobust and interpretable four factor structure and satisfac-tory model fit. The predictive power of the indicators wasalso demonstrated. The clear and meaningful results sug-gest that atmosphere is most likely not as vague anddifficult to measure as commonly believed. The data alsopointed to moderating effects of consumer demographicson atmosphere perception (for example female gueststend to rate the hotels as more atmospheric than is the casefor male guests. However, the difference becomes non-significant as age increases). Theoretical and managerialimplications are highlighted.

Keywords: Atmosphere, common assumptions, per-ception, measurements, stereotypes.

INTRODUCTION

The concept of atmosphere is important both ineveryday life and business. Atmosphere is appreciatedand consciously sought. For example, consumers oftenmake an effort to create atmosphere in their homes. Theyuse time, energy and economic resources to do so, involv-ing costly activities such as decoration, purchase of art andso on. Also, in business, atmosphere is considered vital.For example, in the hospitality sector atmosphere is re-garded as an essential – if not the key factor – to attract andsatisfy guests.

To improve the atmosphere of their establishments,firms frequently make costly – and risky investments.Such activities are difficult to assess in particular sobecause the phenomenon of atmosphere is generally con-sidered as vague and difficult to grasp, which for instancemight complicate investment decisions for managers.

Despite several attempts to define the concept (see forexample, Kotler (1973) who described atmosphere as “theair surrounding a sphere”) the phenomenon apparentlyremains ambiguous.

A review of the literature on atmosphere reveals anemphasis on elements in the environment (e.g., Bitner1992; Mehrabian and Russell 1974). Furthermore, theliterature focuses on how atmosphere is perceived byindividuals (e.g., Baraban and Durocher 2001), and howit may evoke internal responses (e.g., Lin 2004). Severalstudies have examined atmospherics (background fea-tures such as temperature, scent, music, and lighting),social factors, and design factors as drivers of atmosphere(e.g., Baker, Levy, and Grewal 1992; Baraban andDurocher 2001; Bitner 1992; Chebat and Michon 2003;Donovan and Rossiter 1982; Gardner and Siomkos 1986;Martin 1986; Mattila and Wirtz 2001; McElroy, Morrow,and Eroglu 1990; Wilson 2003).

Lin (2004) argues that in order to create the image ofa pleasant physical environment in the mind of customers,it is important to understand customers’ overall evalua-tion process both theoretically and practically. Kotler(1973) also emphasizes that one’s reactions to atmo-spheric stimuli are partly learned and may be influencedby cultural background and nationality. Consequently, itis reasonable to assume that the more dissimilar thepatrons of a particular establishment are, the more variedtheir perceptions of the atmosphere will be.

The purpose of this paper is to examine the role ofatmosphere in relation to the common assumptions dis-cussed above, i.e., (1) whether the phenomenon of atmo-sphere really is vague and difficult to grasp and (2)whether perceptions of the phenomenon of atmospherevary across individuals.

Earlier research provides few directions on how todefine atmosphere conceptually, which is usually consid-ered a prerequisite for developing adequate measure-ments. To overcome this challenge, we found it necessaryto capture how the concept has been used, and what itsubsumes, in the actual context. Consequently, we de-parted by collecting descriptions of how atmosphere ischaracterized or described in the hospitality industry (ormore precisely in relation to hotels). Through such de-scriptions, we can get an understanding of how the con-cept of atmosphere is used in the studied context and whatit means in this setting (Rosch 1976). Based on thecollected descriptors, which were refined into a set of


empirical indicators, we were able to capture importantdimensions of atmosphere, which also implies that atmo-sphere is a multidimensional concept. We chose thehospitality industry as our empirical context, because ofits clear emphasis on atmosphere for creating a successfulconsumer experience, which has been highlighted by bothhospitality practitioners and researchers.

The remaining part of this paper is organized asfollows: In the next section we describe the process ofdeveloping the atmosphere measurement indicators, fol-lowed by a methodological description of an exploratoryfield survey where these indicators were tested. Afterpresenting and discussing the empirical findings, weconclude by highlighting managerial and research impli-cations and proposing avenues for further research.

METHODOLOGY

Because of the lack of a priori knowledge of whatatmosphere actually is and how to capture the phenom-enon empirically, i.e., what the atmosphere concept en-tails, we proceeded in two phases, an inductive and adeductive phase. In the inductive phase, we examinedhow the concept has been applied in the hospitalitycontext and through a comprehensive search collectedterms that have been used to describe atmosphere as itpertains to hotels. Consequently, the output of the induc-tive phase was an inventory of descriptors. In the subse-quent deductive phase, we tested the relevance of eachdescriptor and reduced the number in order to remain withthe most fundamental ones. The two phases are describedbelow.

Inductive Phase – Collection of Atmosphere Descrip-

tors

To ensure that we included all relevant items foratmosphere as pertaining to our empirical context, it wasdeemed necessary to first collect words that are applicablefor describing atmosphere in hotel settings. Consequently,we started by generating a pool of items, which is standardprocedure to cover the conceptual domain, in line withrecommendations from Churchill (1979).

Several approaches for generating the item pool wereconsidered possible, and it will always be open to debatewhich would be the best. The philosophy we followed wasas follows: If an adjective is relevant for describingatmosphere in hotels, it has most likely already been usedfor this purpose. Consequently, we decided to collect allwords that have been used for describing atmosphere inhotel settings. Thus, a search for the universe of relevantdescriptors was launched. To avoid confusion from mix-ing different languages, it was decided to limit the searchto descriptions in English. An internet search on “hotel”and “atmosphere” in combination yielded several million

hits and a search routine was employed for identifyingadjectives used for describing atmosphere. Every time anew adjective for describing hotel atmosphere was en-countered, this word was registered in an inventory ofhotel atmosphere descriptors. Alternative search engineswere used to expand the search, as well as synonyms(ambience, charm, appeal, and resorts, motels, inns). Tosupplement web-based sources, we also scrutinized alarge number of hotel design books, architectural booksand travel magazines. All new descriptors were recordedregardless if the source was hotel managers, travel re-views, guest experiences or expert consultants/architects.When no more new descriptors were found, we concludedthat the search was exhaustive. At this stage the inventoryhad 600 unique descriptors, which we consider as theuniverse of hotel atmosphere descriptors.

Deductive Phase – Reduction from 600 to 43 Atmo-

sphere Descriptors

In the deductive phase, we assessed and tested therelevance of each descriptor and reduced the number inorder to remain with the most fundamental ones (seeAppendix A for details). The 600 descriptors were firstreduced to 458 by employing judgment-based criteria.Two external experts in hospitality design and manage-ment (an architect and a hospitality manager) were con-sulted as part of this process. Thereafter the 458 descrip-tors were empirically tested using a three-stage random-ized experimental design, where subjects were shownpresentations of a range of hotels and thereafter rated howrelevant each descriptor was for describing the hotel’satmosphere on a seven-point Likert scale. In the firststage, the number of descriptors was reduced from 458 to201. The second stage yielded a further reduction to 135descriptors. These descriptors were tested in the thirdstage, which gave the 43 descriptors that were subse-quently tested in the exploratory field survey.

Exploratory Field Survey

To further test the applicability of the reduced set ofindicators it was deemed necessary to collect data fromactual hotel guests. Since this was the first attempt toempirically test the indicators in the field, a single countryenvironment was chosen. Consequently, an exploratoryfield survey was conducted in collaboration with theleading hotel chain in Norway. Six of the chain’s hotels(see Appendix B for particulars) were selected based on apriori knowledge about the hotel’s atmosphere. The mainpurpose of this selection was to ensure variation in per-ceived atmosphere.

Procedure. The survey was conducted during a two-month period. During check-in guests were asked if theywere willing to participate in a guest survey and if so theiremail addresses were collected. After the guests had


checked-out, they received an email invitation with linksto the electronic guest survey questionnaire (Norwegianor English according to respondent’s choice. Translationwas done with utmost care by using available languageexpertise/dictionary sources. The descriptors loaded onthe same four factors both for Norwegian and foreignguests, which suggest that semantic content was notaltered in the translation process). An electronic systemwas used for data-recording and handling. The hotelchain’s standard incentive for customer surveys (draw fora free weekend for two persons) was used.

Respondents. The email invitation was sent to a totalof 559 individuals of which 369 responded (i.e., a re-sponse rate of 66 percent, which we consider to besatisfactory). Experts in the hotel chain verified that thecomposition of the sample (gender, nationality, age, guesttype) corresponded to the regular guest population of theparticular hotels. Forty percent of the respondents were ona business trip, 25 percent attended a course/conference,27 percent visited in connection with leisure/tourism and8 percent for other reasons. Nearly two-thirds (63%) wereemployed in the private sector, 25 percent in the publicsector, while the remainder reported other type of em-ployer (8%) or did not work (4 percent). The median agegroup was 40–49. The majority (77%) of the respondentswere Norwegians.

Measurements. In addition to characteristics aboutthe respondent and ratings of various aspects of the hotel,the questionnaire included the 43 descriptors (in random-ized order). The respondents were asked to indicate howrelevant each of these was for describing the atmospherein the hotel as experienced during their stay (on seven-point scale as earlier). The respondents were also asked toindicate how much atmosphere the hotel had overall (i.e.,overall atmosphere) and compared to other hotels in thesame price range. Respondents were also given the oppor-tunity to provide comments regarding the hotel’s atmo-sphere.

FINDINGS

Common Assumption 1: The Phenomenon of Atmo-

sphere is Vague and Difficult to Grasp

Exploratory Factor Analysis. To examine this as-sumption, we started out with exploratory factor analysesof the 43 descriptors. Table 1 reports the output from thefour-factor solution chosen. Principal component extrac-tion and varimax rotation was used. In order to choose thecut-off point, we inspected the scree-plot, which showedthat a substantial dip followed after the fourth factor. Thefirst four factors were easy to interpret, had eigenvaluesconsiderably greater than one and explained most of thevariance in the 43 descriptors (adding additional factorsonly increased explained variance marginally). Also, other

rotation (direct oblimin) and extraction (principal axisfactoring and maximum-likelihood) methods were em-ployed. However, the four-factor structure was robustregardless of extraction and rotation method. This factorsolution was also the most stable and robust when testedin subsample factor analyses (split half samples, males vs.females, younger vs. older respondents, etc.).

If the assumption that atmosphere is vague and diffi-cult to grasp were valid, no clear and robust factor struc-ture should be expected. However, the clear and stablefactors identified in our analyses suggest that we havecaptured important dimensions of atmosphere and fur-thermore imply that atmosphere is a multidimensionalconcept.

We chose to interpret the first factor in Table 1 asdistinctiveness (high loadings on descriptors like special,fascinating, and different). In our opinion, this factorimplies that a hotel will only be rated high on atmosphereif it has features that differentiate it from other establish-ments. Thus, truly atmospheric hotels have a certaindistinctiveness that separates them from the herd. Thesecond factor may be labeled hospitability (loaded mainlyon descriptors like welcoming, hospitable, and profes-sional). Related to what was mentioned earlier aboutdrivers of atmosphere, it is reasonable to assume that thisfactor is mainly driven by the social factors (particularlythe staff-guest interactions). The third factor has powerfulloadings on descriptors such as pastoral, resort and holi-day and may consequently be interpreted as relaxation.The fourth factor loaded on descriptors like classical,traditional, and upper-class and was thus labeled refine-ment.

Confirmatory Factor Analysis. The factor structurewas further tested by confirmatory factor analysis usingLISREL (Linear Structural Relations, Jöreskog andSörbom 2004). Again, if the above-mentioned assump-tion were correct, we would not expect to get acceptablefit. In Figure 1, we have presented two LISREL measure-ment models (standardized solutions).

Our point of departure is the model to the left. Thismodel has the same four factors as shown in the explor-atory factor analysis, each with three indicators (descrip-tors). Model fit is fairly satisfactory (RMSEA = 0.055).Additional analyses showed that RMSEA (but not the chisquare statistic) improved slightly (to 0.049) by splittingfactor four into its two facets as shown on the right side ofFigure 1. It should be noted that while these confirmatoryfactor analyses suggest that satisfactory fit can be ob-tained, there is a clear need to repeat the analyses with newdata. We will return to this issue in the discussion section.

Prediction. If atmosphere is such a vague and unclearphenomenon, it could be expected that the indicators used


TABLE 1

Factor Analysis – Four-Factor Solution1

Descriptors h2 Distinctiveness Hospitability Relaxation Refinement

Special 0.81 0.82 0.30 0.15 0.14Fascinating 0.75 0.79 0.29 0.13 0.13Different 0.69 0.79 0.15 0.20 -0.01Unusual 0.73 0.78 0.19 0.27 -0.06One-of-a kind 0.74 0.74 0.38 0.14 0.14Memorable 0.69 0.73 0.29 0.20 0.20Rarefied 0.69 0.72 0.31 0.25 0.09Peculiar 0.59 0.71 0.17 0.12 -0.21Seductive 0.74 0.70 0.39 0.29 0.13Characteristic 0.66 0.70 0.36 0.06 0.18Attractive 0.71 0.70 0.45 0.01 0.13Dream 0.69 0.69 0.30 0.29 0.21Charming 0.69 0.68 0.35 0.21 0.25Welcoming 0.81 0.12 0.87 0.08 0.15Hospitable 0.73 0.20 0.82 0.16 -0.01Professional 0.75 0.27 0.81 0.03 0.13Genial 0.73 0.23 0.81 0.16 0.05Communicating 0.69 0.14 0.79 0.21 0.05Effective 0.65 0.20 0.75 0.09 0.19Civilized 0.60 0.32 0.68 -0.02 0.19True 0.69 0.42 0.68 0.19 0.13Feel-good 0.70 0.47 0.68 0.12 0.06Flexible 0.57 0.25 0.64 0.19 0.24Serious 0.54 0.36 0.64 0.01 0.03Well-being 0.68 0.43 0.63 0.27 0.15Pastoral 0.65 0.19 -0.03 0.78 0.01Resort 0.67 0.38 0.13 0.71 0.00Holiday 0.63 0.45 0.21 0.61 -0.05Family-like 0.56 0.37 0.19 0.60 -0.13Serene 0.69 0.48 0.31 0.60 0.07Quiet 0.52 0.39 0.18 0.58 0.07Easy 0.48 0.13 0.32 0.53 0.27Suburban 0.30 -0.16 0.03 0.51 0.12Simple 0.35 -0.34 0.00 0.48 -0.02Classical 0.73 0.18 0.28 0.05 0.78Traditional 0.64 -0.12 0.15 0.21 0.75Historical 0.64 0.49 0.06 -0.09 0.62Upper-class 0.70 0.54 0.18 -0.08 0.61Rich 0.67 0.59 0.29 0.04 0.48Luxurious 0.66 0.61 0.31 0.03 0.44Modern 0.42 0.26 0.33 0.40 -0.3Tranquil 0.55 0.41 0.47 0.39 0.11Great location 0.34 0.42 0.27 -0.22 0.20

Eigenvalues 18.94 3.40 3.10 2.07Percentage of variance 44.03 7.90 7.20 4.82Cumulative percentage 44.03 51.93 59.13 63.95

in the survey would have limited predictive power. To goone step further, i.e., move from measurement to predic-tion, we tested the predictive power of the atmosphere

descriptors both in relation to a criterion variable (overallatmosphere) and their ability to differentiate betweentarget objects (i.e., the six hotels).


First, regression analysis was employed with overallatmosphere of the hotel as dependent variable and thevarious descriptors as explanatory variables. For all sixhotels combined, there were a total of eight significantexplanatory descriptors (p < 0.05) and combined thesevariables explained 70 percent of the variation in per-ceived overall atmosphere (adjusted R2 = 0.70). Whenseparate regression analyses for each hotel were per-formed, even higher levels of explained variance wereobtained (maximum explained variance was for Hotel #2,adjusted R2 = 0.87). In our opinion, these high levels ofexplained variance indicate that the descriptors havepredictive power in relation to the criterion variable.

Second, a fundamental property of any measurementinstrument is the ability to discriminate between differentobjects that it is intended to measure. To test the ability ofthe descriptors to differentiate between target objects, weused discriminant analysis on the respondents’ ratings onthe atmosphere descriptors to predict what hotel they hadstayed in. The random probability of predicting hotelcorrectly is 16.7 percent (i.e., 1/6). By including theresponses on the atmosphere descriptors the probabilityincreased substantially, i.e., 87.4 percent of the originalgrouped cases were correctly classified. Considering that

a certain proportion of respondents in surveys like thesemost likely are characterized by a careless response style,we regard 87.4 percent as a clear indication of predictivepower and consequently usefulness for measurement pur-poses. Thus, it seems fair to conclude that the indicatorshave predictive power in terms of differentiating betweentarget objects.

Common Assumption 2: There are Significant Differ-

ences in the Way that People Perceive Atmospheric

Stimuli

The clear differences between hotels in terms ofperceived atmosphere and the robust factor structurefound in the sub-sample factor analyses (males vs. fe-males, younger vs. older respondents, etc.) suggest thatpeople’s assessments of atmosphere, at least in hotelsettings, are manifested in dimensions that are sufficientlystrong to hold across individual differences. To furthertest the assumption, we investigated the extent to whichguest characteristics explained variation in perceivedatmosphere. ANOVA analyses were performed to deter-mine whether such characteristics represented sources ofsystematic variation. The following characteristics wereincluded in the analyses: Gender, age (group) and guest

FIGURE 1

LISREL Measurement Models3,4


type (business, course/conference or leisure). Becausesome hotels had very few foreign guests, it was notpossible to assess the impact of nationality across hotels.In Table 2, we have presented a summary table of theANOVA analyses.

Table 2 clearly shows that most of the variation inperceived atmosphere (overall atmosphere and factorscores based on the measurement model) is driven by thehotels. However, there are some moderating effects ofguest characteristics related to gender, age and guest type,which we will address in the discussion section.

DISCUSSION

In our opinion, the analyses point toward a robust andinterpretable factor structure, acceptable model fit, andfurthermore demonstrate the predictive power of theindicators both in relation to explaining variance in thecriterion variable (overall atmosphere) and ability to dif-ferentiate between target objects, i.e., predicting which ofthe hotels the guest had stayed in. Our assessment is thatthese results challenge the common assumption that at-mosphere is a vague and unclear phenomenon which isdifficult to grasp and measure.

If we are correct, it should be possible to use theatmosphere descriptors to construct a diagnostic tool formapping consumer perceptions and thereby identify dif-ferences. We have tested this by producing atmosphereprofiles for the six hotels, which are displayed in Appen-dix B. The atmosphere profiles were produced by plottingthe mean factor scores for each of the six hotels. Theseprofiles give a visual picture of the type of atmosphere thatcharacterizes the hotel (because factor scores are stan-dardized, the means for all hotels are zero for all factors.Consequently, scores higher than zero signify a higherthan average score on the respective factor, and scores lessthan zero the opposite).

A visual inspection of the profiles reveals substantialdifferences between the hotels in terms of atmosphere(there are highly significant differences between the ho-tels regarding factor scores, particularly for each of thehotels’ striking factors. Actual output from the analyses isnot shown here because of space considerations). Inaddition to being clear, the differences between the hotelprofiles are meaningful. For example, the atmosphereprofile shows that Hotel #1 was perceived to have anexclusive and traditional atmosphere. Hotel #4, which isa simpler and more basic establishment located in thesame city, is also considered as traditional. However, theatmosphere is perceived as considerably less exclusiveand scores low on distinctiveness.

As discussed, variation in atmosphere perceptionwas mainly driven by differences between hotels, not

guests, i.e., the main features of the perceived atmospherewere the same regardless of gender, age, and guest type.These findings indicate that perception of atmosphere, atleast in hotel settings, are manifested in assessments thatare sufficiently strong to hold across individual differ-ences. Nevertheless, Table 2 also showed some moderat-ing effects of guest characteristics related to gender, age,and guest type, including a significant gender/age effect inthe guests’ ratings of overall atmosphere, which wasfurther examined.

In the upper part of Figure 2, we have presentedstandardized mean ratings of overall atmosphere accord-ing to gender and age of the guests (because of thesignificant differences in overall atmosphere between thesix hotels, it was necessary to standardize scores for eachhotel. This implies that each hotel will have a mean scoreof zero and standard deviation of one, which means that itwill be possible to compare across hotels, if for examplefemale or older guests tend to rate the atmosphere higherthan male or younger guests).

We notice that across hotels, female guests tend torate the hotels as more atmospheric than is the case formale guests. Furthermore, as guests get older, they tend torate hotels higher in atmosphere. What is interesting is thatwhile the increase is nearly linear for both genders, theincrease is sharper for the males. Consequently, while thedifference between males and females is highly signifi-cant (p < 0.01) in the below 40 group, it becomes non-significant as age increases. One possible explanationmight be that females in general have a more developedability to absorb and appreciate the atmosphere of hotels.However, as age increases, males too might learn torecognize the value of atmosphere and devote the neces-sary time and attention to experience it. However, itshould be noted that this is a tentative explanation and thatthe present data set does not permit a comprehensiveinquiry into alternative interpretations.

There was also a significant age effect when it comesto the perception of hospitability. The lower part ofFigure 2 shows the standardized mean ratings regardinghospitability according to age. We note that across hotels,young guests rate the atmosphere as less hospitable thanolder guests do. The difference between the under 30guests and those over 60 is highly significant (p < 0.01).In our view, this is not necessarily a matter of perceptualdifference. It might actually be that senior guests aretreated in a more hospitable manner, and thus subjected toa different atmosphere than is the case for younger guests.

Some differences were also found between guesttypes. Business travelers rated the hotels significantlylower with regards to overall atmosphere, relaxation andexclusiveness compared to the course/conference partici-pants and leisure travelers. For the other factors, there


TABLE 2

Factor Guest Characteristics2 as Sources of Systematic Variation – Summary

Table of the ANOVA Analyses

Perceived atmosphere Source of Variation df Mean Square F p

Overall atmosphere Hotel 5 4.189 4.119 0.001

Gender 1 2.016 1.982 0.160Age 3 1.387 1.364 0.254Guest type 2 0.926 0.911 0.403Hotel x Gender 5 1.206 1.186 0.316Hotel x Age 15 2.040 2.006 0.016

Hotel x Guest type 10 0.864 0.849 0.581Gender x Age 3 3.045 2.994 0.032

Gender x Guest type 2 0.578 0.568 0.567Age x Guest type 6 2.115 2.080 0.056

Distinct atmosphere Hotel 5 8.624 10.952 0.000

Gender 1 0.177 0.225 0.636Age 3 1.091 1.386 0.248Guest type 2 1.662 2.111 0.123No significantInteraction effects

Hospitable atmosphere Hotel 5 1.342 1.456 0.205Gender 1 1.101 1.195 0.275Age 3 2.487 2.698 0.046

Guest type 2 0.054 0.059 0.943Only significantinteraction effect:Age x Guest type 6 2.933 3.182 0.005

Relaxing atmosphere Hotel 5 7.224 13.184 0.000

Gender 1 0.039 0.071 0.790Age 3 0.696 1.270 0.285Guest type 2 1.857 3.389 0.035

No significantInteraction effects

Exclusive atmosphere Hotel 5 9.288 14.379 0.000

Gender 1 0.563 0.872 0.351Age 3 1.708 2.644 0.050

Guest type 2 1.985 3.073 0.048

No significantInteraction effects

Traditional atmosphere Hotel 5 3.839 4.689 0.000

Gender 1 1.242 1.517 0.219Age 3 0.713 0.871 0.457Guest type 2 0.596 0.728 0.484No significantInteraction effects

were no significant differences. Furthermore, we foundno significant differences between course/conferenceguests and leisure travelers. Our interpretation of this

finding is that available time is more important thanwhether the visit is in connection with work or leisure.Course and conference guests, although they are working,


FIGURE 2

Differences in the Way People Perceive Atmosphere

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

Under 30 30-39 40-49 50-59 60 or o lder

Age Group

Ov

era

ll A

tmo

sp

he

re (

Sta

nd

ard

ize

d M

ea

n R

ati

Female

Male

b) Perception of Hospitability – Effect of Age

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

Under 30 30-39 40-49 50-59 60 or older

Age Group

Ho

sp

ita

ble

Atm

os

ph

ere

(S

tan

da

rd.

Me

an

Ra

ta) Perception of Overall Atmosphere – All Six Hotels by Gender and Age of Guests


generally have more time to experience the atmosphere,than is the case for regular business travelers.

Implications, Limitations, and Further Research

Adequate measurement is a prerequisite for manage-ment. The research presented here represents a step to-ward developing an instrument for measuring atmosphereperceptions. Of course such an instrument needs to betested, further refined and validated before it should enterthe tool box of market research practitioners. As ex-plained earlier, our selected approach for generating theitem pool was to collect all words that have been used fordescribing atmosphere in hotel setting regardless of source.This approach should be supplemented by techniques totap more directly into customers’ perception of atmo-sphere as a concept, through, e.g., a focus group approachor the repertory grid technique (Kelly 1955). As men-tioned, the confirmatory factor analyses presented in thispaper should be seen as a first step. Consequently, to bereally convincing these analyses need to be done with newdata. In this regard, we are planning a follow-up survey tocollect data from a new sample of guests.

Even though our approach could undoubtedly benefitfrom being supplemented by qualitative techniques, webelieve that a quantitative measurement instrument couldbe a useful diagnostic tool for market research and man-agement. Given the considerable interest, and conse-quently the presumed value of atmosphere as an intan-gible asset, there are a number of areas where such aninstrument would be useful. For example, the instrumentcould be employed to assess the extent to which theatmosphere of a particular establishment differs from thatof its competitors and whether the atmosphere offeredsatisfies the demand of the market segment aimed at. Also,the measurement instrument could be used to explore therelationship between atmosphere and key variables likecustomer loyalty, word-of-mouth and repeat visits tomention a few.

Managers often have predefined goals to follow. Onesuch goal might be to improve the atmosphere. To estab-lish and create a desired atmosphere is often left to designexperts (e.g., architects and interior designers) or handledby the owner’s gut feeling/intuition. Being involved inatmosphere measurement can help the manager to im-prove his or her professional judgment and consequentlyreduce the risk of bad investments based on feelings morethan facts. The measurement instrument could also be

useful in identifying gaps (e.g., areas where staff mem-bers’ perception of the atmosphere differs from that of theguests), as well as for measuring the effect of variousinterventions (e.g., pre- and post-measurements in con-nection with investments to improve the atmosphere ortraining programs to enhance hospitability). Finally, theinstrument could be used for testing how effectivelydifferent promotional material can communicate the sa-lient aspects of the establishment’s atmosphere to poten-tial consumers.

Despite considerable interest in the phenomenon ofatmosphere primarily among practitioners, there is still away to go in order to address atmosphere in a systematicway from the perspective of consumer research. Instead oftreating atmosphere as something vague and difficult tomeasure, efforts should be directed toward examiningatmosphere perception more systematically. In our opin-ion, further work should be done to classify the wide rangeof atmospheres that are relevant for various types ofconsumer settings. An avenue which might be useful inthis respect comes from the science of viticulture andenology, where the variety and complexity of wine aromahave led to the development of the aroma wheel, whichhas become a useful tool for wine description and classi-fication (Noble et al. 1987). A similar approach might beuseful for research on atmosphere.

Further studies should also address more specificallythe combined effect of drivers to supplement the mainlymonocausal studies (i.e., investigation into the effects ofa single factor) that have dominated research so far. Thus,studies should be undertaken where different types ofdrivers (atmospheric, social, and design factors) are in-cluded and their effects measured both individually andcombined.

It will be important to gain more insight into therelationship between atmosphere and affective responses,and between affective responses and outcome variables.Building on these insights it should be possible to furtherexamine the relationship between atmosphere and opera-tional variables (e.g., occupancy rates and repeat visits) aswell as financial implications. Atmosphere investmentsare by no means risk free and to further examine risks inrelation to atmosphere management we call for studiesthat assess the relationships between (i) atmosphere andfunctionality, (ii) various types of atmosphere and possi-bilities for rationalization/automation and scale expan-sion and (iii) investments in atmosphere and profitability.

ENDNOTES

|1 The choice of varimax rotation and order of presentation

of the descriptors in Table 1 have been done for the

purpose of clarity and easy reading. In the question-naire the descriptors were presented in random order.

2 Categories: Hotel (six hotels), gender (male/female),age (four age groups), guest type (business, course/


conference or leisure).3 RMSEA is the root-mean-square error of approximation

fit index. Our assessment of fit is based on therecommendations of Browne and Cudeck (1993) andMacCallum, Browne, and Sugawara (1996), whichcharacterize RMSEA < 0.05 as close fit, 0.05–0.08 asfair fit, 0.08–0.1 as mediocre fit and > 0.1 as poor fit.

4 Facet identification was based on recommendationsfrom earlier studies (see e.g., Aaker 1997). To iden-tify facets, the set of descriptors in each of the four

factors (which resulted from the principal compo-nents analysis as shown in Table 1) was factor-analyzed individually. For factor four, two sub-fac-tors/facets had eigenvalues greater than one (defaultcriterion for extraction). The first could be inter-preted as exclusive and the second as traditionalatmosphere. As shown in Figure 1, RMSEA im-proved slightly (from 0.055 to 0.049) when the twofacets were included as separate latent variables.

REFERENCES

Aaker, Jennifer L. (1997), “Dimensions of Brand Person-ality,” Journal of Marketing Research, 34, 347–35.

Baker, Julie, Michael Levy, and Dhruv Grewal (1992),“An Experimental Approach to Making Retail StoreEnvironmental Decisions,” Journal of Retailing, 68,445–60.

Baraban, Regina S. and Joseph F. Durocher (2001), Suc-cessful Restaurant Design, 2nd ed. New York: Wiley.

Bitner, Mary J. (1992), “Servicescapes: The Impact ofPhysical Surroundings on Customers and Employ-ees,” Journal of Marketing, 56, 57–71.

Browne, Michael W. and Robert Cudeck (1993), “Alter-native Ways of Assessing Model Fit,” in TestingStructural Equation Models, Kenneth A. Bollen andJ. Scott Long, eds. Newbury Park, CA: Sage, 136–62.

Chebat, Jean C. and Richard Michon (2003), “Impact ofAmbient Odors on Mall Shoppers’ Emotions, Cogni-tion, and Spending – A Test of Competitive CausalTheories,” Journal of Business Research, 56, 529–539.

Churchill, Jr., Gilbert A. (1979), “A Paradigm for Devel-oping Better Measures of Marketing Constructs,”Journal of Marketing Research, 16, 64–73.

Donovan, Robert J. and John R. Rossiter (1982), “StoreAtmosphere: An Environmental Psychology Ap-proach,” Journal of Retailing, 58, 34–57.

Gardner, Meryl P. and George J. Siomkos (1986), “To-ward a Methodology for Assessing Effects of In-Store Atmospherics,” in Advances in Consumer Re-search, R.J. Lutz, ed. Ann Arbor, MI: Association forConsumer Research, 13, 27–31.

Jöreskog, Karl G. and Dag Sörbom (2004), LISREL 8.7Computer Software. Lincolnwood, IL: Scientific Soft-ware International.

Kelly, George A. (1955), The Psychology of PersonalConstructs. New York: Norton.

Kotler, Philip (1973), “Atmospherics as a MarketingTool,” Journal of Marketing, 49, 48–64.

Lin, Ingrid Y. (2004), “Evaluating a Servicescape: TheEffect of Cognition and Emotion,” International Jour-nal of Hospitality Management, 23, 163–78.

MacCallum, Robert C., Michael W. Browne, and HazukiM. Sugawara (1996), “Power Analysis and Determi-nation of Sample Size for Covariance Structure Mod-eling,” Psychological Methods, 1, 130–49.

Martin, William B. (1986), Quality Service: The Restau-rant Manager’s Bible. Ithaca, NY: School of HotelAdministration, Cornell University.

Mattila, Anna S. and Jochen Wirtz (2001), “Congruencyof Scent and Music as a Driver of In-Store Evalua-tions and Behavior,” Journal of Retailing, 77, 273–89.

McElroy, James C., Paula C. Morrow, and Sevo Eroglu(1990), “The Atmospherics of Personal Selling,”Journal of Personal Selling and Sales Management,10, 31–41.

Mehrabian, Albert and James A. Russell (1974), AnApproach to Environmental Psychology. Cambridge,MA: Massachusetts Institute of Technology.

Noble, A.C., R.A. Arnold, J. Buechsenstein, E.J. Leach,J.O. Schmidt, and P.M. Stern, (1987), “Modificationof a Standardized System of Wine Aroma Terminol-ogy,” American Journal of Enology and Viticulture,38, 143–46.

Rosch, Eleanor, Carolyn B. Mervis, Wayne D. Gray,David M. Johnson, and Penny Boyes-Braem (1976),“Basic Objects in Natural Categories,” CognitivePsychology, 8, 382–439.

Schall, Matthew (2003), “Best Practices in the Assess-ment of Hotel-guest Attitudes,” Cornell Hotel andRestaurant Administration Quarterly, 44, 51–65.

Wilson, Stephanie (2003), “The Effect of Music on Per-ceived Atmosphere and Purchase Intentions in aRestaurant,” Psychology of Music, 31, 93–112.


APPENDIX A

Detailed Description of the Deductive Phase of the Study

As mentioned, the purpose of the deductive phase was to test the relevance of each descriptor and reduce the numberin order to remain with the most fundamental ones.

Judgment-based reduction (from 600 to 458 atmosphere descriptors)

Each descriptor was critically assessed in relation to the following criteria: (a) Combined descriptors: Terms thatcombined several descriptors were simplified. For example the term “good old-fashioned atmosphere” was reduced to“old-fashioned atmosphere.” (b) Location specific descriptors: Terms that were only relevant for specific places werenot considered suitable to identify general dimension and were consequently deleted. For example, the term “Sorrentoatmosphere” has only relevance for the city of Sorrento in Italy and was therefore excluded. (c) Context specificdescriptors: Terms that were strictly context specific were deleted, as these were not considered particularly relevant forgeneral atmosphere dimensions. Examples include “Sultan Ahmed atmosphere.” (d) Vapid descriptors: Adjectives thatgave little information about the quality or type of atmosphere (for example “fine atmosphere”) were deleted. (e).Obscure descriptors: The same applied to descriptors with unclear meaning (such as “principal atmosphere”).

Empirical based reduction (from 458 to 43 atmosphere descriptors)

Due to the labor-intensive and meticulous task of testing the 458 descriptors, we were advised by hospitality experts toavoid using actual hotel guests in this early part of the work and we thus decided to use student subjects instead. A threestage randomized experimental design was employed.

First stage (reduction from 458 to 201 descriptors)

Subjects (n = 78, 80% female, mean age = 23.5 undergraduate students, specializing in hotel/tourism management at aleading university in Norway) were shown a ten-minute presentation of a hotel randomly selected from a total of sixpresentations. The hotels ranged from companies low in atmosphere (such as a Motel 6 establishment in Idaho) to trulyatmospheric establishments like a tranquil countryside hotel in England and the exclusive Burj Al Arab Hotel in Dubai.The presentation was followed by a questionnaire with questions about the subjects (age, gender, etc.) and assessmentsof the hotel, including the extent to which the subjects believed that the hotel had a characteristic atmosphere. The lastpart of the questionnaire consisted of a randomly selected one-fourth of the 458 descriptors (i.e., four versions of thequestionnaire were used. This decision was based on a pretest of respondent burden). The subjects were asked to ratehow relevant each descriptor was for describing the hotel’s atmosphere. Based on previous studies (Aaker 1997 andSchall 2003) a seven-point Likert-type scale was used, ranging from 1= Not at all decriptive to 7 = Extremely descriptive.The procedure was repeated twice, i.e., each subject rated three randomly assigned hotels each followed by aquestionnaire (i.e., 78 subjects x 3 questionnaires = 234 observations).

Manipulation check: The degree to which the hotels were assessed to have a characteristic atmosphere was as expected,with Motel 6 having the lowest mean score and Burj Al Arab the highest. Analysis of variance indicated a highlysignificant difference in scores among the six hotels, F(5, 228) = 66.46, p < 0.001.

Outcome: The low number of ratings for each descriptor-hotel combination (less than 10 on average) represented anobstacle to using multivariate analysis in the first stage. To isolate the most relevant indicators, descriptors that were notamong the top 50 most relevant descriptors for any hotel (which amounted to 257 descriptors) were removed, leaving201 (458–257) for further testing. Consequently, the outcome of the first stage was a reduction in the number ofdescriptors from 458 to 201.

Second stage (reduction from to 201 to 135 descriptors

This stage (n = 77, 82% female, mean age = 24.0, different sample from the same student population) yielded a furtherreduction to 135 descriptors. A total of six new hotel presentations were used, representing a broad range of hotels(difference in perceived atmosphere was confirmed by the same type of manipulation check as in the first stage). Theprocedure was identical to the first phase except that only two versions of the questionnaire were used (i.e., subjects ratedthe relevance of a randomly selected one-half of the 201 descriptors, i.e., 101 descriptors in the first version of thequestionnaire and 100 in the second). The significantly higher number of ratings for each descriptor-hotel combinationallowed for a more refined set of criteria to be employed. It was decided to retain descriptors that: (a) Discriminatedbetween hotels, (b) Explained whether the hotel was perceived to have a characteristic atmosphere, (c) Had highrelevance for at least one hotel (i.e., high mean score).


APPENDIX A (CONTINUED)

Detailed Description of the Deductive Phase of the Study

The rationale for including criterion (a) is that the ability to discriminate is a fundamental property of measurementinstruments. Analyses of variance identified a total of 21 variables with substantial (F > 25) discriminatory power. Theseinclude descriptors like airy and idyllic, which are highly relevant for some hotels and not relevant for others.

The rationale for including criterion (b) was to test the descriptors in relation to a criterion variable (characteristicatmosphere). Separate regression analyses were performed for each version of the questionnaire, where extent to whichthe hotel was perceived to have a characteristic atmosphere was used as dependent variable and the various descriptorswere tested as explanatory variables. There were a total of 10 significant explanatory descriptors (p < 0.05) in the firstversion of the questionnaire and combined these variables explained more than 75 percent of the variation in thedependent variable (adjusted R2 = 0.762). In the second version of the questionnaire there were also 10 significantexplanatory descriptors, with slightly higher combined explanatory power (adjusted R2 = 0.775).

Criterion (c) is the same pragmatic criterion that was used in the first stage. To avoid that too many descriptors were keptfor further testing, the cut-off point was set at 40 (i.e., descriptors needed to be among the most 40 most relevantdescriptors for at least one of the hotels to qualify under this criterion).

Sixty-six descriptors failed to meet any of the three above-mentioned criteria and were therefore eliminated. A total of135 descriptors met at least one of the criteria (several of these satisfied more than one criterion) and were consequentlykept for further testing.

The use of a factor analytic approach was ruled out in the second stage because of the fairly low ratio of number ofobservations to variables (77 subjects x 3 questionnaires = 231 observations further divided by 2 versions of thequestionnaire, meant that there were on average 115.5 observations to test the 100 or 101 descriptors in each version,representing a ratio of 1.15, which we considered inadequate for factor analysis).

As is common in most countries, the majority students who specialize in hospitality management and tourism are female.However, we decided to use this group because their particular interest and expertise in the topic being researched wasconsidered to outweigh the potential problem of gender and subject bias. To check for a possible gender effect, analysesof variance were performed for each of the 201 descriptors, with gender and hotel as independent variables. While almostall variance was driven by hotel, gender had a significant main effect (p < 0.05) on 13 of the 201 descriptors. This is mostlikely mainly due to random effects (expected random number = 201 x 0.05 = 10.05). However, to rule out possiblebiasing effects it was decided to employ a more heterogenic sample in the third and final stage.

Third stage (reduction from to 135 to 43 descriptors)

This stage employed a more varied sample of students (n = 278, 59% female, mean age = 23). To improve gender balance,students from several male dominated areas were included (accounting, IT, petroleum technology, engineering andchemistry). This was also done to increase the heterogeneity of the sample and thus reduce any possible bias fromparticular disciplines of study.

Each subject rated three randomly assigned hotels (from a total of 12, i.e., the six from the first stage and the six fromsecond). After each presentation the subjects completed a questionnaire as in earlier stages, except that this time therewas only one version of the questionnaire, which included the 135 descriptors under scrutiny.

The considerably higher number of observations for each descriptor compared to the earlier stages enabled a factoranalytic approach (the number of observations, i.e., 278 subjects x 3 questionnaires = 834 observations, was consideredadequate for using a factor analytic approach on the 135 descriptors, i.e., ratio of 6.2).

Different factor models and descriptors were tested by confirmatory factor analysis using LISREL. In order not to “throwthe baby out with the bath water” (i.e., accidentally delete descriptors that would be relevant for testing in the field, wedecided on the most elaborate model with acceptable fit (RMSEA = 0.050). In addition to the descriptors included inthis model, six additional descriptors that came out as significant explanatory variables in the regression analysis werealso kept. In total, 43 descriptors were retained in the inventory and subsequently tested in the exploratory field survey.


Hotel #1

This is a deluxe, full-service, hotel, centrally located in thecapital, between the parliament building and the RoyalPalace, within walking distance of the main shopping andcultural areas, as well as its sights. The hotel was openedin 1874 and is one of the best-known hotels in the country.Prices are upper-range and the hotel caters mainly forhigh-class business travelers.

Atmosphere profile:

Atmosphere profile:

Hotel #3

The hotel is situated next to the market, set among charm-ing old timber houses in a coastal town in SouthernNorway, which is the town in Norway with most days ofsunshine per year. The hotel has a high standard of serviceand a particular southern, maritime feel. There is a varietyof restaurants and entertainment available, just a walkaway from the hotel.

Atmosphere profile:

Hotel #2

This is one of the most original hotels in chain, locatednext to a famous amusement park/zoo in the Southern partof the country. Motifs and effects from the animal king-dom are a main feature of the rooms and communal areas.During summer the hotel is geared toward families withyoung children. Outside the summer-season, the hotelmainly focuses on the course and conference market.

APPENDIX B

The Six Hotels that Participated in the Field Study


APPENDIX B (CONTINUED)

The Six Hotels that Participated in the Field Study

Hotel #4

This is a standard business and conference hotel located inthe outskirts of the capital. The hotel is basic but func-tional. Public transport is within walking distance and theairport bus stops five minutes away from the hotel, on itsway to and from the main airport.

Atmosphere profile:

Hotel #5

Located in the North-Western part of the country, thishotel has been called Norway’s most complete and flex-ible hotel for culture and conferences. The hotel is de-signed to look like a huge sail, and thwe rooms have asuperb view to the nearby fjord and surrounding moun-tains.

Hotel #6

This hotel is surrounded by mountain terrain. In winter,there is skiing for all abilities, while the water, wide openspaces and montains make it an ideal place for hiking,riding, and fishing during summer. The hotel is a popularvenue for courses and conferences.

Atmosphere profile:

Atmosphere profile:


For further information contact:Morten Heide

Department of Business AdministrationUniversity of Stavanger

N-4036 StavangerNorway

Phone: 47.51.83.37.51Fax: 47.51.83.37.50


Kjell GronhaugDepartment of Strategy and Management

Norwegian School of Economics and Business AdministrationBreiviksveien 40N-5045 Bergen

NorwayPhone: 47.55.95.94.60

Fax: 47.55.95.94.30E-Mail: [email protected]


RETURN ON SERVICESCAPE INVESTMENTS: DOES REMODELING

MAKE A DIFFERENCE?

Dwayne D. Gremler, Bowling Green State University, Bowling GreenElisabeth Brüggen, Maastricht University, The Netherlands

Bram Foubert, Maastricht University, The Netherlands

SUMMARY

Service managers expend a large amount of resourceson designing, building, and furnishing the servicescapesof their establishments. Although the importance of theservicescape within the service encounter has been estab-lished, little research has examined the impact of a signifi-cant remodeling or redesign of a service facility. Thisstudy examines the impact of the major remodeling of aservice environment (a fast food restaurant); of particularinterest is the short- and long-term impact of significantchanges in the servicescape on customers’ affective re-sponses, cognitive responses, and approach behaviors.

Service providers spend millions of dollars per yearto design, build, and furnish their service establishmentsto create “physical evidence” of the provided servicequality (Bitner 1992). Cutthroat competition has led ser-vice providers to employ the service environment as asource of differential advantage. The importance of theservicescape – defined as the environment where theservice is delivered and where the customer and the firminteract – within the service encounter has been estab-lished (e.g., Bitner 1992). Numerous empirical studieshave investigated different aspects of the servicescapesuch as music (e.g., Baker, Levy, and Grewal 1992;Milliman 1986, 1982), lighting (e.g., Baker, Grewal, andParasuraman 1994; Baker, Levy, and Grewal 1992), andodors (e.g., Mitchell, Kahn, and Knasko 1995;Spangenberg, Crowley, and Henderson 1996). However,most servicescape research is cross-sectional and studiesdifferences in the servicescape in a “timeless” setting; thatis, it does not consider changes in an existing serviceenvironment nor does it investigate the effects of suchchanges over time. In addition, while most studies focuson minor changes in particular servicescape elements,little research addresses the impact of a major remodeling.Finally, extant research is often conducted in merchan-dise-oriented retail contexts, although changes in theservicescape are arguably more important for service-oriented retailers.

The purposes of this study are twofold. First, weextend marketing theory by investigating the effects overtime of a significant remodeling of a service provider’sfacility. The impact of a significant, major remodeling ofthe servicescape has, to our knowledge, not been investi-

gated in the marketing literature. Second, building onBitner’s framework (1992), we develop a complete set ofquestions to measure customer perceptions of ambiencefactors, servicescape design, and related social factors.We test how changes in the servicescape influenceservicescape perceptions, affective and cognitive re-sponses, and approach behaviors by collecting data be-fore a major remodeling effort, right after the remodeling,and five months after the remodeling. Our dependentmeasures include key customer consequences of interestto service firms, including satisfaction, service quality,value, image, loyalty, desire to stay, and to spend moremoney, future share of wallet, repatronage intentions, andpositive word-of-mouth communication.

The impact of significantly remodeling theservicescape is investigated with a field experiment in aEuropean fast-food restaurant. This restaurant (part of aninternational chain) underwent a major remodeling effortin September 2006, in which the store atmosphere, light-ing, spatial layout, furnishings, symbols, artifacts, colorschemes, and other factors were significantly changed.We employ a longitudinal design to capture customerattitudes and behavioral intentions at three points in time,including: two months before the remodeling (T

1), right

after the remodeling (T2), and five months after the remod-

eling (T3). To control for seasonal effects or general

changes in customer attitudes and behavior due to adver-tising campaigns, the economic situation, or other exter-nal factors, data were also collected at another fast-foodrestaurant of the same chain in the same general region ofthe country. This second restaurant is similar in size andcustomer structure, but did not undergo any changes to theservicescape.

Our analysis of data gathered from a field experiment(i.e., a significant remodeling of a servicescape) revealed(1) increases in respondent perceptions of the servicescape,(2) short-term, positive increases in customers’ affectiveand cognitive responses and approach behaviors, and (3)mixed results with responses over a longer period of time.

In particular, we find customer perceptions of theservicescape dimensions to increase as a result of a sig-nificant servicescape remodeling in the short-term (i.e.,from T

1 to T

2) for nearly all of the dimensions we mea-

sured, indicating that respondents find the new environ-


ment more welcoming and pleasing. Five servicescapeitems (external appearance, exterior colors, outside furni-ture, warm atmosphere, and vibrant atmosphere) increaseeven further in the long-term. Thus our findings suggestthat significantly changing such environmental elementsas lighting, music volume, layout, furnishings, paintings,furniture materials, and exterior and interior color schemescan be successful from a design perspective since they arewell perceived by customers. Overall, we also find thatthere is an immediate positive effect of remodeling theservicescape on customers’ affective responses, cognitiveresponses, and approach behaviors. That is, a significantremodeling of a servicescape positively influences cus-tomer affective responses (in terms of their attitudes,pleasure, and arousal) in the short-term, implying that achange in the servicescape activates positive emotionalresponses in consumers. We also find that a remodeledservicescape leads to more positive immediate cognitiveresponses in the short-term. Finally, an improvedservicescape encourages approach behaviors in the short

term, as the data indicate that customers are more likely tospend more money, return to the store, stay longer, andpromote the store to friends and family. Hence, positivechanges in the store environment influence short-termapproach behaviors in a way that is beneficial for the firm.

However, our research shows that positive percep-tions of a remodeled servicescape to do not necessarilylead to strong positive results in the long-term. Surpris-ingly, most affective responses to significant changes inthe servicescape, as well as several cognitive responsesand approach behaviors, return to their original (base)levels over time. Indeed, for variables such as attitude andarousal, overall satisfaction, service quality, value, word-of-mouth communication, and repatronage intentions, itseems that the novelty stimulus from the new design doesnot last very long. We do find that a few variables (brandimage, store image, desire to stay, and future share ofwallet) stay at the new, higher level in the long-term.References are available upon request.

For further information contact:Dwayne D. Gremler

Department of Marketing226 Business Administration

College of Business AdministrationBowling Green State University

Bowling Green, OH 43403–0001Phone: 419.372.0226



THE STRUCTURE AND FUNCTION OF LANGUAGE IN THE

STRATEGIC RITUALS OF SERVICE PROVIDERS

Cele C. Otnes, University of Illinois at Urbana-ChampaignBehice Ece Ilhan, University of Illinois at Urbana-Champaign

Atul Kulkarni, University of Illinois at Urbana-Champaign

ABSTRACT

As part of its “Sleep AdvantageTM” program, guestswho check into the Crowne Plaza hotel find on their bedsa burgundy drawstring bag containing earplugs, an eyemask, a plastic bottle of lavender linen spray, and a CDwith the words “Sleep soundly” on its cover. Inside theCD case, the text explains the Sleep AdvantageTM pro-gram and lists the amenities the hotel provides that aredesigned to help guests sleep. The case also lists ten“Sleep Tips,” and the CD itself includes eight differenttracks combining music and sonorous instructions thatguests are supposed to play in order to help them relax, andultimately, to sleep.

As a marketing program, Sleep AdvantageTM clearlyis designed to differentiate the Crowne Plaza brand in thehighly competitive business-oriented hotel category. Butwith its gifts, elaborate descriptions, and scripted instruc-tions, it is also an excellent example of what we describeelsewhere as the use of a “strategic ritual,” or the use of aritual by a marketer to achieve specific firm- and/orcustomer-oriented objectives (Authors’ citation here,2007). Rook (1985, p. 252) defines rituals as multiplesymbolic and expressive behaviors “that occur in a fixed,episodic sequence and that tend to be repeated overtime . . . [that are] dramatically scripted and . . . performedwith formality, seriousness, and inner intensity.” Al-though Rook’s definition originally pertained to the typesof rituals that consumers create and in which they engage,it is intuitively obvious that marketers often employrituals in their interactions with customers, even if they donot label these activities as such. Such rituals can rangefrom the relatively self-contained greetings offered bydoormen when they welcome customers into a hotel, tomore elaborate offerings such as the “Saturn deliveryceremony” – where staff at the dealership thank, cheer,and take pictures of customers acquiring their new cars(www.saturnfans.com).

Most research that explores the role and impact ofrituals in marketing focuses on consumers’ experiences ofrituals such as holidays (e.g., Wallendorf and Arnould1991), life passages (Bonsu and Belk 2003), gift giving(e.g., Lowrey, Otnes, and Ruth 2004), and extraordinaryconsumption experiences (Kozinets 2001). Fewer studiesdo explore how service providers facilitate or hinderconsumers’ ritual experiences (e.g., Arnould and Price

1993; Sherry and McGrath 1989; Otnes, Lowrey, andShrum 1997; Sandikci and Ilhan 2007). Thus, the empha-sis on rituals in marketing clearly privileges consumer-to-consumer versus business-to-consumer experiences. Aspart of a stream of research that explores how serviceproviders and consumers understand the strategic ritualsthat marketers can employ, in this paper, we pivot the lensof ritual research to examine how service providers stra-tegically employ a specific aspect of strategic rituals toachieve specific customer- and firm-oriented objectives.Specifically, we explore the structure and function of oneritual element that is overlooked in the literature – the roleof language in these rituals. In another paper, we Webelieve this topic is important to academics and practitio-ners for at least three reasons. First, language [STARTHERE] although providers do create and offer strategicrituals, they may not be familiar with the structure, func-tion or boundaries of ritual behavior. Unpacking thenature of strategic service rituals will enable providers tocreate more meaningful and successful experiences forcustomers. Second, research within and outside of mar-keting consistently demonstrates that rituals provide mean-ingful structure to existence, enhance social bonds andcan transform people’s lives (Driver 1991). Exploringstrategic rituals can help retailers assess whether theexperiences they offer customers fulfill such functions.Third, services researchers often hint at – and even di-rectly mention – the potential of rituals and ceremonies toenhance customer relationships (e.g., Czepiel 1990; Siehl,Bowen, and Pearson 1992; Price, Arnould, and Tierney1995). But to our knowledge, ours is the first study tospecifically unpack how service providers implementthese elements into their customers’ experiences.

We began by asking a general, discovery-orientedquestion: “How do service providers use rituals wheninteracting with customers?” Interviews with providerssoon revealed that one key ritual element – and oneoverlooked in the scholarship on ritual – is language.Given that communication is crucial in delivering a satis-factory service encounter, we narrowed the focus for thispaper to address these questions: (1) What types of rituallanguage emerge in the strategic rituals of service provid-ers? (2) What objectives do providers try and achieve withritual language? After explaining our research methods,we will discuss the theoretical frameworks that inform ouranalysis.


METHOD

This study is part of a larger research project thatexplores the strategic use of rituals by service providers.We chose an interpretive approach in order to explain thesocial and cultural events of interest from the perspectivesand experiences of those being studied (Nobit and Hare1988). We conducted depth interviews with 23 serviceproviders in a small Midwest city (pop. about 100,000)and surrounding communities from September to Decem-ber 2006 (see Table 1). We screened providers in variouscategories, selecting those who incorporated customer-oriented rituals into their services. Informants variedacross service target, type of customer relationship, de-gree of customer contact, ability to satisfy peak demandand location of service delivery (Lovelock 1983).

We began by presenting Rook’s definition of ritualsto informants and asking them to describe any specificrituals they used. We then used prompts to elicit informa-tion on the services our informants provide, the structuraland functional elements of the rituals they offer, theirstrategic purpose for including rituals, perceptions of howtheir competition uses rituals, and perceptions of custom-ers regarding what types of rituals they expect fromproviders in this category. Interviews ranged from 45minutes to 2 ½ hours, were audio-taped and transcribed,and yielded approximately 800 pages of text. In analyzingour text, we sought emergent themes while engaging indialectical tacking, immersing ourselves in the interdisci-plinary literature on rituals to seek out consistencies and/or inconsistencies with our text. Once we realized thesalience of ritual language, we broadened our literaturereview to include studies from linguistics. This search ledus to research in speech act theory, which we believe is avalid and useful theoretical scaffold for our interpretation.After describing Rook’s structural ritual framework, weprovide an overview of speech act theory below.

THEORETICAL BACKGROUND

Rook’s (1985) Structural Typology of Ritual Elements

Rook argues there are four essential elements toritual: artifacts, or the tangible objects used (e.g., clothingand candles); scripts, or the norms that “guide the use ofthe various artifactual materials” (p. 253), performanceroles, or the explicitly or vaguely scripted parts peopleplay in rituals, and the audience, or the wider constituencythat may experience the ritual, but may not take part in it.Although this typology has proven to be a useful startingpoint for much ritual scholarship, it does not recognize theimportance of ritual language. Rather, Rook limits thediscussion of language to the section on ritual scriptswhere he observes that norms are sometimes highly codi-fied, and “may appear in a written document” (p. 253). Webelieve our text offers sufficient evidence that language

should be regarded as a stand-alone, highly salient ritualelement whose impact extends beyond merely serving asa vehicle to communicate ritual norms. In the next section,we describe speech act theory, which Robbins (2001)argues is highly applicable to understanding language inritual contexts, because both speech act and rituals shareperformative qualities.

Speech Act Theory

Speech act theory (SAT) is a reaction to the logical-positivist philosophers who claim language is primarilycomposed of declarative statements that can be assessedas true or false. In their seminal works outlining SAT,Austin (1962) and Searle (1969) argue that many utter-ances (such as prayers, and phrases like “thank you”)cannot be assessed in this manner, and that the realpurpose of an utterance (a term used for both oral andwritten language) is primarily to perform some action orcreate some reality within a social context (Cline 2002).Smith (1990) argues that even utterances that do not seemperformative, such as “Is John sitting down?” do performan act – in this case, stating indirectly, “I am askingwhether John is sitting down.”

Speech act theory is rooted in pragmatics, a branch oflinguistics that explores “how people comprehend andproduce a communicative act or speech act in a concretespeech situation” (Liu 2007). Speech acts are best under-stood as deliberate performances by speakers or writers,designed to accomplish specific objectives and to elicitsome reaction in the receivers of messages (Saeed 2003).Speech acts are composed of three structural elements: thelocutionary act (the “grammatical” or actual utterance),the illocutionary act (the sender’s intent when making anutterance), and the illocutionary effect (the receiver’sperceived understanding of the intent of the utterance;Hamermesh 1981). This emphasis on understanding thesocial function of utterances, and the possible contradic-tions between the intent and interpretation of an utterance,distinguishes speech act theory from more structural ap-proaches to language that have been applied to marketing(e.g., semiotics; Mick 1986).

Most SAT scholars adopt Searle’s (1969) typologythat classifies utterances into one of five categories. Withhypothetical examples from service contexts, these are:(1) assertives: acts that “commit the speaker to the truth ofan expressed proposition” (e.g., if a clerk says, “I have thatdress in a Size 5,” she must produce the dress when asked);(2) directives: acts that try to convince the recipient to actin some way (e.g., “Please refrain from smoking in therestaurant”); (3) commissives: acts committing the speakerto a future course of action (e.g., “I’ll have your car readytomorrow by 5 p.m.).”; (4) expressives: acts expressing apsychological state such as regret (e.g., “I’m sorry yourcar wasn’t ready as promised”) and (5) declarations: acts


TABLE 1

Summary of Service Providers Interviewed

Type of Service Yrs. In No. of

Provider Services Offered Business Type of Provider Locations

Art Museum Exhibition of art works, 40+ State institution 1tours, workshops,lectures, concerts

Bank Financial services, 139 Corporation 22wealth management

Beauty Salon/Spa Hair, nails, massages, 32 Sole proprietor 3 to 4aesthetics

Bed and breakfast Accommodations, meals 16 Sole proprietor 1

Bookstore Selling books, CDs, 19 Family partnership 1DVDs, gifts.

Car dealership New/used cars; 14 Franchise Chain –maintenance and service Multiple

locations

Destination Spa Spa services, yoga, 23 Partnership 1recreation, massage.

Home cleaning Housecleaning 28 Sole proprietor 1

Limousine/bus Transportation services 21 Sole proprietor 1

Massage therapist/ Massage therapy/private 3 Sole proprietor 1yoga instructor yoga sessions

Premium dessert Ice cream, cakes, shakes, 1 (this location); Franchise 1,400+retailer smoothies. 15 whole (co.)

Private middle school Education 13 Non-profit; 1for girls board-governed

Realtor Assist buyers/sellers 20 Sole proprietor 1

Restaurant Full-service fine dining 3 Partnership 1

Restaurant Full service restaurant 30 Partnership 1

Restaurant Steakhouse dining Approx 12 Franchise Chain

Tree service Tree care; landscaping 16 Sole proprietor 1

Univ. athletic dept. In-game promotions 100+ State institution 1

Univ. office of Advise undergraduate Unspecified State institution 1undergraduate affairs students


that create a new state of reality (e.g., “I appoint youCustomer of the Month”) (Winograd and Flores 1986).

SAT scholars are also interested in determining theconditions under which utterances are felicitous, or under-stood by the receiver in the way the sender intends (Austin1962). If this understanding does not emerge, it is impor-tant to identify the type of breakdown that occurs (e.g., a“misfire,” where a recipient dismisses an utterance, or abreakdown because an utterance is inappropriate in aparticular social context; Austin 1962). Unpacking thecontradictions between a sender’s illocutionary act andthe illocutionary effect on the receiver within ritual con-texts should enable service providers to better understandthe power of language.

INTERPRETATION

Because our data focuses on services providers, andnot provider/customer dyads, this paper explores only thefirst two aspects of speech act theory – the actual(locutionary) a speech act and the intended (illocutionary)act. We organize our interpretation around unpackingthese two elements. Table 2 summarizes the locutionaryand illocutionary speech acts that emerge in our text. Weorganize our discussion around these linguistic unitsbecause our informants identify them as the units ofanalysis that were performative (e.g., that contain theritual-related meaning they are trying to impart).

Neologistic Phrase

A neologistic phrase contains an entirely new wordthat the service provider creates. The use of neologisms inmarketing is a relatively common practice, because it isassumed coining a novel word for a product, service orexperience can differentiate these entities from the com-petition immediatel y. For example, “Fourthmeal,” a

neologism currently touted by Taco Bell, actually encour-ages consumers to add a new meal, one “between dinnerand breakfast” (www.fourthmeal.com), to their existingdining rituals. One informant, Kay, prominently featuresa neologistic phrase in her greeting ritual:

I: Our brochure says, “Welcome to Superbia.” And wewant [guests] to understand that this town has the bestof the big towns and the little towns mixed together.And . . . we like to share our home and our town withother people.

R: You said “superbia,” what’s that?

I: It’s a combination of . . . Superb-ia, so they under-stand it’s like the suburbs, but it’s a superb suburb. . . .

R: Is that something you coined?

I: Yes . . . that’s just our coined phrase.

Kay’s greeting ritual not only accomplishes theillocutionary act of welcoming her guests to her Bed andBreakfast, but also helps her craft a unique, premiumidentity for her town, and hopefully instill positive atti-tudes in visitors about the locale. As such, this neologismhelps Kay sacralize her community (Belk, Wallendorf,and Sherry 1989).

Mantra

While the power of a neologistic phrase stems fromits novelty, a mantra typically is a word or phrase with old(perhaps even ancient) spiritual or religious roots. Repeat-ing a mantra is supposed to connect an individual tospiritual beliefs or a spiritual entity. Because spiritualityand religion are deeply personal, providers recognize thattheir decision to incorporate mantras into their service

TABLE 1 (CONTINUED)

Summary of Service Providers Interviewed

Type of Service Yrs. In No. of

Provider Services Offered Business Type of Provider Locations

Veterinary clinic Preventive wellness, 48 Family 3vaccinations, boarding, partnershipeuthanasia

Voice teacher Singing instruction 30 Sole proprietor 1

Wine merchant Retail, wine bar, café 4 Sole proprietor 1

Yoga Studio Owner Yoga classes, private 10 Sole proprietor 1instruction, massage


could be controversial. Donna describes how she says thesacred Hindu syllable Aum (pronounced “Ohm”) in heryoga classes. She recognizes not everyone will want toreciprocate, as is the yogic custom:

We say “Aum,” and that is a ritual in a sense . . . I donot do it in all classes because some . . . intro classes,it is a little too early in their experience – people havesome conflict with their religion, you know . . . [inclasses where she says “Aum”], there is probablyhalf, maybe two thirds that still won’t do it. But I closemy eyes [when uttering “Aum”] . . . I feel it is yourown personal experience that you don’t need to haveme looking. I have no idea how many people aresaying it . . . you know, you’ve got some loud folksand [in] other classes, they’re real soft, but I can tellthey are doing it . . . And I figure over time let themstart that . . . even if they just think it.

Donna’s reflection on “Aum” reveals she uses themantra to try and help her customers become open to thespiritual and possibly transforming nature of yoga. Nev-ertheless, she is clearly cognizant of the fact that thisillocutionary act may not have the desired illocutionaryeffect. Likewise, as the owner of a day spa, Megancomments that she cannot say religious-holiday mantrasin greeting rituals, because she might offend her custom-ers: “I can’t say Merry Christmas, but I can say HappyHolidays.”

Thus, while service providers might find mantrasappealing to achieve specific goals, the decision to in-clude them must take into account whether customers willdeem a particular utterance as appropriate in a particularsocial context (Austin, 1962) – and if not, how using amantra might impact provider/customer relations.

The Slogan

A slogan shares particular characteristics with a man-tra, in that it is often short, repeated over time, and maytake on important and maybe even transformational cul-tural connotations (e.g., Nike’s “Just Do It.”). However,most slogans are created by an advertiser to position aproduct or service (Slogan: Define). Advertisers recog-nize slogans can help them solidify rituals in a culture, andenact changes in ritual participation (e.g., De Beers’ “ADiamond Is Forever” campaign; Otnes and Scott 1996).Similarly, Joe uses slogans to assist him in buildingcustomer interest in sports rituals:

We’ve got written up there [on his whiteboard] “Tail-gating is game day, all day” and “The greatest tailgat-ing environment in [the conference].” So clearlythese are two tag lines . . . that we feel are a big partof what a football game day is all about. . . . So, it’s

a big focal point for us . . . So obviously we are reallypushing the tailgating atmosphere. . . .

I: What benefits are you hoping the program will gleanfrom this?

R: The revenue aspect, the attendance aspect, the brand-ing and image . . . and it makes it more of a destinationfor a football Saturday . . . and it becomes even agreater point not even for the Athletic department butfor the University as well.

Joe’s attempt to boost the tailgating ritual reveals hisclear understanding that within American culture, sportsare valorized occasions that possess their own uniquerituals. Belk, Wallendorf, and Sherry (1989, p. 12) ob-serve sports fall into the venue of sacred experiences,where “stadiums are temples that may be the site ofpilgrimages . . . and fans participate in pre- post, andduring-game rituals.” In addition, Joe recognizes that byissuing a directive designed to persuade consumers of thequality of the tailgating experience, they are likely to wantto repeat this event, and such desires will be reflected inincreased ticket sales and revenue for the University.

Prescription

Prescriptions are rules or laws for behavior, whichcan both be used to encourage a particular type of behav-ior, or to discourage activities. As previously stated, thisspeech act is the only element of ritual language Rook(1985) mentions, and it is synonymous with ritual scripts.Our informants often create “initiation” rituals (thosedesigned to familiarize customers with the scope andquality of their services), such as tours and orientationsessions. Although these often fulfill the roles of educat-ing consumers, it is equally important to providers thatthese rituals serve as vehicles to teach customers tocomply with particular rules or requirements the providerwishes to enforce. Joe mentions he stresses the rule “Wear[A Color]” to football games. In so doing, he believes theidentity of the university is collectively enforced, because“the thing about the [color] is that within [Conference] . . .we are the [color]. If you wore blue, you could be [Univer-sity X], you could [also] be [University Y] . . . nationallyit is a rare color . . . it makes our [fans] stand out moreclearly.” By emphasizing this rule, Joe specifically directshis customers’ use of clothing, a key artifact in the ritualcelebration (Rook 1985), while targeting key institutionalobjectives.

With respect to prescriptions that forbid behavior,service providers recognize such rules may conflict withconsumers’ own sense of agency, a problem naturallycompounded by the fact that customers are paying for aservice. Meg, who runs a destination day spa, notes she


TABLE 2

Examples of Speech Acts in Service Provider Rituals

Locutionary Example from Type of Illocutionary Marketing

Act Definition Text Ritual Act Objective

Neologistic Phrase containing “Welcome to Greeting Expressive Build Identity;

phrase any new word Superbia” Reassure Customer,

introduced into a (B&B Differentiate

language by Owner)

whatever process.1

Mantra A commonly “Namaste” Parting Expressive Enhance

repeated word or (Yoga Relationship/

phrase, often with instructor) Loyalty

sacred or spiritual

connotations. 2

Slogan “Memorable motto/ “Tailgating Pilgrimage Assertive Increase repeat

phrase used in is Game behavior/loyalty/

political, religious Day, All Build excitement

or commercial Day.”

contexts…repetitive

of an idea or

purpose.”3

Prescription The act of “Wear Orange”; Grooming, Directive Encourage customer

establishing official “No swimming Orientation compliance

rules, laws, or alone” Rituals

directions.4

Song Relatively short Sing songs Ritual Expressive, Enhance

musical composition when receive socialization but also relationships

for the human tips; directive with customers;

voice…the words (Premium (encourages enhance employee

of which are Dessert future tips) morale

typically of a poetic, Store)

rhyming nature.6

Text Levels of linguistic Poem from Rite of Expressive Relationship

analysis higher than veterinarian Passage building/loyalty

a sentence… after pet

connected to euthanasia

explicit and often

formulaic linguistic

representation,

whether written or

verbally ritualized.5

Conversation Social action that is Discussion of Orientation Directive, Enhance provider

spontaneous and not physical/ Commissive effectiveness,

constructed, but Emotional which enhances

governed by some state of satisfaction/decreases

rules (e.g., turn- customers complaining

taking, sequences (Spa, voice behavior.

of utterances).7 teacher)

1. Matthews, P.H. (1998), The Concise Oxford Dictionary of Linguistics.

2. “Mantra,” Wikipedia.com, June 26, 2007;

3. “Slogan,” Wikipedia.com, June 26, 2007;

4. dictionary.com, June 26, 2007;

5. International Encyclopedia of Linguistics 2003. (International Encyclopedia of Linguistics 4 volumes, Second Edition Edited by

William Frawley),

6. http://neohumanism.org/s/so/song.html;

7. Strazny (2005), Encyclopedia of Linguistics.


conducts a thorough orientation with her guests in order tofamiliarize them with the rules. By couching these rules inan initiation ritual of orientation, and in customer-ori-ented language, she tries to make them more palatable. Atthe same time, she makes clear that the illocutionary effectshe desires goes beyond delivering a good customerexperience. She begins her rules with these words:

R: “For your safety, for your best interest, for your well-being. To create a positive vacation, please follow thefollowing instructions. . . . ” When they come into thedoor we do an orientation . . . sit down with all ofguests at the same time, and say “welcome. . . . ” Andbecause we’re small . . . we pre-schedule [massagesand or facials] . . . Some will say, “Well, maybe I don’tknow if I want it then. . . . ” Once in a while they’ll go,“there might be too many rules.” Well, that’s how Ikeep my price down and I can be affordable . . . anddo positive things. I always try to put a spin on it, butif I don’t run efficiently, I won’t be here next year.

In sum, the locutionary act of the prescription is oftensubsumed within a ritualized context in order to soften theblow of asking consumers to give up freedoms (Author’scite here). By using greetings and orientation as vehiclesfor communicating rules, service providers believe theygain consumer compliance, and achieve corporate-relatedgoals such as minimizing costs and running efficiently.

Song

Another form of ritual language – one involving moreof a performative element – is the song. In other studiesthat identify service rituals, songs are seen as linguisticdevices that can cast spells and enhance bonds betweenproviders and customers (e.g., singing “Kodachrome” soclouds will disappear while river rafting; Arnould, Price,and Otnes 1999). However, our providers use songs bothas teaching tools and as means of enhancing atmospher-ics. Bill, who owns a dessert store, observes:

We sing for tips, but . . . [also] for the ambience andfor the experience. If people have not been tipping fora long time, people sing a song, and . . . systematicallythey are supposed to say “Thanks for the tip” at theend . . . everyone who is potentially standing in linewill know that the crew can earn an extra dollar tip . . .It’s a corporate initiative that helps with crew mem-ber satisfaction . . . and it helps build a happy, enthu-siastic environment in the store, and an entertainmentfactor.

In this case, songs not only serve as elements of ritualsocialization (teaching customers that tips are expectedand acceptable) but also as a way to effect the mood andpersonality of the provider as an element of atmospher-ics – itself deemed a form of language (Kotler 1973).

Text

A more complex locutionary act that providers incor-porate into rituals is text, typically a one-way communica-tion to a recipient that is not altered or alterable onceappearing in its final, mediated form. Our informants relyon many types of commercially-available text to enhancethe quality of their consumers’ ritual experiences. Donnaoften offers readings as part of her parting ritual in heryoga class, noting that after the final relaxation posture,“the body, the spirit, is in an open state and kind of readyto listen a little bit more . . . it is a teaching experience . . .I have had people in class cry during the readings . . . andsay, “Oh my God, that was so perfect for me.” As such,Donna’s intent is to use an expressive text to create apsychological connection with her students, which shehopes will then translate into increased customer loyalty.

Likewise, Stan’s veterinary practice sends both a giftand a poem to customers when after euthanizing a pet:“There is a ‘Rainbow Bridge,’ it’s a poem about death. . . .We attach that to the paw print. So we feel it sort of says,‘That relationship is over, the memories are there, maybehopefully someday we will see that pet again.’” Interest-ingly, over time this ritual has grown from sending a card,to sending the more elaborate paw print and poem. Whenasked to explain this transition, Stan observes:

I: You see, the client bond has grown so much since Ibegan practicing.

R: With you guys?

I: No, with their pet . . . it really is truly part of thefamily. . . . When my dad started practicing in 1959 . . .basically they would drag the pet on a chain from theback yard . . . he’d give a rabies shot, and they’d gohome . . . [They’ve] go[ne] from this backyard pet tofamily member.

By coupling a meaningful commercial text with agift, Stan’s intent is to demonstrate that his practice isaware of the pet-as-family metaphor (Hirschman 1994),and that this bond deserves ritual commemoration. Hefurther hopes the ritual enables him to build trust with hiscustomers, and build loyalty among them: “If this . . . ishandled properly, then when they get a new pet they’ll stillfeel like we did a good job even with that aspect of it . . .they [customers] could easily leave you and go.”

Conversation

The final locutionary act we discuss is the conversa-tion, which is a two-way, relatively more spontaneousinteraction between the service provider and the cus-tomer. Although conversations may obviously occur innon-ritualized contexts between parties, some service


providers specifically mention they deliberately incorpo-rate ritualized (e.g., highly structured, important, often-repeated) conversations in their customer interactions.The importance of conversation as a ritual element iscaptured in a recent ad campaign by Hair Cuttery, thenation’s largest franchise hair salon, which features theheadline: “A good haircut starts with a good conversa-tion.”

Many informants believe conversations to be keyelements in greetings and orientations. Jane, a massagetherapist, keeps the frontstage/backstage nature of theconversational act (Goffman 1959) squarely in mind,noting when her customers arrive, “I always walk themback and we talk. I ask them how they’re doing. I don’treally ask them out here. I sort of keep this [living roomfor] public talk. Essentially keep the [massage] roomprivate talk because sometimes my husband is here andthat’s private conversation.” Likewise, Don observes theimportance of having a conversation with his voice stu-dents when they arrive: “That ritual . . . begins with agreeting . . . and most often, an inquiry into how thestudent is feeling. Because if they’re not well, eitherphysically or emotionally, it has an impact on how wellthey can sing. So, that’s kind of an invitation to sharewhat’s going on in their lives.”

Sometimes, the conversation with a customer is evenmore formalized, and the service provider uses specific,often elaborate sequence of utterances within the conver-sation in order to perform desired objectives. Jan, arealtor, describes this behavior:

R: [With] either buyer or seller, I like to sit down withthem for at least an hour and say, just to get to knowthem, “What are you looking for?” “How do you liketo do this?” and “How do you like to do that?” . . . soit’s really good for me to be in contact with thembeforehand, and in a way that’s a ritual because I haveto do it with every client.

I: Is there a defined set of questions you make them gothrough?

R: Yes . . . but I do not make them go through the list. Ihave the list but I cover them in conversation.

In sum, ritualized conversations can help serviceproviders accomplish specific objectives. First, they can

help providers better enhance the service they offer, andprovide a level of customization that would be impossiblewithout such ritual conversation. Second, they can helpproviders demonstrate their sincere desire to build andmaintain relationships with customers.

FUTURE RESEARCH

One limitation of our data stems from our sample,which consists primarily of one-location service provid-ers. To build on our findings, future research shouldinclude a greater range of providers in various geographicsettings. A more diverse sample – e.g., one with globalfranchises, or internet-only providers – is likely to offerricher insights in the strategic use of rituals to increase thequality and quantity of our knowledge of the use of speechacts in service rituals. Future research should also take adyadic approach of analyzing interactions between bothservice providers and customers to unpack the under-standing of illocutionary effects. This approach can en-able us to reach beyond the service organizations wherestrategic rituals are rooted, and unveil the positive andnegative effects of characteristics of rituals on consumers.

Another interesting avenue for future research wouldbe to study the boundaries of strategic rituals. The issue ofritual boundaries with speech acts emerges numeroustimes in our data (e.g., mantras). Future research couldspecifically focus on studying the ways service providersmay navigate possibly controversial speech acts, and howcustomers perceive these vis-à-vis their service experi-ence. Finally, future research may investigate the differ-ences in speech acts across customer segments. For ex-ample, providers may interact differently with experi-enced customers as compared to novices because differ-ent assumptions may govern the use or non-use of utter-ances as familiarity with a customer increases.

In conclusion, we unpack how service providersimplement strategic, customer-oriented rituals. The find-ings suggest language is a highly salient, impactful ele-ment of strategic rituals, and an element that can helpservice providers accomplish (or fail to accomplish) spe-cific marketing objectives. Although we identify the struc-ture and function of specific speech acts, our goal is tocontinue exploring the relationship between the utter-ances service providers execute in ritual contexts, and theimpact of these utterances on customer/service providerrelationships.

REFERENCES

Arnould, Eric J. and Linda L. Price (1993), “River Magic:Extraordinary Experience and the Extended ServiceEncounter,” Journal of Consumer Research, 20

(June).____________, ____________, and Cele Otnes (1999),

“Making Consumption Magic: A Study of White-Water River Rafting,” Journal of Contemporary Eth-nography, 28 (February), 33–68.


Austin, J.L. (1962), How to Do Things with Words.Cambridge: Harvard University Press.

Author’s citation (2007).Belk, Russell (1989), “Materialism and the Modern U.S.

Christmas,” Interpretive Consumer Research, Eliza-beth Hirschman, ed. Provo: Association for Con-sumer Research, 136–147.

____________, Melanie Wallendorf, and John F. Sherry,Jr. (1989), “The Sacred and Profane in ConsumerBehavior: Theodicy on the Odyssey,” Journal ofConsumer Research, 16 (June), 1–38.

Bonsu, Samuel K. and Russell W. Belk (2003), “Do NotGo Cheaply into That Good Night: Death-RitualConsumption in Asante, Ghana,” Journal of Con-sumer Research, 30 (June), 41–55.

Cline, Richard A. (2002), “Understand and Act: ClassicalRhetoric, Speech Acts, and the Teaching of CriticalDemocratic Participation,” doctoral dissertation,University of Missouri – Kansas City.

Crystal, D. (1987), The Cambridge Encyclopaedia ofLanguage. Cambridge: Cambridge University Press.

Czepiel, John A. (1990), “Service Encounters and ServiceRelationships: Implications for Research,” Journalof Business Research, 20 (January), 13–21.

Driver, Tom (1991), The Magic of Ritual. San Francisco:HarperCollins.

Goffman, Erving (1959), The Presentation of Self inEveryday Life. Garden City, NY: Doubleday.

Hamermesh, Madeline (1981), “Sharpening the Old Saws:Speech-Act Theory and Business Communication,”Journal of Business Communication, 18 (2), 15–22.

Hindunet.org, “Aum,” accessed June 27, 2007.Hirschman, Elizabeth C. (1994), “Consumers and Their

Animal Companions,” Journal of Consumer Re-search, 20 (3), 616–32.

Kotler, Philip (1973), “The Major Tasks of MarketingManagement,” Journal of Marketing, 37 (October),42–49.

Kozinets, Robert V. (2002) “Can Consumers Escape theMarket? Emancipatory Illuminations from BurningMan,” Journal of Consumer Research, 29 (June),20–38.

Liu, Shaozhong (2007), “What is Pragmatics? [www.gxnu.edu.cn/Personal/szliu/definition.html], accessedJune 23, 2007.

Lovelock, Christopher H. (1983), “Classifying Servicesto Gain Strategic Marketing Insights,” Journal ofMarketing, 47 (Summer), 9–21.

Lowrey, Tina M., Cele C. Otnes, and Julie A. Ruth (2004),“Social Influences on Dyadic Giving Over Time: ATaxonomy from the Giver’s Perspective,” Journal ofConsumer Research, 30 (March), 547–58.

Mick, David Glen (1986), “Consumer Research andSemiotics: Exploring the Morphology of Signs, Sym-bols, and Significance,” Journal of Consumer Re-search, 13 (September), 196–213.

Noblit, George W. and R. Dwight Hare (1988), Meta-Ethnography: Synthesizing Qualitative Studies,Qualitative Research Methods, Vol. 11, NewburyPark: Sage Publications.

Otnes, Cele and Linda M. Scott (1996), “Something Old,Something New: Exploring the Interaction BetweenRitual and Advertising,” Journal of Advertising, 25(Spring), 33–50.

____________, Tina M. Lowrey, and L.J. Shrum (1997),“Toward an Understanding of Consumer Ambiva-lence,” Journal of Consumer Research, 24 (June),80–95.

____________ and Elizabeth H. Pleck (2003), CinderellaDreams: The Allure of the Lavish Wedding. Berke-ley: University of California Press.

Price, Linda L., Eric J. Arnould, and Patrick Tierney(1995), “Going to Extremes: Managing Service En-counters and Assessing Provider Performance,” Jour-nal of Marketing, 59 (April), 83–97.

Robbins, Joel (2001), “Ritual Communication and Lin-guistic Ideology,” Current Anthropology, 42 (De-cember), 591–614.

Rook, D. (1985), “The Ritual Dimension of ConsumerBehavior,” Journal of Consumer Research, 12, 252–64.

Saeed, John (2003), Semantics, 2nd ed. Blackwell Publish-ing.

Sandikci, Ozlem and Behice E. Ilhan (2007), “Rituals andthe Marketplace: The Impact of Market Agents onConsumption Rituals,” Presented at the ConsumerCulture Theory (CCT) Conference, Toronto,Canada.Saturn [www.saturnfans.com/forums/showthread.php?:t=3062; accessed June 1, 2007].

Searle, J.R. (1969), Speech Acts. London: CambridgeUniversity Press.

Sherry, Jr., John F. and Mary Ann McGrath (1989),“Unpacking the Holiday Presence: A ComparativeEthnography of Two Gift Stores,” in InterpretiveConsumer Research, Elizabeth C. Hirschman, ed.148–67.

Siehl, Caren, David E Bowen, and Christine M Pearson(1992), “Service Encounters as Rites of Integration:An Information Processing Model,” OrganizationScience, 3 (November), 537–55.

Slogan:Define (google search), [http://www.google.com/search?hl=en&defl=en&q=define:SLOGAN&sa=X&oi=glossary_definition&ct=title, June 27, 2007].

Smith, Barry (1990), “Towards a History of Speech ActTheory,” in Speech Acts: Meanings and Intentions:Critical Approaches to the Philosophy of John R.Searle, A. Burkhardt, ed. Berlin/New York: deGruyter, 29–61.

Wallendorf, Melanie and Eric J. Arnould (1991), “WeGather Together: Consumption Rituals of Thanks-giving Day,” Journal of Consumer Research, 18(June), 13–31.


Winograd, T. and F. Flores (1986), Understanding Com-putation and Cognition: A New Foundation for De-sign. New York: Addison-Wesley.

www.fourthmeal.com, accessed June 27, 2007.www.haircuttery.com, accessed June 30, 2007.www.saturnfans.com, accessed June 22, 2007.

For further information contact:Cele C. Otnes

Department of Business Administration350 Wohlers Hall

University of Illinois at Urbana-Champaign1206 S. Sixth St.

Champaign, IL 61820E-Mail: [email protected]


COMMUNICATION OF PRICE INCREASES: HOW CAN NEGATIVE

CONSUMER REACTIONS BE REDUCED?

David M. Woisetschläger, University of Dortmund, GermanyHeiner Evanschitzky, University of Strathclyde, United Kingdom

Hartmut H. Holzmüller, University of Dortmund, Germany

SUMMARY

Price increases seem to be an adequate way to im-prove the earnings of companies. This fact becomesespecially crucial because of an increased price competi-tion in many markets. Price increases might lead to nega-tive customer reactions such as a lower perceived utility ora lower loyalty intention. Therefore, the question formanagers remains how prices can be increased withoutlosing customers. Results of our experimental study sug-gest that customers of energy suppliers rate the perceivedutility of the offer relatively better, when the price in-crease is combined with an additional modification of theproduct or accompanied by a new service.


In retailing and behavioral pricing literature mostlyprice reductions are discussed and very little research hasbeen devoted to price increase strategies (e.g., Homburg,Hoyer, and Koschate 2005; Homburg, Koschate, andHoyer 2005). Managers need to decide through whichstrategy price increases are least harmful in terms ofnegative reactions from consumers. When customers ex-perience an increase in the outcome for the supplier asconsequence of a price increase, they perceive a reducedutility. This gap in perceived utility increases, the biggerthe difference is. The perceived inequity produces nega-tive affective states that motivate people to take actions toreestablish equity (Adams 1963, 1965; Ajzen 1982; Aus-tin and Walster 1975). Therefore, we can assume thatperceived inequity affects behavioral intentions nega-tively, because consumers may change their input toreestablish equity. It can be assumed that (1) passing onthe responsibility for the price increase to a third party, (2)a modification of the product offer, and (3) an additionalinnovative service offer are suitable instruments to reducethe perceived inequity due to a price increase.

Methodology

An experimental study was conducted, examiningthe impact of communication of an external cause, amodified product, and an additional service combinedwith a price increase on perceived utility and loyaltyintentions with a 2x2x2 between-subjects design. Thebetween-subjects factors were the communication of an

external reason for the price increase (two levels – com-municated vs. not communicated), a modified product(two levels) and an additional service (two levels). Theexperimental conditions were manipulated with scenariosthat were sent out by means of a written questionnaire tocustomers of a large energy supplier. We decided againstusing a student sample because we want to ensure externalvalidity of our results by using real customers of an energysupplier. Therefore, only subjects between the age of 40and 60 who currently have a contract with the particularsupplier were considered in the study. The sample for theexperiment comprises 254 adult subjects, equaling a re-turn rate of 9.5 percent. Thirty nine-point eight percentfemales and 60.2 percent males between 40 and 60 yearsof age (average age = 50.33 years) participated in thisstudy. The dependent variables are measured using seven-point Likert-type scales, which range from “1” = “stronglyagree” to “7” = “strongly disagree.” Utility was measuredwith three items stating “the new offer is attractive,” “thenew offer is better than the old offer,” and “the change inthe product offer is positive.” The measure of repurchaseintention incorporates three items, ”I would intend to staya customer of ENERGY Inc.” “I would not search for analternative energy supplier,” “I would not boycott EN-ERGY Inc.” All scales display good reliability, and dis-criminate validity is given as well.

