design science in information systems research - Temple Fox MIS

Hevner et al./Design Science in IS Research

MIS Quarterly Vol. 28 No. 1, pp. 75-105/March 2004 75

RESEARCH ESSAY

DESIGN SCIENCE IN INFORMATIONSYSTEMS RESEARCH1

By: Alan R. HevnerInformation Systems and Decision

SciencesCollege of Business AdministrationUniversity of South FloridaTampa, FL [email protected]

Salvatore T. MarchOwn Graduate School of ManagementVanderbilt UniversityNashville, TN [email protected]

Jinsoo ParkCollege of Business AdministrationKorea UniversitySeoul, [email protected]

1Allen S. Lee was the accepting senior editor for thispaper.

Sudha RamManagement Information SystemsEller College of Business and Public

AdministrationUniversity of ArizonaTucson, AZ [email protected]

Abstract

Two paradigms characterize much of the researchin the Information Systems discipline: behavioralscience and design science. The behavioral-science paradigm seeks to develop and verifytheories that explain or predict human or organi-zational behavior. The design-science paradigmseeks to extend the boundaries of human andorganizational capabilities by creating new andinnovative artifacts. Both paradigms are founda-tional to the IS discipline, positioned as it is at theconfluence of people, organizations, and techno-logy. Our objective is to describe the performanceof design-science research in Information Sys-tems via a concise conceptual framework andclear guidelines for understanding, executing, andevaluating the research. In the design-scienceparadigm, knowledge and understanding of aproblem domain and its solution are achieved inthe building and application of the designed arti-fact. Three recent exemplars in the researchliterature are used to demonstrate the application


76 MIS Quarterly Vol. 28 No. 1/March 2004

of these guidelines. We conclude with an analysisof the challenges of performing high-qualitydesign-science research in the context of thebroader IS community.

Keywords: Information Systems research meth-odologies, design science, design artifact, busi-ness environment, technology infrastructure,search strategies, experimental methods,creativity

Introduction

Information systems are implemented within anorganization for the purpose of improving theeffectiveness and efficiency of that organization.Capabilities of the information system and char-acteristics of the organization, its work systems,its people, and its development and implemen-tation methodologies together determine theextent to which that purpose is achieved (Silver etal. 1995). It is incumbent upon researchers in theInformation Systems (IS) discipline to �furtherknowledge that aids in the productive applicationof information technology to human organizationsand their management� (ISR 2002, inside frontcover) and to develop and communicate �knowl-edge concerning both the management ofinformation technology and the use of informationtechnology for managerial and organizational pur-poses� (Zmud 1997).

We argue that acquiring such knowledge involvestwo complementary but distinct paradigms,behavioral science and design science (Marchand Smith 1995). The behavioral-science para-digm has its roots in natural science researchmethods. It seeks to develop and justify theories(i.e., principles and laws) that explain or predictorganizational and human phenomena sur-rounding the analysis, design, implementation,management, and use of information systems.Such theories ultimately inform researchers andpractitioners of the interactions among people,technology, and organizations that must bemanaged if an information system is to achieve itsstated purpose, namely improving the effective-

ness and efficiency of an organization. Thesetheories impact and are impacted by designdecisions made with respect to the systemdevelopment methodology used and the functionalcapabilities, information contents, and humaninterfaces implemented within the informationsystem.

The design-science paradigm has its roots inengineering and the sciences of the artificial(Simon 1996). It is fundamentally a problem-solving paradigm. It seeks to create innovationsthat define the ideas, practices, technical capa-bilities, and products through which the analysis,design, implementation, management, and use ofinformation systems can be effectively andefficiently accomplished (Denning 1997;Tsichritzis 1998). Such artifacts are not exemptfrom natural laws or behavioral theories. To thecontrary, their creation relies on existing kerneltheories that are applied, tested, modified, andextended through the experience, creativity,intuition, and problem solving capabilities of theresearcher (Markus et al. 2002; Walls et al. 1992).

The importance of design is well recognized in theIS literature (Glass 1999; Winograd 1996, 1998).Benbasat and Zmud (1999, p. 5) argue that therelevance of IS research is directly related to itsapplicability in design, stating that the implicationsof empirical IS research should be �implemen-table,�synthesize an existing body of research,�[or] stimulate critical thinking� among IS practi-tioners. However, designing useful artifacts iscomplex due to the need for creative advances indomain areas in which existing theory is ofteninsufficient. �As technical knowledge grows, IT isapplied to new application areas that were notpreviously believed to be amenable to IT support�(Markus et al. 2002, p. 180). The resultant ITartifacts extend the boundaries of human problemsolving and organizational capabilities by pro-viding intellectual as well as computational tools.Theories regarding their application and impactwill follow their development and use.

Here, we argue, is an opportunity for IS researchto make significant contributions by engaging thecomplementary research cycle between design-


MIS Quarterly Vol. 28 No. 1, pp. /March 2004 77

science and behavioral-science to address funda-mental problems faced in the productive applica-tion of information technology. Technology andbehavior are not dichotomous in an informationsystem. They are inseparable (Lee 2000). Theyare similarly inseparable in IS research. Philo-sophically these arguments draw from the prag-matists (Aboulafia 1991) who argue that truth(justified theory) and utility (artifacts that areeffective) are two sides of the same coin and thatscientific research should be evaluated in light ofits practical implications.

