154
CONSEQUENCES OF REWARDS: THE CREATION, PERPETUATION, AND EROSION OF SOCIAL INEQUALITY A DISSERTATION SUBMITTED TO THE DEPARTMENT OF SOCIOLOGY AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY Sarah Katherine Harkness May 2011

CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

CONSEQUENCES OF REWARDS:

THE CREATION, PERPETUATION, AND EROSION OF SOCIAL INEQUALITY

A DISSERTATION

SUBMITTED TO THE DEPARTMENT OF SOCIOLOGY

AND THE COMMITTEE ON GRADUATE STUDIES

OF STANFORD UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

Sarah Katherine Harkness

May 2011

Page 2: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/zx554nf4443

© 2011 by Sarah Katherine Harkness. All Rights Reserved.

Re-distributed by Stanford University under license with the author.

This work is licensed under a Creative Commons Attribution-Noncommercial 3.0 United States License.

ii

Page 3: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.

Karen Cook, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.

Shelley Correll

I certify that I have read this dissertation and that, in my opinion, it is fully adequatein scope and quality as a dissertation for the degree of Doctor of Philosophy.

Cecilia Ridgeway

Approved for the Stanford University Committee on Graduate Studies.

Patricia J. Gumport, Vice Provost Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file inUniversity Archives.

iii

Page 4: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  iv  

ABSTRACT

This dissertation focuses on how status and rewards jointly impact the creation,

perpetuation and erosion of social inequality. Rewards are objects or positions that

come to have differential levels of prestige when they are affiliated with groups of

varying status, such as certain types of educational degrees, technologies, awards, and

the like. Expectations about who we are and what we should be able to achieve are

formed based on a combination of both our characteristics and displayed status

markers. The first study experimentally tests whether rewards have the power to

create entirely new status characteristics and bases of inequality. The second study is

an examination of how assessments of competence and trustworthiness systematically

bias the distribution of rewards and, thereby, the perpetuation of inequality, by

examining how lenders perceive loan applicants and make funding decisions in

experimentally created lending markets. The third study explores whether rewards

have the power to neutralize status-based inequality when low status individuals are

rewarded with markers of a much higher honorific value than members of high status

groups.

Page 5: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  v  

ACKNOWLEDGEMENTS

Being at Stanford as a member of this remarkable intellectual community has

been one of the most rewarding times in my life. I am eternally grateful to both my

immediate and academic family for all of their encouragement and guidance. It is

only with their support that this dissertation could have been completed. Endless

thanks are due to my primary advisers, Karen Cook and Cecilia Ridgeway, whose

incisive critiques and insights I could not have done without. Every time we met

about this research project, I would come away with new ideas and renewed

motivation. This impressed upon me the importance of being a creative, conscientious

scholar and that there are always questions to be asked. The clarity and joy with

which they teach is inspirational. It is an absolute honor to be their student.

I am also indebted to Shelley Correll who always found aspects of this

dissertation that could be improved. Shelley’s intellectual rigor and dedication are

truly impressive, and I thank her for being such a motivating role model. Many thanks

are also due to Joe Berger and Buzz Zelditch. The foundational theoretical work of

this dissertation is theirs, and I am incredibly lucky to have received their guidance on

this project. I am also grateful to two of my undergraduate professors, Alison Bianchi

and Amy Kroska, who first introduced me to sociological social psychology. It is

their love for and dedication to this discipline that compelled me to become a

sociologist; without them I never would have applied to graduate school. I cannot

fathom what I would be doing instead.

I am also extremely fortunate to have been immersed in such an outstanding

department and to have met such creative and inspiring friends and colleagues. Many

Page 6: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  vi  

thanks are due to the members Social Psychology Workshop for their invaluable

feedback, support, and friendship. Being in this workshop taught me to be a more

complete scholar, and I will miss our weekly meetings tremendously. I would also

like to thank Sara Bloch, Susan Fisk, Alex Gerbasi, Jonathan Haynes, Sharon Jank,

Yan Li, Elizabeth McClintock, Stephen Nunez, Amanda Sharkey, Traci Tucker, and

Alec Watts for all of their help finishing this dissertation. In particular, I would also

like to thank Lynn Chin, not only for her wealth of insights and enthusiasm, but also

because without her peerless work for the Research Experience Program I never

would have been able to run the majority of the experiments for this dissertation. I

will always be in her debt.

A big thank you also goes to all of our administrators, especially Sarah

Giberman, Susan Martin, Randy Michaud, Chrissy Stimmel, and Susan Weersing.

With kindness, understanding, and laughter, you kept my world running.

I would also like to thank my parents, John and Suzanne, who have always

given me the freedom and support to explore life’s possibilities. I am grateful for all

of their trust, assurance, and encouragement. Thank you for instilling the sociological

imagination in me, and I am proud to carry on the family’s sociological tradition into

its third generation.

And I could not have completed this dissertation and degree without my

amazing husband, Michael—we make a fantastic team. Thank you for all the love,

laughter, patience, understanding, flexibility, strategizing, planning, and the thousands

of things you do every day that make us that much happier. I dedicate this dissertation

to you.

Page 7: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  vii  

TABLE OF CONTENTS

ABSTRACT…………………………………………………………………………...iv

ACKNOWLEDGEMENTS…………………………………………………………....v

TABLE OF CONTENTS……………………………………………………………..vii

LIST OF TABLES…………………………………………………………………...viii

LIST OF FIGURES…………………………………………………………………....x

1. INTRODUCTION: THE CONSEQUENCES OF REWARDS…………………….1

2. SPREAD OF STATUS VALUE: THE CREATION OF STATUS

CHARACTERISTICS………………………………………………………………6

References…………………………………………………………………….45

3. THE PERPETUATION OF INEQUALITY: STATUS DYNAMICS IN

LENDING MARKETS…………………………………………………………….49

References…………………………………………………………………….99

4. REWARD INTERVENTIONS: THE EROSION OF SOCIAL INEQUALITY?..106

References…………………………………………………………………...142

Page 8: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  viii  

LIST OF TABLES

2.1 Means or Proportions of Demographic and Bargaining Variables…………...27

2.2 Mean Bargaining Profit Differences by Reward Level...……………………..28

2.3 Mean Ratings of the Value of the Rewards Relative to In-Group Status…….30

2.4 Mean Ratings of “Most People’s” Evaluations of the Rewards’ Possessors

by End Reward Level…………………………………………………………31

2.5 Mean Ratings of “Most People’s” Evaluations of Personal Response Style

Relative to In-Group Status…………………………………………………...33

2.6 Mean Ratings of “Most People’s” Evaluations of Personal Response Style

by End Reward Level………………………….……………………………...34

2.7 Mean Ratings of Personal Evaluations of Personal Response Style

Relative to In-Group Status…………………………………………………...36

2.8 Mean Ratings of Personal Evaluations of Personal Response Style

by End Reward Level…………………………………………………………37

2.9 Mean Difference in the Proportion of Stay Responses……………………….39

2.10 Proportion of Stay Responses ANOVAs……………………………………..40

3.1 Means or Proportions of Demographic Characteristics and Funding

Decision Variables by Study and Participant Group…………………………66

3.2 Study 1: Estimated Mixed-Effects Linear Regression Coefficients for the

Effects of Gender, Race, and Writing Ability on the Competence and

Trustworthiness Scale………………………………………………………...69

3.3 Study 1: Estimated Mixed-Effects Linear Regression Coefficients for the

Effects of Gender, Race, Writing Ability, and Assessments of Competence

and Trustworthiness on Funding Assessments……………………………….77

3.4 Study 1: Estimated Mixed-Effects Linear Regression Coefficients for the

Effects of Gender, Race, Writing Ability, and Assessments of Competence

Page 9: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  ix  

and Trustworthiness on Loan Amount Given………………………………...78

3.5 Study 2: Estimated Mixed-Effects Linear Regression Coefficients for the

Effects of Gender, Race, and Writing Ability on Funding Assessments……..86

3.6 Study 2: Estimated Mixed-Effects Linear Regression Coefficients for the

Effects of Gender, Race, and Writing Ability on Loan Amount Given………87

4.1 Means or Proportions of Demographic Characteristics by Study Medium

and Condition………………………………………………………………..125

4.2 Mean Ratings of Reward Traits by Condition………………………………127

4.3 Mean Ratings of Reward Levels’ Status and Competence by Condition…...129

4.4 Estimated Mixed-Effects Linear Regression Coefficients for the Effects of

Reward Level on the Status Value Scales…………………………………..130

4.5 Estimated Mixed-Effects Logitstic Regression Coefficients for the Effects

of Condition on the Number of Trials Participants Changed Their Initial

Answer………………………………………………………………………133

4.6 Mean Ratings of Partners’ Status and Competence by Condition…………..135

4.7 Estimated Mixed-Effects Linear Regression Coefficients for the Effects of

Partners’ Reward Level, and the Education Status Characteristic on the

Status Value Scales………………………………………………………….136

4.8 Mean Ratings of the Education Status Characteristic’s Status and

Competence by Condition…………………………………………………...138

Page 10: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  x  

LIST OF FIGURES

3.1 Study 1: Predicted Assessment of Competence and Trustworthiness by

Applicants' Gender and Racial Background……………………………….....70

3.2 Study 1: Predicted Assessment of Competence and Trustworthiness by

Applicants' Gender, Racial Background, and Writing Ability………………..74

3.3 Study 2: Predicted Likelihood of Repaying the Loan by Applicants'

Gender and Racial Background………………………………………………89

3.4 Study 2: Predicted Likelihood of Funding Loan by Applicants' Gender,

Racial Background, and Writing Ability……………………………………..90

3.5 Study 2: Predicted Loan Amount Given by Applicants' Gender, Racial

Background, and Writing Ability……………………………………………..93

4.1 Proportion of Trials Participants Changed Their Initial Answer to Agree

with Either Partner by Condition……………………………………………132

Page 11: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  1  

1. INTRODUCTION: THE CONSEQUENCES OF REWARDS

This dissertation is an examination of the interrelations between status, social

rewards, and resources and how these processes create and shape inequality. In

everyday interaction, we are defined not only by our master statuses, such as our

gender and ethnic backgrounds, but also by our awards and positions, which may have

differential social value. Many of these distinctions combine to create an

understanding of who we are, what kinds of behaviors can be expected of us, how

valuable our contributions should be, and the like. These expectations undergird our

shared and accepted social reality, with many using status and reward distinctions as

shorthand for assessments of competence and worth.

Questions therefore arise as to how individuals are affected by the

configuration of valuable rewards in their environment, namely how rewards are

normatively distributed between and within social groups. Can rewards create new

status groups? If we use status information when deciding who should receive more

resources, as prior theory and research suggests, what social groups are most

disadvantaged by this association in contemporary markets? Can the perpetuation of

inequalities based on status distinctions be impeded through the use of reward-based

interventions? This dissertation endeavors to begin to answer these questions.

Paper 1

In the first study, I seek to demonstrate how power, exchange, and status

mutually reinforce each other to create new bases for social inequality. Veblen argues

Page 12: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  2  

that the accumulation of social esteem and honor is the driving force behind all

behavior, and in order to achieve higher levels of status one must consume visible,

honorific goods, such as unpaid leisure. To this day honorific goods have social

significance and are highly esteemed, sought after objects. But what does the

ownership of these rewards really mean to others? It may be possible for these status

symbols to confer or spread their status onto those who possess them, such that people

gain or lose status by virtue of the rewards they have. If we do process the status

information that rewards connote in this fashion, is this spread enough to produce

entirely new status groups by virtue of their association with esteemed or devalued

status objects?

Once the rewards’ value is formed through the association of these markers

with extant status characteristics, this project experimentally tests whether the status

value conveyed by these rewards spreads onto a nominally distinct characteristic of

those who come to possess this reward. The results generally indicate that the states

of the nominal characteristic gain or lose status and influence through their association

with differentially valued rewards. For example, if a new immigrant group,

organization, or college department becomes affiliated with high status objects, the

overall perception of these groups might become more prestigious by virtue of this

association alone. We therefore process the valuations conveyed by status markers in

the same manner as status characteristics, and differential rewards can create new

status distinctions with resulting behavioral expectations, such as divergent

assessments of competence, worth, and influence.

Page 13: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  3  

Paper 2

The second research project explores how rewards affect the perpetuation of

status and wealth inequality by examining status discrimination in lending markets.

One of the key ways individuals accumulate new wealth is first through the receipt of

credit, yet research has documented pervasive group-level disparities in funding

outcomes, such as in obtaining mortgages and venture capital funding. New forms of

credit market have arisen wherein borrowers are financed by their peers and not by

traditional banking institutions with the partial intent of mitigating these inequities.

Even in these peer-to-peer markets, however, borrowers tend to display their personal

information, which conveys many indicators on which bias and discrimination can be

based.

Evidence suggests that these disparities are at least partially due to the actions of

lenders, yet we lack explanations for the mechanisms behind these disparities. In the

second paper, I put forth one possible mechanism. When lenders assess each borrower,

they are at least in part, and perhaps implicitly, assessing the relative competence and

trustworthiness of the borrower against the field of possible borrowers. These

appraisals systematically bias funding decisions, even when the borrowers’ have

commensurate financial histories. Lenders may feel as though they can entrust their

resources to those of higher status in that they assume that these borrowers have the

competence necessary to use the funds faithfully and responsibly and are able to repay

the loan.

The results indicate that these indicators strongly influence a wide array of

funding decisions such that status becomes a means by which lenders compare

Page 14: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  4  

applicants to determine whether and how to fund them, even when the applicants have

similar financial histories. These assessments also systematically vary by gender and

racial categories, such that the funding that African American females receive is

similar to that of white males. Not only are African American females rewarded at the

same level, they are also rated as having slightly higher levels of competence and

trustworthiness than white males. Possible explanations for the combined effect of

gender and race are explored and what this suggests for status characteristics theory

are discussed at the conclusion of this paper.

Paper 3

The continuation of status-based inequality is one of our most important and

pervasive social problems because this source of inequality is founded on essentially

illegitimate assumptions of competence and worth, which can cloud ability

assessments and alter opportunity structures and resource distributions. While rewards

might have the power to create and maintain social inequality when they are relevant

to preexisting status distinctions, rewards may have the capability to strip away the

power of status characteristics as well. This project offers a preliminary step toward

assessing the extent to which a redistribution of socially valued rewards alters status-

based inequality.

In the first two papers, the reward distributions are congruent with the status

groups, such that high status actors are associated with highly desired rewards and low

status actors with devalued rewards. When the situation is reversed, whereby low

status individuals are rewarded with markers of a much higher honorific value than

Page 15: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

  5  

members of high status groups, the relative status advantages of these groups may be

neutralized. The process of using rewards as an intervention can be fraught with

complexity, however. While some types of rewards convey a definite sense of

prestige, value, and ability regardless of their possessors, such as Nobel Prizes, the

meaning of numerous others may not be as immune to the status of those with whom

they are affiliated. Thus, if the reward becomes associated with a group of conflicting

status value, particularly when the reward is relatively novel, the estimation of the

reward and what it connotes may gain or lose value according to the status of its new

possessors.

The third study of this dissertation offers a preliminary experimental test that

assesses the extent to which a redistribution of socially valued rewards alters status-

based inequality. The results tentatively suggest that the meaning of the rewards

becomes contaminated when they are inconsistently affiliated to status groups. Thus,

at least under the particular conditions of this study’s experiment, rewards do not seem

to intervene in status processes. In this paper’s conclusion, future research designs are

discussed to further explore the potential of the reward intervention mechanism.

Page 16: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

6

2. SPREAD OF STATUS VALUE: THE CREATION OF STATUS CHARACTERISTICS

Status and power have been prominent features of theory and research since

the turn of the last century. Weber ([1916] 1946) made one of the earliest distinctions

between power and status, specifying that they are distinct features of the social world

that can have a reciprocal, mutually reinforcing relationship. Veblen ([1899] 2005)

argues that the accumulation of social esteem and honor is the driving force behind all

behavior, and in order to accumulate this status one must consume visible, honorific

goods, such as unpaid leisure. To this day, honorific goods have social significance

and are highly esteemed and coveted. These objects or positions become markers of

prestige, honor, and competence when affiliated with groups who have high status,

and they can make others feel as though they are “in the know” and have distinction

and taste when they come to possess these items.

Obtaining these valued possessions can be an end in and of itself, but these

social rewards may also produce something more. What does the ownership of these

rewards really indicate to others? Evidence tends to suggest that status symbols confer

or spread their status onto those who possess them, such that people gain or lose status

by virtue of the rewards they have (Bierhoff, Buck and Klein 1986; Cook 1975;

Harrod 1980; Lerner 1965; Stewart and Moore 1992). For instance, levels of

attractiveness, including variation due to grooming and clothing, can negatively

impact assessments of credibility and competence (Lennon 1986; Lerner 1965; Patzer

1985; Quereshi and Kay 1986; Rhode 2010). Indeed, almost half of American

Page 17: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

7

workers believe that it is reasonable to not hire someone on the basis of their clothing

(Rhode 2010).

As a separate example, there appears to be a form of a generalized Matthew

effect (Hunt and Blair 1987; Merton 1968, 1988) related to faculty hiring decisions.

Graduating from prestigious departments tends to have a direct impact on assessments

of job candidates’ overall competence and value while even altering hiring committees

appraisals of the candidates’ academic records (Bedeian and Feild 1980; Bedeian,

Cavazos, Hunt, and Jauch 2010; Carson and Navarro 1988; Crane 1965; Long,

Allison, and McGinnis 1979). In this case, the prestige of the degree-granting

institution, a type of status marker, confers status onto its students such that evaluators

expect these students to be more proficient and thrive in the future, holding constant

actual ability and accomplishments.

If we do process the status information that rewards connote in this fashion, is

this spread enough to produce entirely new status groups by virtue of their association

to esteemed or devalued status objects? For example, if new organizations come to

use and be affiliated with valued status markers, such as certain technologies,

esteemed occupations, or prestigious awards, the perception of this group overall may

become more revered through this association alone. As an illustrative example of

this process, consider the case of the creation of a new research center. In academic

communities, the use of certain varieties of methodological techniques and statistical

technologies generally come to be imbued with status as members of the community

tend to believe that those who use them are highly proficient, even if the various

methods accomplish essentially the same task. When a new center is founded, one of

Page 18: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

8

the possible ways through which it can gain status in the academic community may be

through which technologies its predominant members use, thereby helping to create a

new status group. Over time, this new status distinction may diffuse throughout the

community and this center’s members may enjoy status advantages even if they begin

to use a much wider array of technologies.

This research seeks to test this possible mechanism by which new status

groups are created through their association with rewards of various levels. As

Bourdieu argues, symbolic elements of our culture are not only a structural product

but should also have the power to construct new social structures (Bourdieu 1990,

1991). This research offers a test of this contention as well as an extension of status

construction theories by using the status value of possessions as an additional

mechanism in the production of status. More broadly, this is a study of the impact

status symbols can have on social hierarchies.

THEORETICAL BACKGROUND

This research is firmly entrenched within the expectation states research

program of structural social psychology. This body of theory and research seeks to

explicate status processes and their effect on interpersonal relations and inequality.

Within this program, status characteristics theory (hereafter SCT) elucidates how

existing status differences pattern behaviors related to having differential levels of

power and influence in small groups (Berger, Cohen and Zelditch 1966, 1972). This

theory applies to groups who are working together to reach a mutually valued goal

Page 19: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

9

(i.e., who are collectively and task-oriented), such as juries or students working on a

group project for their class.

There are two main kinds of personal characteristics that can distinguish group

members: diffuse and specific status characteristics. Diffuse status characteristics are

culturally defined, socially significant characteristics (e.g., gender or race/ethnicity)

that have varying states (e.g., male-female, White-African American). These various

attributes have differential esteem, honor, and prestige valuations as defined by the

dominant culture that correspond to the level of performance ability a person with a

particular state is assumed to have (Berger and Fisek 2006). Specific status

characteristics are associated with the ability to perform particular tasks, such as

computer skills or business aptitude.

Once a problem-solving group is differentiated by at least one diffuse status

characteristic or by a characteristic that is relevant to the task, individuals will assign

expectations about the performance and potential contributions of group members

based on the valuation of the states of their status characteristics. The characteristic

will be relevant to expectations regarding the individual’s performance on the group’s

task unless it is directly challenged (the burden of proof principle). Behavioral

inequalities favoring the actors who have highly-valued status characteristics will then

emerge with respect to opportunities granted to speak, actual level of participation,

evaluations of others' performance, and the influence members have to change the

mind of others while solving the problem. The power and prestige hierarchy and

expectation hierarchy are mutually reinforcing, and they will remain highly stable over

time (Berger and Conner 1974).

Page 20: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

10

While SCT is mainly concerned with the status of people, another branch of

EST, the status value theory of distributive justice, is concerned with the status of

possessions (Berger, Zelditch, Anderson and Cohen 1972). The theory is based on a

reformulation of Veblen's ([1899] 2005) notion of honorific value that desired objects,

tangible or not, can symbolize status and social standing. Components of the social

world, such as people or objects, come to have status value when they are uniquely

related to elements that do have this type of valuation. If these elements are

consistently appreciated, the related non-status valued object will acquire the same

level of status value. Thus, people expect those with positively valued states of status

characteristics to possess highly status-valued objects.

Status Construction Theories

Traditionally in the expectation states research program, the characteristics that

carry performance expectations, thereby affecting power and prestige behaviors, are

well-known, culturally defined, preexisting distinctions (e.g., age, education, and

gender); however, the question of how nominally distinct characteristics obtain status

value, emerge as status characteristics, and diffuse throughout a society is also within

the purview of this program (Ridgeway 1991, 1997; Berger and Fisek 2006).

Ridgeway (1991, 1997) was the first to posit how initially non-valued

characteristics come to have status value and performance expectations. Nominally

distinct characteristics, such as artistic preference, distinguish only a "mere difference"

between group members, and social identity theorists maintain that this minimal

difference is sufficient to create in-group bias (Tajfel 1978). When a state of this

Page 21: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

11

characteristic becomes more positively valued and more influential in a localized

setting, individuals in the valued group are more likely to believe this judgment

because it reinforces their established in-group bias. A nominally distinct

characteristic becomes a status characteristic when even the disadvantaged group

members believe that most people perceive their group to be less worthy or competent

than the opposing group. It is this assumption of consensus that makes a newly valued

characteristic a social fact (Berger and Luckmann 1966; Ridgeway 1997).

Ridgeway (1991) contends that one environment in which status characteristics

are initially constructed is through goal-oriented, small-group interactions. In these

settings, one way that a non-valued characteristic can obtain status value is though an

association with a structural advantage. Structural advantage here refers to different

levels of material resources, such as wage disparities. When various states of a

nominal characteristic are associated with different levels of resources, the conditions

are set for the status construction process. These situations are "doubly dissimilar"

interactions because individuals are distinguished by their characteristics and by the

amount of resources they acquire. Research demonstrates that when people possess

differential resources, they create performance expectations about themselves and

other group members in direct relation their unequal resources (Bierhoff, Buck and

Klein 1986; Cook 1975; Harrod 1980; Lerner 1965; Stewart and Moore 1992), such

that competence, assertiveness, certainty, and confidence are positively related to

control over resources.

These behaviors form the interactional hierarchy of influence and esteem in a

group working together on a socially important task. Bolstered by these differential

Page 22: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

12

resources, a consensual, valid interactional hierarchy will form in the group. People

then generate performance expectations based on this interactional hierarchy virtually

instantaneously. Resources, therefore, create structural advantages for the rich, and,

consequently, they will gain high levels of influence, esteem, and elevated

expectations for their performance.

After the interactional hierarchy and power and prestige order is formed,

members are likely to attribute the influence and esteem garnered from their relative

structural advantages with their corresponding states of the novel characteristic. This

association creates a valued status characteristic in this immediate situation. Over

successive social encounters, the behavioral expectations come to be associated with

the characteristic itself and not with the original structural advantage as others behave

as though the expectations surrounding the states of the new characteristic are social

facts. Interactions that confirm this new conception help to diffuse this belief

throughout the society, thereby creating a stable status characteristic (Ridgeway, 2006;

Ridgeway, Backor, Li, Tinkler, and Erickson 2009; Ridgeway and Balkwell 1997).

Recently, Berger and Fisek's (2006) have proposed a new mechanism

motivating the construction of diffuse status characteristics. They argue that, aside

from control over resources, nominal characteristics' association with characteristics

that have preexisting status value and related performance expectations can also be a

first step in the creation of a diffuse status characteristic. The status value of the

originating salient characteristic(s) is predicted to spread to the corresponding states of

the nominal characteristic as long as the initial status valued characteristics are

consistently appraised.

Page 23: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

13

This spread of status value process is virtually the same as that which is found

in the status value theory of distributive justice (Berger et al. 1972). The degree to

which the non-valued states of the nominal characteristic acquire positive or negative

status value varies as a function of the number, task relevance, consistency, and

valence of the valued status characteristics a person embodies. This can also vary by

the number of people in the task situation, in addition to other possible relational

factors, such as sentiments among group members. Once this spread has occurred,

performance expectations are associated with the divergent states of the characteristic,

and successive validating social encounters will establish the trait as being a stable,

widely agreed-upon status characteristic.