Results

Results of MANCOVA show that all direct effectsexcept the external reason are significant as well as twofirst-order interaction effects. The second-order interac-tion effect is not significant. The follow-up ANCOVAsmore clearly show a significant interaction effect of exter-nal reason and product as well as a significant interactioneffect of external reason and service for “utility.” None ofthe effects is significant for the dependent variable “loy-alty.” The interaction effect between external reason andpersonalized tariff on utility shows that the utility of theoffer is rated most favorably if the price increase iscombined with the personalized tariff or it is combinedwith both the tariff and the external reason. Communica-tion of the external reason only leads to a significantlyworse evaluation of the utility, in comparison to the offerof the personalized service. Hence, we see an overallsupport for a positive influence of the product manipula-tion on the perceived utility, whereas the external reason


alone does not show a significant effect. The interactioneffect between external reason and service on utilityindicates that the additional service offer is evaluatedsignificantly better than the sole communication of theexternal reason. There is no benefit of combining theexternal reason and the additional service offer. Custom-ers evaluate additional offers (product, service) morefavorably and rate the perceived utility of the offer rela-tively better than when no reason for the price increase isgiven.

Implications

The above findings suggest that the implementationof a price increase in conjunction with appropriate instru-ments (e.g., product and service offers) might reduce the

risk of negative consumer reactions. It could be demon-strated that the additional value offered by a productchange or additional service offer may compensate to anextent the decreased utility due to a price increase. Resultsindicate that companies should, however, avoid blamingexternal reasons alone for the price increase, since this isnot seen as a plausible reason by the customers. The factthat loyalty intentions are not affected by the manipula-tions is two edged. On the one hand, this could be anindicator for loyalty that is resistant against short termchanges in pricing. On the other hand, existing or per-ceived switching barriers could be due to the fact thatcustomers are contractually tied and not emotionallyconnected to their existing energy suppliers. Referencesare available upon request.

For further information contact:David M. Woisetschläger

Service ManagementUniversity of Dortmund


Phone: +49.231.755.4611Fax: +49.231.755.3271



MEDIA COMPETITION FOR NATIONAL ADVERTISING IN

THE ERA OF NEW MEDIA, 1997–2006

John Dimmick, Ohio State University, ColumbusOsei Appiah, Ohio State University, Columbus

Matthew S. Eastin, University of Texas at Austin

SUMMARY

The authors are grateful to Robert J. Coen, SeniorVice President, Universal-McCann for providing the ad-vertising expenditure data used in this paper.

Since the beginning of what we call in this paper the“newer media” era, 1997–2006, media channels availableto consumers have increased significantly. Beginning inthe mid 1990’s, Internet, cable channels, and satelliteradio services have added complexity to the media land-scape. This abstract seeks to ascertain the effects ofcompetition between newer and older media for advertis-ing revenues, beginning in 1997, the first year for whichInternet advertising revenues are available. The diffusionof cable and satellite TV and the evolution of the Internetmay have made an impact during the past decade onadvertising spending in the U.S. The framework contain-ing the set of constructs appropriate for analyzing suchcompetition is the theory of the niche. Our study focusedon media purveying information or entertainment con-tent, but we will include advertising -only media such asyellow pages or direct mail in our analyses where appro-priate.

The Niche

The theory of the niche (Dimmick 2003) states thatwhen a new medium – sometimes called an invader –enters a guild of industries which use the same resourcescompetition for resources become more intense. Themedia compete for a number of resources including ad-vertising, the subject of this paper, as well as for consumertime and attention and consumer dollars. If resourcesincrease when an invasion occurs, the result will be theindustries continue to coexist without mutual harm. Ifresources do not increase, the result is intensified compe-tition which may eventuate in displacement or exclusionof one or more industries. In a situation of constrainedresources the new, if they succeed, succeed at the expenseof the old. Displacement is the appropriation of someresources formerly utilized by an older industry by a newcompetitor. Exclusion is the extinction of one or morecompetitors. In the history of media industries, displace-ment is the common result of the entry of a new competi-tor. Based on this, we ask the following:

RQ1: Has the rise of the newer media, cable and theinternet, resulted in the displacement of older me-dia on the national advertising resource dimen-sion?

In the case of advertising resources (McCombs 1972),researchers suggest the state of the national economy is astrong influence on the availability of advertising dollars.McCombs found what he called “relative constancy” inadvertising spending, that advertising represented a con-stant proportion or percentage (around 2%) of the GNP.Thus, this body of research leads to the second researchquestion:

RQ2: Has the growth of the national economy in the lastdecade, when both cable and the internet weregrowing, resulted in a greater number of dollarsbeing allocated to advertising and, if so, to whichmedia have these dollars been allocated?

Method

Data were obtained from Robert J. Coen, Senior VicePresident of Universal McCann. Universal McCann isresponsible for compiling advertising expenditures onmedia. The media analyzed in this paper include oldermedia such as newspapers, network radio, network TVand magazines as well as newer media, cable and theInternet.

Results

Addressing RQ1, time-series regression analyses wereperformed using both internet and cable network shares asthe independent variables and magazine and network TVshares individually as the dependent variables. This dis-placement analysis indicates that while the internet andcable networks have played a role in the decline ofnetwork TV shares, only cable is related to the downturnin magazine shares. Hence, the results indicate that bothcable and the Internet have had a significant displacingeffect in the last decade on TV networks share of nationaladvertising while cable has had a significant displacinginfluence on magazine shares. Addressing RQ2, the firststep in answering the question was to replicate McCombs’(1972) analysis for national advertising. Data indicate the


trend for national advertising as a share of GDP is flat overthe period 1997–2006, hence, advertising as a share ofGDP does not increase during the decade under study.Further, the majority of the “new” national advertisingdollars made available by economic growth have beenspent on cable networks and the internet, although maga-zines also show a gain in constant dollars. Apparently,economic growth during the period of strong growth ofcable and the internet has cushioned the effect of the cablenetworks and the internet on magazines and networktelevision in constant dollars. The share displacementcoupled with economic growth has resulted not in fewer(constant) dollars allocated to network TV but, instead,has allowed network TV budgets to remain rather stable.

Discussion

While the internet and cable networks demonstrateda statistically significant impact on network television, apattern consistent with displacement, only cable wasfound to significantly demonstrate potential displacementpatterns with magazines. Additional analyses demon-strated that while economic growth between 1997 and2006 allowed network television advertising expendi-tures to remain stable, the competition for new ad dollarswas won by newer media. This suggests that if the economystalls, media such as network television may be faced withfewer constant dollars. References are available uponrequest.

For further information contact:John Dimmick

Ohio State University154 N. Oval Mall3045A Derby Hall

Columbus, OH 43210Phone: 614.292.0168



EXPLORING THE MARKET STRUCTURE OF ONLINE AUCTIONS

Stephen France, Rutgers University, NewarkJ. Douglas Carroll, Rutgers University, Newark

SUMMARY

When developing and marketing a product, it is ofgreat use to know how the product relates to the otherproducts in the marketplace. One may want to know howclose a product is to the competing products in the market.This closeness can be modeled in terms of “distances”between items, and then using the distances to create somesort of visualization of the relative brand positions. Alter-natively one could group the most closely related productsinto clusters or latent segments, where competing prod-ucts are assigned so that the products competing mostclosely with one another are in the same group. Productsare not the only entities that can be segmented; e.g., we cancreate latent classes of consumers, attempting to groupconsumers with the most homogeneous product prefer-ences/purchase patterns.

Market structure can be modeled from a variety ofsources, including perceptual data, behavioral data, andproduct attributes. Perceptual data will typically be gath-ered using questionnaires or a user panel; information canbe elicited and recorded in different ways, including directpreferences between brands, comparing brands on differ-ent attributes, and using a Likert scale questionnaire formeasuring latent attributes. Behavioral data will typicallybe gathered from Point of Sale (POS) purchase data, fromwhich we can derive brand loyalty and brand switchingdata.

Both the gathering of behavioral data and the gather-ing of perceptual data are expensive in terms of time andmoney. A retailer may have data from POS terminalsavailable and have some method (i.e., a loyalty cardscheme) of tracking customer purchases. A manufacturermay not have available purchase information across com-peting brands, and may have to purchase such informationfrom a data provider such as AC Nielsen, who gather datafrom multiple retailers. The uses of behavioral data todetermine market structure is usually limited to highvolume, frequent purchase items; academic examplesinclude coffee and frozen foods. Given a high value, rarepurchase items such as a car, it is not feasible to usepurchase information to model market structure. Yearsworth of historical purchase information would be re-quired to get information for multiple purchases by asingle consumer, and a consumer’s purchase intensionand behavior may have changed drastically in the inter-vening period between purchases. The gathering of per-ceptual data, usually via marketing research, requires

great effort and expense to run the surveys or panels, andonly a small number of consumers can participate, possi-bly leading to a small, unrepresentative sample of con-sumers.

Market structure research typically concentrates onthe sale of new products, i.e., the primary market. Formany products, typically high value, infrequent purchaseitems, there is an important resale or secondary market.Online auctions provide an important part of the second-ary market.

In this paper we propose a method of using onlineauction data to determine market structure. By usingavailable auction data we create a visual product mapping.We concentrate on the visualization of market structure;we also briefly discuss how other types of structure can bedetermined. By using freely available auction data thecost of obtaining market structure analyses is much lowerthan when perceptual or behavioral data are used. Aspreviously described, it is difficult to analyze primarymarket structure for infrequent purchase items. Even ifprimary market structure data are available (i.e., newproduct purchases data or a questionnaire on productpreferences), the secondary data provides valuable extrastructure information. In this paper we concentrate onauctions processed through one particular website, that ofeBay.com. eBay.com has the largest auction volume ofany online auction website. We choose a particular highvolume product class, that of mobile phones, on which todemonstrate the analysis of market structure.

For our analysis of auction data, we explore the ideaof bidding patterns for brands. We define a similarityrelation between each pair of brands. The relation issymmetric and we do not calculate same-brand similarity,so if there are brands 1, . . .,k then there are k×(k-1)/2relations. We define S as a k×(k-1)/2 lower triangularsimilarity matrix and denote sij as the similarity relationbetween brand i and brand j. If a user bids on two itemsthen these items are both in his or her consideration set.We consider this to be a measure of similarity between theitems. This is analogous to the idea of similarity usedwhen analyzing brand switching matrices. Rather thansimilarity between two brands being defined by a con-sumer purchasing these brands in subsequent periods,similarity is defined by a consumer bidding on thesedifferent brands within a certain time period. If a user bidson a pair of products then we add one point of similarityto those pairs of items. We then normalize the similarities


relative to the expected similarities given independentheterogeneous probabilities of bidding for different brands(i.e., there is no association pattern). The normalizedsimilarities and be converted into dissimilarities and thus

distances. The resulting distance matrix can be used asinput to a multidimensional scaling procedure, or anyother procedure that requires a distance matrix (e.g.,clustering, latent class analysis etc.).

For further information contact:J. Douglas Carroll

Management and Psychology MarketingMEC 125

Rutgers University111 Washington Street

Newark, NJ 07102Phone: 973.353.5814



EXPLORING CONSUMER MOTIVATIONS FOR CREATING

USER-GENERATED CONTENT

Laura F. Bright, University of Texas at AustinTerry Daugherty, University of Texas at Austin

Matthew S. Eastin, University of Texas at Austin

SUMMARY

Over the last several decades, the media landscapehas evolved into a complex and dynamic conglomerationof both traditional and interactive media that seek to servethe needs of today’s fast-paced lifestyles. While tradi-tional media are struggling under the weight of increasedsegmentation, the interactive environment has shown thecapacity to capitalize on this fragmented market by offer-ing niche media vehicles that give consumers a voiceamidst the whirlwind of information and advertising. Inthe online world, these niche media markets are beingdriven increasingly less by publishers and more so byuser-generated content (UGC) (Morrissey 2005). UGCrefers to media content that is created or produced by thegeneral public rather than by paid professionals and isprimarily distributed on the Internet. Examples of promi-nent websites and web-based applications that support thecreation and consumption of UGC include, but are notlimited to, YouTube, Facebook, Wikipedia, Flickr,Blogger, and personal Web pages.

Beginning in 2004, with the explosion of the Web 2.0market, UGC has formed a plethora of markets within themedia landscape, attracting more than 69 million usersand generating $450 million plus in advertising revenues(Verna 2007, p. 2). To date though, little research has beenundertaken on UGC and its perceptions amongst both theconsumers and creators of such content. This exploratorystudy is designed to test the core relationship betweenconsumer attitudes and behaviors toward experiencingUGC. Specifically, we sought to confirm the relationshipbetween attitude and behavior within the realm of UGC aswell as identify the motivational functions for creatingUGC.

Method

A survey was conducted using an online panel withdata was gathered from 325 participants solicited over aseven-day period. The panel is an opt-in subject poolrecruited for Web-based research. One thousand emailswere sent on day one, 1500 emails on day three, and 500were sent on day five. The survey was closed once 325completed surveys were recorded (day seven). The cur-rent study was designed as a purposive sample based onsample size, and thus, a response rate was not calculated

as the survey was closed after being open for the desig-nated duration. Nonetheless, the completion rate, which isdefined as those who completed the survey divided bythose who accessed it, was 77 percent.

Results

A series of regression analyses were undertaken todetermine if a relationship exists between attitude andbehavior within the realm of the creation and consump-tion of UGC. Hypothesis 1 predicts that a consumer’sattitude toward UGC would be positively related to theirconsumption of UGC, i.e., attitude would predict behav-ior. In this case, a regression analysis proved to be signifi-cant in that a consumer’s attitude toward UGC success-fully predicts their experiencing of UGC (β = .37, t(71) =7.52, p < .01, R2 = .15). Therefore, this data furtherconfirms the positive relationship that exists betweenattitude and behavior as it can be applied to UGC. Hypoth-esis II posits that a consumer’s attitude toward UGC willmediate the relationship between their consumption andcreation of UGC. To determine whether attitude mediatesthe relationship between the consumption and creation ofUGC, a mediation analysis was conducted, as specified byBaron and Kenny (1986). The first analysis indicated thatthe type of experience positively influenced exposure toUGC (β = .33, t(71) = 2.93, p < .01, R2 = .15). The secondanalysis confirmed that the type of experience positivelyinfluenced a consumer’s attitude toward UGC (β = .34,t(71) = 3.02, p < .01, R2 = .11). Finally, the third analysissupports attitude as a mediating variable between theconsumption and creation dimensions for UGC (β = .25,t(2,68) = 2.01, p < .05, R2 = .16). Accordingly, the effectof consumption on the creation of UGC weakens whenincluded in the analysis with attitude (β = .23, t(2,68) =1.88, p > .05) indicating attitude indeed serves as amediator in the relationship. Hypothesis III predicts that aconsumer’s attitude toward UGC would vary by theirfunctional source of motivation for the creation of UGC.Multiple regression was employed to test this relationshipwithin the realm UGC and found to be significant(F(5,70) = 5.19, p < .01; R2=.29), supporting the hypoth-esis. Examining the contribution of each individual func-tion though revealed significant relationships for onlythree out of the five motivational sources. Specifically,creators of UGC rely predominantly upon the ego-defen-sive function (β = .42, t(5,70) = 2.38, p < .05) and social


function (β = .34, t(5,70) = 2.54, p < .01) as motivationalsources of influence in forming their attitude towardUGC. In contrast, a negative relationship was identifiedwith the value-expressive function (β = -.43, t(5,70) =-2.89, p < .01).

Conclusions

With the consumption and creation of UGC becom-ing more prevalent, understanding why consumers aredrawn to consume and create such content becomesincreasingly important. As such, understanding people’sattitudes toward UGC, and especially the functional sourcesof their attitudes, will result in a better model for predict-

ing behavior, which is increasingly important to bothscholars and industry professionals interested in UGC.The findings of this research were successful in connect-ing functional theory and the subsequent sources of mo-tivation to attitudes toward creating and consuming UGC.Opportunities abound for both advertisers and marketersin this burgeoning information space as evidenced by aforecasted $4.3 billion in advertising revenues by 2011.Consumers are increasingly in control of their medialandscape with UGC becoming a larger part of the puzzle.As such, marketers must seize the opportunity to commu-nicate relevant content to niche audiences via this chan-nel. References are available upon request.

For further information contact:Terry Daugherty

Department of AdvertisingUniversity of Texas at Austin1 University Station A1200

Austin, TX 78712Phone: 512.471.8917



AN INNOVATIVE DATA COLLECTION METHOD FROM THE

INTERNET FOR MARKETING RESEARCH:

COMPUTER SPIDERS

Yanbin Tu, Robert Morris University, Moon Township

SUMMARY

Data collection is a fundamental step in marketingresearch. For certain marketing research projects, particu-larly, those which are empirical, data collection is criticalfor testing research hypotheses, and/or validating propo-sitions. Traditional methods for data collection includesurveys, data recordings, observations, and experiments,etc. Among these, surveys and data recordings are themost popular. Surveys include personal interviews, tele-phone interviews, mail surveys, self-administrated sur-veys, fax surveys, e-mail surveys, and online surveys.Data recordings refer to the processes of collecting busi-ness transaction data over time, such as sales, costs,market shares, etc. Collecting and analyzing businesstransaction data is crucial for understanding consumerbehavior, measuring marketing effectiveness, and fore-casting market demands, which ultimately enhance theeffectiveness and efficiency of marketing strategies.

As the Internet is widely used, and e-commerce isembraced by businesses and individuals, more businessactivities and consumer behaviors are carried out in thevirtual environment of the Internet. Correspondingly,more business transactions and consumer data (generally,e-commerce data) is being generated online. The Internetand information systems allow businesses to accuratelyrecord business transactions and consumer behavior datain real-time. Unlike online experiment data, the e-com-merce data is actual business transaction data without thepresence of data collection design bias.

However, we are now facing new challenges in theinformation age – collecting and taking advantage of theoverwhelming e-commerce data online. Although re-searchers in computer science and information systemsunderstand information technology (IT) very well, theyoften do not understand business and marketing lan-guages. On the other hand, marketing researchers and

educators realize the existence of huge amounts of valu-able e-commerce data online for many potential researchtopics and marketing education, but they are unlikely to befully equipped with sufficient IT knowledge to collectthese data-sets, or train their students to do that. As far aswe know, little research has been done about how toeffectively and efficiently collect e-commerce data fromthe Internet in marketing research.

In this study, we introduce an innovative tool foronline data collection to marketing researchers and educa-tors. We wish to share our experience in creating andusing computer spiders to collect online e-commerce data.At least two primary reasons may be given for collectinge-commerce data online: (1) collecting data for currentusage in marketing research projects to test researchhypotheses, and validating marketing theories; and (2)collecting online data that is temporarily available forfuture usage in potential marketing research. We empha-size the effectiveness and efficiency in data collectionsince, even though the data is freely available online, theworkload associated with manual data collection might besubstantial. For example, every day, millions of onlineauctions take place on eBay.com and Amazon.com. Manu-ally collecting the data would be time-consuming and, attimes, impossible. The spiders discussed in this study arepowerful tools for collecting online data. Within a fewminutes, the spiders can collect thousands of online auc-tion lists. These thousands of auction lists might require aresearcher to spend a month collecting them manually.We believe that our approach for online data collectionthrough computer spiders can promote marketing re-search, especially for research projects involving e-mar-keting or research projects requiring huge data-sets suchas Web-based database marketing studies. For marketingeducation, we believe that using spiders for data collec-tion in class will benefit students in their data collectionskills.

For further information contact:Yanbin Tu

Department of MarketingSchool of Business

Robert Morris UniversityMoon Township, PA 15108

Phone: 412.397.4261 ♦ Fax: 412.262.8672E-Mail: [email protected]


A MEASUREMENT SCALE FOR CUSTOMER CO-PRODUCTION

(CCP) OF SERVICES

Mohammad Ali Zolfagharian, University of Texas – Pan American, Edinburg

SUMMARY

It is now widely recognized that customers play anactive role throughout the marketing process (Vargo andLusch 2004; Wikstrom 1996a), defined as activities thatmake possible the conception, design, production, deliv-ery, purchase, consumption, and disposal of offerings.Value is co-created by customers, marketers, and otherstakeholders during these stages (Wikstrom 1996b). Thisresearch developed a culturally informed, reliable, andvalid scale to measure Customer’s Co-Production (CCP)role in the production and delivery stages.

Marketing literature contains multiple scales tappingconcepts that are similar to CCP. The closest conceptual-ization is incorporated in the servuction framework(Langeard et al. 1981). Since all of these scales consist offormative measures, there is an opportunity to develop anew CCP scale that (a) is culturally informed; (b) includesformative as well as reflective measures; and (c) could beapplied in various service industries with minimal modi-fication. Starting from the extant scales, this researchemployed ethnography, EFA and CFA in three servicesincluding self check-out at grocery stores, toy assemblyworkshops, and meal assembly centers.

The ethnographic project started with semi-distancedparticipant observation and shadowing of customer expe-riences during naturally occurring service encounters(Kvale 1983). After these research activities were com-pleted, a brief analysis of the data necessitated a set of in-depth interviews and focus groups. The chief result ofethnography was a 5-dimension, 19-item CCP scale.Since almost all of the participants alluded to a frame ofreference when discussing their CCP role, the scale in-volved a comparison between two comparable versions ofthe same service core. Accordingly, the 19 items wereanchored on a 5-point semantic differential scale repre-senting a lot less, somewhat less, to the same extent,somewhat more, and a lot more. The comparison pointsfor self check-out, toy assembly workshops, and mealassembly centers were regular check-out, regular toystores, and frozen meals, respectively.

Data for EFA was collected from 50 graduate and 50undergraduate students per service. The instrument, whichwas customized per service, included (a) the 19-item CCPscale; (b) the 25-item version of “self-empowerment”

scale (Bramucci 1977); (c) the 9-item “authoritarian per-sonality” scale (Adorno et al. 1950); and (d) the 19-itemsubmissive patient behavior. Relying on the reviewedliterature, it was posited that the CCP scale would corre-late significantly positively with the self-empowermentand authoritarian scales and significantly inversely withthe modified submissive scale. These posits helped exam-ine construct and nomological validities of the CCP scale(Hair et al. 1998).

In all three services, EFA identified five underlyingdimensions and confirmed the findings of the ethno-graphic investigations. Each of the 19 items showed highloadings (.78–.93) on one of the five factors and lowloadings (< .43) on the other four factors. Across the threeservices, the patterns of loadings and communalities ex-hibited adequate stability; and Cronbach α’s ranged be-tween .88 and .96. Therefore, the five CCP dimensionsenjoyed a sufficient degree of reliability. All of the 19items were retained.

The average of each set of concentric scale items wascomputed and a complete correlation matrix was pre-pared. Since the correlations between each two factorsunder the same construct were significantly lower thancorresponding factor reliabilities represented by α values,the convergent validity of each construct and its dimen-sions was reasonably satisfied (Churchill 1979). Compar-ing within-construct correlations to inter-construct corre-lations indicated that the former were significantly largerthan the latter. This pattern held for all of the within- andinter-construct comparisons across the three services.Therefore, the CCP dimensions had acceptable discrimi-nate validity (Churchill 1979). The correlation of CCPwith self-empowerment, authoritarian personality, andmodified submissive scales were .76, .69, and -.71, pro-viding support that the CCP construct as construed andmeasured here enjoys nomological validity within thenetwork of related constructs.

Data for CFA was collected from 218, 199, and 187professional adults. The resultant λ, δ, and φ estimates andtheir respective standard errors were all in order, and eachof the three CFA runs exhibited an acceptable fit. CFAfindings were generally in agreement with those in EFA.In conclusion, these empirical investigations provide sup-port that the CCP scale is a culturally informed, reliable,and valid measurement tool.


For further information contact:Mohammad Ali Zolfagharian

University of Texas – Pan American1201 W. University Drive BUSA 222A

Edinburg, TX 78541–2999Phone: 956.381.3389



CUSTOMER CO-DESIGN COMPETENCE: MODEL DEVELOPMENT

AND INFLUENCE ON CUSTOMER SATISFACTION WITH MASS

CUSTOMIZATION OFFERINGS

Melanie Müller, Technische Universität München, GermanyRalf Reichwald, Technische Universität München, Germany

Frank Piller, RWTH Aachen, Germany

ABSTRACT

Mass customization is a concept that describes theproduction of goods and services to meet an individualcustomer’s needs. The production and provision of thecustom solution takes place with an efficiency that can becompared with that of mass production. From the custom-ers’ perspective, the defining element of masscustomization is the interactive sales and co-design pro-cess. Within this process customer and company (repre-sented by a salesperson) create or specify the individualcustomer solution. Based on research from the area ofservice marketing, we argue in this paper that the qualityof fulfillment of the co-design task of a customer is highlydependent on the competence of the particular customerperforming this task. We assume that this customer co-design competence comprises four facets: the customers’product and application competence, social skills, meth-odological knowledge, and personal motivation. Basedon a large-scale customer survey in four industries (517respondents) we show that customer co-design compe-tence is multi-dimensional and is an important antecedentof perceived service quality and customer satisfactionwith mass customization offerings.

INTRODUCTION

Mass customization is a concept that describes theproduction of goods and services to meet an individualcustomer’s needs. The production and provision of thecustom solution takes place with an efficiency that can becompared with that of mass production (Tseng and Jiao2001; see also Duray 2002; Piller 2003; Pine 1993;Rangaswamy and Pal 2003). This degree of efficiency ispossible solely because all operations are conducted withina fixed solution space characterized by stable, yet flexibleand responsive processes. As a result, the costs associatedwith customization allow a price level that does not implyswitching to an upper market segment. A masscustomization strategy seems to be promising in variousmarkets, as many of today’s customers want to expresstheir personality, e.g., by means of a custom product, andbecause of their simultaneous price consciousness (seee.g., Prahalad and Ramaswamy 2004).

This paper addresses the mass customization conceptfrom the perspective of the customer who experiences theinterface of the mass customization sales system by par-ticipating in a co-design process. This process is thedefining element of a mass customization strategy (Piller2003); it allows customers to configure their customproduct from a list of pre-defined options and components(Tseng and Du 1998; von Hippel 1998). The customer co-design of (tangible) products corresponds to a phenom-enon that has been previously discussed in service mar-keting and B2B literature, but it is new for consumerproduct markets: the active integration of customers in thesupplier’s value creating activities (“customer integra-tion,” Prahalad and Ramaswamy 2004). Co-design activi-ties are performed in an act of interaction and cooperationbetween the customer and the sales system of the productprovider (Franke and Piller 2003, 2004; Franke andSchreier 2002; Khalid and Helander 2003; von Hippel1998).

Customer related aspects of mass customization of-ferings have only recently been addressed in the literature(see Dellaert and Stremersch 2005; Franke and Piller2004; Huffmann and Kahn 1998; Liechty et al. 2001;Wind and Rangaswamy 2001). Therefore, we argue inthis paper – based on research from the area of servicemarketing – that the quality of fulfillment of the co-designtask is highly dependent on the skills and competence ofthe particular customer performing this task. We assumethat this customer co-design competence comprises fourfacets: the customers’ product and application compe-tence, social skills, methodological knowledge, and per-sonal motivation. Using data (customer surveys, 517respondents) from four different mass customization pro-viders, we show that customer co-design competence ismulti-dimensional and is an important, not yet recognizedantecedent of perceived service quality and customersatisfaction with mass customization offerings.

After the description of the underlying principles ofmass customization in the introduction, we present therelevant theoretical background. In a next step, we buildon our understanding of customer co-design competence.The results of the construct validation process and the


influence of customer co-design competence on the per-ceived service quality and customer satisfaction are thenelaborated on. The paper concludes with fundamentalimplications for theory and practice derived from ouranalysis.

LITERATURE REVIEW AND MODEL

DEVELOPMENT

In service marketing, company-to-customer interac-tion and cooperation during the co-design process isintensively discussed (see e.g., Bettencourt 1997; Bitneret al. 1997; Bowen 2000; Honebein 1997; Mills andMorris 1986; Gouthier 2003). Authors agree that custom-ers contribute to the service and influence the success ofthe offering (see e.g. Honebein 1997; Mills et al. 1983).Several authors emphasize the importance of the cus-tomer as an important resource of the company in thiscontext (Bowen and Schneider 1985; Canziani 1997).Bitner et al. (1997), for example, call the customer a“productive resource.” The same argumentation is rel-evant for the mass customization context: because cus-tomers are deeply involved in the co-design process oftheir individual product, they can be regarded as a re-source of the firm. The supplier has to make sure that itscustomers are prepared to properly fulfill the co-designprocess. They must be able to provide the necessaryinformation in regard to their customization demands,transfer this demand to a product specification, and also beinformed about the possible solution space of the offer-ings and the range (and constraints) of the configurationpossibilities available. Furthermore, customers need tohave certain social and/or methodological skills and mo-tivation if they are to make successful contributions to theprocess.

We use the term “competence” to refer to a consumer’sknowledge, skills, and motivation relevant for the suc-cessful provision of a mass customization offering (for thedefinition of the term competence, see e.g., Gouthier2003). Based on the understanding that competence is anindividual factor, Kaiser (1982) describes competence asthe ability of a person to solve a certain problem. Follow-ing this understanding, we define customer co-designcompetence as the competence of a customer to performco-design tasks. Using a framework from service market-ing (Gouthier 2003), we distinguish between four dimen-sions of customer co-design competence:

• Product and application competence is based oncustomers’ knowledge of and experience with theoffering. This facet includes knowledge about theproduct, as well as knowledge that supports custom-ers in successfully dealing with the requirements ofthe co-design process. We also include product andprocess experience in this dimension.

• The facet social skills comprises customers’ skills inregard to the task, respectively social skills to interactwith the employees of the mass customization pro-vider.

• Methodological competence is a secondary skillscategory. This includes the skill to use technicalequipment, e.g., a configurator.

• The facet personal motivation comprises all aspectsthat motivate the consumer to participate in the co-design process, e.g., product and purchase involve-ment or shopping enjoyment. We are aware that theconsideration of motivation into competence modelsis controversial; however, consider it as importantbesides knowledge, experience and skills.

We regarded perceived service quality, customersatisfaction with the individual product and customersatisfaction with the supplier as dependent variables inour study, as a customer’s integration competence mighthave direct consequences for both aspects. Service qualityis defined as the customers overall impression of therelative inferiority/superiority of the organization and itsservices (Bitner and Hubbert 1994). Satisfaction is de-fined as the result of a customer’s comparison of expecta-tions and experiences (cf., e.g., Oliver 1980, 1981;Olshavsky and Miller 1972). A consumer perceives satis-faction if a given standard is reached or exceeded by theconsumer’s actual experiences (see e.g. Oliver 1980,1981; Westbrook and Reilly 1983). In this study customersatisfaction is considered as a consequence of perceivedservice quality and is influenced by all the experiences acustomer has with the company and its offering: product,sales process, after sales services or other experiences(Anderson and Fornell 1994; Giering 2000; Homburg andGiering 2001; Yi 1990). We assume that customers witha higher degree of competence, i.e., more product andapplication competence, higher social and methodologi-cal skills, and a higher personal motivation, perceive abetter quality and are more satisfied than consumers withless competence. This leads to an initial set of threehypotheses:

H1: The higher a customer’s integration competence, thehigher the perceived service quality is.

H2: Higher customer co-design competence leads to morecustomer satisfaction with the customized product.

H3: Higher customer co-design competence leads to ahigher customer satisfaction with the supplier.

Furthermore, we considered one antecedent of acustomer’s integration competence on the consumer sidethat seems especially important in the context of mass


customization offerings: the desire for unique consumerproducts. It can be described as consumers’ motivation tofind offerings that truly meet their needs and to differen-tiate themselves from others, for example via the productsthey wear (Lynn and Harris 1997; Snyder and Fromkin1977; Tian, Bearden, and Hunter 2001). Consumers’desire for unique products influences their willingness toadopt new products and processes like interacting in a co-design environment. We presume that consumers with agreater desire for unique products are more likely to havea higher level of integration competence, and thus proposethe following hypothesis:

H4: The greater a consumer’s desire for unique consumerproducts, the higher the customer codesign compe-tence is.

Furthermore, we assume a relation between servicequality and customer satisfaction:

H5: The higher a customer’s perceived service quality,the higher the satisfaction with the custom product.

H6: Higher perceived service quality leads to more satis-faction with the supplier.

H7: The higher the satisfaction with the custom product,the higher the satisfaction with the supplier.

MEASUREMENT DEVELOPMENT

Based on our definition of customer co-design com-petence, we conceptualized and operationalized this con-struct using existing scales from the literature. The fourfacets of customer co-design competence are the anchorof our theory-driven approach. However, we conductedexploratory research because existing scales had to beadapted for the special case of a mass customizationoffering (see next chapter). Table 1 summarizes all con-sidered scales from literature used in our study (see alsoAppendix 1 for a description of the different factors).

Perceived service quality and satisfaction as depen-dent variables, consumers’ desire for unique consumerproducts as an antecedent.

As mentioned above, service quality and customersatisfaction are the dependent variables in our study. Tomeasure service quality, we used six items of the scale ofParasuraman, Zeithaml, and Berry (1988, we reduced theinitial set of 21 items based on the results of the pre-test).We measured satisfaction with the supplier and the cus-tom-made product with four items for each aspect basedon the scale developed by Giering (2000). Consumers’desire for unique products is considered an antecedentvariable in our model by using the 8-item scale developedby Lynn and Harris (1997).

RESEARCH METHOD

Several steps were applied to develop a scale ofcustomer co-design competence. First of all, qualitativesemi-structured interviews with mass customization ex-perts were conducted in order to discuss the phenomenonof mass customization offerings and the necessary cus-tomer competences.

Next, three exploratory surveys with customers ofdifferent mass customization companies were organizedin order to consider the customer perspective of thephenomenon. The starting point of interviews and focusgroup discussions was the four-dimensional classifica-tion of customer co-design competence adapted fromGouthier (2003) and the existing scales introduced above.

Based on the literature review and the exploratoryresearch activities, an initial set of 60 indicators formeasuring customer co-design competence was devel-oped and evaluated in the item-reduction process. Asurvey with 157 students from a German business schoolwas conducted to pretest the scale. Multi-item measure-ments of the constructs were applied. All measurementswere conducted with a 7-point rating scale (“stronglyagree” and “strongly disagree” as anchors) except for thefour items for measuring the attitude toward technology.For these items, a 7-point semantic differential scale wasused. Because all of the implemented scales were origi-nally in English, they were translated to German and thentranslated back to English in order to ensure that the originalmeaning was not distorted through the translation.

By means of exploratory factor analysis, coefficientalpha and item-to-total correlation, the scale was purified.This led to a set of 45 items to measure customer co-designcompetence (plus 22 items to measure the dependentvariables and the antecedent). Based on this final item set,we conducted a survey of the customers of four real-lifemass customizers that operate in an offline (brick-and-mortar) retail environment. We surveyed all customerswho had bought a mass customized item during the lasttwo years. Our fields of research were:

• A company offering individual shirts: At first, cus-tomers choose the shirt’s color and size. Then theyselect a motif from a collection and add a personalmessage.

• A retailer offering mass-customized suits: Consum-ers are measured (using a 3D scanner) and are able toco-design a classic men’s suit according to theirfabric and cut preferences.

• A company offering custom bicycles: Customers canselect different colors and functional options for acustom mountain, trekking, city or racing bike.


Custom-made house components: Customers can config-ure custom-made winter gardens and sun blinds accord-ing to their needs.

RESULTS

We received completed questionnaires from a total of530 customers. Response rates were very satisfying andover 24 percent in all cases. Because of some missingvalues, we had to exclude 13 questionnaires, which led toa total of 517 customers who were ultimately included inthe final study. All items related to customer co-designcompetence were integrated in an exploratory factor analy-sis in SPSS 13.0 using a principal axis analysis with directoblimin rotation in order to identify the data structure andrelevant factors. The measure of sampling adequacy forthe overall data set achieved a score of 0.903 (Kaiser andRice 1974). Nine factors could be identified using factoranalysis (eigenvalue >1). Next, exploratory factor analy-sis and coefficient alpha were calculated for each factor.The results lie in all cases above the threshold value of50% (AVE) and 0.70 (Cronbach Alpha) respectively. Ina next step, a confirmatory factor analysis was conductedfor all factors using the common criteria. The calculationswere repeated for the assumed four dimensions and theoverall model. Finally, exploratory and confirmatory fac-tor analyses led to a final set of 18 items and five factors(overall model fit: GFI: 0.929; AGFI: 0.903; RMSEA:0.058; Chi-Square/df: 2.823; NFI: 0.949; CFI: 0.967):

• The factor product knowledge could be confirmed.

• The factors product experience and involvementformed one common factor called product experi-ence and involvement.

• The new factor process knowledge and experienceevolved from the two factors process knowledge andprocess experience.

• The factors need for personal interaction and com-munication style formed a new factor social skills.

• The factor purchase involvement was the secondfactor that could be confirmed.

The factor attitude toward technological productscould not be confirmed in the item reduction processwhich can be traced back to the fact that technologicaltools and systems were mainly steered by the employeesin our research setting. Furthermore, in our study shop-ping enjoyment seems to be unimportant for a consumer’sintegration competence.

In a next step, we analyzed the influence of customerco-design competence in the context of the dependentvariables and the antecedent. The influence of customerco-design competence on perceived service quality andcustomers’ satisfaction with the custom product and the

TABLE 1

Overview of Used Scales to Conceptualize Customer Co-Design Competence

(1) Product and Product knowledge (Flynn, Goldsmith 1999): 5 itemsapplication competence

Product experience (Griffin et al. 1996): 4 items

Process knowledge (Adapted by Flynn, Goldsmith 1999): 5 items (+ 1 specificitem)

Process experience (Adapted by Griffin et al. 1996): 4 items (+ 1 specific item)

Communication skills (Adapted by Williams and Spiro 1985): 6 items

(2) Social skills Need for interaction with salesperson (Dabholkar 1996): 4 items

(3) Methodological Attitude toward technology (Dellaert et al. 2004): 4 itemsknowledge

Product involvement (Laurent, Kapferer 1985): 16 items (+ 1 specific item)

(4) Personal motivation Purchase involvement (Lockshin et al. 1997): 5 items

Shopping enjoyment (Sproles, Kendall 1986): 4 items


FIGURE 1

Causal Model of Customer Co-Design Competence

supplier could be confirmed. Furthermore, our analysesshowed an influence of consumers’ need for individualproducts on customer co-design competence (overallmodel fit: GFI: 0.860; AGFI: 0.839; RMSEA: 0.053; Chi-Square/df: 2.451; NFI: 0.911; CFI: 0.945). Figure 1 showsthe relationships between the different aspects. All hy-potheses could be confirmed except of hypothesis H3.

CONCLUSION

Customer interfaces have been a source of innovativebusiness models at all times. Configuration or co-designsystems for mass customization are a recent example ofsuch an innovative sales interface; they shift developmentand design tasks from the locus of the manufacturer to thecustomer. In contrast to existing literature on masscustomization, we have focused on sales systems that arecharacterized by human interaction between a sales per-son and a consumer. The objective of this paper was to

provide a conceptual model of an important antecedentvariable of the success of operating such an interface:customer co-design competence. Our analysis has shownthat the different variables behind these models are indeedan important driver for the perceived service quality andthe overall customer satisfaction achieved through themass customization offering. There is growing anecdotalevidence and academic research on the design and successfactors of such codesign systems. However, there are stillmany questions open as to how a successful front-end formass customization can be built. Our results confirm theimportance of steering the customer’s contribution. Formass customization providers, it seems essential to regardcustomers as a necessary part of the co-design process.This includes providing them with all relevant capabilitiesand information enabling them to become a co-designer.In addition, it seems important to create an interactivesales system that also provides the consumer with anenjoyable, rewarding experience.

REFERENCES

Alba, J.W. and J.W. Hutchinson (1987), “Dimensions ofConsumer Expertise,” Journal of Consumer Research,13 (March), 411–54.

Anderson, E. and C. Fornell (1994), “A Customer Satis-faction Research Prospectus,” in Service Quality,R.T. Rust and R.L. Oliver, eds. Thousand Oaks.

Babin, B.J., W.R. Darden, and M. Griffin (1994), “Workand/or Fun: Measuring Hedonic and Utilitarian Shop-


ping Value,” Journal of Consumer Research, 20 (4),644–56.

Beatty, S.E. and S.M. Smith (1987), “External SearchEffort: An Investigation Across Several Product Cat-egories,” Journal of Consumer Research, 14 (June),83–95.

Bettencourt, L.A. (1997), “Customers Voluntary Perfor-mance: Customers as Partners in Service Delivery,”Journal of Retailing, 73 (3), 383–406.

Bitner, M.J., B.H. Booms, and M.S. Tetreault (1990),“The Service Encounter: Diagnosing Favorable andUnfavorable Incidents,” Journal of Marketing, 54(1), 71–84.

____________ and A.R. Hubbert (1994), “EncounterSatisfaction Versus Overall Satisfaction Versus Qual-ity, in Service Quality, R.T. Rust and R.L. Oliver, eds.Thousand Oaks, 72–94,

____________, W.T. Faranda, A.R. Hubbert, and V.A.Zeithaml (1997), “Customer Contributions and Rolesin Service Delivery,” International Journal of Ser-vice Industry Management, 8 (3) 193–205.

Bowen, D.E. and B. Schneider (1985), “Boundary-Span-ning-Role Employees and the Service Encounter:Some Guidelines for Management and Research,” inThe Service Encounter, John A. Czepiel et al. eds.Lexington, MA: Lexington Books 127–47.

____________ (2000), “The ‘Customer as Employee’Revisited,” QUIS-Konferenz, (June), Karlstad,Schweden.

Canziani, B.F. (1997), “Leveraging Customer Compe-tency in Service Firms,” International Journal ofService Industry Management, 8 (1), 5–25.

Clarke, K. and R.W. Belk (1979), “The Effects of ProductInvolvement and Task Definition on AnticipatedConsumer Effort,” Advances in Consumer Research6(1), 313–18.

Dabholkar, P.A. (1994), “Consumer Evaluations of NewTechnology-Based Self-Service Options: an Investi-gation of Alternative Models of Service Quality,”International Journal of Research in Marketing, 13,29–51.

____________ (1996), “Consumer Evaluations of NewTechnology-Based Self-Service Options: an Investi-gation of Alternative Models of Service Quality,”International Journal of Research in Marketing, 13(1), 29–51.

Dellaert, B.G.C., J. Goebel, and P.A. Dabholkar (2004),“Consumer Decisions to Use Online Mass Custom-ized Retailing,” forthcoming in Journal of Retailing.

____________ and S. Stremersch (2005), “MarketingMass Customized Products: Striking the BalanceBetween Utility and Complexity,” Journal of Mar-keting Research, 43 (2), 219–27

Duray, R. (2002), “Mass Customization Origins: Mass orCustom Manufacturing?” International Journal ofOperations and Production Management, 22 (3),314–30.

Flynn, L.R. and R.E. Goldsmith (1999), “A Short ReliableMeasure of Subjective Knowledge,” Journal of Busi-ness Research 46 (1), 57–66.

Franke, N. and M. Schreier (2002), “EntrepreneurialOpportunities with Toolkits for User Innovation andDesign,” International Journal on Media Manage-ment, 4 (4), 225–34.

____________ and F. Piller (2003), “Key Research Is-sues in User Interaction with Configuration Toolkitsin a Mass Customization System,” International Jour-nal of Technology Management, 26, 578–99.

_____________ and ____________ (2004), “Toolkitsfor User Innovation and Design: An Exploration ofUser Interaction and Value Creation,” Journal ofProduct Innovation Management, 21(6), 401–15.

Giering, A. (2000), Der Zusammenhang Zwischen Kun-denzufriedenheit Und Kundenloyalität: Eine Unter-suchung Moderierender Effekte (The RelationshipBetween Customer Satisfaction and Loyalty). Wies-baden: Gabler.

Gouthier, M. (2003), Kundenentwicklung Im Dienstleis-tungsbereich (Customer Development in ServiceMarketing). Wiesbaden: Gabler.

Griffin, M., B.J. Babin, and J.S. Attaway (1996), “Antici-pation of Injurious Consumption Outcomes and itsImpact on Consumer Attributions of Blame,” Jour-nal of the Academy of Marketing Science, 24 (4),314–27.

Homburg, C. and A. Giering (2001), “Personal Character-istics as Moderators of the Relationship BetweenCustomer Satisfaction and Loyalty,” Psychology andMarketing, 18 (1), 43–66.

Honebein, P. (1997), “Strategies for Effective ConsumerEducation,” Chicago: Irwin.

Huffman, C. and B. Kahn (1998), “Variety for Sale: MassCustomization or Mass Confusion,” Journal of Re-tailing, 74, 491–513.

Kaiser, P. (1982), Kompetenz Als Erlernbare FähigkeitZur Analyse Und Bewältigung von LebenssituationenAuf Mehreren Ebenen (Competence as LearnableSkill to Analyse and Manage Life Situations),Universität Oldenburg, Dissertation.

Kaiser, H.F. and J. Rice (1974), “Little Jiffy, Mark IV,”Educational and Psychological Measurement, 34,111–17.

Khalid, H.M. and M.G. Helander (2003), “Web-BasedDo-it-Yourself Product Design,” in The CustomerCentric Enterprise, M. Tseng and F. Piller, eds. NewYork und Berlin: Springer, 247–66.

Kleiser, S.B. and S.P. Mantel (1994), “The Dimensions ofConsumer Expertise: A Scale Development,” Ameri-can Marketing Association, (Summer), 20–26.

Laurent, G. and N. Kapferer (1985), “Measuring Con-sumer Involvement Profiles,” Journal of MarketingResearch, 22, 41–53.

Liechty, J., V. Ramaswamy, and S.H. Cohen (2001),“Choice Menue for Mass Customization: An Exper-


imantal Approach for Analyzing Customer Demandwith an Application to a Web-Based InformationService,” Journal of Marketing Research, 38 (May),183–96.

Lockshin, L.S., L. Spawton, and G. Macintosh (1997),“Using Product, Brand, and Purchasing Involvementfor Retail Segmentation,” Journal of Retailing andConsumer Services, 4 (3), 171–83.

Lynn, M. and J. Harris (1997), “The Desire for UniqueConsumer Products: A New Individual DifferencesScale,” Psychology & Marketing, 14 (6), 601–16.

Mason, K., T. Jensen, S. Burton, and D. Roach (2001),“The Accuracy of Brand and Attribute Judgements:The Role of Information Relevancy, Product Experi-ence, and Attribute-Relationship Schemata,” Jour-nal of the Academy of Marketing Science, 29 (3),307–17.

Mills, P.K., R.B. Chase, and N. Margulies (1983), “Mo-tivating the Client/Employee System as a ServiceProduction Strategy,” Academy of Management Re-view, 8 (2), 301–10.

____________ and J.H. Morris (1986), “Clients as ‘Par-tial’ Employees of Service Organizations: Role De-velopment in Client Participation,” Academy of Man-agement Review, 11 (4), 726–35.

Nunnally, J.C. (1967), Psychometric Theory. New York:McGraw-Hill.

Oliver, R.L. (1980), “A Cognitive Model of the Anteced-ents and Consequences of Satisfaction Decisions,”Journal of Marketing Research, 17 (4), 460–69.

____________ (1981), “What Is Customer Satisfaction?”The Wharton Magazine, 5 (Spring), 36–41.

Olshavsky, R. and J. Miller (1972), “Consumer Expecta-tions, Product Performance, and Perceived ProductQuality,” Journal of Marketing Research, 9 (Febru-ary), 19–21.

Parasuraman, A., V.A. Zeithaml, and L.L. Berry (1988),“SERVQUAL: A Multiple-Item Scale for MeasuringConsumer Perceptions of Service Quality,” Journalof Retailing, 64 (1), 5–6.

Peterson, R. and W. Wilson (1992), “Measuring Cus-tomer Satisfaction: Fact and Artefact,” Journal of theAcademy of Marketing Science, 20 (1), 61–71.

Philippe, A. and P.V. Ngobo (1999), “Assessment ofConsumer Knowledge and its Consequences: A Multi-Component Approach,” Advances in Consumer Re-search, 26 (1), 569–75.

Piller, F. (2003), Mass Customization, 3d ed. Wiesbaden:Gabler.

Pine, B.J. (1993), Mass Customization. Boston: HarvardBusiness School Press.

Prahalad, C.K. and V. Ramaswamy (2004), The Future ofCompetition: Co-Creating Unique Value With Cus-tomers. Boston: Harvard Business School Press.

Prakash, V. (1984), “Validity and Reliability of the Con-firmation of Expectations Paradigm as a Determinantof Consumer Satisfaction,” Journal of the Academy

of Marketing Science, 12 (4), 63–76.Rangaswamy, A. and N. Pal (2003), “Gaining Business

Value from Personalization Technologies,” in ThePower of One: Gaining Business Value from Person-alization Technologies, N. Pal and A. Rangaswamy,eds. Victoria: Trafford Publishing, 1–9.

Selnes, F. and K. Gronhaug (1986), “Subjective andObjective Measures of Product Knowledge Con-trasted,” Advances in Consumer Research, 13, 67–71.

Sauer, N. (2003), Consumer Sophistication. Wiesbaden:Gabler.

Shim, S. and K.C. Gehrt (1967), “Hispanic and NativeAmerican Adolescents: An Exploratory Study ofTheir Approach to Shopping,” Journal of Retailing,72 (Fall), 307–24.

Slama, M.E. and A. Tashchian (1985), “Selected Socio-Economic and Demographic Characteristics Associ-ated with Purchase Involvement,” Journal of Mar-keting, 49 (Winter), 72–82.

Snyder, C.R. and H.L. Fromkin (1977), “Abnormality asa Positive Characteristic: The Development and Vali-dation of a Scale Measuring Need for Uniqueness,”Journal of Abnormal Psychology, 86 (October), 518–27.

Spence, M.T. and M. Brucks (1997), “The ModerationEffect of Problem Characteristics on Experts’ andNovices’ Judgments,” Journal of Marketing Research,34, 233–47.

Sproles, G.B. and E.L. Kendall (1986), “A Methodologyfor Profiling Consumer’s Decision-Making Styles,”The Journal of Consumer Affairs, 20 (2), 267–79.

Tian, K.T., W.O. Bearden, and G.L. Hunter (2001), “Con-sumers’ Need for Uniqueness: Scale Developmentand Validation,” Journal of Consumer Research, 28(1), 50–66.

Tseng, M. and X. Du (1998), “Design by Customers ofMass Customization Products,” Annals of the CIRP,47, 103–6.

____________ and J. Jiao (2001), “Mass Customization,”in Handbook of Industrial Engineering, 3d ed. G.Salvendy, ed. New York: Wiley, 684–709.

von Hippel, E. (1998), “Economics of Product Develop-ment by Users: The Impact of ‘Sticky’ Local Infor-mation,” Management Science, 44 (5), 629–44.

Westbrook, R. and M. Reilly (1983), “Value-PerceptDisparity: An Alternative to the Disconfirmation ofExpectations Theory of Consumer Satisfaction,” inAdvances in Consumer Research, R. Bagozzi and A.Tybout, eds. Vol. 10, Ann Arbor.

Williams, K.C. and R.L. Spiro (1985), “CommunicationStyle in the Salesperson-Customer Dyad,” Journal ofMarketing, 22 (4), 434–42.

Wind, Y. and A. Rangaswamy (2001), “Customerization:the next Revolution in Mass Customization,” Jour-nal of Interactive Marketing, 15 (1), 13–32.

Yi, Y. (1990), “A Critical Review of Consumer Satisfac-


tion,” Review of Marketing, Chicago: American Mar-keting Association, 68–123.

Zaichowsky, J. (1985), “Measuring the Involvement Con-struct,” Journal of Consumer Research, 12 (Decem-

ber), 341–52.Zeithaml, V.A. and M.J. Bitner (2003), Services Market-

ing. Integrating Customer Focus Across the Firm.New York: McGrwaw-Hill.

APPENDIX 1

Description of Used Scales to Conceptualize Customer Co-Design Competence

(1) Product and application competence

(i) Product knowledge refers to a certain product category and describes product-related information stored in thecustomers’ mind (e.g., Selnes and Gronhaug 1986). The degree of product knowledge that customers possess (withregard to the product that is being customized) will influence their skills to deal with the co-design process. The 5-itemscale of Flynn and Goldsmith (1999) is used to measure consumer’s self-reported familiarity and expertise with a productcategory.

(ii) Product experience comprises all experiences customers have in using a product (see e.g., Mason et al. 2001, whouses the term “product familiarity”). This concept is related to product knowledge (see e.g., Phillipe and Ngobo 1999).Customers who already have a high level of experience with the product (category) being customized might be able tobetter anticipate their role during the co-process (e.g., the role of selecting the color of different product attributes)compared to consumers who lack such previous experience. We use the 4-item scale of Griffin et al. (1996) to measurethe product experience.

(iii) Process knowledge seems to be important in addition to product knowledge due to the paramount role of the co-design process. Because customers are active partners of the company in the co-design process, their knowledge aboutthe process influences the desired outcome (Dellaert and Stremersch 2005; Huffmann and Kahn 1998, see also servicemarketing literature, e.g., Bowen 2000; Honebein 1997; Mills and Morris 1986). We define process knowledge asknowledge related to the co-design process and measure it by adapting the 5-item scale of Flynn and Goldsmith (1999)for the process. We included one additional item to consider some specifics of the mass customization context.

(iv) Process experience is also relevant for the targeted performance. We define this as all of the customer’s experienceswith the co-design process. Based on the scale adapted from Griffin et al. (1996), we measure process experience withfour items. It is relevant because compared to customers without any process experience, customers who alreadyexperienced a co-design process have a huge informational advantage in regard to their role and contribution. Again,we include one mass customization specific item.

(2) Social skills

(v) Communication skills seem to be relevant for customers’ integration competence, as they have to transfer their wishesin regard to the desired product and interact directly with an employee at the POS if the co-design process is conductedat a physical location. This is the case in our empirical study, and we used the customer interaction-related part ofWilliams and Spiro’s (1985) scale (six items) to measure communication style in the salesperson-customer interaction.

(vi) We further considered the customers’ need for interaction with the service employee. In many service encounters,personal interaction is an important factor for evaluation (Bitner et al. 1990). To measure this construct, we used the 4-item scale by Dabholkar (1996), which measures consumers’ evaluation of new technology-based self-service options.This scale seems to be appropriate because in a mass customization context, consumers can often substitute the personalinteraction at the POS with online configuration processes. Consumers with a high need for personal interaction,however, are more likely to select mass customization offerings with a personal sales system.

(3) Methodological competence

(vii) Attitude toward technology seems to be important for the desired outcome since a configurator or other technicalequipment, e.g., a scanner, might be used during the co-design process. Dabholkar (1996) shows that attitude toward


using technological products has positive effects on the expected quality of technology-based self-services. In our case,technology is often steered by the retailer. However, the openness of customers toward technology might influence thesuccess of the process. The scale by Dabholkar (1994 and 1996; Dellaert et al. 2004 use the same scale for masscustomization) is used to measure attitude toward technology with four seven-point semantic differential scales.

(4) Involvement and personal motivation

(viii) Product involvement was measured with the scale by Laurent and Kapferer (1985). Product involvement refers tothe interest or importance of the object of consumption for a consumer in a certain product category (Laurent andKapferer 1985; Sauer 2003; Zaichowsky 1985). It influences decision processes and information search behavior (Sauer2003), and may also be relevant for the consumers’ decision to participate in a co-design process. We use the five-dimensional scale of Laurent and Kapferer (1985) to address the fact that involvement has several dimensions. Inaddition, one new mass customization-specific item was considered. (After the pretest, the six items with the highestfactor loading were used for the final study.)

(ix) Purchase involvement can be described as customers’ involvement in purchasing. It is related to purchasing activities(Slama and Tashchian 1993) and information search behavior (Beatty and Smith 1987; Clarke and Belk 1979). As theco-design process is a purchasing activity, we consider the purchase involvement relevant for a customer’s decision toco-design his or her custom product. We measure this using the scale of Lockhin et al. (1997) that comprises five items.

(x) Shopping enjoyment is the extent to which a consumer views shopping as an enjoyable activity (Shim and Gehrt1996). We use the 4-item scale of Shim and Gehrt (1996), a modified version of Sproles and Kendall’s (1986) scale. Wefollow this approach because shopping enjoyment might motivate consumers to participate in a co-design process. Thisaspect reflects the fact that customers may consider a mass customization process as a hedonistic shopping experience.

APPENDIX 1 (CONTINUED)

Description of Used Scales to Conceptualize Customer Co-Design Competence

For further information contact:Melanie Müller

Technische Universität MünchenLeopoldstr 13980804 Munich

GermanyPhone: +49.89.289.24800

Ralf ReichwaldTechnische Universität München

Leopoldstr. 13980804 Munich


Frank Piller, RWTH AachenTemplergraben 64

52062 AachenGermany

Phone: +49.241.80.93577E-MaiL; [email protected]


WHO SHOULD DO THE WORK: THE DIFFERENT ROLES OF

CUSTOMER PARTICIPATION AND ITS IMPACTS

ON SATISFACTION

Lan Xia, Bentley College, WalthamXiucheng Fan, Fudan University, China

ABSTRACT

With the economy shifting from a good-dominantlogic to a service-dominant logic (Vargo and Lusch 2004),the role of customer participation becomes more impor-tant. Does customer participation enhance or decreasesatisfaction? Research seems to offer conflicting answers(Bendapudi and Leone 2003; Cermak et al. 1994). In thisresearch we distinguish the role of participation as “co-production” or “co-creation of value.” Using an experi-ment we showed that the impact of participation onsatisfaction depends on both the role of participation andservice outcome. When participation plays the role of “co-creation of value,” participation enhanced satisfactionwhen outcome was positive and decreased satisfactionwhen outcome was negative.

INTRODUCTION

Inseparability has long been identified as a uniquecharacteristic of service (Zeithalm 2000). Very often,consumers are part of the service creation and deliveryprocess. For example, consumers order and serve theirown food in a fast food restaurant. More recently with thedevelopment of new technologies, even manufacturerscan get customers involved by offering productcustomization feature. For example, on Levi jeans website,consumers can take their own measures and enter the datato get a specific pair of jeans manufactured for them. Bydoing this, consumers actually participate in the manufac-turing process. The distinction between services and goodsis increasingly blurred. With the economy shifting from agood-dominant logic to a service-dominant logic (Vargoand Lusch 2004), the role of customer participation be-comes more important.

More importantly, customer participation is no longermerely “co-production” where firms shift part of the workto customers in order to increase productivity and reducecost. Consumers can participate in the design, manufac-ture, and delivery of a product or service and become “co-creator of value” (Bolton, Grewal, and Levy 2007). Areconsumers motivated to participate? Are consumers moreor less satisfied with participation? Do they blame them-selves when products or services fail because they are partof it? Or do they blame the service providers or manufac-turers who make them do the work? Few researchers make

clear distinctions on the different roles of customer par-ticipation and examines different consequences of partici-pation in terms of consumers’ psychological experiencesand evaluations of the product or service. In this research,we try to investigate these questions. We first examine thenature of customer participation by distinguishing partici-pation as co-production (i.e., participation as a source ofproductivity gain for the service provider) versus partici-pation as co-creation of value (i.e., participation as anecessary input for the services or products). Further, wedevelop hypotheses regarding consumers’ psychologicalexperiences and satisfaction with the service providerbased on different roles of participation.

NATURE OF CUSTOMER PARTICIPATION

In a broad sense, customer participation refers to thedegree to which the customer is involved in producing anddelivering the service (Bendapudi and Leone 2003;Dabholkar 1990). Research has tried to classify customerparticipation along different dimensions. For example,Bitner, Faranda, Hubbert, and Zeithaml (1997) catego-rized customer level of participation into low (mere pres-ence, e.g., airline travel), moderate (customer input, e.g.,hair cut), and high (customer co-creates, e.g., personaltraining). They argued that level of participation varies bydifferent types of services. Cermak et al. (1994) catego-rized participation by type of input such as physical andcognitive. Similarly, Ennew and Blinks (1999) catego-rized customer input as three types: information share,fulfill their duty and responsibility, and exert effort tofacilitate trust, support, corporation, and commitment.

In this research, we broadly distinguish customerparticipation by the role it plays. We suggest that customerparticipation can be distinguished based on whether par-ticipation is absolutely required and whether the quality ofconsumer participation bears important consequences onservice outcome. Kelly et al. (1990) suggested that formany services, customers are required to contribute in-formation or effort before the service can be consum-mated. This type of participation fits in contexts such asproduct/service customization where consumers’ input iscrucial in product design and service delivery. By partici-pating, consumers actually co-create values that theyacquire (Bolton, Grewal, Levy 2007). This type of partici-pation is generally not substitutable. For example, it may


take consumers’ effort to articulate what kind of jeans he/she likes but he/she is the only and best person that can dothe work. In some cases such as weight-loss or health care,no service could be possibly successfully delivered with-out consumers’ active participation. Such customer par-ticipation brings many benefits to the firm includingincreased revenue by offering better value-added servicesand consumers spending more on products and services(Wind and Rangaswamy 2001).

However, in other situations, the customer is only oneof the parties that can do the job while the same work canbe done by the product or service provider. For example,while consumers can serve their own food, pump theirown gas, or self-check in at an airport, the same work canbe done by the service employees, saving consumers’effort. In these situations, consumers are involved in theco-production process and participation does not createadditional value for themselves. Typically firms use suchcustomer participation as a way to gain productivity andreduce cost.

Overall, we argue that customer participation playstwo different roles and each involves different goals andbenefits on the service providers’ side and motivationsand expectations on the consumers’ side. While participa-tion plays different roles in different services, it is alsopossible that both roles can be applied to the same servicewith different levels and types of effort input. We expectthat they will have different impacts on consumer satisfac-tion.

DIFFERENT ROLES OF PARTICIPATION AND

CONSUMER SATISFACTION

One of the primary interests of researchers and prac-titioners on the issue of customer participation is the effectof participation on consumer satisfaction. Will customerparticipation enhance or decrease customer satisfaction?Empirical research offered mixed results (Bendapudi andLeone 2003; Cermak et al. 1994). Bendapudi and Leone(2003) showed that participation did not lead to highersatisfaction when service outcome is negative. Whenoutcome is positive, participation actually decreased sat-isfaction due to consumer self-serving bias. However, inthe context of professional legal and financial services,Cermak et al. (1994) showed that participation led toincreased satisfaction. We argue that the differences inthese findings are due to the different roles of participa-tion. In Bendapudi and Leone’s (2003) study, all partici-pation described are efforts that are not absolutely re-quired and the outcomes are independent of participation.Customer participation played the role of “co-produc-tion” and reduced product or service provider’s work(e.g., customer calling the hotel to reserve the roominstead of the travel agent calling). Participation does notcreate additional value to consumers. On the other hand,

in Cermak et al.’s (1994) study, customer participationplayed the role of “co-creation of value” and was crucialto the quality of the service outcome. Customer participa-tion was used to customize and produce the unique serviceto the customer instead of reducing the service provider’seffort. Hence, we propose that participation has differentimpacts on satisfaction depending on different roles itplays.

Customer Participation as Co-Production

An examination of literature showed that the conceptof customer participation can be traced to two differentroots, one from management and operations research andone from marketing. The two roots correspond to the twodifferent roles that customer participation plays. Researchin management and operations research conceptualizedcustomer participation as a source of productivity gainand customers are treated as a productive resource(Fitzsimmons 1985; Lovelock and Young 1979). Firmsmay treat clients as “partial” employees (Mills and Morris1986) and customer participation replaces all or part ofwork that are previously done by the service employees.For example, having customers write the postal codes onthe mail will reduce the work of postal service employees.Many of the self-service technologies such as ATMmachines and self-service gasoline stations replaced em-ployees with technologies and consumers are required todo some of the work that are used to be done by serviceemployees.

When customer participation merely plays the role ofenhancing service provider’s productivity, consumers’work and effort are typically substitutable and participa-tion does not create additional value to customers. There-fore, consumers are not particularly motivated to partici-pate and involvement is low. Research showed that satis-faction judgment is influenced by equity considerationswhere consumers consider what they obtain relative towhat they put in (both money and non-monetary effort)(Oliver and Swan 1989; Swan and Oliver 1991). Sinceparticipation mainly contributes to service providers byreducing cost and increasing productivity, the more inputfrom the customers part means less input from the serviceprovider’s part and the input does not increase consumers’output (i.e., value obtained). Hence, when outcome ispositive, the more customers participated, the less they aresatisfied according to the prediction of equity theory. Self-perception bias theory offers a similar prediction(Bendapudi and Leone 2003). When the outcome ispositive, those who participated are likely to credit them-selves for the success and the amount of the work willreduce satisfaction with the service provider.

When outcome is negative, consumers may rational-ize who is to blame. Attribution theories suggest thatpeople are more likely to attribute failure to external


sources than internal sources (Weiner 1985). In addition,since such participation is typically substitutable and doesnot influence service outcome, consumers may see thatthe service provider actually has the control of not havingcustomers do the work. Hence, consumers are reluctant toattribute the failure to themselves and we would notobserve the difference between participation versus non-participation. Hence, we predict:

H1a: When outcome is positive, participation leads tomore internal attribution (i.e., self-credit) than non-participation.

H1b: When outcome is negative, there is no difference ofinternal attribution between participation and non-participation.

H2a: When outcome is positive, participation leads tolower satisfaction than non-participation.

H2b: When outcome is negative, there is no difference insatisfaction level between participation and non-participation.

Customer Participation as Co-Creation of Value

Research in marketing started to re-examine the roleof consumers in the 90s. Firat and colleagues (Firat et al.1995; Firat and Venkatesh 1995) argued that consump-tion becomes the means through which individuals definetheir self-images for themselves as well as to others.Instead of playing a passive role, consumers actively takethe elements of market offerings and craft a customizedconsumption experience for themselves. They are moti-vated to become a participant in the customization of his/her experiences instead of merely being a recipient that isdetached from the process. In such customer participa-tion, the goal of service providers and manufacturers is tooffer consumers a unique product/service and consump-tion experience. Customer participation becomes a pro-cess of co-creation value for consumers. For example, forhealth care providers to offer the right diagnosis andtreatment, the patient has to be able and willing to explainhow he/she feels and what their preferences are for treat-ment procedures and patients have to carefully followinstructions to obtain the desired outcome. Since partici-pation is not substitutable and bears important conse-quences on service outcome, consumers are more moti-vated to do their part of the work as doctors instructed.

When customer participation plays the role of co-creation of value and used as a product/service differen-tiation strategy through customization or personalization(Song and Adams 1993), consumer input (i.e., participa-tion) is directly related to the outcome (i.e., the quality ofthe service obtained). Therefore, when outcome is posi-

tive, participating consumers will pride themselves on theachievement (e.g., I did it myself) (Bitner et al. 1997;Franke and Schreier 2006) and the enhanced satisfactionwith the outcome may spillover to satisfaction with theservice provider since it is a joint effort. For example,research showed that when consumers comply with par-ticipation requirements and reach their goals, complianceleads directly to satisfaction (Dellande et al. 2004). Hence,when service outcome is positive, we expect that satisfac-tion with the service provider will be higher for participat-ing consumers.

Since service tends to be inconsistent, service out-come could be negative even with customer participation.In situations where a service outcome is based on jointactions of the service provider and the customer, level ofcustomer participation may influence attributions whennegative outcome occurs. Participating consumers mayblame the service provider since they did their part of thework and there is no excuse for the service provider not todeliver the expected service outcome. Hence satisfactionwith the service provider will be lower comparing to non-participating consumers. On the other hand, non-partici-pating consumers may blame themselves for not doingtheir part of the work (Yen, Gwinner, and Su 2004), hencethe negative outcome should have a lesser effect ondissatisfaction comparing to participating consumers.Therefore, we hypothesize that

H3a: When outcome is positive, participation will leadto more credit to self than non-participation.

H3b: When outcome is negative, non-participation willlead to more blame to self than participation.

H4a: When outcome is positive, non-participation leadsto higher satisfaction than participation.

H4b: When outcome is negative, participation leads tolower satisfaction than non-participation.

SATISFACTION WITH OUTCOME, PROCESS,

AND SERVICE PROVIDER

Consumer satisfaction has been shown to be influ-enced by both the outcome and the process (Tax et al.1998; Bendapudi and Leone 2003). Drawing on justicetheory, Tax et al. (1998) showed that consumers’ satisfac-tion judgment on their complaint incident based on theoutcome as well as the process of how the complaint ishandled. However, the relative strength of outcome andprocess satisfaction in predicting service provider satis-faction may vary in the context of customer participation.Bendapudi and Leone (2003) showed that process satis-faction has a larger impact when consumers choose toparticipate comparing to when they choose not to partici-


pate. Self-perception bias explains this effect since con-sumers more or less take responsibility for what theychoose.

While Bendapudi and Leone’s research (2003) oper-ated in the context of “co-production,” we propose that therelative impact of outcome and process satisfaction willbe different when participation is closely related to serviceoutcome (co-creation of value). Consumers who partici-pated may perceive that they did what they are supposedto do and have a clearer expectation of what the outcomeshould be. In addition, they may perceive that there areless excuses on the service provider’s part for servicefailures (I did my work and you should deliver). Hence,consumers will rely more on service outcome to formservice provider satisfaction. On the other hand, consum-ers who did not participate may realize that they did notcomply with service provider’s requirement and they aremore likely to blame themselves for the service failure.Therefore, outcome satisfaction has less impact on satis-faction with the service provider compared to when con-sumers participated. Instead, process satisfaction mayplay a bigger role. Therefore, when participation plays therole of co-creation of value, we expect that,

H5a: Outcome satisfaction is more predictive of satis-faction with the service provider when consumersparticipated compared to when not participated inthe service process.

H5b: Process satisfaction is more predictive of satisfac-tion with the service provider when consumers didnot participate compared to when participated inthe service process.

THE STUDY

Design, Stimuli, and Measures

To test the interaction effects of level of participation,role of participation, and service outcome, we designed a2 x 2 x 2 between subjects experiment. We operated thedesign in the context of a weight loss program. Role ofparticipation is manipulated as “co-creation of value”(participation is not substitutable and crucial to the out-come) or “co-production” (participation is substitutableand has no impact on service outcome). Participation as“co-creation of value” is described as the amount of effortin following the instructions in order to lose weight (i.e.,exercise and stick to the diet). Participation as “co-pro-duction” is described as the logistics in buying food itemsthat supplement the exercise for weight loss. Level ofparticipation is manipulated as participation vs. non-participation on the customers’ part. In the participation as“co-creation of value” condition, participants were toldthat Pat (the customer described in the scenario) eitherstrictly follows the exercise and diet requirement of the

program (participate) or skips exercises and eats out often(non-participate). In the participation as “co-production”condition, participants were told that Pat either buys thefood items specified by the program herself from a localsuper market (participate) or the center purchases theitems for her so she can pick up the items convenientlywhen she attends the program (non-participate) and paysthe center the cost of the items. Service outcome isdescribed such that the weight loss outcome is either“much better than expected” or “much worse than ex-pected.” Undergraduate students (n = 150) from a busi-ness school in Northeast U.S. participated in the experi-ment with 56.8 percent male and average age of 20.4years.

A scenario-based approach was used. Students weregiven a booklet where they read a scenario describingPat’s experience with a weight loss program. The scenariodescribed the process of the program (level of participa-tion and role of participation) as well as the outcome of theservice by the end of the program. After reading thescenario, participants answered a set of questions. Wemeasured participants’ satisfaction with the outcome,satisfaction with the process, and satisfaction with theservice provider using a 7-point semantic differentialscale anchored on dissatisfied and satisfied. We alsomeasured participants’ attribution of blame (in case ofnegative outcome) or credit (in case of positive outcome).In addition, we measured customers’ preference for con-trol, ability to do the work, and perception of social normsin terms of who should do the work and used them ascontrol variables. Finally, we measured consumers’ mo-tivation of participation, expectation of outcome, andfuture purchase intentions and used them as some processmeasures. All measures are presented in Table 1.

Results

Manipulation checks were successful. Participants inthe participation condition perceived that “Pat devoted alot of work in the process of weight loss” more than thosein the non-participation condition (F(1,143) = 10.22), p =0.002, mean = 5.26 vs. 4.45). Participants in the participa-tion as “co-creation of value” condition perceived that“whether Pat does this work herself has a big influence onthe result of the weight loss program” more than those inthe “co-production” condition (F(1,143) = 8.55), p =0.007, mean = 5.58 vs. 4.93). Also as we expected,participants in the “co-creation of value” condition weremore motivated to participate than those in the “co-production” condition (F(1,143) = 11.09), p = 0.001,mean = 3.96 vs. 3.27).

We conducted a MANOVA analysis using level ofparticipation, role of participation, and service outcome asindependent variables and satisfaction with the outcome,process, and service provider as dependent variables. We


controlled customers’ preference for control, ability to dothe work, and perception of social norms by using them ascovariates. Analysis showed a significant 3-way interac-tion on satisfaction with the service provider (F(1,139) =13.05, p < 0.001) as well as on satisfaction with theoutcome and process. We will first discuss the results onsatisfaction with the service provider and then examinethe relationships among different types of satisfaction.

For participation as “co-production,” when outcomeis negative, participation did not influence satisfaction(F(1,33) = 1.97, p = 0.16) which is as expected in H2b.However, participation did not influence satisfaction whenoutcome is positive (F(1,33) < 1). So H2a is not supported.

For participation as “co-creation of value,” when outcomeis positive, participation enhanced satisfaction (F(1,32) =4.48, p = 0.04). When outcome is negative, participationdecreased satisfaction (F(1,34) = 7.03, p = 0.01). So bothH4a and H4b are supported.

We further examined the effects of participation invarious conditions on attributions. Factor analysis on thefour attribution items revealed two factors: attribution tothe service provider and attribution to self. Hence we usedthem as two separate dependent variables. We conductedanother MANOVA analysis using level of participationand role of participation as independent variables andconsumers’ attribution to self and service provider as

TABLE 1

Measures and Reliability

Measures Reliability

Satisfaction

How satisfied is Pat with the weight loss result? (dissatisfied – satisfied)How satisfied is Pat with the process of trying to lose weight? (dissatisfied – satisfied)How satisfied is Pat with this weight loss service? (dissatisfied – satisfied)

Attributions to the service provider (positive outcome)

I believe that the weight-loss center did an excellent job r = 0.72I will give the weight-loss service credit for the result

Attributions to the self (positive outcome)

It is Pat’s work that made the weigh-loss program a success r = 0.42I think it is Pat’s credit

Norms/Expectations

I think it is normal practice that weight-loss centers asks customers to do this work themselves

Capability

Pat is capable of doing this work herself

Control

Typically, I would like to do my own things α = 0.74I like to take things in my own handsIf there is an option, I usually like to serve myself instead of having other people do it

Motivation of participation

Pat is motivated to do this work herself r = 0.55I don’t think Pat is motivated to do this work herself (R)

Expectation of outcome

Doing this work makes Pat expect a better weight loss result

Future purchase intention

It is likely Pat will use this weight loss service again in the future r = 0.78Pat probably will not use this weight loss service in the future


dependent variables. Separate analysis was conducted onpositive (attribution of credit) and negative (attribution ofblame) service outcome.

When outcome is negative, results showed an inter-action effect of level of participation and role of participa-tion on attribution (F(1,69) = 7.80, p = 0.007). For partici-pation as “co-creation of value,” when outcome is nega-tive, people who participated tended to put more blame onthe service provider (F(1,34) = 2.85, p = 0.1; mean = 3.4vs. 2.6) than those who did not participate. On the otherhand, those who did not participate tended to blamethemselves ((F(1,34) = 16.43, p < 0.001; mean = 5.1 vs.3.4) more than those who participated. For participationas “co-production,” participation had no effect on eitherself or service provider attributions. Hence, H1b and H3bare supported. Also as we expected, attributions mediatedthe effects of participation on satisfaction with the serviceprovider. When we used attributions as covariates, theeffects of participation on satisfaction (H4) became statis-tically non-significant. In addition to the interaction, wealso observed a main effect of participation on self-attribution (F1,69) = 10.35, p = 0.002). Non-participingconsumers put more blames on themselves than partici-pating consumers. When outcome is positive, only a maineffect of participation on self attribution is observed(F(1,67) = 8.93, p = 0.004). Regardless role of participa-tion, participating consumers attributed more credits toself than non-participated consumers, supporting H1a andH3a.

Finally, we compared the measures on outcome sat-isfaction, process satisfaction, and service provider sat-isfaction. Similar to satisfaction with the service provider,significant 3-way interactions were observed (F(1,139) =3.71, p = 0.06 on outcome satisfaction; F(1,139) = 4.3, p =0.04 on process satisfaction). As expected, regressionanalysis showed that both outcome satisfaction (b = 0.68)and process satisfaction (b = 0.26) influenced overallsatisfaction with the service provider (R

adj = 0.71). Sepa-

rate regressions suggested that the relative weight ofoutcome satisfaction and process satisfaction varied un-der different participation conditions. For participation as“co-creation of value,” outcome satisfaction played amore important role when consumers participated (r =0.85) than when not participated (r = 0.51), supportingH5a (Chow test F(1,74) = 4.22, p = 0.04). On the otherhand, process satisfaction played a more important rolewhen consumers did not participate (r = 0.30) comparedto when participated (r = 0.17). However the Chow test isnot statistically significant (F(1,72) = 2.39, p = 0.13) soH5b is not supported. Since we argued that the relativelyhigher influence of outcome satisfaction is due to out-come expectations, we checked the expectation measureand found that participating consumers had a higherexpectation of service outcome than non-participatingconsumers (F(1,143) = 10.10), p = 0.002, mean = 4.97 vs.

4.24). No significant differences between participationvs. non-participation condition were found when partici-pation is “co-production.”

DISCUSSIONS AND FUTURE RESEARCH

Results of the study supported most of our expecta-tions. For participation as “co-production,” consistentwith Bendapudi and Leone (2003), we found that partici-pation had no effect when outcome is negative. However,when outcome is positive, we did not find the expectednegative effect of participation. One possible reason couldbe that participants may perceive that buying the fooditems is part of the work they should do. It is not clear thatthe service provider is shifting its work to consumershence the manipulation may represent a weak case ofparticipation as co-production. In fact, participants per-ceived that consumers typically do the work (mean = 5.5on the social norm measure). We did a median split onperception of who should do the work and results showedthat people who think the service provider should do thework were less satisfied with the service provider thanthose who think consumers should do it (mean = 6.6 vs.6.0) when the outcome is positive. Therefore, it is likelythat consumer perception of social norms in terms of whoshould do the work could serve as a moderating factor ofthe impact of participation as co-production on satisfac-tion. This is consistent with literature suggesting thatconsumers gradually adapt to changes such as self-servicegas stations (Mills and Morris 1986).

For participation as “co-creation of value,” we foundthe opposite effect compared to participation as “co-production.” When outcome is positive, participationenhanced satisfaction. When outcome is negative, partici-pation decreased satisfaction. Examination of consumerattribution showed that when outcome is negative, con-sumers who participated tended to blame the serviceprovider than those who did not participate (i.e., “I did mywork so you must be responsible for the failure!”). Con-sumers who did not participate tended to blame them-selves (i.e., “I can’t complaint to the service provider sinceI didn’t do my job.”). When outcome is positive, consum-ers may pride themselves on the accomplishment and thecompliance to the service provider further enhanced sat-isfaction with the service provider. Our results are consis-tent with research which showed that gaining consumercompliance enhances satisfaction (Dellande et al. 2004).

Overall results supported most of our hypotheses.However, there are several limitations that constrain thegeneralization of the findings. First, we used the contextof a weight loss program and not all participants arefamiliar with the context. Second, the person in the sce-nario is described as a female and it would be interestingto incorporate gender as a factor. Third, we only investi-gated the role of participation in one service context while


nature of participation varies by different types of ser-vices. Therefore, future research should overcome theselimitations by utilizing different service scenarios andidentify additional moderating factors. For example, theamount of value created through participation may vary indifferent service contexts which influence consumers’motivation to participate. In addition, consumers may ormay not realize their influences or are more or less certainabout their influences on the end results. These factorsnaturally influence their attributions of credit or blame,hence satisfaction.

Finally, it is worthwhile to examine service provider’srole in motivating and facilitating customer participationand enhancing satisfaction. Research showed that en-hancing customer role clarity and ability to participatehelp to motivate and enable them to participate as well asenhances satisfactions. Different roles of participationmay be associated with different motivation factors andshould be further examined.

CONCLUSION AND IMPLICATIONS

With the development of more advanced technologyand consumers’ desire to get involved in the product/service design and delivery process, customer participa-tion is becoming more important. However, the impact ofsuch participation has not been carefully examined. Ourresearch attempts to examine such impacts. We startedwith the proposition that participation plays differentroles for service providers. On one hand, service provid-ers can use it as a way to “co-produce” and shift part of thework to consumers to enhance productivity and reducecost. On the other hand, service providers can use it as away to “co-create value” delivered to consumers. It servesas a product/service differentiation tool to facilitate cus-tomized and personalized service design and delivery.While goals from the service providers’ perspective dif-fer, consumers’ motivation to participate and perceivedbenefits also vary. Hence, consumers’ satisfactions withthe service providers vary depending on the role of partici-pation and service outcome.

The implications of these results are that when par-ticipation is necessary and quality of participation directlyinfluence service outcome (i.e., participation as “co-cre-ation of value”), the more effort expended by the customer(i.e., participation), the more likely customers will besatisfied and remain with the service provider whenoutcome is positive. Therefore, managers should try togain consumers’ compliance in participation (Dellandeet al. 2004) and at the same time strive to deliver theexpected results. However, participation led to less satis-faction when outcome is negative. Our results furthershowed that when consumers participated but the out-come turned out to be negative, consumers are less likelyto patronize the service provider in the future (F(1,34) =6.79, p = 0.01). This is probably due to the perception thatservice provider is incompetent. Therefore, it is importantthat service providers offer consumers a clear explanationof the service failure. When the explanation helps con-sumers to attribute the failure to causes external to theservice provider and factors that are uncontrollable, it mayhelp to restore consumers’ confidence with the serviceprovider and make them less likely to switch.

When participation is used as a way to shift part of thework to consumers, consumers are less involved andmotivated to do the work. Since the work is substitutableand participation has the potential to decrease satisfac-tion, managers should enhance consumers’ willingness toparticipate by emphasizing the benefits of participationsuch as convenience or efficiency. In addition, managerscan segment consumers by willingness to participate andoffer different versions of the product/service (with dif-ferent price levels) and let consumers choose (Silpakit andFisk 1985). Our research suggested that consumers’ per-ception whether they should do the work may moderatethe negative effect of participation on satisfaction whenoutcome is positive. Therefore, managers may use thesesocial norms to assess whether they can shift some of thework to consumers and anticipate consumers’ reactions tothese participations.

REFERENCES

Bendapudi, Neeli and Robert P. Leone (2003), “Psycho-logical Implications of Customer Participation in Co-Production,” Journal of Marketing, 67 (1), 14–28.

Bitner, Mary Jo, William T. Faranda, Amy R. Hubbert,and Valarie A. Zeithaml (1997), “Customer Contri-butions and Roles in Service Delivery,” Interna-tional Journal of Service Industry Management, 8(3), 193–205.

Bolton, Ruth, Dhruv Grewal, and Michael Levy (2007),“Six Strategies for Competing Through Service: An

Agenda for Future Research,” Journal of Retailing,83 (1), 1–4.

Cermak, Dianne S.P., Karen Maru File, and Russ AlanPrince (1994), “Customer Participation in ServiceSpecification and Delivery,” Journal of Applied Busi-ness Research, 10 (2), 90–100.

Dabholkar, Pratibha Ed (1990), How to Improve Per-ceived Service Quality by Improving Customer Par-ticipation. Cullowhee, NC: Academy of MarketingScience.

Dellande, Stephanie, Mary C. Gilly, and John L. Graham(2004), “Gaining Compliance and Losing Weight:


The Role of the Service Provider in Health CareServices,” Journal of Marketing, 68 (3), 78–91.

Ennew, Christine T. and Martin R. Binks (1999), “Impactof Participative Service Relationships on Quality,Satisfaction and Retention: An Exploratory Study,”Journal of Business Research, 46, 121–32.

Firat, A. Fuat, Nikhilesh Dholakia, and Allandi Venkatesh(1995), “Marketing in a Post Modern World,” Euro-pean Journal of Marketing, 29 (1), 40–56.

____________ and Alladi Venkatesh (1995), “LiberatoryPostmodernism and the Reenchantment of Consump-tion,” Journal of Consumer Research, 22 (3), 239.

Fitzsimmons, James A. (1985), “Consumer Participationand Productivity in Service Operations,” Interfaces,15 (3), 60–67.

Franke, Nikolaus and Martin Schreier (2006), “I MadeThis Myself! Exploring Process Utility in MassCustomization,” in AMA Summer Educators’ Con-ference, Dhruv Grewal, Michael Levy and R.Krishnan, eds. Vol. 17. Chicago: American Market-ing Association.

Kelley, Scott W., James H. Donnelly, Jr, and Steven J.Skinner (1990), “Customer Participation in ServiceProduction and Delivery,” Journal of Retailing, 66(3), 315.

Lovelock, Christopher H. and Robert F. Young (1979),“Look to Consumers to Increase Productivity,”Harvard Business Review, 57 (May/June), 168–78.

Lusch, Robert F., Stephen L. Vargo, and Matthew O’Brien(2007), “Competing Through Service: Insights fromService-Dominant Logic,” Journal of Retailing, 83(1) 5–18.

Mills, P.K., R.B. Chase, N. Margulies (1983), “Motivat-ing the Client/Employee System as a Service Produc-tion Strategy,” Academy of Management Review, 8,301–10.

Mills, Peter K. and James H. Morris (1986), “Clients as‘Partial’ Employees of Service Organizations: RoleDevelopment in Client Participation,” The Academyof Management Review, 11 (4), 726–35.

Oliver, Richard L. and John E. Swan (1989), “Equity andDisconfirmation Perceptions as Influences on Mer-chant and Product Satisfaction,” Journal of Con-sumer Research, 16 (3), 372.

Silpakit, P. and R.P. Fisk (1985), “‘Participatizing’ TheService Encounter: A Theoretical Framework,” inService Marketing in a Changing Environment, T.M.Block, G.D. Upah, and V.A. Zeithaml, eds. Chicago,IL: American Marketing Association, 117–21.

Song, Jae H. and Carl R. Adams (1993), “DifferentiationThrough Customer Involvement in Production orDelivery,” Journal of Consumer Marketing, 10 (2),4–12.

Swan, John E. and Richard L. Oliver (1991), “An AppliedAnalysis of Buyer Equity Perceptions and Satisfac-tion with Automobile Salespeople,” Journal of Per-sonal Selling & Sales Management, 11 (2), 15.