The realm of IS research is at the confluence ofpeople, organizations, and technology (Davis andOlson 1985; Lee 1999). IT artifacts are broadlydefined as constructs (vocabulary and symbols),models (abstractions and representations),methods (algorithms and practices), and instan-tiations (implemented and prototype systems).These are concrete prescriptions that enable ITresearchers and practitioners to understand andaddress the problems inherent in developing andsuccessfully implementing information systemswithin organizations (March and Smith 1995;Nunamaker et al. 1991a). As illustrations, Markuset al. (2002) and Walls et al. (1992) presentdesign-science research aimed at developingexecutive information systems (EISs) and systemsto support emerging knowledge processes(EKPs), respectively, within the context of �ISdesign theories.� Such theories prescribe �effec-tive development practices� (methods) and �a typeof system solution� (instantiation) for �a particularclass of user requirements� (models) (Markus etal. 2002, p. 180). Such prescriptive theories mustbe evaluated with respect to the utility provided forthe class of problems addressed.

An IT artifact, implemented in an organizationalcontext, is often the object of study in IS behav-ioral-science research. Theories seek to predictor explain phenomena that occur with respect tothe artifact�s use (intention to use), perceivedusefulness, and impact on individuals and organi-zations (net benefits) depending on system,service, and information quality (DeLone andMcLean 1992, 2003; Seddon 1997). Much of thisbehavioral research has focused on one class of

artifact, the instantiation (system), although otherresearch efforts have also focused on theevaluation of constructs (e.g., Batra et al. 1990;Bodart et al. 2001; Geerts and McCarthy 2002;Kim and March 1995) and methods (e.g., Marakasand Elam 1998; Sinha and Vessey 1999).Relatively little behavioral research has focusedon evaluating models, a major focus of researchin the management science literature.

Design science, as the other side of the ISresearch cycle, creates and evaluates IT artifactsintended to solve identified organizational prob-lems. Such artifacts are represented in a struc-tured form that may vary from software, formallogic, and rigorous mathematics to informalnatural language descriptions. A mathematicalbasis for design allows many types of quantitativeevaluations of an IT artifact, including optimizationproofs, analytical simulation, and quantitativecomparisons with alternative designs. The furtherevaluation of a new artifact in a given organi-zational context affords the opportunity to applyempirical and qualitative methods. The richphenomena that emerge from the interaction ofpeople, organizations, and technology may needto be qualitatively assessed to yield an under-standing of the phenomena adequate for theorydevelopment or problem solving (Klein andMeyers 1999). As field studies enable behavioral-science researchers to understand organizationalphenomena in context, the process of constructingand exercising innovative IT artifacts enabledesign-science researchers to understand theproblem addressed by the artifact and thefeasibility of their approach to its solution(Nunamaker et al. 1991a).

The primary goal of this paper is to inform thecommunity of IS researchers and practitioners ofhow to conduct, evaluate, and present design-science research. We do so by describing theboundaries of design science within the ISdiscipline via a conceptual framework for under-standing information systems research and bydeveloping a set of guidelines for conducting andevaluating good design-science research. Wefocus primarily on technology-based designalthough we note with interest the current explora-



tion of organizations, policies, and work practicesas designed artifacts (Boland 2002). FollowingKlein and Myers (1999) treatise on the conductand evaluation of interpretive research in IS, weuse the proposed guidelines to assess recentexemplar papers published in the IS literature inorder to illustrate how authors, reviewers, andeditors can apply them consistently. We concludewith an analysis of the challenges of performinghigh-quality design-science research and a call forsynergistic efforts between behavioral-scienceand design-science researchers.

A Framework for IS Research

Information systems and the organizations theysupport are complex, artificial, and purposefullydesigned. They are composed of people, struc-tures, technologies, and work systems (Alter2003; Bunge 1985; Simon 1996). Much of thework performed by IS practitioners, and managersin general (Boland 2002), deals with design�thepurposeful organization of resources to accom-plish a goal. Figure 1 illustrates the essentialalignments between business and informationtechnology strategies and between organizationaland information systems infrastructures (Hender-son and Venkatraman 1993). The effective transi-tion of strategy into infrastructure requires exten-sive design activity on both sides of the figure�organizational design to create an effectiveorganizational infrastructure and informationsystems design to create an effective informationsystem infrastructure.

These are interdependent design activities thatare central to the IS discipline. Hence, IS researchmust address the interplay among businessstrategy, IT strategy, organizational infrastructure,and IS infrastructure. This interplay is becomingmore crucial as information technologies are seenas enablers of business strategy and organiza-tional infrastructure (Kalakota and Robinson 2001;Orlikowski and Barley 2001). Available andemerging IT capabilities are a significant factor indetermining the strategies that guide an organiza-tion. Cutting-edge information systems allow

organizations to engage new forms and newstructures�to change the ways they �do busi-ness� (Drucker 1988, 1991; Orlikowski 2000). Oursubsequent discussion of design science will belimited to the activities of building the IS infrastruc-ture within the business organization. Issues ofstrategy, alignment, and organizational infrastruc-ture design are outside the scope of this paper.

To achieve a true understanding of and appre-ciation for design science as an IS researchparadigm, an important dichotomy must be faced.Design is both a process (set of activities) and aproduct (artifact)�a verb and a noun (Walls et al.1992). It describes the world as acted upon (pro-cesses) and the world as sensed (artifacts). ThisPlatonic view of design supports a problem-solving paradigm that continuously shifts perspec-tive between design processes and designedartifacts for the same complex problem. Thedesign process is a sequence of expert activitiesthat produces an innovative product (i.e., thedesign artifact). The evaluation of the artifact thenprovides feedback information and a betterunderstanding of the problem in order to improveboth the quality of the product and the designprocess. This build-and-evaluate loop is typicallyiterated a number of times before the final designartifact is generated (Markus et al. 2002). Duringthis creative process, the design-science re-searcher must be cognizant of evolving both thedesign process and the design artifact as part ofthe research.