THEORETICAL EXTENSION

This research explores the process by which the status value of resources is

created through their association with preexisting status characteristics, and how these

rewards possibly create new status characteristics via their ownership. The presumed

mechanism by which initially non-valued states of objects and characteristics obtain

their differential levels of prestige and worth is through the spread of status value

process; therefore, this work is a theoretical extension of Berger and Fisek’s (2006)

status construction theory.

When the actors who possess status valued goods or positions enter into task

situations with others, if the only thing that actors know about each other is the status

value of their possessions and that they differ on a nominal state of a characteristic, the

status value of their rewards should spread to the corresponding states of the new

Page 24: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

14

characteristic for the actors in the situation. The initially non-valued states of a

characteristic acquire status value as a function of the number, task relevance,

consistency, and valence of the original characteristics used to form the status value of

the object (Berger and Fisek 2006; Thye and Witkowski 2003). The resulting value of

the newly created characteristic is predicted to be commensurate with the valence of

this initial reward valuation. Actors should then attach differential performance

expectations in accordance with the newly acquired status value of the states of the

novel characteristic as a function of this esteem.

Once states of an initially non-valued characteristic are assigned a consistent

status evaluation and the associated performance expectations about the actors are

activated, this characteristic will become a status characteristic in the local setting.

The proposed effect of status-valued rewards should work independently of any

structural advantage conferred by monetary inequities. Again following Berger and

Fisek's (2006) reasoning, this new status characteristic will only become stable

through subsequent, validating encounters in the absence of the initial reward

disparities. These new beliefs achieve legitimacy through multiple encounters in

which a social consensus is formed about the valuations and expectations attached to

the characteristic.

Hypotheses

For this proposed research project, I hypothesize that in the spread of status

value from valued status characteristics to resources:

Hypothesis 1: The rewards controlled by high-status actors will have greater status

Page 25: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

15

value than those held by low-status actors.

When the status value of rewards and an initially non-valued characteristic of

two states are the only qualities that differentiate a task-group, I hypothesize:

Hypothesis 2: As the status value of actors' rewards increases, the corresponding status value of their nominal characteristic will also increase.

Hypothesis 3: As the status value of actors’ nominal characteristic increases, the

greater the behavioral expectation advantage attributed to these actors will be.

The first hypothesis pertains to the spread of status value from valued

characteristics to resources, thereby creating rewards of differential esteem. The

second set of hypotheses concerns whether the status value of rewards spreads to an

initially non-valued characteristic. When this nominal characteristic is ascribed status

value by members of the task group, they are predicted to assign greater performance

expectations to those with characteristics of greater status value. This process must

occur for members with both valued and devalued rewards and characteristics as a

status characteristic does not indicate in-group bias but widely held beliefs about the

capabilities and worth for those with certain qualities (Ridgeway 1991, 1997).

The ensuing experiment therefore tests whether status value spreads from

status characteristics to initially non-valued objects, and then onto a nominally distinct

characteristic of the new possessors of these now status-valued objects. With the

absence of any distinguishing information about a interaction partners, except that

they control an objects of differential status value and have a different states of an

initially non-valued characteristic, the objects’ status value should spread to the

Page 26: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

16

nominal characteristic to create a new status characteristic with related performance

expectations.

EXPERIMENTAL DESIGN

The experiment has two phases, which correspond to the two sets of

hypotheses above. In the first phase, the participant works with a partner with known

status characteristics on a bargaining task through a computer. In the second phase the

participant works with a different partner, and they are only differentiated by status

valued rewards and a novel, initially non-status valued characteristic. In each phase,

participants negotiate and interact with a partner, and the objects that they use to

exchange, bargaining chips of two different colors, should become status valued

rewards during the first phase of the experiment. These dyads are “doubly dissimilar”

in that partners always differ with respect to their characteristics (status valued in the

first phase and initially nominal in the second) and the level of their rewards’ status

value. The partners are actually simulated actors to control the task cues given by the

“partner,” the relative status of the “partner,” and the way the interaction unfolds.

This study employs a 2x2 factorial design in which conditions are divided by the

relative status of the participants’ first partners (higher status or lower status) and the

status value of the items controlled by the participants’ second partners (high status

value or low status value)1.

At the start of the first phase of this study, participants have either a status

advantage or disadvantage over their partners according to the states of their salient

1 The partners were always matched by gender. 2 The experimenter does not know the color of the second chip allocation either, as this portion of the

Page 27: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

17

status characteristics. Over the course of the first phase of the study, the initially

nominally distinct chips should take on status value that is commensurate with the

valuation of the states of their characteristics. Newly status-valued chips should then

retain their meaning throughout the second phase. There are four resulting conditions

in this study. Participants may have high status-valued rewards in the first phase (i.e.,

they have a status advantage over their first partner resulting in controlling a particular

type of chip) and then control the chips of high status value in the second phase (i.e.,

the color of chips they control does not change throughout the study) (“High High”).

Second, participants can begin with high status-valued rewards and then control low

status-valued rewards in the second phase (i.e., the color of chips they control changes

when they negotiate with their second partner) (“High Low”). Third, participants

begin with devalued rewards (i.e., they have a status disadvantage relative to their first

partner) and are then allotted high status-valued rewards in the second phase (i.e., the

color of the chips they control changes when working with the second partner) (“Low

High”). Finally, participants can control devalued rewards throughout the study (i.e.,

the chip color does not change from the first to the second phase) (“Low Low”).

These conditions are then collapsed to create the two master conditions of this study

by the reward level with which the participants began the study and the reward level

they had in the second phase of the study.

Phase 1: Creation of Status Valued Rewards

The procedures for the first phase closely follow those used by status value

researchers (Thye 2000; Thye and Witkowski 2003). Upon arrival to the lab, the

Page 28: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

18

participants go into their individual study rooms. They complete a brief survey in

which they indicate their gender, age, race/ethnicity, how many quarters they have

taken courses at Stanford and their grade point average. Participants then take a

“meaning insight” ability test. Expectation states researchers frequently use this

meaning insight task as a valid way to create an ability that only has meaning in the

experiment’s immediate setting (Berger and Conner 1974; Berger, Fisek, Norman and

Zelditch 1977). In each round, participants and their partners must match an English

word to a supposed root word from an ancient language. In reality, there is never a

correct answer to the task. Upon completion of this test, the experimenter personally

informs the each participant of the first partner’s demographic information and

meaning insight score.

In the “high beginning status” condition, the partner is a 24-year-old

engineering graduate student with a 4.0 grade point average and has very high

meaning insight ability relative to the participants. In the “low beginning status”

condition, the participants work with a sixteen-year-old vocational-track high school

student with a 2.3 grade point average and who has lower meaning insight ability than

their own. These traits are then made relevant to the two different chip colors

immediately prior to the first phase of negotiations to create the differential rewards.

Phase 1: Negotiated Exchange

Participants and their partners are then allocated thirty color-coded chips,

either purple or orange, according to their relative status. For instance, it could be that

when the participants have a higher status relative to their partner, they control purple

Page 29: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

19

chips and their partners use orange chips. People do not tend to show a personal

preference for the color purple or orange when they exchange for resources, so these

objects should have no initial valuations (Thye 2000). Nevertheless, the computer

randomly assigns the color of the chips associated with the relative status of the

participants in each session to ensure that color preference is not conflated with the

status value of the chips.

To ensure that the participants are aware of the relevance between the chip

color assignment and their relative status, the experimenter informs them that they

have earned the right to use purple or orange chips according to their performance on

the meaning insight task, their age, and educational background. If the participants

question why this is the case, the experimenter responds that the color assignment

helps to keep track of the differences between the participants and their partners.

Thus, the experimenter authorizes a link between the chips and the status

differences among the participants and their partners, which grants the spread of status

value process. It is at this point that the value of the states of the players’ status

characteristics should spread to the two different chip colors. When this occurs, a

status-valued reward is created with two differential levels commensurate to the

estimation of the initial states of the characteristics.

Although the color of the chips varied, their instrumental value is kept constant

due to the payment structure of the subsequent negotiations, as payment is directly tied

to the number of chips that the participants accrue, regardless of the color of the chips

they have accumulated. This also ensures that the participants are motivated to

accumulate as many chips as possible. Additionally, because the chips are not

Page 30: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

20

assigned an exact monetary value, participants should treat the chips as purely

resources that are utilized to obtain higher pay at the end of the study but do not have a

defined value of their own. In actuality, all participants are paid the same amount for

their participation.

The participants then engage in twenty rounds of negotiations, with each round

lasting no more than two minutes. At the start of each exchange round, the participant

and partner have the opportunity to send an initial offer. The exact person who begins

the round is determined on a first-come, first-serve basis. If, after a random amount of

time, the participant has not given an initial offer, the partner sends an offer. The

partner’s behavior in the negotiations is manipulated through pre-programmed

exchange rules meant to simulate reasonable and feasible bargaining behavior.

Once an initial offer is sent, the recipient can accept it, forfeit the round, or

send a counteroffer. Thus, the person who begins the round is relatively

undetermined, but the round proceeds with both players alternating decisions until the

round is over.

When giving offers (counter or initial), players can increase, decrease, or retain

the last exchange offer. The only constraint is that they must request between 0 and

30 chips (i.e., they can increase/decrease their asking price by 1 chip, 2 chips, 3 chips,

etc.). The round finishes and the next begins when either player agrees to the

immediate offer, forfeits the round, or time runs out.

In the negotiations, the participants can only send requests to the partner

according to how many of the partner’s chips they want (i.e., “I want [x] number of

my partner’s purple chips”). The participants do not know how many of their own

Page 31: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

21

chips their partner receives, but they do know that generally the more chips they

request, the fewer chips the partner obtains. Power-dependence exchange theorists

have used this negotiation procedure to reduce equity-seeking behavior among their

players (Cook and Emerson 1978; Lawler and Yoon 1996; Molm, Peterson and

Takahashi 2001). If participants were to know that they were only dividing a fixed set

of chips between themselves and their partner, they tend to quickly and continually

divide the resource pool equally due to concerns of fairness.

The importance of this first phase of negotiations is that over the course of

these exchanges, the participants should associate the color of the chips with the

relative status of their partner. Thus, the status of their partner spreads to the color of

their chips, which confers a certain level of status value to the chips, thereby making

them rewards of relative value.

Phase 2: Creation of a Nominal Characteristic

At the start of the second phase, the participants change partners, and these

new partners are also simulated. The final time the experimenter interacts with the

participants before the debriefing process is to inform them that they are going to work

with a new partner in the last portion of the study. Thus, the remainder of the study is

double-blind. To create the nominal status characteristic that differentiates the

participants from their second partners, participants complete a “Personal Response

Style” test wherein they indicate their preference for a series of Klee and Kandinsky

paintings (Ridgeway, Boyle, Kuipers and Robinson 1998; Ridgeway and Erickson

2000). The computer script informs the participants that researchers have recently

Page 32: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

22

discovered a new individual trait, personal response style, and people around the

world tend to personify one of the two states of this characteristic, S2 and Q2. The

program then notifies the participants that their test results indicate that they embody

one type of response style, while the results of their partners’ test show that they typify

the opposite style.

Participants could not have had any prior experience with the qualities of

people with the S2 or Q2 personal response style as this characteristic does not exist

outside of the laboratory. Therefore, participants should not have a prior valuation or

expectations for S2s and Q2s. Nevertheless, the computer randomly assigns the

participants the S2 or Q2 response style in each session to ensure that artistic

preference is not conflated with the status value of the chips.

The program then allocates chips to the participant and the new partner for a

second phase of bargaining and the computer script creates a relevance bond between

the states of the novel characteristic and the assigned chip colors, again either purple

or orange. This link is made using the same language that was used in the first phase

of the experiment. The program informs the participants that because they embody

one response type, they will be using a particular chip color, while their partner

typifies the opposing response style and will be using the other color of chips. For

instance, the program may state that because the participant exemplifies the S2

response style, they will be using purple chips, whereas the partner has the Q2

response style and will be using orange chips. The computer again randomly selects

Page 33: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

23

the colors associated with this second chip allocation, although the participant is not

made aware of this2.

Participants may retain the same type of status-valued objects throughout the

study or the status value may change in this second phase. In this second exchange

task, the chip’s status value is divorced from its original direct association with high or

low status exchange partners. They then exchange chips with their new partners until

they complete a total of fifteen exchanges. Participants should have enough time

during this task to form an association between the second chip color assignment and

the nominal characteristic distinction between themselves and their new partner.

In this second set of negotiations, the computer’s bargaining rules are the same

as those of the first exchange exercise. All participants, regardless of whether they

control the same or different rewards in the second phase, experience the same actions

and reactions by their partners. Therefore, the partners do not enact an influence

hierarchy based on relative status.

Status Value of New Characteristic

Prior to assessing the relative influence between the participants and their

partners, the participants complete a brief questionnaire in which they evaluate various

qualities of S2’s and Q2’s. This is done at this particular point in the experiment to

give the participants an opportunity to reflect upon and draw connections between the

reward differences and the new response style characteristic. It also gives them an

2 The experimenter does not know the color of the second chip allocation either, as this portion of the study is double-blind. Because of this, the experimenter cannot give task cues about the second partner, and, therefore the experimenter does not prompt an interactional hierarchy between the participants and their second partners.

Page 34: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

24

opportunity to be made aware of and articulate any beliefs they have about the

characteristic prior to interacting with their partner again. In this questionnaire, the

participants are given a chance to associate the behavior of their new partners with the

differences in their nominal characteristic. This survey contains semantic differential

questions pertaining to participants’ perceptions about S2s’ and Q2s’ status,

competence, and considerateness according to their own opinions and to what they

think most people believe (Ridgeway et al. 1998). This questionnaire assesses

whether personal response style has become a status characteristic, as the participants

must indicate the breadth of their status evaluations.

The following adjective pairs anchor the status and power questions:

respected/not respected, low status/high status, leader/follower, and

powerful/powerless. The competence questions pertain to the pairs of:

competent/incompetent, capable/incapable, and knowledgeable/not knowledgeable.

The questions related to the states’ considerateness include: considerate/inconsiderate,

pleasant/unpleasant, likable/unlikable, cooperative/uncooperative3. The participants

answer these questions with a virtual slide-rule that measures their responses along a

continuous 100-point scale. The status, competence, and considerateness questions

were averaged to create scales related to these three components of status value.

3 Considerateness is an important dimension of status value to explore as many have argued (Conway, Pizzamiglio, and Mount 1996; Fiske, Cuddy, Glick, and Xu 2002) that status has two sides: the agentic/competent/high status face and the deferential/reactive/low status face.

Page 35: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

25

Influence

To measure the degree of influence the partners have over the participants,

they then complete a contrast sensitivity task with their partners. Researchers in the

expectation states tradition frequently use this test as a valid behavioral measure of

influence. In each round, participants and their partners view a picture that contains

black and white tiles. Their task is to decide whether the picture contains a greater

area of black or white. These pictures were created so that they truly contain an equal

proportion of black and white space; therefore there is never a correct answer. The

participants indicate their initial choice before being informed of their partners’

preliminary decision. The simulated partners disagree with the participants on all but

five rounds so as to give the impression of an actual social interaction while still

providing a large number of critical disagreement trials. The participants must decide

whether to stay with their original selection or defer to the selection of their partners.

If the participants change their selection to that of their partners, their partners and,

presumably, their partners’ relative status advantage are influencing them. The

proportion of the participants’ stay responses (p(s)) indicates the extent to which they

are influenced by their partners, with a smaller proportion indicating that the

participants are heavily influenced.

Status Value of the Rewards

Lastly, the participants complete a series of questions related to their

estimation of the monetary and status valuation of the two colors of chips. The

participants first indicate the overall value of the chips and whether they thought the

Page 36: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

26

chips had any monetary value4. They then rate how they and subsequently how “most

people” would rate a person who is signified in this study with purple and with orange.

The aforementioned status, competence, and considerateness questions are used for

this series of questions and are again averaged to create three scales of the rewards’

status value.

RESULTS

Participants and Descriptive Statistics

A total of 88 participants completed this study. Of these, four revealed

comprehension issues and nine were deemed too suspicious during the debriefing

process, and they were dropped from the analyses. The final sample size for the

subsequent analyses is 75. 51 females and 24 males participated in the study, and the

average age of participants is 19, with a range of 18 to 23 years of age (please see

Table 1). The average profit in the first phase of bargaining is 22.93 chips, and

participants tended to increase the amount of the partners’ chips they requested as they

proceeded through this first phase. In this first bargaining period, participants or the

computer forfeited almost a third of the time, and about twenty percent of the rounds

timed-out. The average profit in the second phase of bargaining is 23.52, and there

were about the same number of forfeits (27%) and timeouts (15%). The average

proportion of stay responses (p(s)), the behavioral measure of interpersonal influence,

is .66. Participants did not change their final decisions on the contrast sensitivity test

about two-thirds of the time. 4 Questions related to the value of the first partner’s chips are asked prior to the start of the second phase of the study.

Page 37: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

27

Table 2.1. Means or Proportions of Demographic Characteristics and Bargaining Variables

Variables Mean (Std. Dev.) Minimum Maximum

Female .64 Age 19.12 (1.26) 18 23

Bargaining Phase 1 Profit (All Rounds) 22.93 (4.87) 8 28 Profit (Last 10 Rounds) 23.80 (5.12) 8 29 Forfeits .29 Timeouts .21

Bargaining Phase 2 Profit (All Rounds) 23.52 (5.03) 7 28 Profit (Last 10 Rounds) 24.09 (4.91) 8 28

Forfeits .27 Timeouts .15

P(s) .66 .35 1.00 Note: Sample size for this analysis = 75.

Bargaining Differences

Overall the average profit increases over the course of each bargaining phase

according to t-test analyses of the average differences. When comparing the first few

rounds to the last ten rounds, there is a significant increase in profit in both the first

phase of negotiations (t = -4.99, p < .000, two-tailed test) and the second (t = -5.51, p

< .000, two-tailed test) (results not shown). There are no significant differences in the

number of forfeits or timeouts by condition (results not shown). Table 2 displays the

differences in the two phases of bargaining between the two master conditions. Profit

does not differ by status difference in either the first or second rounds of negotiations;

therefore, participants did not behave differently toward their partners by condition.

Page 38: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

28

Regardless of the qualities of participants in either phase of the study, they

appear to “test the waters” with their partners to see how many chips they can garner

from them. This behavior produces the only significant result related to profit: the

profit in the final rounds of negotiation is significantly higher than in the beginning. It

is important to note that since profit did not differ by status level, participants did not

develop any consistent structural advantages in any of the conditions. Additionally,

because the participants’ negotiating behavior did not vary by condition, any

differences in influence and status value in the second phase could not have been the

result of a behavioral influence hierarchy present during the exchanges.

Table 2.2. Mean Bargaining Profit Differences by Reward Level

Rewards: Start with High Rewards End with High Rewards Variables Mean T-Statistic Mean T-Statistic Bargaining Phase 1 Profit (All Rounds)

Low Rewards 22.69 -.55

High Rewards 23.30

Bargaining Phase 1 Profit (Last 10 Rounds)

Low Rewards 23.69 -.32

High Rewards 24.06

Bargaining Phase 2 Profit (All Rounds)

Low Rewards 23.32 -.36

High Rewards 23.74

Bargaining Phase 2 Profit (Last 10 Rounds)

Low Rewards 23.86 -.50

High Rewards 24.42

Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 39: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

29

Behaviorally, participants attempted to garner as many chips as the simulated

partner would allow. The status of the chips did not alter this pattern. This might

indicate that the participants only thought the chips had pecuniary value. Their

assessments of the monetary value of the chips suggest that this is not the case. This

appraisal hovers around the midpoint of the scale, which indicates that the participants

were unsure of whether the chips had monetary value at all.

In the first bargaining phase, a relevance bond is created between the status

characteristics of the participants and partners and the chip color, whereas in the

second this link is formed between the chip color and the nominal characteristic,

personal response style. It is important to therefore assess whether the participants’

valuation of the chips persists throughout these two phases. As the t-test analyses in

Table 3 demonstrate, this valuation does tend to endure. When participants begin the

study with a status advantage, they view their chips as having significantly higher

value than those controlled by their partner (t = 2.59, p < .05, two-tailed test); whereas

when the initial partners is of higher status than the participants, the participants rate

their own chips as having lesser value than those of their partners (t = -2.44, p < .05,

two-tailed test). This demonstrates that the chips obtain reward valuations consistent

with the status differences apparent between the participants and their first partners.

In the second phase, the valuation of the resources is not quite as distinct. The

participants who end the study with the devalued reward also view their chips as

having less value than those of their second partners (t = -1.86, p < .10, two-tailed

test), but there is no significant difference in the chip valuation of those who end the

study with the chip that had high valuation in the first phase. These results also

Page 40: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

30

support the first hypothesis, although those who have a reward advantage at the end of

the study do not tend to acknowledge this advantage.

Participants also gave their impressions of how they expect most people would

perceive those that possessed the two different chips, thereby assessing what the

rewards indicate about their possessors. These questions were coded according to

whether the two types of chips were associated with either high or low status in the

first phase of the study. As displayed in Table 4, chips that are purported to have a

high reward level convey a greater degree of status and competence (t = 3.50, p <

.000 and t = 3.03, p < .01 respectively, two-tailed tests) about those who control them.

These results also support the first hypothesis.

Table 2.3. Mean Ratings of the Value of the Rewards Relative to In-Group Status (with Standard Deviations)

Beginning Reward Group Level End Reward Group Level

Variables High Low High Low

Overall Value

Own Group 63.12 (24.95) 51.68 (26.54) 54.30 (24.41) 48.06 (24.86)

Other Group 47.71 (28.28) 62.29 (24.38) 55.25 (26.54) 56.89 (22.76)

Difference 15.41 (34.73)* -10.61 (27.81)* -.95 (21.30) -3.88 (28.15)+

Note: Sample size for this analysis = 75; t-test analysis; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 41: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

31

Table 2.4. Mean Ratings of “Most People’s” Evaluations of the Rewards’ Possessors by End Reward Level

Variables Mean T-Statistic

Status Scale

High Reward Level 62.69 3.59***

Low Reward Level 55.75

Competence Scale

High Reward Level 64.08 3.03**

Low Reward Level 58.62

Considerateness Scale

High Reward Level 56.76 -.93

Low Reward Level 58.43

Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Overall, this analysis generally shows support for the first hypothesis that once

the chips become relevant to the various status characteristics apparent in the first

phase of the study, the chips come to have differential levels of status value. This also

replicates the general findings of previous status value investigations (Thye 2000;

Thye and Witkowski 2003). Importantly, these valuations persist over the course of

the study and across conditions. Only those who are benefitted by their rewards in the

second phase do not fully acknowledging their advantage. Generally, however, these

results indicate the strength of these beliefs regarding the chips’ valuation.

Status Value of Personal Response Style

In addition to perceiving that the chips have differential levels of status value,

participants also tend to differentially value the previously unknown characteristic,

Page 42: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

32

personal response style, according to predicted reward dissimilarities. To assess the

status value of the personal response style groups, status, competence, and

considerateness scales were created from the exit questionnaire5. The expectation is

that when participants have high rewards during the second phase of the study, they

will evaluate the third-order beliefs about how most people would evaluate their own

personal response style as being higher in status and competence but lower in

considerateness than the other response style; conversely, those with devalued rewards

are expected to report that their own group has lower status and competence but higher

considerateness according to what most people believe. Analyses of the average

differences between the study’s conditions generally support this conclusion (see

Table 5).

5 The alpha-levels of these scales range from .77 to .89.

Page 43: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

33

As predicted, those with high rewards see their own state of the characteristic

as having higher status (t = 2.69, p < .01, two-tailed test) and competence (t = 2.04, p

< .05, two-tailed test) but lower considerateness (t = -2.93, p < .01, two-tailed test).

Those with devalued rewards do not report there to be a significant difference between

the status and competence of the response styles, though the relationship is in the

predicted direction. Additionally, they view their group as being slightly more

considerate (marginally significant, two-tailed test). These results generally indicate

that participants believe most people would say that states of the novel characteristic

Table 2.5. Mean Ratings of “Most People’s” Evaluations of Personal Response Style Relative to In-Group Status (with Standard Deviations)

Final Reward Level

Variables Ends with High

Rewards Ends with Low

Rewards

Status Scale

Own Group 61.69 (14.28) 59.74 (9.23)

Other Group 55.29 (13.48) 58.25 (9.47)

Difference 6.39 (15.01)** -1.49 (9.62)

Competence Scale

Own Group 62.16 (16.05) 61.88 (11.47)

Other Group 58.83 (15.05) 62.04 (10.82)

Difference 3.33 (10.35)* -0.16 (8.92)

Considerateness Scale

Own Group 56.27 (13.82) 59.83 (11.03)

Other Group 61.98 (13.76) 55.75 (12.71)

Difference -5.71 (12.33)** 4.08 (14.15)+

Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 44: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

34

have differential status value by reward level, suggesting that personal response style

has become, or at least is becoming, a status characteristic.