Tax, Stephen S., Stephen W. Brown, and MuraliChandrashekaran (1998), “Customer Evaluations ofService Complaint Experiences: Implications for Re-lationship Marketing,” Journal of Marketing, 62 (2),60–76.

Vargo, Stephen L. and Robert F. Lusch (2004), “Evolvingto a New Dominant Logic for Marketing,” Journal ofMarketing, 68 (1), 1–17.

Weiner, Bernard (1985), “An Attributional Theory ofAchievement Motivation and Emotion,” Psychologi-cal Review, 92 (4), 548–73.

Wind, Jerry and Arvid Rangaswamy (2001), “Customer-ization: The Next Revolution in Mass Customeriza-tion,” Journal of Interactive Marketing, 15 (1), 13–32.

Yen, Hsiu Ju Rebeca, Kevin P. Gwinner, and Wanru Su(2004), “The Impact of Customer Participation andService Expectation on Locus Attributions Follow-ing Service Failure,” International Journal of Ser-vice Industry Management, 15 (1), 7–26.

Zeithaml, Valarie A. and Mary Jo Bitner (2000), ServiceMarketing. Boston: McGraw-Hill.

For further information contact:Lan Xia

Bentley College175 Forest Street

Waltham, MA 02452Phone: 781.891.2468



COOPERATIVE VS. REBELLIOUS CUSTOMERS: HOW THEY

EVALUATE CONSUMER PENALTY

Young “Sally” Kim, Shenandoah University, Winchester

SUMMARY

Background and Hypotheses

Service industry’s popular use of fees and penaltieshas received vast attention from public policy makers asconsumers’ complaints about unfair penalty practiceshave escalated over the years. For example, Congressrecently commissioned a study investigating fees andinterest rates of 28 popular credit card companies andreleased a report in 2006 that late payment penalty hadincreased over the years and averaged $34, up from $13 in1995. Consumer penalties have gained attention not onlyfrom public policy makers but also the companies who arestruggling to find the happy medium between the pressureto increase revenue through penalties and fees and thecompetitive urge to keep customers happy. Surprisingly,very little research has been done to understand the impactof penalty on customers and how different customers usedifferent evaluation processes. In order to fill the gap inthe literature, this study first investigates the dynamicrelationships among penalty-related evaluation (transac-tion-specific evaluation), overall evaluation (i.e., evalua-tion based on overall experiences with the firm), andfuture behavioral intentions. Then, the study examinestwo different customer groups based on their prior rela-tionship with the company (i.e., labeled as “cooperative”and “rebellious” customer groups, which were createdbased on their intent to follow or break the company’srules in the future) and tests whether the model works thesame way for the two different customer groups. Priorresearch indicates that customers who have developed apositive relationship with the company tend to be coop-erative and functional, while the customers who have notoften become dysfunctional and, in some cases, rebel-lious. It is expected that these two different types ofcustomers will use different evaluation processes andbehave differently. Thus, the study examines the differentweights of transaction-specific evaluations and overallevaluation on behavioral intentions for the two customergroups. Overall satisfaction is often used as a mediatorbetween the transaction-specific evaluations and behav-ioral intentions. This study investigates the nature of themediating role (full vs. partial) of overall satisfaction forthe two customer groups.

The following hypotheses are tested in this study.

H1: Overall transaction-specific evaluation (i.e., dissatis-faction with the penalty) will have a significant

negative influence on overall evaluation of the firm(i.e., overall satisfaction) after accounting for cogni-tive antecedents to dissatisfaction (positivedisconfirmation, perceived fairness).

H2: Both overall transaction-specific evaluation and over-all evaluation of the firm will have significant influ-ences on behavioral intentions to repurchase afteraccounting for cognitive antecedents to dissatisfac-tion.

H3: For cooperative customers, overall evaluation willexplain more variances in repurchase intentions thantransaction-specific evaluations.

H4: For rebellious customers, transaction-specific evalu-ations will explain more variances in repurchaseintentions than overall evaluation.

H5: Overall satisfaction will fully mediate the relation-ship between transaction-specific evaluations andrepurchase intentions for cooperative customers.However, a partial mediating effect of overall satis-faction is expected for rebellious customers.

Research Method

Data were collected using cross-sectional surveys oncustomers of various service organizations. The criticalincident technique was employed in the survey question-naire design so that customers’ evaluations of penaltiesbased on a critical penalty incident could be captured.Respondents were asked in the survey to recall anddescribe a recent penalty incident before being subjectedto a battery of structured questions related to evaluationsof the penalty. The final sample after the responses withmissing information were discarded amounted to 201.Question items were generated based on the scales used inprevious studies and all multiple items were subjected toreliability analysis. Cronbach’s alpha coefficients rangedfrom 0.79 to 0.84, suggesting reliabilities of the measures.

Results

The hypothesized structural model was tested usingLisrel for path analysis. The overall model fit is excellentand variances of dependent variables are well explainedby the predictors used in the model. Hypothesis 1 and 2 aresupported. In order to test hypotheses 3 and 4, the samplewas median-split into two groups based on their compli-


ance intentions (willingness to comply with the company’srules in the future). As expected, overall satisfaction playsa more significant role in repurchase intentions thantransaction-specific evaluations do for cooperative cus-tomers. On the other hand, rebellious customers tend torely on transaction-related evaluations more heavily indetermining their behavioral intentions of repurchasethan overall evaluation. Also, overall satisfaction fully(partially) mediates the relationship between transaction-specific evaluations and repurchase intentions for coop-erative (rebellious) customers.

Discussion

The results of the study suggest that both overallevaluation and transaction-specific evaluations be in-cluded and measured in a customer evaluation study tocapture the dynamic relationship among different types of

evaluations and behavioral intentions. The findings of thestudy support that the two different types of evaluationscan be identified and measured separately, and that theyplay different roles in the prediction of repurchase inten-tions for different types of customers. The results suggestthat rebellious customers are more sensitive to a negativepenalty episode than cooperative customers. This meansthat effective management of penalty experiences forrebellious customers (e.g., education program helpingcustomers break away from recidivism) is crucial forcustomer retention. Given that positive overall evaluationis an important determinant of customer retention espe-cially for cooperative customers, companies should culti-vate an environment, in which customers have opportuni-ties to build a positive relationship with the companybefore being subjected to a penalty. References are avail-able upon request.

For further information contact:Young “Sally” Kim

Shenandoah University1460 University DriveWinchester, VA 22601Phone: 540.678.4473



WHAT ROLE DOES CUSTOMER RELATIONSHIP MANAGEMENT

PLAY IN MARKETING STRATEGY?

Martin Reimann, Stanford University, StanfordOliver Schilke, Stanford University, Stanford,

Jacquelyn Thomas, Northwestern University, Evanston

SUMMARY

Customer relationship management (CRM) has re-ceived a growing amount of attention from both practitio-ners and academics for more than a decade. There aremany documented definitions of CRM (see the Appendixof Payne and Frow 2005 for a review). Across the numer-ous definitions, the common idea is that CRM is aboutleveraging information to derive customer insights thatcan be used to implement strategies which enhance theperformance of the organization and add value to thecustomers. Thus, CRM has a clear link to marketingstrategy (Varadarajan and Jayachandra 1999).

However, the issue has been raised as to whetherCRM is a new marketing paradigm or is simply therepackaging of existing marketing thoughts (Bouldinget al. 2005). After extensive observation and interactionwith CRM scholars, Boulding et al. (2005) posited thatCRM is neither a new paradigm nor a simple repackagingof existing marketing ideas. Rather, it is “the outcome ofthe continuing evolution and integration of marketingideas and newly available data technologies, and organi-zational forms” (p. 156). This position suggests that, asopposed to investigating CRM strategies in isolation,there is an opportunity to integrate CRM into a generalstrategic framework and assess its relative impact on firmperformance. Thus, in the current research we focus on thefollowing basic question: “How does CRM relate to otheraspects of marketing strategy and relatively drive firmperformance?” In order to assess this research question,we: (1) integrate CRM with two existing perspectives ofmarketing strategy, differentiation, and focus; (2) reflectmarketing strategy with operational efficiency as an op-posed perspective; and (3) analyze and compare aspectsof marketing strategy in specific environmental settings;that is in highly and less commoditized industries.

Methodology

We interviewed 141 marketing executives, which weidentified through a commercial database. At these busi-ness units, key informants were contacted by phone, afterwhich they were provided with access information to thequestionnaire. Firms were affiliated with one of followingten industries: agriculture and hunting, forestry and log-ging, extraction of stones and soil and other mining,

manufacture of outerwear, manufacture of underwear,manufacture of wearing apparel and accessories, manu-facture of pharmaceuticals, manufacture of furniture,manufacture of toys, and energy supply. These industrieswere chosen to capture a variety of firms that range fromhigh to low commodity environments.

Results

For each construct, indicator reliability was analyzedby looking at the factor loadings, which should be signifi-cant and preferably above a .4 parameter value (Hulland1999). Significance tests were conducted using the boot-strap routine with 500 resamples (Chin 1998). Compositereliability (CR) and average variance extracted (AVE)were analyzed to test construct reliability and validity.Bagozzi and Yi (1988) recommend threshold values of .6for CR and .5 for AVE. Finally, Cronbach’s alpha wasexamined for each construct, which should exceed aminimum level of .6. After having gained confidenceabout the appropriateness of the measurement model, thenext step was to examine the structural model. Given ourresearch goals, we needed separate analysis for the highand low commodity markets. Based on a thorough reviewof the literature, some industries were identified as ex-amples of commodity environments (e.g., Hambrick1983a; Hambrick 1983b; Maes, Guttman, and Moukas1999; Narver and Slater 1990; Stanton and Herbst 2005).We used this precedence to assign industries to the highand low commodity groups. In order to confirm theappropriateness of the assignment of industries to the twogroups, a t-test was performed with commodity environ-ment level as the differentiating factor. This test con-firmed a significantly higher degree of commodity envi-ronment for the high versus low commodity environment.The factor score represents the degree of commoditizationof an industry. It is obtained from PLS reflecting aweighted average of the indicators of the four dimensionsof the commodity environment level. We constructed astructural model and obtained results from two differentdata sets, one from low and the other from high commod-ity environments.

We examined the structural model by looking at theexplanatory power of the entire model and the predictivepower of the independent variables (Chin 1998). Theexplanatory power can be analyzed by looking at the


squared multiple correlations (R2) of the main dependentvariable (Chin 1998), performance. In the model with datafrom the low commodity group, 16.4 percent of thevariation in performance is explained by the marketingstrategy and operational efficiency constructs, while R2 is12.3 percent for the data from the high commodity group.Thus, our model helps highlight some important factorsthat are associated with more successful firms.

The predictive power of the independent variablescan be tested by examining the magnitude and signifi-cance of the standardized parameter estimates betweenthe constructs. The findings provide solid support for theconceptualizations in both subsamples. Except for CRMin low commodity environments, all of the dimensions ofdifferentiation, marketing strategy, and performance aresignificant and show the expected positive sign. At the

same time, the path coefficients between marketing strat-egy and performance as well as between operationalefficiency and performance are significant.

Findings

The first key finding is that CRM seems to be insig-nificant in less commoditized industries with high switch-ing costs, low price sensitivity, low product homogeneity,and high industry dynamism. The second key result isthat, as industries become more commoditized, the man-agement of a brand becomes less important than themanagement of the customers. The third key outcome isthat marketing strategy has a greater effect on a firm’sperformance in a highly commoditized market than op-erational efficiency. References available upon request.

For further information contact:Martin Reimann

Department of PsychologyStanford University

470 Jordan HallStanford, CA 94305

Phone: +1.650.723.7431Fax: +1.650.725.5699


Oliver SchilkeInstitute for Research in the Social Sciences

Stanford University482 Galvez Street

Stanford, CA 94305Phone: +1.650.736.1137


Jacquelyn ThomasIntegrated Marketing Communications

Northwestern University1870 Campus DriveEvanston, IL 60208

Phone: +1.847.491.2196Fax: +1.847.491.5925



ARE NEWSPAPER SUBSCRIBERS CLOSELY CONNECTED OR JUST

BOUND BY CONTRACT? AN EMPIRICAL ANALYSIS OF

MODERATING EFFECTS IN THE FOUR-STAGE

LOYALTY MODEL

Patrick Lentz, University of Dortmund, GermanyDavid M. Woisetschläger, University of Dortmund, Germany

Heiner Evanschitzky, University of Strathclyde, United Kingdom

SUMMARY

Customer relationship management and customerloyalty are known to be important issues in many differentindustries. We adapt Oliver’s (1997) four-stage loyaltymodel to a local newspaper setting and identify factorsthat influence customer loyalty in this specific context.Moreover, we analyze direct and moderating effects ofswitching barriers on customer loyalty. Results indicatethat switching barriers and habitual media usage have anegative impact on the link between satisfaction andintention to repurchase.

Introduction

Circulation of most local daily newspapers in devel-oped countries like France, Germany, Italy, Japan, theUnited States, and the United Kingdom is declining.Against this background, customer retention is an impor-tant topic for publishing companies. Their challenge is toshift from transaction-based marketing to relationshipmarketing (Grönroos 1997; Hennig-Thurau and Hansen2000). However, despite the relevance of loyalty to pub-lication houses, only very few academic articles havebeen identified that particularly focus on customer reten-tion in this specific case (Chen and Xie 2007; Deming andGlasser 1968; Keane and Wang 1995). Hence, it is un-clear, if existing loyalty concepts can be directly trans-ferred to this context. In the present paper, we (1) adaptOliver’s (1997) four-stage loyalty model to a regionalnewspaper setting, (2) identify factors that influence cus-tomer loyalty in this specific context, (3) analyze thedirect and moderating effects of switching barriers oncustomer loyalty, and (4) conclude with implications forcustomer relationship management based on our empiri-cal findings. We test our hypotheses using partial leastsquares modeling and multiple-group analysis on a sampleof 440 subscribers of a local newspaper in Germany.

Four-Stage Loyalty Model

We use Oliver’s (1997) definition of loyalty, becauseit includes both attitudinal and behavioral aspects. Heintroduces a four-stage loyalty model, implying that dif-

ferent aspects of loyalty do not emerge simultaneously,but rather consecutively over time (Oliver 1999). Hence,customer loyalty can be seen as a process involving fourphases (cognitive, affective, and conative loyalty as wellas actual purchase behavior). We hypothesize and inves-tigate the interrelations between these different stages.Also, we extend the existing model using both economicand social switching barriers and habitual media usage asinfluencing factors on loyalty, but also as potential mod-erators of the relationships in the four-stage loyalty model.

Empirical Results and Discussion

Using partial least squares modeling, we find that allaspects of cognitive loyalty exert strong significant influ-ences on affective loyalty (H1a – H1b). Specifically, bothproduct functions (β = .20, p < .05) and product attributes(β = .51, p < .01) significantly and positively influencecustomer satisfaction. Second, we find support for thehypothesis focusing on the influence of affective loyaltyon conative loyalty. This can be further documented bythe strong positive influence of customer satisfaction onintention to repurchase (β = .43, p < .01) and on intentionto recommend (β =.37, p < .01). Third, social and eco-nomic switching barriers are found to positively influenceintention to repurchase. Specifically, we find a strong andpositive significant influence of social switching barrierson intention to repurchase (β = .24, p < .05) as well as aneven stronger positive influence of economic switchingbarriers on intention to repurchase (β = .34, p < .01). Inaddition to these direct effects, our model also incorpo-rates a number of moderating effects. We find the influ-ence of affective loyalty on conative loyalty to be signifi-cantly lower (p

diff = .002) with high levels of social

switching barriers. In particular we observe a relativelyminor influence of satisfaction on intention to repurchaseunder conditions of high social switching barriers (β = .24,p < .05), which becomes almost twice as large underconditions of low social switching barriers (β = .46, p <.01). Similarly, we encounter a strong moderating effect(p

diff < .001) of economic switching barriers on the influ-

ence of satisfaction on intention to repurchase. Specifi-cally, while under conditions of high economic switchingbarriers, the effect of satisfaction on intention to repur-


chase is rather low and only marginally significant (β =.17, p < .1), this effect more than doubles under conditionsof low economic switching barriers (β = .52, p < .01).Finally, our analyses reveal a significant moderatingeffect (p

diff = .001) of habitual media usage on the influ-

ence of satisfaction on intention to repurchase. Whileunder conditions of high habitual media usage, the effect

of satisfaction on repurchase intention is rather moderate(β = .21, p < .05), this effect largely increases whenhabitual media usage is perceived to be low (β = .43, p <.01). Specifically the moderating effects provide someinteresting implications for both academics and practitio-ners, which will be discussed in more detail at the confer-ence. References available upon request.

For further information contact:Patrick Lentz

Department of MarketingUniversity of Dortmund


Phone: +49.231.755.3277Fax: +49.231.755.3271



THE MEDIATING EFFECT OF CUSTOMER SATISFACTION BETWEEN

SWITCHING COSTS AND CUSTOMER LOYALTY

Markus Blut, University of Münster, GermanyDavid M. Woisetschläger, University of Dortmund, Germany

SUMMARY

Switching costs are discussed to be crucial for cus-tomer retention and consequently for firms’ financialperformance. Reviewing literature, most studies onlyinvestigate two effects of switching costs. The first one isa direct effect on customer loyalty and the second one isan interaction effect with customer satisfaction on cus-tomer loyalty, although most of these studies argue thatswitching costs may also influence customer satisfactionnegatively. This paper fills that void by analyzing themediating effect of customer satisfaction between switch-ing costs and customer loyalty, using a sample of 179customers of a German energy provider.


ment

Numerous studies have linked customer satisfactionto financial outcomes (e.g., Anderson, Fornell, andLehmann 1994; Anderson, Fornell, and Rust 1997;Bernhardt, Donthu, and Kennett 2000; Ittner and Larcker1998; Keiningham et al. 1999). Despite the obviousmanagerial relevance, most research has examined cus-tomer satisfaction as the main antecedent of customerloyalty (Anderson and Mittal 2000; Anderson and Sullivan1993; Reichheld and Shefter 2000; Szymanski and Henard2001), while largely neglecting the impact of switchingcosts (Fornell 1992), although most studies on switchingcosts argue that they may also affect customer satisfactionnegatively. Switching costs are the onetime costs thatcustomers associate with the process of switching fromone provider to another (Burnham, Frels, and Mahajan2003). Burnham et al. (2003) distinguish between threehigher-order switching costs: (1) Procedural switchingcosts consist of uncertainty costs, pre-switching searchand evaluation costs, post-switching behavioral and cog-nitive costs, and setup costs. (2) Financial switching costsinclude lost performance costs and sunk costs. (3) Rela-tional switching costs comprise personal relationship losscosts and brand relationship loss costs. Whereas the firsttype of switching costs primarily derives from negativesources of constraint, the second and third types deducefrom positive sources of constraint (Jones et al. 2007).Therefore, the routes of switching costs influencing cus-tomer loyalty should differ. Most studies suppose switch-ing costs to influence customer satisfaction negatively,but do not empirically test this effect for different types of

switching costs. Even reactance theory suggests that adecrease of customer control may result in negative emo-tions (Brehm 1989), whereas equity theory suggests thatpositive switching costs may increase customer satisfac-tion (Adams 1965).

Our study fills this void by (1) discussing the impactof different types of switching costs on customer loyaltyand (2) investigating the mediating effect of customersatisfaction between switching costs and customer loy-alty. The following hypotheses are tested against a sampleof 179 customers of a company in the recently liberalizedGerman energy market:

H1: Customer satisfaction has a positive effect on cus-

tomer loyalty.

H2a/b

: Perceived procedural (financial and relational)switching costs have a positive effect on customerloyalty.

H3a/b

: Perceived procedural (financial and relational)switching costs have a negative (positive) effect oncustomer satisfaction.

H4a/b

: As perceived procedural, financial and relationalswitching costs increase, the link between cus-tomer satisfaction and customer loyalty will beweaker.


Results indicate a two-factor, higher order switchingcost typology, allowing researchers and practitioners todistinguish between different kinds of switching costs,which influence customer loyalty through alternativeroutes. It can be noted that procedural switching costsdirectly affect customer loyalty, whereas financial andrelational switching costs influence customer loyaltythrough customer satisfaction. Contrary to our hypoth-eses, no negative effect of switching costs on customersatisfaction was found, although literature recommendsestablishing switching costs carefully, due to their nega-tive effect on customer satisfaction. This might be ex-plained by cognitive dissonance theory, suggesting thatcustomers who are “locked-in” will rationalize their be-havior and view themselves as satisfied (Festinger 1957).


As a conclusion, the specific direct effects of switch-ing costs on customer satisfaction indicate that researchon the alternative routes through which different types of

switching costs constructs affect relational outcomes is anarea for further fruitful research (Jones et al. 2007).References available upon request.

For further information contact:Markus Blut

University of MuensterAm Stadtgraben 13–15

48143 MuensterGermany

Phone: +49.251.8331431Fax: +49.251.8322032



A SOCIAL MARKETING MODEL OF RELIGIOUS ORGANIZATIONS

Robert M. Schindler, Rutgers University – Camden

SUMMARY

By applying Rothschild’s (1999) conceptual frame-work for social marketing to the functional theory ofreligion, it is proposed that the entire institution of religionin human society can be considered to be a form of socialmarketing. Over the millennia, it was recognized that thepull of the baser side of human nature (e.g., self-centeredness and emotional indulgence) could not bewell-managed by exhortation or governmental force alone.To fill the gap, there evolved organizations that, in a sense,seduced people into being good by offering immediatebenefits, such as the feeling of being protected againstmalicious forces or the feeling of being destined forsalvation in the next life, in return for some involvementin the organization. This involvement enabled the organi-zation to get into people’s heads in numerous ways to helpthem behave so that, in effect, they were saved in this life –in other words, were helped to avoid the long-term nega-tive consequences of self-centeredness and emotionalindulgence.

A model of social marketing is proposed to helpclarify how an organization may use the satisfaction ofpeople’s immediate needs as a means to help satisfy thesepeople’s long-term needs. This model is based on the ideathat marketing typically involves nested exchanges – aninner exchange oriented toward immediate needs and anouter exchange oriented toward long-term needs. In com-mercial marketing, the inner exchange involves custom-ers satisfying immediate needs such as quenching thirst orenjoying the excitement of a new car and the sellingorganization satisfying its immediate needs for cash tomeet payrolls and pay suppliers. The occurrence of thisinner exchange also satisfies the selling organization’slong-term needs for profit. Note, however, that there isnothing in this conception of the commercial marketingexchange that ensures any satisfaction of the customer’slong-term needs.

In the proposed social marketing model, this neglectof the customer’s long-term needs is remedied. In socialmarketing, the inner exchange is embedded within anouter exchange that involves the long-term needs of the

customer, rather than the long-term needs of the sellingorganization. The social marketing organization can besaid to offer “direct products” in the inner exchange, and“emergent products” in the outer exchange. In terms ofRothschild’s example of midnight basketball games of-fered to students by colleges to provide an alternative tobinge drinking, the inner exchange involves the collegeoffering the direct product of the enjoyable game. Theoccurrence of this inner exchange also contributes to anouter exchange where the emergent product offered to thestudent is the habit of responsible drinking.

In terms of this social marketing model, a religiousorganization can be considered to draw in congregants byproviding direct products such as inspiring buildings,comforting rituals, and the uplifting sense of being blessed.In return for these direct products, congregants supporttheir churches with money and volunteer labor that satis-fies the religious organization’s immediate needs.Congregants’ continued consumption of the religion’sdirect products help enable the religious organization tooffer to congregants emergent products, such as the habitof forgiveness or the experience of detachment fromharmful behavioral impulses. In return for these emergentproducts, congregants become loyal “adherents,” raisetheir children in the church, and make other contributionsso as to help the church continue to offer the incentives ofits direct products. The paper presents a taxonomy of boththe direct products of religious organizations and of theiremergent products – i.e., the spiritual devices that reli-gious organizations use to exert long-term beneficialinfluences on people’s behavior.

It is argued that the development of religious organi-zations over the course of human history has involved anintuitive use of social marketing principles to great soci-etal benefit. For example, there are no over-arching orga-nizations required here to sponsor social marketing “inter-ventions.” Religious organizations evolved to be self-replicating, so that once formed, they could accomplishbroad social marketing objectives simply by being turnedloose on the world. The paper concludes with discussionof the implications of this analysis for the role and futureof organized religion in modern society.

REFERENCE

Rothschild, Michael L. (1999), “Carrots, Sticks, and Prom-

ises: A Conceptual Framework for the Managementof Public Health and Social Issue Behaviors,” Jour-nal of Marketing, 63 (October), 24–37.


For further information contact:Robert M. SchindlerSchool of Business

Rutgers University – Camden227 Penn Street

Camden, NJ 08102–1656Phone: 856.225.6716



“WHAT ABOUT US? I MEAN, HEY, WE’RE GROWING, WE’RE HERE”:

WOMEN LIVING WITH HIV/AIDS

Yelena Tsarenko, Monash University, Australia

ABSTRACT

This discovery-oriented study examines various as-pects of life of HIV-positive women in Australia. Depthinterviews conducted with seven women revealed thatissues related to support services have much broaderimplications; namely, a need to develop social marketingprograms. The nature of these programs should be recastfrom prevention to intervention through involving peoplewho live with HIV and ameliorating the impact of HIV ontheir everyday lives.

INTRODUCTION

When people hear of another’s suffering – fromcancer, Alzheimer’s, hemophilia, or heart disease – thisimmediately gives rise to sympathetic and supportivefeelings. However, when people hear of someone af-flicted with HIV/AIDS what first comes to mind is drugs,homosexuality or sexually promiscuous behavior. Onoccasions, the idea of medical negligence or mishap mightcross our minds. All of these themes focus on causesrather than consequences. They are strongly reinforced inthe media and lay on the surface of the informationallandscape associated with HIV/AIDS. The focal point ofconcern is infectious disease per se, which often leads tofeelings of fear, denial, avoidance, and ignorance. But thisis only one side of the story. So, what, then, tells the otherside of this story? Of course, the answer is Who, not What.The answer is People, People living with HIV/AIDS(PLWHA), people who suffer from and struggle againstthis terminal condition regardless of how they acquiredthis virus. The reality is as it is understood by PLWHAVictoria – whether Positive or Negative: HIV is part of ourlives.

The experience of PLWHA already has quite a longhistory in our society; however, the virus is stronglyassociated with the predominantly high risk groups. Oneparticularly vulnerable market segment of individualssuffering from HIV/AIDS – but lacking the high visibilitycommon to other risk groups – is heterosexual women.This largely overlooked segment of the HIV populationreceives minimal attention in the literature. This can beexplained by a number of reasons. First, access to the HIVsector is very limited as clients and their confidentialityare highly protected by various bodies, and second, womenare a comparatively small group within the HIV-positivepopulation. To date, little investigation has been con-

ducted on this emerging segment of consumers, and nonehas focused attention on the needs of women living withHIV/AIDS (Ciambrone 2001; Reilly and Woo 2004).This gives rise to a new segment of consumer whose needsare not fully met through traditional medical avenues andwho, instead, require support services that provide bothmedical and non-medical assistance. Consequently, anincreased research focus on enhancing consumer welfareis considered an urgent imperative in mainstream market-ing and consumer behavior literature (Mick 2006; Shethand Sisodia 2006).

Thus, this exploratory study was initiated to addressthe following research objectives:

• To examine the range of needs of HIV-positive women;

• To determine perceptions of HIV-positive womenabout their social representations within broader so-ciety.

OVERVIEW OF CURRENT TRENDS IN HIV

Due to recent medical advancements, people livingwith HIV/AIDS (PLWHA) are able to continue working,maintain existing relationships, and to start families; ineffect they are maintaining aspects of a pre-HIV life.

An increasing proportion of people living with HIVin Australia are heterosexual females. A significant rise innew HIV notifications among women occurred between1999 and 2004 with the latest available figures revealingapproximately 1,600 registered HIV-positive women inAustralia (NCHECR 2004). Significantly, however, in-fections among women do not pertain to an identifiableage or risk group (NCHECR 2004) and consequently, thespecific needs of individual women vary considerably.The main issues that affect the quality of life of HIV-positive people are two-fold: the personal and socialstigma associated with HIV, and the support servicesavailable for those diagnosed with HIV.

HIV and Stigma

People who acquire HIV differ significantly frompeople with other chronic or terminal illnesses in terms ofhow they cope with medical, social, economic and per-sonal pressures. Feelings of guilt at becoming infectedand fear of disclosing their status due to the perceived


risks of being stigmatized and marginalized in society areunique for PLWHA and influence almost every aspect oftheir lives (Molassiotis et al. 2002).

Gender difference also plays a significant role inrelation to stigma felt by PLWHA. For example, womenare inclined to feel victimized if the infection is caused byunprotected sex with men who did not divulge their HIVstatus (Lichtenstein et al. 2002).

HIV and Social Support Services

Increased rates of life expectancy (following initialHIV-positive diagnosis) mean that those infected are nowfacing a different set of challenges that centre on how tolive with HIV. In particular, the health care system,community care networks and similar agencies, now needto develop strategies to provide adequate assistance forpeople with an HIV-positive diagnosis in order to helpthem maintain good mental health and remain indepen-dent and productive members of the community for aslong as possible. These circumstances raise questions asto the accessibility and relevance of available supportservices, which have been primarily designed for homo-sexually active men and injecting drug users.

In the peak years of the HIV/AIDS epidemic, it wasthe gay community that was predominantly afflicted. Thespread of HIV/AIDS to other groups, however, has neces-sitated some urgency in establishing organizations thataccommodate the needs of a more demographically di-verse population. People outside the gay community andother cultural groups identified in the earlier stages of theepidemic (e.g., needle users, sex workers) often lack thevarious forms of emotional, economic, and political sup-port that are required to promote resilient attitudes. Theyalso tend to be susceptible to higher degrees of isolation,stigma, and marginalization.

Women, in particular, are marginalized within theAustralian HIV population and have few support re-sources targeted directly to their needs as a group. This isbecause the specific needs of women vary considerablydepending on their current health condition, family status,financial situation, and their degree of emotional fulfill-ment. Within Australia, several community-based organi-zations target women’s needs. These organizations offera limited range of information resources and services,such as a drop-in service, hospital and home visits, accessto funding for personal development, and food vouchers.

A substantial body of research in mental health pro-vides evidence that use of social support services has thepotential to reduce levels of stress and risk behavior(Bruce et al. 2002; Waddell and Messeri 2006). Findingsfrom Tsunekawa et al. (2004) demonstrate that positive

attitude changes occur when PLWHA encounter otherswho share the same burden. However, while many studiesemphasize that peer support and participation in socialgroups enables HIV-positive people to reinforce theircoping skills, restore emotional balance and increase self-esteem (Molassiotis et al. 2002; Bruce et al. 2002), onlyone study conducted in the United States pointed out thatmany people diagnosed with HIV do not receive regularsupport (Uphold and Mkanta 2005).

In summary, an overview of academic and practitio-ner literature in the area of HIV support services leads usto conclude that, to date, there are no well groundedprograms or soundly designed strategies for tailoringsupport services to HIV-positive people in general andwomen in particular.

METHOD

This project has discovery-oriented goals and em-ploys a qualitative approach for the data collection pro-cess. The sensitive nature of this study, the ability toguarantee confidentiality to respondents and the desire tomake the research as participatory as possible require aface-to-face, individual approach from the researcher. Forthis purpose our research methodology adopts a groundedtheory approach (Strauss and Corbin 1998) embeddedwithin a constructivist paradigm (Denzin and Lincoln2005) in which the data collection method is informal,based on semi-structured in-depth interviews (Strauss andCorbin 1998) and complemented by observation andrecording of field notes (Marshall and Rossman 1999).The research employs the purposive (or judgment) sam-pling techniques (Albright 2002).

Our hermeneutically grounded interpretive researchframework focuses on unfolding personal stories andexperience. From a hermeneutic perspective, the storiesinformants tell about their experience is a prime locus ofdiscovery (Thompson 1997). The informants – as “self-narrators” – relate their life stories with an emphasis ontheir experiences and views on both a personal under-standing of living with HIV and how HIV is understood –or stigmatized – in the wider community.

Interview Schedule

We explored individuals with one illness to attemptto minimize extraneous factors. It is recognized that HIVis potentially different to other illnesses as HIV is oftenstigmatized and those affected often face discrimination,both of which restrict the well-being of people living withthis condition. The health and support sectors are there-fore frequently very protective of their clients and thoseaffected are suspicious of “outsiders.” These factors posea challenge for academic research, especially when re-


searchers are not based in health disciplines. This projecthas been undertaken over two years and we have engagedwith a range of HIV community-based support organiza-tions, medical practitioners and governmental agencies.Individuals who were interested in participating in thestudy phoned or emailed researchers directly to arrange atime and location for interview. The semi-structured inter-views lasted from one to three hours and were conductedat a place selected as convenient for the informants. Allinterviews were audio taped after written consent wasgranted.

Data analysis was performed in two stages employ-ing the hermeneutic (ideographic) approach (Thompsonet al. 1994; Packer 1985; Fournier 1998). The first stageinvolved analyzing the verbatim transcript with the goalof gaining insight to the personal hardship of an HIVdiagnosis in which psychological tendencies were re-vealed as well as an understanding of how women utilizedavailable support services and resources. The secondstage involved comparative analysis of the responses ofeach interviewee.

Sample

The researchers approached a community-based or-ganization in Australia to seek its assistance with recruit-ment. The key informants of this research were sevenwomen living with HIV/AIDS. All our respondents areAustralian citizens, with English as their native languageand all between 35 and 55 years of age. The time lapsesince diagnosis of these women was from 20 months to 20years. Four women have children. Overall, there wassufficient variation in the life circumstances of thesewomen to reflect some diversity of lived experiences andlife perspectives.

FINDINGS

The interviews focused on the impact on personal andsocial relationships that the women experienced duringthe period of their diagnosis, the subsequent commence-ment of treatment, and the ongoing management of theircondition. This approach elicited three key themes thatemerged during the data analysis – stigma and disclosure,peer support, and education.

Stigma and Disclosure

From the onset of HIV/AIDS, women identifiedstigma and disclosure as the most critical issues pertainingto life with HIV. These factors have a great impact uponfurther patterns of behaviors and inevitably affect the wayhow people cope with HIV/AIDS. When asked whetherthey experienced stigma, the respondents’ initial responsesfocused on medical practitioners.

IF5: I have, but not from people who I love. Not fromfamily or friends. From medical profession yes. WhenI had [name of child], she’s 15 months, and themidwives, you know, it wasn’t – it was ignorance.They, you know, a couple of days after she was bornin the hospital they asked me how long I had to livebeing HIV positive. You’d think working in the medi-cal profession they’d have an idea and they had noidea.

IF1: So many different things even to the point in thehospital where they were all watching me, five peoplein a room saying “don’t breast feed,” I had a naturaldelivery and even to the point where they’ve said wewill come in at 10.00 o’clock and bring formula, theydidn’t come. My baby wasn’t nourished so I even leftit to the universe to work with me to show me what todo.

Two informants recalled their experience during preg-nancy, while another informant recalls her first visit to theobstetrician:

IF4: So I’ve handed him my records he’s read it he’sjumped back and said “you’re HIV- positive, I can’ttreat you.” So I just gone . . . just with my mouthopen. . . . so I’m not used to this kind of treatment andto me not . . . being denied medical treatment and hisreaction to jump away from me in horror just mademe feel disgusting, like I was so disgusting wasunworthy of medical treatment.

IF6: Saw a different specialist but her diagnosis evoked avery similar reaction:

IF6: I had a [name] Certificate to be able to go to a dentistfor him to do the dentures and I went to one and hequestioned as to why I had the Scheme form. I toldhim that I had HIV and he outright refused to treat me.He said “It’s just not worth my while – financially it’snot worth my while.” But yes, I actually found itreally hard to challenge him at the time because I wasso hurt and I was nearly in tears. I just felt awful.

While it would be expected that medical or healthpractitioners are fully informed about misplaced socialperceptions about HIV, the negative experiences of womenhad a significant impact on their perception of themselvesand how they are perceived by broader society. All theabove passages relate encounters with non-HIV special-ists. A contrasting perspective is revealed when womendeal with HIV– specialists.

IF2: I think he’s excellent. He gave me so much informa-tion, so much support, so I highly recommend him. Iput him on a pedestal, you know.


Social Sources of Stigma

Stigma is strongly correlated with disclosure of theserostatus. Some of our respondents had strongly negativeexperiences

IF1: . . . and then in partnerships, on the romance, can therelationship find having to deal with disclosure to amale partner. I have never been rejected before andhaving a younger man run off on me one time, afterfive dates I disclosed to him and he choof, gone andthat really hurt my ego . . . because there is noappropriate time to say I have HIV, there never everis.

IF6: I have certainly had people step back from shakingmy hand.

. . . while others chose do not disclose their status andthus protect themselves from any potential stigma

IF3: It’s not a cancer thing, cancer is fine, you can tellpeople – with HIV people see you differently, so it’snothing that you want to disclose to people, becausepeople will treat you differently, believe me.

IF2: It’s not like saying “I’ve got diabetes,” and peopleunderstand it, they’ve got sympathy. With HIV – it’sa big thing.

Concealing HIV status also has negative consequencesin terms of informants carrying a psychological burdenthat they feel cannot be easily shared.

IF2: Most difficult? I think I mentioned it before – like,telling people. It is. I think even . . . Yeah, it is,thinking about it. Like, none of my friends know.How am I supposed to tell them? They’re just goingto judge you. They think you’re bad.

IF4: Secrecy; not being able to talk to anyone else aboutit.

Our informants commonly drew a contrast betweenthe reactions to HIV they experienced in comparison withsocial perceptions of other illnesses.

Peer Support

Our informants were asked about various servicesprovided by government and non-government agenciesand aimed at assisting in normalizing the life of HIV-positive people. Siegel and Lekas (2002) observed twodominant reactions by women to a diagnosis of HIV: toeither engage with the HIV community or to continue withthe pre-illness self, which often results in social and/or

emotional isolation. Encouraging women to engage withcommunity-based organizations through peer supportworkers can reduce this isolation.

IF1: I liked [name], I wish she was still there because Imiss her. She was a very good, down to earth womanthat I could feel I could be very open, be myself withand discuss the issues.

IF5:And as I said, meeting other women who were posi-tive, I needed to see, I need – because I never met, youknow, I’ve never known anyone who was HIV posi-tive so I needed to see that they were actually real andthat they were living.

IF3:Because it was all new to me – you kept thinkingabout your other life that you had and this is all new,but I am so happy that I made the move. Look, evenjust coming here for lunch, I met some really nicepeople; we all have our problems and whatever, butthey’re nice people. And they are people that you cantalk to. . . .

IF4:When you are newly diagnosed the best thing for youis to talk to someone else who is living with it.

Despite the fact that, at the present moment, not allinformants use peer support service, they unanimouslyexpressed the opinion that such services are crucial,particularly at the immediate post-diagnosis stage or attimes when some specific needs arise. It is important tonote that the most effective support-providers may besimilar others; that is, individuals who themselves havesuccessfully faced the same stressful circumstances thatare currently experienced by the subject (Thoits 1995).The importance of peer support is strongly emphasized bywomen in our study and, in general, respondents seekmore communication with other HIV-positive women –to share information and experience, to maintain socialcontact and for emotional support.

However, when respondents reflected on instrumen-tal support services they feel that those services are under-funded and under-developed.

IF1:No they’re [HIV-positive people] definitely not le-gally represented. In the PLC [positive living center]there’s only one or two lawyers . . . but their phoneline is consistently busy from 2.00pm to 4.00pm andI have been advised, keep trying, keep trying. Youcan’t keep picking up the phone and it’s consistentlyengaged. So I have given up on that.

Another informant has serious concerns about thelack of assistance provided for PLWHA who have chil-dren


IF5: For me, you know, I think that the issue needs to beaddressed about children and about disclosing toyour children. I think that’s a very big thing. In thebeginning of these community organizations peopleeither didn’t have children or they, you know, theirchildren were very young, and now 10, 20 years laterwe’ve got these new ones being born and we’ve gotolder ones who need, they need support and they needto – and I think a lot of parents don’t have the, theydon’t know how to tell them or when to tell them.

Two other respondents mentioned the problem of thegeographical concentration of services in one particularcity area as a significant barrier to accessing and usingthose services

IF2: It takes about an hour to get there. By the end of theday you’re tired, you have to drive back or whatever.

IF6: I think the fact that they are all on the south of town.I certainly feel like I am missing out. For example, thePLC is a brilliant place. They have yoga classes, theyhave computer classes, they do a food bank once afortnight. I actually don’t access any of those servicesbecause it’s just too far for me.

Opinions that our respondents expressed toward vari-ous services demonstrate the heterogeneity of HIV-posi-tive population and the diverse needs of females in par-ticular. Their needs vary to a significant degree and theseare invariably linked to the specific stage of life whileliving with HIV.

HIV and Society

The final theme is the most intriguing one. Whenrespondents were asked what should be done to improveconditions of people living with HIV, the respondentsturned their attention to more global issues. At the end ofthe interview when respondents were asked if they wouldlike to add anything, several of them emphasized theimportance of education on HIV as a social issue.

IF3: They are still not educated – even though people saythey are more educated these days, I still don’t seepeople as educated. … We are separated in a sensebecause society is stupid, they don’t understand it.They are the ones that make us separate, but we arejust normal like everyone else. If they get moreeducation maybe they will learn more about it.

IF2: They need that education. I don’t think many peoplehave that unless they’re confronted with it. So I thinkpeople should be more educated.

IF4: I would like to see education being general, not gayspecific. They get the main bulk of the funding for

education and yet there is basically no general com-munity education. I’d like to see that.

As mentioned by IF3, it is a perceived lack of educa-tion that forces HIV-positive people to construct theirworld in terms of “us” versus “them.” It is a chasm whichforces people not only to live in isolation and to feelexcluded from society but has much more broader impli-cations such as loss of self-esteem, employment difficul-ties, lack of and resources, and generally placing limits ontheir opportunities to contribute to society. As IF5 men-tioned:

IF5: I think that it [Australia] really has no idea about thefact that there are people living with HIV. I think thatwe always need to be involved in the decision makingprocess because we are the ones who understandwhat it is to live with it. And yes, and we’re the oneswho can help talk about things like, you know, to stoptransmission and to talk to our own community. Soyes, I think we always need to be involved.

This gives us an insight into a segment of the popu-lation which may have significant economic, social, cul-tural and religious differences, and offers both an under-standing of and impetus for the examination of experi-ences of individual life experiences with HIV. Both theresponses from the interviewees and the surveillance datajointly support the contention that it is extremely difficultto identify any commonality across the female HIV-positive population; however, where they constitute aunited front is in their expressed concern for promotingHIV related educational campaigns in society.

These narratives illustrate the evolving and dynamicnature of support services needed for women through theevolution of HIV/AIDS from the pre-treatment era to thecurrent situation where widely available antiretroviraltreatments are available. Of particular importance is theevidence that reveals a need for counseling and non-medical support services following the commencementand balancing of antiretroviral treatments.

IMPLICATIONS: RESEARCH IN HIV AREA

This study has introduced the contemporary issuesrelating to the experience of life with HIV in Australia andspecifically highlights the dearth of information relatingto women as a minority group within this population. Byadopting a more flexible approach to the marketing view-point that underscores the design of this study, a contribu-tion has been made to better understanding of (1) theissues of stigma and disclosure, (2) range of women’sneeds in support services and (3) the necessity of interven-tion programs aimed at education to promote positiveimages in society of PLWHA.


Engagement with a supportive HIV community whichwas more closely oriented to women’s particular needswas found to be of great importance. However, in realityit is not so straightforward. Firstly, the population of HIV-positive women in Australia is very heterogeneous, andwhile women want to communicate with others withsimilar experience, the diversity of the group means thatthey may not share any aspects of their identity beyondtheir HIV status. The second issue is that women perceivethat they have little in common with the majority of thecommunity of PLWHA, given that support networks havebeen principally organized by and on behalf of the gaymale community.

The strategies employed by peer-support organiza-tions require extreme flexibility and adaptability in deal-ing with the diverse population of HIV-positive women.HIV support organizations are, by their very nature,limited in resources and tend to operate in a similar way toeach other rather than offering distinctive services to theirclients. There is no competitive incentive to sharpen thequality of services available and women who use theseservices do not have many options from which to choose.Some of the key aspects mentioned in this study suggestthe need for further development of marketing strategiesand service-centered social marketing approaches to sat-isfy the needs of HIV-positive women.

The problems associated with HIV/AIDS are tooreadily perceived as relating purely to the domain ofphysical health. However, the contribution that can bemade from the services marketing and social marketingparadigms is to overcome this major “short-circuit” in thedesign and promotion of health care services caused bythe failure to take into account consumer variables such asindividual personality, socioeconomic status, existingknowledge and interests, mass media habits, and interper-

sonal communication networks (Bratic et al. 1981). Wemust add gender to this list, as it has been long overlooked.As this study reveals, such an approach will improve thedesign of support services and facilitate the developmentof gender-sensitive marketing strategies to enhance theconsumption of support services by HIV-positive women.

CONCLUSION

The process of data collection was marked by objec-tive difficulties in recruiting women with HIV/AIDS whowere willing to participate in this study. The major ob-stacle was that women consider their privacy to be a highpriority and they were therefore reluctant to disclose theirHIV status. Subsequently, the number of respondentsselected to participate in this study was constrained by thescope of the project and the closed nature of the targetcommunity. However, even though the number of respon-dents was small, their comments and responses revealhigh levels of consistency, which is important for newresearch directions.

Future studies could also investigate the needs ofwomen who are caring for HIV-positive partners or chil-dren, individuals from cultural and diverse backgrounds,or for a growing elderly population who are affected byHIV/AIDS. Finally, similar research could examine theconsumption practices of the heterosexual male commu-nity who, like their female counterparts, suffer from beinga minority within a minority.

Greater effectiveness in reaching the growing num-ber of Australian women with HIV/AIDS will assist inmaximizing good health for longer periods, minimizingdiscrimination and reducing the stigma associated withHIV/AIDS, thereby enhancing the well-being of peopleliving with the debilitating condition.

REFERENCES

Albright, R.E. (2002), “What Can Past Technology Fore-casts Tell Us About the Future?” Technological Fore-casting & Social Change, 69 (5), 443–64.

Bratic, E., R. Greenberg, and P. Petersen (1981), “HMTS:Improving the Quality of Public Service Announce-ments Through Standardized Pretesting,” Academyof Marketing Science, 9 (1), 40–51.

Bruce, M.L., W. Smith, J. Miranda, K. Hoagwood, andK.B. Wells (2002), “Community-Based Interven-tions,” Mental Health Services Research, 4 (4), 205–14.

Ciambrone, D. (2001), “Illness and Other Assaults onSelf: The Relative Impact of HIV/AIDS on Women’sLives,” Sociology of Health & Illness, 23 (4), 517–

40.Denzin, N. and Y. Lincoln (2005), “Introduction: The

Discipline and Practice of Qualitative Research,” inThe Handbook of Qualitative Research, N. Denzinand Y. Lincoln, eds. Third Edition, Thousand Oaks,CA: Sage Publications Inc.

Fournier, S. (1998), “Consumers and Their Brands: De-veloping Relationship Theory in Consumer Re-search,” Journal of Consumer Research, 24 (4), 343–73.

Lichtenstein, B., M.K. Laska, and C.J. M (2002), “ChronicSorrow in the HIV-Positive Patient: Issues of Race,Gender, and Social Support,” AIDS Patient Care andSTDs, 16 (1), 27–38.

Marshall, C. and G. Rossman (1999), Designing Qualita-tive Research. Thousand Oaks, CA: Sage Publica-


tions.Mick, D.G. (2006), “Meaning and Mattering Through

Transformative Consumer Research,” Advances inConsumer Research, 33 (2006), 1–4.

Molassiotis, A., P. Callaghan, S.F. Twinn, S.W. Lam,W.Y. Chung, and C.K. Li (2002), “A Pilot Study ofthe Effects of Cognitive-Behavioral Group Therapyand Peer Support/Counselling in Decreasing Psy-chologic Distress and Improving Quality of Life inChinese Patients with Symptomatic HIV Disease,”AIDS Patient Care and STDs, 16 (2), 83–96.

NCHECR (2004), Australian HIV Surveillance ReportSydney. University of New South Wales.

Packer, M.J. (1985), “Hermeneutic Inquiry in the Study ofHuman Conduct,” American Psychologist, 40 (10),1081–93.

Reilly, T. and G. Woo (2004), “Social Support and Main-tenance of Safer Sex Practices among People Livingwith HIV/AIDS,” Health and Social Work, 29 (2),97.

Sheth, J. and R. Sisodia (2006), Does Marketing NeedReform? Armonk, NY: London: M.E Sharpe.

Siegel, K. and H.M. Lekas (2002), “AIDS as a ChronicIllness: Psychosocial Implications,” AIDS Care, 16(4), 69–76.

Strauss, A.L. and J. Corbin (1998), Basics of Qualitative

Research, 2nd ed. Newbury Park, CT: Sage Publica-tions.

Thoits, P.A. (1995), “Stress, Coping, and Social SupportProcess: Where are We? What Next?” Journal ofHealth and Social Behavior, 35 (Extra Issue), 53–79.

Thompson, C.J., H.R. Pollio, and W.B. Locander (1994),“The Spoken and the Unspoken: A HermeneuticApproach to Under,” Journal of Consumer Research,21 (December), 432.

____________ (1997), “Interpreting Consumers: AHermeneutical Framework for Deriving MarketingInsights from the Texts of Consumers’ ConsumptionStories,” JMR, Journal of Marketing Research, 34(November), 438–55.

Tsunekawa, K., S. Moolphate, H. Yanai, N. Yamada, S.Summanapan, and J. Ngamvithayapong (2004), “Carefor People Living with HIV/AIDS: An Assessmentof Day Care Centers in Northern Thailand,” AIDSPatient Care and STDs, 18 (5), 305–14.

Uphold, C.R. and W.N. Mkanta (2005), Use of HealthCare Services Among Persons Living with HIV In-fection: State of the Science and Future Directions,”AIDS Patient Care and STDs, 19 (8), 473–85.

Waddell, E.N. and P.A. Messeri (2006), “Social Support,Disclosure, and Use of Antiretroviral Therapy,” AIDSand Behavior, 10 (3), 263–72.

For further information contact:Yelena Tsarenko

Department of MarketingMonash University

S 7.18, Chisholm Tower26 Sir John Monash Drive

P.O. Box 197, Caulfield East, VIC 3145Australia

Phone: +61.3.9903.2354Fax: +61.3.9903.1558



THE MODERATING ROLE OF MANAGERIAL ETHICAL

PREDISPOSITIONS IN DYADIC

MARKETING EXCHANGE

Kelly D. Martin, Colorado State University, Fort CollinsJean L. Johnson, Washington State University, Pullman

SUMMARY

Marketing decision makers’ cognitive styles andmoral development have been shown to influence theirability to interpret the volatile and turbulent environmentto determine which markets the firm will enter, whichcustomers they will serve, and which competitors theywill challenge (e.g., Fraedrich and Ferrell 1992). Natu-rally, these decisions have a dramatic effect on consider-able resource allocations by the firm, which heightens theimportance of marketing decision making. To date, noresearch has investigated the managerial considerationsinvolved with leveraging a product’s ethical attributes inthe marketplace. Specifically, the pressing question ofhow ethical situations and personal predispositions mightinfluence managers’ decisions made at the market inter-face has been largely neglected. Thus, how a marketingmanager’s personal ethical predispositions affect his orher willingness to leverage ethics in the marketplace,given certain conditions, is the research question centralto this study. We are interested, moreover, in how ethicalinformation about a manager’s interfirm exchange part-ner also may influence managerial investments in ethics.

In the context of dyadic marketing relationships, weare interested in the role of managers’ general ethicalpredispositions as well as their interfirm-specific ethicalpredispositions. Rather than playing a central and pre-dominant role in decision making, the literature informsthat personal ethical predispositions may interplay differ-ently depending on the availability of relevant ethicalinformation between interfirm exchange partners (Ferrelland Gresham 1985; Fraedrich and Ferrell 1992). As such,we predict that when information about an interfirmpartner’s ethical behavior is unknown to a marketingmanager, the decision process necessarily will be com-prised of greater levels of uncertainty (Gundlach andMurphy 1993). In the context of uncertainty, moreover,personal values and beliefs may emerge as more predomi-nant and influential in the managerial decision context(Hunt, Wood, and Chonko 1989). For instance, to theextent that a marketing manager believes interfirm-spe-cific ethics are an important consideration in decisionmaking, the less likely he/she may be to leverage theethical attributes of product offerings in the marketplacewhen ethical information is unavailable. Similarly, thegreater the influence of a manager’s general business

ethics beliefs, the less likely he/she may be to leverageethics at the customer interface when information isunavailable.

To test our questions, we employed a bargaininggame from experimental economics. The study contextinvolved three categories of varying ethical informationusing the investment game, also referred to as the trustgame (Berg, Dickhaut, and McCabe 1995). Dyads ofpracticing managers invested actual cash resources withone another in the form of a new product developmentventure. The participants consisted of 234 managers com-prising 117 dyads. By replicating a one-shot arm’s lengthtransaction between managers, we isolated the effects ofthe ethical information treatment as well as the effects ofpersonal predispositions. Managers indicated their likeli-hood of leveraging ethics based on the information theypossessed as well as the exchange conditions. Finally,participants responded to questions regarding their per-sonal ethical beliefs and demographics information.

Due to the dyadic design of the research and the non-parametric nature of experimental economics data,bootstrapping using Partial Least Squares (PLS) analysiswas employed for measure purification and validation, aswell as hypothesis testing. The results demonstrate thatavailable ethical information influenced both managers’investment in ethics and their stated likelihood to leverageethics in the marketplace. Neither managerial ethicalpredispositions about interfirm exchange nor general busi-ness ethics predispositions had a direct effect on manag-ers’ investment decisions. Regarding managers’ willing-ness to leverage ethics in the marketplace, however, bothinterfirm-specific ethical beliefs and general businessethics beliefs had strong, significant effects on statedwillingness to leverage. Further, depending on the ethicalinformation available, interfirm-specific ethical beliefshad a moderate effect on ethics investments, whereasgeneral ethical beliefs had no significant effect on ethicsinvestments with ethical information availability. Con-versely, both interfirm-specific ethical beliefs and generalethical beliefs had strong and substantive moderatingeffects on stated likelihood to leverage ethics in themarketplace given the available ethical information.

Interestingly, managers often failed to follow throughin their ethical investments consistent with their stated


willingness to leverage ethics for competitive advantage.For dyads in the ethical information context, managerslow in both types of ethical beliefs invested less than thosemanagers high in ethical beliefs. Although not surprisingin itself, what did draw our attention was the reversaldemonstrated by low ethical beliefs managers having amarkedly higher willingness to leverage ethics for com-petitive advantage than those high in ethical beliefs. Low

ethical beliefs managers did not want to invest as much inethics, yet hoped to garner a greater return, whereas highethical beliefs managers demonstrated behavior consis-tent with investing in ethics for the sake of doing the rightthing, rather than hoping to gain profits. The resultsdemonstrate the complex role of managerial ethical pre-dispositions on interfirm investments in ethics and thewillingness to leverage ethics in the marketplace.

For further information contact:Kelly D. Martin

Department of MarketingColorado State University

Fort Collins, CO 80523–1278Phone: 970.491.7269



PREANNOUNCING PIONEERING VS. FOLLOWER PRODUCTS: WHAT

SHOULD THE MESSAGE BE?

Christian Homburg, University of Mannheim, GermanyTorsten Bornemann, University of Mannheim, Germany

Dirk Totzek, University of Mannheim, Germany

SUMMARY

New product launches constitute one of the mostchallenging tasks managers are confronted with. Thedifficulty of this task is also reflected by the fact thatnearly 50 percent of the new products brought to marketfail to gain market acceptance. In order to maximize thechance of a successful new product launch, strategic aswell as tactical decisions have to be made and coordi-nated. One important strategic question pertains to whetherto lead or to follow in a market. Once the strategicboundary conditions for the new product launch aredetermined, the firm has to design tactical measuresrelated to the introduction of the new product. Amongthese tactical measures is the preannouncement of the newproduct. Preannouncements mainly aim at acceleratingmarket penetration once the product is launched.

Despite its practical relevance, however, prior em-pirical research has neglected whether the effectiveness ofnew product preannouncements is contingent upon orderof entry. In particular, the question whether and howpreannouncements for pioneering products should bedifferent from those for follower products in order toenhance new product success is of great interest.

Against this background, we examine potential mod-erating effects of the message focus with regard to therelationship between the intensity of preannouncing andnew product success, i.e., the effectiveness of new prod-uct preannouncements. Specifically, we investigate howdifferent message contents influence preannouncementeffectiveness for pioneers who create a new market, earlyfollowers into a still dynamic and evolving market, andlate followers into an established market.

In this context, industry observations and interviewswith managers revealed two important message empha-ses: a risk reduction focus that addresses the extent towhich preannouncements focus on reducing the per-ceived risk associated with the new product and a relativeadvantage focus that refers to the degree to whichpreannouncements focus on the product advantage com-pared to existing alternatives.

Based on diffusion theory and on the literature onorder of entry, we hypothesize that the appropriateness of

these message emphases varies with order of entry. Spe-cifically, we expect that for pioneers, a risk reductionfocus increases preannouncement effectiveness whereas,due to a lack of reference products, a relative advantagefocus does not influence preannouncement effectiveness.The aim here is to induce trial and to create primarydemand.

For early followers, we expect that both, a riskreduction focus and a relative advantage focus, increasepreannouncement effectiveness. For late followers, how-ever, we propose that only a relative advantage focus hasa positive effect on preannouncement effectiveness ascustomers are already familiar with these products, mak-ing a risk reduction focus redundant.

A cross-industry study investigating more than 150new product launches shows that depending on order ofentry, different message foci are suited to positivelyinfluence the relationship between preannouncement in-tensity and new product success. Results for pioneeringproducts show that the stronger the message focuses onrisk reduction, the stronger is the effect of preannouncementintensity on new product success. Furthermore, a highrelative advantage focus exerts a marginally significantnegative influence on the relationship betweenpreannouncement intensity and new product success,indicating that a comparison with existing products israther counterproductive for pioneers. For early follow-ers, however, results only indicate a marginally signifi-cant interaction effect for a risk reduction focus and,against our expectations, no interaction effect for a rela-tive advantage focus.

Finally, for late followers, a message focusing on therelative product advantage exerts a positive influence onthe relationship between preannouncement intensity andnew product success. In particular, results indicate that forlate followers only preannouncements which stronglyemphasize the relative product advantage lead to a posi-tive effect of preannouncements on new product success.

The key implication that arises from the present studyis that the effectiveness of preannouncements is contin-gent upon order of entry in the market and on the specificcontent of the preannouncement. Therefore, before en-gaging in preannouncements, managers have to carefully


evaluate whether their product is a pioneering product, anearly follower, or a later entrant. Results of the presentstudy provide managers with precise recommendations

on how to preannounce new products, depending on theirrespective market situation.

For further information contact:Christian Homburg

Marketing DepartmentUniversity of Mannheim

68131 MannheimGermany

Phone: +49.621.181.1555Fax: +49.621.181.1556



HOW DOES DISTANCE MATTER? THE IMPORTANCE OF

PRE-LAUNCH STAKEHOLDERS IN SPATIAL DIFFUSION

OF NEW MEDICINES

David Dekker, Erasmus University Rotterdam, The NetherlandsPaul H. Driessen, Radboud University Nijmegen, The Netherlands

SUMMARY

This paper takes the pharmaceutical industry as anempirical context to study how proximity of a market tostakeholders impacts spatial diffusion of a product inno-vation. There is an extensive body of literature showingthat geographical location is of importance in understand-ing business performance in general, and product innova-tion in particular (Bradlow et al. 2005). Geographicalaspects have been found to influence the development ofinnovations, because proximity of various stakeholdersimproves the flow of information between those stake-holders that is necessary to acquire new knowledge (Jaffeet al. 1993; Rosenkopf and Almeida 2003). Also, geo-graphical aspects have been found to influence the spatialdiffusion of innovations, because firms, retailers, andcustomers are geographically interdependent (Bronnen-berg and Mahajan 2001). Previous literature, however,has failed to identify whether the stakeholders that areimportant during the development phase can have animpact on the spatial diffusion after the product has beenlaunched.

Theory

For our theoretical framework, we draw on threetheories that explain the effects of knowledge transfer toand from innovating firms on new product developmentoutcomes (such as sales): social network theory (Colemanet al. 1966; Burt 1987; Van den Bulte and Lilien 2001;Reagans and McEvily 2003), cluster theory (Porter 1990;Audretsch and Stephan 1996), and stakeholder theory(Freeman and McVea 2001; Roome and Wijen 2006). Wepropose that geographical proximity enables tacit knowl-edge transfer between stakeholders and the firm duringthe pre-launch process, and thus can be an importantexplanation for spatial patterns of diffusion of new med-icines. Specifically, we hypothesize that proximity ofpatients (to whom the new medicines) to the locations ofpre-launch stakeholders will positively (negatively) im-pact sales of a new medicine.

Method

In order to study the hypotheses, we obtained archi-val data about the pharmaceutical market in the Nether-lands from various sources. A spatial model is estimated

for three new medicines that were selected: one radicalinnovation, one technological breakthrough, and one in-cremental innovation. We use Mantel-like tests (permuta-tion tests) that are robust against spatial autocorrelation(Mantel 1967; Oden and Sokal 1992; Anderson andLegendre 1999; Anderson and Robinson 2001; Dekkeret al. 2007). Furthermore, these tests do not require anyspecific assumptions about the nature of the spatialautocorrelation.

Results

We find that geographical location of clinical trialshas significant negative effects on the geographical salesfor the incremental innovation. For the technologicalbreakthrough and the radical innovation, clinical triallocation no statistically significant effects were found.The geographical distance to the nearest academic hospi-tal has a significant negative effect on the geographicalsales for the technological breakthrough, but a significantpositive effect for the radical innovation. The geographi-cal location of the nearest health care insurance companyhas no significant effects on the geographical sales for anyof the new medicines. The distance to the nearest regionalpatient consumer platform (uniting local patient associa-tions) has significant negative effects on the geographicalsales for the incremental innovation and the technologicalbreakthrough. Furthermore, the effect of the control vari-able geographical dispersion on geographic sales is sig-nificant and positive for the incremental innovation andthe technological breakthrough. This means that geo-graphically larger markets have higher sales for these twoinnovations.

Discussion

The results show that geographical distance betweenusers and stakeholders matter for the spatial diffusion ofan innovation in the pharmaceutical industry. For theincremental innovation, geographic proximity to clinicaltrials and patient associations increases sales. For thetechnological breakthrough, geographic proximity to anacademic hospital and patient associations increases sales.For the radical innovation, the effect is counterintuitive,as geographic proximity to an academic hospital de-creases sales, which may be explained by the ambiguitysurrounding radical innovations. An implication of our


study is that pre-launch activities should receive seriousattention. During the pre-launch process, the seeds for thespatial diffusion of innovation can be sown. Furthermore,

the study suggests that some stakeholders may have moreclout than they think.

REFERENCES

Anderson, Marti J. and Pierre Legendre (1999), “AnEmpirical Comparison of Permutation Methods forTests of Partial Regression Coefficients in a LinearModel,”Journal of Statistical Computation and Simu-lation, 62 (3), 271–303.

____________ and John Robinson (2001), “PermutationTests for Linear Models,” Australian & New ZealandJournal of Statistics, 43 (1), 75–88.

Audretsch, David B. and Paula E. Stephan (1996), “Com-pany-Scientist Locational Links: The Case of Bio-technology,” American Economic Review, 86 (3),641–52.

Bradlow, Eric T., Bart Bronnenberg, Gary J. Russell,Neeraj Arora, David R. Bell, Sri Devi Duvvuri,Frenkel Ter Hofstede, Catarina Sismeiro, RaphaelThomadsen, and Sha Yang (2005), “Spatial Modelsin Marketing,” Marketing Letters, 16 (3), 267–78.

Bronnenberg, Bart J. and Vijay Mahajan (2001), “Unob-served Retailer Behavior in Multimarket Data: JointSpatial Dependence in Market Shares and PromotionVariables,” Marketing Science, 20 (3), 284–99.

Burt, Ronald S. (1987), “Social Contagion and Innova-tion: Cohesion Versus Structural Equivalence,”American Journal of Sociology, 92 (6), 1287–335.

Coleman, James S., Elihu Katz, and Herbert Menzel(1966), Medical Innovation: A Diffusion Study. In-dianapolis: Bobbs-Merrill.

Dekker, David, David Krackhardt, and Tom Snijders(2007), “Sensitivity of MRQAP Tests to Collinearityand Autocorrelation Conditions,” Psychometrika,Forthcoming.

Freeman, R. Edward and John McVea (2001), “A Stake-holder Approach to Strategic Management,” in Hand-book of Strategic Management, Michael A. Hitt, R.Edward Freeman, and Jeffrey S. Harrison, eds. Ox-ford: Blackwell, 189–207.

Jaffe, Adam B., Manuel Trajtenberg, and RebeccaHenderson (1993), “Geographic Localization ofKnowledge Spillovers as Evidenced by Patent Cita-tions,” The Quarterly Journal of Economics, 108 (3),577–675.

Mantel, Nathan (1967), “The Detection of Disease Clus-tering and a Generalized Regression Approach,”Cancer Research, 27 (2), 209–20.

Oden, Neal L. and Robert R. Sokal (1992), “An Investiga-tion of 3-Matrix Permutation Tests,” Journal of Clas-sification, 9 (2), 275–90.

Porter, Michael E. (1990), The Competitive Advantage ofNations. London: Macmillan.

Reagans, Ray and Bill McEvily (2003), Network Struc-ture and Knowledge Transfer: The Effects of Cohe-sion and Range,” Administrative Science Quarterly,48 (2), 240–67.

Roome, Nigel and Frank Wijen (2006), “StakeholderPower and Organizational Learning in CorporateEnvironmental Management,” Organization Stud-ies, 27 (2), 235–63.

Rosenkopf, Lori and Paul Almeida (2003), “OvercomingLocal Search Through Alliances and Mobility,” Man-agement Science, 49 (6), 751–66.

Van den Bulte, Christophe and Gary L. Lilien (2001),“Medical Innovation Revisited: Social ContagionVersus Marketing Effort,” American Journal of So-ciology, 106 (5), 1409–35.

For further information contact:David Dekker

Econometric InstituteErasmus University Rotterdam

P.O. Box 17383000 DR Rotterdam

The NetherlandsPhone: +31.10.40.81.278


Paul H. DriessenNijmegen School of Management

Radboud University NijmegenP.O. Box 9108

6500 HK NijmegenThe Netherlands

Phone: +31.24.36.15.460E-Mail: [email protected]


THE EFFECT OF PRIOR EXPERIENCE AND PRODUCT TYPE ON

CHANGES IN PERCEIVED RISK FOR AN INNOVATION:

THE ROLE OF FEELINGS

Arjun Chaudhuri, Fairfield University, FairfieldCamelia C. Micu, Fairfield University, Fairfield

ABSTRACT

In the context of discontinuous innovations, the au-thors found that for utilitarian products, low experienceusers report higher change in perceived risk (CPR) thanhigh experience users. For hedonic products, there was noeffect of experience on CPR. Negative feelings mediatethe effect of prior experience and product type on CPR.

INTRODUCTION

Do new products increase or decrease the level ofperceived risk in a product category? Perceived risk hasbeen identified as a key determinant in the successfulintroduction of continuous and discontinuous innova-tions (Herzenstein, Posavac, and Brakus 2007; Hoeffler2003; Ram and Sheth 1989; Rogers 2003). A discontinu-ous innovation (DI) is defined as a newly introducedproduct which is created with new technology and whichrequires substantial changes in consumer behavior (Chandyand Tellis 1998; Robertson 1967). Thus, a hybrid carwhich involves new technology, but no changes in con-sumer behavior, would be a continuous innovation, butnot a DI. However, a car using a new technology whichproduced no gasoline emissions, but required consumersto charge the car at night and limit their acceleration toseventy miles an hour, would be a DI. Thus, DIs (alsoreferred to in the innovation literature as radical innova-tions, true innovations and really new products) have ahigh potential benefit for the consumer, but they alsoinvolve a potential high cost (in terms of price, safety,social standing, changes in behavior, etc.). The prospectof this potential for loss is perceived risk and it may createan unwillingness to adopt depending on the level of theperceived potential benefit of the DI (Moreau, Lehmann,and Markman 2001).

In this paper we use change in perceived risk (CPR)as our dependent variable. CPR indicates the directionalbent in the perceived risk of a DI and, thus, may serve asa better indicator of the likelihood of adoption thanperceived risk alone. We define CPR as the differencebetween innovation risk (the risk associated with theinnovation) and product category risk (the risk associatedwith the existing product category). These two types ofrisk are consonant with the two types of risk alreadyestablished in the risk literature in marketing – inherent

risk (risk in the product category) and handled risk (risk inchoosing among brands in the category) (Bettman 1973).Innovation risk can be higher or lower than productcategory risk. Thus, an innovation could reduce risk in theproduct category or increase it. CPR identifies the risk dueonly to the innovation after taking into account the riskdue to the existing product category. It is a useful conceptbecause it can inform managers if the innovation willincrease or decrease perceived risk, in various segmentsof consumers, for example. Even when an innovationtranscends the product category and becomes a whollynew and different product category (Gregan-Paxton andJohn 1997), the concept of CPR is still relevant. This isbecause even in a new category it is worthwhile tounderstand the relative risk in the newer category ascompared to a pre-existing category. The notion of CPRprovides a comparison standard or benchmark againstwhich to assess the loss potential to consumers of a DI.

This study contributes to the literature on DIs inseveral ways. First, we look at differences in perceivedrisk for different product types, that is, hedonic andutilitarian products. To date, no DI research has examinedthe effect of these product types on perceived risk. Sec-ond, we predict in this paper that different people mayhave different perceptions of CPR based on the amount ofprior experience they may have had with the existingproduct category. Third, we conjecture that differences inprior experience will differentially affect CPR contingenton the type of product category. Finally, we posit thatnegative feelings mediate the effect of prior experienceand product type on change in perceived risk.

The remainder of our paper is structured as follows.In the next section, we provide a review of relevantliterature and state our hypotheses. The methodology,including experimental procedures, and measurements,follows. Next, we provide the result and conclude withdiscussions and managerial implications.

HYPOTHESES

Research in consumer behavior has contemplated avariety of ways to classify products. As noted, this re-search is particularly interested in products which areeither hedonic or functional in nature (Mittal 1989; Vaughn1980, 1986). Hedonic or affective products (e.g., candy


bars, games, sports cars) are consumed primarily foraffective and sensory gratification benefits, and providemore experiential consumption, fun, pleasure and excite-ment. Functional or utilitarian products (e.g., hair dryers,washers/dryers, microwaves) deliver more cognitivelyoriented benefits and are primarily instrumental(Hirschman and Holbrook 1982; Strahilevitz and Myers1998; Woods 1960). Products have in general both ahedonic and a utilitarian side (Voss, Spangenberg, andGrohmann 2003), but a distinction between products thatare dominant on either the hedonic dimension or theutilitarian dimension can be made (Dhar and Wertenbroch2000). Thus, sodas may be both hedonic and utilitarian butthey may also be relatively more hedonic than light bulbsthat, in turn, are relatively more utilitarian than sodas (seealso Dhar and Wertenbroch 2000).

According to Mandler (1982), when an object that isdiscrepant with expectations impinges on an individual, itproduces either negative or positive feelings dependingon the extent of discrepancy and the ability of the indi-vidual to assimilate or accommodate the discrepancy. Ifthe discrepancy is slight, individuals will be able toassimilate this into their existing schema (knowledgestructure) for the product category and mild positivefeelings ensue. If the discrepancy is severe (as is likelywith a DI), one of two scenarios will ensue: (1) theindividual will switch to an alternate schema that appro-priately explains the discrepancy or (2) the individual willbe forced to make structural changes to the existingschema in order to accommodate the discrepancy. In theformer scenario, positive feelings are engendered as aresult of the “delayed congruity” (Mandler 1982, p. 23).In the latter case, negative feelings usually occur becauseof a lack of fit between schema and object. According toMandler (1982), a sense of helplessness develops whenappropriate schemas are not available and this leads tonegative feelings such as anxiety. We posit that suchnegative feelings are contingent upon consumers’ priorexperience with the product and the type of the product,and tend to increase change in perceived risk. Specifi-cally, for reasons explained below, we expect that lowexperience users, compared to high experience users, willreact more adversely to discontinuously new utilitarianproducts, but more favorably to hedonic products.

In the context of hedonic products, it is more likelythat low experience users, those with a low level ofknowledge about an existing product category resultingfrom using the product, will successfully switch to analternate schema and undergo positive feelings. Hedonicproducts have been found to be positively related to bothpositive and negative emotions (while utilitarian productshave not) suggesting that they have the capacity to engen-der both pleasurable and painful consequences (Chaudhuri2002). Consider, for example, that hedonic products such

as candy and alcohol can have both happy and unhappyoutcomes and, thus, have the potential for greater ambiva-lence in perceptions of risk given the requisite level ofknowledge in the product category. Less experiencedusers have less knowledge of the product category and areless aware of the potential for painful consequences inhedonic products. Accordingly, they are likely to be lessambivalent about a hedonic DI and switch with morefacility to another schema that is also consistent with thehedonic aspect of the product. This will produce lessnegative emotions and lower CPR. High experience users,on the other hand, because of their greater knowledge ofa hedonic product (in terms of potentially pleasant andunpleasant outcomes) will be mired to the schema thatthey are well informed about. This will create greaterambivalence to the hedonic DI and stall their processingof alternate schemas, resulting in greater negative feelingsand greater perceptions of risk.

Utilitarian products such as household cleaners donot, however, usually produce pleasure. Therefore, thereis less scope for ambivalence (due to a lack of pleasurableoutcomes). High experience users, because of their greaterknowledge of the product category, will feel more posi-tively towards the new product, leading to less perceivedrisk than for low experience users. A utilitarian DI willcreate fewer negative feelings with these users than withthe low experience users who do not possess the necessaryschemas to absorb the “newness.” These low experienceusers are hampered by their lack of schematic knowledgefor utilitarian products, resulting in greater negative feel-ings towards the DI and greater CPR than for high expe-rience users.

Thus, we expect an interaction between productexperience and product type, such that:

H1: High experience users will report greater CPR thanlow experience users for hedonic products

H2: Low experience users will report greater CPR thanhigh experience users for utilitarian products.

Finally, we expect the effect of prior experience andproduct type on change in perceived risk to be due to thenegative feelings engendered by the innovation. Specifi-cally:

H3: Negative feelings will mediate the effect of producttype and prior experience on CPR.

METHOD

In this section, we discuss the experimental proce-dure, manipulation checks, confound checks, dependentand independent variables.


Stimulus Pretest

In keeping with our definition of a DI, we chose twoinnovations which we considered to involve new patternsof consumer behavior. The two innovations chosen for thestudy were the automated highway and the AIBO. Theautomated highway is a hands free technology that allowsdrivers to read, eat, and chat while they drive their ve-hicles, while the AIBO is an electronic dog that is capableof simulating a wide range of doglike behavior andemotions. Both the automated highway and the AIBOwere expected to be sufficiently different in terms of theirutilitarian and hedonic values.

We tested the nature of the innovation in a pretestwith 25 graduate students from a northeastern university.All students read newspaper articles (see later) about bothproducts and answered several questions about them. Theorder of the newspaper articles was varied (13 studentsread the highway information first and 12 students readthe AIBO information first).

Discontinuous Innovations. Participants were firstgiven the definition of continuous/discontinuous innova-tions and then were asked to answer several questions.The extent to which the products were considered to be adiscontinuous innovation was assessed by asking subjectswhether “The product involves a new technology” and“The product requires consumers to change their con-sumption/ usage behavior” on a 7-point scale with 1 =“Strongly disagree” and 7 = “Strongly agree.” The twoitems were combined to give one discontinuous innova-tion score for the automated highway and one score for theAIBO. A MANOVA with the order in which participantsread the articles (automated highway first versus AIBOfirst) as the independent variable and the two discontinu-ous innovations scores as the dependent variables indi-cates no main effect of the article order (Wilk’s Lambda =.96, F(2, 22) = .43, p = .66). Therefore the data werecombined in further analyses. A t-test indicates that bothproducts were above the midpoint (M = 4.00) of thediscontinuous innovation scale [automated highways:M = 5.78, t(24) = 8.63, p < .001; AIBO: M = 5.08, t(24) =5.42, p < .001]. Thus, both products were considered to bediscontinuous innovations.

Hedonic/Utilitarian Dimensions. In a second pretestwith 54 undergraduate students, we examined the extentto which highaways and pet dogs are considered to behedonic or utilitarian. The HED/UT scale (Voss,Spangenberg, and Grohmann 2003) was used. The scaleincludes the following ten semantic differential responseitems measured on a 7-point scale: “Appealing/Unappeal-ing,” “Fun/Not fun,” “Exciting/Unexciting,” “Interest-ing/Uninteresting,” “Pleasurable/Not pleasurable,” “Help-ful/Unhelpful,” “Necessary/Unnecessary,” “Effective/

Ineffective,” “Functional/Not functional,” “Practical/Im-practical.”

The pretest revealed significant differences in termsof their utilitarian values (F(1, 53) = 53.7, p < .001) aswell as their hedonic values (F(1, 53) = 12.63, p < .001).Respondents rated highways (M = 5.73) to be of higherutilitarian value than pet dogs (M = 4.48) and rated petdogs (M = 5.83) to be of higher hedonic value than thehighways (M = 3.83). These results confirmed the choiceof the automated highway as more utilitarian than theAIBO and the AIBO as more hedonic than the automatedhighway.

Experimental Procedures

Participants. One hundred seventy-five individualson a small northeastern campus participated in the study.Thirty-nine percent were undergraduate students while 28percent identified themselves as white collar workers(secretaries, librarians, etc.) and the rest were blue collar(maintenance staff, etc.), professional (IT, etc.), faculty,etc. Sixty-five percent were female and the mean age was34 years.

Procedures. We used a 2 (product type: AIBO/hedonic versus automated highway/utilitarian) X 2 (expe-rience: low versus high) between subjects design. Allsubjects were approached on a one on one basis by twotrained student assistants. Subjects were approached atrandomly picked buildings on campus and asked to par-ticipate in an academic study on new products. In bothconditions (AIBO and automated highway) subjects weregiven a self-administered questionnaire. The question-naire contained instructions, a description of the innova-tion in an article format, and the measures of the depen-dent and control variables. The questionnaire concludedwith several demographic questions.

Participants were first asked to evaluate their priorexperience with the product category (highways/pet dogs)and the perceived risk with the product category. Next,they read an excerpt from a news article that offered adescription of a DI (automated highway/AIBO). Immedi-ately after reading the description, they were asked onceagain to complete a series of measures – perceived riskwith the innovation, familiarity with the innovation, andnegative feelings, and other measures not pertinent to ourstudy. We used actual news reports, instead of constructedreports, to generate greater external validity for the studyin terms of professionally written, “real world” stimuli.Both reports were from The New York Times.

Measurement: Independent Variables

Prior Experience. Prior experience was measuredusing three 7-point scale items (i.e., “I consider myself


knowledgeable about highways/pet dogs,” “I have fre-quently used/owned highways/pet dogs,” and “I am fa-miliar with highways/pet dogs”). The three items weresummed to give a prior experience score (Cronbach’s α =.86). The independent variable of prior experience wascreated with a median split (median = 16).

Product Type (Hedonic versus Utilitarian). Producttype was a manipulated variable (AIBO = hedonic andhighway = utilitarian). The number of participants per cellvaried from 37 to 47 (low experience-hedonic product =44; low experience-utilitarian product = 44; high experi-ence-hedonic product = 37; high experience-utilitarianproduct = 47).

Measurement: Dependent Variables

Change in Perceived Risk (CPR). Consumers mayperceive a combination of various types of risk (physical,financial, performance) when faced with a DI for the firsttime (Ram and Sheth 1989). Thus, we measured per-ceived risk with the product category with the followingfour 7-point scale items, where 1 = “Completely disagree”to 7 = “Completely agree” (“Highways/pet dogs couldcause me physical harm,” “Highways/pet dogs couldcause me financial loss,” “Highways/pet dogs could per-form poorly and let me down,” and “Overall, highways/pet dogs are risky”) (Kaplan, Szybillo, and Jacoby 1974).We adapted the same items to measure perceived risk withthe innovation (automated highway/AIBO). We used thedifference between perceived risk with the DI and per-ceived risk with the category as the score for CPR. Thus,a greater score indicates a higher perceived risk with theinnovation relative to the risk with the product category.

Negative Feelings. Negative feelings were measuredusing two 7-point scale items, where 1 = “Not at all” and7 = “Very much” (i.e., “The automated highway/AIBOmakes me feel anxious,” “The automated highway/AIBOmakes me feel worried”). The two items were summed togive a negative feelings score (r = .67, p < .05).

Measurement: Covariates

A review of the innovation literature revealed severalvariables that have been found to affect individuals’response to innovations.

Familiarity. Familiarity with the DI could causepositive or negative feelings and increase or reduce CPR.Familiarity was measured using two 7-point scale items(i.e., “I had knowledge of the automated highway/AIBObefore today,” and “I have heard about the automatedhighway/AIBO before today”) with 1 = “Completelydisagree” and 7 = “Completely agree.” A product famil-iarity score was computed by summing the two items (r =.84, p < .001).

Age. Experience and age may be confounded to-gether. Previous research indicates that age is an impor-tant variable used to predict new-product adoption behav-ior and response to innovativeness (see Im, Bayus, andMason 2003). Thus, we attempted to reduce the “noise”associated with its effects by measuring participants’ age(in years).

The covariates were selected on a theoretical basis,but were included in the final analysis only if they werestatistically significant in adjusting the dependent vari-able (Tabachnick and Fidell 1989, p. 345–346).

RESULTS

H1 and H2

To test H1 and H2, we ran a 2 (product type: hedonicversus utilitarian) X 2 (prior experience: high versus low)analysis of variance of subjects’ change in perceived risk(CPR). Familiarity and age were not included as covariatesin the analysis because of their non-significant effect. Theanalysis indicated a non-significant effect of prior experi-ence [F(1, 164) = .48, p = .49], a significant effect ofproduct type [F(1, 164) = 6.46, p < .05], and a significantinteraction [F(1, 164) = 7.09, p < .01].

We followed up with post-hoc comparisons to test thespecific hypotheses. H1 hypothesized that high experi-ence users will report higher CPR than low experienceusers for hedonic products. The results do not support ourfirst hypothesis. Even though the difference is in theexpected direction [high experience users perceiving morerisk with the hedonic innovation relative to the risk withthe product category (M = -2.86) than the low experienceusers (M = -5.47)], the difference is not significant [t(77) =1.37, p = .18].

H2 stated that, for utilitarian products, low experi-ence users will report greater CPR than high experienceusers for utilitarian products. The post-hoc comparisonindicates support for our second hypothesis. Specifically,for utilitarian products, low experience users reportedmore risk with the innovation relative to the risk with theproduct category (M = 1.42) than the high experienceusers [M = -3.02, t(87) = 2.42, p < .05] (see Figure A-1).

H3

To test our third hypothesis (negative feelings willmediate the effect of product type and prior experience onCPR), we conducted three steps using regression analysis.First, we conducted a regression with CPR as the depen-dent variable and product type and prior experience as thepredictor variables. The last two variables were bothdummy coded (utilitarian = 0, hedonic = 1; low experi-ence = 0, high experience = 1). Only the effect of product


Figure A-1

Change in Perceived Risk (CPR)

Low Experience (Dashed Line) and High Experience (Solid Line) Means by Product

Type

-5.47

-2.86

1.42

-3.02

Utilitarian product Hedonic product

CP

R

type was significant (Beta = -.19, p < .05) indicating thatthe mean CPR of the utilitarian product was significantlyhigher than the mean of the hedonic product. Second, weconducted the same analysis but with negative feelings asthe dependent variable. In both cases, once again, onlyproduct type was significantly related to negative feelings(Beta = -.43, p <. 001). Finally, we regressed CPR onceagain on the two dummy coded independent variables, butthis time we also included negative feelings as a predictor.Negative feelings was significantly related to CPR (Beta =.23, p < .01) and, most importantly, the effect of innova-tion type now became non-significant (Beta = -.09, p >.05). This indicates that the effect of product type on CPRwas completely mediated by negative feelings, providingsupport for H3.

DISCUSSIONS AND MANAGERIAL

IMPLICATIONS

This research contributes to the literature on discon-tinuous innovations in several ways. First, unlike previ-ous DI research, we examine the differences in perceivedrisk for hedonic versus utilitarian products. Second, we

predict that consumers’ prior experience with the innova-tion affects change in perceived risk and expect an inter-action effect of prior experience and product type on CPR.Finally, we hypothesize that negative feelings mediate theeffect of prior experience and product type on change inperceived risk. In this section, we summarize the hypoth-eses and the findings, and introduce explanations for theresults that were not supported.

We hypothesized an interaction between producttype and prior experience. Specifically, high experienceusers, compared to low experience users, should expresshigher change in perceived risk (CPR) for hedonic prod-ucts (H1), but lower CPR for utilitarian products (H2).The difference between high and low experience users inchange in perceived risk was not significant for hedonicproducts, thus providing no support for H1. First, anexplanation for this may be that hedonic products areconsumed for sensory gratification purposes, and theyprovide fun, pleasure and excitement, regardless of theconsumer’s level of experience with that product cat-egory. These products may indeed engender painful con-sequences, but because of the pleasurable nature of the

FIGURE A-1

Change in Perceived Risk (CPR)

Low Experience (Dashed Line) and High Experience (Solid Line)

Means by Product Type


product, the painful consequences may not be more sa-lient for high experience users than for low experienceusers, resulting in similar reactions to the hedonic innova-tion. Second, the negative signs associated with change inperceived risk for the hedonic product indicate that par-ticipants found the AIBO as being less risky than thegeneral product category (i.e., pet dogs). Research isneeded to examine the effect of prior experience on CPRwhen the risk with the discontinuous innovation is higherthan the risk with the hedonic product category.

Consistent with H2, we found that, for utilitarianproducts, low experience users reported more perceivedrisk with the innovation relative to the risk with theproduct category than high experience users. Finally, andconsistent with H3, we found that prior experience andproduct type may not have a direct effect on change inperceived risk, but that the effect may be mediated bynegative feelings that arise after reading about discontinu-

ous innovations. Various strategies, such as providingwarranty and company reputation information (seeBearden and Shimp 1982) could help reduce the negativefeelings and perceived risk that consumers report inrelation to a discontinuous innovation.