March and Smith (1995) identify two designprocesses and four design artifacts produced bydesign-science research in IS. The two processesare build and evaluate. The artifacts are con-structs, models, methods, and instantiations.Purposeful artifacts are built to address heretoforeunsolved problems. They are evaluated withrespect to the utility provided in solving thoseproblems. Constructs provide the language inwhich problems and solutions are defined andcommunicated (Schön 1983). Models use con-structs to represent a real world situation�thedesign problem and its solution space (Simon1996). Models aid problem and solution under-standing and frequently represent the connection



Business Strategy

Information Technology

Strategy

Organizational Infrastructure

Information Systems

Infrastructure

Strategy Alignment

Infrastructure Alignment

Organizational Design

Activities

Information Systems Design

Activities

Business Strategy

Information Technology

Strategy

Organizational Infrastructure

Information Systems

Infrastructure

Strategy Alignment

Infrastructure Alignment

Organizational Design

Activities

Information Systems Design

Activities

Figure 1. Organizational Design and Information Systems Design Activities(Adapted from J. Henderson and N. Venkatraman, �Strategic Alignment: LeveragingInformation Technology for Transforming Organizations,� IBM Systems Journal(32:1), 1993.)

between problem and solution componentsenabling exploration of the effects of designdecisions and changes in the real world. Methodsdefine processes. They provide guidance on howto solve problems, that is, how to search thesolution space. These can range from formal,mathematical algorithms that explicitly define thesearch process to informal, textual descriptions of�best practice� approaches, or some combination.Instantiations show that constructs, models, ormethods can be implemented in a working sys-tem. They demonstrate feasibility, enabling con-crete assessment of an artifact�s suitability to itsintended purpose. They also enable researchersto learn about the real world, how the artifactaffects it, and how users appropriate it.

Figure 2 presents our conceptual framework forunderstanding, executing, and evaluating ISresearch combining behavioral-science anddesign-science paradigms. We use this frame-work to position and compare these paradigms.

The environment defines the problem space(Simon 1996) in which reside the phenomena ofinterest. For IS research, it is composed of

people, (business) organizations, and theirexisting or planned technologies (Silver et al.1995). In it are the goals, tasks, problems, andopportunities that define business needs as theyare perceived by people within the organization.Such perceptions are shaped by the roles,capabilities, and characteristics of people withinthe organization. Business needs are assessedand evaluated within the context of organizationalstrategies, structure, culture, and existing busi-ness processes. They are positioned relative toexisting technology infrastructure, applications,communication architectures, and developmentcapabilities. Together these define the businessneed or �problem� as perceived by the researcher.Framing research activities to address businessneeds assures research relevance.

Given such an articulated business need, ISresearch is conducted in two complementaryphases. Behavioral science addresses researchthrough the development and justification oftheories that explain or predict phenomena relatedto the identified business need. Design scienceaddresses research through the building andevaluation of artifacts designed to meet the iden-



Additions to the Knowledge Base

Environment IS Research Knowledge Base

People�Roles�Capabilities�Characteristics

Organizations�Strategies�Structure & Culture�Processes

Technology�Infrastructure�Applications�Communications Architecture�Development Capabilities

Foundations�Theories�Frameworks�Instruments�Constructs�Models�Methods�Instantiations

Methodologies�Data Analysis Techniques�Formalisms�Measures�Validation Criteria

Develop/Build�Theories�Artifacts

Justify/Evaluate�Analytical�Case Study�Experimental�Field Study �Simulation

Assess Refine

Business Needs

Applicable Knowledge

Application in the Appropriate Environment

Relevance Rigor

Additions to the Knowledge Base

Environment IS Research Knowledge Base

People�Roles�Capabilities�Characteristics

Organizations�Strategies�Structure & Culture�Processes

Technology�Infrastructure�Applications�Communications Architecture�Development Capabilities

Foundations�Theories�Frameworks�Instruments�Constructs�Models�Methods�Instantiations

Methodologies�Data Analysis Techniques�Formalisms�Measures�Validation Criteria

Develop/Build�Theories�Artifacts

Justify/Evaluate�Analytical�Case Study�Experimental�Field Study �Simulation

Assess Refine

Business Needs

Applicable Knowledge

Application in the Appropriate Environment

Relevance Rigor

Figure 2. Information Systems Research Framework

tified business need. The goal of behavioral-science research is truth.2 The goal of design-science research is utility. As argued above, ourposition is that truth and utility are inseparable.Truth informs design and utility informs theory. Anartifact may have utility because of some as yetundiscovered truth. A theory may yet to be devel-oped to the point where its truth can be incorpor-ated into design. In both cases, research assess-ment via the justify/evaluate activities can result inthe identification of weaknesses in the theory or

artifact and the need to refine and reassess. Therefinement and reassessment process is typicallydescribed in future research directions.

The knowledge base provides the raw materialsfrom and through which IS research is accom-plished. The knowledge base is composed offoundations and methodologies. Prior IS researchand results from reference disciplines providefoundational theories, frameworks, instruments,constructs, models, methods, and instantiationsused in the develop/build phase of a researchstudy. Methodologies provide guidelines used inthe justify/evaluate phase. Rigor is achieved byappropriately applying existing foundations andmethodologies. In behavioral science, methodol-ogies are typically rooted in data collection andempirical analysis techniques. In design science,computational and mathematical methods are

2Theories posed in behavioral science are principledexplanations of phenomena. We recognize that suchtheories are approximations and are subject to numer-ous assumptions and conditions. However, they areevaluated against the norms of truth or explanatorypower and are valued only as the claims they make areborne out in reality.



primarily used to evaluate the quality and effec-tiveness of artifacts; however, empirical techni-ques may also be employed.

The contributions of behavioral science anddesign science in IS research are assessed asthey are applied to the business need in anappropriate environment and as they add to thecontent of the knowledge base for further researchand practice. A justified theory that is not usefulfor the environment contributes as little to the ISliterature as an artifact that solves a nonexistentproblem.