When the estimations of the response styles are assessed more inclusively by

the reward-level each state was associated with at the end of the study, the resulting

averages are also in the predicted direction (see Table 6). Participants report that most

people believe the response style associated with valued rewards has more status and

competence (both marginally significant, two-tailed tests) but lower considerateness (t

= -3.25, p < .01, two-tailed test) than the reward-disadvantaged response style. These

results tend to support the second hypothesis.

Table 2.6. Mean Ratings of “Most People’s” Evaluations of Personal Response Style by End Reward Level

Variables Mean T-Statistic

Status Scale

High Reward Level 60.08 1.77+

Low Reward Level 57.37

Competence Scale

High Reward Level 62.10 1.64+

Low Reward Level 60.25

Considerateness Scale

High Reward Level 56.03 -3.25**

Low Reward Level 60.97

Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 45: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

35

Turning to personal assessments of the relative status, competence, and

considerateness of the two personal response styles, the expectation is again that those

who are highly rewarded will see their style as being higher in status and competence

but lower in considerateness, whereas the opposite is expected for those with a reward

disadvantage6. Table 7 presents the t-test analyses of this prediction. Similarly to the

results for the third-order beliefs, highly rewarded participants’ personal assessments

of the novel characteristic are in the expected direction. These participants believe

that their response style is higher in status (t = 3.14, p < .01, two-tailed test) and

competence (t = 2.58, p < .05, two-tailed test) than the opposing style. They also

view their style as having lower considerateness, but this result is not statistically

significant. Conversely, those with devalued rewards view their own response style as

having slightly higher status (marginally significant, two-tailed test). This result runs

contrary to predictions and is, interestingly, also inconsistent with the general direction

of their third-order beliefs. This suggests that they acknowledge that there is a

common, accepted belief that their style is of lower status, but they do not personally

subscribe to this belief. The difference between assessments of considerateness is

marginally significant (two-tailed test), and is, however, in the predicted direction.

6 The alpha-levels for these scales range from .76 to .93.

Page 46: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

36

When the personal evaluations are evaluated by reward level alone, we see that

personal evaluations are not as differentiated by condition as the assessments of third-

order beliefs (see Table 8). As expected, the status and competence of the advantaged

response style is higher than that of the disadvantaged state, but these results are only

marginally significant in a one-tailed test (p < .07 and p < .06, respectively). Personal

assessments are open to social-desirability bias as it is not always appropriate to state

that one has an advantage, as is possibly also the case when those with high rewards in

the second phase stated that their chips did not have significantly higher status value.

It also may be easier for those who are reward-disadvantaged to state that they believe

Table 2.7. Mean Ratings of Personal Evaluations of Personal Response Style Relative to In-Group Status (with Standard Deviations)

Final Reward Level

Variables Ends with High

Rewards Ends with Low

Rewards

Status Scale

Own Group 64.43 (14.52) 61.94 (11.54)

Other Group 57.38 (14.24) 58.84 (10.81)

Difference 7.05 (14.19)** 3.10 (10.69)+

Competence Scale

Own Group 65.98 (15.75) 65.19 (12.49)

Other Group 60.68 (16.63) 63.65 (12.68)

Difference 5.29 (12.97)* 1.54 (9.40)

Considerateness Scale

Own Group 59.88 (15.67) 61.94 (12.88)

Other Group 60.12 (13.19) 58.32 (13.81)

Difference -0.24 (12.33) 3.62 (13.55)+

Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 47: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

37

their characteristic to not also be devalued. It is important to note, however, that even

though these personal assessments are not very distinct, the participants report that

“most people” do see these differences, thereby acknowledging the existence of a

broader social reality attached to the states of the novel characteristic.

Table 2.8. Mean Ratings of Personal Evaluations of Personal Response Style by End Reward Level

Variables Mean T-Statistic

Status Scale

High Reward Level 61.82 1.47+

Low Reward Level 59.50

Competence Scale

High Reward Level 64.89 1.53+

Low Reward Level 62.79

Considerateness Scale

High Reward Level 59.15 -1.22

Low Reward Level 60.97

Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001 (one-tailed tests).

Overall, these results tend to indicate that the novel characteristic obtained

status value and partially support Hypothesis 2. In addition to social desirability bias,

another possible explanation for the lack of significant results could be that these

assessments may have been affected by the bargaining behavior of the simulated

partners. In all conditions and phases, the partners evidenced poor bargaining skills,

as the participants were able to extract the same amount of chips from both of their

partners regardless of purported status or reward differences. Thus, before rating the

Page 48: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

38

novel characteristic, the participants had just come from a situation in which their

second partners’ competence at bargaining was demonstratively deficient. The

partners’ modest bargaining ability could have generalized to the assessments of the

novel characteristic’s competence and status. Nevertheless, even if this was the case,

there is still evidence to suggest that those who ended the study with devalued rewards

stated that both markers convey a differential degree of status to those who possess

them.

Influence

Do the relative reward levels lead to expectation advantages and the enactment

of a new status hierarchy based off of the states of the novel characteristic? The

results indicate that this is the case as those with a reward advantage deferred less

frequently to their partner than those with a reward disadvantage (t = -2.00, p < .05,

two-tailed test) (see Table 9). Importantly, the measure of influence does not

significantly vary by gender, personal response style, or the reward level with which

the participants began the study. ANOVA analyses also show that the final reward

level is the only significant predictor of the variation in the proportion of stay

responses (F = 5.01, p < .05) (see Table 10). Thus, although the valuation of the

response styles was not always sharply defined, participants did behave differently by

the final reward-level.

Page 49: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

39

Table 2.9. Mean Difference in the Proportion of Stay Responses

P(s)

Variables Mean T-Statistic

Gender

Male .69 1.26

Female .64

Personal Response Style

S2 .65 -.09

Q2 .66

Beginning Rewards

Low Rewards .65 -.51

High Rewards .66

End Rewards

Low Rewards .62 -2.00*

High Rewards .69 Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 50: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

40

Table 2.10. Proportion of Stay Responses ANOVAs

Model 1 Model 2 Model 3 Model 4 Variables SS MS F SS MS F SS MS F SS MS F

Main Effects Female .04 .04 1.56 .03 .03 1.33 .03 .03 1.17 .00 .00 .15 Personal Response Style (PRS) .00 .00 .14 .00 .00 .00 .00 .00 .11 .01 .01 .43 Start with High Rewards .00 .00 .11 .00 .00 .06

End with High Rewards .08 .08

3.34+ .12 .12

5.01* Interactions Start High x Female .01 .01 .22 Start High x PRS .00 .00 .05 End High x Female .01 .01 .50 End High x PRS .02 .02 .69

Female x PRS .02 .02 .92 .08 .08

3.28+

Residual 1.74 .02 1.70 .03 1.66 .02 1.58 .02

R2 .03 .04 .07 .11 Note: Sample size for this analysis = 75; +p < .10; *p < .05; **p < .01; ***p < .001.

Page 51: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

41

There is an important caveat, however, as to whether this fully supports the

third hypothesis. Because personal response style is always associated with reward

differences in the second phase of this study, the influence disparity may have been

due simply to these reward inequalities. There is no ultimate way of showing that the

new characteristic obtains enough valuation to independently cause these behavioral

differences. This is especially a concern given the moderate support for hypothesis 2

in that participants generally only formed third-order beliefs about the novel

characteristic. Not being able to know for certain whether participants would act

according to only their beliefs about the new characteristics in the absence of reward

disparities is a flaw in this study's design, and one that should be rectified in future

research.

DISCUSSION

The results from this study show preliminary support for the argument that

status markers have the power to create altogether new status groups and structures of

behavioral inequality. The proposed spread of status value mechanism has multiple

stages in that the status value of preexisting status characteristics first create stable

reward disparities before the rewards’ status value can spread to novel characteristics.

Cleanly manipulating and teasing apart each step in this process is a difficult

experimental task, and one at which this study has only partially succeeded.

This design’s main strength is that it divorced the rewards created by the initial

status characteristic differences from the novel characteristic introduced in the second

phase. This was accomplished by having two unique partners for the participant to

Page 52: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

42

interact with and by reversing the reward levels in half of the conditions between the

first and second phase. Additionally, the two simulated partners were programmed to

use the same bargaining rules in the first and second phase of the study, which means

that the partners did not convey divergent behaviors that could have caused the

resulting influence differences. Thus, there could not have been a transfer of

behavioral inequalities between the first and second phase across all of the conditions

(Markovsky, Smith, and Berger 1984; Pugh and Wahrman 1983).

The negotiations may have introduced a different type of behavioral dynamic,

however. Across conditions and phases, not only did participants negotiate similarly

with their partners, they were also quite successful at obtaining large amounts of chips

from them. Participants’ competence assessments may have been confounded due the

association between the partners’ feeble bargaining skills, their associated reward

level, and personal response style.

One of the biggest faults of this study’s design is that final reward levels were

never separated from the nominal characteristic. While the novel characteristic did

obtain a degree of third-order status evaluations, it is unclear whether these valuations

are enough to produce behavioral inequalities without the aid of associated reward

disparities. It is therefore uncertain whether the clear influence differences revealed

with this design could be found if the partners only differed by the newly valued

characteristic.

A future study could rectify these two main failings by eliminating the two

phases of negotiated exchange and by including a final phase wherein the participants

interact with a new partner who only differs from the participants by the state of the

Page 53: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

43

novel characteristic. To create the reward, participants could view a document that

lists past participants and their extant status characteristics. This information would be

color-coded such that those who have high status are highlighted with one color and

those who have lower status are marked with an opposing color. The study would also

have two phases wherein the participant works on an influence task, such as a joint

meaning insight task, with a first partner who differs from the participant by their

reward level and state of a novel characteristic. In the second phase, the participants

would obtain a new partner, presumably from a separate participant pool, who only

differs from them by the state of the novel characteristic. They would then take a

different influence task with the second partner, such as the contrast sensitivity task.

If influence inequalities are found in this second phase, we would be more certain that

these differences are due to the status value of the new characteristic alone.

CONCLUSION

In everyday interaction, we are defined not only by our master statuses, such as

our gender and ethnic backgrounds, but also by our awards and positions, which may

have differential social value. Many of these distinctions combine to create an

understanding of who we are, what kinds of behaviors can be expected of us, how

valuable our contributions should be, and the like. These expectations undergird our

shared and accepted social reality, with many using status and reward distinctions as

shorthand for assessments of competence and worth.

Veblen ([1899] 2005) was one of the first to argue that honorific goods have

social significance and are highly esteemed, sought after possessions. Since then,

Page 54: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

44

researchers have continued to note the relative importance these status markers have

on social interaction (Bourdieu 1990, 1991), namely how their differential allocation

impacts assessments of status, competence, importance, and influence (Cook 1975;

Ridgeway 1991, 1997; Stewart and Moore 1992). This project expands this line of

research by examining how the underlying meanings conveyed by differential rewards

impact the creation of new status groups.

This paper puts forth a new extension of status construction theories, more

specifically Berger and Fisek (2006), by using the status value of objects as an

additional feature in the process of status creation. This study tests whether status

value spreads from preexisting status characteristics to initially non-valued resources

(see Thye 2000) and then onto a nominally distinct characteristic of the new

possessors of this object. In the absence of any distinguishing information about an

interaction partner, except that she controls status valued rewards and has a different,

initially non-valued characteristic, the results generally indicate that the states of the

nominal characteristic will gain status and influence through their association with the

reward. Notably, even after this new association is made between the reward and the

new status characteristic, the results also suggest that we will continue to evaluate

additional, unspecified possessors of this reward in the same fashion.

We therefore process the valuations conveyed by status markers in the same

manner as status characteristics, and under certain conditions differential rewards can

create new status distinctions with resulting behavioral expectations. Rewards can

play an important part in the status construction process in that groups do appear to

obtain differential levels of status and influence from the status markers they display.

Page 55: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

45

References

Alves, Wayne M. and Peter H. Rossi. 1978. “Who Should Get What? Fairness Judgments of the Distribution of Earnings.” American Journal of Sociology 84:541-64.

Bedeian, Arthur G. and Hubert S. Feild. 1980. “Academic Stratification in Graduate Management Programs: Departmental Prestige and Faculty Hiring Patterns.” Journal of Management 6:99-115. Bedeian, Arthur G., David E. Cavazos, James G. Hunt, and Lawrence R. Jauch. 2010.

“Doctoral Degree Prestige and the Academic Marketplace: A Study of Career Mobility Within the Management Discipline.” Academy of Management Learning and Education 9:11-25.

Berger, Joseph, Bernard P. Cohen, and Morris Zelditch Jr. 1966. “Status

Characteristics and Expectation States.” Pp. 29-46 in Sociological Theories in Progress, vol. 1, edited by J. Berger, M. Zelditch Jr., and B. Anderson. Boston: Houghton Mifflin.

------. 1972. “Status Characteristics and Social Interaction.” American Sociological

Review 37:241-55. Berger, Joseph and Thomas L. Conner. 1969. “Performance-Expectations and Behavior in Small Groups.” Acta Sociologica 12:186-98. ------. 1974. "Performance Expectations and Behavior in Small Groups: A Revised

Formulation." Pp. 85-109 in Expectation States Theory: A Theoretical Research Program, edited by J. Berger, T. L. Conner, and M. H. Fisek. Cambridge, MA: Winthrop.

Berger, Joseph and M. Hamit Fisek. 2006. “Diffuse Status Characteristics and The Spread of Status Value: A Formal Theory.” American Journal of Sociology 111:1038-79. Berger, Joseph, M. Hamit Fisek, Robert Z. Norman, and Morris Zelditch Jr. 1977.

Status Characteristics and Social Interaction: An Expectation States Approach. New York: Elsevier.

Berger, Joseph, Morris Zelditch Jr., Bo Anderson, and Bernard P. Cohen. 1972.

"Structural Aspects of Distributive Justice: A Status Value Formulation." Pp. 119-46 in Sociological Theories in Progress, vol. 2, edited by J. Berger, M. Zelditch Jr., and B. Anderson. Boston: Houghton Mifflin.

Berger, Peter L. and Thomas Luckmann. 1966. The Social Construction of Reality.

Page 56: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

46

New York: Doubleday. Bierhoff, Hans W., Ernst Buck, and Renate Klein. 1986. “Social Context and

Perceived Justice.” Pp. 165-185 in Justice in Social Relations, edited by H. W. Bierhoff, R. L. Cohen, and J. Greenberg. New York: Plenum.

Bourdieu, Pierre. 1990. In Other Words: Essays Towards a Reflexive Sociology. Stanford, CA: Stanford University Press. ------. 1991. Language and Symbolic Power. Cambridge, MA: Harvard University Press. Carson, Richard and Peter Navarro. 1988. “A Seller’s (and Buyer’s) Guide to the Job

Market for Beginning Academic Economists.” Journal of Economic Perspectives 2:137-48.

Conway, Michael, M. Teresa Pizzamiglio, and Lauren Mount. 1996. “Status,

Communality, and Agency: Implications for Stereotypes of Gender and Other Groups.” Journal of Personality and Social Psychology 71:25-38.

Cook, Karen S. 1975. "Expectations, Evaluations, and Equity." American Sociological Review 40:372-88. Cook, Karen S. and Richard M. Emerson. 1978. “Power, Equity and Commitment in Exchange Networks.” American Sociological Review 43: 721-39. Crane, Diana. 1965. “Scientists at Major and Minor Universities: A Study of Productivity and Recognition.” American Sociological Review 30:699-714. Fiske, Susan T., Amy J. C. Cuddy, Peter Glick, and Jun Xu. 2002. “A Model of (Often

Mixed) Stereotype Content: Competence and Warmth Respectively Follow from Perceived Status and Competition.” Journal of Personality and Social Psychology 82:878-902.

Harrod, Wendy J. 1980. “Expectations from Unequal Rewards.” Social Psychology Quarterly 43:126-30.

Hunt, James G. and John D. Blair. 1987. “Content, Process, and the Matthew Effect Among Management Academics.” Journal of Management 13:191-210. Lawler, Edward J. and Jeongkoo Yoon. 1996. “Commitment in Exchange Relations:

Test of a Theory of Relational Cohesion.” American Sociological Review 61: 89-108.

Lennon, Sharon J. 1986. “Additivity of Clothing Cues in First Impressions.” Social

Page 57: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

47

Behavior and Personality 14:15-21. Lerner, Melvin J. 1965. “Evaluation of Performance as a Function of Performer’s

Reward and Attractiveness.” Journal of Personality and Social Psychology 1:355-60.

Long, J. Scott, Paul D. Allison, and Robert McGinnis. 1979. “Entrance into the Academic Career.” American Sociological Review 44:816-30. Markovsky, Barry, Le Roy F. Smith, and Joseph Berger. 1984. “Do Status Interventions Persist?” American Sociological Review 49:373-82. Merton, Robert K. 1968. “The Matthew Effect in Science: The Reward and Communication Systems of Science Are Considered.” Science 159:56-63. ------. 1988. “The Matthew Effect in Science, II: Cumulative Advantage and the Symbolism of Intellectual Property.” Isis 79:606-23. Molm, Linda D., Gretchen Peterson, and Nobuyuki Takahashi. 2001. “The Value of Exchange.” Social Forces 79:159-85. Patzer, Gordon L. 1985. The Physical Attractiveness Phenomena. New York: Plenum Press. Pugh, Meredith D. and Ralph Wahrman. 1983. “Neutralizing Sexism in Mixed-Sex

Groups: Do Women Have to Be Better than Men?” American Journal of Sociology 88:746-62.

Quereshi, M. Y. and Janet P. Kay. 1986. “Physical Attractiveness, Age, and Sex as

Determinants of Reactions to Resumes.” Social Behavior and Personality 14:103-12.

Rhode, Deborah L. 2010. The Beauty Bias: The Injustice of Appearance in Life and Law. New York: Oxford University Press. Ridgeway, Cecilia L. 1991. "The Social Construction of Status Value: Gender and Other Nominal Characteristics." Social Forces 70:367-86. ------. 1997. "Where Do Status Beliefs Come From? New Developments." Pp. 137-58

in Status, Networks, and Structure: Theory Development in Group Processes, edited by J. Szmatka, J. Skvoretz, and J. Berger. Stanford, CA: Stanford University Press.

------. 2006. “Status Construction Theory.” Pp. 301-23 in Contemporary Social

Page 58: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

48

Psychological Theories, edited by P. J. Burke. Stanford, CA: Stanford University Press.

Ridgeway, Cecilia L., Kristen Backor, Yan E. Li, Justine E. Tinkler, and Kristan G.

Erickson. 2009. “How Easily Does a Social Difference Become a Status Distinction? Gender Matters.” American Sociological Review 74:44-62.

Ridgeway, Cecilia L. and James Balkwell. 1997. “Group Processes and the Diffusion of Status-Value Beliefs.” Social Psychology Quarterly 60:14-31. Ridgeway, Cecilia L., Elizabeth Heger Boyle, Kathy J. Kuipers, and Dawn T.

Robinson. 1998. "How Do Status Beliefs Develop? The Role of Resources and Interactional Experience." American Sociological Review 63:331-50.

Ridgeway, Cecilia L. and Kristan Glasgow Erickson. 2000. "Creating and Spreading Status Beliefs." American Journal of Sociology 106:579-615. Stewart, Penni A. and James C. Moore, Jr. 1992. "Wage Disparities and Performance

Expectations." Social Psychology Quarterly 55:70-85. Tajfel, Henry. 1978. Differentiation between Social Groups: Studies on the Social Psychology of Intergroup Relations. New York: Academic Press. Thye, Shane R. 2000. "A Status Value Theory of Power in Exchange Relations."

American Sociological Review 65:407-32. Thye, Shane R. and Christine Witkowski. 2003. "The Status Value Theory of Power:

The Effect of Status and Resource Differentiation on Power in Exchange." Paper presented at the annual meeting of the American Sociological Association, Atlanta, August.

Veblen, Thorstein. [1899] 2005. The Theory of the Leisure Class. Gloucestershire, UK: Dodo Press. Weber, Max. [1916] 1946. From Max Weber: Essays in Sociology. Translated and

edited by H. H. Gerth and C. W. Mills. Reprint, New York: Oxford University Press.

Page 59: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

49

3. THE PERPETUATION OF INEQUALITY: STATUS DYNAMICS IN

LENDING MARKETS

The accumulation of wealth continues to be one of the predominant economic

disparities groups experience. One of the key ways that we begin to accumulate

wealth is first through the receipt of credit as people generally do not have the funds

needed to purchase houses, businesses, tuition, and the like outright. However, the

distribution of credit, such as in the mortgage and business loan market, can be fraught

with bias (Carter, Brush, Greene, Gatewood, and Hart 2003; Munnell, Tootell, and

McEneaney 1996; Oliver and Shapiro 1995; Ross and Yinger 1999, 2002). What

remains unclear is whether there are possible interpersonal mechanisms that are a

source of such bias. This study is an examination of one such possible mechanism:

how assessments of status and trustworthiness systematically bias funding allocations

in lending markets, and, thereby, the future accumulation of new wealth.

Past research documents the presence of disparities in a wide array of lending

markets, including the receipt of mortgage and business loans (Munnell et. al 1996;

Ross and Yinger 1999, 2002; Williams, Nesiba, and McConnell 2005). Using a wide

array of statistical and methodological techniques, researchers have demonstrated the

persistence of a systematic minority-white gap in the receipt of housing loans

(Blackburn and Vermilyea 2006; Munnell et al. 1996; Ross and Yinger 1999, 2002).

According to data from the 1991 Home Mortgage Disclosure Act, African Americans

and Hispanics are about 82% less likely to receive housing loans than comparable

whites (Munnell et al. 1996). Ross and Yinger (2002) conclude that legitimate

Page 60: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

50

business practices cannot fully account for this disparity.

Additionally, audit studies demonstrate that black testers are less likely to

receive a quote for a loan, are often quoted higher interest rates, are given less

coaching from loan officers than white testers, and loan officers are less likely to look

for factors to justify accepting their loans (Lawton 1996; Siskin and Cupingood 1996;

Smith and Cloud 1996; Smith and DeLair 1999; Williams et al. 2005). Existing

evidence therefore suggests that lenders’ underwriting standards systematically differ

by the racial and ethnic background of the applicants (Ross and Yinger 2002).

Racial and ethnic minorities are also more likely to receive subprime loans,

which are mortgages associated with higher rates, fees, and closing costs, and,

consequently, greater risks to the borrower (Apgar & Calder 2005; Calem, Gillen, and

Wachter 2004; Immergluck & Wiles 1999; Williams et al. 2005). This appears to be

at least partially due to the conduct of the lenders themselves. For instance, audit

studies by Williams et al. (2005) demonstrate that white applicants are more likely to

be led to lenders’ prime mortgage division than African American applicants.

While minorities generally receive subprime loans overall, these loans also tend

to be concentrated in minority housing areas, thereby leading to reverse redlining.

This creates a “dual mortgage market” (Immergluck and Wiles 1999) where favorable

mortgages are concentrated in white areas and subprime loans in minority

neighborhoods. Lest we think that these findings are really the result of an underlying

income effect, research suggests that even at the highest income level, African

Americans are about three times more likely to receive subprime mortgages than

comparable whites (Calem et al. 2004; Williams et al. 2004).

Page 61: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

51

Beyond mortgage markets, racial and ethnic minorities are also

disproportionately more likely to receive additional dealer markups on car loans and

generally pay a greater price for the vehicles (Ayres 1995; Ayres & Siegelman 1995;

Charles, Hurst, and Stephens 2008; Cohen 2007). For example, even after controlling

for applicants’ background information, bargaining strategy, and credit background,

Ayres and Siegelman (1995) have shown that dealers less flexible in their negotiations

with African American consumers, which leads to higher car prices. There is also

evidence to suggest that although women begin over half of the startup companies in

the United States, only about 7% of their businesses receive venture capital funding

(Brush, Carter, Gatewood, Greene, and Hart 2004; Carter et al. 2003; Greene, Brush,

Hart, and Saparito 2001).

In recent years, new lending markets have been created that allow individuals

to request loans from others without using traditional banking institutions as an

intermediary. Known as peer-to-peer lending markets, these arenas provide an online

forum in which individuals publicly display their lending requests and credit

information in the hopes of finding lenders who will finance portions of the loan.

Although peer-to-peer lending markets are a relatively nascent, they already account

for nearly half a billion dollars in loans, and this figure continues to grow. Peer-to-

peer markets, such as prosper.com and lendingclub.com, tout this funding method as

one free from the restrictions and uncertainties of traditional banking institutions,

wherein the lack of overhead costs leads to lower service fees, interest rates are

competitively set, and the only barrier to obtaining a loan are the judgments of fellow

peers. In these markets, those who desire loans are able to tell their own story

Page 62: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

52

regarding who they are and the reasons for their request, in addition to providing more

conventional information related to their credit history.

While the ability to give such information may make the applicants feel as

though they are not devoid of personal identity, thereby creating a greater sense of

control over their self-presentation and connection to potential lenders, these stories

also convey many indicators on which bias and discrimination can be based. Much as

in traditional credit markets, lenders are not only examining the credit histories and

basic loan information, which in theory should be the only things they use in making

their decisions, but lenders can also see pictures and personally written descriptions

detailing the purposes of the loan and who the applicants’ are. Recent research into these peer-to-peer markets indicates that there are several

processes that impact the likelihood of receiving funding, in addition to those that are

purely financial. Ravina (2008), for instance, finds that there is a large beauty

premium in that those who are rated as more beautiful in their application profiles are

more likely to receive loans. In an important work examining racial discrimination

apparent in prosper.com’s market, Pope and Sydnor (2011) conclude that, controlling

for credit history, African Americans are 25-35 percent less likely to receive loans,

and, when they do, their interest rates are significantly higher than those of

comparable whites.