Our findings also imply that prior experience may bean effective marketing and segmentation tool in the pro-motion of hedonic versus utilitarian products. For hedonicproducts, both high and low experience users respondsimilarly to the same marketing mix. However, for utili-tarian products marketers should differentiate betweenthe two groups. Specifically, marketers should focusstrategies to reduce risk (e.g., testimonials from credibleendorsers, free trial, company reputation information) onlow experience users first, since they were the onesreporting increased risk with the innovation relative to theproduct category.

REFERENCES

Bearden, William O. and Terence A Shimp (1982), “TheUse of Extrinsic Cues to Facilitate Product Adop-tion,” Journal of Marketing Research, 19 (May),229–39.

Bettman, James R. (1973), “Perceived Risk and Its Com-ponents: A Model and Empirical Test ,” Journal ofMarketing Research, 10 (May), 184–90.

Chandy, Rajesh and Gerard J. Tellis (1998), “Organizingfor Radical Product Innovation: The OverlookedRole of Willingness to Cannibalize,” Journal ofMarketing Research, 35 (November), 474–87.

Chaudhuri, Arjun (2002), “A Study of Emotion andReason in Products and Services,” Journal of Con-sumer Behaviour, 1 (3), 267–79.

Dhar, Ravi and Klaus Wertenbroch (2000), “ConsumerChoice Between Hedonic and Utilitarian Goods,”Journal of Marketing Research, 37 (February), 60–71.

Gregan-Paxton, Jeniffer and Deborah Roedder John(1997), “Consumer Learning by Analogy: A Modelof Internal Knowledge Transfer,” Journal of Con-sumer Research, 24 (December), 266–84.

Herzenstein, Michal, Steven S. Posavac, and Josko J.Brakus (2007), “Adoption of New and Really NewProducts: The Effects of Self-Regulation Systemsand Risk Salience,” Journal of Marketing Research,44 (May), 251–60.

Hirschman, Elizabeth C. and Morris B. Holbrook (1982),“Hedonic Consumption: Emerging Concepts, Meth-ods, and Propositions,” Journal of Marketing, 46(Summer), 92–101.

Hoeffler, Steve (2003), “Measuring Preferences for Re-

ally New Products,” Journal of Marketing Research,40 (November), 406–20.

Holbrook, Morris B. and Elizabeth C. Hirschman (1982),“The Experiential Aspects of Consumption: Con-sumer Fantasies, Feelings, and Fun,” Journal ofConsumer Research, 9 (September), 132–40.

Im, Subin, Barry L Bayus, and Charlotte H. Mason (2003),“An Empirical Study of Innate ConsumerInnovativeness, Personal Characteristics, and New-Product Adoption Behavior,” Journal of the Acad-emy of Marketing Science, 31 (1), 61–73.

Kaplan, Leon B., George J. Szybillo, and Jacob Jacoby(1974), “Components of Perceived Risk in ProductPurchase: A Cross-Validation,” Journal of AppliedPsychology, 59 (3), 287–91.

Mandler, George (1982), “The Structure of Value: Ac-counting for Taste,” in Cognition and Affect, Marga-ret Sydnor Clark and Susan T. Fiske, eds. Hillsadale,NJ: Lawrence Erlbaum Associates, Publishers, 3–36.

Mittal, B. (1989), “A Theoretical Analysis of Two RecentMeasures of Involvement,” in Advances in Con-sumer Research, Thomas K. Srull, ed. Provo, UT:Association for Consumer Research, 16, 697–702.

Moreau, C. Page, Donald R. Lehmann, and Arthur B.Markman (2001), “Entrenched Knowledge Struc-tures and Consumer Response to New Products,”Journal of Marketing Research, 38 (February), 14–29.

Ram, S. and Jagdish Sheth (1989), “Consumer Resistanceto Innovations: The Marketing Problem and its Solu-tions,” Journal of Consumer Marketing, 6 (Spring),5–14.

Robertson, Thomas S. (1967), “The Process of Innova-tion and the Diffusion of Innovation,” Journal of


Marketing, 31 (January), 14–19.Rogers, Everett M. (2003), Diffusion of Innovations. New

York: Oxford University.Strahilevitz, Michal and John G. Myers (1998), “Dona-

tions to Charity as Purchase Incentives: How WellThey Work May Depend on What You Are Trying toSell,” Journal of Consumer Research, 24 (4), 434–46.

Tabachnick, G. Barbara and Linda S. Fidell (2001), UsingMultivariate Statistics. Harper & Row Publishers,New York.

Vaughn, Richard (1980), “How Advertising Works: A

Planning Model,” Journal of Advertising Research,20 (October), 27–34.

____________ (1986), “How Advertising Works: A Plan-ning Model Revisited,” Journal of Advertising Re-search, 26 (February/March), 57–63.

Voss, Kevin E., Eric R. Spangenberg, and BiancaGrohmann (2003), “Measuring the Hedonic and Utili-tarian dimensions of Consumer Attitude,” Journal ofMarketing, 40 (August), 310–20.

Woods, W. (1960), “Psychological Dimensions of Con-sumer Decision,” Journal of Marketing, 24, 15–19.

For further contact information:Arjun Chaudhuri

Charles Dolan School of BusinessFairfield University

1073 North Benson RoadFairfield, CT 06824

Phone: 203.254.4000, Ext. 2823Fax: 203.254.4105



THE ROLE OF CONSUMERS’ NEED FOR UNIQUENESS IN THE

INNOVATION ADOPTION PROCESS

Machiel J. Reinders, Agricultural Economics Research Institute, The NetherlandsRuud T. Frambach, VU University Amsterdam, The Netherlands

Mirella H.P. Kleijnen, VU University Amsterdam, The Netherlands

SUMMARY

Although much research has focused on understand-ing the drivers of innovation adoption, relatively little isknown about the determinants underlying rejection ordiscontinuance of innovation adoption. In this study weinvestigate the role of consumers’ need for uniqueness inthe innovation adoption process as potential driver ofdiscontinuance once more widespread diffusion has oc-curred. We do so in the context of different consumptiongoals.

Adoption Process

New products are first adopted by a relatively smallgroup of consumer innovators who then influence lateradopters (Rogers 1995). Adopting new products beforeothers do is one way to satisfy someone’s need to beunique (Roehrich 2004; Lynn and Harris 1997a, 1997b;Burns 1993). However, people who experience a need foruniqueness not only adopt new products earlier than otherpeople, but they also will lose interest in, or discontinueuse of innovations that become too commonplace in orderto reestablish their distinctiveness from other people(Snyder 1992; Tian et al. 2001). More widespread diffu-sion of an innovation is typically triggered by the point oftake-off (Golder and Tellis 1997). It refers to the pointwhen innovations enjoy a dramatic increase in sales andbegin to spread throughout society. We expect that con-sumers highly seeking uniqueness will be attracted by theinnovation before its widespread diffusion (marked by thepoint of take-off), but will be more likely to avoid ordispose the innovation after take-off. We therefore hy-pothesize that need for uniqueness has a positive effect onthe adoption of new products before the take-off point anda positive effect on (a) the rejection and/or (b) the discon-tinuance of new products after the take-off point.

Consumption Goals

The main effects that we hypothesize may be affectedby the goals consumers pursue with respect to innovationadoption. We therefore investigate the potential moderat-ing effects of consumption goals on the relationshipbetween need for uniqueness and the adoption process. Asconsumers do not operate in isolation but often are func-tioning within social groups, they may make choices in

which group interests prevail over individual interests(Mackie and Goethals 1987). This implies that we candistinguish between individual-oriented goals and group-oriented goals. Furthermore, we can classify consumptiongoals as functional, i.e., consumers buy products becauseof the pragmatic benefits that can be derived, or symbolic,i.e., consumers buy products because of what they mean(Escalas and Bettman 2005). Depending on which goalsare predominantly present in a particular decision makingprocess, we expect diverging effects of need for unique-ness on decisions related to adoption, rejection, and(dis)continued use of an innovation. First of all, whenfunctional goals are activated, consumers focus more onthe benefits that an innovation could provide in helpingthem to accomplish specific acts rather than the fact thatinnovations could act as a vessel of uniqueness (Holt1995; Ligas 2000). In addition, when functional groupgoals are activated consumers focus more on collectiveachievements or tasks instead of emphasizing their unique-ness. Therefore, we expect that when both individual-oriented and group-oriented functional goals are acti-vated, the effect of a consumer’s need for uniqueness onthe adoption process is attenuated. In contrast, on anindividual level, psychosocial consumption goals are re-lated to one’s self-concept. In order to create a unique self-concept the need to differentiate oneself from other peopleis heightened (Escalas and Bettman 2005). Therefore, wepredict that psychosocial individual consumption goalsenhance the effect of need for uniqueness on adoption,rejection, and discontinuance of new products. On a grouplevel, psychosocial consumption goals are related to so-cial identification (Turner 1986; Bhat and Reddy 1998).The main difference between individual psychosocialgoals and group psychosocial goals is that when the firsttype of goals are activated a consumer’s focus is on theirself-concept whereas when the second type of goals areactivated a consumer’s focus is on their social group andthey tend to identify with and conform to the norms oftheir group. Therefore, we argue that when this type ofgoals is activated the effect of a consumer’s need foruniqueness on discontinuance is attenuated.

Empirical Studies

We conduct two empirical studies to test our hypoth-eses. In the first study we test the impact of the consumer’sneed for uniqueness on the adoption process. Therefore,


respondents first have to examine a number of innovativeproducts. In order to incorporate the right stimuli in ourstudy, we composed a list of innovations that were as-sessed by seven experts in the field (both people fromacademia as well as industry) on their rate of diffusion.Based on the expert ratings we composed a list of tenproducts. The main dependent variables of our study arethe different outcomes of the adoption process includingrejection and discontinuance (Bagozzi and Lee 1999).Our focal independent variable is Need for Uniqueness.The original Need for Uniqueness scale of Tian et al.(2001) contained over 30 items distributed over threedimensions (Creative Choice Counterconformity, Un-popular Choice Counterconformity, and Avoidance ofSimilarity). We conducted a pre-test among 160 fourth-year business students to reassess the performance of thescale in the current setting and re-assess the relevance of

each single item in the scale. We applied a rather strictprocedure in maintaining only items that loaded higherthan .60 on the main factor and less than .30 on any otherfactor. In total, this resulted in a shorter scale containing11 items distributed over the three dimensions. In addi-tion, we incorporated a number of consumer traits ascontrol variables. The results are analyzed by means of amultinomial logit model. The data collection process forthe first study is currently in process and will be finalizedOctober 2007.

In the second study we further elaborate on theconsumption motives that determine a consumer’s dis-continuance or continuance decisions by means of acontrolled (real-life) experiment in which the differentgoals are primed. References are available upon request.

For further information contact:Machiel J. Reinders

LEI Agricultural Economics Research InstituteP.O. Box 29703

2502 LS The HagueThe Netherlands

Telephone: +31.70.335.81.04Fax: +31.70.335.81.99



AN EMPIRICAL INVESTIGATION OF THE INFLUENCE OF

CORPORATE SOCIAL CAPITAL ON CONSUMERS’

COMPANY AND PRODUCT PERCEPTIONS


Barbara Gassmann, Leibniz University of Hannover, Germany

SUMMARY

The topic of social capital in general and in thecontext of organizations has gained growing interest inthe strategic management literature. Considering this back-ground, this paper aims at an empirical investigation of thedirectionality and degree of the influence of corporatesocial capital on company and product perceptions andconsumer behavior. Our research addresses several im-portant issues: First, the impact of a company’s identifica-tion as “a highly interlinked non-economic and economicnetwork member” on the directionality and magnitude ofconsumers’ perceptions of both the company and itsproducts and services are investigated. Secondly, weinvestigate the degree of transferability to the consumerbehavior level. That is, to what degree do consumersreally perceive and appreciate that a company possesses ahigh amount of social capital.


We suggest that the number, content, and quality ofa company’s network relations constitute its social capitalwith regard to relational and reputational aspects whichinfluence customer perceived value in terms of credibil-ity, reliability, responsibility, and trustworthiness as wellas consumer behavior in terms of customer loyalty andsatisfaction. In this context it has to be stated that consum-ers often transfer the perceptions of corporate reputation,trustworthiness, credibility, etc. onto the company’s prod-ucts or services what can possibly change opinions andinfluence purchase intentions. To investigate customervalue, it is important to comprise its functional (quality,price/value), emotional (affective states), and social valuedimension (customer‘s social network). Describing a com-pany as a network member with various economic andnon-economic network links and for that reason withrelational as well as reputational capital may especiallyenhance the emotional and social dimension of a customer’sglobal attitude toward both the firm and its products/services.

Methodology

To measure the behavior, attitudes, and perceptionsof consumers in relation to our multidimensional frame-

work, we developed a questionnaire with three differentcompany descriptions as stimuli and related questions toevaluate the company. Each company description wasidentical except for the fact that, the company was de-scribed as (1) a company that is successfully involved ina variety of different effective and crucial networks aswell as a high level of social responsibility engagement,which provides credibility for labeling it a “non-economicand economic network member,” or (2) a company that is,compared to its competitors, on the leading edge in theconsumer marketplace, referring to the quantity and qual-ity of the economic network relations, which providescredibility for labeling it as an “economic network mem-ber,” or (3) simply briefly described with the same back-ground text without either cue information (the controlcondition) and is labeled as “neutral.” In a total number of240 interviews the respondents were initially given one ofthe three company descriptions and evaluated the com-pany on five-point bi-polar evaluative scales linked toperceived customer value as well as buying and recom-mending intention.

Main Results

Focusing on the link between corporate relationaland reputational dimensions, this paper examines the roleof social capital in determining customer perceived value.We hypothesize that corporate social capital is rooted inthe presence and quality of a certain set of networkrelations in which a company is embedded and thatrelational and/or reputational capital dimensions have asignificant influence on what is perceived and expectedfrom the company by the customers. In order to test ourhypotheses, an ANOVA was examined for each of thethree experimental conditions for the dependent measuresrelated to company and product attitude. Our experimen-tal research shows that when a company is described as aneconomic and non-economic network member with ahigh amount of relational as well as reputational socialcapital, attitude toward the company is enhanced, and thiseffect does impact favorably upon the evaluation of theproduct, but does not impact the perceived quality of thecompany and its products. Nevertheless, social capital asa unique relational and reputational resource has a signifi-cant impact on consumer behavior in terms of buying andrecommendation intentions. The results showed the stron-


gest effect with reference to the evaluation of the company’sreputation, reliability, social responsibility, and trustwor-thiness as well as customers’ intention to recommend thecompany. Therefore, a company’s social capital resources

effectively enhance customer perceived emotional andsocial value and contribute positively to consumer behav-ior intentions. References are available upon request.


Institute of Marketing and ManagementLeibniz University of Hannover


GermanyPhone: ++49.511.762.4862

Fax: ++49.511.762.3142E-Mail: [email protected]


ASSESSING LOYALTY PROGRAM EFFECTS IN A COMPETITIVE

SETTING: IMPACT OF MARKET SATURATION, MARKET

SHARE, AND CATEGORY EXPANDABILITY

Yuping Liu, Old Dominion University, NorfolkRong Yang, SUNY – College at Brockport, Brockport

SUMMARY

Although more than 26 years have passed sinceAmerican Airlines introduced the first modern loyaltyprogram, the enthusiasm from consumers to embracesuch programs has not abated but rather has rapidly grownover the recent years. It is estimated that more than 75percent of U.S. consumers are enrolled in at least oneloyalty program (Knowledge@Wharton 2006). In 2006,total loyalty program enrollments in the United Statesreached 1.5 billion, representing an increase of 35.5percent from 2000 (Ferguson and Hlavinka 2007). Thefavorable responses from consumers in turn encouragedmany businesses to install loyalty programs as a corecomponent of their marketing strategy.

The proliferation of loyalty programs in the market-place, however, has spawned intense competition amongrival programs. Under such competitive situations, theeffectiveness of a loyalty program is likely to depend notonly on the design and management of the program itselfbut also on the market climate and the offering firm’scompetitive position. In contrast with this market reality,existing studies of loyalty programs have often consid-ered a single program in isolation and have ignored theeffects of competition on a program’s performance. Ad-dressing this gap in the literature, we conducted twostudies to examine the moderating effects of market- andfirm-level factors on loyalty program success. Our resultsshowed an overall positive effect of a loyalty program onthe offering firm’s business. However, this positive effectdoes not benefit every firm equally. Firms with a largermarket share gain more from their loyalty programs thando firms with a smaller market share. Combined with the“double jeopardy” phenomenon documented in the litera-ture (Ehrenberg et al. 1990), this suggests that not only aresmaller firms at a disadvantage in terms of customerpurchase frequency, but they may also gain less fromstrategic moves such as loyalty programs. This creates a“triple jeopardy” situation for smaller firms.

Our research also revealed that the level of saturationof loyalty programs in the marketplace negatively affectsthe performance of individual programs. However, thiseffect is contingent on the expandability of the product

category. While market saturation has a negative effectunder low category expandability, its effect disappearsunder high category expandability. In this latter situation,loyalty programs allow firms to compete not only withrivals in the same industry but also with competitors inalternative product categories. As a result, saturationbecomes less of a threat to the success of each individualprogram.

Our findings offer useful guidance to managers ofloyalty programs and suggest the importance of consider-ing competitive forces in assessing the effects of suchprograms. In deciding whether to offer a loyalty programand in subsequent management of such a program, amanager should take into account the relative marketposition of the firm in order to accurately gauge thepotential benefits that the firm will gain from the program.Smaller firms may have to try special tactics, such asincreasing program reward ratio and offering more re-ward redemption options, to remain competitive withtheir loyalty program offerings. Larger firms, on the otherhand, may want to build on their share advantage and offerloyalty programs as a preemptive move to fend againstsmaller competitors.

The saturation effect found in our study suggests aneed to taper the urge to launch even more loyalty pro-grams in an already saturated marketplace. In sectors witha large number of loyalty programs, such programs maybe more of a defense mechanism and a necessary cost ofdoing business, and their incremental contribution tofirms’ bottom-line will be limited. An important contin-gency factor in this scenario, however, is the expandabilityof the general market. In industries with highly expand-able demands, such as air travel, convenience store, andcertain financial services sectors, the saturation effect isweakened, and a loyalty program can be effective evenunder a high level of saturation. Thus, managers consid-ering the launch of a new loyalty program should examinethe extent to which their products and services can satisfyneeds fulfilled by other industries or subsectors. Whensignificant expanding opportunities exist, incrementalcontribution of a loyalty program can still be expectedeven in relatively saturated markets.


REFERENCES

Ehrenberg, Andrew S.C., Gerald J. Goodhardt, and T.Patrick Barwise (1990), “Double Jeopardy Revis-ited,” Journal of Marketing, 54 (3), 82–91.

Ferguson, Rick and Kelly Hlavinka (2007), Quo Vadis:Sizing Up the U.S. Loyalty Marketing Industry.

Milford, OH: Colloquy.Knowledge@Wharton (2006), “The Lowdown on Cus-

tomer Loyalty Programs: Which Are the Most Effec-tive and Why,” (accessed on July 13, 2007), [avail-able at http://knowledge.wharton.upenn.edu/article.cfm?articleid=1545].

For further information contact:Yuping Liu

Department of MarketingCollege of Business and Public Administration

Old Dominion UniversityNorfolk, VA 23529



USING SEGMENT ATTRACTIVENESS TO IMPROVE SEGMENT

SELECTION IN THE CREDIT CARD BUSINESS

Ranjeet Khaira, Monash University, Australia

ABSTRACT

The traditional approach to segmentation has beenfocused on homogenizing customers based on common-alities. However, being able to homogenize customersegments does not necessarily mean that segments thathave been identified are of value or are attractive to thebusiness. To ensure that segments that are developed areof value or attractive to the business a different approachbased on relative-efficiency, through Data EnvelopmentAnalysis, is proposed. The approach is new and can beadopted by businesses that have innovative data manage-ment systems. It is also timely as there is an increasingneed to ensure that resources are used efficiently inmarketing.

INTRODUCTION

Markets are extensively heterogeneous and it hasbeen suggested that, “. . . markets are defined jointly alonga customer dimension that describes the collection ofpossible customers, and a product and service dimensionthat arrays the competing choice alternatives.” Given theheterogeneous nature of the market, businesses haveturned to segmentation to identify target markets, since itwas first developed in 1954 by Wendell R Smith when hehighlighted that,

“market segmentation, consists of viewing a hetero-geneous market as a number of smaller homogenousmarkets in response to differing product preferenceamong most important market segments . . . is attrib-uted to the desires of consumers or users for moreprecise satisfaction of their varying wants” (Smith1956, p. 6).

It has been suggested that while markets are hetero-geneous, customers that share common needs can beaggregated or homogenized, and that “individual differ-ences in choice behavior can be used to group customersinto homogenous subsets referred to as market segments.”Businesses utilize segmentation to identify homogenoussegments in markets that are heterogeneous. This enablesbusinesses to direct resources more aptly and gives seg-mentation normative or prescriptive capabilities.

However, being able to identify homogenous seg-ments does not necessarily mean that the segments that aredeveloped are of value to the business. Often homogene-ity and the value of customer and customer segments are

two different dimensions, and are not linked. Therefore itis suggested that when conducting segmentation it isimportant to recognize that,

I. Not all segments that are homogenized are equally ofvalue or attractive to the business,

II. Not all businesses share the same definition of valueand this could mean that the value of a segment tobusinesses could be different,

III. Customers that share similar characteristics within asegment could differ in terms of value or attractive-ness to the business, and

IV. There could be numerous different segments thatexist among customers that are similar in terms ofvalue or attractiveness to the business.

This suggests that when conducting segmentationand segment selection, there are additional steps thatshould be integrated that addresses the value or attractive-ness of the customer or customer segment.

The importance of customer segments values orattractive to businesses is becoming more important incurrent times. This is evident even in the manner in whicha segment is being described, which is “a relativelyhomogenous grouping where the members of that seg-ment share some similar characteristics, which is of valueto the organization.” The emphasis on the value of seg-ments is becoming more apparent as there is increasingfocus on efficient resource allocation. This probably stemsfrom the overall pressure that marketing as a disciplinehas been facing, where it has been suggested that,

“. . . the marketing function is under intense scrutinyand escalating criticism from many quarters. CEO’sare questioning whether marketing adds value to thecorporation and its shareholders commensurate withits costs.”

Identifying customer and customer segments that areof value to the business is not an easy task. However, thereare two developments that could possibly make this taskless formidable.

The first is the development in IT and Data Ware-housing, which categorizes the market into two maingroups that consist of new prospective customers and


existing customers. For new customers, existing cannedsegmentation schemes such as Personicx, Mosaic, Prizm,that are developed on the market as a whole can be usedto target new customers. However, these approaches arenot exclusive to any business, and businesses can avail ofthe same information from vendors. For existing custom-ers, there are internally built segmentation schemes basedon data from existing customers. The second is to recog-nize that there has been increasingly more research onmetrics that businesses measure and track. There aremany different groups of metrics, such as financials,competitive markets, consumer behavior, consumer inter-mediate, direct trade customer and innovation and thesemetrics can be used to depict on the business needs that areof value or important to the business. While these metricsdo not represent the needs of the customers directly, someof these metrics provide the business insights on the needsof the customer. For example, data related to credit cardpurchases can be used to understand the behavior of thecustomer and the needs of the customer that has been metby the business.

By using these two developments (which are thedevelopment of IT and the availability of metrics) to-gether with a quantitative approach that focuses on deriv-ing the value of customer segments, it is possible toidentify segments that are attractive or of value to thebusiness. To demonstrate this approach, an effort will bemade to conduct this in a credit card business, whichtraditionally collates extensive customer data.

OBJECTIVE

The objective of this research is to add a differentslant to the traditional approach of segmentation, whichfocuses on homogenizing customers based on commonal-ity. In this approach, an effort is made to focus on the valueor attractiveness of the customer and customer segment tothe business.

This research differs from traditional needs-basedsegmentation, as it focuses more explicitly on the needs ofthe business. As it is conducted on existing customers,there is an indirect assumption that the business has beensuccessful in meeting the needs of the customers. In thisway, it can be used to depict the needs of the customer thathas been met. This research focuses on dimensions thatare critical to the business when making decisions relatedto the targeting of existing customers. By being able todifferentiate the attractiveness or value of the customer tothe business based on different dimensions simultaneously,the business is able to direct resources in a more efficientmanner.

In this research, we will also see how the concept ofrelative-efficiency, derived through Data EnvelopmentAnalysis, lends itself as a measure of attractiveness. This

technique has not been used in segmentation to-date andis selected as it allows for multiple behavioral and perfor-mance metrics simultaneously when determining the rela-tive efficiency of a customer. By understanding the rela-tive-efficiency or attractiveness of the customer to thebusiness, customers that are similar in terms of attractive-ness can be grouped before commonalities are used toidentify the different segments that are similar in terms ofattractiveness to the business. This approach allows forbusinesses to be able to distinguish the relative-efficiencyof the customer or the attractiveness of the customer to thebusiness.

SEGMENTATION IN THE CREDIT

CARD MARKET

The credit card is perhaps one of the most popularpayment instruments available today. In the U.S. alone itis believed that there are over 641 million credit cards incirculation, which account for $1.5 trillion of consumerspending. Most consumers have more than one credit cardand the competition among credit card issuers is so exten-sive that there are constantly credit cards being developedoffering unique features, such as no fee, balance transfer,additional loan extensions (through phone verification),rewards for shopping, air travel, cash usage, internetusage and other benefits. As the competition among creditcard issuers interest the ability to be profitable is oftenmore challenging. One way to overcome this is to becomemore selective on the customers that are targeted.

Traditionally profit is derived based on revenue lesscost. In the credit card business, revenue is earned frominterest and fees. Interest is earned based on customerbehavior, such as the customer spending and revolving onthe credit extended to the customer. Fees, on the otherhand, is earned in the form of annual fee, late fee, cashusage fee, over-limit fees and interchange fee, which is asmall portion of the retail spend that the merchant pays tothe business. Cost on the other hand can be in the form offixed and variable cost. Fixed cost encompasses all thefixed expenses such as IT, marketing, staff, premises thatthe business incurs to keep the business running. Variablecost includes cost of funds that is required by the businessto cover the cost of providing credit and the cost of creditlosses, in the form of write-offs from credit decisions andfraud.

The profitability of a credit card business is veryclosely linked to customer behavior and the economy.Changes to both customer behavior and the economy canhave a direct impact on profitability of a credit cardbusiness. To be profitable, it is critical for the business toacquire and maintain the right customer and customersegments. Traditionally, credit card businesses depend onsegmentation schemes to achieve this. For new custom-ers, businesses often depend on canned segmentation


schemes such as Prizm, Claritas, Personicx, Mosaic toidentify customers from a specific target market. How-ever, these segmentation schemes are available to all theissuers in the business and do not provide the businesswith competitive advantage in terms of targeting. In termsof existing customer, most credit card issuers utilize dataon existing customers to develop internal segmentationschemes, such as early engagement segmentation thatattempts to identify the early needs of the customers andplace them into different segments, lifestyle segmentationthat’s focus on usage patterns and place customers intodifferent usage or lifestyle segments, risk segments thatplace customers into different risk segments after evalu-ating the risk behavior of customers.

However, it is important to recognize that most seg-mentation schemes focus on homogenizing customersbased on commonalities into segments. The approach thatis used is cluster analysis where customers that are similarare clustered together, and each cluster is dissimilar fromother clusters. The dissimilarity is often measured througha meaningful index, such as Euclidian or some otherdistance, some kind of probabilistic index, or in terms ofwithin and between cluster variability. While clusteringallows segments to be created by homogenizing custom-ers based on commonalities, it is important to note that thisapproach does just that and does not address the valuedimension. Segments that are created could vary in termsof value or attractiveness to the business. Likewise, cus-tomers within the segments that share similar characteris-tics could be very different in terms of value or attractive-ness to the business. It is also important to note thatdifferent businesses could also have different perceptionson the value or attractiveness of customers and customersegments.

To be able to ensure that businesses select segmentsthat are attractive to the business, it is imperative tomodify the existing segmentation approach, by incorpo-rating an approach where the value of the customer isdetermined first. The approach should also allow formultiple dimensions, such as financial dimensions, riskand attrition, to be assessed simultaneously. This ap-proach is able to rank customers and customer segments

based on value or attractiveness to the business, and thisin turn allows the business to direct their resources in anappropriate manner.

PROPOSED FIVE STEPS IN SEGMENTATION

The approach that is being proposed consists of fivesteps. The first step is to understand the value of thecustomer or Decision Making Unit (DMU) to the busi-ness. The value of the customer to the business is drivenby a variety of factors, such as behavior and expectedbehavior of the customer and the resources that the busi-ness directed to the customer, and these factors should betaken into account when deriving the value of the cus-tomer. The approach that can be used to evaluate the valueor attractiveness of a customer is relative efficiency throughData Envelopment Analysis (DEA).

The second step is to rank order customers based onrelative-efficiency. The third step would be to categorizecustomers that share similar relative-efficiency scoresinto bands of relative-efficiency. Having done this, thenext step would be to identify clusters of customers thatexist within a particular relative-efficiency band. This willensure that clusters that are created consist of customersthat are not only share commonalities, but are similar interms of value or attractiveness to the business. The fifthstep would be to develop marketing strategy to the differ-ent segments within the different relative efficiency bands.

This approach is different from the traditional seg-mentation process, as it first focuses on defining the valueor attractiveness of the customers. By developing seg-ments based on clustering in the different relative-effi-ciency bands, the business will be able to make resourceallocation decisions in a more efficient manner and directresources to segments that are similar in terms of value orattractiveness to the business.

IDENTIFYING RELATIVE-EFFICIENCY

THROUGH DEA

As mentioned earlier, one approach that lends itself toderiving the value or attractiveness of the customer is Data

Proposed Approach: Five Steps Approach in Segmentation

Step 1 Step 2 Step 3 Step 4 Step 5

Derive Rank order Categorize Define Developcustomer/ customers DMU into clusters marketingDMU value based on Relative within the strategy forthrough relative- efficiency different theDEA efficiency bands Relative different

Efficiency segments


Envelopment Analysis (DEA). The concept of Data En-velopment Analysis was first proposed by Charnes, Coo-per, and Rhodes (1978) and central to their depiction is theconcept of efficiency as proposed by Farrell (1958). Indiscussing the efficiency of a firm, Farrell (1958) suggeststhat efficiency consists of two components: technicalefficiency (which reflects the ability of a firm to obtainmaximum outputs from a given set of inputs) and allocativeefficiency (that demonstrates the ability of a firm to useinputs in optimal proportions, given their respective pro-cess). Technical and allocative efficiency are then com-bined to provide a measure of economic efficiency. Indefining efficiency, Charnes and Cooper (1985) suggestthat 100 percent efficiency is attained for a unit, only when(a) none of its outputs can be increased without eitherincreasing one or more of its inputs, or decreasing some ofits other outputs; and (b) none of its inputs can be de-creased without either decreasing some of its outputs, orincreasing some of its other inputs (Charnes and Cooper1985 quoted in Norman and Stoker 1991). In DEA termi-nology, the units that are assessed, in this case customers,are usually referred to as Decision Making Units (DMU).First coined by Charnes et al. (1978), it is suggested thatDMUs should be “homogenous entities in the sense thatthey use the same resources to procure the same outcomesalbeit in varying amounts.” The term “decision making”implies that the unit of assessment has “common inputsand outputs which are being assessed for efficiency” andthat it is possible to control or influence the processesemployed to convert resources into outcomes. The objec-tive of measuring efficiency of a DMU is “for the purposeof resource conservation or for output augmentation.” Tobe able to achieve this, it is imperative to be able todifferentiate between the performances of various units ofassessment, where “a unit of assessment is the entity wepropose to compare on performance with other entities ofits kind.” A performance indicator, which is typically aratio of output to input pertaining to the unit that is beingassessed, is often developed to gauge efficiency and whenthe performances of various units are compared simulta-neously against each other, comparative or relative effi-ciency can be attained. This provides a sort of a ranking ofthe efficiency performance of all the units that are beingassessed.

DEA generalizes a single-output/single-input effi-ciency measure to the multiple-output/multiple-input caseby constructing a ratio of a single “virtual” output to asingle “virtual” input. DEA does not require any assump-tion about the functional form of the inputs and outputs. Itcalculates a maximal performance measure for each deci-sion making unit, or DMU, in relation to all the otherDMUs in the observed population with the sole require-ment that each DMU lie on or below the external frontier,and that each DMU not on the frontier is scaled against theconvex combination of the DMUs on the frontier facetclosest to it. Norman and Stoker (1991) emphasize that

“the efficiency measure of the DMU is defined by itsposition relative to the frontier of best performance estab-lished mathematically by the ratio of weighted sum ofoutputs to weighted sum of inputs.” DEA calculates amaximum performance measure for each DMU relative toall other DMUs in the observed population. This results ina relative efficiency score being derived and each DMU isplaced accordingly to its efficiency in a production possi-bility frontier (or PPS), where the most efficient DMU’soccupy the external frontier or are placed on the efficientfrontier.

The DEA method determines the relative efficiencyof a DMU by maximizing the ratio of weighted outputs toinputs, subject to the condition that ratio for every DMUis not larger than one. This results in an efficiency scorethat is less than or equal to one. When outputs are held ata constant, this method of achieving relative efficiency isknown as the input-orientated method, whereby inputshave to be decreased or improved in order for a DMU tobe considered efficient. An efficient DMU has no poten-tial improvement, whereas inefficient DMUs yield effi-ciency scores that demonstrate improvement can be madeor achieved based on the performance of other DMUs.Assume a set of observable DMU’s, {DMU j: j = 1,…n)have m inputs {x ij ;i = 1,....m} and s outputs,{yrj;r =1,.......s}. The original method proposed by Charnes,Cooper, and Rhodes suggests that relative efficiency of aDMU j

0can be measured in the following manner: Min h

Where:xij = the amount of the ith input at DMU j, yrj = the amountof the rth output from DMU j and j

0= the DMU to be

assessed.

The ability of DEA to handle a wide range of vari-ables and relations (constraints) relaxes the conditions onthe number of variables that can be used in calculating thedesired evaluation measures. This makes it easier dealingwith complex problems and considerations likely to beconfronted by managers, as it could include “influencingfactors,” which is more that just profits and costs but alsoextends to factors that could influence the performance orefficiency. Hirschberg and Lye (2001) also highlight thisflexibility in the type of inputs that can be used in DEA, bypointing out that “unlike traditional stochastic frontiermethods of production and cost function estimation . . .DEA does not require monetary valued inputs, a single

s.t.

0ijiijj

j hxSx i

0rjrrjj

j ySx r

0, rj ss ri,0j j


output, nor does it rely on assumptions if a particularfunctional form or a particular statistical distribution.”This clearly implies that variables do not have to bedirectly related and there is flexibility in variable selec-tion.

In the next section, an effort will be made to under-stand how DEA is used to assess the value or attractive-ness of the customer and customer segment in a credit cardbusiness.

DERIVING RELATIVE-EFFICIENCY

THROUGH DEA IN A CREDIT

CARD BUSINESS

This research has been carried out on a sample of4,159 credit card accounts. As customers can have mul-tiple credit cards, these accounts when consolidated at acustomer level belong to a sample of 3,000 customers,from a major credit card business operating in SoutheastAsia. The DMU in this research is the individual cus-tomer, which subsequently forms a segment. A constantreturn to scale input orientated DEA model, where inputscan be decreased or improved in order for a DMU toimprove in terms of efficiency, is selected for this re-search.

Among financial metrics that are focused on in thisresearch is revenue, which consists of fee income, interestincome and interchange. The cost taken into accountconsists of both variable and fixed cost, where fixed costconsists mainly of overheads and variable cost is linked tothe behavior of the customer. Besides financial metrics,two key behavior indicators, which are risk and attrition,have also been used. Risk scores that depict the propensityof a customer defaulting is measured at the customer leveland is tracked closely, as credit losses incurred from riskycustomers has the capability of dissolving hard earnedrevenue. Attrition scores might seem to be less importantthan risk scores as it does not directly impact revenue.However, attrition has the potential of eroding potentialrevenue from a customer and potential value is critical toall businesses. There are also many different metrics thatcan be used, such as customer lifetime value, RecencyFrequency Scores and other metrics that the businessdeems as being as important. However, in this research themetrics that are being used are revenue, cost, risk andattrition scores.

In attempting to conduct segmentation that focuseson the value or attractiveness of the customer, the first stepis to derive the relative-efficiency of every DMU (in thiscase the customer) based on the various inputs and outputsmentioned above. By running the DEA algorithm, it isapparent that every observation has a relative-efficiencyscore of between 0 and 1, which is derived based on thethree dimensions highlighted above. The score ranges

from 0.0 to 1.0, whereby 0.0 is the lowest relative effi-ciency score and 1.0 is the highest possible score.

These relative-efficiency scores depict the value orattractiveness at a customer level. It compares the value ofeach customer against the other and rank order the cus-tomers based on the concept of relative-efficiency. Thisscore is used not only to understand the value or attractive-ness of the customer, but it also allows the business togroup together customers that share similar relative-effi-ciency scores. These scores are used to create both effi-cient and inefficient groups. This can be done by decidingon a cut-off point on what is considered attractive. Thecut-off point can be changed as the business sees fit. Themain objective of having cut off scores is to allow thebusiness to create groups of customers that share the sameefficiency. This will be valuable to the business when thebusiness devises strategies and has to make resourceallocation decisions. By being able to group customersthat are similar in terms of value or attractiveness, thebusiness will be able to direct the appropriate amount ofresources to the group. If the business was not able toassess the value or attractiveness of the customer and thengroup customers that are similar together, the businesswould probably require more resources as it would haveto address a larger group of customers.

In this research, the cut-off has been set at 0.5, as it isan impartial mid-point. This cutoff can be determined bythe business and if it only had resources to target the top20 percent of the efficient customers, then it could deter-mine the cut-off at 0.8. However, if the business has lessresource, the business could increase the cut off to 0.7 andif it had more resources it could lower the cut-off score to0.4. The factor that drives the cut-off is the amount ofresources that the business has allocated for the effort.

In the current research, customers that have a score ofgreater than 0.5 are grouped together as the efficientgroup. In the same manner, customers that had less than0.5 were grouped in the inefficient group. From thesample it was found that 1,137 accounts were found tohave a relative efficiency score of greater than 0.5 and3,022 accounts had a relative efficiency score of morethan 0.5. This indicates that 27 percent of the portfolio isefficient from the three dimensions, which are risk, attri-tion and revenue and 73 percent of the portfolio had arelative efficiency score of less than 0.5.

Having identified that 27 percent of the portfolio isefficient and 73 percent is inefficient, the next step is tounderstand the efficient portfolio from several dimen-sions, to verify if these customers really are efficient.

In the credit card business, two dimensions that arecritical are Outstandings and revenue. Outstandings is acritical metric as it represents the amount that is owed to


Portfolio Break dow n - Efficient and Inefficient

73%

27%

the credit card business, by its customers. A portfolio mayhave many customers that spend but do not owe theportfolio anything, as these customers pay off their bal-ances.

In this sample portfolio, it is evident that efficientcustomers constitute a majority of customers that have anoutstanding of greater than $1,000 per month. By carryingbalances on their accounts, these customers would alsogenerate more revenue. This is also evident in the revenuedistribution among efficient and inefficient customers,where it is evident that there are more inefficient custom-

ers generating between $0–$250 in revenue and moreefficient customers that generate more than $1,000. Thedistribution of outstanding and revenue accurately depictsthe distribution of efficient and inefficient customers inthe portfolio. This indicates that the method is able to splitthe efficient and inefficient customers based on a varietyof factors, such as risk, attrition and revenue, simulta-neously.

The efficient and inefficient portfolio can also beanalyzed from other performance dimensions, such ascredit limit distribution, recency of transactions, utiliza-

Portfolio Breakdown Efficient and Inefficient

Proportion of Revenue Distribution by Revenue

Band

0%

20%

40%

60%

80%

100%

a.0-250 b.250-500 c.500-750 d.750-1000 e.>1000

0

Revenue Band

RE score <= 0.5 RE score > 0.5

Distribution of Outstanding by Relative Efficiency

Bands

0%

20%

40%

60%

80%

100%

a.0-250 b.250-500 c.500-750 d.750-1000 e.>1000

Outstaning Bands ($)

RE score <= 0.5 RE score > 0.5

Distribution of Outstandiant by

Relative Efficiency Bands

Proportion of Revenue Distribution

by Revenue Band

TABLE

Variables That Describe the Efficient, Innefficient and Total Portfolio

Efficient Group Inefficient Group Total Portfolio

Attrition Score Attrition Score Attrition Score

Age Credit Limit Months on Books

Number of Times Past Due Months on Books Number of Times Delinquent

Marital Status Delinquency State

State of Residenc Gender Months Since Last Transaction

Gender State of Residence Months Since Last Credit Limit Increase

Months Since Last Transaction Months Since Last Transaction

Marital Status


tion of credit limit, months on books or tenure and fre-quency of delinquency. The portfolio can also be re-viewed from demographic variables such as age, maritalstatus and state of residence. By conducting principalcomponent analysis, it is possible to identify variables thatare different in the efficient and inefficient portfolio, aswell in the total portfolio.

From the results above, it is evident that the threedifferent cuts of the portfolio all have different principalcomponents that describe the populations. This is impor-tant as it demonstrates that without splitting the portfoliointo efficient and inefficient segments, it will not bepossible to distinguish between efficient and inefficientsegments in the portfolio. By not being able to distinguishthe efficient and inefficient segments, it is possible that thebusiness will direct resources to inefficient customersegments. This, in turn, would be inefficient use of re-sources that are often scarce. However by being able todistinguish the efficient and inefficient customers, basedon multiple dimensions simultaneously through DEA, abusiness is able to direct its resources more aptly. In thenext section, an effort will be made to demonstrate howthis approach can be used by businesses in ensuringefficient resource allocation in two different areas, whichare loyalty and risk.

DEVELOPING STRATEGY BASED ON

RELATIVE-EFFICIENCY

Being able to identify efficient and inefficient cus-tomers allows a business to develop strategies and ensurethat resources are allocated to customers that it deems areefficient. In the example below, the overlay of efficiencyand attrition score, which are derived through attritionmodels, it is evident that the business should focus moreon customers that are highly likely to attrite, which are inattrition Band 1 and 2, as well as customers that are moreefficient than others.

In this example, the business should allocate itsresources to reduce attrition among all its most efficientcustomers, which consist of 2,694 accounts. It then shouldfocus on targeting customers in relative-efficiency Band 2,which consist of 328 accounts. The business should avoiddirecting resources to its less efficient customers in rela-tive-efficiency Band 3 and 4. This constitutes 27 percentof the population with regards to attrition, and will enablethe business to reduce cost by at least 27 percent. If thebusiness needs to reduce more cost, it might consider if itreally wants to target customers in attrition Band 3 and 4,as these two bands consist of customers with a lowpropensity to attrite.

In the same way, it is also possible to influence theselection of customers for different risk strategy by over-laying relative-efficiency scores and risk scores. Theobjective of doing this is to focus resources on savingcustomers that are deemed to be of higher value to thebusiness. This can be done by providing credit educationto these customers or by helping customer manage theircredit, as losing these customers would also impact thebusiness. In this example, the business should focus firston saving customers that are deemed to have high risk bybeing in Delinquency Band 1 and 2, as well as customersthat are highly efficient in relative-efficiency Band 1and 2. This consists of approximately 1,607 accounts.

By being able to select customers the business wouldlike to target, the business does not have to incur the costof targeting all its customers. If a customer is deemed to beinefficient based on various dimensions, the businessmight decide that it does not want to direct its resources onthose customers. In short, recognizing that not all custom-ers are equal enables a business to make decisions onwhere to direct its resources to. Targeting efficient cus-tomers will help the business to manage its resources in amore efficient manner.

TABLE

RelativeEfficiency and Attrition Cross Tabulation

Attrition Attrition Attrition Attrition

Band 1 Band 2 Band 3 Band 4 Total

Rel-Effi Band 1 651 1428 615 0 2694

Rel-Effi Band 2 0 0 326 2 328

Rel-Effi Band 3 0 0 99 175 274

Rel-Effi Band 4 0 0 0 863 863

Total 651 1428 1040 1040 4159


There are other examples that can be used to demon-strate how businesses can use relative efficiency score todifferentiate its customers based on efficiency. The rela-tive efficiency scores enable a business to select custom-ers in a more efficient manner, which in turn results in anefficient allocation of resources. This would not havebeen possible if a business only defined its customersbased on commonalities.

CONCLUSION

In this paper, an effort has been made to present adifferent approach to segmentation. The approach stemsfrom the realization that it is no longer sufficient tohomogenize customers into segments based on similari-ties, as segments have to be of value to the business.Taking this into consideration, it is suggested that it is firstessential to assess the value of the customer to the busi-ness, and group customers together based on their value tothe business. However, assessing the value of the cus-tomer is not a straight forward task, as not all businessesvalue the same customer behavior or performance metrics.

To be able to accommodate the different values thatbusinesses might emphasize, it has been suggested thatrelative-efficiency can be used as a measure of attractive-ness. Through relative-efficiency, which is derived throughData Envelopment Analysis, it is possible to rank custom-

ers in terms of relative-efficiency. Customers that sharesimilar levels of relative-efficiency can be grouped to-gether, and the business is able to select customers that areof similar value to the business. This approach providesbusinesses with a customer and customer segment selec-tion protocol that is especially applicable for businessesthat have moved toward collating extensive data at thecustomer level. By having this data, the approach tosegmentation can radically change from one that onlyfocus on homogenizing customers to one that incorpo-rates the element of value and attractiveness to the busi-ness.

The ability to identify customer and customer seg-ments that are of value and attractive to the business frommultiple dimensions simultaneously will allow businessesto direct resources to customer and customer segmentsthat are of greater value to the business. The importance ofidentifying approaches that focus on value and ensureappropriate use of resources in segmentation is exceed-ingly important, both because there are innovative tech-nological enhancements, such as Data Warehousing, thatallow for it and also due to the increasing emphasis onmetrics that are available currently. Being able to differ-entiate the attractiveness of a customer or customer seg-ment, allows businesses to allocate resources in a moreefficient manner to segments that are of value or attractiveto the business.

TABLE

Cross Tabulations of Relative-Efficiency and Propensity for Delinquency Bands

Del Del Del Del Del Del Del Del

Band 1 Band 2 Band 3 Band 4 Total Band 1 Band 2 Band 3 Band 4 Total

Rel-Effi Bank 1 671 796 606 621 2694 Rel-Effi Bank 1 16.1% 19.1% 14.6% 14.9% 64.8%




Total 935 1161 999 1064 4159 Total 22.5% 27.9% 24.0% 25.6% 100.0%

For further information contact:Ranjeet Khaira

Department of MarketingMonash Universkty

Victoria 3800Austrialia



A MODEL TO INCLUDE THE POTENTIAL IMPACT OF

GOVERNMENTAL PUBLIC AND FOREIGN POLICIES

ON GLOBAL BRAND EQUITY

Peter A. Kaufman, Illinois State University, NormalFrederick W. Langrehr, Valparaiso University, Valparaiso

ABSTRACT

A government’s public and foreign policies mayhave positive, neutral or negative impacts on global publicopinion. These policies may also influence foreign evalu-ations of brands associated with that country. It is pro-posed that brand origin should be included within a brandequity framework.

INTRODUCTION

Following September 11, 2001 much of the worldexpressed an unprecedented level of support for the UnitedStates. This sympathy, however, did not last long asindicated by the fate of Ronald McDonald in Ecuador andAmerican Express in Germany. “In Quito Ecuador, atorched Ronald McDonald statue becomes the latest vic-tim of rage over the U.S. invasion of Iraq” (Business Weekonline 2003). At about the same time, for the same reasonin Germany some restaurants were beginning to refuseAmerican Express cards (Barret 2004).

Polls have shown that attitudes toward the U.S. havedeclined considerably since 2001 and some link theseattitudes too affects on U.S. brands. A Pew Global Atti-tudes Project published in 2005 that surveyed 15 countriesfound that global perceptions of the U.S. were declining.Similarly, a BBC World Service poll of more than 26,000consumers from 25 countries conducted between Novem-ber 2006 to January 2007 concluded that the globalopinion of the United States role in world affairs hadsignificantly declined in the prior year. Supporting thesefindings was an Anholt Quarterly (2007) survey of con-sumers in 35 countries which found that the U.S. as abrand ranked poorly on government related categories.

COUNTRY EFFECTS ON U.S. BRANDS

A brand for many companies is the single mostimportant asset of the company and consumer perceptionsof its brand can be shaped by a multitude of factors suchas its actions in the marketplace (e.g., advertising) and thebehavior of its employees (Interbrand.com 2007). A goodexample of the value of a brand is when Kraft was sold for600 percent above its book value. Perceptions of qualityof a product may also be influenced by the country withwhich it is associated (Jolibert and Peterson 1995). The

value of a brand name may also be influenced by external,environmental variables such as its home country’s for-eign and public policies.

A 2004 NOP World study of perceptions of trustfound that only 35 percent of non-U. S. consumers trustedAmerican brands, and their use of U.S. brands droppedby 3 percent at a time when U.S. use did not change. Adifferent study conducted in 2005 by Edelman TrustBarometer found that there was a “trust discount” foriconic U.S. brands. For example, Procter and Gamble hada 70 percent trust rating in the U.S. but only received 38percent in Germany and 29 percent in Spain. McDonaldswas trusted by 51 percent of US opinion leaders but only24 percent in France, and 28 percent in the U.K. whileHeinz had a 70 percent ranking in the U.S. but only 31percent in France and 22 percent in Italy. Madden (2003)found that 13 percent of respondents in China, 19 percentin Indonesia, and 36 percent in India rejected U.S. brandsbased on their origins.

A 2004 study by Global Market Insite (GMI) polled8,000 consumers in eight countries and one in five Euro-peans admitted to making a concerted effort to avoid U.S.goods and more than 50 percent admitted to an increase intheir negative perception of the U.S. (Barrett 2004). GMIstates on its website “American multinational companieswill need to mount a valiant effort to distance themselvesfrom the image of the U.S. federal government and itsunpopular foreign policies in the New Year or risk contin-ued brand erosion and ongoing boycotting by Europeanand Canadian consumers . . . .”

A 2006 study by Energy BBDO, a Chicago-basedadvertising agency that surveyed 3,300 teens 13–18 yearsold in 13 countries, found that while American brands stillrank highly for brand awareness, overall likeability forglobal titans like McDonalds and Coca-Cola has slippedwhile Sony, Nokia, and Adidas were moving up. AmandaFreeman, a vice president at The Intelligence Groupbelieved that American brands may have been hurt due tothe perception that the U.S. had been interfering in in-ternational affairs (GenWorld Teen Study 2006).

In this paper we will develop some propositions onhow a government’s policies may impact brand imagesand brand equity and influence the willingness to buy


brands associated with that country. First, we will outlinethe country of origin and brand equity literatures. Second,we propose a model that integrates country of origin andbrand equity models taking into the consideration theimportance of foreign policy variables. Next, we willoutline factors outside the model that could influencesalience and importance of brand origin for consumers.

COUNTRY OF ORIGIN AND BRAND ORIGIN

Country of Origin

The country of origin (COO) literature is mainlyconcerned with determining the effects of consumers’perceptions of countries on their evaluations of productquality and choice processes (Thakor and Kohli 1996).Country of origin is defined as the opinions consumershave of products based on their evaluations of a country’sproduction and marketing capabilities (Roth and Romeo1992). In many cases COO has been viewed as an extrin-sic product cue that may influence perceptions of productquality such as French wine or Italian made suits. This cuemay be relied on more heavily when time for decisionmaking is limited or knowledge of the product choices islimited. Parameswaran and Pisharodi (1994) in perhaps inthe most comprehensive assessment of the structure ofCOO proposed that it is a multi-dimensional constructcomprised of general country, product, and product at-tributes.

Early COO studies focused on studying the effect of“Made in X” on consumer product evaluations. Thesestudies often excluded brand names entirely, used ficti-tious names, or failed to control for the brand origininherent in real brand names (Thakor and Kohli 1996).Stereo products made in Germany versus Indonesia wereexpected to show higher product quality ratings and oftendid (Yasin et al. 2007). This research may have beenappropriate when products were designed, assembled,and made in the same country but with globalization COOhas since evolved into a multifaceted construct that recog-nizes the complexity of a global production chain.

Brand Origin

This COO conceptualization raises a practical ques-tion, however. How many consumers know that the prod-uct they are considering purchasing is designed, assembled,or made in all these locations and how many care that aSony MP3 player is made in Indonesia or elsewhere?Researchers in one study found that forty percent ofconsumers cannot identify the country of origin for well-known brands (Saimee et al. 2005) and consumers maycare less if the brand they buy is made in a country that isnot well known for manufacturing quality items as long asits corporate brand is from a country they respect forquality such as Sony and Japan and Nike and the U.S.

Thakor and Kohli (1996, p. 32) reinforce this point bystating “…brand origin should not change with a changein manufacturing location . . . . Hence, the perceived originof the brand need not be the same as the country shown onthe ‘made in label’.”

Testing for the independent affect of the brand and itscountry effect in the same research project could bequestionable given the overlap of country onto brand.Because of this problem researchers have recommendedthat brand origin should be used instead of country oforigin. In other words, brand origin should be a compo-nent or facet of brand equity. A group of consumerbehavior scholars state “Brand origin should realisticallybe viewed as a component of brand equity, because theorigins of many products are very much a part of theircharacters” (Samiee et al. 2005, p. 393–394).

ROLE OF A COUNTRY’S FOREIGN POLICY,

BRAND EQUITY, AND BRAND ORIGIN

Given the globalization of business and brands pro-duced by multinational corporations it is more and morelikely that influences from the home country may affecthow consumers view these companies and brands. Asindicated previously, foreign policy by the U.S. has beenquestioned by many abroad.

Two branding experts, Keller and Lehmann (2006)point out that brand equity is increasingly being deter-mined by activities and factors outside the company’sdirect control. Given the polling data and significantanecdotal evidence and speculation it is important tounderstand if a country’s foreign policy could be affectingthe brand equity of brands associated with its homecountry.

If a country has controversial policies related to theeconomy, military, and the environment that affect con-sumers in other countries, these consumers may developnegative attitudes toward the U.S. “American marketerswith overseas ambitions should not ignore the possibilityof negative attitudes toward the U.S. in some regions ofthe world” (Klein et al. 1998, p. 98).

Brand Equity

Branding has emerged as a central management pri-ority in the last decade due to the growing realization thatbrands are one of the most valuable intangible assets thatcompanies have (Keller and Lehmann 2006). The litera-ture on branding is extensive and can be divided into threebroad domains – customer, company, and financial levels(Keller and Lehmann 2006). Brand personality, brandrelationships, brand alliances, brand equity measurement,corporate image and reputation, and how to build a globalbrand are all aspects of this research. At the present time


no research has evaluated the extent to which countryimage impacts the equity of brands associated with thatcountry when accounting for foreign policy activities ofthat country.

The focus of this discussion is how the consumerviews the brand. Kevin Keller defines consumer basedbrand equity (CBBE) as “the differential effect of brandknowledge on consumer response to the marketing of thebrand” (Keller 1993, p. 2). In the most prominentconceptualization of brand equity, Keller proposes in his1993 seminal definition that brand equity is comprised ofbrand awareness and brand image. Thus, brand equity ispresent when the consumer is (1) familiar with the brandand (2) holds some favorable, strong, and unique brandassociations in memory. Given the valuations of brandsabove and beyond their book value, it is expected thatbrand equity has a direct and positive effect on consumerwillingness to buy these brands.

Brand Awareness. Brand awareness according toKeller (1993) is the strength of the brand node or trace inmemory, as demonstrated by consumers’ ability to iden-tify the brand under different conditions. Said differently,brand awareness relates to the probability that a brandname will be thought of and the facility with which itcomes to mind (Keller 1993). Specifically, Keller pro-poses that brand awareness is comprised of brand recog-nition and brand recall. Brand recall relates to consumers’capacity to retrieve the brand from memory when giventhe product category, the needs fulfilled by the category,or some other type of probe as a cue. Consumers need tobe aware of the brand before they can recognize its origin.If consumers are aware of the brand they may then, ifmotivated, investigate its origin. It is also expected thatbrand awareness impacts brand image directly. Keller(1993, p. 3) states “Brand awareness affects decision-making by influencing the formation and strength ofbrand associations in the brand image. A necessary con-dition for the creation of a brand image is that a brand nodehas been established in memory. . . .” Also, mere exposuretheory suggests that exposure over time to an object isexpected to increased positive attitudes toward the object.Consequently, higher levels of brand awareness couldcontribute to greater brand image levels.

Proposition: There is a positive relationship betweenbrand awareness and brand origin.

Proposition: There is a positive relationship betweenbrand awareness and brand image.

Brand Image. Brand image is defined by Keller(1993, p. 3) as “perceptions about a brand as reflected bythe brand associations held in consumer memory.” Kellerhowever, makes no mention of associations that maypertain to country of origin or brand origin and perhaps

that is what motivated him and his coauthor to raise thefollowing question in their 2006 review on the brandequity literature: “To what extent does country image (orequity) impact the equity of brands from that country?”(Keller and Lehmann 2006, p. 750). Other researchershave also identified the need for research to answer thisquestion. Samiee et al. (1995, pp. 393–394) state “Brandorigin should realistically be viewed as a component ofbrand equity, because the origins of the many products arevery much a part of their characters.” Thakor and Lavack(2003) further state that brand origin may play an impor-tant role in determining a brand’s image.

Proposition: There is a positive relationship betweenbrand origin and brand image.

Proposition: There is a positive relationship betweenbrand origin and brand equity.

Willingness to Buy Brand and Perceived Quality.Much research exists that indicates a relationship betweenthe equity of a brand and consumers’ willingness to buy abrand (Interbrand.com 2007). It is expected that brandequity evaluations lead to perceptions of perceived qual-ity. A brand with a higher equity should be perceived ashigher quality. Perceived quality much like perceivedvalue has been found to lead to willingness to purchase(Dodds, Monroe, and Grewal 1991).

Proposition: There is a positive relationship betweenbrand equity and willingness to buy brand.

Proposition: There is a positive relationship betweenbrand equity and perceived quality.

Proposition: There is a positive relationship betweenperceived quality and willingness to buy the brand.

Cognitive and Affective Components of Brand Ori-gin. In a consumer behavior context the cognitive compo-nent focuses on the overall product judgments held by aconsumer for a country associated with the brand. Theaffective component, that is emotional, concerns whetherconsumers like the country that is associated with thebrand. It is proposed that brand origin has both a cognitiveand an affective component. Country of origin has beenmeasured in many ways as cited above. Its measurementhas largely been dominated by a cognitive approach andhas failed to integrate the possible role that emotion(affect) may play. In Parameswaran and Pisharodi (1994,p. 47) comprehensive measurement of COO which wascomprised of general country, general product and spe-cific product attributes, there was very little attentiongiven to emotion. Many items were product focused suchas the product in question is: “long lasting,” “handleswell,” “good value,” etc. In fact, under their generalcountry attribute dimension only one item was considered


affective – people from X country are “friendly andlikable.”

Keller (2003, p. 597) raises the following questionthat is pertinent to the current research: “How mightsource credibility be applied to country of origin effect?”Source credibility theory states that people are more apt tobe persuaded when the source is considered credible andis more of a cognitive evaluation that emphasizes productjudgment characteristics. We believe that positive prod-uct judgments (cognitive component) about the countryof the corporate ownership as indicated by, for example,quality, reliability, and value should be positive related towillingness to buy the brand.

Proposition: There is a positive relationship betweenpositive cognitive thoughts and favorability of brandorigin.

Proposition: There is a positive relationship betweenpositive cognitive thoughts and willingness to buythe brand.

Affective Components of Brand Origin. The role ofemotion in consumer decision-making in this context ishighlighted in the work by Klein et al. (1998). They foundthat Nanjing, China residents respected Japanese madeproducts for their advanced nature, quality, reliability,design, and color, and value for the money, but because oftheir animosity of the Japanese as a result of the Japanesearmy slaying 300,000 Chinese civilians during WorldWar II, the Chinese willingness to purchase Japanesemade goods was negatively affected. The authors concep-tualized animosity as a construct with war and economicanimosity dimensions (facets) that were historical orlongstanding in nature. Note, they did not evaluate Chi-nese opinions in a brand theory framework but ratherfocused on Japanese products in general and their foreignpolicy variables were limited to war and economics. It isexpected that affect or emotion by foreign consumerstoward a country may be engaged or triggered throughreactions to much more recent foreign and public policiesand not just to events that occurred decades earlier.

Proposition: There is a positive relationship betweenpositive affective thoughts and favorability of brandorigin.

Proposition: There is a positive relationship betweenpositive affective thoughts and willingness to buy thebrand.

Non Brand Origin Factors. It is important to pointout that there may be many other factors that drive brandimage apart from brand origin such as: types of brandassociations, product and non-product related attributes,and functional, experiential, and symbolic benefits (Keller

1993). However, non-governmental factors are not thefocus of the present discussion.

An Integration of Existing Models

As discussed, models have been developed to sepa-rately capture country of origin effects and brand equity.Each however, does not adequately capture the potentialeffect of a country’s foreign and public policies on con-sumer perceptions of brands closely associated with thatcountry. What follows is a proposed model integratingbrand equity and country of origin models.

Dimensions of Governmental Policies. The percep-tions of people outside a country may be greatly shaped bygovernmental policies and by association may affectperceptions and valuations of brands that are associatedwith that country. As mentioned previously, there is adearth of research in business and specifically marketingthat has evaluated the impact that foreign consumers’animosity toward a country can have on their willingnessto buy products from that country. After consulting nu-merous books written on foreign and public policies, aliterature search on academic articles, and after reviewingsurvey questions from major global polling organizationsand reviewing the U.S. Council of Foreign Relationswebsite on “Issue Areas,” the following domains wereconsidered to represent the most important areas of for-eign and public policy to include in a brand equity model:military, trade, and the environment.

Military actions taken by a country have life anddeath consequences and as such can radically affectopinions of consumers toward the country initiating theaction. For example the invasion of Iraq by the UnitedStates was not supported by most of the world andattitudes were reinforced when information surfaced thatIraq did not have weapons of mass destruction and evi-dence making the case for invasion may have been ma-nipulated. Protests against this action were widespread.

A country’s trade policy is the set of national policieswhich affect the quantity and value of a nation’s exportsand imports. Examples of policies include tariffs, import/export quotas, and import/export subsidies. On the onehand United States officials continue to pursue and defendfree trade while on the other they succumb to protectionistpressure (Art and Brown 1993). As an example somequestion the true intentions of the U.S. How could acountry that believes so strongly in free trade subsidize itsagricultural sector to such an extent (Hook 2005)?

With increased growth of developing countries thereis greater concern that the world’s environment will not beable to sustain anticipated pollution levels. According toHook (2005) the U.S. has been less assertive in seekingglobal remedies for environmental problems. Much to the


dismay of many people, the United States has not signedand ratified into law what many consider the most impor-tant environmental agreement in history with a goal tocontain greenhouse gases, the Kyoto Protocol. Manyleading nations saw this as a selfish act and this actioncaused great resentment toward the U.S. (Art and Brown1993).

Proposition: Positive attitudes toward the country’smilitary, trade, and environment positions should bepositively related to positive overall affect (liking)toward the country.

Proposition: There is a positive relationship betweenaffective thoughts (absence of animosity) and favor-able attitude toward brand origin.

Consumer Ethnocentricism. Consumer ethnocen-trism (CE) originates from the more general construct ofethnocentrism which is defined as people viewing theirown in-group as primary and possessing proper standardsof behavior and as offering protection against potentialthreats from those outside the in-group (Brislin 1993).Consumer ethnocentrism is measured by the CETSCALE(Shimp and Sharma 1987). Research on ethnocentrismhas shown that the more ethnocentric a consumer is, themore he/she will have less favorable attitudes toward andlower intention to buy foreign products (Han 1988). HighCE people had strong beliefs about the appropriatenessand morality of purchasing foreign-made items (Kleinet al. 1998). Brodowsky (1998) discovered that low eth-nocentric consumers were more likely to use objectiveinformation about product quality than high ethnocentricconsumers. Other studies found a negative relationshipwith high CE and perceptions of quality of imported items(Klein et al. 1998; Netemeyer et al. 1991). Consumerethnocentrism is also expected to moderate the relation-ship between affect and willingness to buy the brand. Ifconsumers like the brand’s country this affect may lead toa willingness to buy the brand however, an ethnocentrictendency will tend to erode that link because of the desireto support the domestic groups.

Proposition: CE is expected to reduce the positiveassociation (relationship) of affect to willingness tobuy the brand.

Proposition: CE is expected to be negatively relatedto the cognitive component.

Proposition: CE is expected to be related negativelyto willingness to buy the brand.

Factors Outside the Model. It is acknowledged thatpolicies with primarily a domestic impact could have aneffect on foreign consumers’ opinions of the country andpossibly its brands. For example, Europeans may disagreewith the United States use of capital punishment and itsliberal gun ownership policies. There are also situationswhere a country’s domestic laws may affect other coun-tries as in the U.S. extending the Dolphin Sea turtle caseextraterritorially. This means that exporters in other coun-tries wishing to export tuna to the U.S. need to fish for itin a way that does not harm dolphins and sea turtles.

Further, there are likely to be non-governmentalaspects of the country such as culture that foreign consum-ers may not like. For example, the Barbie doll was bannedfrom Russia due to its “outrageous curves” that werethought to be “corrupting minds of children” (FinancialTimes 2002, p. 11). The focus, however, of this researchis on the impact of what could be considered formalgovernmental policies on brands associated with the coun-try creating those policies.

CONCLUSION

Brand knowledge is what most consumers will usewhen they make product choice decisions and brandinformation in many cases carries with it country of origininformation. We suggest that a brand’s origin is thus acomponent of brand image and its origin should be inte-grated into a brand equity model. Brand origin is concep-tualized to comprise both cognitive and affective compo-nents. Cognitive aspects deal with product judgmentinformation such as quality, reliability, value, associatedwith a country, and the affective elements involve judg-ments about a country’s policies on military, trade, and theenvironment. It is proposed that if foreign consumers havenegative opinions of a country’s foreign policies thesefeelings may have a negative impact on the equity ofbrands and the willingness to buy the brands closelyassociated with that country. In this case the brands maybe “guilty by association.”

REFERENCES

Anholt Quarterly Survey (2007), Accessed April 20,2007, Available from [http://www.nationbrandindex.com/faq.phtml].

Art, Robert J. and Seyom Brown (1993), U.S. ForeignPolicy – The Search for a New Role. New York:MacMillan Publishing Company.

BBC World Service Poll (2007), Accessed May 23, 2007,Available from [http://news.bbc.co.uk/2/hi/americas/4185205.stm].

Barret, Lucy (2004), “Yankee Go Home,” MarketingWeek, (November 11), 24–26.

Brislin, Richard (1993), Understanding Culture’s Influ-ence on Behavior. Orlando: Harcourt BraceJovanovich.


Brodowsky, Glen H. (1998), “The Effects of Country ofDesign and Country of Assembly on EvaluativeBeliefs about Automobiles,” Journal of InternationalConsumer Marketing, 10 (3), 35–114.

Business Week Online (2003), “Wars and Boycotts, BothFade Away,” 8 (April), Available from BusinessSource Elite.

Dodds, William B., Kent B. Monroe, Dhruv Grewal(1991), “Effects of Price, Brand, and Store Informa-tion on Buyers’ Product Evaluations,” Journal ofMarketing Research, 28 (August), 307–19.

Edelman Trust Barometer (2005), Accessed May 15,2007, Available from [http://www.edelman.com/news/ShowOne.asp?ID=102].

Financial Times (2002), “Barbie is Banned from Russia,Without Love,” (August29), 11.

GenWorld Teen Study (2006), Accessed April 2, 2007,Available from [http://www.businessfordiplomaticactions.com/learn/articles/genworld_leave_behind.pdf].

Global Marketing Insite (2004), Accessed April 29, 2007,Available from [http://www.gmi-mr.com/gmipoll/release.php?p=20041227].

Han, C. Min (1988), “The Role of Consumer Patriotism inthe Choice of Domestic Versus Foreign Products,”Journal of Advertising Research, 28, (June/July),25–31.

Hook, Steven W. (2005), U.S. Foreign Policy – TheParadox of World Power. Washington D.C.: Con-gressional Quarterly.

Interbrand (2007), Accessed May 3, 2007. Available from[http://www.ourfishbowl.com/images/surveys/IB_2007BrandMarketersReport.pdf].

Jolibert, Alain J.P. and Robert Peterson (1995), “A Meta-Analysis of Country of Origin Effects,” Journal ofInternational Business Studies, 26, (4), 883–900.

Keller, Kevin Lane (1993), “Conceptualizing, Measur-ing, and Managing Customer-Based Equity,” Jour-nal of Marketing, 57 (January), 1–22.

____________ (2003), “Brand Synthesis,” Journal ofConsumer Research, 29 (March), 595–600.

____________ and Donald R. Lehmann, “Brands andBranding: Research Findings and Future Priorities,”Marketing Science, 25 (6), 740–59.

Klein, Jill Gabrielle, Richard Ettenson, and Marlene Morris(1998), “The Animosity Model of Foreign ProductPurchase: An Empirical Test in the People’s Repub-

lic of China,” Journal of Marketing, 62 (January),89–100.

Madden, N. (2003), “Brand Origin Not Major Factor forMost Asians,” Advertising Age, 74 (14), 33.

Netemeyer, Richard G., Sroniivas Durvasula, and DonaldR. Lichtenstein (1991), “A Cross-National Assess-ment of the Reliability and Validity,” Journal ofMarketing Research, 28 (August).

NOP World 2004 Global Advocacy Study (2004), Ac-cessed June 5, 2007, Available from [http://www.prnewswire.com/cgi-bin/stories.pl?ACCT=104&STORY=/www/story/10-05-2004/0002266581&EDATE=].

Pew Global Attitudes Survey (2005), Accessed April 20,2007, Available from [http://pewglobal.org/commen-tary/display.php?AnalysisID=1019].

Parameswaran Ravi and R. Mohan Pisharodi (1994),“Facets of Country of Origin Image: An EmpiricalAssessment,” Journal of Advertising, 23 (March),43–56.

Roth, Martin S. and Jean Romeo (1992), “MatchingProduct Category and Country Image Perceptions: AFramework for Managing Country of Origin Ef-fects,” Journal of International Business Studies, 23(3), 477–97.

Samiee, Saeed, Terence Shimp, and Subhash Sharma(2005), “Brand Origin Recognition Accuracy: ItsAntecedents and Consumers’ Cognitive Limitations,”Journal of International Business Studies, 36, 379–97.

Shimp, Terence A. and Subhash Sharma (1987), “Con-sumer Ethnocentrism: Construction and Validatationof the CETSCALE,” Journal of Marketing Research,24 (August), 280–89.

Thakor, Mrugank V. and Chiranjeev S. Kohli (1996),“Brand Origin: Conceptualization and Review,” Jour-nal of Consumer Marketing, 13 (3), 27–42.

____________ and Anne Lavack(2003), “Effect of Per-ceived Brand Origin Associations on Consumer Per-ceptions of Quality,” Journal of Product & BrandManagement, 12 (6), 394–407.

U. S. Council on Foreign Relations (2007), Accessed June5, 2007, Available from [http://www.cfr.org].

Yasin, Norjaya Mohd, Mohd Nasser Noor, and OsmanMohamad (2007), “Does Image of Country of OriginMatter to Brand Equity?” Journal of Product &Brand Management, 16 (1), 38–48.

For further information contact:Peter A. Kaufman5590 Marketing

Illinois State UniversityNormal, IL 61790-2200



NATIONAL CULTURE AND ADOPTION OF TECHNOLOGY-BASED

NEW PRODUCTS

Abhijit Patwardhan, University of Mississippi, OxfordDouglas W. Vorhies, University of Mississippi, Oxford

SUMMARY

During the last two decades, considerable researchhas explored the impact of national culture in the domainof international business. Both managers and academi-cians have witnessed the impact of globalization on eco-nomic activities. To find new markets, firms are increas-ingly trying to pursue new sales opportunities outsidetheir home markets. Therefore, it is important to developa better understanding of factors that impact how consum-ers in different countries adopt new products. Among theacademic studies in the global marketing area, researchershave investigated the impact of national culture on con-sumer behavior focused on the purchase of technology-based products (Mooij and Hofstede 2002; Ganesh, Kumar,and Subramanium 1997; Roth 1995). In this research, weuse the term “technology-based new products” suggest-ing high technology (consumer) electronics products suchas home appliances, photographic equipments, digitalproducts, etc. Levitt (1983) argues that a firm can marketidentical products and services around the globe becausenational markets have become homogenized. However,more recent studies demonstrate that response to newproducts differs from country to country (Mahajan andMuller 1994; Tellis, Stremersch, and Yin 2003; Kumarand Krishnan 2002; Gatignon, Eliashberg, and Robertson1989).

According to Mooij and Hofstede (2002, p. 61),“ignoring culture’s influence has led many companies tocentralize operations and marketing which resulted indeclining profitability instead of increasing efficiency.”Therefore, understanding adoption of technology-basednew products by a particular culture is crucial to marketersas it can provide some idea to what extent, these newproducts can be successful in various countries. Variouscross-cultural researchers have attempted to study cul-tural impact on consumer behavior and suggested thatconsumer behavior is more heterogeneous due influenceof national culture (Mooij and Hofstede 2002). A reviewof literature also reveals much agreement among scholarson the use of Hofstede’s empirical definition of thedimensions of national culture while conducting cross-national and cross-cultural research. Therefore, this paperadopts Hofstede’s method of operationalization of the“national culture” construct.

In addition to national culture, other factors mayimpact consumers’ adoption of technology-based new

products. Bauer (1967) introduced the notion of “per-ceived risk” in the context of consumer behavior andperceived risk has become well established in the con-sumer behavior literature. While reviewing the literatureon consumer perceived risk, Mitchell (1999, p. 163) statesthat “perceived risk theory has intuitive appeal and playsa role in facilitating marketers seeing the world throughtheir customers’ eyes.” Mitchell (1999, p. 164) highlightsversatility of perceived risk studied in the literature andfurther argues that “perceived risk is more powerful atexplaining consumers’ behavior since consumers are moreoften motivated to avoid mistakes than to maximize utilityin purchasing” and “examining risk perceptions can gen-erate new product ideas.” In addition to perceived risk,other potentially important factors in understanding theadoption of technology-based new products are “per-ceived usefulness” and “perceived ease of use” (Davis,Bagozzi, and Warshaw 1989). As we attempt to study thisin relation to consumer markets, it is crucial to know howconsumers perceive the new products’ usefulness andtheir ease of use and therefore, the suggested model in thispaper includes both the variables “perceived usefulness”and “perceived ease of use.”

It is important to note that the development of tech-nology-based new products itself can be a costly businessfor any organization and, therefore, any factor that mightincrease the product’s chance of success is worthy ofresearch. Surprisingly, there seem to be only a few studiesthat have examined the linkages among national culture,perceived ease of use, perceived usefulness, perceivedrisk and adoption or consumption of technology-basednew products. The purpose of this paper is to fill this gapin the literature. To do so, we first review the literaturerelating to technology acceptance model, perceived risk,and national culture and further derive various proposi-tions based on the literature review. Next we develop amodel based on the modification of the technology accep-tance model (Davis 1986) which is widely used to under-stand the factors influencing users’ decision to adopt newtechnology. Thus, this study contributes to the new prod-ucts adoption literature in two ways. First, we introduce atechnology acceptance model (TAM) to consumer behav-ior literature dealing with product adoption. Second, weextend and broaden TAM by introducing other importantrelevant factors such as national culture and perceivedrisk. After a brief discussion, we conclude with potentialcontributions of the suggested research to theory andpractice. References are available upon request.


For further information contact:Abhijit Patwardhan

University of MississippiP.O. Box 2609

University, MS 38677Email: [email protected]


STANDARDIZATION WITH RESPECT TO WHAT?

A CONCEPTUALIZATION OF ADVERTISING

STANDARDIZATION MEASUREMENT IN

PAST RESEARCH

Fernando Fastoso, Bradford University School of Management, United KingdomJeryl Whitelock, Bradford University School of Management, United Kingdom

SUMMARY

In contrast to the question of how standardizationshould be defined, which has long been discussed in theliterature (cf., Walters 1986; Onkvisit and Shaw 1987;Ryans, Griffith, and White 2003), the question of how –i.e., with respect to what – standardization should bemeasured appears to have remained unnoticed, both interms of its existence and its consequences. To our knowl-edge, a recent review of studies of international advertis-ing standardization (IAS) conducted among managers(Fastoso and Whitelock 2007) was the first to clearlyidentify this issue as it showed a lack of consensus inmeasuring IAS in past research. The authors concludedthat differences in measurement make a comparison ofresults between studies de facto impossible and thus limitthe ability of standardization research in general to con-tribute to knowledge advancement. The main purpose ofthis paper was therefore to conduct a comprehensivereview of past standardization research in order to shedmore light onto this neglected issue by exploring how pastresearch studies measured IAS, how they defined theirgeographic focus, how the two issues relate to each other,and whether there were differences on these issues basedon the methodologies employed.

We selected the 17 most influential internationalbusiness, marketing, and advertising journals on which tofocus our analysis based on journal quality and impact asassessed by previous reviews in the fields of marketing(Hult, Neese, and Bashaw 1997), international business(DuBois and Reeb 2000) and advertising (Henthorne,LaTour, and Loraas 1998; Royne Stafford 2005).1 Subse-quently, we conducted several broad keyword searches(e.g., “marketing AND standardization,” “advertising,and standardization”) in the EBSCO host and Proquestdatabases and ended up with a set of 55 past studies whichhad measured the extent of IAS.

In order to identify the way in which this measure-ment had been conducted, we conceptualized what wereferred to as the measurement paths to IAS using anadaptation of Wind, Douglas and Perlmutter’s (1973)E.P.R.G. schema, a long established framework in theinternational business and marketing literature. We showedthat in the same way that a company can be organized

ethnocentrically, standardization measurement can beconducted in an ethnocentric manner when a study takesthe headquarters (HQ) market practices as the point ofreference and compares them with practices in othercountries individually. A regiocentric measurement ofstandardization, in turn, is one that takes a geographicallystrict view of a specific region and measures standardiza-tion as the level of similarity between national practices inthe countries that form the region. Further, a geocentricIAS measurement is one that views the whole world – orat least all countries where the company is operating – asa unity and therefore assesses the level of similarity inadvertising practices across all these countries. Finally,the polycentric category does not apply to standardizationmeasurement given that, by definition, polycentric activi-ties are local activities. We termed this adaptation of theoriginal schema the E.R.G. framework for IAS measure-ment.

The main finding of our analysis was that standard-ization researchers have taken different approaches to-ward measuring the phenomenon without showing aware-ness of the consequences that this can have for the inter-pretation and cross-study comparability of results. Over-all, we found that although past research has relied in themain on the geocentric measurement path, a substantialnumber of studies conducted ethnocentric measurements.Our analysis showed that for studies with an international/global geographic focus the methodological soundness ofethnocentric measurement can be questioned, as randompaired comparisons of country practices appear moresensible. For studies with a regional focus, in turn, weshowed that ethnocentric IAS measurement can bias re-sults toward the localized end of the standardizationcontinuum when the HQ market lies outside of the regionin question. Our study therefore concluded that standard-ization research should relate to company strategy, rely-ing on geocentric measurements when this is interna-tional, and on regiocentric measurements when it is re-gional.

This study also found a tendency to focus on theinternational/global level of standardization independentlyof the research method used. However, our results founda statistically significant association between the researchmethod used and the E.R.G. measurement path to IAS


taken, as ethnocentric measurement was significantlymore frequently applied in studies of managers than inthose of advertisements. We concluded that in order forknowledge advancement to take place in the field (as

called for by for example Ryans, Griffith, and White2003), inconsistencies in terms of IAS measurement bothwithin and across methods must be avoided.

ENDNOTE

1 Based on Hult et al. (1997) we analysed the Journal ofMarketing, International Journal of Research in Mar-keting, Journal of the Academy of Marketing Science,Journal of Business Research, Journal of ConsumerResearch, Journal of Marketing Research, Journalof Retailing, and Marketing Science; based on DuBoisand Reeb (2000) we also considered the Journal ofInternational Business Studies, Journal of World

Business (formerly Columbia Journal of World Busi-ness), International Marketing Review, Journal ofInternational Marketing and Management Interna-tional Review; based on Henthorne, LaTour andLoraas (1998) as well as Royne Stafford (2005) weadded Journal of Advertising, Journal of AdvertisingResearch, Journal of Current Issues and Research inAdvertising (formerly called Current Issues and Re-search in Advertising) as well as International Jour-nal of Advertising to the list of journals of interest.

REFERENCES

DuBois, Frank L. and David Reeb (2000), “Ranking theInternational Business Journals,” Journal of Interna-tional Business Studies, 31 (4), 689–704.

Fastoso, Fernando and Jeryl Whitelock (2007), “Interna-tional Advertising Strategy: The StandardizationQuestion in Manager Studies,” International Mar-keting Review, 24 (5), 591–605.

Henthorne, Tony L., Michael S. LaTour, and Tina Loraas(1998), “Publication Productivity in Three LeadingU.S. Advertising Journals: 1989 Through 1996,”Journal of Advertising, 27 (2), 53–63.

Hult, G. Tomas M., William T. Neese, and R. EdwardBashaw (1997), “Faculty Perceptions of MarketingJournals,” Journal of Marketing Education, 19 (1),37–52.

Onkvisit, Sak and John J. Shaw (1987), “StandardizedInternational Advertising: A Review and CriticalEvaluation of the Theoretical and Empirical Evi-

dence,” Columbia Journal of World Business, 22 (3),43–54.

Royne Stafford, Marla (2005), “International ServicesAdvertising (ISA): Defining the Domain and Re-viewing the Literature,” Journal of Advertising, 34(1), 65–86.

Ryans, John K., David A. Griffith, and D. Steven White(2003), “Viewpoint: Standardization/Adaptation ofInternational Marketing Strategy Necessary Condi-tions for the Advancement of Knowledge,” Interna-tional Marketing Review, 20 (6), 588–603.

Walters, Peter G.P. (1986), “International MarketingPolicy: A Discussion of the Standardization Con-struct and its Relevance for Corporate Policy,” Jour-nal of International Business Studies, 17 (2), 55–69.

Wind, Yoram, Susan P. Douglas, and Howard V.Perlmutter (1973), “Guidelines for Developing Inter-national Marketing Strategies,” Journal of Market-ing, 37 (2), 14–23.

For further information contact:Fernando J. Fastoso

Bradford University School of ManagementEmm Lane

Bradford, BD9 4JLUnited Kingdom

Phone: +44.1274.233192Fax: +44.1274.546886



LOVING A BRAND: CONCEPT AND CULTURE

Noël Albert, University of Grenoble (IAE-CERAG) and University of Lyon 1 (IUT), FranceDeanne Brocato, Iowa State University, Ames

Dwight Merunka, University Paul Cézanne Aix-Marseille (IAE-CEROG) and EUROMEDMarseille School of Management, France

Pierre Valette-Florence, University of Grenoble (IAE-CERAG) and CERAM Nice, France

SUMMARY

Introduction

This study explores the nature and dimensions of lovefor brands among French consumers. This concept isrelatively recent (Ahuvia 2005b; Fournier 1998) and haspredominately been studied in the U.S. Love for a brandis defined as “the degree of passionate emotional attach-ment that a person has for a particular trade name” (Carrolland Ahuvia 2006, p. 5). Ahuvia (1993, 2005a, 2005b)provided empirical proof of the existence of this feeling oflove and demonstrated similarities and differences withinterpersonal love. Main studies dealing with the brandlove construct (Ahuvia 1993, 2005b; Shimp and Madden1988; Whang et al. 2004) introduce the concept of lovebased on one given theory of interpersonal relationships(among many theories), which leads to the exclusion ofthe conceptualization found in other theories. We studythe concept of love toward brands and propose a method-ology that attempts to circumvent this limitation.

Research Methodology

A possible limitation of the study of love in a market-ing context is the explicit use of the term love. We use aprojective method in which participants are given a set ofimages enabling them to express the type of feelings theyhave toward brands. Nineteen images were selected andthree of them were chosen because of their abilities tosymbolize a feeling of love (circled in Figure 1). We usedan internet survey and collected data from 843 respon-dents. The word love only appears at the end of the survey.Previously, different questions enabled an understandingof the relationship between consumers and their brandswithout using the term “love.” The results were analyzedthrough correspondence analysis (CA) which helps sum-marize and visualize the lexicon used by respondents foreach open-ended question used during the survey. Re-sponses with high agreement with the love relationshipbetween the consumer and the brand were analyzed throughCA to uncover the dimensions of the feelings of lovetoward the brand. Cluster analysis of words projected in a

FIGURE 1

Images Submitted to the Survey Participants


multi-dimensional space via multiple correspondenceanalysis also enabled identifying the main dimensions ofthe love construct.

Results

Several dimensions of love appeared in the explana-tions given by respondents. The dimensions of love to-ward a brand found in the French market were: (1)passion, (2) duration of the relationship, (3) self-congru-ity, (4) dreams, (5) memories, (6) pleasure, (7) attraction,(8) uniqueness, (9) beauty, (10) trust, (11) declaration ofaffect. The passion, self-congruity, dream, attraction,uniqueness, beauty, trust, declaration of affect, duration,pleasure dimensions were also found to be linked, if notexplicitly, to current interpersonal literature on the feelingof love. The memories dimension did not seem to berelated to current extant literature on interpersonal theo-ries of love, this provided a new dimension for future

consideration when investigating the brand love con-struct.

Conclusion

Our approach employs an exploratory methodologyrather than applying a selected interpersonal theory oflove to the marketing field. Evidence shows that dimen-sions of love toward a brand among French consumers fallin line with many dimensions identified in interpersonalstudies, which differs with U.S. samples. This exploratorystudy suggests several extensions for further research: thedevelopment of a measurement scale, the study of therelations between brand love and other concepts (e.g.,attachment, commitment, trust) and finally the proposaland empirical testing of a conceptual model that can assessthe influence of a feeling of love on dependent attitudinaland behavioral variables (e.g., brand loyalty, resistance tochange or positive word of mouth).