One issue that must be addressed in design-science research is differentiating routine designor system building from design research. Thedifference is in the nature of the problems andsolutions. Routine design is the application ofexisting knowledge to organizational problems,such as constructing a financial or marketinginformation system using best practice artifacts(constructs, models, methods, and instantiations)existing in the knowledge base. On the otherhand, design-science research addresses impor-tant unsolved problems in unique or innovativeways or solved problems in more effective orefficient ways. The key differentiator between rou-tine design and design research is the clear iden-tification of a contribution to the archival knowl-edge base of foundations and methodologies.

In the early stages of a discipline or with signifi-cant changes in the environment, each newartifact created for that discipline or environmentis �an experiment� that �poses a question tonature� (Newell and Simon 1976, p 114). Existingknowledge is used where appropriate; however,often the requisite knowledge is nonexistent(Markus et al. 2002). Reliance on creativity andtrial-and-error search are characteristic of suchresearch efforts. As design-science researchresults are codified in the knowledge base, theybecome best practice. System building is then theroutine application of the knowledge base toknown problems.

Design activities are endemic in many profes-sions. In particular, the engineering profession

has produced a considerable literature on design(Dym 1994; Pahl and Beitz 1996; Petroski 1996).Within the IS discipline, many design activitieshave been extensively studied, formalized, andbecome normal or routine. Design-scienceresearch in IS addresses what are considered tobe wicked problems (Brooks 1987, 1996; Ritteland Webber 1984). That is, those problemscharacterized by

� unstable requirements and constraints basedupon ill-defined environmental contexts

� complex interactions among subcomponentsof the problem and its solution

� inherent flexibility to change design pro-cesses as well as design artifacts (i.e.,malleable processes and artifacts)

� a critical dependence upon human cognitiveabilities (e.g., creativity) to produce effectivesolutions

� a critical dependence upon human socialabilities (e.g., teamwork) to produce effectivesolutions

As a result, we agree with Simon (1996) that atheory of design in information systems, ofnecessity, is in a constant state of scientificrevolution (Kuhn 1996). Technological advancesare the result of innovative, creative designscience processes. If not capricious, they are atleast arbitrary (Brooks 1987) with respect tobusiness needs and existing knowledge.Innovations, such as database management sys-tems, high-level languages, personal computers,software components, intelligent agents, objecttechnology, the Internet, and the World WideWeb, have had dramatic and at times unintendedimpacts on the way in which information systemsare conceived, designed, implemented, andmanaged. Consequently the guidelines wepresent below are, of necessity, adaptive andprocess-oriented.



Guidelines for Design Sciencein Information SystemsResearch

As discussed above, design science is inherentlya problem solving process. The fundamentalprinciple of design-science research from whichour seven guidelines are derived is that knowl-edge and understanding of a design problem andits solution are acquired in the building andapplication of an artifact. That is, design-scienceresearch requires the creation of an innovative,purposeful artifact (Guideline 1) for a specifiedproblem domain (Guideline 2). Because theartifact is purposeful, it must yield utility for thespecified problem. Hence, thorough evaluation ofthe artifact is crucial (Guideline 3). Novelty issimilarly crucial since the artifact must beinnovative, solving a heretofore unsolved problemor solving a known problem in a more effective orefficient manner (Guideline 4). In this way,design-science research is differentiated from thepractice of design. The artifact itself must berigorously defined, formally represented, coherent,and internally consistent (Guideline 5). The pro-cess by which it is created, and often the artifactitself, incorporates or enables a search processwhereby a problem space is constructed and amechanism posed or enacted to find an effectivesolution (Guideline 6). Finally, the results of thedesign-science research must be communicatedeffectively (Guideline 7) both to a technicalaudience (researchers who will extend them andpractitioners who will implement them) and to amanagerial audience (researchers who will studythem in context and practitioners who will decideif they should be implemented within theirorganizations).

Our purpose for establishing these sevenguidelines is to assist researchers, reviewers,editors, and readers to understand the require-ments for effective design-science research.Following Klein and Myers (1999), we adviseagainst mandatory or rote use of the guidelines.Researchers, reviewers, and editors must usetheir creative skills and judgment to determinewhen, where, and how to apply each of the guide-lines in a specific research project. However, we

contend that each of these guidelines should beaddressed in some manner for design-scienceresearch to be complete. How well the researchsatisfies the intent of each of the guidelines isthen a matter for the reviewers, editors, andreaders to determine.

Table 1 summarizes the seven guidelines. Eachis discussed in detail below. In the followingsection, they are applied to specific exemplarresearch efforts.

Guideline 1: Design as an Artifact

The result of design-science research in IS is, bydefinition, a purposeful IT artifact created to ad-dress an important organizational problem. Itmust be described effectively, enabling its imple-mentation and application in an appropriatedomain.

Orlikowski and Iacono (2001) call the IT artifactthe �core subject matter� of the IS field. Althoughthey articulate multiple definitions of the term ITartifact, many of which include components of theorganization and people involved in the use of acomputer-based artifact, they emphasize theimportance of �those bundles of cultural propertiespackaged in some socially recognizable form suchas hardware and software� (p. 121), i.e., the ITartifact as an instantiation. Weber (1987) arguesthat theories of long-lived artifacts (instantiations)and their representations (Weber 2003) arefundamental to the IS discipline. Such theoriesmust explain how artifacts are created andadapted to their changing environments andunderlying technologies.

Our definition of IT artifacts is both broader andnarrower then those articulated above. It isbroader in the sense that we include not onlyinstantiations in our definition of the IT artifact butalso the constructs, models, and methods appliedin the development and use of informationsystems. However, it is narrower in the sense thatwe do not include people or elements of organi-zations in our definition nor do we explicitlyinclude the process by which such artifacts evolve



Table 1. Design-Science Research GuidelinesGuideline Description

Guideline 1: Design as an Artifact Design-science research must produce a viable artifact in theform of a construct, a model, a method, or an instantiation.