This brief summary of lending market disparity brings us to the question of

why do these disparities persist. If the credit histories and financial worthiness of

applicants are held constant why do we continue to see these inequalities in the receipt

of funding across a wide array of arenas? And is this simply the effect of one factor,

Page 63: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

53

such as gender or race, or do the various demographic characteristics that borrowers

exhibit act in concert to influence lenders’ decisions?

The evidence suggests that these disparities are at least partially due to lender

discrimination. Some of the possible causes of these disparities may be general

prejudice and stereotypes of the lenders’ (see Pager and Shepherd 2008 for a review).

The organizational conditions of the lending environment may also amplify the

opportunities to discriminate wherein the very rules of the organization may be

interpreted or written in such a way that they favor those who are already advantaged.

While many researchers have proposed many potential explanations for why lenders

appear to discriminate, we do not yet know whether and how any of these beliefs

translate into action (Pager and Shepherd 2008). Uncovering the presence of

discrimination does not allow us to then infer back to account for what led someone to

behave in that fashion, and this leaves us with the important question of what the

causal mechanisms may be.

I propose that lenders use status as an indicator of creditworthiness, such that

even among those with comparable credit histories, borrowers’ status characteristics

further differentiate applicants. These characteristics insinuate differential levels of

competence and trustworthiness, which affect applicants’ funding. This study,

therefore, addresses one potential path through which preexisting status divisions and

social inequalities perpetuate themselves by predicting that those who are advantaged

by only their characteristics, including their ascribed traits, will have their economic

and social circumstances enhanced by virtue of their status distinctions. By examining

the decision-making processes behind lending, this project also helps illuminate how

Page 64: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

54

implicit bias and assessments of competence that vary by status groups, such as by

diffuse racial and gender categories, affect funding decisions in traditional credit

markets more broadly. Thus, this research seeks to expand and give theoretical

context to this line of research by examining how indicators of status bias the

likelihood of receiving loans and at what amount.

STATUS AND ASSESSMENTS OF CREDITWORTHINESS

The central argument of this paper is that status distinctions systematically bias

assessments of creditworthiness and are one of the mechanisms driving lending

discrimination. Credit applicants’ status characteristics help to form a prevailing

impression of their competence and trustworthiness to lenders. These perceptions

guide the decision whether to fund the applicant and to what extent even for those of

comparable credit histories. To further explicate the role of status characteristics and

reward distributions, this discussion will now turn to a discussion of the relevant

theory and research on status processes.

Status characteristics theory (hereafter SCT) elucidates how existing status

differences pattern power and prestige behavior in small groups (Berger 1958; Berger,

Cohen and Zelditch 1966, 1972). There are two main types of personal characteristics

that can distinguish group members: diffuse and specific status characteristics.

Diffuse status characteristics are the ubiquitous, socially significant characteristics of a

given culture (e.g., gender or race/ethnicity) that have varying states (e.g., male-

female, white-African American). Specific status characteristics are associated with

the ability to perform particular tasks, such as having differential computer skills or

Page 65: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

55

business aptitude. The various states of either type of characteristic have differential

esteem, prestige, and competence valuations as defined by the dominant culture that

correspond to the level of performance ability a person with a particular state is

assumed to have (Berger, Fisek, Norman, and Zelditch 1977).

Once a characteristic differentiates people in a setting or is relevant to the task,

group members will develop differential expectations related to the performance of the

each person according to whether the states of their status characteristics are esteemed

or devalued. Differential performance expectations pertain to which person the group

expects to be more competent at the task and make more valuable contributions.

These expectations then implicitly direct behavior and determine outcomes and

evaluations in a self-fulfilling manner.

In addition to affecting assessments of worth and competence and behavioral

expectations, the distribution of rewards and resources is also influenced by status

distinctions. Reward expectations theory (hereafter RET), (Berger, Fisek, Norman,

and Wagner 1983, 1998), a theoretical branch of SCT, predicts that individuals,

regardless of their relative status, will allocate larger amounts of valued resources to

high-status actors, while low-status actors will be allocated fewer rewards or more

devalued objects. The more divergent the status differences between actors are, the

greater the reward inequalities will be. The reverse process also operates so that those

who are allocated more rewards in a task situation (e.g., those who are paid more)

come to have more influence in the group (Bierhoff, Buck and Klein 1986; Cook

1975; Harrod 1980; Lerner 1965; Stewart and Moore 1992).

Page 66: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

56

Culturally embedded referential structures govern this allocation process.

Referential structures are sets of widely held beliefs regarding how reward levels are

normatively associated with status-valued characteristics. RET recognizes three

different types of referential structures: categorical, ability, and outcome structures.

Categorical structures pertain to the relationship between diffuse status characteristics

and rewards. Ability structures relate specific task abilities and the determination of

reward levels. Outcome structures associate past achievement with rewards. These

three types of referential structures together form a basis for allocation norms, and

situations differ with regards to which structures are more relevant in the regulation of

reward distributions. For example, survey research indicates that there is a high-

degree of consensus in American society that earnings across a wide variety of

occupations should be based on at least outcome and ability structures (Alves and

Rossi 1978; Jasso and Rossi 1977).

Referential structures affect reward allocations when differential states of at

least one status characteristic are salient in an interaction and these states are the basis

of a normative referential structure. If at least one status characteristic is salient in an

interaction and if this characteristic is included in a referential structure, this reward

structure will also become salient. Expectations about “who should get what” are then

relevant in the situation when reward allocations are made. For instance, consider a

group whose members must make funding decisions. Further assume that educational

differences are apparent in the target group, and, consequently, status differences

based on educational background are salient in the situation. If a referential structure

also exists within the culture that associates higher educational attainment with higher

Page 67: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

57

reward levels, this referential structure will become relevant to the situation. Group

members will then form reward expectations in line with the referential structure and

should fund accordingly.

As related to lending markets, people must evaluate a series of loan applications

and decide whom among them to fund. Within any profile, multiple status indicators

may be present and salient, such as gender, race/ethnicity, education level, and writing

aptitude, among others. Lenders should form reward expectations in line with the

referential structure and, therefore, will be more likely to fund applicants of higher

overall status. Because these applicants are assumed to be highly competent and more

able to repay the loan, high status applicants have an advantage over low status

applicants with the same financial histories in obtaining funding. For example, to the

extent that there is the cultural conception that African Americans have lower status,

and thereby lower competence, than whites, lenders will be less likely to give funding

to African Americans than to comparable whites.

Trust and Trustworthiness

I further contend that assessments of trust and trustworthiness are related to

status and aid in the lenders’ decision-making process. Many have argued that trust

has at least two forms: cognitive trust, which is based on assessments of competence

and expectations of reliably reaching a goal, and affective trust stemming from

discerning whether someone is considerate and mindful of others’ welfare (Cook and

Gerbasi 2009; Jeffries and Reed 2000; Ouchi 1981; Peters, Covello, and McCallum

1997; Rempel, Holmes, and Zanna 1985). While both high and low status individuals

Page 68: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

58

can garner affective trust under certain conditions, cognitive trust should vary by

status because it is rooted in perceptions of competence. The expectation of higher

status people is that they can be entrusted with the group’s welfare: to make beneficial

decisions and accomplish the task with a positive outcome. Individuals can trust those

of higher status to succeed at a given task (e.g., starting a new business venture or

earning a degree) because these individuals are assumed and expected to have either

the specific or generalized ability necessary to do so.

Thus, high status individuals are believed to be more likely to display and

engender this type of trustworthiness, especially in highly uncertain environments like

lending markets wherein the opportunity for malfeasance is rather high. Potential

lenders can trust a high status person to be responsible with the loan, to use it

faithfully, and pay it back on time, thereby resulting in the greater likelihood of the

request being funded. Assessments of trustworthiness should therefore operate in

concert with evaluations of competence, based on status characteristic differences, in

affecting funding decisions.

Argument Summary

Lending markets have a high level of uncertainty tied to them. There is no way

of really telling whether a particular borrower will be willing or able to pay back the

loan, and many of the traditional indicators of creditworthiness, such as credit scores,

are merely guides and are not foolproof. There is no way for lenders to predict the

future. In this uncertain situation, I propose that when lenders assess each borrower,

they are at least in part, and perhaps implicitly, assessing the relative status and

Page 69: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

59

trustworthiness of the borrower against the field of possible borrowers. These

appraisals systematically bias funding decisions, even when the borrowers’ have

commensurate financial histories. Lenders may feel as though they can entrust their

resources to those of higher status in that they assume that these borrowers have the

competence necessary to use the funds faithfully and responsibly and are able to repay

the loan. Additionally, in these markets multiple status indicators can be salient as

lenders generally do not only examine one applicant in isolation and each application

itself can quite naturally contain multiple status indicators. This study therefore also

assesses how various status signals, especially those related to gender, race, and

education, combine to affect funding decisions.

EMPIRICAL TEST

In order to analyze the role status plays in determining which loan applicants

receive funding, I recreated key aspects of lending markets in two separate studies.

The first study tests the proposed theoretical mechanism for how status, competence,

and trustworthiness affect the lending process. The second study employs a relatively

new experimental method in the social sciences, conjoint analysis (Caruso, Rahnev,

and Banaji 2009; Green, Krieger, and Wind 2001; Orme 2009), to examine whether

and to what degree extant status characteristics combine to bias funding decisions.

To better reflect the diverse backgrounds of lenders in various types of credit

markets, I obtained participants from two sources. The first are alumni from a private

university on the West Coast, and the second are from a community college also on

the West Coast. Two different samples from each population were taken for each

Page 70: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

60

study. Because these participants are not necessarily trained lenders, both studies

mimicked the peer-to-peer lending model, including its environment, conditions, and

borrower profiles. Most importantly, all participants took the study online using their

own computers without being fettered by time constraints. It should be mentioned,

however, that this environment is not wholly unlike that of more traditional lending

scenarios wherein lenders are still evaluating multiple borrowers and risking certain

assets, though the effect of the organizational context in this particular study is

virtually eliminated.

STUDY 1

The first study addresses how assessments of status, competence and

trustworthiness affect the lending process, thereby examining this project’s proposed

theoretical mechanism. The profiles used in both this and the subsequent study were

created for the purposes of this research project but actual listings posted in

prosper.com’s market were used to guide the content of each application. Peer-to-peer

loan applications generally contain a description of the purpose of the loan and a

summary of the applicants’ financial viability, which are both provided by the

borrower. Borrowers also select a picture for their listing, with about half choosing a

their own photograph and two-thirds of this group providing a picture of at least one

adult (Pope and Sydnor 2011)1. The organization also provides the applicants’ credit

1 Prosper recently altered this practice so that borrowers are now not allowed to include their own pictures to display along with their loan descriptions. Borrowers can and do, however, provide their own pictures in their personal profile pages, which are linked to their loan requests. Borrowers can also personalize their profile names. Thus, there are still multiple ways in which important social clues can be supplied in these profiles and listings.

Page 71: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

61

report and a letter grade that designates the average net loss rate of a listing. To

control for financial history, no credit report was provided to the participants. Instead,

all applicants had the second highest ranking of “B” to control for financial variability

in risk assessments related to these borrowers. To maintain the sense that these are

actual credit applications, the loan amount requested by each borrower and the loan’s

interest rate randomly varied within a very limited range. Additionally, the borrowers

loan requests were always greater than the amount that the participants were allowed

to lend to reflect the fact that lenders in these particular markets tend to only finance a

fraction of the total loan request.

To vary the gender (male or female) and race/ethnicity (white or African

American) of applicants, I developed a series of avatar pictures that were included in

each listing. In this way, I could control the presence of various status cues, including

the applicants’ age, level of attractiveness and happiness2, so that the information in

the pictures clearly differed by just gender and ethnic background. Participants were

informed that the avatars were based off of the pictures provided by the real

borrowers, with the avatars being used instead to protect the identity of the applicants.

Additionally, the use of this type of manipulation is not unique to this study as fMRI

researchers who investigate stereotype activation regularly employ cartoon stimuli in

their research (Mitchell, Ames, Jenkins, and Banajii 2009).

Finally, the borrowers’ writing ability, a proxy for their educational aptitude

and social capital, was varied in the descriptions of the loans’ purpose and the

applicants’ financial viability. In more traditional, face-to-face lending situations, this

2 The pictures were all evaluated equally along these dimensions by a group of independent raters.

Page 72: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

62

would be akin to speaking very colloquially, using slang language, and not writing

well in correspondences with the lender. These descriptions were either all

grammatically correct or contained more informal grammatical errors, such as using

abbreviations, not using proper capitalization or punctuation, and making basic

spelling errors. The content of these descriptions and the overall presence of these

errors reflect the content of actual peer-to-peer listings. In fact many of the

descriptions were taken directly from borrowers’ public listings. Nevertheless, these

descriptions were kept relatively brief (generally a sentence or two) due to time

constraints, the desire to maintain comparability across profiles, and to help ensure

that no other characteristics became salient.

The purpose of the loan also varied with respect to whether the person was

applying for a debt consolidation or business loan, but all loans had a three-year term.

Thus, the profiles presented differed by the gender, race/ethnicity, writing ability, of

the loan applicants, as well as by the type of loan. This produced a total of 16

different profiles.

Procedure

In this first study, participants assessed the qualities of two loan applicants and

then made a series of funding decisions. One applicant was always a white male with

strong writing ability (i.e., the borrower that had the most possible status advantage)

who was applying for a debt consolidation loan, which is the most common type of

loan in peer-to-peer markets. Whether this applicant was evaluated first or last was

randomly designated to help prevent ordering effects. Using the same applicant in all

Page 73: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

63

of the conditions helped constrain the number of conditions to a total of 163 in

addition to providing a consistent applicant profile against which participants would

compare the other borrowers.

After viewing the each profile, the participants were asked to make various

funding decisions. They first indicated the extent to which they expected the borrower

to likely repay the loan and how likely they would be to fund the applicant on 100-

point scales. They were also asked to put themselves in the position of being a

potential lender with a total of $1000 to either lend out, partially or fully, for potential

future gain if the loan is repaid or to keep for other purposes of their own choosing.

Under this hypothetical situation, they indicated whether they would like to fund this

particular applicant, and, if they did, how much of this pool of money they would like

to lend. Because this question asks participants to put themselves in a hypothetical

situation, the risk of providing funding is much lower than when lenders experience an

immediate cost to supplying a loan that may not be repaid. This is a strong test of

whether status indicators bias funding decisions because the participants are not

risking their own money and have no financial reason to not lend the full amount since

the applicants all have the same credit histories.

After making these preliminary lending decisions, the participants were then

asked to evaluate various qualities of the borrowers. To measure the applicants’

perceived competence – a traditional measure of status – participants rated the

3 This number also includes two sub-control conditions wherein both applicants were white males with strong writing ability applying for a debt consolidation loan. In these control conditions, the order or which borrower appeared varied to ascertain whether being evaluated first or last advantaged borrowers. Analysis (not shown) revealed that the order with which the applicants are evaluated did not influence assessments of competence, trustworthiness, or the various funding decisions.

Page 74: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

64

borrowers according to whether they are capable, organized, skilled, competent,

confident, responsible, independent, and intelligent on 100-point scales (Correll,

Benard, and Paik 2007; Cuddy, Fiske, and Glick 2004). They were also asked to

report how trustworthy they viewed each applicant to be on the same scale.

Participants then listed the applicant’s pros and cons, which compelled the participants

to more fully process the information provided in the profiles. Finally, participants

reported their demographic information, including their gender, race/ethnicity, and

age.

Hypotheses

The intention of this study is to assess how aspects of gender, ethnic

background and perceived educational aptitude combine to influence assessments of

competence and trustworthiness and whether these assessments influence funding

decisions independent of the borrowers’ financial background. Therefore, this study

evaluates the proposed theoretical mechanism that status characteristics bias reward

distributions through differential assessments of competence and trustworthiness4. To

this end, I hypothesize:

4 SCT’s traditional scope pertains to groups who are working together to solve a valued task, such as juries and certain types of work groups. The circumstances under which these funding decisions are made, however, do not necessarily meet these original scope conditions. Researchers have recently begun to broaden the scope of SCT by arguing that status processes also operate in situations where one is asked to evaluate the work and contributions of others who differ on at least one salient, valued characteristic (e.g., Correll et. al 2007; Foschi, Lai, and Sigerson 1994; Foschi, Sigerson, and Lembesis 1995; Foschi and Valenzuela 2007; Webster and Driskell 1983). Given the evaluative setting of determining the creditworthiness of loan applicants, this study falls under these more encompassing scope conditions. Lenders must take into account all of the information provided by borrowers, including their credit histories as reported by external and legitimate agencies, in order to successfully perform their roles. Borrowers have an incentive to present themselves in the best possible fashion to attract lenders, and savvy lenders should carefully process this information to make profitable decisions. Additionally, lenders do not simply evaluate one borrower over the course of their careers but multiple

Page 75: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

65

Hypothesis 1: The greater the status advantage of the applicant, as defined by

prevailing status characteristics, the more competence and trustworthiness will

be ascribed to the applicant.

Hypothesis 2: The more competent and trustworthy applicants are perceived to

be, participants will be more likely to expect that they will repay their loan.

Hypothesis 3: The more competent and trustworthy applicants are perceived to

be, participants will be more likely to fund their loan requests.

Hypothesis 4: As perceptions of the applicants’ competence and

trustworthiness increase, greater sums of money will be lent to the applicants.

STUDY 1 RESULTS

Sample

A total of 463 individuals participated in this first study: 154 undergraduates

and 309 alumni (please see Table 1). There are some differences between the

undergraduate and alumni samples. Expectedly, the average age of alumni is higher

than that of the undergraduates (t = 59.52, p < .000, two-tailed test), but the alumni

sample also has proportionally fewer females (t = 5.14, p < .000, two-tailed test) and

Asians (t = 22.76, p < .000, two-tailed test) but more whites (t = 25.91, p < .000, two-

tailed test). The alumni’s funding decisions are also more conservative: they believe different applicants. This should increase the salience of the various status characteristic differences between the applicants in the lenders’ portfolio.

Page 76: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

66

applicants are less likely to repay their loans (t = 9.70, p < .000, two-tailed test), less

likely to fund the loan requests (t = 19.46, p < .000, two-tailed test), and lend less

money to the borrowers (t = 22.42, p < .000, two-tailed test). Finally, the

undergraduates are more likely to rate the borrowers as competent and trustworthy (t =

7.51, p < .000, two-tailed test).

Table 3.1. Means or Proportions of Demographic Characteristics and Funding Decision Variables by Study and Participant Group Study 1 Study 2 Variables Undergraduates Alumni Undergraduates Alumni

Female .73 .63** .67 .59***

Age 22.64 (5.69)

50.73*** (14.12)

25.35 (8.75)

49.51*** (15.11)

Ethnic background White .43 .81*** .37 .80***

African American .04 .03 .04 .05

Latino .09 .08 .09 .06

Asian .36 .09*** .40 .05***

Other .09 .00 .10 .04

Likelihood of applicant repaying loan

69.45 (21.65)

57.65*** (21.34)

63.48 (26.12)

55.32*** (23.63)

Likelihood of funding each loan request

49.09 (29.36)

24.55*** (24.42)

50.89 (30.37)

38.57*** (28.51)

Loan amount 464.84 (265.10)

254.61*** (245.20)

477.80 (294.01)

250.44*** (244.62)

Proportion selected to receive funding

.83 .50***

Competence/trust rating

5.93 (1.40)

5.16*** (1.18)

N = 154 309 170 103

Note: SDs in parentheses; 463 individuals participated in Study 1 and 273 individuals participated in Study 2; +p<.10; *p<.05; **p<.01; *** p<.001 (two-tailed tests).

Page 77: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

67

Multivariate Analysis

To test this study’s hypotheses, I employed mixed-effects linear regression as

this study used a repeated-measures design in which the participants answered the

same series of questions for both of the applicants (Bates 2010; Jiang 2007). This form

of regression allows for proper specification of the variance and correlation of non-

independent observations while more effectively handling missing data as compared

to repeated measures ANOVA modeling (Gueorguieva and Krystal 2004; Krueger and

Tian 2004).

In all of these models, the characteristics of the borrowers (i.e., gender, race,

and writing ability), the type of loan, and the participants’ demographic information

(i.e., their graduate standing, age, and minority status) were included as fixed-effects

parameters, with a random effect designated by each participants’ ID. The applicants’

gender, race, writing ability, loan type were all coded as dummy variables, with males,

whites, those with poor writing ability and those applying for debt consolidation loans

coded as 1. The participants’ graduation and minority status were also coded as

dummy variables, with alumni and minority participants coded as 1. The competence

and trustworthiness variables are highly correlated and were averaged to create an

overall scale of competence and trustworthiness (mean = 5.45, SD = 1.32, alpha =

.95)5. This provides support for this paper’s contention that status and trustworthiness

processes work together, at least in this particular context.

5 Exploratory factor analysis also reports one underlying factor for these measures, with trustworthiness having a moderately high factor loading.

Page 78: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

68

Competence and Trustworthiness

The first hypothesis predicts that higher status applicants, as defined by the

contemporary status distinctions, will be associated with greater levels of competence

and trustworthiness. Table 2 displays the effects of the applicants’ characteristics on

the competence and trustworthiness scale; the three models add interaction effects of

the applicants’ characteristics onto prior models, with the third model containing the

three-way interaction between the applicants’ gender, racial background, and

presumed writing ability. As illustrated in the third model of Table 2, applying for a

debt consolidation loan is negatively associated with assessments of competence and

trustworthiness (b = -0.41, p < .01). For this group, applicants have already

demonstrated a lack of care and financial acumen regarding their spending practices,

thereby showing a general lack of competence in this arena. As one participant

described, “I think I’m biased because I think the fact that he’s asking for a loan to

consolidate credit card debt makes him irresponsible.” The extent to which

participants viewed the application for a debt consolidation loan as indicative of a lack

of financial ability and responsibility demonstrates that loan type can be an indicator

of status in these markets.

Page 79: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

69

Table 3.2. Study 1: Estimated Mixed-Effects Linear Regression Coefficients for the Effects of Gender, Race, and Writing Ability on the Competence and Trustworthiness Scale

Competence and Trustworthiness Scale

Variables Model 1 Model 2 Model 3

Male applicant -0.14 (0.13)

0.00 (0.25)

-0.34 (0.29)

White applicant -0.24+ (0.13)

-0.32 (0.25)

-0.68* (0.30)

Poor writing ability -1.03*** (0.13)

-0.47+ (0.26)

-0.85** (0.31)

Debt consolidation loana -0.25+ (0.13)

-0.37** (0.14)

-0.41** (0.14)

Applicant Interactions: White x male

0.21 (0.27)

0.75* (0.37)

White x poor writer -0.23 (0.27)

0.48 (0.42)

Male x poor writer -0.65* (0.27)

0.03 (0.41)

White x male x poor writer -1.21* (0.54)

Participant Controls: Alumi participant

-0.43* (0.18)

-0.45* (0.18)

-0.45* (0.19)

Age of participant

-0.01* (0.01)

-0.01* (0.01)

-0.01* (0.01)

Minority participant 0.13 (0.13)

0.15 (0.13)

0.15 (0.13)

Intercept 6.81*** (0.23)

6.65*** (0.27)

6.87*** (0.29)

Random Effect Variance Termb: Intercept

0.69 (0.07)

0.70 (0.07)

0.71 (0.07)

Note: Models fit using restricted maximum likelihood (REML); SEs in parentheses; N = 463 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Comparison group are those requesting a business loan. b Expressed as a standard deviation.

Page 80: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

70

Loan type, therefore, may be a more legitimate indicator of competence and

trustworthiness than the applicants’ ascribed characteristics, whose combined effects

on status are illustrated in the third model of Table 2. The interaction of gender and

race is significant and positive (b = 0.75, p < .05), which shows that being male

increases assessments of competence and trustworthiness for whites but not for

African Americans. To get a better sense of this interaction, Figure 1 displays the

levels of competence and trustworthiness by the applicants’ gender and racial

background as predicted from model 3. The statistically significant average group

differences are also reported in this figure.

a Letters signify statistically significant t-test differences; Capital letters represent that the difference is significant in a two-tailed test (p < .05); lowercase letters represent that the difference is marginally significant in a two-tailed test (p < .10); subscript letters represent that the difference is marginally significant in a one-tail test (p < .10).