REFERENCES

Aron, A. and E.N. Aron (1986), Love as the Expansion ofSelf: Understanding Attraction and Satisfaction. NewYork: Hemisphere.

Ahuvia, A.C. (1993), “I Love It! Towards a UnifyingTheory of Love Across Divers Love Objects,” Doc-toral Dissertation, Northwestern University.

____________ (2005a), “Beyond the Extended Self: LoveObjects and Consumer’s Identity Narratives,” Jour-nal of Consumer Research, 32 (June), 171–84.

____________ (2005b), “The Love Prototype Revisited:A Qualitative Exploration of Contemporary FolkPsychology,” Working Paper. University of Michi-gan – Dearborn.

Carroll, B.A. and A.C. Ahuvia (2006), “Some Anteced-ents and Outcomes of Brand Love,” Marketing Let-

ters, 17, 79–89.Fournier, S. (1998), “Consumers and Their Brands: De-

veloping Relationship Theory in Consumer Re-search,” Journal of Consumer Research, 24 (March),343–73.

Hatfield, E.C. (1988), “Passionate and CompanionateLove,” in The Psychology of Love, R.J. Sternberg andM.L. Barnes, eds. New Haven, CT: Yale UniversityPress, 191–217.

Shimp, T.A. and T.J. Madden (1988), “Consumer-ObjectRelations: A Conceptual Framework-Based Analo-gously on Sternberg’s Triangular Theory of Love,”Advances in Consumer Research, 15, 163–68.

Whang, Y.O., J. Allen, N. Sahoury, and H. Zhang (2004),“Falling in Love with a Product: The Structure of aRomantic Consumer-Product Relationship,” Ad-vances in Consumer Research, 31, 320–27.

For further information contact:Noël Albert

University of Grenoble (IAE-CERAG)University of Lyon 1 (IUT)

FrancePhone: 33.676.461.216



STRESS-INDUCED WORD-OF-MOUTH SEEKING: A MISSING

DIMENSION IN WORD-OF-MOUTH RESEARCH

Selina Cui, The Chinese University of Hong Kong, Hong KongHaksin Chan, The Chinese University of Hong Kong, Hong Kong

SUMMARY

We address a missing dimension in WOM researchby conceptualizing the dissatisfied consumer as an anx-ious WOM seeker, who interacts with others in order tocope with the stress induced by negative consumptionexperiences. Drawing on the stress affiliation and grouppolarization literatures, we propose that stress-inducedWOM seeking may alleviate or aggravate dissatisfactiondepending on various conditions. Specifically, we ad-vance propositions regarding the moderating effects ofseveral factors on the likelihood of stress-induced WOMseeking and on the relative strength of alleviation versusaggravation.

Coping With Negative Consumption Emotions

Dissatisfied consumers may experience a range ofnegative emotions, such as anger, disappointment, regret,fear, and distress. Conceptually, these are emotional mani-festations of stress and anxiety due to perceived harm orloss. In response, consumers try to find ways to cope withthese emotions. The stress affiliation literature providesthe rationale for WOM seeking as a coping response bydissatisfied consumers. Interestingly, the group polariza-tion literature suggests that WOM seeking by coping-minded consumers may in fact exacerbate dissatisfaction.

Stress Affiliation

According to stress affiliation research, there is atendency for people under stress to affiliate with otherswho undergo similar experiences. Affiliation may serve(1) a self-validation function – as a means to gauging thelegitimacy of the stress-induced feelings, (2) a psychoso-cial function – as a means to receiving comfort and sup-port, or (3) a cognitive function – as a means to obtainingdisambiguating information. Hence, seeking out and com-municating with similar others through WOM is effectivefor coping with negative consumption emotions and re-ducing stress. Dissatisfied consumers may feel betterbecause they secure emotional validation, receive socialsupport, or make sense of a stressful situation.

Group Polarization

Group polarization refers to the tendency for peopleto become more extreme in attitude after communicationwith similar others. The informational influence argu-

ment suggests that interaction with (mostly) like-mindedothers opens the door to new information in support ofprior thoughts and feelings, thereby reinforcing the initialinclination. The normative influence argument, on theother hand, concerns the social motivation to adjust andamplify one’s responses in the direction of the majorityposition. Regardless of the underlying process, the polar-ization that occurs as a result of WOM seeking shouldhave an aggravating effect on dissatisfaction. This iscontrary to the general conclusion of stress affiliationresearch.

Propositions

As a step toward disentangling these opposing ef-fects, we identify four moderators pertinent to stress-induced WOM seeking. The following propositions sum-marize the influences of these moderators on WOM-seeking tendency by dissatisfied consumers (P1, P3, P5,P7) and on the relative strength of the alleviating affilia-tion effect versus the aggravating polarization effect (P2,P4, P6, P8).

Problem Severity

P1: Dissatisfied consumers are more likely to seek WOMfrom similar others when problem severity is high(vs. low).

P2: WOM seeking from similar others results in morepronounced dissatisfaction alleviation when prob-lem severity is high (vs. low).

Attribution Uncertainty

P3: Dissatisfied consumers are more likely to seek WOMfrom similar others when attribution uncertainty ishigh (vs. low).

P4: WOM seeking from similar others results in morepronounced dissatisfaction aggravation when attri-bution uncertainty is high (vs. low).

Affect Intensity

P5: Dissatisfied consumers are more likely to seek WOMfrom similar others when affect intensity is high (vs.low).


P6: WOM seeking from similar others results in morepronounced dissatisfaction alleviation when affectintensity is high (vs. low).

Source Similarity

P7: Dissatisfied consumers are more likely to seek WOM

from similar others when source similarity is high(vs. low).

P8: WOM seeking from similar others results in morepronounced dissatisfaction aggravation when sourcesimilarity is high (vs. low).

For further information contact:Haksin Chan

Department of MarketingThe Chinese University of Hong Kong

Shatin, Hong KongPhone: 852.2609.7637

Fax: 852.2603 5473E-Mail: [email protected]


THE CONSUMER STRIKES BACK: ASSESSING ANTECEDENTS TO

BOYCOTT PARTICIPATION

Hans H. Bauer, University of Mannheim, GermanyCarmen-Maria Albrecht, University of Mannheim, Germany

Melchior D. Bryant, University of Mannheim, GermanyTobias E. Haber, University of Mannheim, Germany

SUMMARY

Consumer boycotts have become increasingly im-portant for marketing decision-making, as they are apervading and effective instrument of expressing theconsumers’ discontent with a company and its businesspractices (e.g., Sen, Gürhan-Canli, and Morwitz 2001;Tyran and Engelmann 2005). In today’s marketplace,companies are more and more judged by consumersaccording to their being environmentally and/or ethicallyresponsible (Klein, Smith, and John 2004). If companiesdo not act in a socially responsible way, they will riskbeing the target of consumer resistance behavior (Klein,Smith, and John 2002). One way of opposing the respec-tive company is to refuse to buy its brands. If consumersengage in organized and collective acts of anticonsumption,they are said to boycott (Kozinets and Handelman 1998).

Generally, a consumer boycott can be defined as “anattempt by one or more parties to achieve certain objec-tives by urging individual consumers to refrain frommaking selected purchases in the marketplace” (Friedman1985, p. 97). Through the consumers’ decision not to buyproducts of the respective company, boycotting can leadto loss in sales (Miller and Sturdivant 1977). Moreover,the corporate reputation as well as the brand image cansuffer severe damage if the boycott organization publiclypillories the company’s business practices (Klein, Smith,and John 2004).

For the mentioned reasons, it is important for compa-nies to know why consumers participate in boycotts. Notonly is it important for companies to gain deeper insightinto the individual’s motivation to participate in a boycott,but also for boycott organizations to know about thedrivers of individual boycott behavior in order to arrangea boycott in a more successful way.

In the latest literature (e.g., Ettenson and Klein 2005),researchers are requested to identify further antecedentsto boycott participation, since one has not found the entirerange of antecedents yet. Thus, the objective of our studyis to fill this research gap by unveiling further antecedentsto the consumer’s intention to participate in a boycott.Therefore, we draw on results of prior research as well as

insights gained from an exploratory study with 12 infor-mants.

Taking all findings into account, our model consistsof seven hypotheses: We assume that the perceived par-ticipation of others in the boycott and the credibility of theboycott call determine the consumer’s perceived successlikelihood of the boycott which, in turn, influences theintention to participate in the boycott. Further, the cred-ibility of the boycott call as well as the involvement in theboycott cause drive the intention to participate in theboycott. Contrary, the consumer’s commitment to thebrand to be boycotted negatively influences his or herintention to participate in a boycott. And finally, theconsumer’s refusal to buy the particular brand is affectedby the intention to participate in the boycott.

We used data of 234 study participants (average ageof 27.6 years, 48.1% female, 51.9% male) to assess ourassumptions. Subjects were exposed to a fictitious news-paper article about a boycott call against a soft drinkcompany that was accused of infringing upon environ-mental protection policies in India. Afterwards, subjectsanswered questions regarding the model constructs. Theresults confirmed all but one hypothesis. The credibilityof the boycott call does not influence the intention toparticipate in the boycott.

The findings reveal several drivers of consumer’sboycott participation that are of significant importance toboycott organizers. The credibility of the call to partici-pate in the boycott as well as the consumer’s involvementin the boycott cause influence the intention to participatein the boycott. Therefore, boycott organizers should seekbroad media coverage of their intended boycott withcredible information sources to enhance the credibility ofthe boycott. Further, it is important to evoke the conster-nation of consumers to enhance involvement in the boy-cott cause. In addition, the perceived boycott participationof others has a strong effect on the perceived successlikelihood of the boycott which, in turn, determines theintention to participate in the boycott. Thus, boycottorganizers should strengthen the consumers’ perceptionof others being engaged in the boycott to enhance theindividual’s intention to participate.


Of course, managerial implications for companiesrefer to any activity that prevents the boycott cause frombeing credible or from evoking consternation. Basically,press releases could act as a counterbalance to the mediaactivities of boycott organizers. However, the results ofthis study show another effective way to prevent consum-ers from participating in boycotts. The consumer’s com-mitment to the brand to be boycotted negatively influ-ences the consumer’s intention to participate in the boy-

cott. Therefore, a strong brand image that leads to highlycommitted consumers appears to be a protective barrier toboycott calls.

In summary, our research shows how important it isfor managers and boycott organizers to understand ante-cedents to and consequences of boycott participation tosuccessfully act in the marketplace.

REFERENCES

Ettenson, Richard and Jill G. Klein (2005), “The Falloutfrom French Nuclear Testing in the South Pacific: ALongitudinal Study of Consumer Boycotts,” Interna-tional Marketing Review, 22 (2), 199–224.

Friedman, Monroe (1985), “Consumer Boycotts in theUnited States, 1970–1980: Contemporary Events inHistorical Perspective,” Journal of Consumer Af-fairs, 19 (1), 96–117.

Klein, Jill G., N. Craig Smith, and Andrew John (2002),“Exploring Motivations for Participation in a Con-sumer Boycott,” Advances in Consumer Research,29 (1), 363–69.

____________, ____________, ____________ (2004),“Why We Boycott: Consumer Motivations for Boy-cott Participation,” Journal of Marketing, 68 (July),

92–109.Kozinets, Robert V. and Jay Handelman (1998),

“Ensouling Consumption: A Netnographic Explora-tion of the Meaning of Boycotting Behavior,” Ad-vances in Consumer Research, 25 (1), 475–80.

Miller, Kenneth E. and Frederick D. Sturdivant (1977),“Consumer Responses to Socially Questionable Cor-porate Behavior: An Empirical Test,” Journal of Con-sumer Research, 4 (1), 1–7.

Sen, Sankar, Zeynep Gürhan-Canli, and Vicki Morwitz(2001), “Withholding Consumption: A Social Di-lemma Perspective on Consumer Boycotts,” Journalof Consumer Research, 28 (December), 399–417.

Tyran, Jean-Robert and Dirk Engelmann (2005), “To Buyor Not to Buy? An Experimental Study of ConsumerBoycotts in Retail Markets,” Economica, 72 (285),1–16.

For further information contact:Carmen-Maria AlbrechtUniversity of Mannheim

L 5, 1Mannheim, 68131




SELECTIVE CONSUMER COMMUNICATIONS: THE ROLE OF

RELATIONSHIP STRENGTH

Yu Hu, Virginia Polytechnic Institute and State University, Blacksburg

SUMMARY

It has been well-documented in marketing and con-sumer literature that interpersonal, word-of-mouth com-munications (WOM) have a powerful impact on consum-ers’ product evaluations and purchase decisions (e.g.,Frenzen and Nakamoto 1993; Herr, Kardes, and Kim1991). However, despite the extensive research efforts,one important topic in WOM communications has largelyescaped researchers’ attention: the WOM communicationcontents. Most WOM communications happen after thecommunicators have acquired the product information.The construction of the communication messages shoulddepend on how the communicators access, search, re-trieve, and present the information to the audience.

Social psychology research in interpersonal commu-nications suggests that with the same product informa-tion, consumers might communicate different aspects ofthe knowledge to different audiences to suit the audience’scharacteristics or the communicators’ own goals (Higginsand Rholes 1978). In addition, evidence from memoryresearch has long established that people tend to retrieveonly a subset of information available in memory due tovarious cognitive or motivational constraints in differentsocial contexts (Wyer and Srull 1986). Therefore, wehave reasons to doubt that consumers can or are willing tofaithfully transmit all the production information theypossess to any audience that listens to them. Given com-munication messages’ malleable nature, it is important tounderstand the process of the WOM message generationand to investigate factors that might influence consumers’memory-based message production.

The current study explores consumers’ selective com-munication process by using the relationship strengthbetween the communicator and recipient (i.e., tie strength)as the defining character of the communicator-audiencerelationship (Granovetter 1973). Specifically, I investi-gate how the different degree of tie strength between thecommunicator and the audience would influence the waythat the communicator produces and presents the mes-sage.

Given the informational superiority of negative in-formation over positive information (e.g., Skowronskiand Carlston 1987) and the findings from relationshipnorms theory (e.g., Clark 1984), it is reasoned that whencommunicating product information with strong ties, thecommunicators should be more likely to disclose the

negative information in the hope that the information willhelp recipients avoid potential losses or dissatisfactionbecause the communicators are genuinely concerned aboutrecipients’ welfare. On the contrary, when communicat-ing with weak ties, communicators are more concernedwith balanced exchange and their willingness to sharevaluable information should be contingent on the likeli-hood of getting back something equivalent in value.Because people are generally not familiar with weak ties,it is not clear how and when the payback will happen.Thus, the communicators should be more likely to with-hold the negative product information unless there is aneminent possibility of reciprocity.

Therefore, I postulate that, given the existence ofpositive and negative product information or experiences,consumers, when engaging in WOM communications,tend to describe more of the negative features of theproduct to strong ties than to weak ties. Conversely,consumers tend to describe more of the positive featuresof the product when offering product information to weakties than to strong ties. Results from two studies supportedthe research hypotheses. In a laboratory-based advice-giving survey, subjects were found to generate more proarguments for several suggested everyday behavior whenthe intended audience was casual acquaintances, whereasthey gave more con arguments against those behaviorwhen the intended audience was supposed to be bestfriends. In the second study, subjects’ pre-communicationproduct knowledge was experimental controlled and itprovided a more stringent test of the selective communi-cation process in a memory-based WOM situation. Re-sults clearly demonstrated that when both positive andnegative product attributes were available in memory,subjects mentioned more positive attributes to casualacquaintances than to best friends, whereas they men-tioned more negative product attributes to best friendsthan to casual acquaintances.

For future research, selective WOM communica-tions should be studied beyond the influences of interper-sonal relationship strength. Many other situational orpersonal factors can be expected to affect thecommunicator’s message constructions. For example,researchers have long evinced that consumers form per-sonal relationships with the products they use (Fournier1998). From a social distance perspective, I suspect thatbecause the communicators see some products closer tothemselves while some other products more distant, theymight be less likely to share with others information about


closely-related products, while more likely to disclosedistant product information to others. In other words, wecan extend the current research by looking into the inter-

play between brand relationship and interpersonal rela-tionship and its joint effect on consumers’ WOM commu-nications.

REFERENCE

Clark, Margaret S. (1984), “Record Keeping in TwoTypes of Relationships,” Journal of Personality andSocial Psychology, 47 (September), 549–57.

Fournier, Susan (1998), “Consumers and Their Brands:Developing Relationship Theory in Consumer Re-search,” Journal of Consumer Research, 24 (March),343–73.

Frenzen, Jonathan and Kent Nakamoto (1993), “Struc-ture, Cooperation and the Flow of Market Informa-tion,” Journal of Consumer Research, 20 (Decem-ber), 360–75.

Granovetter, Mark S. (1973), “The Strength of WeakTies,” American Journal of Sociology, 78 (May),1360–80.

Herr, Paul M., Frank R. Kardes, and John Kim (1991),

“Effects of Word-of-Mouth and Product-AttributeInformation on Persuasion: An Accessibility-Diagnositicity Perspective,” Journal of ConsumerResearch, 17 (March), 454–62.

Higgins, E. Tory and William S. Rholes (1978), “SayingIs Believing: Effects of Message Modification onMemory and Liking for the Person Described,” Jour-nal of Experimental Social Psychology, 14 (July),363–78.

Skowronski, John J. and Donal E. Carlston (1987), “So-cial Judgment and Social Memory: The Role of CueDiagnosticity in Negativity, Positivity, and Extrem-ity Biases,” Journal of Personality and Social Psy-chology, 52 (April), 689–99.

Wyer, Robert S. and Thomas K. Srull (1986), “HumanCognition in Its Social Context,” Psychological Re-view, 93 (July), 322–59.

For further information contact:Yu Hu

Virginia Polytechnic Institute and State University2016 Pamplin Hall

Blacksburg, VA 24061Phone: 540.231.1967



UNDERSTANDING ATTITUDE-BEHAVIOR GAP IN GREEN BUYING

AS A SOCIAL DILEMMA

Shruti Gupta, Pennsylvania State University – AbingtonDenise T. Ogden, Pennsylvania State University – Lehigh Valley, Fogelsville

SUMMARY

For marketers of green products, the gap betweenpro-environmental attitudes and green purchase behaviorof the green consumer segment is a daunting challenge.An attitude is defined as “an enduring set of beliefs aboutan object that predisposes people to behave in particularway toward the object”1 (p. 257). Theory in the area ofconsumer attitude argues that individuals behave in waysconsistent with their attitudes. However, research in envi-ronmental consumerism has produced inconclusive evi-dence in support of consumer attitude theory with mixedresults that support both positive relationships betweenattitude toward the environment and behavior as well asweak relationships. A more recent study by Mintel2 (2006)resonates earlier results and found that despite pro-envi-ronmental attitudes, intention to recycle, concern aboutcar pollution and willingness to pay more for environmen-tally-friendly products, few consumers translated theseattitudes into regular green buying behavior.

This paper draws on social dilemma theory andreference group theory to understand the attitude-behav-ior gap in environmental consumerism. This researchargues that the attitude-behavior gap in environmentalconsumerism exists because it presents a social dilemmato the consumers. For a small subset of consumers who are“true believers” in environmental protection and conser-vation, the personal importance of the environmentalissue is likely to ensure unconditional participation. How-ever, most consumers, despite holding a positive attitudetoward environmental conservation make purchase deci-sions to maximize self-interest because in their view, thecosts of cooperation outweigh the uncertain utility ob-tained from it. Therefore, the decision to buy (collectivesocial gain) or not buy (self-interest) the green productdespite positive attitude toward environmental conserva-tion may be conceptualized as a social dilemma. Thispaper also draws on reference group theory that suggeststhat consumer decision to make the trade-off betweensuch self and collective group interests may also bedependent on the pressure to comply with the expecta-tions and behaviors of significant reference groups (e.g.,environmentally friendly group’s choice to cooperate andtherefore buy green products). Consequently, we draw onboth social dilemma and reference group theories toinvestigate the determinants of and the mechanisms toexplain the rationale behind the attitude-behavior gap as

it pertains to a specific environmental issue – energyconservation.

A model is presented that depicts the conditionsunder which consumers’ pro-environmental attitudes sub-sequently translate into positive purchase behavior forenergy efficient products. We posit that the attitude-behavior gap in green buying may be bridged by address-ing consumers’ individual attributions regarding coop-eration, their susceptibility to normative reference groupinfluence (i.e., social pressure to comply), in-group iden-tification, social value orientation, trust, personal effi-cacy, and the costs they incur in purchasing the product(i.e., preference for the conventional product and theirperception of substitutability between conventional andgreen products) on the predictive influence of pro-energyconserving attitude on their decision to buy energy effi-cient products (i.e., cooperation in a social dilemma). Thefollowing propositions are suggested:

H1: The positive influence of pro-environmental atti-tude on purchase behavior will be greater in thecase of cooperators than in the case of individual-ists or competitors.

H2: The positive influence of pro-environmental atti-tude on purchase behavior will be greater with hightrusters than with low trusters.

H3a: The positive effect of pro-environmental attitudeon purchase behavior will be greater when in-group identification is high than when it is low.

H3b: The positive effect of pro-environmental attitudeon purchase behavior will be greater for coopera-tors when in-group identity is high than when it islow.

H3c: The positive effect of pro-environmental attitudeon purchase behavior will be greater for individu-alists when in-group identity is high than when it islow.

H4a: The positive effect of pro-environmental attitudeon purchase behavior will be greater when expec-tation of others’ cooperation is high than when it islow.


H4b: The positive influence of expectation of others’cooperation will be greater for individuals who aremore susceptible to normative social influencethan for those who are less susceptible.

H5: The moderating influence of the effect of expecta-tion of others’ cooperation will be greater whenperceived efficacy is low than when it is high.

H6a: The positive effect of pro-environmental attitudeon purchase behavior will be higher when consum-ers perceive green and conventional products assubstitutes than when they don’t.

H6b: The positive effect of pro-environmental attitudeon purchase behavior will be higher when con-sumer preference for the conventional product islow than when it is high.

H6c: The positive effect of expectation of others’ coop-eration on purchase behavior will be greater whenconsumers’ perceive conventional and green prod-ucts as substitutes than when they don’t.

H6d: The positive effect of expectation of others’ coop-eration on purchase behavior will be greater whenconsumers’ preference for conventional product islow than when it is high. References are availableupon request.

For further information contact:Shruti GuptaMarketing

Pennsylvania State University – Abington1600 Woodland Rd.Abington, PA 19001Phone: 215.881.7649


ENDNOTES

1 Weigel, R.H. (1983), Environmental Attitudes and thePrediction of Behavior. New York: Praeger.

2 Mintel (2006), Green Living – U.S. Marketing ResearchReport.


DRIVERS OF SERVICE RECOVERY PERFORMANCE: PERCEIVED

ORGANIZATIONAL SUPPORT, LEARNING AND

PSYCHOLOGICAL JOB OUTCOMES

Christiana Raquel Lages, Boston College, Chestnut HillNigel F. Piercy, The University of Warwick, United Kingdom

SUMMARY

Given that services are simultaneously produced andconsumed and there is co-production (Berry 1980), it isimpossible to guarantee a 100 percent error-free service(Brown, Fisk, and Bitner 1994).

Service recovery performance follows failures inservice delivery. Limited knowledge exists regarding thefactors which have an impact on the service recoveryperformance, as perceived by frontline employees (Boshoffand Allen 2000). Therefore, service recovery perfor-mance is defined as the individual frontline serviceemployee’s perceptions of his/her own skills and actions,whilst resolving a service failure so as to guaranteecustomer’s satisfaction (Babakus et al. 2003; Boshoff andAllen 2000).

This research draws on goal orientation, perceivedorganizational support, psychological job outcomes andservice recovery literature, in order to investigate thedrivers of service recovery performance. The major goalsof this study are to determine the effect of learning goalorientation on service recovery performance and the im-pact of perceived organizational support on learning goalorientation. Additionally, the effect of both learning goalorientation and perceived organizational support on emo-tional exhaustion is investigated.

Based on the literature review, an integrative concep-tual framework comprising eighteen hypotheses are pro-posed and empirically tested next to frontline serviceemployees from the restaurant industry. A total of 740employees (representing a response rate of about 32%)provide the data for the analysis. Confirmatory factoranalysis (CFA) is employed to assess the fit of the sevenmeasurement components of the model. CFA results

provide evidence of high internal consistency reliabilityas well as convergent and discriminant validity. The finalseven-factor model comprising twenty-four items fits thedata reasonably well (χ²= 423.41; df = 231; p-value =0.000; GFI = 0.92; NNFI = 0.88; CFI = 0.90; IFI = 0.90;RMSEA = 0.034). Structural equation modeling (SEM) isused to test the hypothesized path model. All the indicessuggest a good fit of the final model to the data: χ² =320.05 (df = 171, p-value = 0.000), CFI = 0.91, IFI = 0.91,GFI = 0.92, and NNFI = 0.89.

The findings provide academic insights that mayopen new fruitful avenues for further research. The resultssupport all the hypotheses regarding the determinants ofpersonal learning, mastery orientation, job satisfactionand organizational commitment. With regard to the driv-ers of emotional exhaustion, the results are only partiallysupported, given that both personal learning and masteryorientation were found to be unrelated to emotional ex-haustion. The findings further indicate partial support ofthe hypotheses related to the antecedents of service recov-ery performance. Two unexpected results emerge: jobsatisfaction has a negative impact on service recoveryperformance and emotional exhaustion has a positiveeffect on service recovery performance. The new empiri-cal results reveal that learning goal orientation positivelyimpacts service recovery performance. Additionally, per-ceived organizational support has a positive effect onlearning goal orientation. Finally, whereas perceived or-ganizational support has a negative impact on emotionalexhaustion, learning goal orientation is unrelated to emo-tional exhaustion. These findings provide several mana-gerial implications for service marketing practitioners byoffering them practical guidelines to develop and imple-ment effective service recovery programs. Suggestionsare provided in terms of the best practices when recruitingand training frontline service employees.

REFERENCES

Babakus, Emin, Ugur Yavas, Osman M. Karatepe, andTurgay Avci (2003), “The Effect of ManagementCommitment to Service Quality on Employees’ Af-fective and Performance Outcomes,” Journal of theAcademy of Marketing Science, 31 (3), 272–86.

Berry, Leonard L. (1980), “Services Marketing is Differ-ent,” Business Week, 30 (May/June), 24–29.

Boshoff, Christo and Janine Allen (2000), “The Influenceof Selected Antecedents on Frontline Staff’s Percep-tions of Service Recovery Performance,” Interna-tional Journal of Service Industry Management, 11(1), 63–90.


Brown, Steven W., Raymond P. Fisk, and Mary Jo Bitner(1994), “The Development and Emergence of Ser-

vices Marketing Thought,” International Journal ofService Industry Management, 5 (1), 21–48.

For further information contact:Nigel Piercy

Department of Marketing and Strategic Management GroupRoom C2.09

Warwick Business SchoolThe University of Warwick

Coventry, CV4 7ALUnited Kingdom

Phone: +44(0)24.7652.3911E-Mail: [email protected]


FAILURE AND RECOVERY IN INTERNET PURCHASING: THEIR

IMPACT ON CONSUMER TRUST AND LOYALTY TO THE

COMPANY’S SITE AND TO THE ONLINE

SHOPPING ENVIRONMENT

Daniel von der Heyde Fenandes, UFRGSCristiane Pizzutti dos Santos, UFRGS

ABSTRACT

This paper aims to examine the impact of recoveryefforts on consumer trust and loyalty following an unsat-isfactory Internet purchasing experience. The authorsdevelop and test a theoretical model focusing on interre-lationships among specific evaluations of complainingprocess, prior experience, familiarity, trust, perceivedvalue, and loyalty. Findings indicate an enhancing role ofthe interpersonal treatment by the e-retailer on the con-sumer perceptions of the online recovery process. Con-sumer trust is influenced by post-complaint satisfaction,familiarity, and quality of prior experiences with thespecific website and with Internet purchasing. Repur-chase intention and word-of-mouth communication areinfluenced by dimensions of trust.

INTRODUCTION

Since its expansion in the second half of the 1990s,there have been a number of innovations in the Internet,such as e-mail, free content, e-Banking, free music, in-stant messenger, and e-commerce. The latter has arousedgreat interest due to its rapid growth and great impact inthe business context. The prominent growth of electronicretailing has been accompanied by the emergence ofunique and inherent characteristics of the online environ-ment, such as the lack of physical contact between firmand customer and the perception of insecurity regardingthe Internet (Warrington, Abgrab, and Caldwell 2000), aswell as the diagnosis of recurrent failures, some of whichare new (when compared to the offline context) (Forbes,Kelley, and Hoffman 2005). For every four purchasesmade online, at least one results in some kind of dissatis-faction involving undue information, delays in uploadsessions, difficulty in payment, delayed delivery, extracharges related to the purchase or delivery, receipt ofincomplete orders or damaged products (Zemke andConnellan 2001).

Exacerbating this situation, electronic retailing seemsto be an almost perfect market, as the information isinstantaneous and the buyers can compare offers from allover the world. So, with the competition just one mouseclick away, the exodus of customers becomes much easier(Shankar, Smith, and Rangaswamy 2003). Within this

context, the correction of mistakes, that is “fixing” whatwent wrong the first time round, becomes essential forretaining customers and consequently, for the firm sur-vival in the business environment.

A consistent theoretical framework sustained byempirical evidence suggests that satisfactory complainthandling contributes to customer retention, loyalty andcommitment (Kumar 1996; Tax, Brown, and Chandrashek-aran 1998; Smith and Bolton 1998; Santos and Fernandes2005). On the other hand, failure to revert customerdissatisfaction represents a double failure and could leadto reduced trust, loss of customers and negative word-of-mouth (Berry and Parasuraman 1991). However, the vastmajority of research carried out on this topic has involvedthe traditional offline retail environment.

Accordingly, it is possible to consider some of thedifferences associated with the complaint handling inthese two business environments. Firstly, the interaction,mainly face-to-face, that is so central in failures andrecovery processes in traditional retailing, in the onlineenvironment, is substituted by technology. Another factorthat appears to differentiate recovery online is the factthat, while the experience of consumers in the traditionalretail environment is frequent and longstanding, onlineretailing is a relatively new phenomenon, with the con-tinuous entry of new buyers. Earlier research has shownthat, when the customer has little or no previous experi-ence of consuming, he/she forms an assessment based onwhat he considers fair to receive (Berry and Parasuraman1991) and not on realistic expectations, formed on thebasis of previous real performance (Woodruff, Cadotte,and Jenkins 1983). It is believed, therefore, that it isparticularly important to make rapid and efficient recov-ery efforts when failures occur in online retailing, as thecustomers in this environment expect fair results, morethan in the traditional retail environment (Ahmad 2002).

The incipient nature of the area is shown by the smallnumber of studies available, which contribute to theformation of knowledge about recovery in electronicpurchasing, offer insights into the topic and open up awhole new avenue of research. Descriptive studies havesought to identify and classify failures that occur inelectronic retailing, the recovery strategies most com-


monly used and the consequences of such strategies(Ahmad 2002; Holloway and Beatty 2003; Nasir 2004;Forbes et al. 2005). Coming closer to what has beendeveloped in the offline environment, Holloway, Wang,and Parish (2005) sought to empirically investigate themoderating influence of cumulative online purchasingexperience by testing a “traditional” service recoveryframework within the context of online retailing, usingdistributive justice, experience, satisfaction with com-plaint handling, and word-of-mouth communication.However, other hand, other important variables have beenneglected (e.g., other dimensions of justice, value, famil-iarity, and trust).

Based on this scenario, the present study aims toinvestigate the impact of the complaint handling processrelated to online purchasing on consumer trust and loy-alty. More specifically, a theoretical model will be con-structed and tested, in which inter-relations between thespecific aspects of the complaints process will be ana-lyzed together with the aforementioned constructs per-ceived value, prior experience, familiarity, trust, andloyalty. This article consists of three main parts. The firstdeals with the theoretical foundation underpinning thestudy, and the framework. The second discusses theaspects related to the research method, and the thirdpresents the results.

THEORETICAL FOUNDATION AND

RESEARCH HYPOTHESES

Theory of Justice and Satisfaction with Complaint

Handling

A complaint handling process begins with a com-plaint made by a customer, that, typically, produces (a)interaction between the claimant and representatives ofthe firm and (b) outcomes (Tax et al. 1998). Contempo-rary studies on complaints management has offered sub-stantial evidence on the suitability of the concept offairness as a base for understanding the process of com-plaining and its outcomes (Goodwin and Ross 1992;Blodgett et al. 1997; Tax et al. 1998; Smith et al. 1999).Distributive Justice refers to the allocation fo benefits andcosts between the parties to a transaction. Within thecontext of the complaint, distributions are seen as atangible outcome offered to the claimant by the firm (e.g.,exchanging the product or reimbursing the money). Thesecond dimension, Procedural Justice, concerns policiesand procedures used by firms and have six sub-dimen-sions: flexibility, accessibility, process control, decisioncontrol, speed of response and acceptance of responsibil-ity (Thibaut and Walker 1975; Tax et al. 1998). Interac-tional Justice is involving the manner in which the em-ployees treat and communicate with the customer duringthe complaint. Six sub-dimensions have been studied:courtesy, honesty, empathy, effort, explanations, and

apologizing (Clemmer 1988; Tax et al. 1998). It is impor-tant to highlight that the latter may appear of little signifi-cance in the case of online purchasing, due to the absenceof personal contact with employees. However, thoughirrelevant at the time of purchase, this dimension of justiceappears to become an important element for an efficientrecovery, due to the consumer need, either through e-mailor over the phone, to feel well.

Post-purchase satisfaction has been considered acentral mediator that links prior beliefs to post-purchasecognitive structures, communication and repurchase be-havior (Westbrook 1987). Similarly, satisfaction withcomplaint management can be considered a central ele-ment mediating the relationship between assessmentsmade regarding such management and post-complaintattitudes and behavior. According to the literature onsocial justice, satisfaction is linked to assessments offairness in various conflict situations (Messick and Cook1983). Extending this logic to complaint handling, it iswidely recognized that the consumer satisfaction with thecomplaint episode results from the assessment of aspectsregarding the final outcome, the process that led to theoutcome and the manner in which the consumer wastreated and informed during the episode, that is, how fairthese aspects were (Blodgett et al. 1997; Tax et al. 1998;Smith et al. 1999; Mattila and Patterson 2004; Liao 2007).Thus, this study presents the following hypotheses:

H1: Perceptions of interactional justice will have a posi-

tive impact on satisfaction with complaint handling.

H2: Perceptions of distributive justice will have a posi-

tive impact on satisfaction with complaint handling.

H3:

Perceptions of procedural justice will have a positiveimpact on satisfaction with complaint handling.

Satisfaction with Complaint Handling in Online Pur-

chasing and Consumer Trust

The construct of trust has been widely studied in theelectronic commerce literature. It has been seen as funda-mental in explaining the consumer repurchase behavioron the Internet (Pavlou 2003; Sultan et al. 2005), as wellas outside it (Morgan and Hunt 1994; Nooteboom et al.1997; Garbarino and Jonhson 1999). Trust is a conceptstudied in several disciplines, and, as a result, there aredifferent definitions. Consumer trust in online purchase isdefined as “the willingness of the consumer to be vulner-able to actions of the electronic retailer during an Internettransaction, based on expectation that the retailer willbehave in a suitable manner, independently of the abilityof a consumer to monitor or control that electronic re-tailer” (Lee and Turban 2001, p. 79). In the same way thatthe trust of the consumer in a service supplier is based ontwo distinct facets – trust in the frontline employees and


in the business practices (Sirdeshmukh et al. 2002), con-sumer trust in an Internet purchase would have twodimensions: trust in the purchase site and in the Internet asa whole (Chen and Dhillon 2003). These facets are, then,distinct nodules on which the customer makes indepen-dent judgments during the course of the exchange. Thisidea is corroborated by Grabner-Krauter and Kalusha(2003), who dealt with two dimensions of trust in theonline environment: trust based on the system (system-based trust) that would be trust in the Internet as a mean ofpurchase and trust in the particular retailer (e-tailer trust).According to the same authors, while the latter has beenwidely studied, the former has been largely neglected.

The role of satisfaction as a central element, linkingperceptions regarding recovery processes to future atti-tudes and behavior, has been confirmed in the offlineenvironment (Bitner et al. 1990; Tax et al. 1998; Dube andMaute 1998). However, the focus has been on immediateintentions of behavior (purchase intentions, for example)than in variables such as trust, that reveal the potential forlong-term relationships. The study developed by Tax et al.(1998) is one of the few that explore the relationshipbetween post-complaint satisfaction and relational vari-ables. They found a significant and positive relationbetween satisfaction and post-complaint trust, while em-phasizing the central role of managing conflict in theproducing (or reducing) trust between the parties in-volved.

In the online environment, some authors confirm theinfluence of satisfaction on trust in the online retailer(Pavlou 2003; Ribbink et al. 2004). This is due to thenotion that, when the consumer perceives the perfor-mance of the firm as fair and satisfactory, his/her feelingsof trust tend to be strengthened. Conflict situations appearto be even more critical, as it is in adversity that partner-ships are put to the test. Accordingly, satisfactorily solu-tion of a failure in delivering a product, for example, cansay more about the credibility of a certain site than aroutine situation. Similarly, given that the experience ofmaking purchases on the Internet is a relatively new, anefficient recovery process can help increase the trust inelectronic commerce as a whole. Ribbink et al. (2004), infact, state that satisfactory experiences with a specificelectronic retailer inspire consumer trust in the virtualmedium (“system-based trust”). Therefore, it can be sug-gested that:

H4: Consumer-perceived satisfaction with complaint han-

dling in online purchasing will have a positive impacton consumer trust in the company’s site(a) and inonline retailing(b).

Trust is a critical factor in any relationship in whichthe trustor (consumer) does not have direct control over

the actions of a trustee (store), the decision is importantand the environment is uncertain (Mayer 1995). Drawingfrom cognitive consistency arguments, trust on Internetshopping is proposed to directly influence consumer truston the firm’s web site. When the salesperson is absentfrom or peripheral to the selling and buying process, as isgenerally the case with Internet stores (Lohse and Spiller1998), then the primary target of the consumer’s trust isthe Internet as a whole (Chow and Holden 1997). More-over, when consumers have little or no prior purchasingexperience with a specific web site, their trust in Internetpurchasing and their experiences with other Internet sitesare likely to have an impact on their trust in a specific website through a process of affect transfer: trust in theInternet shopping may influence trust in a company’s sitebecause offers a guarantee of the consistent and compe-tent performance of the companies operating in the envi-ronment. The literature also offers support for a reciprocaleffect between trust in the firm’s and in Internet. Accord-ing to dispositional approach, consumers develop atti-tudes and tendencies to respond to industry-context situ-ations in a particular, predetermined manner (Nijssenet al. 2003). These dispositions emerge as consumerssynthetize information and experiences across multipleexchange within an industry to develop higher-level in-ferences. Within an industry context, such sense makingis aggregated across sellers, experiences and time througha process of assimilation, resulting in an industry disposi-tion. Thus, it is proposed:

H5: Consumer trust in Internet retailing as a whole will

have a reciprocal impact on the consumer trust in thecompany’s site.

Prior Experiences, Familiarity, and Consumer Trust

In an online environment, prior experience withonline purchases plays an important role, as it linksprevious behavior with the probability of that behaviorbeing repeated (Cho 2004). The consumer experience canbe either direct such as searching for information, pur-chase and consumption of products, or indirect such asexposure to propaganda and observation of other consum-ers (Alba, Hutchinson, and Wesley 1987). The constructof experience of consumption is not limited to a specificcategory of product, but is also related to diverse purchasesituations, as well as the assessment of purchase alterna-tives (Suk and Mitchell 2004). Therefore, in the same wayof trust operationalization, cumulative consumer experi-ence involves two dimensions: (1) specific experienceswith the site in question and (2) experiences with purchas-ing online. Consumer prior experience with the firm’s siteor with the online environment in general can be valencedpositively with good, pleasant and valuable experiencesor negatively with bad, unpleasant, unvaluable experi-ences. Thus, by affecting the consumer’s ability to make


predictions based on benefits and value received, priorexperiences are expected to positively impact consumertrust.

Drawing on attribution theory we posit that positiveprior experiences will lead to beneficial and benevolentattributions of firm behaviors enhancing feelings of trust.Positive prior experiences with the firm’s website providea cue for its charitable action. Negative prior experiencegenerates suspicions and doubts about the website, dete-riorating consumer trust on the site. In sum, as trust isformed not as a result of a single episode but based onsuccessive good experiences with the firm, it is expectedthat experiences prior to the complaint, in addition to theinfluence of satisfaction with the recovery process, sig-nificantly influences trust in the firm’s site. Therefore, itis proposed that:

H6: Prior experience with company’s site will positively

influence consumer trust in the site.

It is also proposed that prior experience on onlineretailing environment will impact consumer’s trust on themedium. When prior experiences with Internet purchas-ing are bad, consumers may develop uncertainty regard-ing the Web. This effect could be explained by theTechnology Acceptance Model. Rogers (1995) incorpo-rated prior experience as a component of the knowledgestage that impacts the formation of attitudes to adopt aninnovation. Satisfactory prior practice enhances trialabilityand observability, which is important for a system adop-tion. Thus, consumers tend to accumulate knowledge andbeliefs about the new system, which will increase theiroverall attitudes toward it (Yoh et al. 2003). Based on thisreasoning, it is proposed that:

H7: Prior experience with online purchasing will posi-

tively influence consumer trust in online retailing asa whole.

While prior experience has to do with the valence ofthe experience (good, pleasant, satisfactory or bad, un-pleasant, dissatisfactory), familiarity is related to theknowledge acquired and stored in memory, and with theability to make decisions (Bettman and Park 1980; Alba,Hutchinson, and Wesley 1987). Familiarity on the firm’ssite reflects the consumer’s knowledge about purchasingfrom a specific online retailer. Consequently, it deals withknowing and understanding the trading partner, its workprocedures, goals, beliefs and values. It builds trust bycreating appropriate frameworks within which the trustorcan place its beliefs about the trustee and know what toexpect from it (Gefen 2000). Another way in whichfamiliarity builds trust is that familiarity results in lesscommunicational misunderstandings (Kumar 1996) thatwould otherwise ruin trust, even unintentionally. There-fore, it is proposed that:

H8: Familiarity with the company’s site will positively

influence consumer trust in the site.

Familiarity could also be related to Internet shoppingexperience in general and not only to a specific web site(Sultan et al. 2005). From this, consumers will accumulateknowledge about purchasing on the Internet and, there-fore, will become more skillful. Familiarity on the Internetthus reflects the ability of the consumer to understandinformation based on the Internet (Moreau et al. 2001). Anexpert user is more likely to have greater confidence onthe web than a novice user (Sultan et al. 2005). Therefore,trust may be higher for an expert or web-savvy consumer:

H9: Familiarity with online shopping will positively in-

fluence consumer trust in the online retailing as awhole.

Antecedents of Loyalty

In the words of Reichheld and Schefter (2000, p. 113):“loyalty is not won with technology, but through theconsistent delivery of greater value to the customer.”Loyalty is usually composed of two dimensions: intentionto repurchase and positive word-of-mouth (WOM). Withregard too word-of-mouth, some differences become ap-parent comparing online and offline media. The Internethas given rise to new forms of communication amongconsumers and extended the possibilities for consumers toreceive and supply unbiased information (Hennig-Thurauand Walsh 2004). Using the Internet, consumers can shareinformation and experience in chats, emails and commu-nities. Therefore, virtual WOM appears to be faster andreach further than that made offline.

Conceptual (Nooteboom et al. 1997) and empirical(Morgan and Hunt 1994; Tax et al. 1998; Agustin andSingh 2005) studies emphasize trust as a fundamentalingredient for the development of long-lasting relation-ships between consumers and firms. In online commerce,trust appears to be even more relevant since there is a lackof the physical contact and tangible aspects which arepresent in the traditional retailing. Therefore, the firmsuccess depends on building trust among consumers anda favorable attitude toward the web. This idea is rein-forced by Pavlou (2003) and Ribbink et al. (2004), thoughthe latter authors note that there is still a lack of empiricalevidence for the link between trust and loyalty in theelectronic environment. Thus, it is proposed that:

H10

: Consumer trust in the site will have a positive impacton (a) repurchase from the site intention and (b)positive WOM regarding it.

H11

: Consumer trust in online retailing will have a positiveimpact on (a) the repurchase from the Internet inten-tion and (b) positive WOM about it.


In addition, this study proposes the existence of asecond precursor of loyalty: the perceived value. Per-ceived value refers to the consumer’s assessment of thebenefits and costs of maintaining a relationship with a firm(Sirdeshmukh et al. 2002). The logic here is that, evenwhen the consumer extracts benefits from the exchange,if high maintenance costs are demanded, the relationshipwill be less attractive and there will be less evidence ofloyalty. The findings from Sirdeshmukh et al. (2002) offerempirical evidence of this reasoning. In their study, valueemerged as a significant antecedent of loyalty. In theonline context, Broekhuizen (2006) found the effect ofperceived value items (time and effort) on repurchaseintentions. Based on this reasoning, it is proposed that:

H12

: The value perceived of the site will have a positiveimpact on (a) repurchase from the site intention and(b) positive WOM regarding the site.

Based on the theoretical framework and establishedhypotheses, Figure 1 presents the theoretical model to beinvestigated.

METHOD

This study consisted of two parts, using a mix ofqualitative and quantitative approaches. Firstly, we em-ployed personal interviews to raise relevant questions

regarding complaint handling processes involving onlinepurchasing and refine items for the quantitative phase.Next, we administered a cross-sectional research withrespondents with recent complaints about online pur-chases. During the exploratory phase we have done twentyin-dept interviews. Using a mail with colleagues names, aconvenience sample of e-consumers with complaint ex-periences was found. Through the interviews with themwe tried to answer questions, such as: what dimensions dothe customers use to assess the process of failure recov-ery? Are the dimensions of justice perceived? Whendealing with online shopping, do experience, trust andloyalty refers to two different levels – the site and onlineretailing as a whole, as foreseen in the theory? Do vari-ables at the micro level (site) influence the variablesrelated to the macro level (electronic retailing)? If so, forwhat does this influence take? Does perceived valueinfluence only re-purchasing and WOM to the site, or alsorepurchase and WOM regarding the Internet as a whole?

For the quantitative phase, 3,339 customers from allover Brazil who were engaged in complaint processeswithin the past six months responded to an online ques-tionnaire. This was achieved in partnership with the E-bit(www.ebit.com.br), a company specialized in researchinvolving Internet purchasing. Only those consumerswho had experienced some failure in an online purchaseand had complained to the e-company in the last six

FIGURE 1

Theoretical Model

Figure 1: Theoretical model

H4ADistributive

Just ice

Trust in the Online

Retailing

Loyalty (Positive Word-of-mouth about the Site)

Loyalty (Retention to the Site)

PerceivedValue

H1

H10A

H11A

Trust in the Firm’s Site

H4B

Experience with the site H6

H5H3

Satisfaction with the

Complaint Handling

Experience with Online Purchasing

H2

Loyalty (Positive Word-of-mouth

about the Internet)

Loyalty (Retention to the Internet)

H12B

H12A

ProceduralJust ice

Interactional Justice

Familiarity with the site

H8

Familiarity with Internet Purchasing

H9

H7

H10B

H11B


months were asked to answer the research. In exchangethey received 7,000 bits, to be used to compete for prizeson the site. Data was examined to ensure that it abides tothe multivariate assumptions of normality, homoscedas-ticity and linearity. Furthermore, we seek to remove caseslabeled as outliers. After making these examinations, weretained 3,206 cases in our sample. The respondence ratewas 16.5 percent, which can be considered high, consid-ering that only those that had had a problem with an onlinepurchase could respond and participation was spontane-ous.

The initial portion of the instrument contained ques-tions that required subjects to report the details of a recentonline purchase experience leading to their lodging acomplaint. Details included a description of what causedthe complaint, the media used to lodge the complaint andthe time of the year the complaint was made. It provideduseful information and helped the respondent focus on theencounter. Following those initial open- and closed-endedquestions, scaled measures adapted from previous studieswere applied. Interactional (6 items), procedural (6 items),distributive justice (4 items), and satisfaction with com-plaint handling (3 items) were taken from Tax et al.(1998); perceived value (4 items), consumer trust (8items; 4 related to trust in the firm site and 4 related to trustin online environment) were adapted from Sidershmukhet al. (2002); repurchase intention (4 items) and positiveword-of-mouth (3 items) from Zeithaml et al. (1996). Theonly measure taken from studies deriving from researchon the online environment was familiarity (6 items; 3related to familiarity with the website and 3 to familiaritywith the online environment) from Sultan et al. (2005).Although there is a plenty of research on online trust andmeasures from it (e.g., Sultan et al. 2005; Chen andDhillon 2003; Lee and Turban 2001), Sidershmukh et al.(2002) scale reflects important aspects of trust – compe-tence, integrity and problem solving orientation. Mea-sures were translated using the back-translation techniqueand a pre-test, gathering 15 people was done.

RESULTS

First, the measurement model will be examined.Then the examination of the structural model will beperformed.

Validity and Reliability

Based on several authors (e.g., Hair et al. 1998;Churchill 1999), the validity of this model is basicallysupported if: (a) the measurement model adapts to the datareasonably well that is, within the adjustment indexesconsidered satisfactory; (b) the factor loadings of items inthe corresponding factors are large and significant; (c)indicators of the same construct produce reliability in-dexes higher than 0.70 and extracted variance over 0.50;

(d) the correlations between indicators (or factors) of thesame construct produce convergent validity; (e) the analy-sis of correlations indicates discriminant validity. Theconvergent validity was supported by the fact that everyitem, with no exceptions, presented high and significantfactorial coefficients in the constructs they measured(between .59 and .96, t-values over 10.61). Furthermore,items associated to the same construct presented a signifi-cant correlation. Thus, the convergence of measures, thatis, the existence of a strong correlation between itemsmeasuring the same construct, was detected (Churchill1999).

Evidence of discriminant validity, which is particu-larly important when constructs are similar by nature, wasfound through the correlation levels among constructs.Correlation over 0.80 would indicate lack of discriminantvalidity, that is, constructs would be measuring the samephenomenon. Constructs were found distinct from oneanother, with the higher correlation between trust in thefirm website and repurchase intention (.78). Measuresprovided satisfactory levels of reliability and extractedvariance. Composite reliability was between .81 and .91(procedural and distributive justice, respectively). Ex-tracted variance was between .46 and .73 (proceduraljustice and WOM communication, respectively). It dem-onstrates internal consistency of the constructs’ measures.

Hypotheses Examination

The investigation of the set of hypotheses will beprimarily made through the goodness-of-fit indices of thehybrid model and the significance and magnitude ofestimated regression coefficients. Findings of the struc-tural model are found in Table 1. The chi-square issignificant. Goodness-of-fit indexes CFI, NFI, NNFI, allover 0.90, are satisfactory and the RMSEA of 0.05 isacceptable.

The effects of justice (interactional, distributive andprocedural) on satisfaction, established on hypotheses H

1,

H2, and H

3, respectively, were supported by the results.

The R2 of 0.85 indicates that the three aspects (personaltreatment, process, and tangible results) explained anexpressive proportion (85%) of variance in consumer’sfinal satisfaction. The evaluations on processes during thecomplaint management had a small influence on the finalsatisfaction (regression coefficient of 0.25). The tangibleresults (distributive justice) obtained through complainthad a higher effect on satisfaction (0.70) than interactionalaspects (0.54), which corroborates for the importance ofoffering compensation for the customer, and not onlytreating him/her well. Results also support hypothesis H

4,

in which the relationship between satisfaction and trust isestablished. However, the impact of satisfaction on trustin the firm web site is much higher (0.65) than in onlinepurchasing as a whole (0.14). This is consistent with the


TABLE 1

Estimated Regression Coefficients for the Theoretical Relationships Established

Model Relationships Standardized Regression Coefficient Hypotheses Findings

Dependent Variable: Satisfaction with Complaint Handling

Interactional Justice 0.54 (11.78) H1

AcceptedDistributive Justice 0.70 (14.03) H

2Accepted

Procedural Justice 0.25 (5.85) H3

AcceptedR2 = 0.85

Dependent Variable: Consumer Trust in the Firm Website

Satisfaction with Complaint Handling 0.67 (13.59) H4A

AcceptedConsumer Trust in Online Retailing 0.06 (1.38) H

5No Accepted

Familiarity with the Company’s Web Site 0.35 (25.07) H9

AcceptedPrior Experience with the Company’s Web Site 0.15 (4.09) H

8Accepted

R2 = 0.61

Dependent Variable: Consumer Trust in Online Retailing

Satisfaction with Complaint Handling 0.14 (3.27) H4B

AcceptedConsumer Trust in the Company’s Web Site 0.02 (1.08) H

5No Accepted

Familiarity with the Internet 0.25 (5.25) H7

AcceptedPrior Experience with the Internet 0.36 (7.26) H

6Accepted

R2 = 0.21

Dependent Variable: Repurchase Intention from the Site

Consumer Trust in the Company’s Web Site 0.87 (16.15) H10A

AcceptedPerceived Value 0.03 (0.89) H

12ANo Accepted

R2 = 0.76

Dependent Variable: Positive Word-Of-Mouth about the Site

Consumer Trust in the Company’s Web Site 0.86 (15.68) H10B

AcceptedPerceived Value 0.01 (0.34) H

12BNo Accepted

R2 = 0.74

Dependent Variable: Repurchase Intention from the Internet

Consumer Trust in Online Retailing 0.58 (12.38) H11A

AcceptedR2 = 0.33

Dependent Variable: Positive Word-Of-Mouth about Online Shopping

Consumer Trust in Online Retailing 0.61 (13.09) H11B

AcceptedR2 = 0.38

Goodness-of-fit Indices:

χ2 (Chi-square) 24276,096 (p<0.001) DF (Degrees of Freedom) 1309CFI (Comparative Fit Index) 0.96 NNFI (NonNormed Fit Index) 0.96NFI (Normed Fit Index) 0.96 RMR (Root Mean Sq. Residual) 0.32RMSEA (Root Mean Sq. Error of Approx.) 0.06

a t-values between parentheses. Based on the two-tail test: t-values > 1.65 = p < 0.05; and t-values > 2.33 = p < 0.01.Significant coefficients are in bold (all at level 0.01).

idea that, in conflict situations, specific perceptions on theepisode may affect (enhancing or depleting) trust inrelation to the other party involved in the dispute, but this

effect is not extended to other online retailers. It is impor-tant to reinforce that the impact of satisfaction on con-sumer trust in the web site is much higher than the effect


of familiarity (0.35), prior experiences with the site (0.15)and consumer trust in online purchasing (0.06). Whenconsumers perceive that the company acted appropriatelyin their problem resolution, they updated impressionsabout the company and, thus, feelings of trust are rein-forced. The opposite occurs when consumers perceivethat the company handled their complaint negligentlyand/or ineffectively. The R2 of 0.61 reflects a good ex-planatory power of consumer trust in the firm’s web site.

Regarding consumer trust in online purchasing, fa-miliarity and prior experiences with the Internet exert ahigher effect than satisfaction with complaint handling.Those three variables could explain 21 percent of variancein consumer online trust. Moreover, the reciprocal effectof trust in the firm’s web site on trust in Internet shoppingwas not confirmed since this effect was not statisticallysignificant (0.02). The impact of trust on the consumerrepurchase intention from the same web site, supposed onH

10A was confirmed. Nonetheless, whereas the level of

consumer trust presented an expressive effect on theconsumer repurchase intention (regression coefficient of0.87), the relationship between value and repurchaseintention was null (coefficient of 0.03), rejecting the H

12A.

Concerning word-of-mouth communication about thefirm’s web site, similar effects were encountered, that is,trust exerted a major impact and perceived value did notpresent a significant coefficient estimate, rejecting hy-pothesis H

12B. The proportion of variation of the word-of-

mouth about the web site dimension explained by trust inthe firm’s web site (0.74) support the importance of truston the consumer intention to recommend a firm’s web siteand on the development of ongoing relationships betweenthe consumer and the firm. Finally, the impact of trust inonline purchasing on repurchase intention from the Internetand on WOM about online purchasing, predicted by H

12A

and H12B,

posits trust as an important antecedent of thecustomer’s actions toward the Internet in the future.Moreover, the R2 estimates also reveal that, besides thelarge effect of trust, other predictors may be employed toimprove the proportion of variance explained.

CONCLUSIONS AND IMPLICATIONS

This article aimed at providing and testing a theory-based framework for mapping the mechanisms underly-ing recovery process regarding online exchanges. Thisstudy offers three distinct contributions. First, we concep-tualize consumer trust as a bi-dimensional construct toobtain a fine-grained understanding of its differentialeffects. This conceptualization is sustained by the recog-nition of trust’s multidimensionality in the interorganiza-tional and marketing literature. Second, it proposes amodel with interrelationships among web site-specificrecovery evaluations and cognitions about online pur-chasing as a whole (e.g., intention to repurchase onlineand positive word-of-mouth about Internet shopping).

This provides valuable insights for a dynamic portrayal ofonline purchasing. Third, our results support the view thatjustice concepts, in the online context, provide an effec-tive theoretical framework for explaining satisfactionwith complaint situations, which in turn is strongly asso-ciated with both trust and loyalty, offering empiricalsupport for the proposition that complaint handling is tiedclosely to relationship marketing. Each contribution pro-vides new insights on recovery process involving con-sumers and virtual providers.

While the bi-dimensionality of trust is ultimately anempirical question (not just psychometric), systematicstudies establishing the validity and usefulness of trustdimensions have been lacking in the literature (Singh andSirdeshmukh 2000). While these dimensions of consumeronline trust has been supported in previous research(Chow and Holden 1997), their interrelationships havenot been proposed and tested heretofore. However, differ-ently of the confirmatory results related to the two firstpropositions, we found no influence of the trust incompany’s site on the trust in Internet purchasing, andvice-versa. It offers support for distinct nodes on whicheach dimension of consumer trust is formed. Moreover,the results reveal that, not only the trust in the web site andin the Internet medium are not associated, but also thatconstructs related particularly to the web merchant (e.g.,satisfaction with complaint handling and trust in thecompany’s web site) do not impact more “general” vari-ables such as trust and loyalty to the online environment.This lack of influence shed some light about the existenceof other mechanisms and variables, such as consumeroverall satisfaction (not measured in this study). Theresults reveal that recovery efforts are essential for main-taining long-lasting relationships with the company’sweb site but do not influence customers’ view aboutInternet shopping as a whole. One explanation for this isthat we dealt with initiated consumers since a large part ofour sample is composed by consumers who had reason-able shopping experience on the Internet. These consum-ers know exactly how Internet works and are less inclinedto change attitudes by a singular experience.

Analyzing the links between the dimensions of trustand the dimensions of loyalty, consumer trust emerged asthe strongest antecedent of repurchase and recommenda-tion intentions. This result is in accordance with the logicthat by increasing consumer trust, they start to believe thatthe company will continue to act consistently and compe-tently in the future, thus reducing the risks associated toservice purchase and creating the belief that consumerswill continue to obtain value in further businesses with theprovider. In other words, the higher the consumers trustthe higher the probability to make future exchanges withthis company. An inappropriate complaint handling wouldgenerate a “double deviation,” that is, the company failstwice to meet the customer’s needs. Then, consumer trust


is reduced, or even destroyed, leading the customer to acompeting company.

Another point to emphasize refers to the impact ofinteractional dimensions of justice on satisfaction withcomplaint handling. Both past research into the serviceencounter and Fairness Theory (Folger and Cropanzano1998) suggest that considerations of interactional justiceare crucial in service recovery situations. In the onlineenvironment, the role interpersonal treatment was sup-posed to be weaker or even not exist because there are fewopportunities for employees to be in contact with e-consumers. Contrary to this, the findings establish that inonline purchase recoveries poor personal treatment vio-lates basic human needs such as fairness and a sense ofself-worth and may result in customer defection and highintention to spread bad words in an effort to restore self-esteem. Therefore, the present study advocates for the roleof interpersonal treatment in customer–web merchantinteractions in online recovery situations.

The significant impacts of prior experiences andfamiliarity on trust, lead to the idea that investigates theimpact of the complaints in the virtual context withoutincluding aspects related to the “history” of the consumerwith this context (i.e., prior experiences, familiarity, andperceived value), would mean to omit important casualvariables, and ignore the dynamic that guides marketingexchanges. Besides, it was expected that online consum-ers would value aspects such as time, convenience, andeffort on future exchanges with the same provider, but wefound a very weak impact of value on loyalty to the website. Although it seems surprising, this lack of influencewas already found by other authors (e.g., Baptista 2005;Warrington 2002). One explanation for this is that con-

sumers can easily make comparisons online and do notdistinguish the attractiveness of alternative suppliers. Asthe Internet provides convenience and information almostinstantaneously, consumers do not construct a distin-guished value for a singular web merchant.

We recognize that the proposed model is an initialattempt toward modeling complaint handling in the onlineenvironment phenomenon. We do not view our model asa novel approach for understanding failure-recovery pro-cess related to online exchanges. Rather, our model bringstogether different perspectives and streams that are wellgrounded, and we hope that it will serve to motivateinterest and encourage future researchers to refine andenhance the proposed framework.

Contributions brought by this study should be con-sidered bearing in mind its limitations. This paper used across-sectional approach, based on a non-probabilisticsample, composed of people who were online when datawere collected. Under this perspective, the generalizationof results is limited. We believe that future focus oninteractions between the dimensions of justice and itsimpact directly on consumer trust will promote a betterunderstanding on the service recovery phenomenon. Wealso instigate other researchers to the study the role ofswitching costs and its effect directly on loyalty and as amoderator in the proposed framework. Another recom-mendation for future research is the analysis of the impactof the types of relationships between consumers and thecompany – a more transactional or relational orientation –on those constructs and also moderating the relationshipsestablished in the model, following Gabarino and Johnson(1999).

REFERENCES

Agustin, C. and J. Singh (2005), “Curvilinear Effects ofConsumer Loyalty Determinants in Relational Ex-changes,” Journal of Marketing Research, 42 (1),96–108.

Ahmad, S. (2002), “Service Failures and Customer De-fection: A Closer Look at Online Shopping Experi-ences,” Managing Service Quality, 12 (1), 19–30.

Alba, J., W. Hutchinson, and J. Wesley (1987), “Dimen-sions of Consumer Expertise,” Journal of ConsumerResearch, 13 (4), 411–55.

Berry, L. and A. Parasuraman (1991), Marketing Ser-vices: Competing Through Quality. New York: TheFree Press.

Bettman, J.R. and C.W. Park (1980), “Effects of PriorKnowledge and Experience and Phase of the ChoiceProcess on Consumer Decision Processes: A Proto-

col Analysis,” Journal of Consumer Research, 7 (3).Bitner, M.J., B.H. Booms, M. Tetreault (1990), “The

Service Encounter: Diagnosing Favorable and Unfa-vorable Incidents,” Journal of Marketing, 54 (Janu-ary), 71–84.

Blodgett, J.G., D.J. Hill, and S.S. Tax (1997), “TheEffects of Distributive, Procedural, and InteractionalJustice on Post-Complaint Behavior,” Journal ofRetailing, 73 (2).

Broekhuizen, T. (2006), “Understanding Channel Pur-chase Intentions: Measuring Online E Offline Shop-ping Value Perception,” Doctoral Thesis, GroningenUniversity.

Chen, S. and G. Dhillon (2003), “Interpreting Dimensionsof Consumer Trust in E-Commerce,”InformationTechnology and Management, 4 (2/3).

Clemmer, E.C. (1988), “The Role of Fairness in CustomerSatisfaction with Services,” Doctoral Dissertation,


University of Maryland.Dube, L. and M.E. Maute (1998), “Defensive Strategies

for Managing Satisfaction and Loyalty in the ServiceIndustry,” Psychology and Marketing, 15 (8), 775–91.

E-BIT (2005), Webshoppers. [http://www.e-commerce.org.br. Assess: 15 de abr. 2005].

Forbes, L., S. Kelley, and D. Hoffman (2005), “Typologiesof E-commerce Retail Failures and Recovery Strate-gies,” The Journal of Services Marketing, 19 (5).

Garbarino, E. and M. Johnson (1999), “The Roles ofSatisfaction, Trust and Commitment for Relationaland Transactional Consumers,” Journal of Market-ing, 63, 70–87.

Goodwin, C. and I. Ross (1992), “Consumer Responses toService Failures: Influence of Procedural and Inter-actional Fairness Perceptions,” Journal of BusinessResearch, 25, 149–63.

Grabner-Krauter, S. and E. Kaluska (2003), “EmpiricalResearch in Online Trust: Review and Assess,” Inter-national Journal of Human-Computer Studies, 58(6), 783–812.

Hair, J., R. Anderson, R. Tatham, and W. Black (1998),Multivariate Data Analysis. New Jersey: PrenticeHall.

Holloway, B.B. and S.E. Beatty (2003), “Service Failurein Online Retailing – A Recovery Opportunity,”Journal of Service Research, 6 (1), 92–105.

Holloway, B., S. Wang, and J. Parish (2005), “The Roleof Online Purchasing Experience in Service Recov-ery Management,” Journal of Interactive Marketing,19 (3).

Kumar, N. (1996), “The Power of Trust in Manufacturer-Retailer Relationships,” Harvard Business Review,(November/December), 92–106.

Lee, M.K.O. and E. Turban (2001), “A Trust Model forConsumer Internet Shopping,” International Jour-nal of Electronic Commerce, 6 (1), 75–91.

Liao, H. (2007), “Do it Right this Time: The Role ofEmployee Recovery in Customer-Perceived Justiceand Loyalty after Service Failures,” Journal of Ap-plied Psychology, 92 (2), 475–89.

Mattila, A. and P. Patterson (2004), “Service Recoveryand Fairness Perceptions in Collectivist and Indi-vidualist Contexts,” Journal of Service Research, 6(4), 336–46.

Messick, D. and K. Cook (1983), Equity Theory: Psychoand Socio Perspectives. New York: Praeger.

Mitchell, A. and P.A. Dacin (1996), “The Assessment ofAlternative Measures of Consumer Expertise,” Jour-nal of Consumer Research, 23 (3), 219–40.

Moreau, C., D. Lehmann, A. Markman (2001), “En-trenched Knowledge Structures and Response toNew Products,” Journal of Marketing Research, 38(1), 14–30.

Morgan, R.M. and S.D. Hunt (1994), “The Commitment-

Trust Theory of Relationship Marketing,” Journal ofMarketing, 58 (July), 20–38.

NASIR, A. (2004), “e-Consumer Complaints About On-Line Stores,” Journal of Consumer Satisfaction, Dis-satisfaction and Complaining Behavior. 17, 68–88.

Nooteboom, B., H.E. Berger, and N.G. Noorderhaven(1997), “Effects of Trust and Governance RelationalRisk,” Academy of Management Journal, 40 (2),308–38.

Pavlou, P.A. (2003), “Consumer Acceptance of Elec-tronic Commerce. Integrating Trust and Risk with theTAM,” International Journal of Electronic Com-merce, 7 (3), 101–34.

Reichheld, F.F. and P. Schefter (2000), “E-Loyalty. YourSecret Weapon on the Web,” Harvard Business Re-view, 105–13.

Ribbink, D., A. Van Riel, V. Liljander, and S. Streukens(2004), “Comfort Your Online Customer: Quality,Trust, and Loyalty on Internet,” Managing ServiceQuality, 14 (6).

Savitt, R. (1980), “Historical Research in Marketing,”Journal of Marketing, 44, 52–58.

Shankar, V., A. Smith, and A. Rangaswamy (2003),“Customer Satisfaction and Loyalty in Online andOffline Environments,” International Journal ofResearch in Marketing, 20 (2), 153–75.

Sirdeshmukh, D., S. Singh, and B. Sabol (2002), “Con-sumer Trust, Value, and Loyalty in Relational Ex-changes,” Journal of Marketing, 66 (January).

Smith, A. and R. Bolton (1998), “An Investigation ofCustomer Reactions to Service Failure and RecoveryEncounters: Paradox or Peril?” Journal of ServiceResearch, 1 (1).

____________, ____________, and J. Wagner (1999),“A Model of Customer Satisfaction with ServiceEncounter Failure and Recovery,” Journal of Mar-keting Research, 36, 356–72.

Sultan, F., G. Urban, V. Shankar, and I. Bart (2002),“Determinants and Role of Trust in E-Business: ALarge-Scale Empirical Study,” Working Paper, MITSloan School.

Tax, S., S. Brown, and M. Chandrashekaran (1998),“Customer Evaluations of Service Complaint Expe-riences: Implications for Relationship Marketing,”Journal of Marketing, 62, 60–76.

Thibaut, J. and L. Walker (1975), Procedural Justice: APsychological Analysis. Hillsdale, NJ: LawrenceErlbaum.

Warrington, T., N. Abgrab, and H. Caldwell (2000),“Building Trust to Develop Competitive Advantagein E-Business Relationships,” Competitiveness Re-view, 10 (2), 160–70.

Westbrook, R.A. (1987), “Product/Consumption-BasedAffective Responses and Post-Purchase Processes,”Journal of Marketing Research, 24 (August), 258–70.


Woodruff, R., E. Cadotte, and R. Jenkins (1983), “Mod-eling Consumer Satisfaction Processes Using Expe-rience Norms,” Journal of Marketing Research, 20(3), 296–305.

Zeithaml, V.A., L.L. Berry, A. Parasuraman (2001), “The

Behavioral Consequences of Service Quality,” Jour-nal of Marketing, 60 (April), 31–46.

Zemke, R. and T. Connellan (2001), “Online Customers,”Executive Excellence, 18 (8).

For further information contact:Cristiane Pizzutti dos Santos

Escola de Administração/UFRGSWashington Luis, 855

90010–460 Porto Alegre – RSPhone: 51.3316.3690

Fax: 51.3316.3991E-Mail: [email protected]: [email protected]


MEASURING THE PERCEIVED IMPORTANCE THAT CONSUMERS

PLACE ON THE BENEFIT OF FREE TECHNICAL SUPPORT:

A NEW SCALE

Lee Hibbett, Freed-Hardeman University, HendersonKeith Absher, Union University, Jackson

Joshua Shackman, Touro University International, CypressDarin White, Union University, Jackson

SUMMARY

Today’s consumer durable goods marketplace is abun-dant with brands attempting to increase customer valueand gain competitive advantages by adding intangiblefeatures to their core products. Add-ons such as warranteesand guarantees have received extensive attention in mar-keting literature. Another prevalent type of intangibleoffering is free technical support to aid in product perfor-mance and customer satisfaction. However, despite thepopularity of technical assistance programs, the authorsreveal a significant gap in the research regarding this non-feature benefit. Specifically, little consideration has beengiven to measuring consumers’ perceptions of just howimportant it is to have free technical support included withthe purchase of a product and how the inclusion of freehelp affects purchase decisions. The purpose of this studyis to validate a new scale to measure the importanceconsumers place on the benefit of this type of service fora specific product domain.

It might be assumed that one’s perception of impor-tance is a one-dimensional construct that could probablybe measured adequately with a single, straightforwardquestion. In fact, marketing studies have utilized single-item rating scales to determine consumers’ perceptionsregarding which product or service attributes are more orless important. However, research indicates that single-item scales tend to be highly unreliable. Also, the purposeof this study is not to compare the importance of variousfactors with one another, but instead, is to arrive at ameasurement that meaningfully depicts consumers’ atti-tudes toward the feature of free help/technical assistance.Thus, validating a scale with multiple items should shoreup the scale reliability by making sure the respondentsdisplay consistency in attitude for similar questions re-garding this concept.

This three-study project placed convenience samplesof university students in a purchase situation for a new,digital camera for personal use. This scenario was se-lected primarily because it presented a realistic purchasedecision task for the sample. This product category wasalso deemed appropriate because some brands do, in fact,offer free technical support as a feature. Relevant camera

features were chosen by referencing 34 university stu-dents, two shopping guide Web sites, and two profes-sional photographers. A final list of 18 features wasincluded in various combinations for three to nine un-named brands representing the evoked set for respondentsto evaluate as they made their hypothetical purchasedecisions. After reviewing the camera choices and fea-tures, the participants were asked to answer the surveyquestions regarding the importance of free technical sup-port as a product benefit.

Because the concept of perceived importance is basedon individual attitudes, this study chose attitudinal re-search in the psychology and marketing literature for itemdevelopment. Specifically, twelve Likert-type questionswere either created or borrowed from previous researchwith the goal of tapping into the cognitive, affective, andbehavior components of the sample’s attitudes regardingthe importance of free technical support. Half of theseitems were worded negatively and reverse scored toensure that respondents were not inadvertently led towarda particular attitude. Before progressing to the three vali-dation studies, an assessment of the instrument was con-ducted by two Ph.D., marketing professors and two col-lege students to ensure external and face validity.

The first study, which was completed by 40 under-graduate university students, attained an acceptableCronbach’s Alpha of .79. A factor analysis suggested thattwo of the twelve items should be deleted due to cross-loading concerns, thereby increasing internal reliability to.83. Additionally, convergent and discriminant validitywere confirmed by comparing the behavior of this newscale with various concepts deemed to be related andunrelated.

Using the ten-item scale, the second study was com-pleted by 30 students, and attained a Cronbach’s Alpha of.95. All items loaded onto a single factor, and no singleitem was recommended for omission in order to improveinternal reliability. Once again, this construct’s behaviorsuggested both convergent and discriminant validity.However, retaining ten items is inconsistent with mostattitudinal, marketing scales. So, a step-wise regressionwas performed to isolate the few items that best predicted


the construct. Through this step, four items were elimi-nated, while maintaining an R2 of .993 and a Cronbach’salpha of .93 with the remaining six items.

A final study to authenticate this scale utilized a muchlarger sample of 277 respondents. An acceptableCronbach’s Alpha of .91 was produced, and once again,all six items loaded onto a single factor. These similarresults produced by a much larger sample suggest that thispurified scale is adequate for measuring the importanceconsumers place on the benefit of free technical assistancein future research.

This new scale may serve a number of researchfunctions. First, it can help marketers understand how free

help is valued within their particular product category.This information will aid practitioners in building strate-gies regarding whether or not to offer free help, how muchto offer, and how to position and promote the supportfeature. Second, by collecting demographic and psycho-graphic data from consumers, this scale can be used toinvestigate the type of individual that values free help incontrast to the type that does not. This insight would provevaluable for segmentation purposes. Third, this scale canbe studied in conjunction with other consumer behaviorconcepts and models to determine what relationshipsexist. The findings from such research could prove usefulas marketers seek to understand their target markets andthe situational interactions that affect purchase decisions.References are available upon request.

For further information contact:Lee Hibbett

Freed-Hardeman University158 East Main Street

Henderson, TN 38340Phone: 1.800.348.3481



SPECTATOR RAGE: EXAMINING THE DARK-SIDE OF FAN BEHAVIOR

Scott A. Jones, Clemson University, ClemsonStephen J. Grove, Clemson University, Clemson

Gregory M. Pickett, Clemson University, Clemson

ABSTRACT

Customer rage is a phenomenon which has receivedconsiderable attention from service marketing research-ers. The following paper provides a comprehensive over-view of customer rage in the context of spectator sport.The authors present a myriad of issues which makespectator rage a unique, complex and challenging concernfor the marketers of such services. The discussion isframed by Grove, Fisk, and John’s (2004) 4-T’s of cus-tomer rage. The authors consider the variety of and accessto rage Targets, the psychological and sociological influ-ences on Temperament prevalent in live sport, and thesundry of Triggers which make sport a more likely venuefor aggressive behavior. The concluding Treatment sec-tion presents current issues and trends facing marketersand some of the strategies at work by sport organizationsto address the problem of spectator rage.

INTRODUCTION

The potential impact of customer to customer interac-tion upon one’s assessment of a service encounter hasbeen recognized for some time (Arnould and Price 1993;Bitner, Booms, and Mohr 1994; Grove and Fisk 1997).How organizations handle such incidents is a criticalfactor affecting customers’ view of a service organizationand may affect future patronage (Bitner 1990; Grove andFisk 1997; Lovelock 1994). Consequently, the topic hasreceived significant attention in services marketing litera-ture (e.g., Bitner 1990; Frankel and Tresniowski 2000;Grove, Fisk, and John 2004; Harris and Reynolds 2003;McColl-Kennedy, Daus, and Sparks 2003).

One service industry in which customer rage hasgained a great deal of notoriety is that of spectator sport.While rage in this context has been examined in otherdisciplines, it has received little attention by marketers(Hunt, Bristol, and Bashaw 1999). Spectator sport fallswithin the Sports, Arts, and Entertainment category inFisk and Tansuhaj’s (1985) classification of service fields.As with services in general, spectator sport events arecomprised of four key components that combine to createa service encounter (Fisk, Grove, and John 2004): asetting (the venue), customers (the spectators), employees(the athletes and the venue staff), and the service perfor-mance (the athletic contest and all of the facets of its

staging within the venue). Each component has the poten-tial to impact upon one’s service experience and percep-tion of service excellence (Langeard et al. 1981).

In the subsequent discussion of incidents of rage inthe context of spectator sport, the terms spectator is usedto identify an individual that attends a live sporting event;this conceptualization distinguishes those individuals fromothers which observe a sporting event through otherchannels (e.g., radio, television, Internet). Hence, specta-tor rage represents a particular form of customer rage. Inaddition, it is our contention that spectator sport is charac-terized by a number of features that are unique amongservice contexts and that may ultimately increase thelikelihood and severity of fan rage. An understanding ofthose features and the dynamics of customer rage ingeneral are critical for both the marketers of live sportingevents (i.e., teams and leagues) and those that use specta-tor sport as a vehicle for marketing communications (i.e.,sponsors and media partners). Obviously, any circum-stances that undermine the attractiveness of sportingevents – such as the occurrence of spectator rage – mayhave an adverse effect upon the industry and significantfinancial impact for a sundry of stakeholders.

The term rage throughout our discussion refers to“violent and uncontrolled anger” (Merriam Webster’sOnline Dictionary), and we suggest that the term is syn-onymous with the concept of aggression, defined as“verbal or physical behavior grounded in an intent todominate, control, or do harm to another person” (Coakley1998, p. 175). Acts of spectator rage may include a widerange of behaviors such as taunting opposing players,coaches, and fans; harassing officials; verbal abuse, throw-ing objects on the field of play; tossing a beverage or fooditem at another fan; refusing to move out of another’s lineof sight; obscene gestures; destruction of the physicalsetting; and pushing, shoving, or striking another person.

To better understand incidents such as these and toprovide a means to systematically examine their occur-rence, we utilize a framework introduced by Grove, Fisk,and John (2004) to organize thoughts regarding customerrage in general. The following discussion will apply thatframework to the phenomenon of spectator rage with agoal of providing arguments pertaining to marketing’sinfluence upon its occurrence and its remedies.


THE FOUR T’S OF SPECTATOR RAGE

To examine the phenomenon of spectator rage weadapt a framework that was originally proposed by Grove,Fisk, and John (2004) as a means for analyzing thephenomenon of customer rage. In their treatment, Grove,Fisk, and John (2004) posit four categories of factorsrelated to the occurrence of rage, which they labeled the“4 T’”s ; included among these categories are the targetsof consumer rage behavior, the temperament of customersexpressing rage, the triggers that spark such behavior andthe treatments for preventing and managing consumerrage.

TARGETS

Perhaps more so than any other service encounter, thelive sporting event is characterized by rich and complexinteractions among spectators (i.e., customers) and ser-vice providers, often within a well-defined service setting.As with customer rage in general (Harris and Reynolds2003; Lovelock 1994), both the participants (fans andemployees) and the service organization (particularlyaspects of the physical setting) may become targets ofraging spectators. The consumption experience of a livesporting event involves interactions with a number ofstadium, arena or team representatives responsible forsupport services. Such services include concessionaires,ushers, ticket-takers, and security personnel. Whetherprovoked by their actions or the unfortunate recipients ofbehaviors played out due to other causes, such personnelsometimes become targets for the raging fan. As represen-tatives of the sport venue, spectators who have a “bone topick” with the organization may direct their disdain uponthem.

Sporting events also involve interaction, albeit rela-tively indirect, between spectators and those participatingin the game itself such as players, coaches, and referees.Indeed, the interaction between spectators and game par-ticipants (e.g., cheering, distracting an opponent withnoise, shouting at an umpire, or official) may be bestdescribed as a fundamental component of the serviceexperience. Unfortunately, some spectators who are dis-pleased with the on-field performance or shoddy officiat-ing sometimes cross the line of civility and hurl insults andobjects at those deemed responsible. Recently, throngs ofNASCAR fans who were upset that Jeff Gordon passedbeloved Dale Earnhardt in career victories hurled beercans and debris onto the track fans who were upset withJeff Gordon (Ryan 2007).

In addition, the live sporting event offers the opportu-nity for substantial interaction between spectators. Fansare often seated close to one another, a circumstance thatsometimes makes one a party to another’s unwelcomeverbal and physical behavior, – behaviors that may incite

rage. Innocent bystanders’ spectator experience can eas-ily become ruined by others prompting varied responses(Grove and Fisk 1997; Harris and Reynolds 2003). Fur-thermore, team sporting events and many individual ath-letic contests are typically attended by supporters ofopposing teams or individuals involved in the competi-tion. Consequently, the ideal outcome of the spectatorsport service encounter may vary substantially based onwhich team or individual a spectator supports, and inter-actions among spectators may reflect this conflict ofobjectives.

Sometimes the target of fan rage is not another personor persons, but rather the physical setting housing thesporting event (McColl-Kennedy and Smith 2006). What-ever the cause of their anger, like customers in general,spectators have been known to strike out at their surround-ings when emotions boil over. Ripping seats apart, pound-ing partitions, marring walls or equipment and damagingthe landscape are common expressions of spectator fury.Sometimes the fury is not rooted in anger, but rather itdevelops as a tumultuous celebratory expression such asin the case of goal posts being torn down and the like.Regardless of its underpinnings, in truth, no aspect of theservice setting is safe from such exuberance.

TEMPERAMENT

While a number of physiological factors may in-crease the likelihood of rage across any service context,there also exist a number of powerful psychological andsociological factors unique to the sport consumptionexperience. Consider the phenomenon of identification.Identification has been defined as the process of maintain-ing a positive, self-defining relationship with others onedeems attractive (Kelman 1958, 1961). Sport fans oftenview a team or a player as an extension of their self (Kahle,Duncan, Dalakis, and Aiken 2001). In this respect, iden-tification may be described as a psychological orientationof the self whereby individuals define themselves in termsof their group membership and derive strength and a senseof identity from the affiliation.

Identification often leads spectators to form a linkbetween their favorite sport properties and their self-concept (Hunt, Bristol, and Bashaw 1999). With this linkin place, threats to the sport property (i.e., losing, support-ers of an opponent, etc.) may be viewed as a threat to one’sself. In such instances, fans are likely to engage in aprocess of “blasting” or derogation as a means of enhanc-ing their identity or restoring self-esteem (Branscombeand Wann 1992, 1994). Aggression is one defense strat-egy that spectators of sport may adopt when their self-concept is threatened (Wann, Melnick, Russell, and Pease2001) and a number of studies have linked identificationwith a sports team or athlete and hostile behavior (e.g.,Branscombe and Wann 1994; Wann, Peterson, Cothran,


and Dykes 1999; Wann, Hayes, McLean, and Pullen2003).

There are other explanations for fan rage that emergefrom one’s psychological profile. For example, peoplewith passive-aggressive personalities are prone to antago-nistic verbal displays and may seek retribution for theslightest of perceived transgressions by other fans orstadium personnel (Stone 1993). Similarly, spectatorswith Type A behavioral patterns (i.e., highly competitive,intense achievement orientation) often allow their free-floating hostility to be triggered over minor incidents(Holmes and Will 1985; Friedman 1996) that manifestsitself as fan rage. Even those with borderline personalities(BP) marked by feelings of low self-worth, obsessivebehavior and need for assurance can become hostile andprovocative or push others to act similarly (Mahari 2006).

There also exist sociological explanations for specta-tor rage. For instance, the role often prescribed to specta-tors of live sporting events may encourage such aggres-sive behavior. As noted by Pratt and Cardador (2006),roles tend to dictate behavior and “complying with behav-ior norms and expectations may predispose one to identifywith a role” (p. 177). Role theory is the “study of theconduct associated with certain socially defined positionsrather than of the particular individuals who occupy thesepositions” (Solomon et al. 1985, p. 102). Part of the roleassigned to spectators of sport may be to foster a hostileenvironment for an opponent. Behaviors consistent withthis role include the creation of noise and heckling ortaunting opponents in an attempt to influence their perfor-mance. Such behaviors can provoke or incite the opponent’sfans as well as their intended target. In a sense, then,aggression and its expression as fan rage may be learnedvia socialization as appropriate behavior (Bandura 1973,1986), a circumstance that is underscored to some degreeby the names of the stadiums in which athletic contestsoccur. Stadiums and arenas are frequently branded withnames such as “Death Valley,” “The Pit,” and “TheSwamp – where only Gators come out alive” and the like.Promoting such names may also encourage the accep-tance of aggression as a means of fulfilling the role ofspectator. As noted by Brad Davis, an associate commis-sioner of the N.C.A.A.’s Southeastern Conference, “whatbecomes acceptable is to do your best to create an intimi-dating (home-field) atmosphere. And that has taken a tollon sportsmanship” (Weiberg 2002, p. 1).

Part of the acceptance of such a role for spectators ofsport may be that for both players and spectators, sport isset apart both cognitively and emotionally from the every-day world. Both anthropologists and sociologists haveobserved that sport requires a radical transformation incognition and perception. This characterization suggeststhat moral norms which prescribe equal consideration ofall people are often suspended during sporting events in

favor of a more egocentric moral perspective (Bredemeier,Shields, and Horn 1985).

Sporting events are also characterized by groups ofspectators. A characteristic of group dynamics is that, allelse being equal, social consequences for individuals willbe less in groups than they are when a person acts alone(Guerin 1994, 1998; LeBon 1960). Because an individual’svisibility to outsiders is reduced when he or she is in agroup, groups also usually involve less individual evalu-ation, less accountability, less assignment of responsibil-ity to individuals, and less direct reinforcement or punish-ment. In this respect, spectators may also learn that asmembers of a crowd attending a sporting event the crowdprovides a degree of anonymity which protects them fromrepercussions for their hostile behavior (Wann, Peterson,Cothran, and Dykes 1999; Wann, Hayes, McLean, andPullen 2003). This is perhaps most evident when the targetof the hostility is an athlete or official on the field.Moreover, it may well be that the collective excitement ofthe crowd at a spectator sport event intensifies the emo-tional responses of those among those in attendancemaking them more susceptible to rage behavior (Coakley2006)

TRIGGERS

The context of spectator sport offers many potentialtriggers prompting fan rage. At issue are the causes,expression, and controls pertaining to spectator hostility.Ironically, there may be no service industry where anorganization’s interest in reducing or eliminating the actsof aggressive behavior more directly conflict with theorganizational objectives assigned to marketers. Amongothers, sport marketers are charged with the responsibilityof providing spectators with a fun and exciting entertain-ment experience, developing and managing sponsor rela-tions, and maximizing revenue opportunities. Yet, manyof the same marketing strategies employed to achievesuch objectives may also provide the triggers that increasethe likelihood of rage incidents.