Guideline 2: Problem Relevance The objective of design-science research is to developtechnology-based solutions to important and relevantbusiness problems.

Guideline 3: Design Evaluation The utility, quality, and efficacy of a design artifact must berigorously demonstrated via well-executed evaluationmethods.

Guideline 4: Research Contributions Effective design-science research must provide clear andverifiable contributions in the areas of the design artifact,design foundations, and/or design methodologies.

Guideline 5: Research Rigor Design-science research relies upon the application ofrigorous methods in both the construction and evaluation ofthe design artifact.

Guideline 6: Design as a SearchProcess

The search for an effective artifact requires utilizing availablemeans to reach desired ends while satisfying laws in theproblem environment.

Guideline 7: Communication ofResearch

Design-science research must be presented effectively bothto technology-oriented as well as management-orientedaudiences.

over time. We conceive of IT artifacts not asindependent of people or the organizational andsocial contexts in which they are used but asinterdependent and coequal with them in meetingbusiness needs. We acknowledge that percep-tions and fit with an organization are crucial to thesuccessful development and implementation of aninformation system. We argue, however, that thecapabilities of the constructs, models, methods,and instantiations are equally crucial and thatdesign-science research efforts are necessary fortheir creation.

Furthermore, artifacts constructed in design-science research are rarely full-grown informationsystems that are used in practice. Instead, artif-acts are innovations that define the ideas,practices, technical capabilities, and productsthrough which the analysis, design, implemen-tation, and use of information systems can beeffectively and efficiently accomplished (Denning

1997; Tsichritzis 1998). This definition of theartifact is consistent with the concept of IS designtheory as used by Walls et al. (1992) and Markuset al. (2002) where the theory addresses both theprocess of design and the designed product.

More precisely, constructs provide the vocabularyand symbols used to define problems andsolutions. They have a significant impact on theway in which tasks and problems are conceived(Boland 2002; Schön 1983). They enable theconstruction of models or representations of theproblem domain. Representation has a profoundimpact on design work. The field of mathematicswas revolutionized, for example, with the con-structs defined by Arabic numbers, zero, andplace notation. The search for an effective prob-lem representation is crucial to finding an effectivedesign solution (Weber 2003). Simon (1996, p.132) states, �solving a problem simply meansrepresenting it so as to make the solutiontransparent.�



The entity-relationship model (Chen 1976), forexample, is a set of constructs for representingthe semantics of data. It has had a profoundimpact on the way in which systems analysis anddatabase design are executed and the way inwhich information systems are represented anddeveloped. Furthermore, these constructs havebeen used to build models of specific businesssituations that have been generalized into patternsfor application in similar domains (Purao et al.2003). Methods for building such models havealso been the subject of considerable research(Halpin 2001; McCarthy 1982; Parsons and Wand2000; Storey et al. 1997).

Artifact instantiation demonstrates feasibility bothof the design process and of the designed pro-duct. Design-science research in IT often ad-dresses problems related to some aspect of thedesign of an information system. Hence, theinstantiations produced may be in the form ofintellectual or software tools aimed at improvingthe process of information system development.Constructing a system instantiation that auto-mates a process demonstrates that the processcan, in fact, be automated. It provides �proof byconstruction� (Nunamaker 1991a). The criticalnature of design-science research in IS lies in theidentification of as yet undeveloped capabilitiesneeded to expand IS into new realms �notpreviously believed amenable to IT support�(Markus et al. 2002, p. 180). Such a result issignificant IS research only if there is a seriousquestion about the ability to construct such anartifact, there is uncertainty about its ability toperform appropriately, and the automated task isimportant to the IS community. TOP Modeler(Markus et al. 2002), for example, is a tool thatinstantiates methods for the development ofinformation systems that support �emergentknowledge processes.� Construction of such aprototype artifact in a research setting or in asingle organizational setting is only a first steptoward its deployment, but we argue that it is anecessary one. As an exemplar of design-scienceresearch (see below), this research resulted in acommercial product that �has been used in overtwo dozen �real use� situations� (p. 187).

To illustrate further, prior to the construction of thefirst expert system (instantiation), it was not clearif such a system could be constructed. It was notclear how to describe or represent it, or how wellit would perform. Once feasibility was demon-strated by constructing an expert system in aselected domain, constructs and models weredeveloped and subsequent research in expertsystems focused on demonstrating significantimprovements in the product or process (methods)of construction (Tam 1990; Trice and Davis 1993).Similar examples exist in requirements determi-nation (Bell 1993; Bhargava et al. 1998), individualand group decision support systems (Aiken et al.1991; Basu and Blanning 1994), database designand integration (Dey et al. 1998; Dey et al. 1999;Storey et al. 1997), and workflow analysis (Basuand Blanning 2000), to name a few importantareas of IS design-science research.

Guideline 2: Problem Relevance

The objective of research in information systemsis to acquire knowledge and understanding thatenable the development and implementation oftechnology-based solutions to heretofore unsolvedand important business problems. Behavioralscience approaches this goal through the devel-opment and justification of theories explaining orpredicting phenomena that occur. Design scienceapproaches this goal through the construction ofinnovative artifacts aimed at changing the pheno-mena that occur. Each must inform and challengethe other. For example, the technology accep-tance model provides a theory that explains andpredicts the acceptance of information techno-logies within organizations (Venkatesh 2000).This theory challenges design-science re-searchers to create artifacts that enable organi-zations to overcome the acceptance problemspredicted. We argue that a combination oftechnology-based artifacts (e.g., system concep-tualizations and representations, practices, tech-nical capabilities, interfaces, etc.), organization-based artifacts (e.g., structures, compensation,reporting relationships, social systems, etc.), andpeople-based artifacts (e.g., training, consensusbuilding, etc.) are necessary to address suchissues.