5.41

5.15

5.23

5.48

5.0

5.1

5.2

5.3

5.4

5.5

5.6

Male Female

Com

pete

nce

and

Trus

twor

thin

ess S

cale

Gender

Figure 3.1. Study 1: Predicted Assessment of Competence and Trustworthiness by Applicants' Gender and Racial Backgrounda

White

African American

c, d

A

B, d

A, B, c

Page 81: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

71

The main effect of gender is negative, which would suggest that African

American males are rated as having lower competence and trustworthiness than

African American females, but this effect is not significant (b = -0.34, p = .244). A t-

test analysis of the average difference between African American males and females

reveals, however, that there is significant difference between these two groups, with

African American females having higher assessments of competence and

trustworthiness (please see Figure 1; t = 2.09, p < .05, two-tailed test). Turning back

to the third model of Table 2, we see that the main effect of racial background is

significant and negative, indicating that white females are evaluated as having lower

status than African American females. T-test analysis also bears this out (t = 2.42, p <

.05, two-tailed test). As seen in Figure 1, African American females are also rated as

having marginally higher competence and trustworthiness than white males (t = 1.74,

p = .083, two-tailed test). No significant differences were found between the status

assessments of white females and African American males.

These are somewhat surprising findings and ones that contradict the theory’s

prediction that the states of the status characteristics would basically aggregate (Berger

et al. 1977). SCT would predict that if male and white are the advantaged states of

gender and racial background, then white males should be rated as having the highest

levels of competence and trustworthiness because they are doubly advantaged and

African Americans females would experience what is essentially double jeopardy.

Instead, participants viewed African American females as having marginally higher

competence and trustworthiness than white males, while white males and African

American females have higher relative status and trustworthiness than African

Page 82: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

72

American males and white females. As will be shown, this general trend persists

throughout the analysis of study 1.

Turning now to the effect of writing ability as displayed in model 3 in Table 2,

there is generally a negative effect across the gender and racial categories for

evidencing poor writing ability on assessments of competence and trustworthiness (b

= -0.41, p < .01). Providing a summary of your loan needs that includes grammatical

errors may be a signal of a lack of education or social capital as the borrower is not

savvy enough to understand that one is trying to foster what is essentially a

professional business relationship with the lenders. As one participant stated of an

applicant with poor writing, “…the applicant doesn't seem very intelligent. My first

impression of him is of a person who wrote this as a text to his friend. He doesn't look

very professional.”

While there is an overall negative effect of displaying poor writing on

assessments of competence and trustworthiness, the magnitude of the effect does vary

by gender and racial category as shown in the three-way interaction of model 3 (b = -

1.21, p < .05). To help grasp the details of this interaction, Figure 2 displays the

predicted assessment of competence and trustworthiness by the applicants’ gender,

racial back, and exhibited writing ability. Among those who showed strong writing

ability, African American females were again rated as having higher levels of

competence and trustworthiness than white males (t = 3.04, p < .01, two-tailed test)6.

Overall, white females are the only group for whom writing ability does not

6 While African American females’ have the highest average rating of competence and trustworthiness within this group of superior writers, their differences with white females and African American males do not quite reach marginal levels of significance (white females: t = 1.19, p = 0.117, one-tailed test; African American males: t = 1.28, p = 0.102, one-tailed test).

Page 83: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

73

significantly alter status assessments (t = 0.61, p = .54, two-tailed test). White males,

however, are the most disadvantaged by including grammatical mistakes in their loan

applications, and participants view this group as having significantly lower

competence and trustworthiness than white females (t = 2.00, p < .05, two-tailed test)

and African American females (t = 2.49, p < .05, two-tailed test).

Page 84: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

74

a Letters signify statistically significant t-test differences; Capital letters represent that the difference is significant in a two-tailed test (p < .05); lowercase letters represent that the difference is marginally significant in a two-tailed test (p < .10); subscript letters represent that the difference is marginally significant in a one-tail test (p < .10).

5.51 5.29

5.62

5.92

4.20

5.00 4.83

5.14

4.0

4.5

5.0

5.5

6.0

6.5

White Male White Female African American Male African American Female C

ompe

tenc

e an

d Tr

ustw

orth

ines

s Sca

le

Gender

Figure 3.2. Study 1: Predicted Assessment of Competence and Trustworthiness by Applicants' Gender, Racial Background, and Writing Abilitya

Superior Writing Ability

Poor Writing Ability

A, E, F

E

B, g

A, D, i

B, h, i

C, D

C, F, g

h

Page 85: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

75

White males have the double advantage of their gender and racial background,

and when they evidence poor writing, they are violating a generalized normative

expectation that they should be proficient in other domains. For this group, perhaps

this inconsistent trait has a deeper impact due to these expectations (Berger et al.

1977). It is important to note, however, that the other applicant in this condition was

always a white male who presented strong writing skills; thus the saliency of writing

ability in this condition was the strongest it could be because no other characteristics

were salient. So this effect among white males may at least in part be due to the

extreme salience of this writing characteristic.

Funding Assessments

Are those rated as more competent and trustworthy more likely to be

advantaged when lenders make their funding assessments? The results suggest that

this is indeed the case (please see Table 3 and 4, which contain the funding assessment

analyses). The first models contained in Table 3 and 4 report the basic model with

only the main effects of the applicants’ characteristics for each of the funding

assessment dependent variables: perceived likelihood of the applicant repaying the

loan, perceived likelihood of funding the applicant, and loan amount given. The

second models include the addition of the gender, racial background, and writing

ability interactions. Finally, the third models present the full model with the

assessments of competence and trustworthiness1.

1 The third models reported in Table 3 do not contain the applicant interactions because they were not significant in the second models and were also not significant when added into the full model (analyses

Page 86: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

76

As reported in Table 3, participants reported that men and African Americans

are expected to be the less likely to repay their loans than women and whites (b = -

4.20, p < .05 and b = 3.61, p < .05 respectively), but these expectations do not persist

in the analysis of the other funding decision variables. The interaction of gender,

racial background, and conveyed writing ability is again negative and significant in the

second model for the amount of money lent to the borrowers (b = -210.86, p < .05);

however, this interaction is no longer significant when assessments of the applicants’

competence and trustworthiness are added to the model (b = -86.63, p = .384).

The lenders’ assumptions about the applicants’ competence and trustworthiness have a

consistently positive and significant impact on their funding judgments. As

participants’ perceptions about applicants’ competence and trustworthiness increase,

they are more likely to expect that the borrowers will repay their loan (b = 8.55, p <

.000), they are more likely to lend money to the applicant (b = 9.43, p < .000), and the

more money they will lend (b = 102.62, p < .000, model 4). Additionally,

assessments of competence and trustworthiness are the only significant predictor of

whether a lender is likely to fund an applicant, and at what amount. This provides

strong support for hypotheses two through four.

not reported). The fourth model in Table 4 eliminates these applicant interactions because they were not significant when assessments of competence and trustworthiness were added in model three.

Page 87: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

77

Table 3.3. Study 1: Estimated Mixed-Effects Linear Regression Coefficients for the Effects of Gender, Race, Writing Ability, and Assessments of Competence and Trustworthiness on Funding Assessments

Variables Likelihood of Applicant Repaying Loan

Likelihood of Funding Each Loan Request

Model 1 Model 2 Model 3 Model 1 Model 2 Model 3

Male applicant -5.14** (1.71)

-3.31 (3.92)

-4.20* (1.85)

1.11 (2.07)

-2.66 (4.70)

3.28 (2.33)

White applicant 0.42 (1.71)

0.54 (3.91)

3.61* (1.82)

-1.10 (2.07)

-1.60 (4.69)

0.02 (2.29)

Poor writing ability -10.83*** (1.76)

-9.63* (3.96)

-1.71 (1.93)

-6.68** (2.15)

-1.41 (4.75)

2.10 (2.44)

Debt consolidation loana 0.22 (1.69)

0.16 (1.82)

3.02+ (1.79)

-2.68 (2.04)

-5.06* (2.18)

0.29 (2.26)

Competence and trustworthiness scale

8.55*** (0.61)

9.43*** (0.78)

Applicant Interactions: White x male

-1.15 (4.80)

7.07 (5.75)

White x poor writer

0.35 (5.45)

-4.53 (6.53)

Male x poor writer .

-3.14 (5.42)

2.19 (6.49)

White x male x poor writer

1.06 (7.29)

-11.23 (8.74)

Participant Controls: Alumni participant

-9.48** (2.86)

-9.51** (2.87)

-2.96 (2.81)

-28.89*** (3.62)

-29.27*** (3.64)

-24.67*** (3.78)

Age of participant -0.15+ (0.08)

-0.15+ (0.08)

-0.09 (0.08)

0.19* (0.10)

0.20 (0.10)

0.27 (0.11)

Minority participant -2.83 (2.11)

-2.83 (2.11)

-2.59 (2.00)

3.54 (2.67)

3.64 (2.68)

3.24 (2.69)

Intercept 80.87*** (3.41)

80.09*** (4.06)

21.39*** (5.38)

46.48*** (4.25)

46.42*** (5.00)

-15.73* (6.95)

Random Effect Variance Termb: Intercept

14.27 (0.85)

14.26 (0.85)

11.71 (0.97)

18.57 (1.02)

18.84 (1.02)

16.69 (1.11)

Note: Models fit using restricted maximum likelihood (REML); SEs in parentheses; N = 273 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Comparison groups are those requesting a business or student loan. b Expressed as a standard deviation.

Page 88: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

78

Table 3.4. Study 1: Estimated Mixed-Effects Linear Regression Coefficients for the Effects of Gender, Race, Writing Ability, and Assessments of Competence and Trustworthiness on Loan Amount Given

Variables Loan Amount Given

Model 1 Model 2 Model 3 Model 4

Male applicant 8.30 (22.33)

2.57 (50.08)

48.14 (51.56)

-3.24 (23.98)

White applicant -11.76 (22.94)

-9.62 (49.31)

71.28 (54.17)

-2.92 (24.02)

Poor writing ability -53.89* (23.74)

-2.12 (50.38)

148.69** (56.09)

41.99 (25.49)

Debt consolidation loana -30.70 (22.43)

-54.14* (23.30)

-15.19 (24.46)

2.27 (23.71)

Competence and trustworthiness scale

100.06*** (8.15)

102.62*** (8.17)

Applicant Interactions: White x male

47.05 (60.77)

-44.36 (64.68)

White x poor writer

8.60 (70.24)

-92.16 (76.55)

Male x poor writer .

17.34 (69.80)

-50.77 (72.93)

White x male x poor writer

-210.86* (94.95)

-86.63 (99.46)

Participant Controls: Alumni participant

-231.09*** (27.75)

-234.77*** (38.10)

-217.68*** (38.00)

-213.34*** (37.64)

Age of participant 0.53 (1.00)

0.70 (1.01)

1.75 (1.08)

1.64 (1.07)

Minority participant 7.49 (27.98)

8.96 (28.22)

-4.36 (26.95)

-8.11 (26.72)

Intercept 487.91*** (43.54)

468.42*** (51.46)

-212.53*** (76.75)

-164.33* (70.23)

Random Effect Variance Termb: Intercept

176.74 (11.02)

181.42 (10.90)

155.95 (11.40)

151.11 (11.60)

Note: Models fit using restricted maximum likelihood (REML); SEs in parentheses; N = 273 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Comparison groups are those requesting a business or student loan. b Expressed as a standard deviation.

Page 89: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

79

SUMMARY OF STUDY 1

The results from Study 1 suggest that lenders’ assessments of applicants’

general competence and trustworthiness do significantly alter their funding decisions.

This provides strong support for the proposed status mechanism that lenders appear to

rely in part on assumptions about who these borrowers are even when applicants’

evidence similar levels of creditworthiness. These results also suggest that many of

the qualities of the borrowers affect funding decisions by first impacting the lenders’

assumptions of their relative competence and trustworthiness. Expectations derived

from the applicants’ characteristics then impact the lenders’ funding decisions. Thus,

status and trustworthiness are likely a mechanism behind lending discrimination in

that they mediate the relationship between socially significant categories and lending

outcomes.

Some of the borrowers’ characteristics may be at least partially within their

control, such as the financial reasons behind why they applied for a debt consolidation

loan or the quality and formality of their writing and self-presentation; however, the

lenders also appear to rely on assumptions based on the applicants’ ascribed

characteristics, specifically their gender and racial background, which are illegitimate

bases on which to form funding assessments.

The way in which these traits combined to influence assessments of status was

not, however, always in a manner that would be predicted by SCT. Specifically,

African American females, whom the theory would predict as being the most status-

disadvantaged, were rated as having essentially the highest levels of competence and

trustworthiness. Perhaps in this situation in which people are actively seeking

Page 90: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

80

assistance to improve their lives, stereotypes associated with African American

females as strong, responsible women who are able to get things accomplished no

matter the situation or cost become salient and impact lenders’ funding decisions and

status assessments of the applicants (Browne and Kennelly 1999; Collins 1990;

Harris-Lacewell 2001; Kennelly 1999).

An opposing explanation could be that these findings are at least partially due

to social desirability bias suggesting that participants are not always able or willing to

respond authentically about socially sensitive topics, especially in social research.

Consequently, participants respond in what they believe to be a more socially

acceptable or normative fashion (Greenwald and Banaji 1995; King and Brunner

2000). In Study 1, the procedure made differences in the applicants’ characteristics

extremely salient, as they were always being compared to a white male who evidenced

having a strong writing ability. Due to this salience, social desirability bias is more

likely to pose a problem and influence this study’s results.

STUDY 2

To help verify the combined effects of gender, race, and education, a second

study was run that employed a relatively new technique for the social sciences,

conjoint analysis (Caruso et al. 2009; Green et al. 2001; Orme 2009). This method is

designed to more effectively uncover implicit attitudes while reducing social

desirability bias.

Page 91: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

81

Conjoint Analysis

Conjoint analysis, popularized in marketing research, allows researchers to

uncover implicit attitudes by presenting individuals with sets of profiles that vary in

pre-determined, theoretically important ways. Respondents then select which profile

they most prefer or how greatly they prefer each profile in a given context. Because

people evaluate the various features of the profiles in concert, they tend to be better

able to ascertain what their relative preferences are than when they are asked to report

their preferences without points of comparison (Caruso et al. 2009). Additionally,

since multiple features are manipulated simultaneously, people are not asked to

directly compare socially sensitive characteristics, such as gender or ethnic

background, in isolation. Instead, these potentially sensitive attributes can be

embedded within broader, more complex profiles. Thus, the method generally reduces

social desirability bias while uncovering implicit beliefs.

For instance, Caruso et al. (2009) employed this method to gauge relative

preferences for trivia partners. The potential partners varied according to their

education level, IQ, trivia experience, and body weight. While participants reported

that body weight had an inconsequential effect on their choice, in actuality participants

selected partners with lower IQs, by about 11 points, to have a thin partner. As related

to lending, while respondents may state that they consider, say, a business loan more

fundable than a debt consolidation loan, their behavior may indicate an underlying

preference for relatively high status borrowers as defined by their ascribed

characteristics.

Page 92: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

82

Procedure

There are many different types of conjoint analysis (see Orme 2009 for an

overview), but this study employs the single concepts variation. In single concepts

conjoint analysis, participants examine and rate a random series of individual profiles2.

The general set-up for this study mimics that of the first: participants were asked to

evaluate loan applications that all fell under the same credit ranking, the terms of the

loans were identical, and the amount requested and the loans’ interest rate randomly

varied within a limited range.

The content of these profiles slightly differed from that of the first study. The

applicants again differed by their gender (male or female), race/ethnicity (white or

African American), and writing skill (with and without grammatical errors in the

application materials); however, in the second study, borrowers could also apply for a

school loan in addition to a debt consolidation or business loan3. The combination of

these characteristics created a total of 24 profiles for the participants to evaluate. Each

participant rated half of these 24 profiles4 but all rated the four baseline profiles:

borrowers seeking a debt consolidation loan who used correct grammar but who

varied by gender and race/ethnicity5. Participants therefore evaluated a set of 14

randomly generated profiles.

2 Research indicates that single concepts conjoint analysis produces essentially the same results as those obtained through pairwise comparisons conjoint analysis, wherein participants evaluate a series of paired profiles and must determine the extent of their preferences for one profile relative to the other (Caruso et al. 2009). 3 Subsequent analysis revealed that funding decisions did not significantly vary by between whether the loan was for a school or business purpose. These two loan types are combined in the reported analyses. 4 Participants only viewed half of the profiles due to the time constraints associated with using the alumni sample. 5 Debt consolidation loans are the most prevalent type of loan in peer-to-peer markets; therefore this type of loan was selected as being the most basic.

Page 93: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

83

For each session, participants evaluated one application at a time. They were

asked to carefully read through the application before answering a series of questions

related to each profile. The participants were again asked to indicate how likely they

believed each candidate would be to repay the loan and how likely they would be to

fund each applicant on a 100-point scale. If the participants stated that they would

likely fund the applicant, they were then asked that if they had $1000 to either give to

the applicant or retain for another purpose, how much would they lend to the

applicant. Once they finished evaluating an application, they proceeded to assess next

profile until they sequentially evaluated all 14 applications. Participants also reported

their demographic information, including their gender, race/ethnicity, and age at the

end of the study.

STUDY 2 RESULTS

Sample

170 undergraduates and 103 alumni participated in study 2 for a total of 273

participants (please refer back to Table 1). As in Study 1, the alumni are older, on

average, than the undergraduates (t = 32.11, p < .000, two-tailed test). The alumni

sample also contains proportionately fewer female (t = 2.94, p < .01, two-tailed test)

and Asian participants (t = 10.02, p < .000, two-tailed test) but more white participants

(t = 11.76, p < .000, two-tailed test). Again, the undergraduates generally make more

generous funding decisions: they are more likely to think the applicants will repay

their loans (t = 7.86, p < .000, two-tailed test), are more likely to want to fund the

listing (t = 13.42, p < .000, two-tailed test), lend more money (t = 11.13, p < .000,

Page 94: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

84

two-tailed test), and are more likely to ultimately fund either applicant (t = 10.16, p <

.000, two-tailed test).

Multivariate Analysis

Mixed-effects linear regression was used again for this study’s analyses as

participants repeatedly made the same three kinds of funding assessments across 14

credit applicants (Bates 2010; Jiang 2007). As in Study 1, the characteristics of the

borrowers (i.e., gender, race, and writing ability), the type of loan, and the

participants’ demographic information (i.e., their graduate standing, age, and minority

status) were included as fixed-effects parameters, with a random effect designated by

each participants’ ID. The applicants’ gender, race, writing ability, loan type were all

coded as dummy variables, with males, whites, those with poor writing ability, and

those applying for debt consolidation loans coded as 1. The participants’ graduation

and minority status were also coded as dummy variables, with alumni and minority

participants coded as 1.

Funding Decisions

The purpose of this study is to ascertain whether and how the status indicators

present in Study 1 persist in shaping funding decisions (i.e., perceived likelihood of

the applicant to repay the loan, how likely the participant is to give a loan to the

applicant, and the loan amount given). As such, the funding assessment models

reported in Table 5 and 6 mirror those from Study 1. The only theoretically important,

Page 95: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

85

consistent, and significant predictor in this study was the assessment of the borrower’s

competence and trustworthiness.

These models suggest that there is a persistent negative effect of requesting a

debt consolidation loan and displaying poor and improper writing across all of the

funding assessment variables. Thus, participants find business and school loans more

creditable and are more apt to fund those who have strong writing statements. These

results are generally consistent with those from Study 1, except that this more

sensitive methodological technique reveals a direct effect of these applicant

characteristics on lenders’ funding choices.

Page 96: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

86

Table 3.5. Study 2: Estimated Mixed-Effects Linear Regression Coefficients for the Effects of Gender, Race, and Writing Ability on Funding Assessments

Variables Likelihood of Applicant Repaying Loan

Likelihood of Funding Each Loan Request

Model 1 Model 2 Model 3 Model 1 Model 2 Model 3

Male applicant -0.60 (0.69)

-3.96** (1.17)

-2.87* (1.33)

-2.15** (0.73)

-6.56*** (1.25)

-4.03** (1.41)

White applicant 0.10 (0.69)

-2.52* (1.15)

-1.45 (1.30)

0.88 (0.73)

-2.42* (1.22)

0.06 (1.37)

Poor writing ability -11.40*** (0.70)

-11.54*** (1.22)

-10.30*** (1.42)

-13.18*** (0.75)

-13.78*** (1.30)

-10.88*** (1.51)

Debt consolidation loana -3.67*** (0.70)

-3.70*** (0.70)

-3.69*** (0.70)

-4.52*** (0.75)

-4.55*** (0.75)

-4.53*** (0.75)

Applicant Interactions: White x male

5.88*** (1.39)

3.78* (1.85)

7.22*** (1.48)

2.33 (1.96)

White x poor writer

-0.76 (1.39)

-3.14 (1.96)

-0.74 (1.48)

-6.29** (2.09)

Male x poor writer .

0.99 (1.44)

-1.49 (2.04)

.

1.87 (1.54)

-3.91+ (2.17)

White x male x poor writer

4.80+ (2.79)

11.16*** (2.97)

Participant Controls: Alumni participant

-13.39*** (3.03)

-13.37*** (3.03)

-13.37*** (3.03)

-20.22*** (4.07)

-20.19*** (4.07)

-20.20*** (4.07)

Age of participant 0.16+ (0.09)

0.16+ (0.09)

0.16+ (0.09)

0.08 (0.12)

0.08 (0.12)

0.08 (0.12)

Minority participant -1.11 (2.22)

-1.14 (2.22)

-1.14 (2.22)

3.81 (2.98)

3.76 (2.98)

3.79 (2.98)

Intercept 67.11*** (3.20)

67.91*** (3.25)

68.14*** (3.27)

54.87*** (4.25)

57.03*** (4.30)

55.69*** (4.31)

Random Effect Variance Termb: Intercept

14.27 (0.75)

14.27 (0.75)

14.27 (0.75)

19.60 (0.98)

19.61 (0.98)

19.61 (0.98)

Note: Models fit using restricted maximum likelihood (REML); SEs in parentheses; N = 273 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Comparison groups are those requesting a business or student loan. b Expressed as a standard deviation.

Page 97: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

87

Table 3.6. Study 2: Estimated Mixed-Effects Linear Regression Coefficients for the Effects of Gender, Race, and Writing Ability on Loan Amount Given

Variables Loan Amount Given

Model 1 Model 2 Model 3

Male applicant -13.17+ (7.14)

-40.25** (12.06)

-21.09 (13.62)

White applicant 5.94 (7.14)

-25.96* (11.79)

-7.31 (13.30)

Poor writing ability -114.08*** (7.38)

-109.83*** (12.77)

-87.26*** (14.80)

Debt consolidation loana -59.05*** (7.35)

-59.55*** (7.33)

-59.56*** (7.33)

Applicant Interactions: White x male

63.09*** (14.37)

26.38 (18.84)

White x poor writer

1.62 (14.48)

-41.63* (20.39)

Male x poor writer .

-11.47 (15.03)

-56.91** (21.29)

White x male x poor writer

87.50** (29.09)

Participant Controls: Alumni participant

-267.40*** (39.72)

-267.27*** (39.71)

-267.53*** (39.71)

Age of participant 2.13+ (1.15)

2.14+ (1.15)

2.15+ (1.15)

Minority participant 49.65+ (28.91)

49.48+ (28.90)

49.81+ (28.90)

Intercept 459.93*** (41.28)

473.73*** (41.70)

463.51*** (41.83)

Random Effect Variance Termb: Intercept

189.26 (9.62)

189.27 (9.62)

189.30 (9.61)

Note: Models fit using restricted maximum likelihood (REML); SEs in parentheses; N = 273 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Comparison groups are those requesting a business or student loan. b Expressed as a standard deviation.

Page 98: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

88

Turning now to the impact of the applicants’ gender and racial background, we

see again that the magnitude of the effects that gender, racial background, and writing

aptitude have are conditional on each other. Notably, the effects of each combination

on funding decisions are very similar to the influence they had in predicting divergent

assessments of competence and trustworthiness in Study 1. Beginning with projected

repayment, the interaction between gender and racial background is significant (b =

3.78, p < .05), whereas the three-way interaction with writing ability is only

marginally significant (b = 4.80, p < .10). As displayed in Figure 3, white males and

African American females are viewed as equally likely to repay a loan (t = 0.28, p =

.779, two-tailed test), as are white females and African American males (t = 0.46, p =

.645, two-tailed test). White males and African American females are also perceived

as being more likely to repay a loan than white females and African American males6.

6 A figure for the three-way interaction with writing aptitude was not included, as it is only a marginally significant predictor of anticipated repayment.

Page 99: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

89

a Letters signify statistically significant t-test differences; Capital letters represent that the difference is significant in a two-tailed test (p < .05); lowercase letters represent that the difference is marginally significant in a two-tailed test (p < .10); subscript letters represent that the difference is marginally significant in a one-tail test (p < .10).

The three-way interaction of the applicants’ gender, racial background, and

writing precision is a significant predictor of whether the lender foresees giving a loan

to the particular borrower (b = 11.16, p < .000). To better appreciate the details of this

interaction, Figure 4 reports the predicted likelihood of funding the request by each

combination of this interaction of borrower characteristics. Without taking apparent

writing ability into consideration, participants are about equally likely to fund white

males and African American females (t = 0.89, p = .372, two-tailed test), and African

American males are the least likely to receive a loan (difference with white females: t

= 1.87, p < .10, two-tailed test).