The triggers of spectator rage are essentially situ-ational factors that are present and affect spectators’experience at the sporting event. Like people in general,some spectators are more likely to be affected by situ-ational variables than others (Bandura 1973). Some timeago, Belk (1974) argued that situational factors fell intofive categories: physical surroundings (environmental oratmospheric aspects of the setting), social surroundings(the number, status and characteristics of those present),task definition (the purpose or motivations one brings tothe place), temporal perspective (the impact of time onsubsequent behaviors) and antecedent states (transitoryphysical or psychological conditions that one brings to anevent).


For the purposes of brevity, Table 1 provides a list ofthe situational factors most likely to serve as triggers ofspectator rage. Each of these triggers has been empiricallylinked to aggressive behavior.

It should be noted that these factors often operate incombination (e.g., crowded circumstances and excessivealcohol consumption) to create the propensity to rage. Itshould also be restated that many of the triggers thatprompt rage are, to some degree, a byproduct of a sportorganization’s marketing effort (e.g., seating arrange-ments close to the contest) and/or its lack of foresightregarding the triggers themselves. Fortunately, there areways in which fan rage can be addressed. Some of theseare discussed in the following section.

TREATMENT

Those organizations seeking a more responsible com-petitive environment can successfully address some of thekey issues contributing to inhospitable sport environ-ments. First, organizational capacity to deal with fan ragewill vary significantly depending upon the category offactors that might have contributed to the rage event(Figure 1). For example, events that occur at a sport venueand serve as triggers for a spectator’s wrath offer greaterpotential for control and treatment than other factorsrelated to rage, i.e., targets and temperament. In large part,

trigger events are traceable to a distinct cause and effectrelationship that may be minimized by organizationalaction. Fan rage emanating from this set of factors issomewhat predictable and can be mitigated by sportorganizations wishing to create a less volatile settingthrough their game-day management efforts. Actionssuch as lowering the music volume, curtailing alcoholsales and the like play a role here.

Slightly less treatable are those incidents initiatedunder the target category of rage factors. Targets of fanrage are often institutionalized as part of the contest andcan not be removed effectively without altering the es-sence of the contest itself. For instance, the fans, players,officials, coaches, and support personnel that may be-come targets of spectator rage contribute significantly tothe overall consumption experience. One can not simplyeliminate interaction between them and maintain theexperience that fans expect and desire. In reality, sportorganizations build an infrastructure that often supports ahigh degree of interaction and intimacy that fans relish.Moreover, millions of dollars would be required to correctservicescape dimensions that may contribute to potentialfan rage, e.g., seating sections close to the action orcramped seating arrangements. Organizational changewith respect to these dimensions would likely only ad-dress fringe issues.

TABLE 1

Situational Factors as Triggers of Spectator Rage

Situational Factor Trigger

Physical Surroundings • Close Proximity to the field• Temperature extremes• Noise level in stadium

Social Surroundings • Crowding at the venue• Absence of authority• Verbal/physical behaviors of other fans

Task Definition • Contest is a heated rivalry• Wager made on the contest• Rage Intent

Temporal Perspective • Long waits for refreshments, restrooms, admission• Hunger (length of time since last meal)• Late start (more time tailgating)

Antecedent States • Alcohol consumption• Bad mood from poor service at venue• Aggressive behavior by opposing fans at last contest

Adapted from Belk (1974)


The least addressable category of factors contribut-ing to spectator rage resides in the temperament category.Knowing that some people may be predisposed to rageincidents are certainly unsettling. However, with the ex-ception of transitory intensifiers of rage, enduring person-ality characteristics or less obvious physical states canonly be managed through employees trained to monitorand recognize inappropriate behavior as it becomes mani-fest. Thus, preemptive actions necessary to screen for thepotential disruptive fan are very difficult, and offer theleast opportunity for treatment options. One way to cat-egorize treatment options is to think of actions that mightbe taken by the organization before the event (pre-gameplanning), during the event (game day activities) and afterthe event (post game follow-up). Obviously, the result ofpre-game planning will be evident during the contest, butfor discussion purposes we have tried to categorize activi-ties based on when the substantive work required forimplementation would likely take place.

Pre-Game Planning

Recently, the NCAA held a sportsmanship summitwhere several helpful ideas were put forth by coaches andathletic administrators (Brown 2003) for curbing violenceat NCAA events. Some of the recommendations includedencouraging staff from opposing teams to participate inpre-game planning, to meet with key student groupsbefore/during seasons to review consequences of nega-tive behavior, and to consider equipment issues such asthe installation of breakaway football goal posts. In gen-eral, some of these collegiate examples may be adoptedmore broadly by all sport organizations. Certainly pre-game planning that would include representatives fromthe opposing teams would be helpful in communicatingin-game security enforcement issues and, fines and sanc-tions that might differ from venue to venue. Leagues couldbe proactive in this planning phase as well. Monitoringsystems (e.g., closed circuit cameras), require consider-able planning prior to game day to ensure effective cover-age of potential trouble locations. Such systems haveaided numerous professional teams in attempts to identifydisruptive spectators and more recently have becomemandatory components for all Greek and Italian profes-sional soccer stadiums. (Gatopolous 2007; Sanminiatelli2007) Employee training and incentive programs mightbe developed that emphasized effective management ofservice failure and customer rage incidents (Grove, Fisk,and John 2004); such activities obviously require leadtime. Development of signage and other communicationdevices that remind fans of appropriate behavior andpotential consequences of inappropriate behavior may aidmanagement in realizing results from pre-game initia-tives. Finally, reconfiguring seating options to allow morelogical fan segmentation (i.e., students, opposing fans)might be explored and implemented. Professional teamssuch as the Seattle Seahawks offer family sections inside

the stadium “free from alcohol sales, exposure to alcohol,and foul, suggestive or vulgar language” (www.seattleseahawks.com, accessed June 1, 2007).

Game Day Activities

As previously mentioned, the influx of fans thatconverge on the sport venue the day of a contest are likelyto be exposed to a variety of rage triggers that must bemanaged well by the sport organization to minimizedisruptive actions by a few fans. While not exhaustive, thefollowing categories of action represent areas most likelyto yield positive results in curbing violence during gameday.

Spectators attending a live sporting event expect toencounter security screening at facility entrances. Whileserving the purpose of ticket validation, most sport arenasuse entry gates as an opportunity to screen spectators foritems banned from the facility. Items commonly bannedinclude any object that could be used as a missile orprojectile, alcoholic beverages, weapons/fireworks, la-sers and umbrellas. The N.F.L. recently announced apolicy requiring teams to establish a screening processsubjecting spectators to the possibility of a random search.The policy received legal scrutiny from a fan claiming itviolated his constitutional protection against unreason-able searches (Nagel 2005; Varian 2005). A U.S. FederalAppeals court rejected the fans contention clearing theway for other sport organizations to follow the precedence(Washington Post 2007).

Most sporting events are also staffed by security andlaw enforcement; the presence of such personnel is in-tended to both deter acts of aggression and enforce sta-dium policies regarding breaches of acceptable behavior.Sport organizations have adopted a number of strategiesintended to reduce the probability of hostile behavior onthe part of spectators. Many of the strategies are reactiveand consist of various punishments for spectators whocommit acts of rage. A common approach is to removethose individuals from the stadium or venue. For ragebehavior that is also criminal in nature, spectators mayface charges from local or state authorities. More recently,sport organizations have adopted technologies to assis-tance in the identification of fans engaging in hostilebehavior. In the mid-1990’s the New England Patriots ofthe N.F.L. mounted cameras inside the stadium – anactivity that requires pre-game planning as noted above –to capture images of inappropriate behavior during thecontest. Spectators who were identified as engaging insuch incidents faced sanctions from the team. In 1995, thefirst year of the camera system, a total of 70 Patriot seasonticket holders were refused their request to renew thetickets for violations of the team’s policies on appropriatebehavior (Weisman 1995).


Recently, sport organizing bodies have subjectedteams to significant financial penalties for failing to en-force security policies. For example, in 2005 the interna-tional governing body for soccer (F.I.F.A.) fined threenational football clubs “for lack of proper security ar-rangements for that match” (CNN.com, September 26,2005). Similarly, the Southeastern Conference (S.E.C)fined the University of Arkansas $5,000 when spectatorsstormed the court to celebrate a victory by the men’sbasketball team (Associated Press, February 21, 2006).The Italian Soccer Federation, with the backing of theItalian government, passed several “zero tolerance” rulesintended to curb fan violence at soccer matches. A soccermatch will be abandoned if anything is thrown onto thefield and the team whose fans perpetuated the act will bepenalized with an automatic 3–0 loss. Moreover, local lawenforcement has the right to cancel a game should vio-lence erupt either inside or outside of the stadium beforethe start of a match or if offensive signs, banners, orsymbols are seen inside the venue (Associated Press,April 14, 2005).

Sport organizations use a variety of promotionalmedia to inform spectators of behaviors which are inap-propriate and subject to sanctions. The list of actionableoffenses is commonly found on the backside of gametickets and in game day programs. Stadium signage,scoreboard messages and audio announcements, particu-larly when used to fill wait time during the consumptionexperience, may be used to remind all spectators ofappropriate behavior and the potential sanctions for vio-lations.

Service providers have long discussed the impor-tance of managing wait times as a way of improving acustomer’s service experience. Haynes (1990) offers anumber of guidelines many of which are applicable in asport setting. For example, concession waits might seemshorter if sport venues offered fans closed circuit viewingof the action they are missing or facts about the team and/or opponent. Additionally, creatively reviewing ways tomake the wait seem fairer depending on the purchaseexpectation is possible. Purchasing only a bottle of waterat a concession stand should be much quicker than thepurchase of multiple food items and soft drinks. Perhapshaving separate lines or stationing individuals sellingsingle items from portable packs could be considered.Making the wait as comfortable as possible is also advis-able. During very hot outdoor events having a fan/airconditioner blowing on those inline might lessen the angstfelt by many.

Other proactive strategies attempt to minimize theinfluence of alcohol as a potential trigger for hostility. TheN.C.A.A. prohibits the sale of alcoholic beverages in on-

campus athletic facilities. Further, some universities haverecently placed restrictions on “pass-outs,” a term used todescribe the practice of allowing spectators to exit andreenter the stadium, the occasions for which are fre-quently used as an opportunity to consume more alcohol.The National Football League places strict restrictions onthe sale of alcohol. Most notably, the sale of beer insideN.F.L. stadiums is terminated at the start of the finalquarter of play. However, more recently, sport organiza-tions have taken an even more aggressive approach tocurbing the influence of alcohol as a potential trigger forrage behavior. Governmental officials in the state ofHawaii are considering a ban on alcoholic beverages inthe parking lots outside of Aloha Stadium – home of theUniversity of Hawaii’s football team (Associated Press,November 1, 2005). In November of 2005, prior to theN.F.L. contest between the New York Jets and NewEngland Patriots, the Jets (the host team) and the NewJersey Sports and Exposition Authority (the stadiumowners) agreed to ban the sale of alcohol inside thestadium in an attempt to reduce hostilities during theevent.

Post Game Follow-Up

Significant incidents of spectator rage should becatalogued and reviewed following every sporting event.A structure can be developed wherein official who are aparty to the rage are required to file a written reportdetailing their perspective. These can then be aggregatedand examined for trends and patterns. Management canalso review these “critical rage incidents” and determineif the customer(s) were principally at fault or if theorganization/service employee seemed instrumental intriggering the hostile action. In addition, if the right typeof information is collected, such an activity may offer thesport organization an opportunity to engage in servicerecovery for those parties who have been significantlyimpacted by the rage event. Finally, using the post-gameinformation, sport organizations must embrace an organi-zational learning perspective that welcomes process re-view and improvement.

Since the potential for rage events to morph into afull-blown crisis sometimes exists, sport organizationsshould develop and follow a set of protocols for crisismanagement. Even when not operating in a crisis inter-vention mode these groups should meet frequently toconfirm existing routines and procedures. The upshot issuch an undertaking would elevate the importance ofmanaging rage events and institutionalize the significanceof its control function for all employees. This committeemight also serve as a proactive force charged with regularupgrades aimed at creating a more reasonable competitiveenvironment at the sport venue.


STRATEGIC RECOMMENDATIONS AND

CONCLUSION

Ultimately, sport organizations must be willing torecognize that inappropriate hostile action by fans dimin-ish the game day experience they invest so heavily tocreate. There exists a misbehavioral line for fans that ifcrossed, is likely to erode the organization’s brand equity.Strategically, some sport enterprises will need to revisithow they energize their fans before and during games withmessaging, how they associate competitive play of theirteams with a winning at all cost mentality, and what thefans role should be in creating the shared experience thatreflects key principles of the sport brand. If organizationsare serious about reducing increasingly rowdy fan behav-ior, a repositioning of the brand is likely to be needed formany.

Efforts at repositioning might receive valuable assis-tance by league or governmental organizations. In theearly 1990’s the Canadian government initiated a market-ing campaign aimed at reducing the number of violentbehaviors in amateur sports. The “Settling the Score willnot Even the Score” campaign featured prominent profes-sional athletes urging both players and fans to exhibitgood sportsmanship (McLean’s 1993). The Jupiter-Tequesta Athletic Association in Florida has been re-ferred to as a model for educating parents and spectators.The organization annually packs more than 1,500 parentsinto a stadium to watch a video on how to be a good sportsparent, pick up a handbook describing sportsmanship, andsign a sportsmanship pledge (Cary, Dotinga, and Comarow2004). These and similar campaigns attempt to introducean alternative to the hostile fan role assumed by manyspectators.

When properly positioned, a sport brand can serve asmechanism to encourage civil behavior. To the extent thatthe sport enterprise emphasizes responsible competitionand a positive fan experience, its fans should exhibit thoseproperties as well. Moreover, aggression of hostility to-ward opposing fans or others as a means of restoring self-esteem when poor performances occur should diminish asan acceptable defense mechanism. Hence, building andreinforcing a brand image for the sport organization thatincludes and conveys positive attributes of sportsmanshipand is void of those elements that undermine it has thepotential to curb the likelihood of spectator rage.

Spectator sports present a myriad of unique chal-lenges for those concerned about the potential deleteriousconsequences of aggressive behavior. Such service sce-narios are typically characterized by numerous and acces-sible targets. Consumers, in the form of spectators, arefrequently influenced by a variety of psychological andsociological influences which may promote such behav-

ior. Finally, a wide variety of triggers unique to sport mayhelp to facilitate rage.

Leagues and governing bodies, teams, and facilitymanagers are all impacted by spectator rage and createtheir own strategies for addressing the concern. A chal-lenge in such an environment is communication amongthe various constituents to insure the implementation of acomprehensive plan. As an example, teams are typicallyresponsible for providing information regarding inappro-priate behavior on the tickets and brochures mailed toseason ticket holders while arena managers are mostlikely responsible for reminding spectators with arenasignage and audio messages. As noted by Ron VanDeVeen,V.P. and Associate General Manager for New Jersey’sMeadowlands Sports Complex, each event is unique andrequires the interaction of various parties in the decisionmaking process. “Seton Hall, which plays home games inour Continental Airlines Arena, restricts the sale of alco-hol. However, when we contracted with Duke Universityand the University of Texas to play a game in the samearena, the contract included a stipulation allowing for thesale of alcohol. Many of these and other issues regardingstadium security are key components of contract negotia-tions and the approach taken to address the issue ofspectator rage varies between sports, leagues, and teams”(VanDeVeen, personal communication, May 30, 2006).

In some instances, the policies of one stakeholdermay contradict the policies of other stakeholders. Forexample, the Tampa Sports Authority, the managing bodyfor Raymond James Stadium, has concessionaires removethe plastic caps from water bottles purchased inside thestadium. The policy exists to prohibit fans from using thecaps as projectiles. However, at a game in December of2006, fans entering the stadium were handed a metal keychain as part of a promotional giveaway prior to a TampaBay Buccaneer football game. The metal key chains maybe thought of as a much greater threat as a potentialprojectile than the plastic caps. This contradiction under-scores the importance of collaboration and communica-tion between stakeholders associated with sporting events.

Spectator rage has been persistent throughout thehistory of sport and is likely to remain a concern for theforeseeable future. Further, with greater and greater com-petition for the sport consumer dollar, marketers mustcontinue to employ tactics to enhance the game experi-ence; some of which may increase the likelihood ofincidents involving spectator rage. For example, teamscontinue to seek seating options providing fans with closeproximity to the game. When doing so, it is critical forteams to inform spectators of what is considered inappro-priate behavior and the consequences of engaging in thebehavior. Further, stakeholders must communicate andemploy a variety of tactics to enforce such polices includ-


ing the training and presence of security personnel, the useof camera systems and the implementation of penalties todeter such behavior.

Sport has long served as an invaluable social functionfor societies. It offers an unmatched forum for intensephysical competition, self expression, identification andconnection to one’s past. Today, we have the opportunity

to stem what some have noted as a general erosion ofcivility in society (Brown 2003; Kelley 2007). Sport iswell positioned to serve as a cultural change agent for civilexchange. Might sport properties teach us all what itmeans to engage one’s adversary in a manor that elevatesthe human spirit? Perhaps rugby leagues have it right;compete ferociously for 60 minutes, then lets have a beer.

REFERENCES

Arnould, Eric J. and Linda J. Price (1993), “River Magic:Extraordinary Experience and the Extended ServiceEncounter,” Journal of Consumer Research, 20 (1),24–46.

Associated Press (2005), “Italy Imposes New Anti-Vio-lence Measures for Soccer Games,” (April 14).

____________ (2005), “Aloha Stadium Moves Closer toAlcohol Ban,” (November 1).

____________ (2006), “SEC Fines Arkansas $5,000 forFans’ Jubilation,” (February 21).

Bandura, Albert (1973), Social Learning Theory.Englewood Cliffs, NJ: Prentice Hall.

____________ (1986), Social Foundations of Thoughtand Action. Englewood Cliffs, NJ: Prentice Hall.

Belk, Russell W. (1974), “An Exploratory Assessment ofSituational Effects in Buyer Behavior,” Journal ofMarketing Research, 11 (2), 156–63.

Bitner, Mary Jo (1990), “Evaluating Service Encounters:The Effects of Physical Surroundings and EmployeeResponses,” Journal of Marketing, 54 (2), 69–82.

____________, Bernard H. Booms, and Lois Mohr (1994),“Critical Service Encounters: The Employee’s View-point,” Journal of Marketing, 58 (4), 95–106.

Branscombe, N.R. and D.L. Wann (1992), “Role of Iden-tification with a Group, Arousal, Categorization Pro-cesses, and Self-Esteem in Sports Spectators Aggres-sion,” Human Relations, 45, 1013–33.

____________ and ____________ (1994), “CollectiveSelf-Esteem Consequences of Outgroup DerogationWhen a Valued Social Identity Is on Trial,” Euro-pean Journal of Social Psychology, 24, 641–57.

Bredemeier, Brenda Jo, David L. Shields, and Jack C.Horn (1985), “Values and Violence in Sports Today;the Moral Reasoning Athletes Use in Their Gamesand in Their Lives,” Psychology Today, 19, 22–31.

Brown, Gary T. (2003), “Leaders Seek Behavior Shiftafter Sportsmanship Summit,” The NCAA News,(March 3).

Cary, Peter, Randy Dotinga, and Randy Comarow (2004),“Rescuing Children’s Games from Crazed Coachesand Parents,” U.S. News and World Report, 136 (20),44–52.

CNN.com (2005), “FIFA Fine Swiss After Fans Protest,”(September 26).

Coakley, J.J. (1998), Sport in Society: Issues and Contro-versies, 6th ed. St. Louis, MO: Times Mirror/ Mosby.

____________ (2006), Sport and Society: Issues andControversies, 9th ed. New York: McGraw-Hill.

Fisk, Raymond P. and Patriya Tanshay (1985), ServicesMarketing: An Annotated Biography. Chicago, IL:American Marketing Association.

Frankel, B. and A. Tresniowski (2000), “Stormy Skies,”People Weekly, 5.

Friedman, Meyer (1996), Type A Behavior: Its Diagnosisand Treatment. New York: Plenum Press.

Gatopoulos, Derek (2007), “Security Cameras Promisedfor Greek Soccer Stadiums,” Associated PressWorldstream, (April 12).

Grove, Stephen J. and Raymond P. Fisk (1997), “TheImpact of Other Customers on Service Encounters: ACritical Incident Examination of ‘Getting Along’,”Journal of Retailing, 73 (1), 63–85.

____________, ____________, and Joby John (2004),“Surviving in the Age of Rage,” Marketing Manage-ment, 13 (2), 41–46.

Guerin, B. (1994), Analyzing Social Behavior: BehaviorAnalysis and the Social Sciences. Reno, NV: ContextPress.

____________ (1998), “Religious Behaviors as Strate-gies for Organizing Groups of People: A SocialContingency Analysis,” Behavior Analyst, 21 (1),53–72.

Harris, Lloyd C. and Kate L. Reynolds (2003), “TheConsequences of Dysfunctional Customer Behav-ior,” Journal of Service Research, 6 (2), 144–61.

Holmes, Davis A. and Michael J. Will (1985), “Expres-sion of Interpersonal Aggression by Angered andNonangered Persons with Type A and Type B Behav-ior Patters,” Journal of Personality and Social Psy-chology, 48 (3), 723–27.

Hunt, Kenneth H., Terry Bristol, and R. Edward Bashaw(1999), “A Conceptual Approach to Classifying SportsFans, Journal of Services Marketing, 13 (6), 439–52.

Kahle, L., M. Duncan, V. Dalakas, and D. Aiken (2001).“The Social Values of Fans for Men’s and Women’sUniversity Basketball,” Sport Marketing Quarterly,10 (2), 156–62.

Kelman, H.C. (1958), “Compliance, Identification andInternalization: Three Processes of Attitude Change,”Journal of Conflict Resolution, 2, 51–60.


____________ (1961), “Processes of Opinion Change,”Public Opinion Quarterly, 25, 57–78.

Langeard, Eric, John E.G. Bateson, Christopher Lovelock,and Pierre Eiglier (1981), Services Marketing: NewInsights from Consumers and Managers. Cambridge,MA: Marketing Science Institute.

Lovelock, Christopher H. (1994), Product Plus. NewYork: McGraw-Hill.

LeBon, Gustave (1960), The Crowd. New York: VikingPress.

Mahari, Ace-Jensen (2006), BPD and Rage. PhoenixRising Publications (www.phoenixrisingbooks.com).

McColl-Kennedy, J., C. Daus, B.A. Sparks (2003), “TheRole of Gender in Reaction to Service Failure andRecovery,” Journal of Service Research, 6 (1), 66–83.

McColl-Kennedy, Janet R. and Amy K. Smith (2006),“Customer Emotions in Service Failure and Recov-ery Encounters,” in Research on Emotions in Orga-nizations, Wilfred J. Zerbe, Neal M. Ashkanasky,and Charmine E.J. Hartel, eds. Amsterdam, TheNetherlands: JA1 (Elservier), 2, 243–75.

McLean’s (1993), “Play Fair or Else,” 106 (47), 13.Nagel, Kyle (2005), “NFL Pat-Down Policy at Issue in

Cincy,” Cox News Service, (October 19).Pratt, Michael G. and M. Teresa Cardador (2006), “Iden-

tification Management and Its Bases: Bridging Man-agement and Marketing Perspectives Through a Fo-cus on Affiliation Dimensions,” Journal of the Acad-emy of Marketing Science, 34 (2), 174–84

Ryan, Nate (2007), “Gordon Ventures into EarnhardtTurf,” USA Today, (April 27).

Sanminiatelli, Maria (2007), “Fans Could Be Barred fromItalian Soccer Stadiums If Security NotImproved,” Associated Press FinancialWire, (February 5).

Solomon, Michael R, Carol Surprenant, John A. Czepiel,and Evelyn G. Gutman (1985), “A Role TheoryPerspective on Dyadic Interactions: The ServiceEncounter,” Journal of Marketing, 49 (1), 1–13.

Stone, Michael H. (1993), Abnormalities of Personality:Within and Beyond the Realm of Treatment. NewYork: W.W. Norton.

Varian, Bill (2005), “Fan Challenges Legality of StadiumPatdowns,” St. Petersburg Times, (October 14).

Wann, D.L., R.R. Peterson, C. Cothran, and M. Dykes(1999), “Sport Fan Aggression and Anonymity: TheImportance of Team Identification,” Social Behaviorand Personality: An International Journal, 27, 597–602.

____________, Merrill J. Melnick, Gordon W. Russell,and Dale G. Pease (2001), Sport Fans: The Psychol-ogy and Social Impact of Spectators. Boca Raton, FL:Routledge Publishing.

____________, G. Hayes, B. McLean, and P. Pullen(2003), “Sport Team Identification and Willingnessto Consider Anonymous Acts of Hostile Aggres-sion,” Aggressive Behavior, 29, 406–13.

Washington Post (2007), “Court Allows Pat-Downs toResume at Bucs Games,” (June 28).

Weiberg, Steve (2002), “Rowdy Fans Concern CollegeOfficials,” USA Today, (November 15), 1.

Weisman, L. (1995), “Patriot Fans Must Play by theRules, Too,” USA Today, (August 30), C6.

For further information contact:Scott A. Jones

250 Sirrine HallClemson UniversityClemson, SC 29634


Stephen J. Grove243 Sirrine Hall

Clemson UniversityClemson, SC 29634


Gregory M. Pickett245 Sirrine Hall

Clemson UniversityClemson, SC 29634



AN ARGUMENT FOR CO-OPETITION’S PRO-CONSUMER POSTURE

Michael A. Levin, Texas Tech University, LubbockSabrina Sattler, Texas Tech University, Lubbock

ABSTRACT

Traditional American antitrust enforcement focuseson rents. A recent federal court case shifted that focusfrom rents to deadweight loss. The meaning of dead-weight loss has changed over time. The implication of thiscase affects co-opetition, a form of competition. Thispaper argues for a view of co-opetition as pro-consumer.

INTRODUCTION

The marketing literature champions the concept ofcompetition where firms form alliances to share resources,create a sustainable competitive advantage, and to achievesuperior financial rewards (Hunt 2000; Lambe, Speckman,and Hunt 2002; Williams 2005). Different forms of alli-ances include buyer-seller dyads, and wholesaler-retailer-consumer chains. These different forms of alliances fitwithin the overall context of relationship marketing. Rela-tionship marketing specifies that all firms maintain fourtypes of relations: supplier, lateral, buyer, and internal(Morgan and Hunt 1994). This paper focuses on anexample of a lateral relationship, or alliance, created byindependent firms working together to set a technologystandard. In these alliances, firms create and/or maintainresources to produce a market offering that might nototherwise exist (Levin and McDonald 2006).Brandenburger and Nalebuff (1996) as well as Sheth andSisodia (1997) refer to this type of alliance as “co-opetition.” A co-opetitive is defined here as an alliance ofindependent, competing firms that share resources tocreate a good or a service more effectively and/or effi-ciently.

Public policy makers typically view cartels in generaland co-opetitives in particular as harmful to marketsbecause of certain strategies like erecting barriers tomarket entry, fixing prices, artificially dividing markets,or a combination of these strategies (Lande 1993;Schleicher 1997). Levin and McDonald (2006) show thatthese strategies are inconsistent with the intent of theSherman Antitrust Act, and with resource-based theoriesof competition (Barney 1991; Hunt 2000). Recently,public policy makers adopted the view that consumerchoice, even in absence of these strategies, must beprotected (Lande 2001). Lande (2001, p. 514) writes “Theultimate point to note here is that the Microsoft case isbeing argued in terms of consumer choice and in terms ofprice. The case thus dramatically illustrates how con-sumer choice is emerging as an explicit paradigm for

antitrust.” This view is problematic because equilibrium-seeking economics cannot demonstrate directly how co-opetitives harm consumers. Resource-advantage theory(hereafter R-A Theory) can explain how co-opetitives donot inherently harm consumers. Therefore, the purpose ofthis paper is to address the issue of the co-opetitive’s effecton consumers (Levin and McDonald 2006) and to extendmarketing’s role in public policy (Gundlach 2001;Gundlach and Phillips 2002; Gundlach, Phillips,Desrochers 2002; Sheffman 2002; Sullivan 2002).

The marketing literature has examined different is-sues related to a firm’s strategies and tactics as they relateto antitrust, competition, and public policy. For example,Heil and Langvardt (1994) examine the use of signaling,Guiltinan (2002) argues for the inclusion of consumerbehavior, Gundlach (2001) discusses the nature of ex-change within marketing strategy, and Pertiz (2002) andSullivan (2002) review business strategy within antitrustpolicies. In the antitrust debate, marketing strategy so farremains an afterthought (Foer 2002; Hawker 2002). Over-all, antitrust analysis is underserved because it “ignoreshow customers respond to changes” in products offered inthe marketplace (Levin and McDonald 2006, p. 2). This“third leg of the antitrust stool” (Foer 2002, p. 227)supports Hunt’s (2000) argument that competition is pro-consumer because firms produce more effective and/orefficiently in order to achieve superior financial rewardsthrough the process of competition.

Public policy makers and researchers alike generallyaccept that consumers benefit from firms competing in themarketplace. A firm can respond to what it learns fromcustomers and competitors by offering new or improvedproducts, or by adopting a new or improved process.Through the process of competition, consumers benefitfrom the increased value generated by firms attempting toimprove their competitive position in the marketplace(Hunt 2000).

If consumers benefit from this aspect of competition,then consumers may also benefit from co-opetition, or analliance comprised of competing firms that cooperate toshare resources in order to produce a product that mightotherwise not exist (Levin and McDonald 2006). Con-sumers may benefit through the efficiencies gained whenfirms within an industry cooperate to establish an industrystandard. For example, consumers benefitted when firmsformed a co-opetitive that resulted in a standard forDigital Versatile Disc (hereafter DVD) players (i.e., hard-


ware) and titles (i.e., software). Without this co-opetitive,manufacturers, suppliers, retailers, and consumers wouldhave faced incompatible standards similar to the Betamaxand VHS dichotomy.

Traditionally, public policy makers viewed co-opetitives as inherently harmful to the marketplace be-cause they were thought of as resulting in an inefficientallocation of resources in pursuit of rents, and, ultimately,prevented economies from reaching full potential. Overtime, public policy makers have shifted their stance on co-opetitives and their potential effects. This paper recog-nizes that co-opetitives represent a type of alliance thatleads to benefits for consumers. Recently, the investiga-tion of the harm caused by co-opetitives has shifted fromthe accrual of rents (rectangle ABCD in Figure A-1) to thecreation of deadweight loss (triangle CDE in Figure A-1).“(Consumer choice) would represent a change of empha-sis from traditional emphasis” (Lande 2001, p. 516). Thischange of emphasis represents a specious argument andfurther signifies that antitrust policy has less to do with atheory of competition and more to do with public policymakers’ whims.

To recognize how co-opetitives adopt a pro-con-sumer posture, a theory is needed that explains thedisequilibriating nature of competition while acceptingthat firms form alliances without harming society. Weargue for the adoption of the R-A theory by marketingresearchers, practitioners, and public policy makers be-cause it views co-opetitives as pro-consumer, treats com-petition as a dynamic process, and can incorporate anti-trust immunities created by the Supreme Court. Addition-ally, R-A theory can explain why firms form co-opetitivesand why such alliances are not “naked restraints” of trade.Rather, such alliances may result in per se antitrust viola-tions in the form of “ancillary restraints” (Cartensen andRoth 2000, p. 355), or alternatively may meet the U.S.Supreme Court’s Rule of Reason defense.

REVIEW OF LITERATURE

In 1890, the United States government formalized itspolicy toward competition with the passage of the ShermanAntitrust Act. The Sherman Act consists of two sections.The first section (hereafter Sherman I) states that “everycontract, combination in the form of trust or otherwise, orconspiracy in restraint of trade or commerce among theseveral states or foreign nations is declared to be illegal”(U.S. Code 2005). Since every contract is potentiallycollusionary in nature, the Supreme Court establishes theRule of Reason where the defendant argues for the reason-ableness of the alleged collusionary behavior due toimproved efficiencies and/or creation of consumer ben-efits (Utton 2003). However, the Court does not normallyaccept this defense in cases involving Sherman I viola-tions (Cartensen and Roth 2000).

The second section (hereafter Sherman II) states that“every person who shall monopolize, or attempt to mo-nopolize or combine or conspire with any other person orpersons, to monopolize any part of the trade or commerceamong the several states, or with foreign nations, shall bedeemed guilty of a felony” (U.S. Code 2005). InSherman II, the firm’s behavior in the market remains keyto constituting legality. Sherman II does not define wheth-er the goal is to protect markets or consumers. Instead,public policy makers, including regulatory agencies suchas the Department of Justice, jurists such as the SupremeCourt, and elected officials such as members of Congress,shoulder the responsibility of deciding what needs protec-tions and what constitutes collusion.

Neither Dickson and Wells (2001) nor Hunt andArnett (2001) mention a desire by Congress to protectconsumers in their respective reviews of the ShermanAct’s passage. Rather, as Levitt (1952) and Thorelli(1955) show, the Sherman Act’s direct intent is preservingeconomic freedom, lowering prices, and protecting themarket from perceived increasing power exercised bylarge firms. Consumers’ benefit becomes an indirectresult in the interpretation of the Sherman Act. Because ofSherman’s vagueness regarding protection and collusion,public policy makers (i.e., regulators, jurists, electedofficials) have a wide berth in Sherman’s interpretation(Dickson and Wells 2001; Hunt and Arnett 2001). Theinterpretation of Sherman varies over time (Hunt andArnett 2001; Pertiz 2002; Schlericher 1997). Indeed, fiveperiods of interpretations are delineated from Sherman’senactment to Eastman Kodak Co. v. Image TechnicalServices (1992; hereafter Kodak).

These interpretations rely on theories and argumentscentering on competition, and resource allocation. In thefollowing sections, two theories, (1) cartel and (2) re-source advantage, are discussed and compared.

Cartel Theory

In a cartel, a collection of firms attempts to control themarket (Armentanto 1982; Utton 2003). The process ofseveral firms working together to erect barriers to entry,attempt to maximize profitability by setting prices, limitproduction, or otherwise artificially control the market isknown as collusion. By restricting the quantity of aproduct through these attempts, the cartel seeks to accruecartel rents. In Figure 1, rectangle ABCD represents cartelrents. These rents represent profits from prices that exceedthose allowed under perfect competition, which is repre-sented by line AE. To achieve these cartel rents, thecolluding firms reduce quantity available (Q

competition Q

cartel).

In turn, rents increase (Pcompetition

Pcartel

). Thus, firms havea financial incentive to engage in collusionary behaviors(Varian 2003).


Figure 1 here

FIGURE A–1

Cartel Rents and Deadweight Loss

Source: Varian (2003). Note: Figure A–1 appears in many sources including Armentano (1982), Hunt andArnett (2001), and Viscui et al. (2005).


In order to raise prices, cartels create a deadweightloss (Armentano 1982). In Figure A–1, triangle CDEmarks the deadweight loss. The deadweight loss occursbecause consumers must pay higher prices for the prod-uct, and manufacturers must produce lower levels of theproduct. The impact of the deadweight loss can be illus-trated through an example drawn from the personal com-puter (hereafter PC) industry. If the manufacturers ofmemory chips form a cartel, the harm extends eventuallyto consumers. Because memory chip manufacturers pro-duce below capacity, PC assemblers such as Dell andApple must produce below capacity. This leads to reducedinventory levels and higher retailer prices for consumers.

Since all firms must produce below capacity, all firmsrequire less labor. As a result, the economy will be unableto maximize employment. Because the economy is unable

to maximize its employment resource, many consumerslack the necessary means to purchase products. There-fore, a cartel results in a deadweight loss that affects manyfacets of society (Viscui, Harrington, and Vernon 2005).

The meaning of deadweight loss shifts over time.Historically, the deadweight loss represents the misallo-cation of resources caused by the cartel producing belowefficiency. Cartels embody inefficient production be-cause the members must produce below efficiency inorder to accrue rents (Armentano 1982; Varian 2003).

Recently, the meaning of deadweight loss became aloss of technological innovation. Members of the cartelshave little incentive to innovate because innovations maycause a cartel to collapse (Viscui et al. 2005). Sullivan andGrimes (2006, p. 725) note, “Credible (economic) theory

TABLE A–1

Summary History of Antitrust Enforcement

Interpretation

Periods

Classical Economics(1890 – 1920)

Market Economics(1920 – 1950s)

Pre-Chicago(1950s – 1977)

Chicago(1977 – 1992)

Post-Chicago(1992 – Current)

Exemplary

Cases

United States v. Trans–MissouriFreight Association (1897)

United States v. United States SteelCorp. (1920); United States v. Socony-Vaccum Oil Co. (1940)

Northern Pacific Railway Co. v.United States (1958); Brown ShoeCo. v. United States (1962)

Continental T.V. Inc. v. GTE SylvaniaInc. (1977)

Eastman Kodak Co. v. Image Techni-cal Services (1992); United States v.Microsoft (1999)

Summary

Public policy focuses on freedom to contract, andrestraint of trade. “The more freedom to contract,the more competition there is” (Peritz 2002, p. 237).

Public policy makers shift from “classical eco-nomics” to “market economics” (Pertiz 2002,p. 37). Justice William O. Douglas, cited in Pertiz(2002, p. 238), avows, “competitive pricing theeconomy’s ‘central nervous system’ and pricefixing cartels the worst affliction.”

Public policy makers adopt the Structural-Con-duct-Performance model (Bain 1954, 1968) in anattempt to prevent a concentration of resources(Grengs 2006).

Public policy blends tenets from the price theoryperiod with “increased sophisticated economicanalysis” (Schlecher 1997, p. 310) to create astatic, neoclassical economic model of competi-tion. Thus, “The exclusive goal of antitrust law iseconomic efficiency” (Spivack and Ellis 1993,p. 214).

Public policy accepts imperfections in the marketand rejects a pure static view of competition.Alternate theories “have had little impact in thecourts, due to the difficulty of developing opera-tional rules” (Grengs 2006, p. 127).


has never suggested that concentration (i.e., many firms inthe marketplace) enhances innovation.” The loss of tech-nological innovation represents another form of misallo-cation of resources because technological innovation re-quires the efficient allocation of resources (Hunt 2000).Deadweight loss also encapsulates a lack of consumerchoice (Lande 2001) and a lock in strategy based onswitch costs (United States v. Microsoft Corp. 1999;hereafter Microsoft).

Until Microsoft, public policy makers did not rely ondeadweight loss alone. Sherman II fails to explicitlymention deadweight loss, or harm to consumers caused byfirms’ collusionary behavior. In Microsoft, both sidesagreed that Microsoft did not accrue rents (Cavanagh2003); rather, Microsoft created a deadweight loss suffi-ciently large enough to warrant public policy makers’intervention (Lande 2001).

This intervention is consistent with Harberger’s (1954)and Lande’s (2001) assertion that the aim of public policymust be to protect consumers from collusionary behavior,or minimize and perhaps eliminate deadweight loss. Thisaim is difficult to achieve because of measurement prob-lems.

Sherer and Ross (1990, p. 631–633) estimate thatdeadweight loss represents between 0.5 and 2 percent ofgross national product. To measure the deadweight loss,the researcher must calculate the consumer surplus underperfect and imperfect conditions. The residual, or differ-ence between these two results, represents the deadweightloss. Equilibrium-seeking economics cannot directlymeasure deadweight loss because it relies on a system ofequations to derive a solution. Only resources that aremeasurable such as land, labor, and capital are included inthese systems of equations. Because the residual capturesonly tangible deadweight loss, a deadweight loss includ-ing intangibles becomes open to interpretation. Theseinterpretations of deadweight loss include inefficient allo-cation of resources (Varian 2003), lack of technologicalinnovation (Viscui et al. 2005), reduced consumer choice(Lande 2001), and lock in costs (Microsoft).

Microsoft presents four pressing issues with the U.S.antitrust enforcement, collusionary behavior, and CartelTheory: (1) discussion of deadweight loss appears spe-cious, (2) use of deadweight loss instead of rents toinvestigate if collusionary behavior actually harms con-sumers, (3) equilibrium-seeking economic thought nolonger serves as the dominant language of competition,and (4) marketing practitioners and researchers, as well aspublic policy makers need a theory that accepts co-opetitives as a, possibly even necessary, relationship inthe competitive process. R-A theory fulfils the needs ofmarketing practitioners and researchers, as well as public

policy makers while addressing the issues related to co-opetives and deadweight loss raised in Microsoft.

R-A Theory

Hunt (2001, p. 527) defines R-A theory “as an evolu-tionary, disequilibrium-provoking process theory of com-petition, in which innovation and organizational learningare endogenous, firms and customers have imperfectinformation, and in which institutions and public policyaffect economic performance.” In R-A theory, firms usetheir resources to maintain a position of competitiveadvantage, and, ultimately, achieve superior financialrewards. To maintain this position of competitive advan-tage related to competing firms in the marketplace, a firmallocates its available resources effectively and/or effi-ciently.

Resources, as conceptualized in R-A theory, extendthe traditional view of resources (i.e., land, labor, capital,and management) to include financial, legal, human,information, organizational, and relational (Hunt 2001,p. 529–530). A firm can combine these seven types ofresources into higher order resources. Market orientationexemplifies a higher order resource because it combineselements of human, information, organizational, and rela-tional resources (Hunt and Morgan 1995). A firm relies onits higher order resources to produce a market offeringthat satisfies consumers’ heterogeneous demand.

All firms may possess these seven lower order re-sources. However, firms possess different degrees ofthese resources (Barney 1991). If the firm wants to im-prove its degree of a resource, the firm can improve itinternally, acquire it, or enter into an alliance. Lambe et al.(2002, p. 141) define an alliance as “collaborative effortsbetween two or more firms in which the firms pool theirresources in an effort to achieve mutually compatiblegoals that they not achieve easily alone.” Sheth andSisodia (1999) argue for marketing’s continued researchinto co-opetitives, or when competing firms form analliance to achieve a mutually desirable outcome. Shethand Sisodia (1999, p. 81) note that such alliances “involvemutual dependence and a degree of shared decision mak-ing between separate firms.” Despite the interest in alli-ances comprised of competitors, rarely does the allianceliterature discuss antitrust implications (Fontenot andHyman 2002).

R-A Theory and Antitrust

The Department of Justice’s Antitrust DivisionManual (1998) lists 10 industries that hold immunitiesfrom antitrust enforcement. Many of these immunitiesallow for the creation of alliances that share informationrelated to the marketing mix. For example, in the agricul-


tural industry, growers who belong to the co-operativesexchange resources as part of the development of pricingand promotional strategies. The cooperative of cranberrygrowers that owns Ocean Spray shares information as partof their competitive effort in the market place. By allow-ing competing firms to form alliances to exchange re-sources, the U.S. government takes a pro-competitivepublic policy stance. Yet, immunities such as the one thatallows for the formation of an agriculture cooperativeconflict with equilibrium-seeking economic thought.

Equilibrium-seeking economic thought treats theseimmunities as aberrations because, at their core, the firm(i.e., Major League Baseball) or firms (i.e., technologystandard setting co-opetitives) benefitting from the im-munities are viewed as engaging in collusionary behavior.These immunities are granted by law and, therefore, canalso be removed by law. For example, the National Coop-erative Research and Production Act encourages the for-mation of research-focused co-opetitives (U.S. Code 2005).However, Sullivan and Grimes (2006) write that firmsmay not freely form such co-opetitives; albeit, these co-opetitives must meet review guidelines (i.e., market defi-nition, market power) that are informed by equilibrium-seeking economic thought.

R-A theory, however, allows for precisely thesetypes of co-opetitive arrangements because they enhancecompetition in the marketplace and represent a pro-con-sumer posture. That is, competing firms share informationregarding elements of the marketing mix as part of thecompetitive process. In R-A theory, these exemptions toantitrust enforcement would not be immunities, ratherrecognitions of allowable behavior, because the courtcould recognize that competing firms enter into alliancesto achieve superior financial rewards without malice offorethought to limit competition or raise prices.

R-A theory acknowledges that co-opetitives do notinherently lead to cartel rents. For example, standardsetting alliances, which have been held as an example ofper se violations of antitrust statutes, coordinate withfirms in the marketplace to set a standard for an emergingtechnology. In light of Microsoft, public policy makersmay take a less than favorable attitude toward such alli-ances because they may harm the spurious consumerchoice.

Without these co-opetitives, many firms could hypo-thetically produce and sell market offerings using dispar-ate and incompatible technology. A manufacturer thenmight have difficulty gaining market power because re-tailers can select from many different technologies. Like-wise, consumers can choose to purchase a product frommany different retailers because no retailer will likelystock all available technological formats. Thus, consumersurplus would be maintained. However, without the criti-

cal mass created by a uniform standard, manufacturers ofcomplementary products (i.e., software) hesitate to investin the development of new products. Therefore, consumeshave fewer choices and face a higher price. Prices are highbecause manufacturers cannot achieve economies of scales.Indeed, many consumers may not purchase anything outof fear of choosing the “wrong” technology (i.e., one thatmight be later outdated or not supported by complemen-tary products), which leads to a loss of sales and profit.

The Court ruled that Microsoft benefitted from adeadweight loss by locking in consumers because con-sumers would incur switch costs if they purchased a PCthat deviated from Microsoft’s operating system soft-ware. In the PC industry during the 1980s, several incom-patible operating system locked consumers into one tech-nology standard (e.g., Apple, Atari, Commodore). WhenAtari, and Commodore withdrew from the market, theircustomers were stuck with hardware and software that nolonger received updates (i.e., no technological innova-tion). Further, customers could not use the software onoperating systems from Apple or Microsoft nor couldthese operating systems work on their computers. Thegovernment did not sue Apple, Atari, or Commodore forfollowing the same strategic behavior as Microsoft.

Manufacturers of DVD players avoided the operat-ing system scenario by forming an alliance to coordinatean industry standard. Time Warner, Toshiba, Philips, andIBM among others formed a co-opetitive, which agreedupon technological standards that created a single formatfor the DVD and for the DVD players (Frankel 2007).Hardware suppliers (i.e., Toshiba), intellectual propertyrights owners (i.e., Philips), software suppliers (i.e., TimeWarner), and assemblers (i.e., IBM) exchanged resourcesto create a single product. Many firms supply DVDplayers and several firms release DVDs titles. Retailersselect from various DVD suppliers and assemblers, andconsumers purchase from the retailer they chose. Becauseof the DVD co-opetitive, the DVD quickly gained adop-tion among consumers, providing consumers with in-creased entertainment value compared to the existingvideocassette technology. Unlike in the operating systemsoftware industry, the government has not sued a memberof the DVD co-opetitive for anticompetitive behavior.

In the DVD industry, firms had an incentive to forma co-opetitive in order to use their resources more effi-ciently and effectively. Deadweight loss did not occurdespite the appearance of collusion (Shughart and Tollison1998). The co-opetitive did not lead to inefficient alloca-tion of resources, discourage technological innovation, orlimit consumer choice. Rather, in all likelihood, the co-opetitive led to a more efficient allocation of resources,encouraged technological innovation, and increased con-sumer choice. Instead of many firms allocating resourcesto create duplicate products, members of the co-opetitive


enhance efficiency by sharing resources within the alli-ance. The resulting technology standard encourages firmsto produce a market offering that builds on the industrystandard and improves technological innovation. Toachieve superior financial rewards, it is in the best interestof the co-opetitive’s members to produce an innovativeproduct that offers superior value to consumers.

Finally, a technological standard enhances adoptionas seen through the diffusion of innovation curve. Thediffusion of innovation of the videocassette recorder hasa flatter slope because two incompatible technologicalformats competed in the marketplace. In contrast, thediffusion of innovation of the DVD player has a steeperslope because only one technological format was offeredin the marketplace, facilitating consumers’ adoption ofthis technology. This is due in part to the network exter-nalities of complimentary product availability, movies onDVD. Without a sufficient number of movies to watch onDVD, consumers would not purchase the technology.Without its co-opetitive, it is doubtful that the DVD wouldbe as ubiquitous as it is (Brull 1999). Consumers pur-chased DVD players over a shorter period compared tovideocassette players because regardless of which DVDplayer was purchased, the consumer could watch anyDVD. Assemblers were free to add features such as scan,menu, and other items to differentiate their DVD playerfrom other DVD players but the underlying technologyremains the same.

The DVD co-opetitive illustrates a co-opetitive’spro-consumer posture. Society benefits from a co-opetitivewhen it results in (1) a greater sensitivity to differences inconsumers’ needs, wants, tastes, and preferences, (2)higher quality goods and services, (3) greaterinnovativeness, (4) higher productivity, and (5) greatereconomic growth (Hunt and Arnett 2006; Ellig 2001).Unlike Apple, Atari, Commodore, and Microsoft, theDVD co-opetitive ensured that customers did not facelock-in to a particular manufacturer, nor the risk of obso-lescence due to withdrawn or incompatible hardware andsoftware.

DISCUSSION

The pro-consumer posture of co-opetitives carriesseveral implications for public policy makers, marketingpractitioners, and researchers. Historically, public policy

makers viewed co-opetitives as inherently harmful tosociety because co-opetitives create a deadweight loss.Although deadweight loss has been discussed in theeconomics literature, the interpretation of a deadweightloss has shifted over time. Deadweight loss has beenthought of as a byproduct of collusionary behavior, amisallocation of resources, a loss of technological innova-tion, and as harmful to consumer choice. The reinterpre-tation of deadweight loss illustrates the idea that antitrustenforcement has less to do with theory and more to do withpublic policy makers’ whim.

Public policy makers should view co-opetitives asinherently beneficial to consumers instead of inherentlyharmful. Investigation of co-opetitives should follow aprocedure of naked and ancillary restraints. Public policymakers would prohibit a naked restraint such as a co-opetitive that agrees to set a price for all market offerings.Alternatively, public policy makers would allow an ancil-lary restraint such as a co-opetitive that sets a technologi-cal standard. To allow ancillary restraint such as a co-opetitive, public policy makers need to adopt R-A theorybecause it accepts a co-opetitive’s pro-consumer posture.Unlike equilibrium-seeking economic thought, whichviews a co-opetitive as inherently harmful to consumers,R-A theory can explain how consumers benefit from co-opetition (i.e., the DVD technology standard format).

Co-opetitives benefit consumers as the DVD co-opetitive exemplifies. Firms do not misallocate resources;rather, they share resources to produce a market offeringthat meets the heterogeneous demands from consumersegments. By achieving higher efficiency and greatereffectiveness through co-opetition, firms can achievepositions of competitive advantage in the marketplace,and, ultimately, superior financial rewards. Marketerswould benefit from the resulting steeper diffusion oftechnology.

Finally, researchers should continue to investigateco-opetitives. Many issues remain unsolved, such as whysome firms form and participate in co-opetitives, whysome co-opetitives are more successful than others, andwhy public policy makers must rethink the ten immunitiesif they adopt R-A theory as the prevailing view on co-opetitives. These answers better reflect how businessescompete in the marketplace. Thus, public policy becomesnormative rather than positive in nature.

REFERENCES

Armentano, Dominick T. (1982), Antitrust and Monopoly:Anatomy of a Policy Failure. New York: John Wiley& Sons.

Bain, Joe S. (1954), “Conditions of Entry and the Emer-

gence of Monopoly,” in the Monopoly and Competi-tion and Their Regulation, Edward Chamberlin, ed.London: Macmillan.

____________ (1968), Industrial Organization, 2nd ed.New York: John Wiley.

Barney, Jay (1991), “Firm Resources and Sustained Com-


petitive Advantage,” Journal of Management, 17 (1),99–120.

Brandenburger, Adam M. and Barry J. Nalebuff (1996),Coopetition. New York: Doubleday.

Brown Shoe Co. v. United States (1962), 370 U.S. 317.Brull, Steven V. (July 5, 1999), “DVD and Conquer: Why

One Technology Prevailed,” BusinessWeek, accessedon April 24, 2007, [http://www.businessweek.com/index.html].

Carstensen, Peter C. and Bette Roth (2000), “The Per SeLegality of Some Naked Restraints: A[Re]conceptualization of the Antitrust Analysis ofCartelistic Organizations,” Antitrust Bulletin, 45 (2),349–435.

Cavanagh, Edward D. (2003), “De-Regulation of the AirWaves: Is Antitrust Enough?” St. John’s Journal ofLegal Commentary, 17 (1), 67–80.

Continental T.V., Inc. v. GTE Sylvania Inc. (1977), 433U.S. 36.

Dickson, Peter R. and Philippa K. Wells (2001), “TheDubious Origins of the Sherman Antitrust Act: TheMouse That Roared,” Journal of Public Policy &Marketing, 20 (1), 3–14.

Department of Justice (1998), Antitrust Division Manual,3d ed. Washington D.C.: U.S. Government PrintingOffice.

Ellig, Jerry (2001), Dynamic Competition and PublicPolicy: Technology, Innovation, and Antitrust Is-sues. Cambridge, U.K.: Cambridge University.

Eastman Kodak Co. v. Image Technical Services (1992),504 U.S. 451.

Foer, Albert A. (2002), “The Third Leg of the AntitrustStool,” The Journal of Public Policy & Marketing, 21(2), 227–31.

Fontenot, Renee J. and Michael R. Hyman (2002), “TheAntitrust Implications of Relationship Marketing,”Journal of Business Research, 57, 1211–21.

Frankel, Daniel (2007), “DVD Timeline,” Variety, ac-cessed on April 23, 2007, [http://www.variety.com/article/VR1117963613.html?categoryid=20&cs=1].

Grengs, Christopher M. (2006), “Verizon v. Trinko: FromPost-Chicago Antitrust to Resource-AdvantageCompetition,”Journal of Law, Economics and Policy,2 (1), 105–44.

Guiltinan, Joseph P. (2002), “Choice and Variety inAntitrust Law: A Marketing Perspective,” Journal ofPublic Policy & Marketing, 21 (2), 260–67.

Gundlach, Gregory T. (2001), “Choice as the Focus ofAntitrust: A Marketing Perspective,” University ofPittsburgh Law Review, 62, 527–33.

____________ and Joan M. Phillips (2002), “Marketingin Antitrust: Contributions and Challenges,” Journalof Public Policy & Marketing, 21 (2), 250–53.

____________, ____________, and Debra M. Desrochers(2002), “Antitrust and Marketing: A Primer and Callto Research,” Journal of Public Policy & Marketing,21(2), 232–36.

Harberger, Arnold C. (1954), “Monopoly and ResourceAllocation,” American Economic Review, 44 (2), 77–87.

Hawker, Norman W. (2002), “The Public Policy of Anti-trust and Strategy: An Overview,” Journal of PublicPolicy & Marketing, 21 (2), 257–59.

Heil, Oliver P. and Arlen W. Langvardt (1994), “TheInterface Between Competitive Marketing Signalingand Antitrust Law,” Journal of Marketing, 58 (July),81–96.

Hunt, Shelby D. and Robert M. Morgan (1995), “TheComparative Advantage Theory of Competition,”Journal of Marketing, 59 (April), 1–15.

____________ (2000), A General Theory of Competition.Thousands Oaks, CA: Sage.

____________ (2001), “A General Theory of Competi-tion: Issues, Answers, and an Invitation,” EuropeanJournal of Marketing, 35 (5/6), 524–48.

____________ and Dennis B. Arnett (2001), “Competi-tion as an Evolutionary Process and Antitrust Policy,”Journal of Public Policy & Marketing, 20 (1), 15–26.

____________ and ____________ (2006), “Does Mar-keting Success Lead to Market Success?” Journal ofBusiness Research, 59, 820–28.

Lambe, C. Jay, Robert E. Spekman, and Shelby D. Hunt(2002), “Alliance Competence, Resources, and Alli-ance Success: Conceptualization, Measurement, andInitial Test,” Journal of the Academy of MarketingScience, 30 (2), 141–59.

Lande, Robert H. (2001), “Consumer Choice as the Ulti-mate Goal of Antitrust,” University of PittsburghLaw Review, 62 (2), 503–25.

____________ (1993), “Chicago Takes It on the Chin:Imperfect Information Could Play a Crucial Role inthe Post-Kodak World,” Antitrust Law Journal, 62(1), 193–201.

Levin, Michael A. and Robert E. McDonald (2006), “R-A Theory as a Post-Chicago Argument for LegalCoopetition,” Marketing Management Journal, 16(2), 1–12.

Levitt, Theodore (1952), “The Dilemma of AntitrustAims: Comment,” The American Economic Review,42, 893–95.

Morgan, Robert M. and Shelby D. Hunt (1994), “TheCommitment-Trust Theory of Relationship Market-ing,” Journal of Marketing, 58 (July), 20–38.

Northern Pacific Railway Co. v. United States (1958), 356U.S. 1.

Peritz, Rudolph J.R. (2002), “Antitrust Policy and Ag-gressive Business Strategy: A Historical Perspectiveon Understanding Commercial Purposes and Ef-fects,” Journal of Public Policy & Marketing, 21 (2),237–42.

Scherer, F.M. and David Ross (1990), Industrial MarketStructure and Economic Performance, 3d ed. Boston:Houghton Mifflin.

Schleicher, Tara J. (1997), “The U.S. Supreme Court’s


Use of Post-Chicago Antitrust Theory in EastmanKodak v. Image Technical Services: Implications forMarketing Practice,” Journal of Public Policy &Marketing, 16 (2), 310–18.

Sheffman, David T. (2002), “Antitrust Economics andMarketing,” Journal of Public Policy & Marketing,21 (2), 243–46.

Sheth, Jagdish N. and Rajendra S. Sisodia (1999), “Revis-iting Marketing’s Lawlike Generalizations,” Journalof the Academy of Marketing Science, 27 (1), 71–87.

Shughart II, William F. and Robert D. Tollison (1998),“Collusion, Profits, and Rational Antitrust,” Anti-trust Bulletin, 43 (2), 365–74.

Spivack, Gordon B. and Carolyn T. Ellis (1993), “Kodak:Enlightened Antitrust Analysis and Traditional Ty-ing Law,” Antitrust Law Journal, 62 (Summer), 203–16.

Sullivan, Mary W. (2002), “The Role of Marketing inAntitrust,” Journal of Public Policy & Marketing, 21(2), 247–49.

Thorelli, Hans B. (1955), The Federal Antitrust Policy.Baltimore, MD: Johns Hopkins University Press.

United States Code. Title 15. Commerce and Trade, 15USCS, Chapter 1. Monopolies and Combinations inRestraint of Trade, 2005 [available at http://uscode.house.gov/download/title_15.php].

United States v. Microsoft Corp. (1999), 84 F. Supp. 2d 9.United States v. Socony-Vacuum Oil Co., Inc. et al.

(1940), 310 U.S. 150.United States v. Trans-Missouri Freight Association

(1897), 166 U.S. 290.United States v. United States Steel Corp. (1920), 251

U.S. 417.Utton, Michael A. (2003), Market Dominance and Anti-

trust Policy, 2nd ed. Northampton, MA: Edward Elgar.Varian, Hal R. (2003), Intermediate Microeconomics: A

Modern Approach, 6th ed. New York: W.W. Norton.Viscui, W. Kip, Joseph E. Harrington, Jr., and John M.

Vernon (2005), Economics of Regulation and Anti-trust, 4th ed. Cambridge, MA: MIT Press.

Williams, Trevor (2005), “Cooperation by Design: Struc-ture and Cooperation in Interogranizational Net-works,” Journal of Business Research, 58, 223–31.

For further information contact:Michael A. Levin

Texas Tech UniversityP.O. Box 42101

Lubbock, TX 79409Phone: 806.742.3162



FUNCTIONAL FOOD PRODUCTS: ISSUES AND IMPLICATIONS

FOR SOCIETY AND FIRMS

Shelly Freyn, Cleveland State University, ClevelandSreedhar Madhavaram, Cleveland State University, Cleveland

SUMMARY

In about two decades, the proportion of people whoare obese has increased from one-sixth of the populationto almost one-third (Taylor and Leitman 2002). Further-more, Food Marketing Institute (2006) reports that 59percent of all grocery shoppers are overweight. Also, thisproblem is not limited to adults. Even for the youth in theUnited States, the health statistics do not look very prom-ising. One in eight children has two or more risk factors forheart disease, 10% of teens have total cholesterol levelsabove 200 mg/dL, and as many as 31 percent of the youthare at risk of or are overweight (Sloan 2006). Among otherthings, food is considered a major reason for causing suchunhealthy trends.

As obesity is reaching almost epidemic proportionsin the United States, the food industry is responding withhealthier options for consumers. For example, for theyounger consumers, the industry has come out with milkdrinks with added vitamins and calcium, and less sugar,cereal bars enriched with folate and DHA (omega 3 fattyacids), and baby food products with iron and DHA (Sloan2006). Information Resources Inc. (2006) revealed that,in 2005, over half of the most successful new U.S. con-sumer food and beverage brands had “good-for-you”benefits. Food products that offer health benefits are oftenincluded under the rubric “functional foods” and arebelieved to have the potential to benefit society and firms.For Mark-Herbert (2002), from a societal point of view,

preventing diseases or delaying the onset through certainfood products holds great promise. For firms, the growingself-care movement and the overwhelming scientific evi-dence highlighting the critical link between diet andhealth resulted in the functional foods category’s enor-mous potential (Hasler 2000).

However, three critical issues hinder our society(consumers) and firms from benefitting from functionalfoods. First, there is no clarity on the definition of func-tional foods and, therefore, (1) consumers are often con-fused about the claims of the products that they purchaseand (2) some firms could potentially take advantage of thesituation and introduce products with false or inflatedclaims. Second, current public policy falls short of regu-lating firms’ claims with regards to functional foods. Asfunctional foods are loosely defined, firms have the op-portunities to market several products as functional foods.Third, there is very little medical/nutritional research andmarketing/consumer research with reference to func-tional foods. Addressing these three issues, this paper,first, discusses the various definitions of functional foodsand proposes a classificational schemata for functionalfoods. Second, the adequacy of current regulations in thecontext of functional foods is assessed. Third, the currentmedical/nutritional research and marketing/consumer re-search for functional foods is evaluated. Finally, the paperconcludes with a discussion of implications of our paperfor researchers, firms, and society. References are avail-able upon request.

For further information contact:Sreedhar R MadhavaramDepartment of Marketing

Cleveland State University2121 Euclid Ave. BU 446

Cleveland, OH 44115Phone: 216.687.3797



WHO IS ON YOUR SIDE? AGENCY REPRESENTATION AND

CONSUMER WELFARE IN REAL ESTATE MARKETS

Kristin Rotte, Australian Graduate School of Management, AustraliaMurali Chandrashekaran, Australian Graduate School of Management, Australia

ABSTRACT

Often, the most expensive and emotional purchase anaverage consumer makes is that of a house. Statistics hintat a complex and uncertainty-laden purchase process inwhich a consumer actively views anywhere from eight tosixteen houses over eight to ten weeks (Anglin 1997;Bishop, Hightower, and Bickicioglu 2005), invests over$200,000 in the purchase (median U.S. single-familyhouse in 2006 was $221,900, and U.S. mean was $269,500;National Association of Realtors 2007), borrows a signifi-cant amount of that from a lending institution, and pays offthe loan over several years. New consumers constantlyenter the marketplace, and almost 40 percent of homessold are to first-time homebuyers (Bishop, Bickicioglu,and Hightower 2006).

INTRODUCTION

Like many professional service situations (e.g., healthcare, law, and accounting), homebuyers and sellers end uprelying on someone else to resolve the uncertainty forthem. In 2005, 90 percent of buyers utilized the servicesof a real estate agent in the process of purchasing a home(Bishop et al. 2005). From a marketing strategy perspec-tive grounded in economic theory, the use of agents isrational – constructs such as search costs, relative uncer-tainty among alternatives, expertise, and information asym-metry can be invoked to motivate the discussion. At thesame time, however, uncertainty for some representsexploitable opportunities for others.

Consider the 2000 case of Bonnie Mehner, a realestate agent in the state of Alaska, who represented ahomebuyer in finding a new home. Like all good realestate agents, Ms. Mehner showed the potential homebuyerseveral available properties. Unbeknownst to the poten-tial homebuyer, however, she was also the listing agent(that is, representing the sellers) for several of theseproperties. Following the search process, Ms. Mehnerproceeded to assist the homebuyer with selecting a home,and writing a purchase contract that was later accepted.What was special about the house that was bought? Ms.Mehner also represented the seller, but never disclosedher “dual agency” status to either the buyer or the seller.Later, after discovering this “dual agency” behavior andbelieving that this lack of disclosure led to misrepresenta-tion that resulted in overpaying for his home, the homebuyer

sued Ms. Mehner and the case was taken to court. In court,Ms. Mehner, in response to a question as to why she didnot disclose her relationship with both parties (which atthe time was required by law in the state of Alaska, and byindustry standards), was quoted saying, “I’m afraid thatI’m one of a thousand (real estate agents) who don’t quitelive up to the standard because it is not practical” (Evans2002a). The end result? Ms. Mehner and her real estatecompany were fined more than $200,000 for their roles inthis transaction. The case went to spark legislation thatresulted in the practice of dual agency being outlawed inthe state of Alaska as of January 1, 2005.

From a consumer welfare perspective, it would beideal if cases like Ms. Mehner’s were the exception ratherthan the rule. Ms. Mehner’s quote, however, seems toindicate to the contrary. In the U.S., about 28 percent ofhome sales fall under the same structural description as theones sold by Ms. Mehner, that is, where the same agent/real estate company represents both the buyer and theseller (Turnbull and Dombrow 2005). Consistent withindustry terminology, we refer to such transactions asdual agency transactions. If Ms. Mehner’s sentimentcaptures the pulse of the real estate community in general,there are enormous consumer welfare and policy implica-tions. As we will see shortly, though there is little system-atic academic research to date on the implications of dualagency, burgeoning anecdotal evidence indicates that thisis a real problem that should be addressed.

The goal of this research, therefore, is to understandthe implications, if any, of dual agency on market out-comes in general and on consumer welfare in particular.Specifically, do dual agency transactions result in highersale prices and hurt buyers? The research is motivated byseveral considerations. First, home ownership often rep-resents the life-savings of individual consumers–makinga mistake or a bad investment can lead to financialstruggles, if not financial ruin. Second, the relative inex-perience of many homebuyers, particularly first-timehomebuyers, in navigating the real estate transactionprocess results in agents and agencies having a relativeadvantage because they control most of the information.Third, most of the research on sale prices of homes hasbeen conducted utilizing a “hedonic price” model (Epple1987; Griliches 1961), which operates under thetradistional economic assumption of perfect equilibrium –the sale price of each home is seen as the market clearing


price that is based on specific attributes represented in theproperty. However, given that buyers and sellers havecompeting motivations to secure the “best price,” coupledwith the motivations inherent in real estate agents, thismodel may not fully capture the process through whichhouses are eventually sold.

To address our research objectives, we first describethe dual agency landscape and the specific aspects of suchcontexts that bear potential to harm consumer welfare. Wethen describe our study to determine the current impact ofdual agency on market outcomes. Focusing squarely onthe potential impact of dual agency on real estate prices,we develop models that trace a house from the originalprice that is set by the seller to the final sale price of thehouse. We utilize multiple listing service (MLS) datadescribing home sales in a large U.S. city in the Midwestto address our research objectives. The analysis proceedsin two stages: (a) we estimate an asking price model thatreflects the latent value of the house (represented by theattributes of the house and location differences) andunobserved seller heterogeneity in how far above thelatent value they set the asking price, and (b) we developa sale price model that recognizes that the sale price isarrived at through a process of negotiation in whichbuyers are anchored at some aspect of the asking price andagents help converge the buyer and the seller. We con-clude with a discussion of key findings and the researchand consumer welfare implications.

DUAL AGENCY AND CONSUMER WELFARE

IN REAL ESTATE MARKETS

Agency Representation in Real Estate Markets

Search theory indicates that the need for search arisesonly when there is relative uncertainty about alternativesin the market, that is, when there is uncertainty aboutwhich alternative is the “best” (Moorthy, Ratchford, andTalukdar 1997). Such relative uncertainty about houses isa defining aspect of real estate markets. Real estate agentshave come to be viewed as expert sources of information,and a real information asymmetry exists in the market-place (Levitt and Dubner 2005; Levitt and Syverson2005). There are several forms of relationships between acustomer (either a buyer or a seller) and the real estateservice provider, ranging from full-service representationto merely facilitation (no legal representation). There arealso two primary levels of dual agency relationships,ranging from situations when same agent represents boththe buyer and the seller, to situations when two separateagents from the same real estate company represents theparties. In this research, we are concerned with transac-tions conducted under dual agent relationships (i.e., situ-ations when the buyer and seller have the same individualagent).

Expectations in the Agent-Consumer Relationship.

In most U.S. states, industry-level standards, as well asreal estate law, dictate that a listing agent and firm (that is,the party that represents the home seller) has a fiduciaryduty to the home seller, while the buyer’s agent and firmhas a fiduciary responsibility to the home purchaser. Froma compensation perspective, most real estate agents arecompensated through 100 percent commission and areconsidered independent contractors of the brokerage firm.Therefore, agents are compensated only when a transac-tion closes (when funds and ownership are transferred). Inthe U.S., total commission on a home sale is typically 5–6 percent of the final sales price, and it is split between fourparties: the seller’s agent, the seller’s agent’s company,the buyer’s agent, and the buyer’s agent’s company.

Tensions in the Agent-Consumer Relationship. Inreality, there is some inherent conflict between agents andtheir clients from the outset. Though the total commissionon a house sale may be 6 percent, the listing agent andbuying agent would typically each receive only 1.5 per-cent of the final sales price. As a result, from a self-interested perspective, real estate agents would prefer toclose a sale faster than for more money. Therefore, if anagent is helping a prospective buyer look for a housearound $100,000, that agent will make $1,500 commis-sion regardless of whether they find a house within oneweek or within 10 weeks. Here, the buyer’s objectivewould be to find the best house for $100,000, while theagent’s objective might be to find a “suitable” house asquickly as possible (notwithstanding the ethical require-ments of serving the consumer without such a self-inter-ested preference guiding their behavior).

The relative attractiveness of a faster sale, however,may be lower for dual agents. This is because the potentialfor double commission may lower the time-sensitivity ofreturns. For instance, in the context of the above example,the house that sells for $240,000 will net the dual agent$7,200 rather than the $3,600 from a non-dual agentsituation. Evidence that a real estate agent’s proclivity for“faster” sales is diminished when confronted withsignificanlty higher returns is revealed by Levitt andSyverson (2005), who found that agents who sell theirown homes tend to sell them for more money (on average$7,600 more amounting to about 3.7 percent in theirstudy) and leave them on the market 10 percent longer.This suggests that there exists a monetary value abovewhich the agent is willing to spend more time searchingfor a buyer.

Potential and Evidence for Consumer Harm

In a dual agency situation, it is not only very difficultfor a single agent to fulfill all required services to eitherparty, but she is now faced with earning double commis-


sion, which will only serve to increase the chance of self-interested behavior. Certainly, the dual agent can performwell in these situations and conduct herself in an honestand fair manner by disclosing any known material defectsor other non-confidential information. However, the dualagent cannot truly negotiate or advocate on behalf ofeither buyer or seller or disclose any confidential informa-tion of either party (e.g., why a homeowner is selling, theactual price at which homeowner will sell or the buyer willbe willing to purchase). Instead, she may end up actingmore as a facilitator, essentially leaving the buyer and theseller to represent themselves.

Lack of Disclosure. In 1997, the Massachusetts Of-fice of Consumer Affairs conducted a statewide study totest the compliance of real estate agents with agencydisclosure laws. Among the 45 real estate companiestested, not one complied fully (Cassidy and Curran 1997).Two years later, a study by the Massachusetts Board ofRegistration of Real Estate found that fewer than 10percent of agents fully complied with agency disclosurelaws (Perkins 1999). Likewise, a study by the NationalAssociation of Realtors (NAR) found that while 30 per-cent of buyers were told at their first meeting that a dualagent situation could potentially exist, 28 percent weretold when the contract was actually being written and 42percent of buyers did not receive agency disclosure, didnot know, or were unsure if they received disclosure(Bishop et al. 2005). Among first-time buyers, arguablythe most vulnerable group, only 23 percent were informedof potential dual agency at the first meeting (Evans 2006;the comparable value among repeat home buyers was35%).

Insurers and Policy-Makers Are Anxious. In thereal estate industry, agents and companies are required topurchase Errors and Omissions (E&O) Insurance (quitesimilar to malpractice insurance for doctors). As an indic-tor of a potential problem with dual agency, E&O Insur-ance premiums are higher for real estate companies thatpractice dual agency, with some insurance companiesrefusing to provide insurance if dual agency transactionsexceed 20–25 percent of all sales within the company(Evans 2002b). Policy makers are also concerned aboutdual agency. Among legislation, 40 percent of all new realestate laws introduced in government in 2005 had to dowith agency law (Legal Scan 2005). Further, in the periodfrom 2003–2005, 18 statutes or regulations, among 10U.S. states, were enacted dealing specifically with dualagency, with a few states outlawing dual agency alto-gether (e.g., Colorado, Florida, and Kansas).

Empirical Evidence. Overall, the evidence in theliterature, largely based on “cross-sectional, hedonic meth-ods,” (Evans and Kolbe 2005, p. 290) appears to find littleharm as a result of dual agency (e.g., Black and Nourse1995; Elder et al. 2000. However, Zeitz and Newsome

(2001) find that when buyer agent incentives are offered,buyers agents tend to “steer” buyers to those homes,which reflects the potential for self-interested behavior.

MODELING PRICING IN DUAL AGENCY

TRANSACTIONS

Asking Price (AP)

We start with the extant hedonic price model thatlargely focuses on attributes of a house to influence thevalue of the house (Epple 1987). The attributes can beseen in terms of both the physical attributes of the house(e.g., number of rooms, bedrooms, bathrooms) as well asthe location of the house (e.g., suburb).

We extend the hedonic price model by recognizingthat sellers approach the sale of their house from above the“latent value” (LV) of the house (reflected in attributesand location), that is, they desire to set the asking pricehigher than the latent value of the house. Sellers, however,are likely to evidence heterogeneity in this motivation –sellers are not equally motivated to seek more than thelatent value of the house. We therefore specify the askingprice model as follows:

(1) lnAPij = lnLV

ij+ |α

ij| + ε

ij

where lnAPij

denotes the log of the asking price of housei in suburb j, lnLV

ijdenotes the log of the latent value of

house i in suburb j, α ~ N(0, σ2α) captures the randomly-

distributed tendency of sellers, |.| denotes the absolutevalue, and ε is an error term Nid(0, σ2

ε). The key here is theabsolute value of α

ij – because |α

ij| > 0 and is randomly

distributed across sellers, the model recognizes that thoughsellers have a tendency to set higher asking price (note the‘+’ sign preceding |α

ij|), they differ in this underlying

tendency.1 The latent value of the house is then specifiedas a function of the attributes of the house:

(2) lnLVij

= τSA

ij + λ

Sj

where A captures the attributes of the house, λSj captures

the importance of location j to asking price from theseller’s perspective, τ

Scaptures attributes weights from

the seller’s perspective.

Sale Price (SP)

We start with the recognition that the sale price of ahouse is arrived at through a process of interaction be-tween the buyer and the seller. Clearly, in this interaction,the seller agent and buyer agent play central roles. Ourtheory has three components:

a. The buyer-seller interaction process can be capturedthrough an anchoring-and-adjustment process in


which buyers will first anchor on some price. Theagent will help the buyer formulate the anchor price,and then help the buyer adjust the initial anchor toarrive at a final sale price. The sale price therefore willtypically be less than the asking price.

b. In contrast to the sellers, buyers will approach the salefrom below the anchor price. Buyers, however, arelikely to differ in the intensity of this motivation. Forinstance, some buyers may be less concerned aboutthe final price, instead being governed by the aspectsof getting a loan and the dynamics of dealing withfinancial institutions. Other buyers may be less con-cerned about the dynamics of getting a loan andinstead focus on getting the best price possible.

c. Houses that remain on the market for a longer periodof time are eventually sold for lower prices (Knight2002; Turnbull and Dombrow 2005), and marketingactions in which agents engage (e.g., pictures orvirtual tours of the house on the internet), that canreduce search times for the buyer and/or increase theperceptions of value embodied in the house over andabove that due to the attributes of the house.

The sale price model is therefore expressed as:

(3) lnSPij = DA(η

DAlnAnchor

ij) + )*(1-DA)*(η

NDAln

Anchorij) - |β

ij| + δDOM

ij + γγγγγMKTG

ij + ξ

ij

where lnSPij

denotes the log of the sale price of house i insuburb j, lnAnchor

ij denotes the log of the anchor price

that is adopted by the buyer, DA is a 1/0 dummy variabledenoting whether or not a house sold under) dual agency,DOM denotes the “days on market,” and MKTG repre-sents a vector of variables that describe any marketingactivities associated with the house; β ~ N(0, σ2

β), |.|denotes the absolute value, and ξ is an error term Nid(0,σ2

ξ). Because |βij| > 0 and is randomly distributed across

buyers, the model recognizes that buyers differ in theirmotivation to approach the sale from below the latentvalue (note the ‘-’ sign preceding |β

ij|).

Formulating the Anchor. Which anchor does a buyeruse in arriving at a sale price? Two possible candidatesexist:

ANCHOR 1 – Asking price. In the absence of anymarket-based knowledge informing pricing, the ask-ing price is a readily available anchor, and could beutilized by a naïve buyer. Sellers may even reducetheir asking price following an initial period, which isoften presented to potential buyers as a signal that theprice is closer to market value.

ANCHOR 2 – Latent value. In contrast to individualbuyers who are uncertain about the market, real estate

agents will recognize that even a “revised” askingprice still reflects a tendency on the part of the sellerto be “above” the latent value of the house. Thebuyer’s agent could have a model of latent value ofhouses that may return a value closer to that predictedby asking price model (right-hand side of equation2a). This anchor, would be given by lnLV = τ

SAij +

λSj , which will always be lower than the asking price.

We theorize that non-dual agents are likely to anchorbuyers on the latent value of the house, while dual agentsare likely to anchor buyers on the original asking price (orthe revised asking price, if a revision exists). Conse-quently, a buyer under dual agency will anchor at thehigher asking price, while a buyer under non-dual agencywill anchor at the lower latent value.

These expectations are captured with the followingmodel for sale price:

(4) lnSPij = η

DA(lnAP

ij)*DUAL + η

NDA(lnLV

ij)*(1-

DUAL) - |βij| + δDOM

ij + γγγγγMKTG

ij + ξ

ij where lnLV

ij =

τSAij + λSj , and τ

Sand λSj are estimates from estimating

equation 1–2; ηDA

and ηNDA

are the adjustments made tothe anchor price in arriving at the sale price, for buyersrepresented by dual agents and non-dual agents, respec-tively.

Adjustments to the Anchor. Extensive research docu-ments that once anchored consumers do not adjust “suffi-ciently.” This may occur due to premature cessation of theadjustment process, or the tendency of decision makers tominimize cognitive effort, or even a self-generated pro-cess that increases the selective accessibility of an anchor(see Simonson and Drolet 2004 for a discussion of thesethemes). We would expect the final sale price to be belowthe initial anchor (i.e., both η

DA and η

NDA will be less than

1.0). We, however, believe that is more likely to be thecase for η

NDA than for η

DA. This is because in a dual agency

situation, the agent simply cannot work vigorously to helpthe buyer obtain the “best” possible price. Because dualagents’ ability to deliver on the fiduciary duty to bothparties has been compromised by their dual agency status,buyers are left to fend for themselves. All three factorsdiscussed above may converge to produce little adjust-ment by buyers under dual agency representation. Incontrast, non-dual agents are not laboring under the sametensions as dual agents, and are more likely to activelypursue a better deal for the buyers. Thus, we expect η

DA =

1.0 and ηNDA

< 1.0.

METHOD

Data and Variables

We utilize single-family home sales data, spanningone year (April 2005–April 2006) from a large Midwest-

^

^ ^^ ^

^


ern city in the United States. The data come from theMultiple Listing Service (MLS) of that city. UtilizingMLS data has several advantages: it provides an unbiasedsource of data that will accurately provide information onoriginal price set by the sellers, the final sales price anddays on market; the data identifies the seller’s and buyer’sagents by name and company affiliation, allowing us tocreate a dual agent dummy variable; the dataset includeshome attribute control variables (e.g., size of house,number of bedrooms and bathrooms, number of garagespaces, etc.); the data allow us to trace the journey of ahouse from the initial asking price to final sale price – wecan therefore calibrate both stages of the theorized model;the data includes another important control variable, loca-tion, including the area of town and suburb correspondingto each house, which permits the estimation of fixedeffects regression models that account for sources ofunobserved heterogeneity due to location differences.

Despite the high quality and detailed nature of thedata, it is pertinent to point out that MLS data also havesome disadvantages, including not having any informa-tion about buyer characteristics. Further, because indi-viduals in real estate offices enter MLS data, it is subjectto human error (e.g., not entering the correct number ofbedrooms or entering incorrect room sizes) and poten-tially missing information (e.g., not reporting the lot sizeor number of garages). Finally, data is only available forthose homes that were listed with a real estate agent.Therefore, the data cannot account for homes sold byowner or by non-members of the MLS. On the whole,however, MLS data is the most accurate data for thelargest volume of homes sales. Indeed, MLS data has beenthe primary data source for most real estate researchfocused on home sales. To be consistent with prior realestate research using MLS data, we deleted observationswith missing data on variables of interest, as well asobservations with infeasible values for the variables (e.g.,a house with zero bathrooms). A final sample size of 2351homes was used in the present study. Table 1 presents thevariables used in the research.

RESULTS

Asking Price Analysis

Two sets of results emerge from the analysis ofasking price. First, and of main interest at this stage, wefocus on the unobserved heterogeneity among sellers inapproaching the sale from “above.” We thus focus on thequestion: Is there any the value of including the |α

ij| term

in equation 1? Utilizing a Likelihood Ratio (LR) test, thenull hypothesis H

0: σ2 = 0 was rejected (χ2

1 = 18.48, p <

.0001). This reveals that there is significant heterogeneityamong sellers to approach the sale from “above,” and thatrecognizing this source of heterogeneity results in a better

overall explanation of the asking price dynamics. Second,and of less importance in this research, is the impact ofhouse attributes on asking price. The estimates revealed,not surprisingly, that while asking price decreases withthe age of the house, it increases with the number ofrooms, the number of bathrooms, the number of full-baths, the number of garage spaces, the number of levelsin the house, and the presence of air conditioning.

Sale Price Analysis

Results of estimating equation (4) are presented inTable 2.

Test 1 – σ2 = 0? This test corresponded to the valueof including the |β

ij| term in equation (4). A LR test

rejected the null hypothesis (χ21 = 13.36, p < .0001),

revealing that there is significant heterogeneity amongbuyers to approach the sale from “below” the initialanchor price – recognizing this source of heterogeneityresults in a better explanation of sale price dynamics.

Test 2 – Do dual agents and non-dual agents anchorbuyers at different prices? We first compared the hypoth-esized model to a “no anchor price” model (MR1; inwhich we allowed the house attributes and location tohave direct effects on the sale price of the house). Utilizingnon-nested tests, results indicated the superiority of thehypothesized model (AIC = 621.4 and 1248.6 for thehypothesized model and MR1 model, respectively, andBIC = 661.7 and 1640.5 for the hypothesized and MR1models, respectively). Next, the hypothesized model wasfound to be superior to the “same anchor price” model(MR2; in which both dual and non-dual agents anchoredtheir clients on the lower latent value of the house) – AICand BIC values for MR2 were 1177.3 and 1217.6, respec-tively. This supports our expectation that while non-dualagents are more likely to anchor buyers at the lower latentvalue of the house, dual agents are likely to anchor buyersat the asking price.

Tests 3 and 4 – Adjustments to initial anchor in dualand non-dual agency situations. The results presented inTable 2 indicate that the anchored price is adjusted down-wards in both agency situations to arrive at the final price.Both estimates are less than 1.0 (η

DA= .962 and = .983).

Results indicated that while the adjustment in non-dualagency situations (η

DA= 1 - .962 = .038) was significantly

different than zero (t = 4.774, p < .0001), the adjustmentin dual agency situations (= 1 – .983 = .017) was nodifferent than zero (t = .498, ns.). These results support ourtheorizing.

Stemming from the “higher anchor-no adjustment”“double whammy” in dual agency transactions, we com-puted the implied difference in sale price paid by buyers

^

^


in dual agent and non-dual agent transactions. Based onthe estimation results presented in Table 2, we found thaton average, buyers in dual agent transactions paid$16,486.67 more than their counterparts who were in non-dual agent transactions.

Results also indicate that the marketing variableshave the expected impact – promotion of the house viamultiple pictures or a virtual tour on the Internet increasesthe sale price. And consistent with expectations, houseson the market longer tend to sell at lower prices.

DISCUSSION AND CONCLUSION

The key objective of this paper was to examine whetherdual agency representation had an effect on market out-comes in general and on consumer welfare in particular.We advanced the view that, in transactions where thesame real estate agent represents both the seller and thebuyer, there is a potential for disadvantage for one or bothparties. We articulated a model that recognized the inher-ent tension between the home seller and the homebuyer innegotiating a final sale price. In particular, we theorized

TABLE 1

Variable Definitions

Variable Description

ROOMS The total number of rooms in the house.

BED The total number of bedrooms in the house.

BATH The total number of bathrooms in the house.

FBATH The number of full bathrooms in the house.

BSMT basement type (= 0 for no basement, =1 for crawl space only, = 2 for partial basement; =3 for full basement).

FP 0/1 variable indicating absence/presence of a fireplace in the house.

GARSPC Number of garage spaces in the house.

AIRCON 1/0 variable indicating the presence/absence of airconditioning.

LVL1, LVL15, 1/0 variables representing one storey, 1.5 storeys, 2 storeys, and 3 storeys, respectivelyLVL2 and LVL3 (baseline was the collection of bi-, tri- and quad-level houses).

LOTSZ The total size of the lot in acres on which the house sits.

AGE The age of the house in years.