Formally, a problem can be defined as thedifferences between a goal state and the currentstate of a system. Problem solving can be definedas a search process (see Guideline 6) usingactions to reduce or eliminate the differences(Simon 1996). These definitions imply an environ-ment that imposes goal criteria as well asconstraints upon a system. Business organiza-tions are goal-oriented entities existing in aneconomic and social setting. Economic theoryoften portrays the goals of business organizationsas being related to profit (utility) maximization.Hence, business problems and opportunities oftenrelate to increasing revenue or decreasing costthrough the design of effective business pro-cesses. The design of organizational and inter-organizational information systems plays a majorrole in enabling effective business processes toachieve these goals.

The relevance of any design-science researcheffort is with respect to a constituent community.For IS researchers, that constituent community isthe practitioners who plan, manage, design,implement, operate, and evaluate informationsystems and those who plan, manage, design,implement, operate, and evaluate the tech-nologies that enable their development andimplementation. To be relevant to this community,research must address the problems faced andthe opportunities afforded by the interaction ofpeople, organizations, and information technology.Organizations spend billions of dollars annually onIT, only too often to conclude that those dollarswere wasted (Keil 1995; Keil et al. 1998; Keil andRobey 1999). This community would welcomeeffective artifacts that enable such problems to beaddressed�constructs by which to think aboutthem, models by which to represent and explorethem, methods by which to analyze or optimizethem, and instantiations that demonstrate how toaffect them.

Guideline 3: Design Evaluation

The utility, quality, and efficacy of a design artifactmust be rigorously demonstrated via well-executed evaluation methods. Evaluation is a

crucial component of the research process. Thebusiness environment establishes the require-ments upon which the evaluation of the artifact isbased. This environment includes the technicalinfrastructure which itself is incrementally built bythe implementation of new IT artifacts. Thus,evaluation includes the integration of the artifactwithin the technical infrastructure of the businessenvironment.

As in the justification of a behavioral sciencetheory, evaluation of a designed IT artifactrequires the definition of appropriate metrics andpossibly the gathering and analysis of appropriatedata. IT artifacts can be evaluated in terms offunctionality, completeness, consistency, accu-racy, performance, reliability, usability, fit with theorganization, and other relevant quality attributes.When analytical metrics are appropriate, designedartifacts may be mathematically evaluated. Astwo examples, distributed database design algo-rithms can be evaluated using expected operatingcost or average response time for a givencharacterization of information processing require-ments (Johansson et al. 2003) and searchalgorithms can be evaluated using informationretrieval metrics such as precision and recall(Salton 1988).

Because design is inherently an iterative andincremental activity, the evaluation phase providesessential feedback to the construction phase as tothe quality of the design process and the designproduct under development. A design artifact iscomplete and effective when it satisfies therequirements and constraints of the problem itwas meant to solve. Design-science researchefforts may begin with simplified conceptuali-zations and representations of problems. Asavailable technology or organizational environ-ments change, assumptions made in priorresearch may become invalid. Johansson (2000),for example, demonstrated that network latency isa major component in the response-time perfor-mance of distributed databases. Prior research indistributed database design ignored latencybecause it assumed a low-bandwidth networkwhere latency is negligible. In a high-bandwidthnetwork, however, latency can account for over 90



Table 2. Design Evaluation Methods1. Observational Case Study: Study artifact in depth in business environment

Field Study: Monitor use of artifact in multiple projects

2. Analytical Static Analysis: Examine structure of artifact for static qualities (e.g.,complexity)

Architecture Analysis: Study fit of artifact into technical IS architecture

Optimization: Demonstrate inherent optimal properties of artifact or provideoptimality bounds on artifact behavior

Dynamic Analysis: Study artifact in use for dynamic qualities (e.g.,performance)

3. Experimental Controlled Experiment: Study artifact in controlled environment for qualities(e.g., usability)

Simulation � Execute artifact with artificial data

4. Testing Functional (Black Box) Testing: Execute artifact interfaces to discoverfailures and identify defects

Structural (White Box) Testing: Perform coverage testing of some metric(e.g., execution paths) in the artifact implementation

5. Descriptive Informed Argument: Use information from the knowledge base (e.g.,relevant research) to build a convincing argument for the artifact�s utility

Scenarios: Construct detailed scenarios around the artifact to demonstrateits utility

percent of the response time. Johansson et al.(2003) extended prior distributed database designresearch by developing a model that includesnetwork latency and the effects of parallel pro-cessing on response time.

The evaluation of designed artifacts typically usesmethodologies available in the knowledge base.These are summarized in Table 2. The selectionof evaluation methods must be matched appro-priately with the designed artifact and the selectedevaluation metrics. For example, descriptivemethods of evaluation should only be used forespecially innovative artifacts for which otherforms of evaluation may not be feasible. Thegoodness and efficacy of an artifact can berigorously demonstrated via well-selected evalua-tion methods (Basili 1996; Kleindorfer et al. 1998;Zelkowitz and Wallace 1998).

Design, in all of its realizations (e.g., architecture,landscaping, art, music), has style. Given theproblem and solution requirements, sufficientdegrees of freedom remain to express a variety offorms and functions in the artifact that areaesthetically pleasing to both the designer and theuser. Good designers bring an element of style totheir work (Norman 1988). Thus, we posit thatdesign evaluation should include an assessmentof the artifact�s style.

The measurement of style lies in the realm ofhuman perception and taste. In other words, weknow good style when we see it. While difficult todefine, style in IS design is widely recognized andappreciated (Kernighan and Plauger 1978; Wino-grad 1996). Gelernter (1998) terms the essenceof style in IS design machine beauty. He de-scribes it as a marriage between simplicity and



power that drives innovation in science andtechnology. Simon (1996) also notes the impor-tance of style in the design process. The ability tocreatively vary the design process, within thelimits of satisfactory constraints, challenges andadds value to designers who participate in theprocess.