62.11

59.59 59.16

62.33

57.0

58.0

59.0

60.0

61.0

62.0

63.0

Male Female

Lik

elih

ood

of R

epay

ing

the

Loa

n

Gender

Figure 3.3. Study 2: Predicted Likelihood of Repaying the Loan by Applicants' Gender and Racial Backgrounda

White

African American

A, c

A, D

B, c

B, D

Page 100: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

90

a Letters signify statistically significant t-test differences; Capital letters represent that the difference is significant in a two-tailed test (p < .05); lowercase letters represent that the difference is marginally significant in a two-tailed test (p < .10); subscript letters represent that the difference is marginally significant in a one-tail test (p < .10).

49.51

51.21

47.01

51.25

40.52

34.61 33.26

40.84

32.0

34.0

36.0

38.0

40.0

42.0

44.0

46.0

48.0

50.0

52.0

White Male White Female African American Male African American Female

Lik

elih

ood

of F

undi

ng L

oan

Gender

Figure 3.4. Study 2: Predicted Likelihood of Funding Loan by Applicants' Gender, Racial Background, and Writing Abilitya

Superior Writing Ability

Poor Writing Ability E, K, L

F, K, M

G, L, N

E

G, I, J

H, I

H, M, N

F, J

A

A, C, d

b, d

b, C

Page 101: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

91

Overall, writing ability has a significantly negative impact on whether the

participants fund a request across all of the gender and racial groups. For those whose

applications did not include writing mistakes, African American males are less likely

to receive loans than white and African American females (t = 2.27, p < .05 and t =

1.98, p < .05, respectively in two-tailed tests). Assessing those with poor writing

ability, we see the continuing trend that white females and African American males

are less likely to be funded than white males and African American females.

Finally, the loan amount given is also similarly affected by the three-way

interaction of the applicants’ characteristics (please see Model 3 of Table 6). Figure 5

displays the predicted loan amount given by the applicants’ gender, racial background,

and perceived writing ability. African American males and white females are again

the most disadvantaged regardless of how well their applications are written. White

females are given marginally less money than African American females (t = 1.68, p <

.10, one-tailed test), while African American males are provided with fewer funds than

both white males (t = 2.33, p < .05, two-tailed test) and African American females (t =

2.89, p < .01, two-tailed test).

Across all of the gender and racial groups, writing ability again has a

consistent and negative impact on how much each applicant is credited. Additionally,

whether the applications contained grammatical errors influenced each gender and

racial group in essentially the same manner as in the prior analysis of funding

likelihood. Once more we see that for those with strong writing statements, African

American male applicants are provided with less money than African American

females (t = 1.75, p < .10, two-tailed test) and white females (t = 1.36, p < .10, one-

Page 102: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

92

tailed test). Among the poor writers, African American females are lent about the

same amount of money as white males (t = 0.41, p = .682, two-tailed test), and these

two groups are both given more funds than white females and African American

males.

Page 103: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

93

a Letters signify statistically significant t-test differences; Capital letters represent that the difference is significant in a two-tailed test (p < .05); lowercase letters represent that the difference is marginally significant in a two-tailed test (p < .10); subscript letters represent that the difference is marginally significant in a one-tail test (p < .10).

438.59 433.30

418.19

442.02

352.63

312.02

287.69

360.96

275.00

300.00

325.00

350.00

375.00

400.00

425.00

450.00

White Male White Female African American Male African American Female

Loa

n A

mou

nt G

iven

Gender

Figure 3.5. Study 2: Predicted Loan Amount Given by Applicants' Gender, Racial Background, and Writing Abilitya

Superior Writing Ability

Poor Writing Ability

D, J, K

E, J, L

F, K, M

D

F, h, i

G, h

G, L, M

E, i

A

A, C

b

b, C

Page 104: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

94

SUMMARY OF STUDY 2

The use of the conjoint method uncovers a consistent and significant effect of

the applicants’ characteristics on lenders’ funding assessments. The general finding

across these models is that white males and African American females are similarly

advantaged across all funding decisions, and their applications are evaluated more

favorably than both those of white females and African American males. This result

mirrors that which was reported in the analysis of competence and trustworthiness

from Study 1, excepting that African American females were evaluated as having the

highest levels of competence and trustworthiness. Importantly, the conjoint analysis

technique used in Study 2 was specifically designed to reduce the effects of social

desirability bias to uncover implicit attitudes and preferences; yet, these results

generally replicate those from Study 1 in that the status beliefs associated with gender

and race do not have a simple, additive impact on lending decisions.

DISCUSSION

In both studies we are able to see the combined effect of gender and racial

categories on a wide array of funding decisions and in assessments of competence and

trustworthiness. The intersections of race and gender indicate that African American

males continue to be highly disadvantaged in a wide array of lending decisions, while

the funding that African American females receive is similar to that of white males.

Not only are African American females rewarded at the same level, they are also rated

as having slightly higher levels of competence and trustworthiness than white males.

Page 105: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

95

This is a somewhat surprising result in that SCT would predict that the status

valuations associated with the states of gender and racial background would

essentially aggregate, which is what we see with white males relative to African

American males and white females. In broad terms, the stereotypes associated with

these groups generally conform to SCT’s predictions. White females are generally

prescribed to be communal, such as being more emotional, passive, and ineffectual

than white males (Diekman and Eagly, 2000; Prentice and Carranza 2002; Rudman

and Glick 2001). Consequently, lenders may perhaps implicitly assume that white

females are not able to use the loans appropriately and are therefore less likely to have

their requests fulfilled. African American males tend to be viewed as lazy,

threatening, or overly aggressive, and thus they may not be responsible with a loan

(Devine and Elliot 1995; Eberhardt, Goff, Purdie, and Davies 2004; Lane, Banaji,

Nosek, and Greenwald 2007; Pager 2007; Quillian 2006, 2008).

Indeed, many participants held a harsher standard against African American

males and white females in their descriptions of the applicants’ pros and cons in Study

1 (Foschi 1989, 2000). They are more critical that these two groups did not have the

highest credit rating, even though participants were informed that all of the applicants

have the same credit history. A few participants were openly against funding a white

female. For instance, one participant reported, “Applicant 1 [the white male] sounds

better1.” Another said, “I trust men more than I trust women,” while a third stated that

the only negative about a particular applicant was that she “is a woman.” Participants

were also generally more suspecting of African American males’ employment.

1 Both applicants exhibited strong writing ability in this case.

Page 106: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

96

Various participants made statements along the lines of “his employment is probably

temporary,” that “his job may not be secure,” and that he is generally “too evasive.”

Simultaneously, participants in these conditions made overly positive statements about

the white male applicants. For instance, one participant listed the benefits of funding

the African American male applicant as “none,” but said that the white male “seems

intelligent and hardworking.”

Stereotypes of African American females, however, suggest that people tend to

view this group as being strong black women trying to do it all against the odds;

therefore, they are perhaps more worthy of assistance (Browne and Kennelly 1999;

Collins 1990; Harris-Lacewell 2001; Kennelly 1999). Harris-Lacewell (2001) defines

this concept thusly:

In her contemporary form, the strong black woman is a motivated,

hardworking breadwinner. She is always prepared ‘to do what needs to

be done’ for her family and her people. She is sacrificial and smart.

She suppresses her own emotional needs while anticipating those of

others. She has a seemingly irrepressible spirit unbroken by a legacy of

oppression, poverty, and rejection. (P. 3)

The statements made by the participants from Study 1 demonstrate the salience

of this conception. Many participants reported that the African American female

applicant was “intelligent, smart, seems to be reliable since she has a good credit

history,” and that she seems “well organized and trustworthy” and “dependable.”

Another participant stated, “She has enough self-confidence to go online and request a

Page 107: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

97

loan.” This sentiment was echoed by another, saying, “I think this candidate would

make her best effort to repay the loan.”

These results suggest the persistence of this strong black woman stereotype in

this lending environment. This conception of African American females is certainly

not always beneficial, however. For instance, it can generate great stress and social

pressures, and observers may see their commitment and responsibility to their job as

stemming from a “need to do it” rather than a “wanting to do it,” which is considered

to be a more desirable trait in the labor force (Kennelly 1999). Still, in this particular

environment this stereotype does appear to advantage African American females.

As related to expectation states theories, it is already accepted that status

evaluations are contingent on the group’s task and that different situations can

generate various levels of inequality (Balkwell 1991). This work demonstrates that

under certain conditions the expectations about certain groups may even reverse. It

may then be beneficial for status theorists to allow for a more intersectional approach

to understanding the combined effects of gender and race (Browne and Misra 2003).

Even though the over-arching status mechanism operates as expected, the attitudes and

beliefs about each combination of traits may be relatively unique, as opposed to

simply being characterized as what is essentially multiple jeopardy. Those who work

within this research program have been relatively good about allowing for the

presence of contextual effects, but how these contexts affect the way status

characteristics combine has not been well investigated.

Page 108: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

98

CONCLUSIONS

By examining the decision-making processes of potential lenders with an

experimental framework, this project helps illuminate how assessments of competence

and trustworthiness that vary by status groups may be a driving mechanism behind the

funding variations found in traditional credit markets more broadly. Status becomes a

means by which lenders compare applicants to determine whether and how to fund

them, even when the applicants have similar financial histories. This research offers a

potential cause behind the discrepancies in underwriting standards captured in

previous research, and, thereby, provides one potential solution to the puzzle of

continued lending discrimination and the accumulation of new wealth.

This work also has implications for online markets in that this study

demonstrates how social cues can bias judgments in online settings in which we do not

necessarily have the benefit of face-to-face interaction that perhaps dampens the

saliency of certain status cues. Indeed, relatively small actions that are common in

online settings, such as not proof-reading messages or using texting language, can give

big signals to evaluators about one’s competence and trustworthiness. As online

transactions become an increasingly popular venue for business, schooling, dating, and

the like, this type of bias may become ever more important.

Page 109: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

99

References

Alves, Wayne M. and Peter H. Rossi. 1978. “Who Should Get What? Fairness Judgments of the Distribution of Earnings.” American Journal of Sociology 84:541-64.

Apgar, William C. and Allegra Carter. 2005. “The Dual Mortgage Market: The Persistence of Discrimination in Mortgage Lending.” Pp. 101-26 in The Geography of Opportunity: Race and Housing Choice in Metropolitan America, edited by X. S. Briggs. Washington, DC: Brookings Institute Press. Ayres, Ian. 1995. “Further Evidence of Discrimination in New Car Negotiations and Estimates of Its Cause.” Michigan Law Review 94:109-47. Ayres, Ian and Peter Siegelman. 1995. “Race and Gender Discrimination in Bargaining for a New Car.” The American Economic Review 85:304-21. Balkwell, James W. 1991. “Status Characteristics and Social Interaction: Assessment of Theoretical Variants.” Pp. 135-76 in Advances in Group Processes, vol. 8, edited by E. J. Lawler, B. Markovsky, C. Ridgeway, and H. A. Walker. Greenwich, CT: Jai Press. Bates, Douglas M. 2010. lme4: Mixed-Effects Modeling with R. Springer. Berger, Joseph. 1958. “Relations between Performance, Rewards, and Action- Opportunities in Small Groups.” Ph.D. dissertation, Harvard University.

Berger, Joseph, Bernard P. Cohen, and Morris Zelditch Jr. 1966. “Status

Characteristics and Expectation States.” Pp. 29-46 in Sociological Theories in Progress, vol. 1, edited by J. Berger, M. Zelditch Jr., and B. Anderson. Boston: Houghton Mifflin.

------. 1972. “Status Characteristics and Social Interaction.” American Sociological

Review 37:241-55. Berger, Joseph, M. Hamit Fisek, Robert Z. Norman, and David Wagner. 1983. "The Formation of Reward Expectations in Status Situations." Pp. 127-68 in Equity

Theory, edited by D. M. Messick and K. S. Cook. New York: Praeger. -------. 1998. “Formation of Reward Expectations in Status Situations.” Pp. 121-53 in

Status, Power, and Legitimacy, edited by J. Berger and M. Zelditch. New Brunswick, NJ: Transaction Publishers.

Berger, Joseph, M. Hamit Fisek, Robert Z. Norman, and Morris Zelditch Jr. 1977.

Status Characteristics and Social Interaction: An Expectation States

Page 110: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

100

Approach. New York: Elsevier. Bierhoff, Hans W., Ernst Buck, and Renate Klein. 1986. “Social Context and

Perceived Justice.” Pp. 165-185 in Justice in Social Relations, edited by H. W. Bierhoff, R. L. Cohen, and J. Greenberg. New York: Plenum.

Blackburn, McKinley and Todd Vermilyea. 2006. “A Comparison of Unexplained Racial Disparities in Bank-Level and Market-Level Models of Mortgage Lending.” Journal of Financial Services Research 29:125-47. Browne, Irene and Ivy Kennelly. 1999. “Stereotypes and Realities: Images of Black

Women in the Labor Market.” Pp. 302-26 in Latinas and African American Women at Work: Race, Gender, and Economic Inequality, edited by I. Browne. New York: Russell Sage Foundation.

Browne, Irene and Joya Misra. “The Intersection of Gender and Race in the Labor

Market.” Annual Review of Sociology 29:487-513. Brush, Candida, Nancy Carter, Elizabeth Gatewood, Patricia Greene, and Myra Hart. 2004. Gatekeepers of Venture Growth: A Diana Project Report on the Role

and Participation of Women in the Venture Capital Industry. The Kauffman Foundation.

Calem, Paul S., Kevin Gillen, and Susan Wachter. 2004. “The Neighborhood Distribution of Subprime Mortgage Lending.” Journal of Real Estate Finance

and Economics 29:393-410. Carter, Nancy, Candida Brush, Patricia Greene, Elizabeth Gatewood and Myra Hart. 2003. “Women Entrepreneurs Who Break Through to Equity Financing: The

Influence of Human, Social and Financial Capital.” Venture Capital 5:1-28.

Caruso, Eugene M., Doboromir A. Rahnev, and Mahzarin R. Banaji. 2009. “Using Conjoint Analysis to Detect Discrimination: Revealing Covert Preferences from Overt Choices.” Social Cognition 27:128-37.

Charles, Kerwin Kofi, Erik Hurst, and Melvin Stephens. 2008. “Rates for Vehicle Loans: Race and Loan Source.” The American Economic Review 98: 315-20. Cohen, Mark A. 2007. “Imperfect Competition in Auto Lending: Subjective Markup, Racial Disparity, and Class Action Litigation.” Vanderbilt Law School, Law

and Economics Research Paper. Collins, Patricia Hill. 1990. Black Feminist Thought: Knowledge, Consciousness, and the Politics of Empowerment. New York: Routledge.

Page 111: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

101

Conway, Michael, M. Teresa Pizzamiglio, and Lauren Mount. 1996. “Status, Communality, and Agency: Implications for Stereotypes of Gender and Other Groups.” Journal of Personality and Social Psychology 71:25-38.

Cook, Karen S. 1975. "Expectations, Evaluations, and Equity." American Sociological

Review 40:372-88. Cook, Karen and Alexandra Gerbasi. 2009. “Trust.” Pp. 218-41 in The Oxford

Handbook of Analytical Sociology, edited by P. Hedstrom and P. Bearman. Oxford: Oxford University Press.

Correll, Shelley J., Stephen Benard, and In Paik. 2007. “Getting a Job: Is There a

Motherhood Penalty?” American Journal of Sociology 112:1297-338. Cuddy, Amy J., Susan T. Fiske, and Peter Glick. 2004. “When Professionals Become

Mothers,Warmth Doesn’t Cut the Ice.” Journal of Social Issues 60:701-18. Devine, Patricia G. and Andrew J. Elliot. 1995. “Are Racial Stereotypes Really

Fading? The Princeton Trilogy Revisited.” Personality and Social Psychology Bulletin 21:1139-50.

Diekman, Amanda B. and Alice H. Eagly. 2000. “Stereotypes as Dynamic Constructs: Women and Men of the Past, Present, and Future.” Personality and Social Psychology Bulletin 26:1171-88. Eberhardt, Jennifer L., Phillip Atiba Goff, Valerie J. Purdie, and Paul G. Davies. 2004. “Seeing Black: Race, Crime, and Visual Processing.” Journal of Personality and Social Psychology 87:876-93. Foschi, Martha. 1989. “Status Characteristics, Standards and Attributions.” Pp. 58-72 in Sociological Theories in Progress, Theories in Progress, vol. 3, edited by J. Berger, M. Zelditch Jr., and B. Anderson. Newbury Park, CA: Sage. Foschi, Martha. 2000. “Double Standards for Competence: Theory and Research.” Annual Review of Sociology 26: 21-42. Foschi, Martha, Larissa Lai, and Kirsten Sigerson. 1994. “Gender and Double Standards in the Assessment of Job Applicants.” Social Psychology Quarterly 57:326-39. Foschi, Martha, Kirsten Sigerson, and Marie Lembesis. 1995. “Assessing Job Applicants: The Relative Effects of Gender, Academic Record, and Decision Type.” Small Group Research 26:328-52. Foschi, Martha and Jerilee Valenzuela. 2007. “Selecting Job Applicants: Effects from

Page 112: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

102

Gender, Self-Presentation, and Decision Type.” Social Science Research 37:1022-38.

Green, Paul E., Abba M. Krieger, and Yoram Wind. 2001. “Thirty Years of Conjoint Analysis: Reflections and Prospects.” Interfaces 31:56-73. Greene, Patricia G., Candida G. Bush, Myra M. Hart, and Patrick Saparito. 2001. “Patterns of Venture Capital Funding: Is Gender a Factor?” Venture Capital: An International Journal of Entrepreneurial Finance 3: 63-83. Greenwald, Anthony G. and Mahzarin R. Banaji. 1995. “Implicit Social Cognition

Research: Attitudes, Self-Esteem and Stereotypes.” Psychological Review 102:4-27.

Gueorguieva, Ralitza and John H. Krystal. 2004. “Move Over ANOVA” Progress in Analyzing Repeated-Measures Data and Its Reflection in Papers Published in

The Archives of General Psychiatry.” Archives of General Psychiatry 61:310-7.

Harris-Lacewell, Melissa. 2001. “No Place to Rest: African American Political

Attitudes and the Myth of Black Women’s Strength.” Women and Politics 23:1-33.

Harrod, Wendy J. 1980. “Expectations from Unequal Rewards.” Social Psychology

Quarterly 43:126-30. Immergluck, Daniel and Marti Wiles. 1999. Two Steps Back: The Dual Mortgage Market, Predatory Lending, and the Undoing of Community Development.

Chicago, IL: Woodstock Institute. Jasso, Guillermina and Peter H. Rossi. 1977. “Distributive Justice and Earned

Income.” American Sociological Review 42:639-51. Jeffries, Frank L. and Richard Reed. 2000. “Trust and Adaptation in Relational

Contracting.” Academy of Management Review 25:873-82. Jiang, Jiming. 2007. Linear and Generalized Linear Mixed Models and Their

Applications. New York: Springer. Kennelly, Ivy. 1999. “‘That Single-Mother Element’: How White Employers Typify Black Women.” Gender and Society 13:168-92. King, Maryon F. and Gordon C. Bruner. 2000. “Social Desirability Bias: A Neglected Aspect of Validity Testing.” Psychology and Marketing 17:79-103.

Page 113: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

103

Krueger, Charlene and Lili Tian. 2004. “A Comparison of the General Linear Mixed Model and Repeated Measures ANOVA Using a Dataset with Multiple Missing Data Points.” Biological Research for Nursing 6:151-7. Lane, Kristin A., Mahzarin R. Banaji, Brian A. Nosek, and Anthony G. Greenwald. 2007. “Understanding and Using the Implicit Association Test: IV. What We

Know (So Far) about the Method.” Pp. 59-102 in Implicit Measures of Attitudes: Procedures and Controversies, edited by B. Wittenbrink and N. S. Schwarz. New York: Guilford.

Lawton, Rachel. 1996. “Preapplication Mortgage Lending Testing Program: Lender

Testing by a Local Agency.” Pp.611-22 in Mortgage Lending, Racial Discrimination, and Federal Policy, edited by J. Goering and R. Wienk. Washington, DC: Urban Institute Press.

Lerner, Melvin J. 1965. “Evaluation of Performance as a Function of Performer’s

Reward and Attractiveness.” Journal of Personality and Social Psychology 1:355-60.

Mitchell, Jason P., Daniel L. Ames, Adrianna C Jenkins, and Mahzarin R. Banaji.

2009. “Neural Correlates of Stereotype Application.” Journal of Cognitive Neuroscience 21:594–604.

Munnell, Alicia H., Geoffrey M. B. Tootell, Lynn E. Browne, and James McEneaney. 1996. “Mortgage Lending in Boston: Interpreting HMDA Data.” American Economic Review 86:25-53. Oliver, Melvin L. and Thomas M. Shapiro. 1995. Black Wealth, White Wealth: A New Perspective on Racial Inequality. New York: Rutledge. Orme, Bryan K. 2009. Getting Started with Conjoint Analysis: Strategies for Product

Design and Pricing Research. Madison, WI: Research Publishers LLC. Ouchi, William G. 1981. Theory Z: How American Business Can Meet the Japanese

Challenge. Reading, MA: Addison-Wesley. Pager, Devah. Marked: Race, Crime, and Finding Work in an Era of Mass Incarceration. Chicago: The University of Chicago Press. Pager, Devah and Hana Shepherd. 2008. “The Sociology of Discrimination: Racial Discrimination in Employment, Housing, Credit, and Consumer Markets.” Annual Review of Sociology 34:181-209. Peters, Richard G., Vincnet T. Covello, and David B. McCallum. 1997. “The

Page 114: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

104

Determinants of Trust and Credibility in Environmental Risk Communication: An Empirical Study.” Risk Analysis 17:43-54.

Pope, Devin G. and Justin R. Sydnor. 2011. “What’s in a Picture? Evidence of

Discrimination from Prosper.com” Journal of Human Resources 46:53-92. Prentice, Deborah A., and Erica Carranza. 2002. “What Women and Men Should Be, Shouldn’t Be, Are Allowed to Be, and Don’t Have to Be: The Contents of Prescriptive Gender Stereotypes.” Psychology of Women Quarterly 26:269-81. Quillian, Lincoln. 2006. “New Approaches to Understanding Racial Prejudice and

Discrimination.” Annual Review of Sociology 32: 299-328. -------. 2008. “Does Unconscious Racism Exist?” Social Psychology Quarterly 71:6- 11. Ravina, Enrichetta. 2008. “Love and Loans: The Effect of Beauty and Personal

Characteristics in Credit Markets.” American Law and Economics Association Annual Meetings.

Rempel, John K., John G. Holmes, and Mark P. Zanna. 1985. “Trust in Close

Relationships.” Journal of Personality and Social Psychology 49:95-112. Ross, Stephen L. and John Yinger. 1999. “Does Discrimination in Mortgage Lending

Exist? The Boston Fed Study and Its Critics.” Pp. 43-83 in Mortgage Lending Discrimination: A Review of Existing Evidence, edited by M.A. Turner and F. Skidmore. Washington, DC: Urban Institute.

-------. 2002. The Color of Credit: Mortgage Discrimination, Research Methodology, and Fair-Lending Enforcement. Cambridge, MA: MIT Press. Rudman, Laurie A. and Peter Glick. 2001. “Prescriptive Gender Stereotypes and Blacklash Against Agentic Women.” Journal of Social Issues 57:743-62. Siskin, Bernard R. and Leonard A. Cupingood. 1996. “Use of Statistical Models to

Provide Statistical Evidence of Discrimination in the Treatment of Mortgage Loan Applicants: A Study of One Lending Institution.” Pp. 451-68 in Mortgage Lending, Racial Discrimination, and Federal Policy, edited by J. Goering and R. Wienk. Washington, DC: Urban Institute Press.

Smith, Robin and Michelle DeLair. 1999. “New Evidence from Lender Testing: Discrimination at the Pre-Application Stage.” Pp. 23-41 in Mortgage Lending Discrimination: A Review of Existing Evidence, edited by M.A. Turner and F.

Skidmore. Washington, DC: Urban Institute.

Page 115: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

105

Smith, Shana and Cathy Cloud. 1996. “The Role of Private, Nonprofit Fair Housing Enforcement Organizations in Lending Testing.” Pp. 589-610 in Mortgage Lending, Racal Discrimination, and Federal Policy, edited by J. Goering and R. Wienk. Washington, DC: Urban Institute Press.

Stewart, Penni A. and James C. Moore, Jr. 1992. "Wage Disparities and Performance

Expectations." Social Psychology Quarterly 55:70-85. Thye, Shane R. and Christine Witkowski. 2003. "The Status Value Theory of Power:

The Effect of Status and Resource Differentiation on Power in Exchange." Paper presented at the annual meeting of the American Sociological Association, Atlanta, August.

Veblen, Thorstein. [1899] 2005. The Theory of the Leisure Class. Gloucestershire,

UK: Dodo Press. Webster, Murray Jr. and James E. Driskell Jr. 1983. “Beauty as Status.” American Journal of Sociology 89:140-65. Williams, Richard, Nesiba, Reynold, and Eileen Diaz McConnell. 2005. “The Changing Face of Inequality in Home Mortgage Lending.” Social Problems 52:181-208.