DUAL 1/0 variable indicating non-dual agent/dual agent representation in the sale of the house.

MKTG1_PIC 1/0 variable indicating the presence/absence of multiple pictures of the house on theinternet.

MKTG2_VT 1/0 variable indicating the presence/absence of a virtual tour of the house on the internet.

AP Original asking price, in dollars, when the house was initially listed.

SP Final selling price, in dollars.

DOM Number of days the house was on the market prior to its sale.


that the home seller approaches the sale from “above” thelatent value of the house, the homebuyer adopts an initialprice anchor with the help of the agent and approaches thesale from “below” this anchor, and the real estate agent ismeant to help “converge” the two parties to an agreedsales price. This perspective stands in contrast to extantreal estate research that espouses the view that real estatemarkets are largely efficient, and that the sale price of ahome is not influenced by the real estate agent (Elder et al2000). Within the hypothesized model of sale price, thereal estate agent played a crucial role in anchoring buyersto a price point and then influencing the adjustment fromthis initial anchor to arrive at a final sale price. Wetheorized that both aspects in the convergence process –the selection of the anchor and the amount of adjustment –would be compromised in dual agent situations becausethe agent cannot truly act in both parties’ best interest toachieve a sales price that is fair. Offering a potentialgenerative mechanism for why dual agency homes maysell for more than non-dual agency homes, we tested thehypothesis that dual (non-dual) agents will anchor buyersat a higher (lower) price and work less (more) vigorouslythat result in lower (greater) adjustments to the initialanchor. The results strongly supported our theorizing anddemonstrated that homebuyers who are represented by adual agent are disadvantaged because they will pay morefor a home than homebuyers represented by a buyer agent.In particular, buyers represented by dual agents will, on

average, pay over $16,000 more than buyers representedby buyer agents. The findings reveal a process that maygovern agency representation and contributes to a betterunderstanding of the residential real estate transactionprocess and its potential for consumer harm.

While the results of this research are compelling andimportant, the research is not without limitations. Al-though MLS data is often viewed as the most unbiasedform of real estate transaction data, it does have bound-aries in its ability to describe the entire home selling orpurchase process. For example, from this data, we cannotdetect whether the disadvantage detected is a result ofdeliberate actions on the part of real estate agents, orwhether they are unintentional. In addition, MLS data alsoonly contains information on home sales that were actu-ally listed with a real estate agent/company. Therefore, wecannot account for any homes that sold outside of thisdomain (e.g., 13% of homes sales are For Sale By Owner).A limitation outside of the MLS data is that the sample isdrawn from only one metropolitan area. Consistent withextant real estate research, we chose a metropolitan areathat has achieved a stable growth rate, rather than anexcessive appreciating or depreciating one. Over the pastthree years, the real estate market on which we focused hasachieved growth of 8.6 percent, compared to a nationalaverage of 31 percent (National Association of Realtors2006).

TABLE 2

Estimation Results from the Anchoring-and-Adjustment Process Analysis of Sale Price

Effect Investigateda Independent Variable Parameter Estimate t-ratio p-value

Anchor price under dual agency (lnAP)*DUAL ηDA

0.983 28.42 < .0001

Anchor price under non-dual (lnLV)*(1-DUAL) ηNDA

0.962 120.35 < .0001agency

Days on market DOM δ -0.0004 -4.381 < .0001

Marketing actionsMultiple pictures MKTG1_PIC γ

10.146 10.12 < .0001

Virtual tour MKTG2_VT γ2

0.052 3.221 < .01

Log-likelihood -303.7

R2 0.908

NOTE: The estimates were obtained from the estimation of equation 4.

a: The anchor price under dual agency was the asking price, while the anchor price under non-dual agency was thelatent value obtained from the asking price model (equations 1–2).


A key motivating consideration in this research wasthe consumer welfare implications of dual agency. Thefinding that buyers represented by dual agents pay signifi-cantly more than buyers represented by non-dual agents istroubling. Real estate agents and agencies promote theirethical standards and practices as being industry-leading.However, the results in this study point to a problem withdual agency transactions. From a public policy perspec-tive, the implications of this research are vast. Severalstakeholders, including community groups, the govern-ment and industry groups like National Association ofRealtors, would be able to take steps to lessen the effect ofthis market inefficiency.

In this research, we have demonstrated that one of themost important financial decisions that a consumer canmake can be harmed by whom he chooses to representhim. Although this is an important contribution to both theliterature and to society as a whole, further research canenhance the knowledge of this process. For example,future research can investigate the impact of dual agencyat the company level. Even if public policy eventually

dissolves the dual agent relationship, it is unlikely that realestate firms will be excluded from servicing both buyersand sellers. As such, future research needs to understandthe implication of allowing this market structure to con-tinue.

From a consumer welfare perspective, future re-search should also investigate whether any specific groupsof buyers or sellers are more or less likely to be disadvan-taged by dual agency. For example, researchers shouldseek to understand whether the disadvantage of dualagency differs by socio-economic status of the buyer and/or seller – in other words, do lower income or higherincome customers suffer effects of dual agency differ-ently? Finally, future research should seek to understandwhether there is a relationship between subprime loans,foreclosures and dual agency. If these three market reali-ties converge to create a “perfect storm,” consumers andpublic policy makers need to understand this and takeaction. We look forward to developments in our literatureand in practice.

ENDNOTE

1 The basic structure of the model is similar to that used inthe economics literature focused on modeling pro-duction outputs and costs of firms. Although firmsdesire to minimize costs, inefficiencies, which varyacross firms, lead to costs always being higher than

they could be. Starting with the work of Aigner,Lovell, and Schmidt (1977), the contribution of these“stochastic frontier models” has been to explicitlyrecognize that process outcomes may be “nonoptimal”because of system inefficiencies in addition to ran-dom statistical noise (see Kumbhakar and Lovell2000).

REFERENCES

Aigner, Dennis J., C.A. Knox Lovell, and Peter Schmidt(1977), “Formulation and Estimation of StochasticFrontier Production Function Models,” Journal ofEconometrics, 6 (July), 21–37.

Anglin, Paul M. (1997), “Determinants of Buyer Searchin a Housing Market,” Real Estate Economics, 25(Winter), 567–89.

Bagnoli, M. and N. Khanna (1991), “Buyers’ and Sellers’Agents in the Housing Market,” Journal of RealEstate Finance and Economics, 4 (2), 147–56.

Bishop, Paul C., Shonda D. Hightower, and HarikaBickicioglu (2005), “2005 Profile of Home Buyersand Sellers,” National Association of Realtors, Re-search Division Report.

____________, Harika Bickicioglu, and Shonda D.Hightower (2006), “2006 Profile of Home Buyersand Sellers,” National Association of Realtors, Re-search Division Report.

Black, Roy T. and Hugh O. Nourse (1995), “The Effect ofDifferent Brokerage Modes on Closing Costs and

House Prices,” The Journal of Real Estate Research,10 (1), 87–98.

Blumenthal, Karen (1995), “Could Your Broker Be aDouble Agent?” The Wall Street Journal (EasternEdition), (September 8), B8.

Cassidy, Tina and Karen Curran (1997), “Realtors Break-ing State’s Disclosure Law: Consumer Affairs ‘Sting’Finds Buyers Not Informed,” Boston Globe, (March29).

Elder, Harold W., Leonard V. Zumpano, and Edward A.Baryla (2000), “Buyer Brokers: Do They Make aDifference? Their Influence on Selling Price andSearch Duration,” Real Estate Economics, 28 (Sum-mer), 337–62.

Epple, Dennis (1987), “Hedonic Prices and Implicit Mar-kets: Estimating Demand and Supply Functions forDifferentiated Products,” The Journal of PoliticalEconomy, 95 (February), 59–80.

Epley, Nicholas and Thomas Gilovich (2001), “PuttingAdjustment Back in the Anchoring and AdjustmentHeuristic,” Psychological Science, 12 (5), 391–96.

Evans, Blanche (2002a), “Why Some Dual Agents Delay


Disclosing Their Agency Status,” Realty Times, [http://realtytimes.com/rtapages/20020731_agencystatus.htm], July 31.

____________ (2002b), “E&O Insurers Say Dual AgentsPay More, Coverage Will Tighten,” Realty Times,[http://realtytimes.com/rtapages/20021022_insurers.htm], October 22.

____________ (2006), “Agent Disclosure Worse ThanEver,”Realty Times, [http://realtytimes.com/rtapages/20060124_disclosure.htm], January 24.

Evans, Richard D. and Phillip T. Kolbe (2005),“Homeowners’ Repeat-Sale Gains, Dual Agency andRepeated Use of the Same Agent,” The Journal ofReal Estate Research, 27 (3), 267–92.

Griliches, Z. (1961), “Hedonic Price Indexes for Automo-biles: An Econometric Analysis of Quality Change,”in The Price Statistics of the Federal Government,General Series, 73, New York: National Bureau ofEconomic Research, 137–96.

Jacowitz, Karen E. and Daniel Kahneman (1995), “Mea-sures of Anchoring in Estimation Tasks,” Personal-ity and Social Psychology Bulletin, 21 (11), 1161–67.

Knight, John R. (2002), “Listing Price, Time on Market,and Ultimate Selling Price: Causes and Effects ofListing Price Changes,” Real Estate Economics, 30(2), 213–37.

Kruger, Justin (1999), “Lake Wobegon Be Gone! The‘Below Average Effect’ and the Egocentric Nature ofComparative Ability Judgments,” Journal of Person-ality and Social Psychology, 77 (2), 221–32.

Kumbhakar, Subal C. and C.A. Knox Lovell (2000),Stochastic Frontier Analysis. New York: CambridgeUniversity Press.

Legal Scan: Legal Issues Facing Real Estate Professionals(2005), National Association of Realtors, Legal Af-fairs Division Report.

Levitt, Steven D. and Stephen J. Dubner (2005),Freakonomics. New York: Harper Collins Publish-ers.

____________ and Chad Syverson (2005), “Market Dis-tortions When Agents are Better Informed: The Valueof Information in Real Estate Transactions,” NBERWorking Paper No. W11053, National Bureau ofEconomic Research, Inc., Cambridge, MA 02138.

Moorthy, Sridhar, Brian T. Ratchford, and DebabrataTalukdar (1997), “Consumer Information Search Re-

visited: Theory and Empirical Analysis,” Journal ofConsumer Research, 23 (March), 263–77.

National Association of Realtors (2004), “UnderstandingRepresentation,” [http://www.realtor.org/rmomag.nsf/0/e9724a6ae1f7754a85256e3d00686d5a?OpenDocument].

National Association of Realtors (2006), “2006 Market-by-Market Home Price Analysis Reports,” ResearchDivision Report, Washington, DC: [http://www.realtor.org/Research.nsf/pages/MetroHomePriceAnalysisReports?OpenDocument].

National Association of Realtors (2007), “Single-FamilyExisting-Home Sales and Prices,” Research DivisionReport, Washington, DC: [http://www.realtor.org/Research.nsf/Pages/EHSdata].

Ong, Seow Eng, Poh Har Neo, and Andrew C. Spieler(2006), “Price Premium and Foreclosure Risk,” RealEstate Economics, 34 (2), 211–42.

Perkins, Broderick (1999), “Who’s on Your Side WhenYou Buy a Home?” Realty Times, [http://realtytimes.com/rtcpages/19990218_disclosure.htm], February18.

Rosen, Sherwin (1974), “Hedonic Prices and ImplicitMarkets – Product Differentiation in Pure Competi-tion,” The Journal of Political Economy, 82 (Janu-ary/February), 34–.

Simonson, Itamar and Aimee Drolet (2004), “AnchoringEffects on Consumers’ Willingness-to-Pay and Will-ingness-to-Accept,” Journal of Consumer Research,31 (December), 681–90.

Strack, Fritz and Thomas Mussweiler (1997), “Explain-ing the Enigmatic Anchoring Effect: Mechanisms ofSelective Accessibility,” Journal of Personality andSocial Psychology, 73 (3), 437–46.

Turnbull, Geoffrey and Jonathan Dombrow (2005), “Iden-tifying Agent-Specific Influences in the BrokerageProcess,” National Center for Real Estate Research:National Association of Realtors.

Tversky, Amos and Daniel Kahneman (1974), “JudgmentUnder Uncertainty: Heuristics and Biases,” Science,185, 1124–31.

____________ (2002), “Agency Reprsentation and theSale Price of Houses,” The Journal of Real EstateResearch, 24 (September/October), 165–91.

For further information contact:Kristin Rotte

Australian Graduate School of ManagementAustralia

Phone: 612.993.1928Fax: 612.931.3727


Murali ChandrashekaranAustralian Graduate School of Management,

AustraliaPhone: 612.993.1925

E-Mail: muralic2aagsm.edu.au


MARKETING IN SECOND LIFE: IF YOU BUILD IT, WILL THEY COME?

Lakshmi Goel, University of Houston, HoustonSonja Prokopec, ESSEC Business School, France

ABSTRACT

Virtual worlds present a new channel for communi-cation, information dissemination, and business transac-tions. The purpose of this paper is to provide a startingpoint for a discussion on the value of virtual worlds, andstrategies that organizations may adopt when establishinga virtual presence in them.

INTRODUCTION

Traditionally, three-dimensional (3D) environmentshave been used for modeling in the fields of engineering,architecture (e.g., Whyte et al. 2000), education (e.g.,Bouras et al. 2004), and product development. Morerecently, interactive 3D environments are popular plat-forms for gaming and entertainment. Internet-based 3Dvirtual environments, or virtual worlds, are the next evo-lution in technology that promise to transform the currentmode of communication and information dissemination.While the Internet today is primarily two-dimensional andcommunication is through text, audio, or video files,persistent virtual worlds1,2 provide an immersive environ-ment where “avatars”3 can touch and feel, and interactwith other avatars or objects in real-time. The businessimplications of these virtual worlds, while potentiallyvast, have thus far not received much scholarly attention.Early innovators have established a presence in virtualworlds such as Second Life, and practitioner outletsdiscuss the potential and current problems, faced in theseenvironments. The purpose of this research is to draw onliterature in marketing and strategic management to ana-lyze the impact of the virtual presence of organizations oncustomers. In doing so, we discuss strategic elements thatshould be borne in mind by organizations that wish toestablish a presence in a virtual world.

Virtual Worlds

As 3D virtual environments have become more readilyavailable and open in some aspects, applications of these“virtual worlds” have grown beyond gaming and 3Dmodeling to other forms of entertainment and informationassimilation such as virtual real estate, brand building, andmarketing. Businesses are beginning to see them as op-portunities for collaboration or as a new vehicle forreaching customers or business partners. One reason is“Moore’s Law” which steadily produces cheaper andmore effective hardware (Schaller 1997), resulting in

faster and richer 3D environments. A second could be theevolution of representation of information from basic textto richer, more intuitive forms that include multimediasuch as sound, video, animation, and life-like structures.The “milleneals,” the generation born after 1980, hasgrown up with digital media during their formative yearsand are accustomed to visual and auditory channels suchas high-definition television, podcasts, surround-soundsystems, e-books, email, and instant messaging, and so-cial networking tools such as MySpace and FaceBook.This generation is comfortable using virtual platforms forsocial networking, education, business transactions, andtelework. Virtual networks and crowd-sourcing4 plat-forms (such as Wikipedia, epinions, MySpace, andYouTube) are a prominent source of information for themillinneals. With required technological development tosupport them, it seems intuitive that 3D environments thatsupport networking with richer audio, visual, and textualfeatures will replace the current form of standard two-dimensional information representations on the web.

Early innovators such as IBM have set up “islands” inSecond Life, an internet-based persistent 3D world, ex-clusively for the purpose of providing a virtual meetingplace for its employees to interact. Organizations such asSony and Dell use Second Life as a channel for promotingtheir brand image. STA Travels, a travel agency catered tostudents, has set up an island in Second Life to reach themilleneals, and expand its customer base. Toyota andReebok, amongst others, use Second Life to collect cus-tomer preference information and expand their marketingchannels. Such applications could, in time, lead to seriouseconomic and business transformation (Hemp 2006).

E-Commerce and Virtual Worlds

With the popularity of the Internet, there has been aplethora of research that looks at leveraging virtual oppor-tunities for e-commerce. Advantages of web channelsover physical channels, derived from transaction costeconomics, include economies of scale from reaching awider customer base, lower sunk costs in “brick-and-mortar” set-ups and inventory, lower costs due to bypass-ing retail distribution intermediaries, higher degree oftransaction automation, and availability of the channel24/7 at no extra cost.

The decision to engage in a particular e-commercestrategy can be seen to stem from three factors – Cost


Reduction, Value-Addition, and Market Extension(Steinfield et al. 2001). 3D worlds, by virtue of beinginternet-based, can be hypothesized to enjoy the sameadvantages of traditional web channels. However, thenew environment presents unique opportunities for repre-sentation of products and services. Unlike at a 2D website, customers can be greeted in real-time by a virtualsales representative, can try on clothes to see how they fittheir body specifications, can test-drive cars while sittingat home, and can experience a virtual vacation in Hawaiiwhen deciding to buy from a travel agent. Hence, virtualworlds can be hypothesized to provide a higher level ofValue-Addition and Market Extension than 2D web sites.Given the newness of the concept and the technology,little is known about whether the same strategies that wereused for web-channels will apply to virtual worlds. How-ever, we believe that virtual worlds overcome some limi-tations of traditional web channels.

One major limitation of web-based channels is that ofTrust. Research in e-commerce has recognized that asignificant impediment of virtual stores is the consumers’lack of trust in its legitimacy (for example see Grabner-Kraeuter 2002 or Bramall 2004). Perceived risk is foundto be significantly lower in physical establishments whereconsumers feel they have a place where they can returngoods, register complaints, or speak with someone (Leeand Tan 2003). Additionally, physical businesses areoften embedded in social networks, which enhance trust(Granovetter 1985). Consumers see, and can interact witheach other when shopping through physical channels.Consequently, organizations try to create online socialnetworks through virtual forums where consumers canexchange opinions, chat with customer service personnel,and rate products or services (Jarvenpaa et al. 2000;Stewart 2003; Achrol and Kotler 1999). It can be hypoth-esized that virtual worlds provide a richer platform fordevelopment of social networks, and hence increasedtrust, among consumers.

The Effects of Types of Products and Services

Some strategies used by organizations that havevirtual and physical presences are that of “virtual,” “par-allel,” “mirror,” and “synergy” (Steinfield et al. 2000).5 Avirtual strategy is one where the organization does nothave a physical presence. A parallel strategy is one wherean organization’s physical presence is not explicitly linkedto its physical presence. When adopting the mirror strat-egy, firms develop a web channel that resembles theirphysical channel in terms of the look and feel, and theofferings of products and services. When adopting asynergy strategy, firms explicitly link their virtual andphysical presence, but exploit each channel’s uniquestrengths. While organizations are testing virtual worldsfor different purposes (e.g., brand awareness, customerpreferences), we believe that the type of product that the

firm offers has a moderating effect on the impact of itspresence in the virtual world, on consumers. In particular,we feel that virtual worlds are more suited for certain typesof products, and hence may influence what strategy thefirm should adopt. Hence, our research question is: Whatstrategies are best suited for organizations that wish toestablish a presence in virtual worlds, based on the typeof product they offer?

In terms of service, virtual worlds provide a channelfor adding value to a customer’s experience. Companiessuch as STA Travel have hired full time employees inSecond Life that perform as managers in their virtualshowroom. Websites may offer real-time chat facilities orcustomer support via email or telephone, but having anavatar of an employee sharing the same virtual space withthe customer synchronously while providing informationabout the product or service, enhances the experience ofthe customer. Additionally, such service enhances thelevel of trust as discussed earlier. For a customer based inDallas who wishes to plan a holiday in Australia, interact-ing with a travel agent from Sydney in Second Life addsfar more value than meeting an agent in Dallas in real life.However, for the purpose of this study, we focus solely onthe products offerings of organizations in virtual worlds.

In order to arrive at a classification for products, weutilize a matrix that differentiates between products basedon two-dimensions – product value, and an “experiential”component. The term “experiential” has traditionally beenused to differentiate consumption that emphasizes sen-sory pleasures, playful leisure activities, and emotionalresponses (analogous with “hedonic” consumption) from“utilitarian” consumption, which focuses on problemsolving (Holbrook and Hirschman 1982). However, thisdichotomy of “hedonic” versus “utilitarian” consumptionhas been used for analysis of shopping experience andshopping value (Babin, Darden, and Griffin 1994; Jones,Reynolds, and Arnold 2006), or to analyze the activity ofonline purchases (Maenpaa, Kanto, Kuusela, and Paul2006); to our knowledge, it has not focused on the natureof products. We use the term in the same vein as this priorliterature, but emphasize the nature of certain productsthat have a higher degree of the need to be “experienced”before they are purchased – such as a car, or a newperfume. Based on these dimensions, we classify productsas Low-Value Low-Experiential, Low-Value High-Ex-periential, High-Value Low-Experiential, and High-ValueHigh-Experiential. We discuss these below.

Low-Value Low-Experiential. Examples of suchproducts include text books, paper towels, or pencils.Consumers typically do not need to spend much effort inbuying such products because they are typically inexpen-sive and do not need to be tried-out or experienced beforethe purchase is made.


High-Value Low-Experiential. Examples of suchproducts include office furniture, or the encyclopediabooks. These products are typically more expensive, butthe consumer knows what to expect when purchasing thistype of a product.

Low-Value High-Experiential. Products such as ap-parel, accessories, music, and electronic gadgets such ascell phones would fall in this category. These products,even though typically inexpensive, more often than notrequire consumers to try them on, feel them and/or expe-rience them.

High-Value High-Experiential. Products such ascars, expensive art, or an overseas vacation are productsthat are high in both value and experiential component.They are the most risky purchases because they are noteasily returnable. They also require the most effort andcommitment from the consumer.

PROPOSED RESEARCH METHODOLOGY

This paper is primarily intended to be of a conceptualnature and aims to provide a starting point for a discussionon how companies can use virtual worlds. In order to testand support hypotheses derived from our research ques-tions, we intend to conduct experiments and surveys withconsumers in virtual worlds. A theoretical sampling oforganizations that offer products based on our categoriza-tion will be done. We will use scales validated by priorliterature to assess various outcomes such as customersatisfaction, brand identification, trust and intention tobuy a product.

In this paper, we present some preliminary qualita-tive data that was gathered for an exploratory study ofconsumers in virtual worlds. While we do not claim thatthis data is representative or generalizable, and fullyrecognize the need for an in-depth study to support ourclaims, we use this data to get an initial impression ofthemes and concepts for our research.

Our subjects are avatars who visited three islands inSecond Life – Sony BMG, Scion City (Toyota), andReebok. The avatars interviewed were asked open-endedquestions about their satisfaction with the virtual stores,

perceptions of the brand, and a description of their expe-riences with the products and services the stores offered.While some products and services are designed exclu-sively for avatars in the virtual world rather than forindividuals in real life, we believe that these transactionsplay an increasingly important role in the organization’scommerce. Some virtual worlds such as Second Life havean economy of their own. Currency in Second Life is inLinden Dollars which has an exchange rate of 265 LindenDollars for 1 U.S. Dollar. Avatars in Second Life can own,and create products and services in Second Life and retainthe exclusive rights for them. They can buy and sell theseproducts and services “in-world” for Linden Dollars,which are fully transferable with U.S. Dollars. In May2007, 293,102 customers spent approximately $1,931,011in one day in Second Life. (See more statistics at http://secondlife.com/whatis/economy_stats.php). Second Lifehas a total population of 7,472,404, with 1,677,740 mem-bers active in the last 60 days. Hence virtual worlds havea growing healthy economy that mimics that of the realworld. Experiences of consumers at each store are de-scribed next.

Sony BMG primarily uses their island for showcas-ing artists, latest music and upcoming events. Avatars canbrowse through different areas of the virtual store andlisten to streaming music. The store also offers free goodiebags with clothes and music for the avatar. While thevirtual store looks like a physical store of Sony BMG, itdos not allow avatars to buy CDs or DVDs, but insteadprovides music and videos for enjoying in-world. Thevirtual store is customized to avatars in Second Life. Thissuggests that the purpose of establishing a virtual store inSecond Life was to get avatars (who already are or mightbecome consumers in real world) aware of their presenceand more comfortable with the store. Hence we can inferthat Sony BMG followed a synergy strategy for enhanc-ing company’s brand image when designing their virtualstore in Second Life.

When asked their impressions of the store and thebrand, some comments from avatars in Sony BMG were:

I felt amazed by the multitude of activities that areavailable. And you can definitely tell the huge differ-ence between Sony and other locations. They are

TABLE 1

Product Classification Matrix

Low Experiential High Experiential

Low Value e.g., Textbooks, office supplies e.g., Electronic gadgets, cell phones, CDs

High Value e.g. Office furniture, encyclopedias e.g., Cars, perfumes, art


ahead of the game. I feel that the store (brand)appreciates me as a customer because they offeredme to take me to other web pages for me to learn moredetails about their features. I felt like they reallywanted to do business.

I like the store because I was able to hear and seeactual artist videos and music in a dynamic environ-ment. The image of this store (brand) has improvedbecause they already have a presence on second life.

I like the store because I could dress up my characterand add music.

I feel that the store (brand) appreciates me as acustomer because they made a virtual world for me toexplore with a map for easy exploration.

The brand image is enhanced as consumers feel thatSony BMG is an innovator by establishing a virtual store.The primary product, music, is easily marketed and show-cased in the store.

The second island visited was Scion City (Toyota).This island is used by Scion as a virtual car store whereavatars can see new models and test drive the cars. Whileit is common for avatars to create and own vehicles(though they are not of much use since avatars can fly andteleport in Second Life), being able to test-drive a car froma real world company was unique. This is reflected in thequotes below:

I felt like this could be a place where I could hangoutwith friends and why not organize a race!!!!

The image of this store (brand) has improved becauseI would not think they would have a store on secondlife.They are trying to stay ahead of the game, which Ibelieve is a positive message.

I feel that the store (brand) appreciates me as acustomer because they offered me to test drive aScion. I think this a great way to remind people aboutthe brand.

The image of this store (brand) has improved becausetheir presence shows that they are interested in usingnew technology to get to their potential customers.

The consumers appreciated the fact that Scion is oneof the early adopters of virtual worlds and enjoyed thedriving experience from the perspective of their avatar.However, this did not reflect the strategy of Scion, whichwas to mirror a physical showroom where potential cus-tomers can test drive new models of their cars. Simulatingthe high degree of the sense of experience, and look-and-

feel associated with test-driving a car is difficult. From thequotes below, it is clear that customers felt a sense ofdisappointment with the experience:

[The store] wasn’t attractive in the first place, itlooked dull, the cars were placed here and there anddidn’t have the style or anything to attract me as acustomer. The driving experience was strictly ok; itwas more similar to driving a car in a video game. Idon’t think they made it feel like a virtual store oranything.

I felt disappointed. There’s really nothing to interactwith other than the cars to test-drive them.

The test drive felt like a video game that was hard tocontrol. I test drove different looking cars, but the feelwas the same in each.

I didn’t like the store because they focused on testingthe car but the reality is that the customer won’t reallyknow what the car is like just by testing it on SecondLife. More company and vehicle information or videoads should be available.

As can be seen in the quote above, the problem is notwith the technology, but rather with the inadequacy of theexperience of test-driving a car virtually.

The third island visited for interviews was that ofReebok. Reebok allows avatars to buy “blank shoes”which are basic white shoes in different sizes. Each pair ofshoe costs 50 L$ (approximately 18 cents USD). Theavatar can then customize the shoe by changing colors ofdifferent areas in the shoe. Each additional customizationcosts the avatar 5 L$. Reebok can easily gather data of howconsumers customize their shoes. The virtual store alsoadvertises sports apparel of the brand on posters displayedon the walls of the stores. Hence, the products offered arefor in-world avatars, but the channel is also used for brandimaging and marketing real-world offerings. This sug-gests that the company follows a combination of mirrorand synergy strategy. Excerpts from some interviews inReebok are given below:

I like the store because I was able to customizeReeboks product real time and was able to zoom infor detail.

The image of this store (brand) has improved becausethey already have a presence in second life eventhough it is a somewhat new system.

It was how real life shoe stores are and they took adetailed step to make the customer feel at home, likethe poster and the way it was arranged.


I feel that the store (brand) appreciates me as acustomer because the store was made just for thecustomers, selection is easy, customization also wasquite easy though it was a bit hard initially but thenit was fine. So it did care about the customers.

It was a very easy experience to buy the shoes andwear it, just needed to left click on it and we had theoption to pay it right away and get it, wearing theshoes also was really easy so over the purchase atReebok was pretty much similar to what was in thereal life world so it was a very neat kind of thing whichsecond life provided in the Reebok store

Hence, customers appreciated the fact that Reebokhad a presence in the new environment, and that therewere products targeted for in-world avatars.

DISCUSSION

Since virtual worlds are relatively new and untestedin terms of their potential as a main channel of sales andmarketing, it is early to draw conclusions on their applica-bility. However, from previous literature, and some pre-liminary insights, we can arrive at the following proposi-tions:

P1: Virtual worlds provide a greater degree of valueaddition, a new channel for market expansion, andhigher trust, than traditional web channels such aswebsites.

Given the volume of traffic in virtual worlds today,early adopters are able to reach a greater customer base.By customizing products and services to be inline with thenew environment, and by providing a different way forcustomers to experience these products virtually, there isadded value to the product or service. The nature of thesevirtual worlds enables virtual co-location of multiple“avatars” in the same virtual space. This is not possible intraditional online forums where communication is mainlythrough asynchronous text between people representedthrough nick names. Avatars are customizable in severaldimensions, based on body size, shape, hairdo, makeup,clothes, accessories, and possessions (such as virtualhouses or vehicles). Having such a virtual “life” andinteraction with other avatars in this persistent worldincreases the “embededness” of individuals in the socialnetworks, increasing their sense of trust. There is previousresearch that shows online retailers that use avatars havea competitive advantage because they provide customerswith enhanced perceptions of human connections and theformation of emotional bonds (Wang, Baker, Wagner,and Wakefield 2007). This proposition is also in line withresearch that studies economic leverage gained by activi-ties associated with being embedded within virtualcommunities (which is difficult to achieve through

websites) (Balasubramanian and Mahajan 2001). Draw-ing on this as well as literature in social networks in reallife (Granovetter 1973), we posit that organizations maythus, find it easier to build trust in brands in virtual worlds,as compared to websites.

P2: The type of product offered must guide the strategyemployed by an organization seeking to establish apresence in a virtual world.

We draw upon the classification we had introducedearlier to differentiate products. We believe that organiza-tions can draw greater benefit from virtual worlds if theychoose a strategy based on their type of product offering.

Low-Value Low-Experiential. Consumers typicallydo not need to spend much effort in buying products thatbelong to this category (see Table 1). Virtual channels –websites or virtual worlds – may provide only a CostReduction benefit of not having a physical presence (forexample Amazon.com versus Barnes and Noble). Virtualchannels for such products may use a “parallel” strategyor a “virtual” strategy where physical and virtual channelsare not explicitly linked.

High-Value Low-Experiential. Products belongingto this product category (Table 1) also benefit from theCost Reduction afforded by virtual channels. However,consumers are usually more likely to trust physical chan-nels when buying high-value products. Having socialnetworks that enhance trust helps companies market suchproducts virtually. Examples are websites like e-pinions,which provides a forum for consumers to discuss elec-tronics such as laptop computers and Amazon.com’s orEbay’s features of rating of suppliers and shopping expe-riences. A parallel strategy would be most suited for thesetypes of products. With virtual channels becoming moreaccepted by consumers, it has become more common-place for a pure virtual strategy, even for high-valueproducts. Virtual worlds may provide Value-Additionand Market-Extension for such products.

Low-Value High-Experiential. These products, webelieve, benefit most from virtual worlds. Consumersshopping for music or cell phones can experience a richerrepresentation of the product in virtual worlds. Hencestores such as Sony BMG and Reebok are well positionedto adopt a mirror strategy which extends their existingmarketing channels. Avatars can easily make a purchasefor shoes or music CDs to be delivered home after theyhave experienced the products on their avatars. Thefunctionality of the virtual channel would work exactly inthe same manner as an e-commerce website, with theadded advantage of the consumer experiencing the prod-uct in a 3D virtual world. Musicians and painters fre-quently hold shows and exhibition on Second Life wherethey can showcase their art at minimal costs. Avatars pay


to attend some shows and may choose to buy from theartist.

High-Value High-Experiential. Products such ascars, expensive art, or an overseas vacation are difficult tosell through virtual channels. Hence adopting a mirrorstrategy may not be optimal for companies such as Scion.Instead, by employing the strategy of synergy, an organi-zation can customize offerings in the virtual worlds toboost their brand image in real world. For example,creating a virtual café where avatars can “hang-out,” drinkcoffee, and interact may help a restaurant’s image morethan a website that lists the prices of its products. Scioncan provide automobiles for avatars that fly and are morein line with the environment in the virtual world, ratherthan try to imitate their real life showrooms. Similarly,virtual art can be made available for use by avatars in theirvirtual offices or homes. Table 2 summarizes our recom-mendation for the strategy that organizations should adoptbased on the value and experiential attributes of theirproducts.

We would like to point out that adoption of onestrategy does not necessarily preclude the adoption ofanother. For example, for Low-Value High-Experientialproducts, a mirror strategy may be combined with asynergy one – as done by Reebok. Also, this categoriza-tion is by no means static. As virtual channels and virtualworlds become more mainstream and consumers knowwhat to expect when making purchases, more types of

products are bought online. For example, it is now notuncommon for consumers to book hotel rooms for vaca-tions online. Hence a combination of synergy and mirrorstrategies may be adopted for even High-Value High-Experiential products in virtual worlds.

MANAGERIAL IMPLICATIONS

This study was exploratory in nature. The purpose ofthis research was to draw on literature in marketing andstrategic management to analyze the impact of the virtualpresence of firms on customers. While early innovatorssuch as IBM, DELL, Sony, BMG, and others have estab-lished a presence in virtual worlds such as Second Life, itseems that many of them do not have a clear idea of howconsumers are responding to it. Additionally, many firmsdo not exactly know what strategy is the optimal one tofollow. We suggest a framework to help firms marketthemselves more effectively in 3D virtual worlds such asSecond Life. The contribution of this research is twofold.First, we present a product classification matrix that canhelp firms place their products more effectively in 3Dvirtual worlds. Second, we suggest that firms think ofwhat strategy to adopt in virtual worlds based on theproducts they offer.

This research should be taken as the beginning stagesof a research program that shows promise in regards to therole of firm’s virtual presence on its customers.

TABLE 2

Recommended Strategy based on Product Classification

Low Experiential High Experiential

Low Value Parallel or Virtual Mirror

High Value Parallel Synergy

ENDNOTES

1 A virtual environment (or virtual world) is “a computer-based simulated environment intended for its users toinhabit and interact via avatars. . . . The world hassome real world rules such as topography, real-timeactions, and communication. Communication has,until recently, been in the form of text, but now real-time voice communication using VOIP is available.”(from http://en.wikipedia.org/wiki/Persistent_world).

2 A persistent world is a virtual world that is alwaysavailable and world events happen continually andchange is persistent even if a user is not logged-in.The persistency comes from maintaining and devel-

oping the state of the virtual world around the clock.Characters in these virtual worlds can influence andchange a persistent world. (from http://en.wikipedia.org/wiki/Persistent_world)

3 An avatar is an Internet user’s representation of himselfor herself . . . in the form of a three-dimensional modelused in computer games . . . (http://en.wikipedia.org/wiki/Avatar_(virtual_reality))

4 “Crowd sourcing” is a term employed for leveraging theproductive potential of an undefined group of people,usually connected to the internet to pool resourcessuch as information, images, videos etc. For example,see http://www.wired.com/wired/archive/14.06/crowds.html.


5 Stenfield et. al. (2000) identify two more – “anti-mirror”and pure “physical.” A company that adopts a purephysical strategy does not have a virtual presence,and one that adopts an anti-mirror strategy changes

its physical channel to resemble its virtual one. Sinceour focus is on the virtual channels, we choose toexamine the other strategies, which are concernedwith how organizations use the virtual channels.

REFERENCES

Achrol, Ravi and Philip Kotler (1999), “Marketing in theNetwork Economy,” Journal of Marketing, 63, 146–63.

Babin, Barry J., William R. Darden, and Mitch Griffin(1994), “Work and/or Fun: Measuring Hedonic andUtilitarian Shopping Value,” Journal of ConsumerResearch, 20, 644–56.

Balasubramanian, Sridhar and Vijay Mahajan (2001),“The Economic Leverage of the Virtual Commu-nity,” International Journal of Electronic Commerce,5 (3), 103–38.

Bouras, C., A. Philopoulos, and Th. Tsiatsos (2001), “e-Learning Through Distributed Virtual Environments,”Journal of Network and Computer Applications, 24,175–99.

Bramall, C., K. Schoefer, and S. MsKechnie (2004), “TheDeterminants and Consequences of Consumer Trustin Online Environments: An Exploratory Investiga-tion,” International Journal of Internet Marketingand Advertising, 1 (4), 388–412.

Grabner-Kraeuter, Sonja (2002), “The Role of Consum-ers’ Trust in Online-Shopping,” Journal of BusinessEthics, 39 (1/2), 45–71.

Granovetter, M.S. (1973), “The Strength of Weak Ties,”American Journal of Sociology, 78 (6), 1360–80.

____________ (1985), “Economic Action and SocialStructure: The Problem of Embededness,” AmericanJournal of Sociology, 91 (3), 481–510.

Hemp, Paul (2006), “Avatar-Based Marketing,” HarvardBusiness Review, 84 (6).

Holbrook, Morris B. and Elizabeth C. Hirschman (1982),“The Experiential Aspects of Consumption: Con-sumer Fantasies, Feelings, and Fun,” Journal ofConsumer Research, 9 (2).

Jarvenpaa, Sirkaa, N. Tractinsky, and M. Vitale (2000),“Consumer Trust in an Internet Store,” Information

Technology and Management, 1, 132–40.Jones, Michael E., Kristy E. Reynolds, and Mark J.

Arnold (2006), “Hedonic and Utilitarian ShoppingValue: Investigating Differential Effects on RetailOutcomes,” Journal of Business Research, 59 (9),974–81.

Lee, K.S. and S.J. Tan (2003), “E-Retailing Versus Physi-cal Retailing: A Theoretical Model and EmpiricalTest of Consumer Choice,” Journal of Business Re-search, 56, 877–85.

Mäenpää, Katariina, Antti Kanto, Hannu Kuusela, andPallab Paul (2006), “More Hedonic Versus LessHedonic Consumption Behavior in Advanced InternetBank Services,” Journal of Financial Services Mar-keting, 11 (1), 4–16.

Schaller, R. (1997), “Moore’s Law: Past, Present, andFuture,” IEEE Spectrum, 34 (6), 52–60.

Steinfield, Charles, H. Bouwman, and T. Adelaar (2001),“Combining Physical and Virtual Channels: Impera-tives and Challenges,” 14th Bled Electronic Com-merce Conference, Bled, Slovenia, June 25–26.

____________, D. Wit, T. Adelaar, A. Bruins, E. Fielt, M.Hoefsloot, A. Smit, and H. Bouwman (2000), “Le-veraging Physical and Virtual Presence in ElectronicCommerce: The Role of Hybrid Approaches in theNew Economy,” Presented to the International Tele-communication Society, Buenos Aires, Argentina,July 2–5.

Stewart, Katherine (2003), “Trust Transfer on the WorldWide Web,” Organization Science, 14, (1), 5–17.

Wang, Liz C., Julie Baker, Judy A. Wagner, and KirkWakefield (2007), “Can a Retail Web Site Be So-cial?” Journal of Marketing, 71 (3), 143–57.

Whyte, J., N. Bouchlaghem, A. Thorpe, and R. McCaffer(2000), “From CAD to Virtual Reality: ModelingApproaches, Data Exchange and Interactive 3d Build-ing Design Tools,” Automation in Construction, 10(1), 43–55.

For further information contact:Sonja Prokopec

Marketing DepartmentESSEC Business School

Avenue Bernard Hirsch B.P. 5010595021 Cergy-Pontoise Cedex

FrancePhone:+33(0)1.34.43.36.19

Fax:+33(0)1.34.43.32.11E-Mail: [email protected]


PROPOSING AND CONCEPTUALIZING A SERVICE-DOMINANT

STRATEGIC ORIENTATION

Ingo O. Karpen, The University of Melbourne, AustraliaLiliana L. Bove, The University of Melbourne, Australia

Alexander P. Josiassen, The University of Melbourne, Australia

ABSTRACT

In a seminal article Vargo and Lusch (2004) presenta service-dominant (S-D) logic for marketing that pro-vides strategic implications for competing through ser-vice (see also Lusch, Vargo, and O’Brien 2007). Takinga strategy perspective, this study conceptualizes an S-Dstrategic orientation, proposes its dimensions, and looksat potential benefits.

INTRODUCTION

The service-dominant (hereafter S-D) logic of mar-keting (Vargo and Lusch 2004) as an emerging marketingparadigm (see Webster Jr. 2006), challenges the existingperspective and assumptions that marketing inheritedfrom economics. For example, under an S-D logic, valueis not added to a product but is what customers get out ofa product and, thus, value arises within individual cus-tomer (usage) experiences (e.g., Grönroos 2006; Prahaladand Ramaswamy 2004; Vargo and Lusch 2004). Serviceprovision rather than goods represent the core of ex-change processes (Vargo and Lusch 2004). As such, theS-D logic has been characterized as a perspective orframework that promotes a service-based understandingof marketing phenomena (e.g., Lusch, Vargo, and Malter2006; Vargo and Lusch 2004) and rests on nine founda-tional premises (see, e.g., Lusch, Vargo, and O’Brien2007).

Managers might adopt the S-D logic as their ownmental model or mindset for viewing the business worldand identifying new ways of (co-) creating value in lightof competitive advantage. However, academia currentlydoes not provide a strategic orientation that focuses on asimultaneous and coherent set of organizational practicesconsistent with the S-D logic (see also Winklhofer Palmerand Brodie in press). That is, despite the significance andapplicability of S-D principles in theory and practice,there are no respective strategic orientation addressingservice-dominant organizational behaviors, activities, andprocesses.

Considering the call for a fresh contribution of mar-keting to the strategy dialog and the criticism of existingstrategic orientations (see, e.g., Day 1992; Hunt andLambe 2000), this is especially relevant. The primary

purpose of this paper is thus to conceptualize an S-Dstrategic orientation and its dimensions. The focus will bedirected toward organizational practices because a strate-gic orientation in general concerns broad strategic choicesas manifested in respective behaviors/activities that en-able an organization to achieve long-term success (seeGatignon and Xuereb 1997). Furthermore, we will presentpotential benefits of implementing an S-D strategic orien-tation as a combined effect of the proposed dimensions ofthis behavioral concept.

To this end, we will first highlight the relevance of anS-D strategic orientation in today’s challenging environ-ment to achieve a competitive advantage. Second, we willoutline the nature of S-D logic and synthesize its strategicimplications. Third, we will briefly review marketing’skey contribution to the strategy dialog and emphasize thenecessity for a new strategic orientation that addresses thefundamentals of S-D logic concurrently. Fourth, we willpropose an S-D strategic orientation that is consistent withS-D logic but focuses on activities, behaviors, and pro-cesses. Fifth, the five proposed dimensions of the S-Dstrategic orientation will be discussed and specified. Fi-nally, potential benefits of the presented S-D strategicorientation will be suggested.

STRATEGY IN TODAY’S BUSINESS REALITY

The reasons as to why and how firms succeed or failin business present the central question in strategy (Fréry2006; Porter 1991). From a strategy perspective, achiev-ing and sustaining competitive advantage in today’s highlycompetitive, proliferated, and dynamic market environ-ments is increasingly challenging (e.g., Court, French,and Knudsen 2006). For instance, strategists have toincreasingly deal with intensified global networked com-petition, segment-of-one demand characteristics, and moreempowered customers (see, e.g., Cares 2006; Oliver,Rust, and Varki 1998; Pires, Stanton, and Rita 2006). Asthe nature of the business arena is transforming, thetraditional roles of economic actors are changing andbecoming blurred (Prahalad and Ramaswamy 2004). Forexample, customers progressively take on employee rolesand perform activities in value creation processes. Assuch, more informed, connected, and active customersprogressively demand participation in service provisionprocesses even in traditional manufacturing industries,


while organizations encourage and enable customers todo so (e.g., Bendapudi and Leone 2003; Firat andVenkatesh 1995).

Consequently, the central element of marketing rep-resents joint processes (of providing benefit) and serviceflows rather than finished units of output (Vargo andLusch 2004). Organizations thus constantly strive to pro-vide better service outcomes and seek to compete basedon service principles, independent of their industry or“pure” service business nature (Berry et al. 2006). Ac-cordingly, it is here argued that organizations could ben-efit from implementing a strategic orientation consistentwith the principles of the S-D logic as it guides organiza-tional strategy-making activities and service-like interac-tions with the market, both internally and externally. Italso provides a way of responding not only to the chal-lenges but the opportunities of the current business envi-ronment.

THE NATURE OF S-D LOGIC

The S-D logic of marketing has been described as alens, mindset or philosophy for understanding marketingphenomena (e.g., Lusch, Vargo, and Malter 2006; Vargoand Lusch 2004). As such, the S-D logic represents acertain perspective for viewing strategically relevant as-pects like customers, goods, value creation, among othersin light of service provision. For example, service provi-sion rather than goods constitute exchange processes,while interactions and operant resources form the basisfor meaningful experiences and co-value creation (LuschVargo, and O’Brien 2007; Prahalad and Ramaswamy2004). To date, there exists neither a theory nor anorientation of the firm explicitly based on the S-D logic(see, e.g., Vargo in press; Vargo and Lusch 2006). Never-theless, S-D logic may lead to a general theory of market-ing and, as such, is “[…] implicitly normative and thus canpoint managers toward practical actions and organiza-tions to standards for ethical interaction and social well-being” (Lusch and Vargo 2006, p. 407). Adopting S-Dlogic as an informing perspective for strategy allows thisstudy to merge the strategic implications of S-D logic andrelated literature streams such as service, relationship, andnetwork marketing into a coherent S-D strategic orienta-tion of the firm.

THE CORE STRATEGIC IMPLICATIONS OF

THE S-D LOGIC

S-D logic assumes that service is the vantage pointand core of every exchange, rather than goods being thecentral element of a market offering (Vargo and Lusch2004). As such, “[i]nteractivity, integration, customization,and coproduction are the hallmarks of a service-centeredview and its inherent focus on the customer and therelationship” (Vargo and Lusch 2004, p. 11). Recently,

Lusch et al. (2007, p. 8) provided a set of propositions forcompeting through service grounded in the premises ofthe S-D logic. Based on these and the associated S-D logicliterature (e.g., Ballantyne and Varey 2006; Flint 2006;Grönroos 2006; Kalaignanam and Varadarajan 2006;Lusch et al. 2006; Lusch et al. 2007; Rust and Thompson2006; Vargo and Lusch 2004), the following core strate-gic implications are proposed: (1) focusing on individualcustomers’ own value creation processes and experi-ences; (2) taking a long-term relational and partneringapproach for interacting with value creation partners; (3)offering and engaging value creation partners in co-production activities; (4) developing and leveraging oper-ant resources (capabilities) among value creation part-ners; and (5) aligning and coordinating value networkwide processes and service flows. Accordingly, we sug-gest that these five aspects should represent the funda-mental dimensions of an S-D strategic orientation that isconsistent with S-D logic.

STRATEGIC ORIENTATIONS IN LIGHT OF

THE S-D LOGIC

Both the marketing and management literature pro-vide various examples of strategic orientations andtypologies (see, e.g., Gatignon and Xuereb 1997; Lumpkinand Dess 1996; Miles and Snow 1978; Noble, Sinha, andKumar 2002; Porter 1980). Despite conceptual differ-ences, a common characteristic of strategic orientations isthat they are assumed to play an essential role in thepursuit of competitive advantage as they are concernedwith the decisions that businesses make to achieve acompetitive edge (e.g., Slater, Olson, and Hult 2006). Assuch, they can be seen as informing and directing a firm’sstrategy-making activities while guiding interactions withthe marketplace (Hunt and Morgan 1995; Noble et al.2002). As an essential contribution of the marketingdiscipline to the strategy dialogue (Hult, Ketchen, Jr., andSlater 2005; Hunt and Lambe 2000), market orientation(Kohli and Jaworski 1990; Narver and Slater 1990) hasalso been considered a strategic orientation (Hunt andMorgan 1995; Zheng, Zhou, Yim, and Tse 2005). Accord-ingly, market orientation is assumed to guide and affectstrategy selection. However, our in-depth literature re-view shows that the existing marketing and managementliterature is lacking a strategic orientation that addressesthe above identified five strategic implications of the S-Dlogic at once.

Contemporary strategic orientations seem to be lim-ited in light of emerging academic thought and businesspractice. For instance, it has been shown that marketorientation does not sufficiently explain the variance infirm performance and that companies often adopt mul-tiple orientations at the same time (Fritz 1996; Gray,Matear, and Matheson 2002). Furthermore, Hunt andLambe (2000) observe that a “partnering orientation”


(p. 28) is missing in the current conceptualization ofmarket orientation, as is the notion of co-production. Bothare increasingly important for co-creating value and co-producing market offerings with customers and othervalue creation partners. On the other hand, the relevanceof collaboration, relationships, and pan-organization-widestrategies has long been promoted in, for example, theservice, relationship, and network literature (e.g., Achroland Kotler 1999; Anderson, Håkansson, and Johanson1994; Grönroos 2000; Gummesson 1995).

In summary, contemporary (strategic) orientations(see, e.g., Gatignon and Xuereb 1997; Helfert, Ritter, andWalter 2002; Kohli and Jaworski 1990; Lytle, Hom, andMokwa 1998; Zheng, Zhou et al. 2005) only addresscertain elements of the S-D logic and of a service-basedcompetition respectively. Furthermore, the applied no-tions of service in these orientations are typically groundedin traditional understandings of a goods-dominant (G-D)logic and, thus, do not respond to today’s marketingthought and practices which are mirrored in the S-D logic.

CONCEPTUALIZATION OF AN S-D


Based on this paucity in the literature in regards to astrategic orientation that is consistent with S-D logic, wewill present such a strategic approach that is grounded onthe fundamentals of competing through service. In thisstudy, we adopt a conceptualization of a strategic orienta-tion that concerns broad strategic choices as manifested inrespective behaviors/activities that enable an organiza-tion to achieve long-term success (see Gatignon andXuereb1997). That is, rather than focusing on an underly-ing relatively abstract philosophy or mindset, this paperemphasizes organizational behaviors, activities, and pro-cesses reflecting a strategic orientation that is in line withS-D logic. In view of empirical testing, our behavioralconceptualization reduces the difficulties of capturing adominant managerial logic as cognitive processes or sche-mata that management scholars have issued caution against

(see, e.g., Lampel and Shamsie 2000; Prahalad and Bettis1986). Based on Vargo and Lusch (2004) and the abovediscussion, we define an S-D strategic orientation as:

Organizational behaviors, activities, and processesconcerned with the joint creation of superior valueand service experiences with value network partners,while focusing on mutually beneficial relationships,operant resources, and coordinated service flows.

Grounded in the aforementioned strategic implica-tions derived from the S-D associated literature, we pro-pose five corresponding dimensions of an S-D orienta-tion. These are hypothesized to be correlated and to reflectthe complex notion of interactive value creation for supe-rior service outcomes. We label the dimensions as fol-lows: “Experience Orientation,” “Relational Orientation,”“Co-Production Orientation,” “Capabilities Orientation,”and “Process Flow Orientation.” These are depicted inFigure 1 and will be discussed and defined in the follow-ing.

Experience Orientation

As Flint (2006) puts it: “The S-D logic of marketingseems to be largely about the marketer’s role in helping[individual] customers to create valuable experiences atall stages of the ‘consumption’ process including plan-ning, selection, purchase, consumption, and disposal. Thekey concept is servicing an experience” (p. 350). As such,experiences are the loci of value creation (Berry 1995;Prahalad and Ramaswamy 2003) where service partnersdirectly or indirectly interact through touch points such asgoods, employees or IT infrastructures. Organizations donot create value for and deliver value to customers butinstead assist customers in their own value creation andsolve problems (e.g., Grönroos 2006; Vargo and Lusch2004). The center of attention and, thus, the unit ofanalysis shifts from products to customers and with that toservice processes and individual value creation partners’experiences (e.g., Lusch et al. 2007; Rust and Thompson

FIGURE 1

Proposed Components of a Service-Dominant Strategic Orientation


2006). Accordingly, we propose that firms adopting an S-D strategic orientation have an “Experience Orientation,”which we define as:

Organizational behaviors, activities, and processesthat reflect a firm’s emphasis on understanding andenhancing value creation partner experiences overtime.

Relational Orientation

A second key aspect of the S-D logic refers to arelational approach when interacting with value creationpartners such as customers and suppliers (Vargo andLusch 2004). Customers are hence treated as partners andoperant resources rather than targets (Lusch et al. 2007).In general, firms “[…] move away from transaction ori-ented marketing strategies and move toward relationshiporiented marketing strategies for enhanced performance”(Sheth and Sharma 1997, p. 91). The creation, mainte-nance, and cultivation of mutually beneficial interactionsand relationships with a long-term focus has been high-lighted in the context of the S-D logic (Bolton et al. 2004;Vargo and Lusch 2004) and in the marketing and manage-ment literature in general (e.g., Grönroos 2000;Gummesson 1995; Pine II, Peppers, and Rogers 1995).However, emphasizing a relational approach does notmean having deep/strong relationships with every cus-tomer, since this might not be desired or beneficial for oneof the involved relationship parties (e.g., Grönroos 1997).Under this condition of mutual interest, firms implement-ing an S-D strategic orientation have a “Relational Orien-tation,” which we define as:

Organizational behaviors, activities, and processesthat reflect a firm’s approach in creating and main-taining relational exchanges with value creationpartners over time.

Co-Production Orientation

In line with S-D logic, organizations increasinglyencourage and engage customers and other value creationpartners in co-production activities. That is, they em-power customers and strive to maximize the involvementof value creation partners in the customization of serviceoutcomes (Lusch et al. 2007). For example, they mightopen up their service production processes during theideation, innovation, production, and/or distribution phaseof the core of the market offering. However, some cus-tomers might not necessarily want to be highly involvedin service processes (see, e.g., Kalaignanam andVaradarajan 2006) as this participation might increasetheir perceived costs. Therefore, competing strategicallythrough service entails offering “service co-productionopportunities and resources consistent with the customer’sdesired level of involvement [. . .]” (Lusch et al. 2007,

p. 8). Technology plays a critical role in facilitating co-production activities and managing the trade-off betweenoffering an individualized service and the respective costsassociated of it (Lusch et al. 2007; Rust and Espinoza2006; Rust and Thompson 2006). We define a “Co-Production Orientation” as:

Organizational behaviors, activities, and processesthat reflect a firm’s enabling of value creation part-ners to participate in co-constructing market offer-ings.

Capabilities Orientation

“A service-centered view identifies operant resources,especially higher-order, core competences, as the key toobtaining competitive advantage” (Vargo and Lusch 2004,p. 12). The relevance of such effect-driving core compe-tencies and dynamic capabilities for achieving competi-tive edge has been emphasized in the past (e.g., Day 1994;Prahalad and Hamel 1990; Teece, Pisano, and Shuen1997). Yet, while the focus previously has often been ondeveloping, as well as sharing core competencies andcapabilities across departments or divisions (e.g., Narverand Slater 1990), firms with an S-D strategic orientationaim to cultivate and leverage these within the valuenetwork. In a value network environment that is increas-ingly characterized by mutual dependencies, what a firmis able to offer and the outcome of a service does not onlydepend on one’s own operant resources but is a functionof available and useable capabilities of the value creationpartners (e.g., Håkansson and Ford 2002; Lusch et al.2007). Internal and external operant resources are thusdeployed for superior service provision. We define the“Capabilities Orientation” as:

Organizational behaviors, activities, and processesthat reflect a firm’s sharing and leveraging of oper-ant resources with value creation partners.

Process Flow Orientation

Under S-D logic, the central elements of marketingare processes and service flows rather than units of output(Vargo and Lusch 2004). Accordingly, market offeringsbecome processes of providing benefit in business realityrather than finished products. These value creating pro-cesses can be viewed within a value creating network,where customers are an integral part of the system (e.g.,Normann and Ramírez 1993). In such a connected envi-ronment, organizations move away from unidirectional,sequential value chains to multidirectional, synchronic,interactive, and dynamic value networks (Lusch et al.2007; Prahalad and Ramaswamy 2004). However, wheremore than one party is involved in service provisionprocesses, these need to be coordinated and synchronizedfor smooth operating value systems and process flows that


facilitate flexible service on demand (see, e.g., Flint andMentzer 2006; Lambert and García-Dastugue 2006). Theconstant fine-tuning, streamlining, and re-configurationof the value network supports the attainment and mainte-nance of efficient and effective end-to-end service pro-cesses (Normann and Ramírez 1993). Hence, we define a“Process Flow Orientation” as:

Organizational behaviors, activities, and processesthat reflect a firm’s alignment and coordination ofservice provision processes among value creationpartners.

POTENTIAL BENEFITS OF AN S-D


The five dimensions function as cumulative proxyfor an overall S-D strategic orientation. Since the compo-nents of an S-D orientation as such are seen as being ofstrategic nature and relevance, the superior implementa-tion and execution of this strategic orientation relative tocompetitors should support the achievement of competi-tive advantage. Figure 2 accordingly highlights exem-plary activities reflecting the underlying dimensions of anS-D strategic orientation and summarizes the potentialbenefits and outcomes. These outcomes are groupedaround: (1) improved service experiences; (2) improvedservice interactions; (3) improved service offerings; (4)improved service potential; and (5) improved serviceflows. It is assumed that all dimensions contribute to atleast one or a multiple of these expected potential im-provements.

In general, S-D logic is grounded in and largelyconsistent with the resource-based view (RBV) and theresource-advantage (R-A) theory. The two perspectivesbasically maintain that resources influence performanceas they are not quantitatively and qualitatively equallydistributed or accessible among firms. Idiosyncratic anddifficult-to-imitate resources thus create heterogeneousresource pools among firms that provide an importantfoundation for superior value creation with value networkpartners, relative to competitors (e.g., Barney 1991; Dasand Teng 2000). The R-A theory complements the RBV,for instance, by assuming not only resource but alsodemand heterogeneity from the customer’s perspective(Hunt 2002). In line with these frameworks, organizationsfocus on core competencies and dynamic capabilities,which are commonly seen as higher-order resources,rooted in casually ambiguous, idiosyncratic, and tacitroutines/processes (e.g., Hunt 2002; Prahalad and Hamel1990; Teece et al. 1997). In turn, this emphasis supportsthe superior co-creation of subjective value and contrib-utes to the achievement of competitive advantage (Huntand Madhavaram 2006). In S-D logic, the effect-prompt-ing operant resources such as knowledge, receive specialconsideration as they are seen to be the critical source ofcompetitive advantage (Lusch et al. 2007; Vargo andLusch 2004). Thus, the RBV and R-A theory serve astheoretical frameworks for the following discussion.

The first potential benefit of the implementation of anS-D strategic orientation relates to improved service ex-periences. Lusch et al. (2007) state that “[c]ompetitiveadvantage is a function of how one firm applies its operant

FIGURE 2

Dimensions, Exemplary Activity Foci, and Potential Benefits of an S-D Strategic Orientation


resources to meet the needs of the customer [. . .]” (p. 8)relative to competitors. As experiences result from director indirect interaction and are the loci of value creation(Berry 1995; Prahalad and Ramaswamy 2003), a firm’sfocus on superior application of its critical resources and/or the provision of superior resources that customersinteract with during their own value creation processes,contributes to better value creation partner experiences.On the other hand, a firm’s service provision processes arehere seen as core competencies themselves that assistcustomers’ value creation and problem solving. The im-provement of these processes and the involved resourcesare expected to lead to better value creation partnerexperiences. In the long-run, higher customer satisfac-tion, and in turn higher loyalty, will result in positiveoutcomes that support business success.

The second potential benefit of an S-D strategicorientation concerns improved relationships with valuecreation partners. As such, long-term firm relationshipswith customers and suppliers are considered to be essen-tial resources of a firm. Treating customers as long-termpartners and, thus, as effect-triggering and dynamic oper-ant rather than passive operand resources (Vargo andLusch 2004), while fostering two-way communicationand information symmetry, supports the development ofbetter relationships among the value creation partners.Long-term relationships foster trust which facilitates de-riving deeper insights, mutual learning, and better transferof valuable knowledge (e.g., Pine II et al. 1995). Theestablishment of relational rather than transactional inter-actions with partners who prefer this type of exchangethus promotes the cultivation of critical operant resourcesfor the firm and among its value creation partners. Fur-thermore, good relationships may also contribute to betterexperiences as they can generate positive emotional ben-efits and thus influence perceived value. Again, moresatisfied value creation partners, and in turn loyalty, leadto positive outcomes that support a firm’s success.

The third potential benefit of an S-D strategic orien-tation addresses improved or more effective market offer-ings. The involvement of customers in different stages ofthe service chain, such as during ideation or production ofa good, allows them to generate more realistic expecta-tions about the outcome as well as directly influence theactual outcome itself. Via this higher personalizationcustomers can ensure that they get what they really want,and therefore achieve a closer fit with their unique prefer-ences (Kalaignanam and Varadarajan 2006; Lusch et al.2007; Wind and Rangaswamy 2001). However, it iscritical to “[d]esign a mutually beneficial operating rolefor customers” (Frei 2006, p. 100) at their desired level ofinvolvement (Lusch et al. 2007). Through co-productionactivities, organizations can thus capitalize on value cre-ation partner competencies (e.g., Prahalad andRamaswamy 2000) and even lower labor costs (Bendapudi

and Leone 2003; Mills and Morris 1986), leading toinnovation and financial gain of the firm.

The fourth potential benefit of an S-D strategic orien-tation pertains to improved service potential. S-D orientedfirms focus on their own core competencies while sharingand leveraging operant resources among value creationpartners (Vargo and Lusch 2004). In fact, competenciesand resources are the basis for superior service provisionand with higher access to valuable operant resourcesorganizations can improve the service potential of thevalue network (Lusch et al. 2007). Furthermore, by pro-viding operant resources to value creation partners, firmscan leverage these into beneficial outcomes in the spirit ofmutual service provision. Organizations that foster thesuperior cultivation, sharing and application of a valuenetwork’s idiosyncratic and difficult-to-copy capabilitieswill take a significant step toward competitive advantage(Lusch et al. 2007). Service-dominant organizations thusunderstand their own competencies and operant resourcesin terms of value creation as well innovation potential thatcan be enhanced throughout a value creation network (seealso Golfetto and Gibbert 2006).

The fifth potential benefit of an S-D strategic orien-tation tangents improved service flows in the value net-work. As such, the ability to conceive the entire valuecreating system and make it work is a critical aspect ofnetworked service processes (Duncan and Moriarty 2006;Normann and Ramírez 1993). An implemented processflow orientation supports effective and efficient serviceprocesses. On the one hand, “lean” service processes thatare coordinated and aligned consequently waste fewerresources, which in turn reduce the costs of the involvedvalue network partners. On the other hand, customers arebuying and participate evermore intensely in service flows(Lusch et al. 2006). Since customers are part of theseservice provision processes, smooth and efficient work-ing processes contribute to better experiences and, thus,higher satisfaction. Furthermore, improved service flowsrepresent an important leverage for optimized resourcedeployment among value creation partners.

In summary, the aforementioned examples highlightpotential benefits of implementing an S-D strategic orien-tation. While S-D logic rationalizes exchanges, markets,organizations, and competition based on service (Luschet al. 2007), it directs managerial focus toward operantresources such as competencies, relational collaboration,and alignment of customers’ and other value-networkpartners’ value-creating processes. Accordingly, superiorservice flows, service potential, service offerings, servicerelationships, and service experiences are expected. As acombined effect of the five dimensions of the S-D strate-gic orientation, firms strengthen their competitive posi-tion, which is supported in this context by the RBV and R-A theory.


CONCLUSION

The service-dominant logic of marketing focuses on“interactivity, integration, customization, and coproduction[. . .]” while highlighting the centricity of customers andrelationships (Vargo and Lusch 2004, p. 11). These as-pects have been considered, synthesized, and conceptual-ized here into a coherent strategic orientation that fostersservice-based competition. Accordingly, we fill the iden-tified gap of the lack of an existing strategic orientationthat corresponds to the identified managerial implicationsof the S-D logic. As such, we propose and specify fivedimensions of an S-D strategic orientation grounded inthe S-D literature. We also provide insights into thepotential benefits of an implementation of the S-D strate-gic orientation. Thereby, we focus on five beneficialoutcomes that improve a firm’s overall competitive posi-tion, which is largely grounded in the RBV and R-Atheory. These lay the foundation for further knowledgeadvancement. For instance, through the empirical devel-

opment of a respective scale and testing of potentialrelationships such as its impact on customer satisfactionand organizational performance.

An S-D strategic orientation reflects a different ap-proach in comparison to the traditional differentiationstrategies based on added service (Lusch et al. 2007). Assuch, competing through service by adopting an S-Dstrategic orientation focuses strategy on superior pro-cesses of providing benefit and, thus, on joint outcomesrather than finished output. Service is not just added to themarket offering to increase the value potential; rather, it isthe critical process of benefit provision in concert withvalue creation partners that is the foundation for service-based competition (Vargo and Lusch 2004). As experi-ences, relationships, interactions, collaborations, and in-tegrations within these processes in today’s networkedbusiness arena gain significance, an S-D strategic orien-tation provides a response to managerial challenges in thepursuit of competitive advantage.

REFERENCES

Achrol, Ravi S. and Philip Kotler (1999), “Marketing inthe Network Economy,” Journal of Marketing, 63(4), 146–63.

Anderson, James C., Håkan Håkansson, and Jan Johanson(1994), “Dyadic Business Relationships Within aBusiness Network Context,” Journal of Marketing,58 (4), 1–15.

Ballantyne, David and Richard J. Varey (2006), “CreatingValue-in-Use Through Marketing Interaction: TheExchange Logic of Relating, Communicating, andKnowing,” Marketing Theory, 6 (3), 335–48.

Barney, Jay B. (1991), “Firm Resources and SustainedCompetitive Advantage,” Journal of Management,17 (1), 99–121.

Bendapudi, Neeli and Robert P. Leone (2003), “Psycho-logical Implications of Customer Participation in Co-Production,” Journal of Marketing, 67 (1), 14–28.

Berry, Leonard L. (1995), “Relationship Marketing ofServices–Growing Interest, Emerging Perspectives,”Journal of the Academy of Marketing Science, 23 (4),236–45.

Berry, Leonard L., Venkatesh Shankar, Janet TurnerParish, Susan Cadwallader, and Thomas Dotzel(2006), “Creating New Markets through Service In-novation,” MIT Sloan Management Review, 47 (2),56–63.

Bolton, Ruth N., George S. Day, John Deighton, DasNarayandas, Evert Gummesson, Shelby D. Hunt,C.K. Prahalad, Roland T. Rust, and Steven M. Shugan(2004), “Invited Commentaries on ‘Evolving to aNew Dominant Logic for Marketing,’” Journal ofMarketing, 68 (1), 18–27.

Cares, Jeff (2006), “Battle of the Networks,” HarvardBusiness Review, 84 (2), 40–41.

Court, David C., Thomas D. French, and Trond R. Knudsen(2006), “The Proliferation Challenge,” in Profitingfrom Proliferation, Allen P. Webb, ed. New York:McKinsey & Company.

Das, T.K. and Bing-Sheng Teng (2000), “A Resource-Based Theory of Strategic Alliances,” Journal ofManagement, 26 (1), 31–62.

Day, George S. (1992), “Marketing’s Contribution to theStrategy Dialogue,” Journal of the Academy of Mar-keting Science, 20 (4), 323–29.

____________ (1994), “The Capabilities of Market-Driven Organizations,” Journal of Marketing, 58 (4),37–52.

Duncan, Tom and Sandra Moriarty (2006), “How Inte-grated Marketing Communication’s ‘Touchpoints’Can Operationalize the Service-Dominant Logic,” inThe Service-Dominant Logic of Marketing: Dialog,Debate, and Directions, Robert F. Lusch and StephenL. Vargo, eds. Armonk, NY: M.E. Sharpe.

Firat, A. Fuat and Alladi Venkatesh (1995), “LiberatoryPostmodernism and the Reenchantment of Consump-tion,” Journal of Consumer Research, 22 (3), 239–67.

Flint, Daniel J. (2006), “Innovation, Symbolic Interactionand Customer Valuing: Thoughts Stemming from aService-Dominant Logic of Marketing,” MarketingTheory, 6 (3), 349–62.

____________ and John T. Mentzer (2006), “Striving forIntegrated Value Chain Management Given a Ser-vice-Dominant Logic for Marketing,” in The Ser-vice-Dominant Logic of Marketing: Dialog, Debate,and Directions, Robert F. Lusch and Stephen L.


Vargo, eds. Armonk, NY: M.E. Sharpe.Frei, Frances X. (2006), “Breaking the Trade-Off: Be-

tween Efficiency and Service,” Harvard BusinessReview, 84 (11), 92–101.

Fréry, Frédéric (2006), “The Fundamental Dimensions ofStrategy,” MIT Sloan Management Review, 48 (1),71–75.

Fritz, Wolfgang (1996), “Market Orientation and Corpo-rate Success: Findings from Germany,” EuropeanJournal of Marketing, 30 (8), 59–74.

Gatignon, Hubert and Jean-Marc Xuereb (1997), “Strate-gic Orientation of the Firm and New Product Perfor-mance,” Journal of Marketing Research, 34 (1), 77–90.

Golfetto, Francesca and Michael Gibbert (2006), “Mar-keting Competencies and the Sources of CustomerValue in Business Markets,” Industrial MarketingManagement, 35 (8), 904–12.

Gray, Brendan J., Sheelagh Matear, and Philip K. Matheson(2002), “Improving Service Firm Performance,” Jour-nal of Services Marketing, 16 (3), 186–200.

Grönroos, Christian (2000), Service Management andMarketing. A Customer Relationship ManagementApproach. Chichester: John Wiley.

____________ (1997), “From Marketing Mix to Rela-tionship Marketing –Towards a Paradigm Shift inMarketing,” Management Decision, 35 (3/4), 322–39.

____________ (2006), “On Defining Marketing: Findinga New Roadmap for Marketing,” Marketing Theory,6 (4), 395–417.

Gummesson, Evert (1995), “Relationship Marketing: ItsRole in the Service Economy,” in UnderstandingServices Management, William J. Glynn and JamesG. Barnes, eds. New York: John Wiley & Sons.

Håkansson, Håkan and David Ford (2002), “How ShouldCompanies Interact in Business Networks?” Journalof Business Research, 55 (2), 133–39.

Helfert, Gabriele, Thomas Ritter, and Achim Walter(2002), “Redefining Market Orientation from a Rela-tionship Perspective: Theoretical Considerations andEmpirical Results,” European Journal of Marketing,36 (9/10), 1119–39.

Hult, G. Tomas M., David J. Ketchen, Jr., and Stanley F.Slater (2005), “Market Orientation and Performance:An Integration of Disparate Approaches,” StrategicManagement Journal, 26 (12), 1173–81.

Hunt, Shelby D. and Robert M. Morgan (1995), “TheComparative Advantage Theory of Competition,”Journal of Marketing, 59 (4), 1–15.

____________ and C. Jay Lambe (2000), “Marketing’sContribution to Business Strategy: Market Orienta-tion, Relationship Marketing and Resource-Advan-tage Theory,” International Journal of ManagementReviews, 2 (1), 17–43.

____________ (2002), Foundations of Marketing Theory:Toward a General Theory of Marketing. New York:

M.E. Sharpe.____________ and Sreedhar Madhavaram (2006), “The

Service-Dominant Logic of Marketing: TheoreticalFoundations, Pedagogy, and Resource-AdvantageTheory,” in The Service-Dominant Logic of Market-ing: Dialog, Debate, And Directions, R.F. Lusch andS.L. Vargo, eds. Armonk, NY: M.E. Sharpe.

Kalaignanam, Kartik and Rajan Varadarajan (2006), “Cus-tomers as Co-Producers: Implications for MarketingStrategy Effectiveness and Marketing OperationsEfficiency,” in The Service-Dominant Logic of Mar-keting: Dialog, Debate, and Directions, Robert F.Lusch and Stephen L. Vargo, eds. Armonk, NY: M.E.Sharpe.

Kohli, A.K. and B.J. Jaworski (1990), “Market Orienta-tion: The Construct, Research Propositions, and Mana-gerial Implications,” Journal of Marketing, 54 (1), 1–18.

Lambert, Douglas M. and Sebastián J. García-Dastugue(2006), “Cross-Functional Business Processes forthe Implementation of Service-Dominant Logic,” inThe Service-Dominant Logic of Marketing: Dialog,Debate, and Directions, Robert F. Lusch and StephenL. Vargo, eds. New York: M.E. Sharpe.

Lampel, Joseph and Jamal Shamsie (2000), “Probing theUnobtrusive Link: Dominant Logic and the Designof Joint Ventures at General Electric,” StrategicManagement Journal, 21 (5), 593–602.

Lumpkin, G.T. and Gregory G. Dess (1996), “Clarifyingthe Entrepreneurial Orientation Construct and Link-ing It to Performance,” Academy of ManagementReview, 21 (1), 135–72.

Lusch, Robert F. and Stephen L. Vargo (2006), “Service-Dominant Logic as a Foundation for a GeneralTheory,” in The Service-Dominant Logic of Market-ing: Dialog, Debate, and Directions, Robert F. Luschand Stephen L. Vargo, eds. Armonk, NY: M.E.Sharpe.

____________, ____________, and Alan J. Malter (2006),“Taking a Leadership Role in Global MarketingManagement,” Organizational Dynamics, 35 (3),264–78.

____________, ____________, and Matthew O’Brien(2007), “Competing Through Service: Insights fromService-Dominant Logic,” Journal of Retailing, 83(1), 5–18.

Lytle, Richard S., Peter W. Hom, and Michael P. Mokwa(1998), “SERV*OR: A Managerial Measure of Or-ganizational Service-Orientation,” Journal of Retail-ing, 74 (4), 455–89.

Miles, Raymond E. and Charles C. Snow (1978), Organi-zational Strategy, Structure, and Process. New York:MaGraw-Hill.

Mills, Peter K. and James H. Morris (1986), “Clients as‘Partial Employees of Service Organizations: RoleDevelopment in Client Participation,” Academy ofManagement Review, 11 (4), 726–35.


Narver, John C. and Stanley F. Slater (1990), “The Effectof a Market Orientation on Business Profitability,”Journal of Marketing, 54 (4), 20–36.

Noble, Charles H., Rajiv K. Sinha, and Ajith Kumar(2002), “Market Orientation and Alternative Strate-gic Orientations: A Longitudinal Assessment of Per-formance Implications,” Journal of Marketing, 66(4), 25–39.

Normann, Richard and Rafael Ramírez (1993), “FromValue Chain to Value Constellation: Designing Inter-active Strategy,” Harvard Business Review, 71 (4).

Oliver, Richard W., Roland T. Rust, and Sajeev Varki(1998), “Real-Time Marketing,” Marketing Man-agement, 7 (4), 28–37.

Pine II, B. Joseph, Don Peppers, and Martha Rogers(1995), “Do You Want to Keep Your CustomersForever?” Harvard Business Review, 73 (2), 103–14.

Pires, Guilherme D., John Stanton, and Paulo Rita (2006),“The Internet, Consumer Empowerment, and Mar-keting Strategies,” European Journal of Marketing,40 (9/10), 936–49.

Porter, Michael E. (1980), Competitive Strategy. NewYork: Free Press.

____________ (1991), “Towards a Dynamic Theory ofStrategy,” Strategic Management Journal, 12 (8),95–117.

Prahalad, C.K. and Richard A. Bettis (1986), “The Domi-nant Logic: A New Linkage Between Diversity andPerformance,” Strategic Management Journal, 7 (6),485–501.

____________ and Gary Hamel (1990), “The Core Com-petence of the Corporation,” Harvard Business Re-view, 68 (3).

____________ and Venkat Ramaswamy (2000), “Co-Opting Customer Competence,” Harvard BusinessReview, 78 (1), 79–87.

____________ (2003), “The New Frontier of ExperienceInnovation,” MIT Sloan Management Review, 44 (4),12–18.

____________ (2004), The Future of Competition: Co-Creating Unique Value with Customers. Boston:Harvard Business School Press.

Rust, Roland T. and Francine Espinoza (2006), “HowTechnology Advances Influence Business Researchand Marketing Strategy,” Journal of Business Re-search, 59 (10/11), 1072–78.

____________ and Debora V. Thompson (2006), “How

Does Marketing Strategy Change in a Service-BasedWorld?” in The Service-Dominant Logic of Market-ing: Dialog, Debate, and Directions, Robert F. Luschand Stephen L. Vargo, eds. Armonk, NY: M.E.Sharpe.

Sheth, Jagdish N. and Arun Sharma (1997), “SupplierRelationships: Emerging Issues and Challenges,”Industrial Marketing Management, 26 (2), 91–100.

Slater, Stanley F., Eric M. Olson, and G. Tomas M. Hult(2006), “The Moderating Influence of Strategic Ori-entation on the Strategy Formation Capability-Per-formance Relationship,” Strategic Management Jour-nal, 27 (12), 1221–31.

Teece, David J., Gary Pisano, and Amy Shuen (1997),“Dynamic Capabilities and Strategic Management,”Strategic Management Journal, 18 (7), 509–33.

Vargo, Stephen L. and Robert F. Lusch (2004), “Evolvingto a New Dominant Logic for Marketing,” Journal ofMarketing, 68 (1), 1–17.

____________ (2006), “Service-Dominant Logic: WhatIt Is, What It Is Not, What It Might Be,” in TheService-Dominant Logic of Marketing: Dialog, De-bate, and Directions, Robert F. Lusch and Stephen L.Vargo, eds. Armonk, NY: M.E. Sharpe.

____________ (in press), “Paradigms, Pluralisms, andPeripheries: On the Assessment of S-D Logic,”Australasian Marketing Journal.

Webster, Jr., E. Frederick (2006), “Foreword,” in TheService-Dominant Logic of Marketing: Dialog, De-bate, and Directions, Robert F. Lusch and Stephen L.Vargo, eds. Armonk, NY: M.E. Sharpe.

Wernerfelt, Birger (1984), “The Resource-Based View ofthe Firm,” Strategic Management Journal, 5 (2),171–81.

Wind, Jerry and Arvind Rangaswamy (2001), “Custom-erization: The Next Revolution in Mass Customiza-tion,” Journal of Interactive Marketing, 15 (1), 13–32.

Winklhofer, Heidi, Roger A. Palmer, and Roderick J.Brodie (in press), “Researching the Service-Domi-nant Logic-Normative Perspective versus Practice,”Australasian Marketing Journal.

Zheng Zhou, Kevin, Chi Kin (Bennett) Yim, and David K.Tse (2005), “The Effects of Strategic Orientations onTechnology- and Market-Based Breakthrough Inno-vations,” Journal of Marketing, 69 (2), 42–60.

For further information contact:Ingo Oswald Karpen

The University of Melbourne4th Floor Alan Gilbert Building

Victoria 3010, AustraliaPhone: +61.383441921 ♦ Fax: +61.393481921



PERFORMANCE IMPLICATIONS OF DEVELOPING CUSTOMER-

LINKING CAPABILITIES

Kevin Trainor, Kent State University, KentAdam Rapp, Kent State University, Kent

Raj Agnihotri, Kent State University, KentNiels Schillewaert, Vlerick School of Management, Belgium

SUMMARY

This research adds to the growing literature on dy-namic capabilities by introducing a model that describeshow firms convert existing marketing capabilities andresources into the dynamic capability of customer linking(i.e., connecting a firm’s business processes to customersthereby leading to durable relationships). To achieve thisgoal, the current study presents and empirically tests amodel of how market-sensing capabilities are leveragedwith the intangible, complementary business resources oftop management commitment, capability strategy, andworkforce flexibility to improve organization perfor-mance. Specifically, the focus is on the creation of cus-tomer-linking capabilities via the implementation of e-business technology. With the dynamic capabilities viewserving as the primary conceptual lens, literature frommarketing strategy and management strategy is used topropose and test a conceptual model of how advantageouscapabilities can be identified, developed and measuredwithin organizations.

Day’s (1994) seminal work posits that a capability,which he defines as the bundling of skills and abilities thatfacilitate the execution of business processes, can developinto a distinctive capability that is difficult to match byrival firms. These distinctive capabilities are what providesuperior customer value in Day’s view of the market-driven organization. In parallel with the marketingliterature’s attention toward the strategic importance ofcapabilities, Teece, Pisano, and Shuen (1997) introducethe dynamic capabilities view in the strategic manage-ment literature. This perspective, which highlights howsome firms develop and sustain competitive advantagesand superior profitability, extends the resource-basedview (RBV) of the firm. The dynamic capabilities viewemphasizes the key management role of appropriatelyadapting, integrating, and reshaping organizational skillsand resources as well as internal and external functionalcompetences.

From a marketing strategy perspective, the dynamiccapabilities view suggests that internal capabilities, suchas the outside-in capabilities of market-sensing, must becombined and integrated with other complementary re-sources or capabilities to generate and sustain a competi-

tive advantage. In our investigation, we focus on thecomplementary capability of customer linking and exam-ine its critical dimensions that lead to performance. Extantliterature suggests that the coordination of activities withcustomers is one of the skills, abilities and processes thatmust be mastered to develop customer-linking capabili-ties. From this we identify the deployment of marketing-related information technology as a specific instance of anactivitycoordinating mechanism that links a company toits customers.

Our conceptual model is informed by an informationtechnology and an innovation capability framework fromthe management literature in addition to the rich body ofmarket orientation literature. Our model suggests thatmarket orientation (viewed as a specific instance of amarket sensing capability) is a critical antecedent to thedevelopment of customer-linking capabilities. Further-more, the model explains how the business resources of e-business strategy are integrated with the complementaryhuman resources of top management team commitmentand a flexible culture to form a new, distinctive capability.It is through the successful development of this newcapability that firms can enhance their performance.

Structural equation modeling was used to analyzesurvey data from 584 top management team members ofBelgium-based firms. The results support our positionthat market orientation is an important driver in thesuccessful deployment and subsequent performance ofnew capabilities that connect a firm’s business processesto its customers. Specifically, we find that when marketorientation is coupled with managerial commitment, anoverall strategy, and a supportive and flexible workforcethat embraces the new capability, firm performance canbe positively influenced.

From a theoretical perspective, this study providesevidence that taking a capabilities-based approach towardthe examination of a market-oriented philosophy canreveal new insight into how firms develop, integrate anddeploy specific capabilities to create sustained competi-tive advantage. To our knowledge there is currently nomodel to describe how market-driven firms identify,develop and measure the marketing capability of cus-tomer linking. Additionally, by viewing marketing capa-


bilities as a form of innovation, as was done in this study,researchers can leverage a rich body of the innovation andtechnology literature found in the management discipline.While a specific instance of a capability is selected for this

study (i.e., customer linking), it is expected that this modelcan be useful for exploring how firms identify and de-velop other marketing-related capabilities. Referencesavailable upon request.

For further information contact:Kevin Trainor

Department of MarketingCollege of Business Administration

Kent State UniversityP.O. Box 5190

Kent, OH 44242Phone: 330.672.1271



WHEN CAN WE MEASURE WILLINGNESS TO PAY DIRECTLY? AN

EMPIRICAL STUDY ON THE ROLE OF CONSUMERS’

INVOLVEMENT IN THE DIRECT ELICITATION

OF RESERVATION PRICES

Klaus Matthias Miller, University of Bern, SwitzerlandHarley Krohmer, University of Bern, SwitzerlandReto Hofstetter, University of Bern, Switzerland

SUMMARY

A key challenge in applied market research is theprecise identification of consumers’ willingness to pay(WTP), which denotes the maximum amount of money aconsumer is willing to pay for a given quantity of a good.Researchers usually focus on consumer’s hypotheticalwillingness to pay (HWTP; i.e., no obligation of respon-dents to buy the product), which may deviate fromconsumer’s actual willingness to pay (AWTP). This de-viation is known as the “hypothetical bias.” Two differentapproaches are used to elicit consumers’ HWTP: theindirect and the direct approach. If the researcher appliesthe indirect approach (e.g., conjoint analysis), then hecalculates a consumer’s HWTP based on his or her pref-erence in alternative product profiles, which vary in priceand other product attributes. For the direct approach, theresearcher asks consumers directly to explicitly state theirHWTP. In direct WTP measurement techniques and pro-cedures of applied market research usually three differentquestion formats are used. For the open-ended (OE)question format, respondents are asked to state theirmaximum WTP for a specific product in an open-endedfield. For the dichotomous choice (DC) question format,respondents are asked whether they will accept or reject astated price for a specific product. For the payment card(PC) question format, respondents are asked to select oneof several proposed prices as their maximum WTP for aspecific product.

The number of studies which compare the differentdirect question formats (OE, DC, and PC) with regard tothe hypothetical bias is limited. Specifically, we know ofonly four studies on this issue. These studies empiricallycompare the DC and OE formats for hypothetical bias.The results of those four studies do not provide a clearpicture of the superiority of the DC over the OE questionformat. To our knowledge, the PC question format has notyet been empirically compared with the other two directquestion formats. The research deficits of the literature onthe direct approach to measure consumers’ WTP can besummarized as follows: prior empirical studies originatein environmental economics (the contingent-valuationliterature) and measure WTP for public goods such as air

quality, groundwater, or forest wildlife resources. In thecontext of such public goods the social desirability ofanswers may play a certain role and hence have an impacton the magnitude of the hypothetical bias. There is a lackof research on the direct approach to elicit consumers’WTP in the context of private (i.e., consumer) goods.Measuring WTP for private goods may be different (e.g.,due to better product knowledge of respondents or lack ofsocial desirability of responses). Another research deficitrefers to the observation that prior empirical studies arerather exploratory and do not provide strong conceptualarguments concerning the existence and antecedents ofhypothetical bias. Furthermore, the validity of alternativedirect question formats is empirically compared withoutdiscussing the effect of contextual factors such as con-sumers’ product-related involvement on the magnitude ofhypothetical bias. This may explain some of the conflict-ing findings with regard to the supremacy of alternativequestion formats. By taking an exploratory empiricalfocus and by neglecting the role of contextual factors,prior studies tend to miss the core issue relevant forapplied market research, that is, under what circum-stances is the direct approach a valid instrument to mea-sure consumers’ WTP? Based on these research deficits,we will answer the following research questions: (1) Doalternative direct question formats differ in their hypo-thetical bias? (2) How does the level of respondents’product-related involvement affect the validity of thedirect approach?