Guideline 4: Research Contributions

Effective design-science research must provideclear contributions in the areas of the designartifact, design construction knowledge (i.e., foun-dations), and/or design evaluation knowledge (i.e.,methodologies). The ultimate assessment for anyresearch is, �What are the new and interestingcontributions?� Design-science research holdsthe potential for three types of research contri-butions based on the novelty, generality, andsignificance of the designed artifact. One or moreof these contributions must be found in a givenresearch project.

1. The Design Artifact. Most often, the contribu-tion of design-science research is the artifactitself. The artifact must enable the solution ofheretofore unsolved problems. It may extendthe knowledge base (see below) or applyexisting knowledge in new and innovativeways. As shown in Figure 2 by the left-facingarrow at the bottom of the figure from ISResearch to the Environment, exercising theartifact in the environment producessignificant value to the constituent IScommunity. System development method-ologies, design tools, and prototype systems(e.g., GDSS, expert systems) are examplesof such artifacts.

2. Foundations. The creative development ofnovel, appropriately evaluated constructs,models, methods, or instantiations thatextend and improve the existing foundationsin the design-science knowledge base arealso important contributions. The right-facingarrow at the bottom of the figure from ISResearch to the Knowledge Base in Figure 2indicates these contributions. Modeling

formalisms, ontologies (Wand and Weber1993, 1995; Weber 1997), problem andsolution representations, design algorithms(Storey et al. 1997), and innovativeinformation systems (Aiken 1991; Markus etal. 2002; Walls et al. 1992) are examples ofsuch artifacts.

3. Methodologies. Finally, the creative develop-ment and use of evaluation methods (e.g.,experimental, analytical, observational,testing, and descriptive) and new evaluationmetrics provide design-science researchcontributions. Measures and evaluationmetrics in particular are crucial componentsof design-science research. The right-facingarrow at the bottom of the figure from ISResearch to the Knowledge Base in Figure 2also indicates these contributions. TAM, forexample, presents a framework for predictingand explaining why a particular informationsystem will or will not be accepted in a givenorganizational setting (Venkatesh 2000).Although TAM is posed as a behavioraltheory, it also provides metrics by which adesigned information system or implemen-tation process can be evaluated. Its implica-tions for design itself are as yet unexplored.

Criteria for assessing contribution focus onrepresentational fidelity and implementability.Artifacts must accurately represent the businessand technology environments used in theresearch, information systems themselves beingmodels of the business. These artifacts must be�implementable,� hence the importance of instan-tiating design science artifacts. Beyond these,however, the research must demonstrate a clearcontribution to the business environment, solvingan important, previously unsolved problem.

Guideline 5: Research Rigor

Rigor addresses the way in which research isconducted. Design-science research requires theapplication of rigorous methods in both theconstruction and evaluation of the designedartifact. In behavioral-science research, rigor is



often assessed by adherence to appropriate datacollection and analysis techniques. Over-emphasis on rigor in behavioral IS research hasoften resulted in a corresponding lowering ofrelevance (Lee 1999).

Design-science research often relies on mathe-matical formalism to describe the specified andconstructed artifact. However, the environmentsin which IT artifacts must perform and the artifactsthemselves may defy excessive formalism. Or, inan attempt to be mathematically rigorous,important parts of the problem may be abstractedor �assumed away.� In particular, with respect tothe construction activity, rigor must be assessedwith respect to the applicability and generali-zability of the artifact. Again, an overemphasis onrigor can lessen relevance. We argue, along withbehavioral IS researchers (Applegate 1999), thatit is possible and necessary for all IS researchparadigms to be both rigorous and relevant.

In both design-science and behavioral-scienceresearch, rigor is derived from the effective use ofthe knowledge base�theoretical foundations andresearch methodologies. Success is predicatedon the researcher�s skilled selection of appropriatetechniques to develop or construct a theory orartifact and the selection of appropriate means tojustify the theory or evaluate the artifact.

Claims about artifacts are typically dependentupon performance metrics. Even formal mathe-matical proofs rely on evaluation criteria againstwhich the performance of an artifact can bemeasured. Design-science researchers mustconstantly assess the appropriateness of theirmetrics and the construction of effective metrics isan important part of design-science research.

Furthermore, designed artifacts are often com-ponents of a human-machine problem-solvingsystem. For such artifacts, knowledge of behav-ioral theories and empirical work are necessary toconstruct and evaluate such artifacts. Constructs,models, methods, and instantiations must beexercised within appropriate environments.Appropriate subject groups must be obtained forsuch studies. Issues that are addressed include

comparability, subject selection, training, time,and tasks. Methods for this type of evaluation arenot unlike those for justifying or testing behavioraltheories. However, the principal aim is to deter-mine how well an artifact works, not to theorizeabout or prove anything about why the artifactworks. This is where design-science andbehavioral-science researchers must complementone another. Because design-science artifactsare often the �machine� part of the human-machine system constituting an information sys-tem, it is imperative to understand why an artifactworks or does not work to enable new artifacts tobe constructed that exploit the former and avoidthe latter.

Guideline 6: Design as aSearch Process

Design science is inherently iterative. The searchfor the best, or optimal, design is often intractablefor realistic information systems problems.Heuristic search strategies produce feasible, gooddesigns that can be implemented in the businessenvironment. Simon (1996) describes the natureof the design process as a Generate/Test Cycle(Figure 3).

Design is essentially a search process to discoveran effective solution to a problem. Problemsolving can be viewed as utilizing available meansto reach desired ends while satisfying lawsexisting in the environment (Simon 1996).Abstraction and representation of appropriatemeans, ends, and laws are crucial components ofdesign-science research. These factors are prob-lem and environment dependent and invariablyinvolve creativity and innovation. Means are theset of actions and resources available to constructa solution. Ends represent goals and constraintson the solution. Laws are uncontrollable forces inthe environment. Effective design requires knowl-edge of both the application domain (e.g., require-ments and constraints) and the solution domain(e.g., technical and organizational).