Page 116: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

106

4. REWARD INTERVENTIONS: THE EROSION OF SOCIAL INEQUALITY?

The perpetuation of social inequality is one of the grand social issues of the

modern era. Many social programs aim to reduce such inequality by rewarding

disadvantaged groups through increasing their access to valued occupations, positions

of authority, monetary resources, and esteemed awards. Some of these programs

include merit-based awards for minority scholars, foundations that provide

microfinance services to women in developing countries (such as the MicroLoan

Foundation and Friends of Women’s World Banking), comparable worth policies, and

affirmative action more broadly. At the core of many of these initiatives is a two-

pronged attack against discrimination: first, due to discrimination we should have

policies in place to better assure fairness and equality, and, second, the hope that the

redistribution of financial, educational, and occupational advancements to

disadvantaged, low status groups will help to increase their social standing and

opportunities over time.

Many have argued that one of the key reasons why inequality remains

pervasive is that those who control access to resources and rewards are primarily

advantaged groups who, intentionally or not, continually reallocate these rewards to

similar, high status others (Jacobs 1989; Kanter 1977; Padavic and Reskin 2002;

Reskin 1988; Williams 1992). By having limited access to these rewards, such as

obtaining college degrees from prestigious schools, high status occupations, and

public acknowledgments of achievements, low status groups continue to be at a

disadvantage. In the ubiquitous case, reward distributions are congruent with status

Page 117: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

107

groups, such that high status actors are associated with highly desired rewards and low

status actors with unattractive or devalued rewards. When the situation is reversed,

however, as is the case in some of the aforementioned social programs, the status

disadvantaged may enjoy an increase in their assumed competence and ability by

virtue of these rewards. Perhaps, even the overall estimation of those who have the

same characteristics may be positively affected over time.

The process of using rewards as an intervention can be fraught with

complexity, however. While some types of rewards convey a definite sense of

prestige, value, and ability regardless of their possessors, such as Nobel Prizes,

McArthur Genius Grants, and some high-ranking political offices, the meaning of

numerous others may not be as concrete and immune to the status of those with whom

they are affiliated. Additionally, once valued objects are more freely bestowed or

obtained and begin to filter throughout a population, a tipping point may be reached

wherein the value of the status symbols begins to be contaminated. With newer

rewards that do not have a strong valuation, there may be an even quicker tendency for

the objects to lose their meaning once they are noticeably coupled with disadvantaged

groups.

Indeed, there are numerous instances of groups trying to protect the value of

their awards by further excluding those who are eligible to receive them. For

example, the Theatre Development Fund has bestowed the Wendy Wasserstein Prize

to young female playwrights every year since 2007; however, in 2010, they initially

concluded that that year’s eligible playwrights were not “worthy” enough for the

reward (Healy 2010). Amid much criticism, the Fund later decided to alter how it

Page 118: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

108

evaluates each candidate so that its members felt as though they could legitimately

give this prize to one of the female candidates, thereby further protecting the meaning

of the Wasserstein Prize in their estimation. As one member of the committee, Ms.

Ettinger, stated, “As a funder, we must be able to insure the integrity of the prize”

(Healy 2010).

Understanding the relationship between rewards and those who control them is

an important step towards fully understanding how rewards might work to reduce

social inequality. This research seeks to ascertain whether the reward alone is enough

to diminish the effect of status differences on processes of inequality, or whether, once

a reward is affiliated with disadvantaged groups, the reward itself looses value and

prestige through this association, thereby diminishing any power it may have had to

affect social change.

STATUS-BASED INEQUALITY

While there are many different types of inequality, this research focuses on

status-based inequality. Inequality of this sort arises when people or groups are

distinguished by certain characteristics that convey divergent levels of status, esteem,

competence, and prestige. Expectation states theory is a research program within

structural social psychology that seeks to explicate these status processes and their

effect on interpersonal relations and inequality. Within this program, status

characteristics theory (hereafter SCT) elucidates how existing status differences

pattern behaviors related to having differential levels of power and influence in small

groups (Berger, Cohen and Zelditch 1966, 1972). This theory applies to groups who

Page 119: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

109

are working together to reach a goal (i.e., who are collectively and task-oriented), such

as juries or students working on a group project for a class. These status processes

also operate when a person evaluates the work and contributions of others who have

differing states of at least one, salient status characteristic (Correll et. al 2007; Foschi,

Lai, and Sigerson 1994; Foschi, Sigerson, and Lembesis 1995; Foschi and Valenzuela

2007; Webster and Driskell 1983).

There are two main kinds of personal characteristics that can distinguish group

members: diffuse and specific status characteristics. Diffuse status characteristics are

culturally defined, socially significant characteristics (e.g., gender or race/ethnicity)

that have varying states (e.g., male-female, white-African American). These various

attributes have differential esteem, competence, and prestige valuations as defined by

the dominant culture that correspond to the level of performance ability a person with

a particular state is assumed to have (Berger and Fisek 2006). Specific status

characteristics are associated with the ability to perform particular tasks, such as

computer skills or business aptitude.

The basic form of SCT is as follows. Once a group is differentiated by at least

one diffuse status characteristic or by a characteristics that is relevant to the task,

individuals will assign expectations about the performance and potential contributions

of group members based on the valuation of the states of their status characteristics.

The characteristic will be relevant to expectations of individual’s performance on the

group’s task unless it is directly challenged (the burden of proof principle). Behavioral

inequalities favoring the actors who have highly-valued status characteristics will then

emerge with respect to opportunities granted to speak, actual level of participation,

Page 120: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

110

evaluations of others' performance, and the influence members have to change the

mind of others while solving the problem. The power and prestige hierarchy and

expectation hierarchy are mutually reinforcing, and they will remain highly stable over

time (Berger and Conner 1974).

Status Interventions

In addition to examining the formation and continuation of the power and

prestige hierarchy, researchers have also investigated the impact of certain

interventions intended to reduce these inequalities. These interventions have entailed

the activation of inconsistent status elements to neutralize or overcome the

consequences of negatively valued states of diffuse status characteristics. Pugh and

Wahrman (1983) demonstrate that when experimenters inform a mixed-gender group

that the female members are highly competent at an ability that is relevant to the

current task, while the male partners are not, the males’ influence is greatly reduced.

This alteration in level of influence also continued through a second identical task with

new partners. Markovsky, Smith, and Berger (1984) also show that a similar type of

ability intervention reduced, and in some cases reversed, the effect of education level

on deference behaviors. This effect also continued when the subjects worked with a

second partner on a different task, although the strength of the effect was not as strong

in the second phase.

According to SCT, when status characteristics differentiate people who are

participating on a task together that has clear goals, people process all relevant status

information practically instantaneously in order to form an overall expectation of

Page 121: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

111

whose performance will contribute the most and be the most valued in terms of

solving the problem at hand. When the characteristics are consistent, the process is

relatively straightforward; however, when the states are inconsistent, people must

process both sets of information to form their performance expectations (Berger,

Fisek, Norman, and Zelditch 1977; Berger, Norman, Balkwell, and Smith 1992). This

process occurs through the combining of organized subsets, wherein the

characteristics’ states of positive and negative valence are sorted into unique sets of

information. An attenuation principle governs the processing of these subsets

wherein each additional piece of information within a subset has marginally less

impact on changing the aggregated performance expectation (Berger et al. 1977;

Berger et al. 1992). If a group member has many positive characteristics but markedly

fewer negative attributes, these negative traits will still have great importance because

the numerous positive characteristics are progressively less consequential in forming a

final expectation of who will be more or less active and influential in the group.

Thus, inconsistent status information is not disregarded regardless of how

much of this information exists, at least in situations in which SCT is applicable.

Status interventions have an impact precisely because they introduce inconsistent

information regarding each of the group members. When these discrepancies are

directly relevant to an imminent task through specific status attributes, these ability

structures can even override the effect of some diffuse status characteristics

(Markovsky et al. 1984). This research has shown the fruitfulness of status

interventions, but the interventions studies thus far are largely based on providing

ability distinctions. In application, providing aptitude information may be one way of

Page 122: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

112

lessening status-based inequality, but there are also many programs that aim for the

same goal by providing rewards and valued acknowledgements to disadvantaged

groups. Theoretically, the impact of rewards should be as consequential to achieving

this end. I will now turn to a discussion of rewards and how they may be relevant to

the diminution of status hierarchies.

Rewards

While SCT is mainly concerned with the status of people, another branch of

expectation states theory, the status value theory of distributive justice (hereafter

SVDJ), is concerned with the status of objects and positions (Berger, Zelditch,

Anderson, and Cohen 1972). The theory is based on a reformulation of Veblen's

([1899] 2005) notion of honorific value in that desired objects, tangible or not, can

symbolize status and social standing. Elements in the social world, such as people,

objects, or positions, come to have status value when they are uniquely related to

elements of the social world that do. If these elements are consistently valued, then

the related non-status valued object will acquire the status value of these elements

through a spread of status value process wherein the preexisting valuation is ascribed

onto the new object. Rewards thereby come to have status value, and are reified

markers that connote prestige and honor (Veblen ([1899] 2005).

Central to the current proposed research project is reward expectations theory

(hereafter RET), which is an interrelation of SCT and SVDJ (Berger, Fisek, Norman,

and Wagner 1983). RET explicates how rewards are allocated according to status

differences. The general prediction derived from RET is that people, regardless of

Page 123: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

113

their relative status, will allocate larger amounts of valued rewards to high-status, as

opposed to low-status actors, while low-status actors will either be allocated fewer

valued rewards or devalued objects. The more divergent the status differences

between actors are, the greater the reward inequalities will be; while the resulting

reward distribution will be more equivalent when the status elements are more

inconsistent. Additionally, the reverse of this process also operates in that those who

are allocated more rewards in a task situation (e.g., those who have higher salaries)

will come to have more influence in the group (Bierhoff, Buck and Klein 1986; Cook

1975; Harrod 1980; Lerner 1965; Stewart and Moore 1992).

Thus, people expect those with positively valued status characteristics to

possess highly status-valued objects and positions due to the ascription of their

characteristics’ assessment to their resources. The effect of rewards is also potentially

quite powerful in that when those who are allocated highly esteemed rewards are also

differentiated by an initially non-valued characteristic, the status value of the reward

spreads to the associated state of the nominal characteristic, thereby beginning the

creation of a new status characteristic (please see Chapter 1). The spread of status

value is an especially important process that may even affect the valuation of

preexisting status characteristics as well.

Reward Allocation and Status Characteristics

Rewards provide status information about their possessors, and those who

acquire highly valued rewards come to have more influence in groups than those who

do not (Bierhoff et al. 1986; Cook 1975; Harrod 1980; Lerner 1965; Stewart and

Page 124: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

114

Moore 1992). If rewards are reified status objects that convey relative status

advantages to their possessors, this implies that when status disadvantaged people

control highly valued rewards, they should receive a status advantage by virtue of this

ownership. When the status value of the reward and salient status characteristics is

consistent, then the possessor of the reward should experience an attenuated increase

in her relative influence in a group.

If the reward and salient status characteristics are inconsistent, however, the

situation becomes unbalanced, which produces tension and prompts the actors to bring

these social elements back to equilibrium (Berger et al. 1972). In this situation, the

status value of the reward should be processed along with that of the characteristics, as

the principle of organized subset combining suggests. As Berger, Fisek, Norman, and

Wagner (1983) state: “allocating a positive state of a goal object to an actor increases

the actor’s overall task expectation, while allocating the negative decreases the actor’s

overall task expectation.” Accordingly, reward allocations that are inconsistent with

the states of active status characteristics may produce at least a change in the particular

actor’s status and influence in the direction of the valence of the reward’s valuation,

relative to similar others without such rewards. Therefore, I hypothesize:

Hypothesis 1: As the status value of the actors’ rewards increases, the actors will be evaluated as having greater overall status assessments.

Hypothesis 2: As the status of the actors increases, the greater the behavioral

expectation advantage attributed to these actors will be. Once the reward is relevant to the state of a status characteristic, the reward

level’s status value may not only affect the actor’s relative status in the immediate

encounter, but the valuation may spread to the actors’ associated characteristics

Page 125: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

115

themselves. If this spread does occur, the evaluation of the characteristic overall

should change congruently to the valuation of the rewards. Pugh and Wahrman’s

(1983) and Markovsky et al.’s (1984) intervention work suggests that this may be the

case in that interventions aimed at altering the ability-level of disadvantaged groups

continued to have a positive effect on the relative influence of those from

disadvantaged status groups in subsequent encounters with new partners. The

experience of inconsistent ability information may therefore have altered the overall

valuation of those from disadvantaged status groups. This leads to the following

hypothesis as related to reward interventions:

Hypothesis 3: As the status value of actors' rewards increases, the corresponding status value of the states of their preexisting status characteristics will also increase.

Interventions that provide esteemed rewards to disadvantaged status groups

may help to increase the relative power and prestige of these groups precisely because

these possessions convey additional status information about those who acquire them

and, consequently, others who are similar to them. Support for these hypotheses

would offer a social and cognitive mechanism that could reduce this form of

inequality.

Reward Contamination

When rewards are inconsistently allocated to status groups, the status and

influence of the disadvantaged actors and their pertinent characteristics may increase,

while the advantages previously experienced by high status actors may decrease as

Page 126: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

116

SVDJ and RET posit. Nevertheless, the spread of status value is a complex process,

and the very act of associating a valued object with a low status person may

contaminate the value of the object instead of increasing the prestige of the actor. For

instance, when occupations and disciplines feminize, the wages, esteem, and resources

that were previously enjoyed by these workers, such as having advancement

opportunities and control over working conditions, tends to decrease (Catanzarite

2003; Cohn 1985; England, Allison, and Wu 2007; Strober and Arnold 1987), and this

may be at least partially due to the devaluation of “women’s work.”

Objects become imbued with status when they are uniquely associated with

status groups (Berger et al. 1972). Once this association is made, the reward becomes

a reified marker of the status of that particular group. However, if the reward becomes

affiliated with a group of conflicting status value, particularly when the reward is

relatively novel and its meaning may still be in flux, the estimation of the reward and

what it connotes may shift to reflect the estimation of this new affiliation. In the case

of interventions, if the reward’s significance is not immutable, its allocation to those of

disadvantaged backgrounds may be of little to no assistance, as the audience would

come to alter only the value of the reward to reflect that of its new possessors. This

possibility leads to the following hypothesis:

Hypothesis 4: When highly valued rewards are made relevant to devalued states of a status characteristic, the status of the reward will decrease.

If support is found for hypothesis 4, we would not expect there to be a

significant alteration in the relative influence associated with the actors in this

situation as compared to those of the same states of the characteristic but without

Page 127: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

117

rewards. Additionally, we would not expect there to be an alteration in the status

value of the associated characteristics. Thus, if inconsistent reward allocations alter

the impressions of the meaning of the reward instead of the person possessing it, the

use of rewards for status interventions may not have a simple mechanism of

combining status information as RET contends1.

EXPERIMENTAL DESIGN

The purpose of this experiment is to create a reward by making a color-coding

scheme relevant to preexisting status characteristics, with the color scheme becoming

the marker of either high or low status by virtue of this association. Depending on the

condition, this reward marker is then consistently or inconsistently associated with

another status characteristic, education level, to assess how the valuation of the marker

affects the influence and qualities of the characteristic. The consistency of the reward

allocation denotes this study’s two conditions.

Cover Story

In this experiment, participants are told that the purpose of the study is to

understand how groups who work together only via a computer and not necessarily

simultaneously, work together to make decisions and come to consensus. The

participants are informed that they are the third members of a three-person team.

Their role in the team is to evaluate the performance of their two teammates who

1 Due to the nature of this study’s experiment and the sensitivity of the measurement tools, hypotheses 1 through 3 are set in opposition to hypothesis 4. A more thorough examination of the reward process may find that the spread of status value affects the impressions of the rewards, their possessors, or both, thereby working together dynamically overtime.

Page 128: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

118

worked together on a language task as well as supplying their own answer to the task.

To help ensure that the participants have a sense of group cohesiveness and a shared

fate with their other teammates, the participants are told that their answers will be

combined with those of their partners to generate a final team score. All participants

were compensated with course credit, but participants are informed that teams that

perform extremely well on the task will receive a bonus credit. The participants’

partners are actually simulated actors to control for the task cues given by the partners,

the relative status of the partners, and the way the interaction unfolds.

Creation of Status Valued Rewards

In the first phase of the study, the rewards are created by virtue of assigning a

color-coding scheme to extant status characteristics. The procedure follows that

which has been used in status construction research (Ridgeway, Backor, Li, Tinkler,

and Erickson 2009; Ridgeway, Boyle, Kuipers, and Robinson 1998; Ridgeway and

Erickson 2000), as they are a parsimonious and effective way of creating such an

association. At the beginning of the session, the participants complete a brief survey

in which they indicate their gender, age, and their grade point average. To introduce a

novel ability trait to further differentiate the associated reward levels, participants then

complete an individual contrast sensitivity test wherein they must select between two

pictures which contains the greater amount of white area for a total of ten rounds. In

actuality, all of the pictures contain the same proportion of white space, but

participants are led to believe that the test is indicative of visual ability.

Page 129: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

119

The participants then examine an information sheet that details the

aforementioned demographic information of previous participants in addition to their

scores on the contrast sensitivity task. This list is color-coded, with each person’s

information highlighted in either green or yellow. The participants are prompted to

carefully examine the information sheet, including the color-coding scheme,

presumably while the computer loads the partners’ responses on the language task.

On the information sheet, the status backgrounds (i.e., age, gender, education

level, and contrast sensitivity test scores) of paired former participants are provided.

Within each pair, one partner always has higher states on the status characteristics than

the other, though not all of the status characteristics are salient within the pair to add

believability; some paired participants may both be high school students or are the

same gender, for instance. The entries for the status-advantaged partners are always

highlighted in green, while the relatively lower status partners are highlighted in

yellow. It is during this point that the association between the valuation of these

characteristics’ states and each color should be made, thereby allowing the colors to

become differentially valued rewards.

This should occur through the spread of status value process wherein the

previously non-valued color scheme takes on a relatively higher or lower status value

as determined by the associated status advantages of the various characteristics. Once

this transpires, the colors should be seen as either esteemed or devalued status

markers. This procedure is similar to that used by Thye (2000) to create status valued

rewards wherein color-coded bargaining chips were made relevant to participants’

differing status attributes prior to negotiating with each other using these chips. In

Page 130: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

120

Thye’s (2000) study, as in this one, the different colors are purportedly used to help

identify the players’ actions.

The participants are then informed about their own two teammates whose

group performance they are to evaluate. In all conditions, one partner is a local

graduate student from a private university and the other is a regional, South Bay high

school student. The color-coding scheme is then assigned to each of these partners by

highlighting this background information and their choices and responses on the

ensuing language task in each of the two status marker colors. In the reward-

consistent condition, each partner’s information and decisions are marked with the

color that connotes the same level of status as their educational level (i.e., the graduate

students are indicated in green and the high school students’ responses are highlighted

in yellow). In the reward-inconsistent condition, the partners are marked with the

color of the opposing valuation (i.e., green for the high school student and yellow for

the graduate student).

Influence Task

The participants then evaluate their partners’ responses on a meaning insight

language task (hereafter MIT). The MIT is a frequently used task by expectation

states researchers to measure influence. On this task, participants are presented with

an English word and two possible ancient language root woods. The overall goal of

the task is to select which root word was used to create the current English word. In

reality, there are no correct answers to this task.

Page 131: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

121

Due to the nature of this three-person group, the traditional MIT’s structure has

been altered. In the classic version of MIT, the participant makes an initial choice,

and subsequently learns what her partner has independently selected before making a

final choice between the two root words. In this study’s revision to this task, the

participants still give an initial answer to each question. The participants then see a

color-coded reproduction of their partners’ own initial responses, as well as a short

justification for why they made their selections. The participants then provide their

own final answer and provide any comments they have about their partners’ reasoning.

The choices and reasoning given by the partners are actually scripted so that

the two partners’ initial selections differ on twenty out of the twenty-five rounds of the

MIT, as in the classic MIT, and to control for the justifications provided by each

partner2. Preferences for the reasoning of one partner over the other cannot be

affected by the veracity of the argument as there is no correct answer, but may provide

further means for the participants to justify their agreement with one partner over the

other. Nevertheless, the root words and justifications were pretested without

information about the partners’ status to ensure that participants equally selected each

word in their initial and final decisions.

In the classic version of the MIT, the proportion of rounds in which the

participants stay with their initial selection in the twenty disagreement rounds

constitutes the measure of influence, p(s). As p(s) decreases, the participant is more

heavily influenced by the initial selection of her partner. For this three-partner

version, there are now two partners who disagree with each other, and the participant

2 “Pum has the same number of letters as ‘eye’” is an example of one such rationalization.

Page 132: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

122

who must decide which partner she believes is “correct.” In this revised manipulation,

the measure of influence instead pertains to the proportion of times the participant

changed her initial choice to agree with particular partner (i.e., p(c)a or p(c)b for

partners a and b). As p(c)x increases, the participants are altering their initial opinion

to agreeing more with one partner over the other and are thereby more heavily

influenced by this partner.

Measuring Status of the Partners, the Color Scheme, and the States of the Status

Characteristic

Finally, subjects fill out an exit questionnaire that contains questions pertaining

to the qualities of their partners, their rewards, and the salient education level status

characteristic. To assess the personal status of the partners, the participants provide

their own assessments of each partner’s status and competence. They also answer the

same questions related to how they think most people would evaluate high school and

graduate students. These questionnaire items are also adapted from those used in

status construction research (Ridgeway et al. 1998). The following adjective pairs

anchor the status questions: respected/not respected, low status/high status, and

leader/follower. The competence questions pertain to the pairs of:

competent/incompetent, capable/incapable, and knowledgeable/not knowledgeable.

The participants answer these questions with a virtual slide-rule that measures their

responses along a continuous 100-point scale. The status, and competence questions

are averaged to create scales to analyze these components of status value.

Page 133: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

123

Lastly, the participants complete a series of questions related to their

estimation of the value of the rewards, i.e., the color-coding scheme. They first

indicate whether they think it is more difficult to have their responses highlighted with

either green or yellow, whether they think a person would get more respect if her

answers were identified with green or yellow, whether a person would have more

influence if their responses were in one of the two colors, and whether most people

would prefer to have their responses highlighted in green or yellow. Finally,

participants answer how they think “most people” would rate a person who is signified

in this study with green and with yellow. The aforementioned status and competence

questions are again used to anchor these scales and are also averaged to create the two

scales related to the rewards’ status value.

RESULTS

Sample

A total of 41 people participated in this study: 22 participants in the reward-

consistent condition and 19 in the reward-inconsistent condition. Due to time

constraints on data collection, this study was run both in-person and online. The main

difference between the two mediums is that while the in-person participants

continually had the information sheet in front of them while taking the MIT and

answering the questionnaire, the online participants only viewed an electronic copy of

the information sheet before learning about their particular partners. Both groups

viewed this electronic version of the information sheet for at least one minute, and all

Page 134: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

124

participants are reminded of their partners’ level of education and associated reward

levels prior to answering the questionnaire.

There are a total of 17 in-person and 24 online participants (please see Table

1). While there are some demographic differences between the two samples, their

responses and behaviors did not differ by the key theoretical variables. The online

sample contains more females (z = 2.9, p < .01, two-tailed test), Latinos, and those of

multiple ethnic backgrounds (z = 2.6, p < .05; z = 11.7, p < .000; two-tailed tests,

respectively); but there are fewer whites and African Americans (z = 2.7, p < .01; z =

4.2, p < .000; two-tailed tests, respectively), and this sample is generally younger than

the in-person sample (t = 2.2, p < .05, two-tailed test).

The demographic background of those in either condition also slightly differs.

When rewards are allocated inconsistently according to the partners’ education level,

there are fewer females (z = 3.3, p < .01, two-tailed test) and Asians (z = 2.1, p < .05,

two-tailed test) but more Latinos (z = 3.9, p < .000, two-tailed test) who participated as

compared to those in the reward consistent condition. Lastly, across conditions and

study medium, participants rated the status and competence of both partners combined

as slightly above the midpoint of the scale and changed their initial MIT answers on

14-17% of the rounds.

Page 135: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

125

Table 4.1. Means or Proportions of Demographic Characteristics by Study Medium and Condition Study Medium Condition

Variables In-Person Online Reward

Consistent Reward

Inconsistent

Female .5 .7** .7 .5**

Age 23.2 (6.3)

21.6* (5.1)

22.3 (5.7)

22.2 (5.7)

Ethnic background White .4 .3** .3 .3

African American .1 .0*** .1 .1 Latino .1 .2* .1 .2*** Asian .4 .5 .5 .4*

Other .0 .1*** .1 .1

Status Scale 64.0 (20.0)

64.3 (17.3)

64.1 (19.3)

64.3 (17.7)

Competence Scale 65.2 (18.5)

64.3 (17.3)

65.8 (19.5)

64.3 (17.7)

Change initial answer .2 .2 .1 .2

N = 17 24 22 19 Note: SDs in parentheses; +p<.10; *p<.05; **p<.01; *** p<.001 (two-tailed tests).