In order to answer our first research question, wecompared the bias of alternative direct question formats:open-ended (OE), dichotomous-choice (DC) and pay-ment-card (PC). We were able to show that all threequestion formats lead to hypothetical bias, which wedefined as the difference between hypothetical willing-ness to pay (HWTP), and actual willingness to pay(AWTP). As hypothesized, we did not find significantdifferences in hypothetical bias in paired comparisons ofeither the PC and OE question formats or the DC and OEquestion formats. We found a significant difference inhypothetical bias only in paired comparison of the DC andPC question formats. However, our results indicate thatthe somewhat larger bias with the DC question format


may be the consequence of what method is used tomeasure AWTP in a within-subject design. In contrast,HWTP values across the three question formats do notdiffer significantly. These results show that in a generalcontext (i.e., not taking contingency variables into consid-eration), the various direct question formats can generatesimilar results for HWTP and hypothetical bias. Hence,we find no generally superior direct question format formeasuring WTP. Our second research question addressesthe issue that seems to have been neglected in priorresearch, that is, under what circumstances is the directapproach a valid instrument to measure consumers’ WTP?After a conceptual discussion of possible drivers of hypo-thetical bias, we focused on the role of respondents’product-related involvement. We found a significant nega-tive linear relation between hypothetical bias and theproduct-related involvement of the consumer, which mayeven eliminate hypothetical bias in the case of extremelyhigh involvement. In other words, when consumers showhigh product-related involvement, HWTP statements seemto be more accurate than are HWTP statements fromconsumers with low product-related involvement.

Our study suggests several opportunities for furtherresearch. For example, future research could empiricallyexamine the causes of hypothetical bias in a more compre-

hensive way. An empirical examination of the underlyingmotivation and ability of the respondents to disclose theirtrue WTP may help managers to better interpret estimatedHWTP values. Our research has several implications formanagement. First, our results indicate that the directapproach, which only measures consumers’ HWTP, mayresult in WTP estimates that are significantly differentfrom AWTP. Hence, when interpreting HWTP estima-tions that were generated with a direct approach, manag-ers should bear in mind that the estimated HWTP may beconsiderably overestimated. A failure to adjust for hypo-thetical bias might lead managers to substantially over-price their products. Second, the direct approach leads tomore valid HWTP estimates (smaller hypothetical bias)when product-related involvement of consumers is high.With consumers that show high product-related involve-ment, hypothetical bias disappears. This finding is rel-evant for applied market research: In the case of highlyinvolved respondents and/or in the case of high involve-ment consumer goods, the direct approach to measureconsumers’ WTP can provide valid results in a cost- andtime-efficient way. In the case of low or medium involve-ment consumer goods, researchers are well-advised toadditionally elicit product-related involvement of theirrespondents in order to better interpret the validity of theresulting WTP data.

For further information contact:Klaus Matthias Miller

University of BernEngehaldenstrasse 4

Bern, 3012Switzerland

Phone: 0041.31.631.4550Fax: 0041.31.631.8032



AN INVESTIGATION OF THE RELATIONSHIP BETWEEN THE

NUMBER OF RESPONSE CATEGORIES AND

SCALE SENSITIVITY

Richard J. Fox, University of Georgia, AthensSunil H. Contractor, Towson University, Towson

SUMMARY

Category rating scales are frequently used in socialscience research as well as in opinion polling to measureattitude or reaction to a stimulus. Hence, the issue ofwhether the number of response categories included in thecategory rating scale affects scale sensitivity, i.e., theability to statistically separate mean ratings for differentstimuli, is important to researchers and practitioners alike.There are two conflicting views regarding the relationshipbetween the number of categories in an itemized ratingscale and scale sensitivity (Parasuraman et al. 2004). Onone hand, increasing the number of scale categories argu-ably enhances the precision of a scale by providing moreoptions for respondents, and thus increases sensitivity(Krosnick and Fabrigar 1997). On the other hand, thegreater the number of categories included in a scale, thehigher the variance or noise in the measurement, and theseincreasing the number of scale categories beyond a certainpoint may actually decrease sensitivity. The paper de-scribes the results of an experiment that was conducted toevaluate and compare the sensitivities of commonly usedcategory rating scales with different number of catego-ries.

The experiment involved requesting consumers, in-tercepted and recruited in shopping malls, to evaluatethree sample products in each of three very distinct prod-uct categories, orange juice, facial tissue, and writinginstruments (pens). The stimuli within each product cat-egory were chosen to represent the range of quality in thecategory so that differences in average acceptability couldbe expected. Subjects were asked to rate each sample foroverall acceptability using one of the four category ratingscales included in the study. The rating task was thenrepeated using the same stimuli (this was not revealed tothe participant), but using a different rating scale. Eigh-teen respondents were randomly assigned to each of thetwenty-four groups, 6 pairs of scales (formed from thefour scales studied) x 2 orders of scale presentation (e.g.,

5-pt. followed by 6-pt. or vice versa) x 2 product categorysequences (orange juice, facial tissue, pens or pens, facialtissue, orange juice) resulting in a total sample of 432responds.

Statistical analyses indicated that the ratings data,within each product category, could be aggregated acrossexperimented conditions, e.g., whether the scale was usedfirst or second, for each of the four scales. The ratings data(aggregated) for each product category and each of thefour scales were analyzed using a repeated measuresmodel (3 measures, one scale rating for each of threestimuli, for each subject). For each of the 12 analyses (4scales and 3 product categories), the base size for thestatistical analyses ranged from 207 to 220.

The results indicate that increasing the number ofscale points/response categories do not increase sensitiv-ity. The Eta² measure of effect size was quite consistentacross scales within each product category. In particular,in the context of labeled category rating scales, the 5-pointand 6-point scales appear to be at least as sensitive as a 9-point scale. In fact, there were several instances where the5-point and 6-point scales indicated significant differ-ences between stimuli means which were not found to besignificantly different using the 9-point and 10-pointscales. It is also noted that the scales with the low numberof response categories take less time to administer than thescales with the large number of categories and are pre-ferred by subjects. The longer the survey, the more expen-sive the data collection and the more likely a subject willterminate an interview before completion. Hence, from apractical point of view as well, the scales with 5 and 6response categories are preferred.

In conclusion, scale sensitivity can be added to the listof constructs, including reliability and validity, not neces-sarily enhanced by increasing the number of responseoptions.


REFERENCES

Krosnick, Jon A. and Leandre R. Fabrigar (1997), “De-signing Rating Scales for Effective Measurement inSurveys,” in Survey Measurement and Process Qual-

ity, Biemer Lyberg, Collins, de Leeuw, Dippo,Schwarz, Trewin, eds. John Wiley & Sons, Inc.

Parasuraman, A., Dhruv Grewal, and R. Krishnan (2004),Marketing Research. Boston, MA: Houghton MifflinCompany.

For further information contact:Richard J. Fox

Terry College of BusinessUniversity of Georgia

Athens, GA 30602Phone: 706.542.3761


Sunil H. ContractorTowson University

8000 York RoadToswon, MD 21252Phone: 410.704.335Fax: 410.704.3772



THE PUZZLE OF “ABOUT SIX” INDICATORS IN LATENT VARIABLES

Robert A. Ping, Wright State University, Dayton

ABSTRACT

The paper investigates the observation that latentvariables have a maximum of about six indicators inpublished substantive studies. An additional explanationfor this “puzzle of about six indicators” is offered, andseveral approaches to utilizing more than about six indi-cators for latent variables are explored. A suggestedadditional step in item judging is offered for theoreticalmodel tests with “weeded” measures used as latent vari-able indicators.

INTRODUCTION

Theoretical model tests in the Social Sciences fre-quently combine unobserved, or latent variables (LV’s),with hypothesized linkages among these variables (amodel), and observed (reflexive) indicators of these LV’s.Commenting on the number of indicators per LV inpublished model tests, authors have noted that the numberof indicators per LV seems to be about six (e.g., Andersonand Gerbing 1984; Bagozzi and Heatherton 1994; Gerbingand Anderson 1993; Ping 2004). This has several impli-cations. Cattell (1973) stated that the resulting latentvariable measures tend to be “bloated specific” instancesof their target construct which tap only one facet of thatconstruct. Larger, well-established measures, developedbefore structural equation analysis became popular, havelargely disappeared from published research. When theydo appear in structural equation analysis, they are fre-quently “shadows of their former selves” because ofextensive item deletion to attain model-to-data fit (e.g.,Comer, Machleit, and Lagace 1989). Because of theimplications for face or content validity of “about 6-item”LV’s, this paper explores the “puzzle of about six indica-tors.” An additional explanation for this apparent puzzleis proposed, and previously proposed approaches to avoid-ing this limitation are explored. The results suggested thatseveral proposed approaches might not always perform.Along the way, difficulties associated with “weeded”subsets of items (i.e., about 6-item subsets that result fromdeleting items from a measure to obtain measurementmodel-to-data fit), such as structural coefficient indeter-minacy, are examined.

ABOUT SIX INDICATORS?

The puzzle of about six indicators is apparently theresult of chronic measurement model-to-data fit1 difficul-ties (i.e., internal inconsistency,2 see Anderson and Gerbing1982). Gerbing and Anderson (1993) commented that

“. . . fit indices indicated less fit as the . . . number ofindicators per factor, increased. . . .” They suggested that“Models with few indicators per factor . . . have fewer(degrees of freedom), leaving more ‘room to maneuver’the parameter estimates so as to minimize the fit function,which in turn is a function of the residuals” (i.e., differ-ences between the model implied covariances and theinput covariances).

AN ADDITIONAL EXPLANATION

Intuitively, lack of (single construct) measurementmodel-to-data fit as caused by “unrelated” items in ameasure – the result of retaining items that do not “cluster”well with other items. With lack of single constructmeasurement model-to-data fit, the input or observedcorrelation between an unrelated item and each item in aset of better-related items cannot be satisfactorily ac-counted for by the (usual) model paths connecting them.3

The Endnote 3 equation for the model-implied covarianceof two unidimensional indicators suggests an additionalexplanation for the puzzle of about six indicators. Includ-ing indicators for an LV without accounting for correla-tions among their measurement errors (e.g., because ofcommon method) may eventually ruin their single con-struct measurement model-to-data fit. In different words,by not specifying correlated measurement errors in an LV,the sum of the resulting residuals (i.e., the differencesbetween the Endnote 3 computed covariances of theindicators without correlated error terms, and the inputcovariances) may eventually become unacceptably large.

CLASSICAL REMEDIES FOR LACK OF FIT

The classical remedy for lack of single constructmeasurement model-to-data fit (SCMM fit) was to corre-late several indicator measurement errors (see for ex-ample Bagozzi 1981a; Byrne and Shavelson 1986; Beardenand Mason 1980; Duncan, Haller, and Portes 1971; Reilly1982). However, this has been criticized (e.g., Bagozzi1983; Fornell 1983; Gerbing and Anderson 1984), and itsuse has become rare in published theoretical model tests.Another remedy for lack of SCMM fit is to delete or“weed” indicators. These and other suggestions will bebriefly discussed next, then their usefulness will be ex-plored later.

CORRELATED MEASUREMENT ERRORS

Strategically including a non-zero correlated mea-surement error can improve the difference between a


model-implied covariance estimate and its input/observedcovariance (see Endnote 3). However, authors have criti-cized correlated measurement errors (e.g., Bagozzi 1983;Fornell 1983; Gerbing and Anderson 1984). They are adeparture from the assumptions underlying classical testtheory and factor analysis. Measurement errors that arecorrelated are typically unhypothesized and thus they arediscovered by chance. The process of finding them istedious, and thus error-prone. And, as we will see, theremay be several sets of correlated measurement errors thatwill produce SCMM fit, and there is no guidance forchoosing the “best” set of correlated measurement errors.

ITEM WEEDING

In recent published LV model tests, single constructmeasurement model-to-data fits are usually attained by“item weeding” – dropping items to be used as indicatorsfrom a measure. However, because the resulting set ofconsistent items is unknown beforehand, the results ofitem weeding capitalize on chance. Weeding is also te-dious, and thus error-prone. And, as we will see, there maybe several subsets of weeded items that fit the data. Itemweeding in valid and reliable measures to attain SCMM fithas been criticized because it can change content or facevalidity 4 (e.g., Cattell 1973, 1978; see Gerbing, Hamilton,and Freeman 1994). As mentioned earlier, Cattell (1973)remarked that the resulting weeded measures tend to beoperationally narrow instances of their construct whichusually tap only one facet of the construct.

SECOND-ORDER CONSTRUCTS

Gerbing and Anderson (1984) proposed using a sec-ond-order construct5 as an alternative to correlating mea-surement errors. They argued that a pair of indicators withcorrelated measurement errors could be re-specified as afactor (Factor 1, without correlated measurement errors).Factor 2, containing the rest of the indicators, along withthe two-indicator Factor 1, could then be specified as the“indicator” constructs of a second-order construct. How-ever, because the Factor 1 indicators are unknown before-hand, the process of identifying the Factor 1 indicatorscapitalizes on chance. The process of identifying theFactor 1 indicators is also tedious and error-prone. Inaddition, there may be several second-order constructsthat will fit the data. A second-order construct also is amore complex model specification, and thus it may bejudged less parsimonious.

AGGREGATION

Variations of an approach attributed to Kenny (1979)and involving summed items, has been used presumablyto avoid item weeding (e.g., Heise and Smith-Lovin 1981;James, Mulaik, and Brett 1982; Williams and Hazer

1986).6 This approach produces a single summed indica-tor for an LV, with a fixed LV variance and a fixedmeasurement error variance for that indicator. This full ortotal aggregation (Bagozzi and Heatherton 1994) alterna-tive to item weeding has many advantages (details omit-ted). However, the indiscriminant use of aggregated indi-cators can mask the psychometric structure of a measure(e.g., a summed indicator that is composed of multidimen-sional items). Aggregated indicators are also non-tradi-tional in structural equation analysis, and their use instructural equation analysis could be viewed as not par-ticularly elegant.

PARTIAL AGGREGATION

Bagozzi and Heatherton (1994) used partial aggrega-tion – items were grouped into subsets and each subsetwas then aggregated. This approach has the benefits anddrawbacks of full aggregation. However, because the waythe items should be aggregated could be unknown before-hand, partial aggregation capitalizes on chance. The pro-cess of finding consistent aggregations of items is alsotedious and thus error-prone. In addition, there may beseveral aggregations of items that will fit the data.

EXTERNAL CONSISTENCY ONLY

Item weeding is typically performed one-measure-at-a-time (i.e., using single construct measurement mod-els – see Jöreskog 1993) in order to establish the internalconsistency of each measure (i.e., each measure fits itsSCMM – see Anderson and Gerbing 1988). Then, toassess the external consistency of a measure (i.e., themeasure fits the data in fuller measurement models – againsee Anderson and Gerbing 1988), the resulting internallyconsistent measure is jointly specified in fuller measure-ment models (i.e., measurement models that contain someor all the measures). However, it could be argued that theultimate objective of item weeding is for the full measure-ment model to fit the data adequately (to isolate anystructural model fit problems to the structural paths amongthe constructs). Thus, it may be possible with real-worlddata to skip the internal consistency step, and attainadequate (full) measurement model-to-data fit by item-weeded in a full measurement model only. This alterna-tive may produce measures with fewer items weeded out.However, because any weeded items are unknown be-forehand, this alternative capitalizes on chance. Weedingis tedious, and there may be several subsets of weededitems that will fit the data. In addition, skipping theinternal consistency step violates the “received view” intheoretical model testing using survey data (Andersonand Gerbing’s 1988 “Two-Step” approach: first verifyinternal consistency, then sequentially verify externalconsistency – see Jöreskog 1993).


MEASURE VALIDATION

Ideally, measure validation uses several data sets;one to show measure adequacy (i.e., acceptable psycho-metrics), and another data set to validate (i.e., disprove)the adequacy of the measure. However, this approach iscomparatively rare in published survey model tests. There,measure adequacy is frequently established in a typicallysmall pretest survey(s). These pretests are usually used todetermine a response rate, as well as to preliminarilyassess the measures and the model. Measure adequacyassessment is then combined with the model test in a largersurvey (with possibly additional changes to the measuresto achieve measure adequacy). Obviously a measurevalidation study is desirable. It would allow the “discov-ery” of, for example, “the” weeded subset, “the” corre-lated measurement error structure, “the” second-order LVstructure, “the” partial aggregation structure, or “the”external-consistency-only structure of a measure in studyone. Then, that structure could be tested (disconfirmed) instudy two. In practice, however, conducting multiplesurveys is expensive and time consuming, which may bemore than sufficient to limit its use.

USING THESE APPROACHES

The balance of the paper investigates the aboveapproaches using a survey data set with more than 200usable responses.8 Among the nine LV’s in the hypoth-esized model for which the data was gathered was thesingle endogenous variable N that had a new 18-itemmeasure.9 The eight other LV’s were judged to be validand reliable, but while the measure for N was judged to becontent or face valid, it was multidimensional (i.e., it hadthree dimensions using maximum likelihood exploratorycommon factor analysis). The items in each of the threefactors were subsequently judged to be valid and reliable,but their single construct measurement models were in-consistent. Thus, item weeding or an alternative wasdesirable. In addition to expecting that each of the abovetechniques would be useful, this exploratory researchinvestigated the following Research Questions (omittedbut discussed later). I also made the following assump-tions in the investigations. LV’s used in model testsshould be content or face valid, and reliable. They alsoshould be convergent valid using a strict criterion, such asAverage Variance Extracted (see Fornell and Larker1981). And, except when noted, LV’s first should beinternally and externally consistent (e.g., Anderson andGerbing 1988).

ITEM WEEDING

Appendix A investigated the use of item weeding toproduce subsets of consistent indicators. Several ap-proaches were used, including maximizing the reliabilityof the 18 item measure for N by successively deleting

items that degraded the reliability of N, Ordered Similar-ity Coefficients (OSC’s) (see Anderson and Gerbing1982),16 and a new procedure (see Ping 2004) that usespartial derivatives of the likelihood function with respectto the measurement errors. A total of 24 consistent subsetsof the 18 item measure for N were found. While there mayhave been more, I discontinued the search after it becametedious to judge which subset had the “best” content orface validity. In structural models specified with 9 of theseweeded subsets, the significance of all but three of themodel’s 8 hypothesized associations with N varied fromsignificant to nonsignificant depending on which weededsubset was selected to itemize N (see Table A). Stateddifferently, most hypothesized associations with N weresignificant using some (weeded) consistent subsets, butnonsignificant using others.

PARTIAL AGGREGATION

Appendix B presents the results of using partialaggregation. In summary, a psychometrically adequate(i.e., reliable, convergent valid, and consistent) logicalpartitioning (i.e., grouping items that appear to tap thesame facet of N, see Bagozzi and Heatherton 1994) of all18 items of N was not found. However, averaged itemsfrom forced 2, 3, and 4 factor factorizations were psycho-metrically adequate in full measurement models with eachof the other LV’s itemized with their multiple indicators.In measurement models with each of the other LV’sitemized with fully aggregated indicators (each LV’sindicators were averaged) these itemizations were notexternally consistent. In addition, two aggregated indica-tors for N using its Factor 1 items (i.e., an indicator thatwas the average of the weeded Factor 1 items, and anotherindicator that was the average of the Factor 1 items thatwere weeded out), produced similar results. In structuralmodels specified using the items that fit the data, thesignificance of all but one of the model’s 8 hypothesizedassociations with N varied from significant to nonsignifi-cant depending on which aggregated subset was selectedto itemize N (see Table B). Stated differently, somehypothesized associations with N were significant usingsome partial aggregations, but nonsignificant using oth-ers.

CORRELATED MEASUREMENT ERRORS

Appendix C investigated the use of correlated mea-surement errors. In summary, two sets of correlated mea-surement errors were found that resulted in the Factor 1items of N fitting their single construct measurementmodel. In addition, a set of correlated measurement errorsthat allowed the full 18-item set to fit its single constructmeasurement model was found. In structural models forthe same itemization of N, significant structural coeffi-cients and their t-values were practically identical acrossdifferent sets of correlated measurement errors (see


TABLE A

Hypothesized Associations with “Weeded” N (Standardized Structural Coefficients,

with Standard Errors and t-values of the Corresponding Unstandardized

Structural Coefficient Beneath)

Exog Vars S A I C L V O E

Subset 0 (n9, n

11, n

13, – n

17) Reliab = .956 AVE = .761

-0.425 -0.041 0.036 0.154 0.040 -0.261 -0.062 0.253(0.091) (0.087) (0.082) (0.062) (0.103) (0.105) (0.070) (0.071)-3.465 -0.350 0.454 1.771 0.480 -3.026 -0.650 2.084

Subset a (n5– n

8) Reliab = .930 AVE = 769

-0.331 -0.362 0.080 0.266 0.065 -0.323 -0.240 -0.098(0.109) (0.106) (0.100) (0.075) (0.124) (0.127) (0.086) (0.086)-2.392 -2.634 0.880 2.652 0.677 -3.255 -2.169 -0.710

Subset b (n10

, n12

, n14

, and n18

) Reliab = .934 AVE = 781-0.373 -0.151 0.203 0.117 0.066 -0.345 -0.106 0.029(0.111) (0.108) (0.103) (0.076) (0.128) (0.130) (0.087) (0.088)-2.795 -1.146 2.291 1.213 0.708 -3.622 -0.999 -0.213

Subset c (n13

, n15

– n17

) Reliab = .952 AVE = .837-0.394 -0.022 0.021 0.149 0.035 -0.259 -0.047 0.252(0.106) (0.102) (0.097) (0.073) (0.121) (0.123) (0.083) (0.084)-3.140 -0.178 0.249 1.662 0.398 -2.913 -0.470 -2.013

Subset 1 (n1, n

12, n

14, n

15, and n

18) Reliab = .785 AVE = .598

-0.317 -0.145 0.152 0.119 0.064 -0.348 -0.059 0.081(0.105) (0.102) (0.097) (0.072) (0.121) (0.123) (0.082) (0.083)-2.353 -1.092 1.699 1.230 0.684 -3.603 -0.550 0.604

Subset 1a (n4, n

12, n

14, n

15, and n

18) Reliab = .896 AVE = .668

-0.320 -0.149 0.153 0.119 0.064 -0.351 -0.060 0.078(0.095) (0.092) (0.087) (0.065) (0.109) (0.111) (0.074) (0.075)-2.389 -1.133 1.723 1.235 0.687 -3.651 -0.566 0.585

Subset 2 (n9, n

13, n

15, n

16, n

17, and n

18) Reliab = .947 AVE = .757

-0.399 -0.033 0.032 0.146 0.043 -0.274 -0.046 0.241(0.104) (0.100) (0.095) (0.071) (0.119) (0.120) (0.081) (0.082)-3.237 -0.278 0.397 1.655 0.508 -3.133 -0.477 1.961

Subset 3 (9, n

10, n

11, and n

17) Reliab = .903 AVE = .716

-0.492 -0.117 0.093 0.162 0.066 -0.272 -0.120 0.210(0.084) (0.077) (0.073) (0.055) (0.091) (0.095) (0.063) (0.064)-3.954 -1.008 1.189 1.898 0.804 -3.159 -1.274 1.769

Subset 4 (n11

, n13

, n16

, n17

, and n18

) Reliab = .935 AVE = 751-0.418 -0.049 0.046 0.157 0.033 -0.284 -0.090 0.265(0.102) (0.098) (0.093) (0.069) (0.116) (0.117) (0.079) (0.080)-4.440 -0.411 0.575 1.807 0.385 -3.292 -0.934 2.179

Table C). However, different itemizations of N produceddifferent sets of significant structural coefficients withcorrelated measurement errors.

EXTERNAL CONSISTENCY ONLY

Appendix D illustrates the omission of the internalconsistency verification step to achieve full measurement


model fit using external consistency only. I itemized Nwith its (unidimensional) Factor 1 items obtained from amaximum likelihood exploratory common factor analy-ses.17 The resulting measurement model fit the data, andthe structural model produced results that wereinterpretationally equivalent to the correlated measure-ment error results for the Factor 1 itemization of N(corresponding associations agreed in direction and sig-nificance between the models) (see Table D). However,ignoring internal consistency (slightly) degraded the “uni-dimensionality” of N in the full model (the loadings of theindicators of N on other LV’s). I also itemized all 9 LV’swith their Factor 1 items that were obtained from singleconstruct maximum likelihood exploratory common fac-tor analyses (i.e., N was itemized with its Factor 1 items,the next construct was itemized with its Factor 1 items,etc.). The resulting full measurement model containing Nand the other LV’s fit the data without deleting anyadditional items. However, LISREL produced a warning

that the number of cases was less than the number ofparameters to be estimated and the parameter estimateswere unreliable.

FULL AGGREGATION

Appendix E summarizes the results of using fullaggregation. First, the full (multidimensional) 18-itemmeasure for N was summed then averaged. Next, a fullmeasurement model with this specification of N, and theother LV’s, was estimated. This process was repeatedwith a single averaged indicator for N composed of theFactor 1 (inconsistent) items, then a weeded subset ofitems. In all cases the full measurement models fit the data,as did the corresponding structural models. However, thesignificant associations varied considerably across thesefull aggregations, and the structural models generallydisplayed fewer significant associations as the number ofitems aggregated increased (see Table E).

TABLE B

Hypothesized Associations with Partially Aggregated N (Standardized Structural Coefficients,

with Standard Errors and t-values of the Corresponding Unstandardized



18 Items with Forced 2 Factors; Reliab = .943(F1), .889(F2), .758(LV)a; AVE = .585(F1), .542(F2), .617(LV)a

-0.209 -0.336 0.046 0.269 0.114 -0.296 0.007 -0.136(0.071) (0.070) (0.065) (0.050) (0.082) (0.085) (0.055) (0.056)-1.478 -2.362 0.502 2.573 1.154 -2.856 -0.067 0.960

18 Items with 3 Factors; Reliab = .943(F1), .917(F2), .749(F3), .758(LV)a; AVE = .585(F1), .708(F2), .605(F3),.617(LV)a

-0.409 -0.100 0.104 0.145 0.058 -0.309 -0.081 0.167(0.090) (0.087) (0.082) (0.062) (0.103) (0.104) (0.070) (0.071)-3.422 -0.853 1.319 1.683 0.699 -3.635 -0.858 1.397

18 Items with Forced 4 Factors; Reliab = .957(F1), .930(F2), .812(F3), .938(F4), .858(LV)a; AVE = .737(F1),.769(F2), .530(F3), .858(F4), .637(LV)a

-0.423 -0.197 0.153 0.183 0.072 -0.349 -0.122 0.036(0.086) (0.083) (0.079) (0.059) (0.098) (0.100) (0.067) (0.068)-3.225 -1.537 1.776 1.950 0.794 -3.731 -1.179 0.274

Partially Aggregated Weeded (Subset 2) and “Weeded Out” Factor 1 Items; Reliab = .947(F1), .923(F2), .933(LV)AVE = .757(F1), .753(F2), .938(LV)a

-0.427 -0.133 0.147 0.151 0.061 -0.320 -0.107 0.136(0.089) (0.085) (0.081) (0.060) (0.101) (0.101) (0.069) (0.070)-3.397 -1.084 1.770 1.686 0.700 -3.590 -1.078 1.087

a Latent variable with aggregated factors as items.


OTHER APPROACHES

Appendix F presents the results of investigating theremaining suggestions, second-order constructs and mea-sure validation, and a comment on factor scores. Severalsecond order specifications were attempted, includingrespecifying correlated measurement errors as secondorder constructs, combining consistent and weeded-outitems, and logically grouping items based on their contentor face validity. In summary, while not every possibilityfor a second-order specification was investigated, I couldnot easily find a second order specification of the items ofN that fit the data. In order investigate the use of measurevalidation, and provide an example of a measure valida-tion study that was comparatively less time consumingand less expensive, I conducted a Scenario Analysis. AScenario Analysis is an experiment in which subjects readwritten scenarios that portray a situation in which thestudy constructs are manipulated. Then they are asked tocomplete a questionnaire containing the study measures.The scenario questionnaire included the Factor 1 itemsfrom the measure for N in the mailed-out study. Unfortu-nately, a maximum likelihood exploratory common factoranalysis of N produced two dimensions for the mailed-outstudy’s Factor 1 items. In addition, none of the Factor 1weeded itemizations from the full study were consistent in

the scenario data. These results were subsequently attrib-uted to N being insufficiently manipulated in the ScenarioAnalysis, and insufficient attention to the details of theexperiment.

RESEARCH QUESTIONS

Returning to the research questions (RQ’s) and RQ1,“Would the investigation produce empirical evidence ofthe “unmodeled-measurement-error” explanation of the“about-six-indicators upper limit,” in the first derivative(Modification Indices) weeding procedure, the sum ofModification Indices for Theta’s (measurement errorsthat should be freed/correlated) declined (see AppendixA), and the process of correlating appropriate measure-ment errors improved model-to-data fit (see Appendix C).

Turning to RQ2a, “Would weeding produce dissimi-lar but internally consistent subsets,” Subset 0 was dis-joint from any of the other weeded subsets, and Subsets band c were disjoint (see Appendix A or the lists of itemsin Table A). The other weeded subsets contained 2 ormore items in common.

For RQ2b, “Would weeded subsets produceinterpretationally dissimilar structural model results,”

TABLE C

Hypothesized Associations with N Specified Using Correlated Measurement Errors (Standardized

Structural Coefficients, with Standard Errors and t-values of the Corresponding Unstandardized



Factor 1 Items with Correlated Measurement Errors; Reliab = .963 AVE = .724

-0.386 -0.056 0.086 0.150 0.053 -0.305 -0.106 0.234(0.089) (0.085) (0.080) (0.060) (0.100) (0.103) (0.068) (0.070)-3.158 -0.472 1.073 1.726 0.627 -3.501 -1.106 1.925

Factor 1 Items with 2nd Set of Corr Measurement Errors; Reliab = .963 AVE = .724

-0.367 -0.076 0.086 0.154 0.054 -0.312 -0.115 0.241(0.090) (0.087) (0.082) (0.062) (0.103) (0.105) (0.070) (0.071)-2.968 -0.635 1.055 1.748 0.632 -3.532 -1.178 1.955

18 Items with Correlated Measurement Errors; (Est.)a Reliab = .947 AVE = 560

-0.173 -0.089 0.020 0.081 0.024 -0.198 -0.111 0.006(0.103) (0.100) (0.094) (0.071) (0.118) (0.121) (0.082) (0.081)-1.843 -0.969 0.324 1.206 0.370 -2.975 -1.480 0.066

a Coefficient Alpha Latent variable with aggregated factors as items. Other estimates of reliability such as coefficientomega have been proposed for multidimensional items.


Subset a produced 5 significant structural model associa-tions, while the other subsets produced fewer significantassociations. In addition, three of the Subset a associa-tions were not observed in any of the other subsets (seeTable A).

Regarding RQ3, “Would structural coefficients forthe same set of items be practically equivalent across theapproaches,” the results were mixed. Weeded Subset aproduced 5 significant associations, and an aggregatedSubset 0 produced structural coefficients that wereinterpretationally equivalent (corresponding structuralcoefficients agreed in direction and significance betweenthe specifications) and the structural coefficients were

trivially different (see Table F). However, weeded Sub-set 1, with lower reliability and convergent validity (Av-erage Variance Extracted), and an aggregated Subset 1produced structural coefficients that were interpretationallyequivalent, but while the significant structural coeffi-cients differed by a few points in the second decimal place,the nonsignificant coefficients differed more as theirmagnitude declined.

Further, the 18 (multidimensional) items with corre-lated measurement errors and a full aggregation of all 18items were not interpretationally equivalent. The sameresult obtained from the forced partial aggregations of thefull 18 item measures. However, for the unidimensional

TABLE D

Hypothesized Associations with N Specified Using “External Consistency Only” (Standardized

Structural Coefficients, with Standard Errors and t-values of the Corresponding

Unstandardized Structural Coefficient Beneath)


Factor 1 Items with External Consistency Only; Reliab = .963 AVE = 724

-0.422 -0.080 0.082 0.153 0.051 -0.296 -0.075 0.199(0.090) (0.086) (0.081) (0.061) (0.102) (0.104) (0.069) (0.070)-3.439 -0.672 1.019 1.750 0.609 -3.400 -0.782 1.641

TABLE E

Hypothesized Associations with Fully Aggregated N (Standardized Structural Coefficients, with Standard

Errors and t-values of the Corresponding Unstandardized Structural Coefficient beneath)


Fully Aggregated 18 Items-0.340 -0.234 0.083 0.223 0.093 -0.329 -0.042 0.022(0.086) (0.084) (0.079) (0.059) (0.099) (0.100) (0.067) (0.068)-2.682 -1.872 0.990 2.443 1.052 -3.649 -0.414 0.170

Fully Aggregated Factor 1 Items-0.417 -0.102 0.106 0.147 0.060 -0.314 -0.083 0.170(0.091) (0.089) (0.084) (0.063) (0.105) (0.106) (0.071) (0.072)-3.425 -0.855 1.321 1.685 0.700 -3.638 -0.860 1.399

Fully Aggregated Factor 2 Items-0.329 -0.347 0.089 0.253 0.057 -0.347 -0.193 0.094(0.108) (0.105) (0.099) (0.074) (0.125) (0.125) (0.085) (0.086)-2.400 -2.548 0.973 2.549 0.588 -3.543 -1.760 -0.769

TABLE F

A Summary of the Hypothesized Associations with N

(Omitted – see http://home.att.net/~rpingjr)


Factor 1 items, full aggregation, partial aggregation thatcombined weeded and “weeded out” items, external con-sistency only, and correlated measurement errors pro-duced interpretationally equivalent structural coefficients,and corresponding standardized structural coefficientsthat were within a few points of each other.

Recalling RQ4, “Would larger consistent subsetsproduce fewer significant structural coefficients,” its an-swer was generally yes. The smallest weeded subsetsproduced 2, 3, or 5 significant associations (Subsets 3, b,c, 3 and a respectively), while the larger Factor 1 itemiza-tion of N with 10 items produced 2 significant associa-tions (see Table A).

This generally negative observed relationship be-tween the number of items and structural coefficient sizealso may be due to unmodeled measurement error correla-tions, in this case between LV’s instead of within them.Using covariance algebra and the usual assumptions, thecovariance between two indicators x

1 and z

1 is λ

xa*

λz1

Cov(X,Z) + Cov(εx1

,εz1

), where X and Z are theindicator’s respective LV’s, λ denotes loading, ε denotesmeasurement error, and Cov denotes covariance. Rear-ranging terms,

Cov(X,Z) = [Cov(x1,z

1) - Cov(ε

x1,ε

z1)]/(λ

xa*λ

z1) , (1)

and the error covariance term (Cov(εx1

,εz1

)/(λxa

*λz1

)) is a“disturbance” term. While I have not worked out thematrix algebra completely, I speculate that as the numberof items in X or Z increase, the estimated (fitted) covari-ance of X and Z is likely to decline because of theincreasing number of these unmodeled “disturbanceterms,” ceteris paribus.

Regarding RQ5, “Would correlated measurementerrors mask internal inconsistency,” and RQ6, “Could fullaggregation be used to accomplish the same thing,” one ofthe objectives in the investigations was to find a specifi-cation of N that used the full 18 item measure for N.However, it was apparent that the items in the full 18 itemmeasure for N were not all measuring the same construct.Partial aggregation suggested Factors 1, 2, and 3 wereexternally inconsistent (they were measuring other con-structs besides N). So, although correlated measurementerrors and full aggregation produced a specification for Ncontaining all the 18 items, these specifications may nothave been appropriate because they masked the externalinconsistency difficulty discovered using partial aggrega-tion.

Finally, for RQ7, “Would a larger model mask exter-nal inconsistency,” the Appendix D results, where theFactor 1 items were specified in a full measurement modelthat fit the data, suggest the answer was yes.

DISCUSSION

Because second order specifications did not perform,and measure validation using scenario analysis was im-properly administered, the useful alternatives to itemweeding in the example data set were external-consis-tency-only, partial and full aggregation, and correlatedmeasurement errors. However, ignoring the criticisms ofthese surviving alternatives, the practical obstacle of theattenuated structural coefficients these alternatives pro-duced likely would have removed them from consider-ation for any itemization alternative in the example dataset.18

ITEM WEEDING

Thus, item weeding may continue as the preferredapproach among substantive researchers to attaining mea-surement model-to-data fit in survey data model tests.Improvements to the current practice of ignoring thecriticisms of item weeding might include reporting bothweeded results and full aggregation results. However, theexample suggested that one or more divergences in sig-nificance (i.e., a structural coefficient is significant in onespecification and nonsignificant in the other) may result,and explaining them might be a challenge. Another im-provement might be to find multiple itemizations, then re-convene the item-judging panel to determine the itemiza-tion that best taps the conceptual definition for themeasure’s construct. This alternative is discussed next.

A SUGGESTED ADDITIONAL STEP IN

ITEM JUDGING

Specifically, the factor structure of the target measurecould be determined using maximum likelihood explor-atory common factor analysis, preferably in a measurevalidation study. If the measure is multidimensional, thedimensions or factors could be considered as candidateitemizations.

Next, weeded subsets of the target measure could befound using the procedures in Appendix A (EXCELtemplates are available from the author for Ordered Simi-larity Coefficients (OSC’s) and the First Derivative (FD)procedures). Specifically, because the number of inter-nally consistent weeded subsets may be large, it may besufficient to generate 1 consistent weeded subset from thefull measure (regardless of its dimensionality) using theusual reliability approach (the procedure usually pro-duces 0 to 1 weeded subset), and at least 1 weeded subsetusing OSC’s and FD also using the full measure. Then, 1weeded subset per factor using reliability and FD could begenerated if the full measure is multidimensional. Theobjective would be to find disjoint weeded subsets, orones with minimal overlap (e.g., Subsets a-c, and 1, 1a,and 3 in the example).


Next, the resulting subsets could be judged by rank-ing them by how well each subset taps the target construct’sconceptual and operational definitions. Experience withweeded subset judging suggests that constructs should bejudged singlely – judging weeded subsets for severalmeasures at once can be burdensome. Weeded subsets forthe endogenous variables could be judged first. There alsowas a slight bias toward small itemizations among thejudges. They reported being accustomed to smaller item-izations, and that they seemed easier to match to concep-tual and operational definitions. Reliabilities and averageextracted variances (AVE)19 should probably not be shownfor judging because they may overly influence the rankings(the judging issue is content validity). Then, the bottomranked subsets could be removed from the list for “tweak-ing” (re-judging) and the weeded subsets could be pre-sented in rank order. If a larger itemization for a targetconstruct is at the top of the judges’ overall ranking, thetweaking subsets could include its (smaller) internallyconsistent subsets, regardless of where they finished in theprevious judging, to be certain the larger itemization ispreferred.

Experience suggests that the Factor 1 items are notalways picked by the judges, and they do not alwaysproduce the most significant associations (e.g., Subset a inthe example, with 5 significant associations, was fromFactor 2). Thus, for multidimensional measures, all fac-tors probably should be weeded and presented as candi-dates. As stated elsewhere, experience suggests RMSEAprobably should be used to judge internal consistency.(Unfortunately, this would have eliminated Subset 0, thereliability itemization from item judging.) External con-sistency judging should probably be postponed until aftertweaking and final subset selection because of its com-plexity with multiple candidate weeded subsets.

LIMITATIONS AND FUTURE RESEARCH


SUMMARY AND CONCLUSION


ENDNOTES

(Most omitted – see http://home.att.net/~rpingjr).3 Using path analysis (Wright 1934), the covariance

between two unidimensional indicators x1 and x

2

(i.e., two indicators with only one underlying con-struct) implied by a model is the product of the pathcoefficients on their loading paths from their com-mon construct (i.e., the product of their loadings),

plus the product of the path coefficients due to corre-lated measurement errors (i.e., λ

1*λ

2 + 1*1*Cov(ε

1,ε

2),

where λ denotes loading, ε denotes measurementerror, 1 is the implied path coefficient on the pathbetween each measurement error and their respectivex, and Cov denotes covariance). Because correla-tions between measurement errors are usually as-sumed to be zero, the second (covariance) term isusually ignored.

REFERENCES

Anderson, J.C. and D.W. Gerbing (1982), “Some Meth-ods for Respecifying Measurement Models to ObtainUnidimensional Construct Measurement,” Journalof Marketing Research, 19, 453–60.

____________ and ____________ (1984), “The Effectof Sampling Error on Convergence, Improper Solu-tions, and Goodness of Fit Indices for MaximumLikelihood Confirmatory Factor Analysis,” Psy-chometrika, 49, 155–73.

____________ and ____________ (1988), “StructuralEquation Modeling in Practice: A Review and Rec-ommended Two-Step Approach,” PsychologicalBulletin, 103, 411–23.

Bagozzi, R.P. (1981a), “Attitudes, Intentions, and Behav-ior: A Test of Some Key Hypotheses,” Journal ofPersonality and Social Psychology, 41, 607–27.

____________ (1981b), “An Examination of the Validity

of Two Models of Attitude, Multivariate BehavioralResearch, 16, 323–59.

____________ (1983), “Issues in the Application of Co-variant Structure Analysis: A Further Comment,”Journal of Consumer Research, 9, 449–50.

____________ and T.F. Heatherton (1994), “A GeneralApproach to Representing Multifaceted PersonalityConstructs: Application to Self Esteem,” StructuralEquation Modeling, 1, 35–67.

Bearden, W.O. and J.B. Mason (1980), “Determinants ofPhysician and Pharmacist Support of Generic Drugs,”Journal of Consumer Research, 7, 121–30.

Bollen, K.A. (1989), Structural Equations with LatentVariable. New York: Wiley.

____________ and J.S. Long (1993), Testing StructuralEquation Models. Newbury Park, CA: SAGE.

Browne, M.W. and R. Cudeck (1993), “Alternative Waysof Assessing Model Fit,” in Testing Structural Equa-tion Models, K.A. Bollen et al. eds. Newbury Park,


CA: Sage.Byrne, B.M. and R.J. Shavelson (1986), “Adolescent

Self-Concept: Testing the Assumption of EquivalentStructure Across Gender,” American EducationalResearch Journal, 12, 365–85.

Cattell, R.B. (1973), Personality and Mood by Question-naire. San Francisco: Jossey-Bass.

____________ (1978), The Scientific Use of Factor Analy-sis in Behavioral and Life Sciences. New York:Plenum.

Comer, J.M., K.A. Machleit, and R.R. Lagace (1989),“Psychometric Assessment of a Reduced Version ofINDSALES,” Journal of Business Research, 18,291–302.

Cronbach, L.J., G.C. Gleser, H. Nada, and N. Rajaratnam(1972), “Dependability of Behavioral Measurements:Theory of Generalizability for Scores and Profiles,”Psychological Bulletin, 52, 281–302.

Duncan, O.D., A.O. Haller, and A. Portes (1971), “PeerInfluences on Aspirations: A Reinterpretation,” inCausal Models in the Social Sciences, H.M. Blalock,Jr., ed. Chicago: Aldane.

Efron, B. (1981), “Nonparametric Estimates of StandardError: The Jacknife, the Bootstrap, and OtherResampling Methods,” Biometrika, 68, 589–99.

Fornell, C. and D.F. Larker (1981), “Evaluating Struc-tural Equation Models with Unobservable Variablesand Measurement Error,” Journal of Marketing Re-search, 18, 39–50.

____________ (1983), “Issues in the Application of Co-variant Structure Analysis: A Comment,” Journal ofConsumer Research, 9, 443–47.

Gerbing, D.W. and J.C. Anderson (1984), “On the Mean-ing of Within-Factor Correlated Measurement Er-rors,” Journal of Consumer Research, 11, 572–80.

____________ and ____________ (1993), “Monte CarloEvaluations of Goodness-of-Fit Indices for Struc-tural Equation Models,” in Testing Structural Equa-tion Models, K.A. Bollen and J.S. Long, eds. NewburyPark, CA: SAGE Publications.

____________, J.G. Hamilton, and E.B. Freeman (1994),“A Large-Scale Second-Order Structural EquationModel of the Influence of Management Participationon Organizational Planning Benefits,” Journal ofManagement, 20, 859–85.

Gerbing, David W. and John E. Hunter (1988), ITAN: A

Statistical Package for Item Analysis with Correla-tional Data Including Multiple Groups ConfirmatoryFactor Analysis, Portland, OR: School of BusinessAdministration, Portland State University.

Heise, D.R. and L. Smith-Lovin (1981), “Impressions ofGoodness, Powerfulness, and Liveliness from Dis-cerned Social Events,” Social Psychology Quarterly,44, 93–106.

Hunter, J.E. and D.W. Gerbing (1982), “UnidimensionalMeasurement, Second-Order Factor Analysis, andCausal Models,” in Research in Organizational Be-havior, Barry M. Staw and L.L. Cummings, eds.Greenwich, CT: JAI Press, IV, 267–320.

James, J.R., S.S. Mulaik, and J.M. Brett (1982), CausalAnalysis. Beverly Hills: SAGE.

Jöreskog, K.G. (1970), “A General Method for Analysisof Covariance Structures,” Biometrika, 57, 239–51.

____________ (1993), “Testing Structural EquationModels,” in Testing Structural Equation Models,K.A. Bollen and J.S. Long, eds. Newbury Park, CA:Sage.

____________ and D. Sörbom (1996), LISREL 8 User’sReference Guide. Chicago: Scientific Software Inter-national, Inc.

Kenny, David (1979), Correlation and Causality. NewYork: Wiley.

Ping, R.A. (2004), “On Assuring Valid Measures forTheoretical Models Using Survey Data,” Journal ofBusiness Research, 57 (2), 125–41.

Reilly, M.D. (1982), “Working Wives and ConvenienceConsumption,” Journal of Consumer Research, 8(March), 407–18.

Rindskopf, D. and T. Rose (1988), “Some Theory andApplications of Confirmatory Second-Order FactorAnalysis,” Multivariate Behavioral Research, 23(January), 51–67.

Steiger, J.H. (1990), “Structural Model Evaluation andModification: An Interval Estimation Approach,”Multivariate Behavioral Research, 25, 173–80.

Williams, L.J. and J.T. Hazer (1986), “Antecedents andConsequences of Satisfaction and Commitment inTurnover Models: A Reanalysis Using Latent Vari-able Structural Equation Methods,” Journal of Ap-plied Psychology, 71 (May), 219–31.

Wright, S. (1934), “The Method of Path Coefficients,”Annals of Mathematical Statistics, 5, 161–215.

APPENDIX A

Item Weeding (Omitted – see http://home.att.net/~rpingjr)

APPENDIX B

Partial Aggregation (Omitted – see http://home.att.net/~rpingjr)


APPENDIX C

Correlated Measurement Errors (Omitted – see http://home.att.net/~rpingjr)

APPENDIX D

External Consistency Only (Omitted – see http://home.att.net/~rpingjr)

APPENDIX E

Full Aggregation (Omitted – see http://home.att.net/~rpingjr)

APPENDIX F

Other Approaches (Omitted – see http://home.att.net/~rpingjr)

For further information contact:Robert Ping

Department of MarketingWright State University

Dayton, OH 45437E-Mail: [email protected].


THE VITAL FEW AND THE USEFUL MANY USING SUPPORT VECTOR

MACHINES TO IDENTIFY FUTURE BEST CUSTOMERS

Markus Wuebben, Technical University Munich, GermanyFlorian Wangenheim, Technical University Munich, Germany

ABSTRACT

Disproportionate marketing activities in the spirit ofcustomer equity management require accurate identifica-tion of high and low value customers. However, currentforecast methods suffer from severe forecasting error andmake their application in managerial practice impossible.In this paper we introduce the support vector machine aspromising methodology for high value customer identifi-cation and test its performance against the Pareto/NBD,BG/NBD, and logit model as well as a managerial heuris-tic on three data sets from three different businessesoperating in non-contractual settings. We find that thesupport vector machine clearly outperforms the compet-ing models and suggest further investigation of the sup-port vector machine in marketing.

INTRODUCTION

In the spirit of customer equity management (Blattbergand Deighton 1996), customer relationships are viewed asassets that require appropriate investments and manage-ment. Valuable customers and customers that are becom-ing increasingly valuable should receive more invest-ments and more management resources than less valuablecustomers (Venkatesan and Kuma 2004).

It is widely known that for many businesses only asmall fraction of all customers accounts for a preponder-ant share of revenues (Mulhern 1999). Very often only 20percent of the customers account for 80 percent of therevenues of a firm, which is yet another reflection of thePareto-principle (Juran 1941). These very profitable 20percent of the customer base should be given specialattention, e.g., being in the focus of marketing campaigns,given special perks, and incentives that aim at retainingthem and increasing their loyalty.

Disproportionate marketing is common in many busi-nesses. Airlines offer shorter checkin queues to their bestcustomers, board them early, give them access to theairline’s bestcustomer-lounge and give them additionalfrequent flier mileage. CD stores, coffee shops, and hotelsmay give away free units, i.e., free CDs/DVDs, freecoffee, free weekend-stay, after a certain purchase amount.However, this strategy is based on a quid-pro-quo ap-proach (Malthouse and Blattberg 2005) that is common inloyalty programs: Customers get special incentives be-

cause they have already given (their money) to the com-pany. Nevertheless, a proactive approach would requirepredicting which current best customer remains futurebest customer and ideally which current less profitablecustomer will turn into a future best customer. For ex-ample, best customers of a hotel may occasionally beupgraded to a better room or have waived access fees forthe hotel’s W-LAN. Car rental companies may occasion-ally upgrade their best customers to a better car, or waiveor lower the deductible. These discretionary marketingactivities aim at triggering customer (positive) surpriseand customer delight, which refers to “a profoundlypositive emotional state generally resulting from havingone’s expectations exceeded to a surprising degree” (Rustand Oliver 2000, p. 86) and ultimately at increasingretention and increased revenues.

However, current methodologies very often sufferfrom severe forecasting error. In fact, using feed forwardneuronal networks and regression analysis, Malthouseand Blattberg (2005) found that of the top 20 percent bestcustomers approximately 55 percent will be misclassified,while of the bottom 80 percent, approximately 15 percentwill be misclassified. This indicates that basing dispropor-tionate marketing activities on current prediction modelsmay result in unnecessary spendings of scarce marketingresources on less unprofitable customers or omittedspendings on more profitable customers who otherwisecould further be developed by special treatment.

The field of marketing has long favored models thatare based on structured parametric statistics, such as logit,probit, hazard, and NBD models, and, at the same time, areeasy to interpret. However, as Bucklin et al. (2002)remark, with today’s diverse data sets, “. . . it may becounterproductive to rely primarily on standard statisticalmethods. Emphasizing scalable methods and predictiveresults may enable us to observe a richer set of behavioralphenomena. . . .” Customer bases may often be very large,containing hundreds of thousands of customers and myriadvariables. Additionally, companies’ customer databasesare often incomplete, i.e., not the same information isalways available for all customers. For example, custom-ers may or may not have participated in surveys, and if so,some of them may only have given partial information.Others may have had customer service contact whileothers have not. Last but not least, customer serviceemployees sometimes maintain customer records care-


lessly leaving out important information they should haveentered. Nevertheless, managers request tools and meth-ods that work for all of their customers alike.

In the field of machine learning, researchers havelong sought for methods that are scalable and work onlarge samples (cf., Schölkopf and Smola 2002). Still,recognition and diffusion of these methodologies into themarketing literature has been slow. Just recently Guptaet al. (2006) remark that “Many of these [machine learn-ing] approaches may be more suitable to the study ofcustomer churn where we typically have a very largenumber of variables, which is commonly referred to as the“curse of dimensionality.” The sparseness of data in thesesituations inflates the variance of the estimates, makingtraditional parametric and nonparametric models lessuseful.” In this spirit, it appears inevitable to empiricallyanalyze the applicability, performance and limitations ofthese methods in marketing.

One of the most powerful and promising methodolo-gies that recently emerged out of the field of machinelearning are support vector machines (SVM; e.g., Vapnik1998). SVMs are capable of multi-class classification andregression as well as many derived applications such aslearning hidden markov models, sequence alignments,and context free grammars (Tsochantaridis 2004). Amongothers, they have successfully been applied in cancerdiagnostics (Guyon et al. 2002), bioinformatics, i.e., geneclassification (Furey et al. 2000) and character recogni-tion (Joachims 2002). Nevertheless, in the marketingliterature there exists only one work that deployed SVMs(Cui and Curry 2005). They conduct a Monte-Carlo studyon simulated data and find superior prediction perfor-mance of SVMs in comparison to a multinominal-logitmodel. However, Cui’s and Curry’s work mainly aims attheoretically analyzing SVMs. Empirical evidence of theapplicability, performance and limitations of support vec-tor machines on actual customer behavioral and attitudi-nal data in services marketing is still missing.

The paper suggests SVM as a promising methodol-ogy to best customer classification. We empirically ana-lyze the classification performance of SVM using threedifferent datasets from three different businesses. All ofthese businesses operate in non-contractual settings(Schmittlein, Morrison, and Columbo 1987), i.e., thecustomer-firm relationship is not governed by contractwhich very often predetermines the contract lengths andpurchase volume. Noncontractual settings make it verydifficult for companies to determine whether a customerwill repurchase and if so how much. As a comparativebasis for SVM we also test its performance against thePareto/NBD (Schmittlein, Morrison, and Colombo 1987)and BG/NBD model (Fader, Hardie, and Lee 2005),which are considered state-of-the-art methodologies fornoncontractual settings. We also ran a logistic regression

as well as a very simple management heuristic. Thus, ourcentral contribution is to provide a first empirical test ofthe usefulness of a new approach for customer classifica-tion that has received substantial recognition in otherscientific fields.

The paper is organized as follows. First we introducesupport vector machines and give an introduction tostructural risk minimization, which is the theoretical basisfor SVM. Then, the data and methodological approachused in this study is presented. Finally, we present anddiscuss the result and wrap this work up with a conclusion.

SUPPORT VECTOR MACHINES

In this section, we very briefly introduce SVM. Adetailed introduction to this methodology would be outscope of this paper. The interested reader is referred toBurges (1998).

Given a training set [xi, yi], i [1,... ,l ] , yi [–1,1] ,xi Rd, where xi represents a vector of an individual’scharacteristics, e.g., an individual’s purchase amount,purchase timing etc., and yi signals to which class theindividual belongs, e.g., best customers (yi = 1) or lessvaluable customers (yi = -1). SVM looks for the seperatinghyperplane that has the largest margin between negativeand positive examples. This hyperplane is defined through

[x|wT x + b = 0]

where w is normal to the hyperplane and b R . In order tofind this separating hyperplane two supporting hyper-planes

[x|wT x + b = 1]

[x|wT x + b = -1]

need to be defined that have equal and maximum distanceto the separating hyperplane and that satisfy:

wT xi + b > 1 for yi = 1

wT xi + b < - 1 for yi = -1

This is equivalent to maximizing 2/||w|| which in turn isequivalent to minimizing wT w under constraint:

yi (wxi + b) > 1 for all i [1,.. , l ]

One of the attractive features of SVM is that thisoptimization problem, in contrast to Maximum Likeli-hood Estimation, always has unique and global solution.

However, this formulation of the problem assumesthat the data is linearly separable. In marketing problems

1

2–


this is generally not the case. Customers showing the samepurchase pattern in the past may or may not be future bestcustomers, i.e., one of them might have changed jobs orpassed away while the other continues to pursue hiscurrent purchase pattern. In order to cater for this problem,slack variables ζ

i are introduced that in combination with

a constant cost parameter C allow for non-linearly sepa-rable data. This combination associates certain cost tonon-linearly separable data, i.e., the higher C the higherthe penalty for misclassified individuals. C is parameterthat can be determined prior to the analysis or throughexperimentation. The cost parameter and slack variableschange the optimization problem to

min wT w + CΣζi

under constraints:

xi w + b > 1 - ζi for yi = 1

xi w + b < 1 + ζi for yi = – 1

with ζ i > 0 for all i .

Another strength of SVM is that through the intro-duction of a so-called “kernel-function”: Rn → H, nonlinearly separable data can be made linearly separable(Burges 1998). In fact, this can be implemented very costefficiently in the optimization algorithm and is therefore

called the “kernel-trick.” However, no theory exists whichdetermines the optimal kernel function. Finding the “best”kernel function and its parameters is therefore subject toexperimentation only. One of the consequences of usingkernel functions is that the projection space H may be ofmuch a higher dimensionality than the original train setspace. This makes interpretation of the determinants ofclass membership particularly difficult. The theoreticalfoundation for SVM is called “Structural Risk Minimiza-tion” which is introduced in the next section.

Structural Risk Minimization

Suppose we are given l observations, i.e., l custom-ers. Each observation consists of a pair: a vector xi Rd ,i =1...l and the associated “truth” yi, given to us by a trustedsource. In our case, xi might be a vector of customerbehavior, and yi would be 1 if the customer is a bestcustomer and -1 otherwise. Now it is assumed that thereexists some unknown probability distribution P(x; y) fromwhich these data are drawn, i.e., the data is assumed “iid”(independently drawn and identically distributed). (P isused for cumulative probability distributions and p fortheir densities). This assumption is more general thanassociating a fixed y with every x: it allows for a distribu-tion of y for a given x. In that case, the trusted source wouldassign labels yi according to a fixed distribution, condi-tional on xi.

FIGURE 1

Separating Hyperplane in the Linearly Saperable Case

1

2–


Now suppose we have a machine whose task it is tolearn the mapping xi → yi. The machine is actually definedby a set of possible mappings x → f(x; a), where thefunction’s f(x; a) themselves are labeled by the adjustableparameters a. The machine is assumed to be deterministic:For a given input x and choice of a, it will always give thesame output f(x; a). A particular choice of a generateswhat is called a “trained machine.” The expectation of thetest error for a trained machine is therefore:

R (a) = ∫ | y – f (x , a) | dP (x , y)

Note that, when a density p(x; y) exists, dP(x; y) maybe written p(x; y)dxdy. This is a nice way of writing thetrue mean error, but unless we have an estimate of whatP(x; y) is, it is not very useful.

The quantity R(a) is called the actual risk, to empha-size that it is the quantity that we are ultimately interestedin. The “empirical risk” Remp(a) is defined to be just themeasured mean error rate on the training set (for a fixed,finite number of observations):

Remp (a) = Σli=1

| yi - f (xi ,a)|

Note that no probability distribution appears here.This is an especially interesting property since manydatasets in Marketing do not satisfy distributional require-ments of standard econometric methodologies. Remp(a) is

a fixed number for a particular choice of a and for aparticular training set [xi; yi].

Now choose some η such that 0 < η < 1. Then thefollowing bound holds (Vapnik 1995):

R (a) < Remp (a) + h (log (2l/h + 1) - log (η 4)

where h is a non-negative integer called the VapnikChervonenkis (VC) dimension, and is a measure of thenotion of capacity. The right-hand side of the bound iscalled the “risk bound.” The second term on the right-hand side is called the “VC confidence.”

We note three key points about this bound. First,remarkably, it is independent of P(x; y). It assumes onlythat both the training data and the test data are drawnindependently according to some P(x; y). Second, it isusually not possible to compute the left-hand side. Third,if we know h, we can easily compute the right-hand side.Thus given several different learning machines (recallthat “learning machine” is just another name for a familyof functions f(x; a)), and choosing a fixed, sufficientlysmall η, by then taking that machine which minimizes theright-hand side, we are choosing that machine whichgives the lowest upper bound on the actual risk. This givesa principled method for choosing a learning machine fora given task, and is the essential idea of structural riskminimization.

FIGURE 2

Separating Hyperplane in the Not Linearly Saperable Case

√l

1

21–

1

2–


FIGURE 3

Transforming to Linearly Separable Case Through a Kernel Function

Vapnik (1995) shows that the VC dimension for asupport vector machine is bound by a term that is recipro-cally proportional to the margin of the data. Therefore,maximizing the margin between the two classes reducesthe VC-dimension. By maximizing the margin betweenthe two classes, SVM implements the principle of structralrisk minimization. Therefore, SVM are very resistant toover fitting (Veropoulos 1999).

RESEARCH METHODOLOGY

The objective of this paper is to identify a company’sfuture best customers using SVM on past purchase infor-mation using three data-sets from three different indus-tries with non-contractual settings (Schmittlein, Morrison,and Columbo 1987). We evaluate the SVM performanceagainst a logit model, the Pareto/NBD (Schmittlein,Morrison, and Columbo 1987) and BG/NBD model (Fader,Hardie, and Lee 2005a) and a simple management heuris-tic. The NBD models are limited to a very small amountof input data (number of purchases and purchase timing)and yield forecasts for the number of purchases. There-fore, in order to make the forecasts of the models ascomparative as possible, we (a) use the same input infor-mation for all models, and (b) identify best customers interms of number of purchases. However, because theNBD models do not require “future” information to esti-mate model parameters, we directly forecast purchasebehavior on all customers. SVM and the logit modelrequire n-fold cross-validation to estimate their predictivevalidity. The following sections covers this issue in more

detail. We are aware that this might limit the comparabil-ity of the results to a small extent. However, this is the bestwe can achieve with the focal models and data availableand should suffice to compare the models’ forecast per-formance.

Data

We conducted our study on three different data setsfrom three different industries. The first data set comesfrom an apparel retailer and covers a cohort of 2,330customers that started their buyer-seller relationship withthe apparel retailer in the last week of January 2003.Therefore, for this cohort, the available data covers theinitial and repeat purchases for each customer over aperiod of 80-weeks. To calibrate the models, we usedrepeat purchase data for the 2,330 customers over the first40-weeks of the 80-week period leaving a 40-week hold-out period to validate the models.

The second data set comes from a major global airlineand covers 146,961 customers and their purchases fromJanuary 1999 through December 2002. The data availableonly provided aggregated quarterly transactions for eachcustomer and did not include the exact purchase dates.Our analysis of this data set focused on a cohort of 2,891customers who conducted their initial purchase from theairline in the first quarter of 1999. For this cohort, wechose a calibration period of eight quarters (January 1999through December 2000) leaving eight quarters for theholdout period (January 2001 through December 2002).


The third data set covers customers of the online CDretailer CDNOW. The data tracks 23,570 customers andtheir purchases from January 1997 through June 1998(78-weeks) who all initiated their first purchase at CDNOWin the first quarter of 1997. Fader and Hardie have alreadyused this data in multiple studies (see Fader and Hardie2001) for more information of this data set. More pre-cisely, we used the 2,357 customer cohort available onBruce Hardie’s website (Fader, Hardie, and Lee 2005b).This cohort is a 1/10th sample of the 23,570 customers inthe data set. The calibration and holdout periods are 39-weeks each.

The information we use solely consists of customers’past purchase behavior. More precisely, for each cus-tomer the models operate on three values (X = x,t,T),where

• X = x is the number of purchases;

• made in time-frame (0,T];

with the last purchase occurring at time t, with 0 < t < T.We do so because we want to gain comparable results ofSVM, logit model and the focal NBD models. The NBDmodels are not able to incorporate more than the informa-tion above.

Determining Future Best Customers Using Support

Vector Machines

SVM need to be trained on a set of customers, whereeach customer is represented by a vector (x1,...,xn,y). TheSVM yield a model that is subsequently used to classify atest set of customers. The predictions yield class-member-ships (best/less valuable customers), which can then bevalidated using actual class memberships. In this analysis,the xi represent the number of purchases and the timing ofthe last purchase given certain time-frame. y is either -1or 1 depending on whether the customer currently be-longs to the set of best customers (y = 1) or not (y = -1). Inorder to achieve “honest” prediction accuracy, we use 5-fold cross-validation (Efron and Tibshirani 1993). In n-fold cross-validation, the original sample is randomlypartitioned into n subsamples. Of the n subsamples, asingle subsample is retained as the validation data fortesting the model, and the remaining n-1 subsamples areused as training data. The cross validation process is thenrepeated n times (the folds), with each of the n subsamplesused exactly once as the validation data. The n results fromthe folds are then averaged to produce a single estimation.

Determining Future Best Customers Using Pareto/

NBD and BG/NBD Model

Both, the Pareto/NBD and BG/NBD model weredeveloped to model repeat-buying behavior in a setting

where customers buy at a steady (albeit stochastic) rateand eventually become inactive at some unobserved pointof time. In addition, the models require to be calibrated onthe customer base they are applied to. This calibrationprocess yields several model parameters that describe thepurchase and dropout process of the analyzed customerbase. We estimated the models using Matlab. The param-eter estimates can be found in Table 1a and Table 1b.

Among others, the Pareto/NBD and BG/NBD modelyield E(X*|X=x,t,T,T*), the expected number of transac-tions X* of a random customer with purchase patterns(X=x,t,T) (and unknown individual purchase-rate anddropout-rate) in a time-frame (T,T+T*]. These individualcustomer predictions can then be used to determine thecompany’s future best customers. Validation of the NBDmodels does not require n-fold cross-validation. Thereason is that the NBD models do not incorporate any“future” data, i.e., unlike other methodologies, the modelparameters are estimated only on past purchase data butnot on future data that tell the “true” outcomes of param-eter constellations. The procedure is then as follows: First,estimate the model parameters, then generate individualpurchase behavior forecasts and validate these using theholdout data.

Determining Future Best Customers Using the Logit

Model

The Logit model’s parameters were estimated usingR’s glm function (binomial family with logit link func-tion) (R Development Core Team 2007) and a training setof customers. Then using the logistic function:

exp(beta0 + beta

1 *X

1beta

2 *X

2)

P (Y =1 given X I = xi) =

1 + exp(beta0 + beta

1 *X

1 + beta

2 *X

2)

the probability of belonging to the group of best customerscan be computed for each customer in the test set. In orderto dichotomize the probability of belonging to the groupof best customers, a cutoff point needs to be determined,above which customers are classified as belong to thegroup of best customers. Since .5 may not be an optimalcutoff threshold, a sensitivity analysis is run according tothe following scheme:

Determine the cutoff threshold pc which maximizesthe average of the percentages of correctly classified bestand correctly classified low customers.

This balanced approach ensures that errors for falsepositives and false negatives are equally weighted andalso respects the different class sizes. If a metric waschosen that averages the overall, i.e., sum of the numberof correctly classified high and low customers, the differ-ent class sizes may lead to solutions that are optimal onlybecause the number of low customers intrinsically out-


weighs the number of high customers. If such a metricwere used, a simple approach that would classify allcustomers as low customers easily reaches 80 percent(90%) correctly classified customers. Ultimately, we areinterested in a methodology that performs well in bothclassifying low and high customers correctly but superiorin classifying high customers correctly.

In order to compute the optimal cutoff level pc a 5-fold cross validation is run for each p in 0..1 (discretizedin steps of 0.01). In each fold we estimated the model onfour partitions and predicted on the remaining partitions.The results of all folds are then averaged to in order to gaina point estimate of the performance for a single p in 0..1.The p that has the highest balanced overall correctlyclassified hitrate, is chosen as pc.

Determining Future Best Customers Using a Simple

Management Heuristic

If the complex models under consideration should beused to identify future best customers, they need to per-form better than a simple management heuristic. Thesimple management heuristic we are using assumes that:

The past’s best 10 percent (20% resp.) of the custom-ers in a customer base will also be future’s best 10percent of the customers.

The validity of the predictions is then checked using theholdout data.

Analysis Results

The analysis shows consistent superior performanceof SVM compared to the NBD and Logit models as wellas the simple management heuristic in predicting acompany’s future best 10 percent and 20 percent custom-ers. Tables 2a–4b show detailed analysis results for allthree data sets.

It is interesting to note, that SVM not only performsmuch better in classifying best customers correctly, butalso on the balanced overall correctly classified percent-ages, i.e., the average percentage of correctly classifiedbest and less valuable customers. Remarkably, in classi-fying the best 20 percent of the customers correctly, theSVM nearly perfectly identifies these high value custom-ers and also nearly perfectly forecasts the low valuecustomer for all three data sets.

However, the performance of the SVM in classifyingthe best 20 percent of customers might give reason todoubt the results. The analysis might have been run infavor of the SVM. Standard n-fold cross-validation usesn-1 partitions for training and the remaining partition forprediction. Since the sample sizes ranged only from2,330–2,891 customers, the SVM might have been able to“memorize” the patterns that lead to future best customerexistence and then applies this knowledge to the rathersmall test set (the remaining partition). Therefore, itappears mandatory to increase the difficulty for the SVM.Due to data constraints, this was possible only on the

TABLE 1a

Results of the Pareto/NBD Maximum Likelihood Estimation

r ααααα r/ααααα s βββββ s/βββββ LL

Apparel 1.0954 9.2029 0.1190 1.0885 973.782 0.0011 -31338.7

Airline 1.4304 0.7196 1.9877 2.5086 19.6408 0.1277 -2150.2

CDNOW 0.5533 10.5776 0.0523 0.6061 11.6650 0.0519 -9595.0

TABLE 1b

Results of the BG/NBD Maximum Likelihood Estimation

r ααααα r/ααααα a b a/(a+b) LL

Apparel 1.0592 8.8371 0.1198 0.0324 2.6243 0.0122 -31336.6

Airline 1.15186 0.54578 2.11048 0.45663 3.73439 0.10895 -2238.34

CDNOW 0.2426 4.4135 0.0549 0.7931 2.4260 0.2463 -9582.43


TABLE 2a

Best 10% Future Customers – Airline

Statistic BG/NBD Pareto/NBD Heuristic Logit SVM

High, correctlyclassified (%) 61.09 61.09 57.84 35.9375 88.7757

Low, Correctlyclassified (%) 95.22 95.22 94.85 69.70874 82.89552

Overall correctlyclassified (%) 91.76 91.76 90.93 75.85952 83.53605

Overall avg. correctlyclassified (%) 78.16 78.16 76.345 53.80704 85.9171

TABLE 2b

Best 20% Future Customers – Airline

Statistic BG/NBD Pareto/NBD Heuristic Logit1 SVM

High, correctlyclassified (%) 64.26 63.40 63.60 N/A 98.17828

Low, Correctlyclassified (%) 89.99 90.60 89.27 N/A 100.00

Overall correctlyclassified (%) 84.81 85.12 84.05 N/A 99.6195

Overall avg. correctlyclassified (%) 77.125 77.00 76.435 N/A 99.08914

TABLE 3a

Best 10% Future Customers – Apparel






1R’s glm.fit procedure did not converge on the data.


TABLE 3b

Best 20% Future Customers – Apparel


High, correctlyclassified (%) 67.13 67.13 73.72 N/A 1




TABLE 4a

Best 10% Future Customers – CDNOW






TABLE 4b

Best 20% Future Customers – CDNOW


High, correctlyclassified (%) 61.250 61.69 71.78 N/A 1




2R’s glm.fit procedure did not converge on the data.3R’s glm.fit procedure did not converge on the data.


airline data-set. First, we increased the sample size used toall customers available (n = 54365) that purchased at leastonce in the focal time frame, then we inverted the logic inthe cross-validation procedure: One partition was used fortraining the SVM while the remaining n-1 partitions wereused as test data. This procedure ensures that the numberof customers in the test data is much higher than the traindata, which was designed to prevent the SVM frommemorizing too many purchase patterns, i.e., to preventoverfitting. The result of the analysis can be found inTable 5. Again, even on the much larger dataset and usingthe inverted cross-validation procedure, the SVM showssuperior performance in classifying future best custom-ers. Once again in classifying the best 20 percent ofcustomers the SVM nearly perfectly predicts the high andlow value customers. Due to computational infeasibilityof the parameter estimation process, we cannot run theNBD models on this large data set. However, the greatperformance of the SVM is confirmed even on this muchlarger data set and the tightened analysis conditions.

DISCUSSION AND CONCLUSION

Being able to identify the best future customers is animport step for a company in efficiently applying discretedisproportionate marketing activities. Very often only 20percent of the customer base accounts for 80 percent of therevenues of a company. Determining which customer willbe one of these vital few enables companies to identifywhich customers should receive incentives and perks inorder to make or keep these customers loyal. However,current methodologies suffer from severe prediction er-rors. In fact, Malthouse and Blattberg (2005) found that ofthe top 20 percent of the customers, about 55 percent willbe misclassified while of the bottom 80 percent, 15 per-cent will be misclassified resulting in unnecessaryspendings in less valuable customers and omitted spend-ing in very valuable customers.

Using SVM these numbers can significantly be im-proved. In this study, we found that SVM clearly outper-forms NBD and logit models as well as simple manage-ment heuristic. Even on a very large dataset on whichNBD models are currently computationally unfeasible,SVM still performs extremely well.

Conclusion

For companies it may very well be an option to applymachine learning techniques instead of standard econo-metric analysis in order to identify their future best cus-tomers. As companies collect more and more data anddatabases with millions of customers become more com-mon, methodologies that are capable of dealing with thesehuge amounts of data are needed. SVM is one of thesemethodologies. SVM has proven the applicability in manyother classification problems, such as cancer detection orface recognition but diffusion into the field of marketinghas been slow. However, current software implementa-tions make the application of SVM machines straightfor-ward and uncomplicated. In fact, they are sufficiently easyto apply that they can be incorporated into automatedforecasting tools without much effort. However, it shouldbe noted that more studies need to be conducted on otherclassification problems than the one studied here, in orderto determine whether the SVM provides superior perfor-mance for other customer base analyses as well.

One of the main points of the critique of SVM is thateffect of the input variables is not directly interpretable.Especially the use of kernel functions (polynomial kernelin this study) is problematic as they transform the originalinput space into a higher dimensional space. However, asdiscussed by Cui and Curry (2005), much more work hasto be put into understanding how standard econometricmethodologies and SVM may work in tandem rather thanbeing mutually exclusive concepts. For example, if an

TABLE 5

Best 10/20% Future Customers (Airline n = 54365)

Statistic Best 10% Best 20%

High, correctly classified (%) 74.65398 99.9891

Low, Correctly classified (%) 90.27637 96.47366

Overall correctly classified (%) 88.70827 97.2142

Overall avg. correctly classified (%) 82.46518 98.23138


SVM has much better predictive power than a standardeconometric technique such as the logit model, then SVMmay shed light on higher dimensional relationships be-tween the predictor variables. This may lead to refined

logit model structure and therefore a better understandingof the problem, while the SVM delivers superior perfor-mance in the actual classification problem. This is subjectto our current research.

REFERENCES

Blattberg, Robert and John Deighton (1996), “ManageMarketing by the Customer Equity Test,” HarvardBusiness Review, 74 (4), 136–44.

Bucklin, Randolph E., James M. Lattin, Asim Ansari,Sunil Gupta, David Bell, Eloise Coupey, John D.C.Little, Carl Mela, Alan Montgomery, and Joel Steckel(2002), “Choice and the Internet: From Clickstreamto Research Stream,” Marketing Letters, 13 (3), 245–58.

Burges, Christopher J.C. (1998), “A Tutorial on SupportVector Machines for Pattern Recognition,” DataMining and Knowledge Discovery, 2 (2), 121–67.

Cui, Dapeng and David Curry (2005), “Prediction inMarketing Using the Support Vector Machine,” Mar-keting Science, 24 (4), 595–615.

Efron, B. and R.J. Tibshirani (1993), An Introduction tothe Bootstrap. London, Chapman & Hall.

Fader, Peter S. and Bruce G.S. Hardie (2001), “Forecast-ing Repeat Sales at CDNOW: A Case Study,” Inter-faces, 31 (Part 2 of 2), 94–107.

____________, ____________, and Ka Lok Lee (2005a),“‘Counting Your Customers’ the Easy Way: AnAlternative to the Pareto/NBD Model,” MarketingScience, 24 (2), 275–85.

____________, ____________, and ____________(2005b), “A Note on Implementing the Pareto/NBDModel in MATLAB,” [http://brucehardie.com/notes/008/].

Furey T.S., N. Christianini, N. Duffy, D.W. Bednarski, M.Schummer, and D. Haussler (2000), “Support VectorMachine Classification and Validation of CancerTissue Samples Using Microarray Expression Data,”Bioinformatics 16 (10), 906–14.

Gupta, Sunil, Dominique Hanssens, Bruce Hardie, WiliamKahn, V. Kumar, Nathaniel Lin, Nalini Ravishanker,and S. Sriram (2006), “Modeling Customer Lifetime

Value,” Journal of Services Research, 9 (2), 139–55.Guyon, Isabelle, Jason Weston, Stephen Barnhill, and

Vladimir Vapnik (2002), “Gene Selection for CancerClassification Using Support Vector Machines,”Machine Learning, 46 (1/3), 389–422.

Joachims, Torsten (2002), Learning to Classify Text Us-ing Support Vector Machines. Boston: Kluwer.

Juran, Joseph (1951), Quality Control Handbook. NewYork: McGraw-Hill.

Malthouse, Edward and Robert Blattberg (2005), “CanWe Predict Customer Lifetime Value?” Journal ofInteractive Marketing, 19 (1), 2–16.

Mulhern, Frank (1999), “Customer Profitability Analy-sis: Measurement, Concentration, and Research Di-rections,” Journal of Interactive Marketing, 13 (1),25–40.

R Development Core Team (2007), R: A Language andEnvironment for Statistical Computing. Vienna.

Schmittlein, David C., Donald G. Morrison, and RichardColombo (1987), “Counting Your Customers: WhoAre They and What Will They Do Next?” Manage-ment Science, 33 (1), 1–24.

Schölkopf, Bernhard and Alex Smola (2002). Learningwith Kernels. Cambridge: MIT Press.

Ioannis Tsochantaridis, Thomas Hofmann, ThorstenJoachims, and Yasemin Altun (2004), “Support Vec-tor Machine Learning for Interdependent and Struc-tured Output Spaces,” 21st Int. Conf. Machine Learn-ing.

Vapnik, Vladimir (1995). The Nature of Statistical Learn-ing Theory. New York: Springer.

____________ (1998). Statistical Learning Theory. NewYork: Wiley.

Veropoulos, K., C. Campbell, and N. Cristianini (1999),“Controlling the Sensitivity of Support Vector Ma-chines,” Proceedings of the International Joint Con-ference on Artificial Intelligence, Stockholm,Schweden, 55–60.

For further information contact:Markus Wuebben

Technical University MunichGermany


Florian WangenheimTechnical University Munich

GermanyPhone: 49.289.28401


3448002 retniW / noitaicossA gnitekraM naciremA

AUTHOR INDEX

963htieK ,rehsbA514 ,151jaR ,irtohingA151leahciM ,enraehA532reteiD ,trelhA931diraF ,demhA643lëoN ,treblA053airaM-nemraC ,thcerblA922giarC ,swerdnA662iesO ,haippA731.J cirE ,dluonrA661.W leinaD ,kcaaB361knarF ,nnamhcaB532fotsirhC ,suahkcaB19.S.M ,ijalaB61.C dlanoD ,senraB202nevetS ,yenraB053 ,411 ,84.H snaH ,reuaB381 ,221.O mailliW ,nedraeB532sennahoJ ,neztnereB901namhaR-ludbA ,nuodyeB892 ,532sukraM ,tulB113netsroT ,nnamenroB241mailliW ,gnidluoB071.M ,divaD ,hsuoB604.L anailiL ,evoB072.F aruaL ,thgirB643ennaeD ,otacorB252htebasilE ,neggürB053 ,411.D roihcleM ,tnayrB922tocS ,notruB391truK ,noslraC862salguoD .J ,llorraC843niskaH ,nahC093ilaruM ,narakehsardnahC513nujrA ,iruhduahC221 ,811harobeD ,notloC914.H linuS ,rotcartnoC011anitteB .T ,llewnroC802.A niboR ,retluoC82hpesoJ .J ,ninorC843anileS ,iuC072 ,911yrreT ,ytrehguaD971.M igiuL ,acuL eD313divaD ,rekkeD662nhoJ ,kcimmiD84saiboT ,trevennoD

dos Santos, Cristiane Pizzutti611naJ ,rengnerD313.H luaP ,nesseirD441.S radreS ,ulgosumruD

Eastin, Matthew S. 119, 266, 270861nitraM ,dnesiE591odI ,verE

692 ,462renieH ,ykztihcsnavE482gnehcuiX ,naF443odnanreF ,osotsaF

Fenandes, Daniel von der Heyde26samohT ,thcsoF252marB ,trebuoF77.R yerbuA ,III relwoF82.L nivaG ,xoF914.J drahciR ,xoF223.T duuR ,hcabmarF862nehpetS ,ecnarF97.R egroeG ,eknarF983yllehS ,nyerF791hpotsirhC ,shcuF611aiP ,miehhcruF57leinaD yraG ,llertuF423 ,991arabraB ,nnamssaG421.F arsE ,krutcneG901ecnaL ,yrtneG993imhskaL ,leoG03lladneK ,hcirdooG252.D enyawD ,relmerG732llejK ,guahnorG173.J nehpetS ,evorG971oloaP ,izneuG401 ,28ogirdoR ,agalaseuG391 ,241tijihbA ,ahuG453iturhS ,atpuG801rednaxelA ,saaH053.E saiboT ,rebaH811kcirederF lieN ,riaH84kiaM ,tdimhcsremmaH332)ydnA( ieW ,oaH62naiviV ,beltraH591nanrE ,yvuraH181esiuoL ,nassaH612yaM esiuoL ,nassaH732netroM ,edieH423 ,991 ,361enidaN ,sginneH211saerdnA ,nnamrreH963eeL ,ttebbiH141tawanaT ,adapiwaynuriH714oteR ,rettetsfoH462.H tumtraH ,rellümzloH113 ,402naitsirhC ,grubmoH202.B kraM ,notsuoH253uY ,uH452ecE eciheB ,nahlI903.L naeJ ,nosnhoJ173.A ttocS ,senoJ604.P rednaxelA ,nessaisoJ901sirroM ,ynillaK421natseD ,rimednaK

358

358

8002 retniW / noitaicossA gnitekraM naciremA444

604.O ognI ,nepraK633.A reteP ,namfuaK801reteP ,gninneK08oniT ,senöhT-relsseK823teejnaR ,ariahK292”yllaS“ gnuoY ,miK402nitraM ,nnamralK223.P.H alleriM ,nenjielK381dnomyaR ,ikselpudroK08nairolF ,suarK714yelraH ,remhorK77.T leahciM ,hsurK452lutA ,inrakluK202.G nahW-kI ,nowK653leuqaR anaitsirhC ,segaL633.W kcirederF ,rhergnaL121.C.Y nivlA ,eeL641.F aivilO ,eeL692kcirtaP ,ztneL083.A leahciM ,niveL841nixgniB eniloraC ,iL332aruaL ,niL161.P hpesoJ ,elttiL623gnipuY ,uiL941gnimeuX ,ouL332nuJ ,aM983rahdeerS ,maravahdaM001ailataN ,elheaM871hsanivA ,ehslaM41anileteD ,avoniraM911tilaG ,eivaL-romraM903.D ylleK ,nitraM81otokaM ,oustaM451xileF ,odnovaM351.E treboR ,dlanoDcM441.C anigeR ,yllaNcM643thgiwD ,aknureM641.L wehttaM ,retueM532 ,23leunaM ,sileahciM01anitsirC acnA ,uciM513 ,802.C ailemaC ,uciM714saihttaM sualK ,relliM351.S leahciM ,roniM201sakiV ,lattiM121drahciR ,iksreziM98rooyaM ,nahoM1nayggarP ,ytnahoM46inawhsA ,ognoM48.M saticileF ,trahroM48ekliS ,reiemlhüM572einaleM ,rellüM

Mummalaneni, Venkatapparao941 ,21.M semaJ ,hcnuM78salkiN ,rhyM121yrraL ,elaeN411.M sucraM ,nnamueN

21.P ueiH ,neyugN453.T esineD ,nedgO452 ,001.C eleC ,sentO06nicumiT ,naczO

Panagopoulos, Nikolaos G.011ivaR ,uppaP97nuE-gnoeJ ,kraP602kitraK ,itapuhsaP243tijihbA ,nahdrawtaP941kraM ,nosreteP173.M yrogerG ,ttekciP653.F legiN ,ycreiP572knarF ,relliP124.A treboR ,gniP922koorB ,kcalP993ajnoS ,cepokorP19aydivirS ,navahgaR602alakhsuP ,namaR514 ,151madA ,ppaR572flaR ,dlawhcieR492nitraM ,nnamieR223.J leihcaM ,srednieR911ynnoS ,lahtnesoR221.S nitraM ,htoR093nitsirK ,ettoR611aleunaM ,eshcaS85lliJ ,reldaS46hsetiR ,iniaS901uhsnA ,naraS083anirbaS ,relttaS492revilO ,eklihcS514sleiN ,treawellihcS003.M treboR ,reldnihcS402sneJ ,ttimhcS26annaH ,nielK-mmarhcS791nitraM ,reierhcS43.H naJ ,nnamuhcS963auhsoJ ,namkcahS06leinaD ,niniehS612 ,181drawdE ,uihS41pidgaJ ,hgniS871.S teerpivaR ,ihoS011adnamA ,yrpS241drahciR ,nileatS83ennA ,wollefgnirtS202noweaT ,huS132niQ ,nuS001engaM ,nellehppuS201ahtinaV ,nahtanimawS26drahnreB ,adobowS731iynauhC ,gnaT922ztnieH aerdnA ,iragnaT492nyleuqcaJ ,samohT85wehttaM ,nosmohT661.D nairB ,lliT211 ,48netsroT ,kazcmoT

36

106

5448002 retniW / noitaicossA gnitekraM naciremA

113kriD ,keztoT514 ,151niveK ,roniarT971eleirbaG ,oliorT203 ,451aneleY ,oknerasT272nibnaY ,uT381eoJ ,ynabrU643erreiP ,ecnerolF-ettelaV243.W salguoD ,seihroV98.E niveK ,ssoV351nnamlliT ,rengaW612 ,181ocnarfnaiG ,hslaW

Wangenheim, Florian V.211leinaD ,leztneW963niraD ,etihW

443lyreJ ,kcoletihWWiedmann, Klaus-Peter

08naJ ,ekeseiW201egaP neraK ,hciretniW

Woisetschläger, David M.68gnaF ,uW234sukraM ,nebbeuW482naL ,aiX071niX-gnauG ,eiX623gnoR ,gnaY43nilihZ ,gnaY41nuJ ,eY841 ,68gnehZ niveK ,uohZ

Zolfagharian, Mohammad Ali

34, 432

163, 199, 324

32, 26, 264, 296, 298

273

Documents

Marketing Theory and Applications - ama.org · Cross-Cultural Differences in the Development of Trust in Relational Service Exchange: An Empirical Analysis in the Banking Context