Reward Value

In the first portion of this study, the color-coding scheme is associated with

extant status characteristics in order to create the reward. This scheme must become

imbued with status value before it can possibly affect the status hierarchy between the

participants’ partners. It is therefore important to first assess whether the association

between the two types of highlighting and the status characteristics salient in the

information sheet created the color reward. The results indicate that this is the case.

Across conditions, participants are more likely to state that the purported high reward

Page 136: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

126

color is more difficult to obtain than the color that was associated with devalued states

of the status characteristics (z = 11.1, p < .000, two-tailed test), that the high reward

color conveys greater respect and influence (z = 6.3, p < .000; z = 5.2, p < .0000; two-

tailed tests, respectively), and that they believe that most people would prefer to have

the high reward color (z = 9.8, p < .000, two-tailed test) (results not reported).

Participants are more likely to believe in the color scheme’s differential

appraisals when they are in the reward consistent as opposed to the inconsistent

condition, however (please see Table 2). While those in both conditions believe that it

is more difficult to obtain the valued color, those in the inconsistent condition are less

likely to state that the intended high reward conveys a greater level of respect and

influence. They are also less likely to state that most people would prefer to have the

high reward than those in the consistent condition. This preliminarily indicates that

the scheme’s reward value may begin to be redefined by the partners’ education level

such that the valued colors loses status when it is associated with a lower status

partner.

Page 137: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

127

The participants’ responses on the status and competence scales related to the

value of each reward level also support this conclusion3. As displayed in Table 3, the

intended high reward conveys a greater degree of status and competence onto those

who possess it (t = 6.0, p < .000; t = 5.3, p < .000; two-tailed tests, respectively) only

when the reward is consistently assigned according to the partners’ relative education 3 The alpha-levels of these scales are both .9, and exploratory factor analysis also reports one underlying factor for each of these measures.

Table 4.2. Mean Ratings of Reward Traits by Condition

Variables Mean Z-Score

More difficult to obtain high versus low reward

Rewards consistent with partners’ characteristics 0.8

0.7

Rewards inconsistent with partners’ characteristics 0.8

More respect conveyed with high versus low reward

Rewards consistent with partners’ characteristics 0.8

4.4***

Rewards inconsistent with partners’ characteristics 0.5

More influence conveyed with high versus low reward

Rewards consistent with partners’ characteristics 0.7

1.9+

Rewards inconsistent with partners’ characteristics 0.6

Most people prefer to have high versus low reward

Rewards consistent with partners’ characteristics 0.9

4.5***

Rewards inconsistent with partners’ characteristics 0.6

Note: Sample size for this analysis = 41; SDs in parentheses; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 138: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

128

status. In the inconsistent condition, wherein the high school partner receives the

valued reward and the graduate student is given the opposing marker, participants do

not clearly differentiate the status value of the associated rewards.

Page 139: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

129

Mixed-effect linear regression also demonstrates that the qualities conveyed by

the color-coding scheme significantly vary by the consistency of the reward allocation,

even when controlling for the participants’ characteristics (please see Table 4).

Mixed-effect linear regression is used to account for the non-independence of these

status value questions that are repeated across multiple stimuli (Bates 2010; Jiang

2007). These models contain fixed effects related to reward level, condition, and

participant controls. A random effect, as defined by the participants’ identification

code, is also included in these models. The significant main effect of reward level

demonstrates that those in the reward-consistent condition evaluated the reward

associated with the high status partner as conveying greater levels of status (b = 30.8,

p < .000) and competence (b = 28.6, p < .000) as expected. The significant main effect

of condition in each model (b = 21.8, p < .01; b = 13.9, p < .05, respectively) shows

that participants evaluated the status and competence conveyed by the intended

Table 4.3. Mean Ratings of Reward Levels’ Status and Competence by Condition

Condition

Variables Rewards Consistent with

Partners’ Status Characteristics

Rewards Inconsistent with Partners’ Status Characteristics

Mean T-Statistic Mean T-Statistic

Status Scale

High reward 76.7

(11.7) 6.0***

61.9

(19.7) -0.5

Low reward 45.8

(18.6) 65.5

(18.9)

Competence Scale

High reward 77.4

(14.0) 5.3***

65.5

(14.6) 0.7

Low reward 48.7

(19.5) 61.7

(15.5) Note: Sample size for this analysis = 41; SDs in parentheses; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 140: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

130

devalued reward as actually being higher in the inconsistent condition wherein this

particular reward is associated with the high status partner. The significant interaction

term between anticipated reward level and condition in these models (status: b = -34.9,

p < .000; competence: b = -24.8, p < .000) indicates that those in the reward-

inconsistent condition rated the status value of the intended esteemed reward, which is

associated with the low status partner, as being significantly lower than those in the

reward-consistent condition, wherein this marker is linked with the high status partner.

Table 4.4. Estimated Mixed-Effects Linear Regression Coefficients for the Effects of Reward Level on the Status Value Scales

Status Value Scales

Variables Status Scale Competence Scale

Rewards inconsistent with partners’ status characteristics

21.8** (6.5)

13.9* (5.6)

High reward 30.8*** (5.8)

28.6*** (5.3)

Inconsistent x high reward -34.9*** (8.7)

-24.8** (7.9)

Participant Controls: Female participant

4.5 (5.0)

3.2 (4.5)

Age of participant

0.4 (0.4)

0.3 (0.4)

Minority participant 5.3 (5.0)

1.6 (4.7)

Online study 0.1 (4.7)

-0.4 (4.2)

Intercept 30.3* (12.2)

38.3** (11.1)

Random Effect Variance Terma: Intercept

0.0 (0.0)

0.0 (0.0)

Note: Models fit using restricted maximum likelihood (REML); SEs in parentheses; N = 41 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Expressed as a standard deviation.

Page 141: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

131

These results suggest that the color-coding scheme obtains distinct levels of

status value only when the rewards are allocated consistently according to additional

extant characteristics. Even though participants in both conditions viewed the same

information sheet that led to the creation of a stable status marker in the reward-

consistent condition, those in the inconsistent condition did not report a clear

distinction between reward levels. In this inconsistent condition, the reward’s

valuation becomes more vague when it is inconsistently affiliated with the partners’

education background differences.

Behavioral Influence

Even though the status value of the color-coding scheme significantly differs

by condition, does reward-level alter the partners’ influence over the participants?

The data show that when participants changed their initial responses on the meaning

insight task, they are more likely to be influenced by the partners’ educational

background and not their reward level. Figure 1 displays the proportion of trials in

which participants changed their initial responses by condition and the partners’

reward level. When the participants changed their initial responses, they are

significantly more likely to agree with the graduate students’ opinion in both

conditions regardless of reward level. In the consistent condition, participants agree

with the highly rewarded graduate student in 79% of the changed rounds, while in the

inconsistent condition, the reward-disadvantaged graduate student influences the

participants’ in 73% of these particular trials. Mixed-effect logistic regression also

indicates that participants are about equally likely to alter their final choices and, thus,

Page 142: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

132

be influenced in either condition, as demonstrated by the lack of significance of

condition in the first model of Table 5 (b = 0.2, p = .33). When the partners' relative

status differences are taken into account in the second model, participants are

significantly less likely to agree with the highly rewarded high school student partner

than the highly rewarded graduate student (b = -2.6, p < .000). These results indicate

that allocating previously esteemed rewards to those with a disadvantaged status

background does not elevate their relative influence.

a Letters signify statistically significant z-score differences; Capital letters represent that the difference is significant in a two-tailed test (p < .000).

0.8

0.3 0.2

0.7

0.0

0.2

0.4

0.6

0.8

1.0

Consistent with Status Inconsistent with Status Prop

ortio

n of

Cha

nged

Ans

wer

s

Condition

Figure 4.1. Proportion of Trials Participants Changed Their Initial Answer to Agree with Either Partner by Conditiona

High Rewards

Low Rewards

A, C

A, D

B, C

B, D

Page 143: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

133

Table 4.5. Estimated Mixed-Effects Logitstic Regression Coefficients for the Effects of Condition on the Number of Trials Participants Changed Their Initial Answer

Changed Initial Answer

Variables Agreement with Either Partner

Agreement with High Reward Partner

Rewards inconsistent with partners’ status characteristics

0.2 (0.2)

-2.6*** (0.5)

Participant Controls: Female participant

0.1 (0.3)

-0.3 (0.5)

Age of participant

-0.1*** (0.0)

-0.1 (0.1)

Minority participant 0.0 (0.1)

0.1 (0.3)

Online study -0.2 (0.2)

-0.4 (0.5)

Intercept 0.9 (0.7)

3.1* (1.4)

Random Effect Variance Terma: Intercept

0.3 (0.2)

0.0 (1.3)

Note: SEs in parentheses; N = 41 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Expressed as a standard deviation.

Status Assessments of Partners and the Education Status Characteristic

Participants’ evaluations of their partners’ status and competence also indicate

that the rewards did not detectably alter their opinion of their partners4. Instead, the

partners’ relative status and competence is aligned with their educational background.

As reported in Table 6, the highly rewarded graduate student is rated as having

significantly higher status and competence than the high school student with the

devalued reward (t = 3.0, p < .01; t = 3.8, p < .000, respectively). In the reward

4 The alpha-levels for the status and competence scales related to the assessments of each partner and the education status characteristic are .7, .7, .8, and .7, respectively. Exploratory factor analysis also reveals one underlying factor for each of the status and competence scales.

Page 144: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

134

inconsistent condition, however, the reward-disadvantaged graduate student is rated as

having about the same average status as their counterpart in the opposing condition

(consistent graduate student average status score = 71.5; inconsistent graduate student

average status score = 70.9) and significantly higher status than the highly rewarded

high school student (t = 3.0, p < .05). The competence ratings of the two partners in

the inconsistent condition also favors the reward-disadvantaged graduate student,

though this difference is only marginally significant (t = -1.8, p = .08).

Page 145: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

135

Mixed-effect linear regression also supports this conclusion (please see Table

7). The significant, negative interaction terms between condition and the highly

rewarded partner in the first two models demonstrate that the reward advantaged high

school partner is evaluated as having lower status and competence (b = -26.0, p <

.000; b = -24.5, p < .000, respectively), while the main effect for condition in these

models shows that the reward-disadvantaged graduate student has significantly greater

status and marginally higher competence (b = 12.9, p < .05; b = 7.7, p < .10,

respectively). The main effect of the partners’ reward level indicates that the highly

rewarded graduate student in the reward-consistent condition has significantly greater

evaluations of status and competence (b = 14.6, p < .01; b = 17.0, p < .000,

respectively).

Table 4.6. Mean Ratings of Partners’ Status and Competence by Condition

Condition

Variables Rewards Consistent with

Partners’ Status Characteristics

Rewards Inconsistent with Partners’ Status Characteristics

Mean T-Statistic Mean T-Statistic

Status Scale

High reward partner 71.5

(10.7) 3.0**

59.3

(13.2) -2.2*

Low reward partner 57.5

(14.8) 70.1

(15.2)

Competence Scale

High reward partner 76.3

(10.3) 3.8***

60.8

(9.0) -1.8+

Low reward partner 58.9

(15.4) 68.2

(14.0) Note: Sample size for this analysis = 41; SDs in parentheses; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 146: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

136

Table 4.7. Estimated Mixed-Effects Linear Regression Coefficients for the Effects of Partners’ Reward Level, and the Education Status Characteristic on the Status Value Scales

Variables Reward Level of Partner Education Status

Characteristic

Status Scale

Competence Scale

Status Scale

Competence Scale

Rewards inconsistent with partners’ status characteristics

12.9* (5.6)

7.7+ (4.6)

1.3 (4.7)

0.6 (4.4)

High reward partner 14.6** (4.3)

17.0*** (3.3)

Inconsistent x high reward partner

-26.0*** (6.0)

-24.5*** (4.6)

Graduate studentsa

32.6*** (3.5)

28.4*** (3.4)

Inconsistent x graduate students -1.6

(5.1) -0.5 (3.7)

Participant Controls: Female participant

-3.8 (4.7)

-3.4 (4.2)

1.5 (3.9)

-0.5 (3.7)

Age of participant

0.1 (0.4)

0.0 (0.4)

0.4 (0.3)

0.5 (0.3)

Minority participant 0.7 (4.9)

-6.1 (4.4)

8.3* (3.9)

-2.9 (3.9)

Online study 2.2 (4.4)

2.5 (4.0)

-2.2 (3.5)

0.8 (3.5)

Intercept 55.4*** (11.1)

64.5*** (10.1)

31.9*** (9.1)

43.5*** (9.0)

Random Effect Variance Termb: Intercept

8.3 (2.5)

8.6 (2.0)

5.9 (2.3)

6.2 (2.2)

Note: Models fit using restricted maximum likelihood (REML); SEs in parentheses; N = 41 participants; +p<.10; *p<.05; **p<.01; *** p<.001. a Comparison group is high school students. b Expressed as a standard deviation.

Page 147: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

137

In general, these data suggest that the reward may have an attenuated effect on

increasing the graduate students’ and decreasing the high school students’ status and

competence assessments when rewards are consistently assigned to the partners. The

rewards may also have had a negligible impact at reducing the difference between

these evaluations in the reward inconsistent condition, especially related to

competence assessments, but this effect is not strong enough to reverse the status and

competence evaluations.

The average assessment of the education status characteristic also follows this

pattern, though the differences in the reward-inconsistent condition are more distinct

(please see Table 8). In both conditions, the overall evaluation of graduate students’

status and competence are significantly higher than the evaluations of the high school

students’ (status: t = 7.8, p < .000; t = 7.6, p < .000; competence: t = 6.5, p < .000; t =

7.3, p < .000). Mixed-effect linear regression also indicates that the assessments of

this status characteristic are not altered by the consistency of the reward allocation

(please refer back to Table 7). The only significant predictor of these status appraisals

is the state of this education status characteristic (b = 32.6, p < .000; b = 28.4, p <

.000, respectively). Although rewards may have had a slight impact on altering the

participants’ evaluations of their partners, their inconsistent allocation did not have an

effect on altering their assessments of these two states of the educational status

characteristic.

Page 148: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

138

SUMMARY AND CONCLUSIONS

The continuation of status-based inequality is one of our most important and

pervasive social problems because this source of inequality is founded on essentially

illegitimate assumptions of competence and worth, which can cloud ability

assessments and alter opportunity structures and resource distributions. Indeed, one of

the reasons why this type of inequality is so pervasive is that those who control the

distribution of resources and rewards continually reallocate these advantages to

additional high status groups (Jacobs 1989; Kanter 1977; Padavic and Reskin 2002;

Reskin 1988; Williams 1992). When certain traits and characteristics become imbued

with status value, these states become bases on which esteem, prestige, honor,

resources, and influence are given, even implicitly.

Table 4.8. Mean Ratings of the Education Status Characteristic’s Status and Competence by Condition

Condition

Variables Rewards Consistent with

Partners’ Status Characteristics

Rewards Inconsistent with Partners’ Status Characteristics

Mean T-Statistic Mean T-Statistic

Status Scale

Graduate student 79.3

(11.0) 7.8***

79.8

(9.4) 7.6***

High school student 46.4

(13.7) 47.5

(12.5)

Competence Scale

Graduate student 80.8

(9.3) 6.5***

78.1

(8.5) 7.3***

High school student 51.9

(16.8) 52.3

(11.1) Note: Sample size for this analysis = 41; SDs in parentheses; +p < .10; *p < .05; **p < .01; ***p < .001 (two-tailed tests).

Page 149: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

139

Many have sought ways of intervening in this process, either through social

programs that aim to reduce such inequality by distributing resources and honor to

disadvantaged groups, or by introducing differential ability information in immediate

social interactions (Markovsky et al. 1984; Pugh and Wahrman 1983). This project

offers a preliminary step toward assessing the extent to which a redistribution of

socially valued rewards alters status-based inequality. Specifically, when those with

disadvantaged status characteristics hold valued symbols of status, such as certain

awards, occupations, and positions, does their status and influence increase by virtue

of their possession of these valued rewards?

The results of this study suggest that this is not the case, at least under the

particular conditions of this experiment’s one-shot context. When valued rewards are

affiliated with status-disadvantaged individuals, the rewards begin to lose their status

value and, thereby, do not alter the status hierarchy present in the group and the

general status assessments of the group members’ status characteristics more broadly,

as predicted by hypothesis 4. There is some evidence to suggest that the rewards have

an attenuated effect toward further differentiating the particular group members’

overall status and competence when the rewards are consistently distributed, but they

may only have at most a negligible impact on reducing the divergence in these

evaluations when they are inconsistently allocated. Nevertheless, the partners do not

appreciably gain or lose influence by virtue of their reward level, and, importantly, the

status and competence assessments of the group members’ salient status characteristic

do not vary by the type of reward distribution. In sum, differing reward levels are

generally unsuccessful at reducing the inequalities present in this study’s task groups.

Page 150: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

140

These are tentative conclusions, however, since there is reason to believe that

certain aspects of this study’s design may not have allowed the reward to have a strong

enough impact. It may be that this study’s reward does not have a robust enough

valuation to guard it against status contamination when it is inconsistently allocated. In

situations wherein the valuation of the status markers is uncertain or relatively new,

the results suggest that the these rewards are far more susceptible to having their

meaning redefined by those who possess them and may not help reduce inequality.

In this study, multiple status characteristics including age, gender, education

level, and grade point average, are used to create a reward through the spread of status

value process. This has the advantage of creating a reward that only takes its meaning

from how it is defined in the experiment as participants could not have experienced

the study’s color coding reward outside of the laboratory. Even though the reward is

created through a consistent link between the levels of these multiple characteristics, it

may be that this particular reward is too novel to not be susceptible to the particular

status background of the participants’ teammates. Additionally, affiliating the reward

with educational background and then distinguishing the partners by a similar

educational difference may have provided the conditions for a swift status

contamination in the reward-inconsistent condition. Future work could vary the type

of reward used to ascertain whether it is possible for rewards to a greater impact on

reducing status inequality. For example, a more exclusive reward, such as a

university-wide or national award, could be used in place of the color-coding scheme.

Another prominent concern with this study’s design is that the rewards are

associated with the partners’ clearly hierarchically divergent education levels. This

Page 151: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

141

may have overwhelmed the new rewards’ appraisal and any effect it may have had.

Education is also a keenly valued status characteristic that strongly suggests

differences in competence, ability, and development. Moreover, it is more legitimate

to assume ability differences derived from differing educational backgrounds than

based on ascribed characteristics like gender and race, and the status beliefs

surrounding ascribed characteristics tend to be more uncertain (Balkwell 1991).

Notably, these demographic groups are also generally the targets of social programs

that reallocate rewards. Future research could affiliate the color-coding reward with

the partners’ gender or racial background to ascertain whether rewards can reduce

inequalities resulting from their demographic differences.

Understanding how socially valued rewards can alter the relative status of

those who possess them is an important step towards finding additional mechanisms

that intervene in the processes of status-based inequality. Using reward allocations in

this manner is very complex, and the results of this study suggest that when rewards

are relatively nascent, they may not have an appreciable impact on status assessments

and behavioral inequalities. While the results of this study do not show support for this

particular reward intervention mechanism, further research may demonstrate that

rewards can indeed reduce status-based inequality when they are affiliated with

characteristics that are less powerfully linked with expectations of competence or

when stronger forms of rewards are used.

Page 152: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

142

References

Balkwell, James W. 1991. “Status Characteristics and Social Interaction: Assessment of Theoretical Variants.” Pp. 135-76 in Advances in Group Processes, vol. 8, edited by E. J. Lawler, B. Markovsky, C. Ridgeway, and H. A. Walker. Greenwich, CT: Jai Press. Bates, Douglas M. 2010. lme4: Mixed-Effects Modeling with R. Springer. Berger, Joseph, Bernard P. Cohen, and Morris Zelditch Jr. 1966. “Status

Characteristics and Expectation States.” Pp. 29-46 in Sociological Theories in Progress, vol. 1, edited by J. Berger, M. Zelditch Jr., and B. Anderson. Boston: Houghton Mifflin.

------. 1972. “Status Characteristics and Social Interaction.” American Sociological

Review 37:241-55. Berger, Joseph and Thomas L. Conner. 1974. "Performance Expectations and

Behavior in Small Groups: A Revised Formulation." Pp. 85-109 in Expectation States Theory: A Theoretical Research Program, edited by J. Berger, T. L. Conner, and M. H. Fisek. Cambridge, MA: Winthrop.

Berger, Joseph and M. Hamit Fisek. 2006. “Diffuse Status Characteristics and The Spread of Status Value: A Formal Theory.” American Journal of Sociology 111:1038-79. Berger, Joseph, M. Hamit Fisek, Robert Z. Norman, and David Wagner. 1983. "The Formation of Reward Expectations in Status Situations." Pp. 127-68 in Equity

Theory, edited by D. M. Messick and K. S. Cook. New York: Praeger. Berger, Joseph, M. Hamit Fisek, Robert Z. Norman, and Morris Zelditch Jr. 1977.

Status Characteristics and Social Interaction: An Expectation States Approach. New York: Elsevier.

Berger, Joseph, Robert Z. Norman, James W. Balkwell, and Roy F. Smith. 1992.

“Status Inconsistency in Task Situations: A Test of Four Status Processing Principles.” American Sociological Review 57: 843-55.

Berger, Joseph, Morris Zelditch Jr., Bo Anderson, and Bernard P. Cohen. 1972.

"Structural Aspects of Distributive Justice: A Status Value Formulation." Pp. 119-46 in Sociological Theories in Progress, vol. 2, edited by J. Berger, M. Zelditch Jr., and B. Anderson. Boston: Houghton Mifflin.

Catanzarite, Lisa. 2003. “Race-Gender Composition and Occupational Pay

Degradation.” Social Problems 50:14-37.

Page 153: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

143

Cohn, Samuel. 1985. The Process of Occupational Sex-Typing: The Feminization of Clerical Labor in Great Britain. Philadelphia: Temple University Press. Cook, Karen S. 1975. "Expectations, Evaluations, and Equity." American Sociological Review 40:372-88. Correll, Shelley J., Stephen Benard, and In Paik. 2007. “Getting a Job: Is There a

Motherhood Penalty?” American Journal of Sociology 112:1297-338. England, Paula, Paul Allison, and Yuxiao Wu. 2007. “Does Bad Pay Cause

Occupations to Feminize, Does Feminization Reduce Pay, and How Can We Tell with Longitudinal Data?” Social Science Research 36:1237-56.

Foschi, Martha, Larissa Lai, and Kirsten Sigerson. 1994. “Gender and Double Standards in the Assessment of Job Applicants.” Social Psychology Quarterly 57:326-39. Foschi, Martha, Kirsten Sigerson, and Marie Lembesis. 1995. “Assessing Job Applicants: The Relative Effects of Gender, Academic Record, and Decision Type.” Small Group Research 26:328-52. Foschi, Martha and Jerilee Valenzuela. 2007. “Selecting Job Applicants: Effects from

Gender, Self-Presentation, and Decision Type.” Social Science Research 37:1022-38.

Harrod, Wendy J. 1980. “Expectations from Unequal Rewards.” Social Psychology

Quarterly 43:126-30. Healy, Patrick. 2010. “A Do-Over for the Wasserstein Playwriting Prize.” The New

York Times. November 15. Jiang, Jiming. 2007. Linear and Generalized Linear Mixed Models and Their

Applications. New York: Springer. Markovsky, Barry, Le Roy F. Smith, and Joseph Berger. 1984. “Do Status Interventions Persist?” American Sociological Review 49:373-82. Pugh, Meredith D. and Ralph Wahrman. 1983. “Neutralizing Sexism in Mixed-Sex

Groups: Do Women Have to Be Better than Men?” American Journal of Sociology 88:746-62.

Ridgeway, Cecilia L., Kristen Backor, Yan E. Li, Justine E. Tinkler, and Kristan G.

Erickson. 2009. “How Easily Does a Social Difference Become a Status Distinction? Gender Matters.” American Sociological Review 74:44-62.

Page 154: CONSEQUENCES OF REWARDS: A DISSERTATION …zx554nf4443/...this dissertation is theirs, and I am incredibly lucky to have received their guidance on this project. I am also grateful

144

Ridgeway, Cecilia L., Elizabeth Heger Boyle, Kathy J. Kuipers, and Dawn T.

Robinson. 1998. "How Do Status Beliefs Develop? The Role of Resources and Interactional Experience." American Sociological Review 63:331-50.

Ridgeway, Cecilia L. and Kristan Glasgow Erickson. 2000. "Creating and Spreading Status Beliefs." American Journal of Sociology 106:579-615. Stewart, Penni A. and James C. Moore, Jr. 1992. "Wage Disparities and Performance

Expectations." Social Psychology Quarterly 55:70-85. Strober, Myra and Carolyn Arnold. 1987. “The Dynamics of Occupational

Segregation Among Bank Tellers.” Pp. 107-48 in Gender in the Workplace, edited by C. Brown and J. Pechman. Washington, DC: Brookings Institute.

Thye, Shane R. 2000. "A Status Value Theory of Power in Exchange Relations."

American Sociological Review 65:407-32. Veblen, Thorstein. [1899] 2005. The Theory of the Leisure Class. Gloucestershire, UK: Dodo Press. Webster, Murray Jr. and James E. Driskell Jr. 1983. “Beauty as Status.” American Journal of Sociology 89:140-65.