Design-science research often simplifies a prob-lem by explicitly representing only a subset of the



Generate

Design

Alternatives

Test Alternatives

Against

Requirements/Constraints

Generate

Design

Alternatives

Test Alternatives

Against

Requirements/Constraints

Figure 3. The Generate/Test Cycle

relevant means, ends, and laws or by decom-posing a problem into simpler subproblems. Suchsimplifications and decompositions may not berealistic enough to have a significant impact onpractice but may represent a starting point.Progress is made iteratively as the scope of thedesign problem is expanded. As means, ends,and laws are refined and made more realistic, thedesign artifact becomes more relevant andvaluable. The means, ends, and laws for ISdesign problems can often be represented usingthe tools of mathematics and operations research.Means are represented by decision variableswhose values constitute an implementable designsolution. Ends are represented using a utilityfunction and constraints that can be expressed interms of decision variables and constants. Lawsare represented by the values of constants usedin the utility function and constraints.

The set of possible design solutions for anyproblem is specified as all possible means thatsatisfy all end conditions consistent with identifiedlaws. When these can be formulated appro-priately and posed mathematically, standardoperations research techniques can be used todetermine an optimal solution for the specifiedend conditions. Given the wicked nature of manyinformation system design problems, however, it

may not be possible to determine, let aloneexplicitly describe, the relevant means, ends, orlaws (Vessey and Glass 1998). Even when it ispossible to do so, the sheer size and complexity ofthe solution space will often render the problemcomputationally infeasible. For example, to builda �reliable, secure, and responsive informationsystems infrastructure,� one of the key issuesfaced by IS managers (Brancheau et al. 1996), adesigner would need to represent all possibleinfrastructures (means), determine their utility andconstraints (ends), and specify all cost and benefitconstants (laws). Clearly such an approach isinfeasible. However, this does not mean thatdesign-science research is inappropriate for sucha problem.

In such situations, the search is for satisfactorysolutions, i.e., satisficing (Simon 1996), withoutexplicitly specifying all possible solutions. Thedesign task involves the creation, utilization, andassessment of heuristic search strategies. Thatis, constructing an artifact that �works� well for thespecified class of problems. Although its con-struction is based on prior theory and existingdesign knowledge, it may or may not be entirelyclear why it works or the extent of its generaliza-bility; it simply qualifies as �credentialed knowl-edge� (Meehl 1986, p. 311). While it is important



to understand why an artifact works, the criticalnature of design in IS makes it important to firstestablish that it does work and to characterize theenvironments in which it works, even if we cannotcompletely explain why it works. This enables ISpractitioners to take advantage of the artifact toimprove practice and provides a context foradditional research aimed at more fully explicatingthe resultant phenomena. Markus et al. (2002),for example, describe their search process interms of iteratively identifying deficiencies inconstructed prototype software systems andcreatively developing solutions to address them.

The use of heuristics to find �good� design solu-tions opens the question of how goodness ismeasured. Different problem representations mayprovide varying techniques for measuring howgood a solution is. One approach is to prove ordemonstrate that a heuristic design solution isalways within close proximity of an optimal solu-tion. Another is to compare produced solutionswith those constructed by expert human designersfor the same problem situation.

Guideline 7: Communicationof Research

Design-science research must be presented bothto technology-oriented as well as management-oriented audiences. Technology-oriented audi-ences need sufficient detail to enable thedescribed artifact to be constructed (implemented)and used within an appropriate organizationalcontext. This enables practitioners to take advan-tage of the benefits offered by the artifact and itenables researchers to build a cumulative knowl-edge base for further extension and evaluation. Itis also important for such audiences to under-stand the processes by which the artifact wasconstructed and evaluated. This establishesrepeatability of the research project and builds theknowledge base for further research extensions bydesign-science researchers in IS.

Management-oriented audiences need sufficientdetail to determine if the organizational resources

should be committed to constructing (or pur-chasing) and using the artifact within their specificorganizational context. Zmud (1997) suggeststhat presentation of design-science research for amanagerial audience requires an emphasis not onthe inherent nature of the artifact itself, but on theknowledge required to effectively apply the artifact�within specific contexts for individual or organi-zational gain� (p. ix). That is, the emphasis mustbe on the importance of the problem and thenovelty and effectiveness of the solution approachrealized in the artifact. While we agree with thisstatement, we note that it may be necessary todescribe the artifact in some detail to enablemanagers to appreciate its nature and understandits application. Presenting that detail in concise,well-organized appendices, as advised by Zmud,is an appropriate communication mechanism forsuch an audience.

Application of the DesignScience ResearchGuidelines

To illustrate the application of the design-scienceguidelines to IS research, we have selected threeexemplar articles for analysis from three differentIS journals, one from Decision Support Systems,one from Information Systems Research, and onefrom MIS Quarterly. Each has strengths andweaknesses when viewed through the lens of theabove guidelines. Our goal is not to perform acritical evaluation of the quality of the researchcontributions, but rather to illuminate the design-science guidelines. The articles are

� Gavish and Gerdes (1998), which developstechniques for implementing anonymity inGroup Decision Support Systems (GDSS)environments

� Aalst and Kumar (2003), which proposes adesign for an eXchangeable Routing Lan-guage (XRL) to support electronic commerceworkflows among trading partners



� Markus, Majchrzak, and Gasser (2002),which proposes a design theory for thedevelopment of information systems built tosupport emergent knowledge processes

The fundamental questions for design-scienceresearch are, �What utility does the new artifactprovide?� and �What demonstrates that utility?�Evidence must be presented to address these twoquestions. That is the essence of design science.Contribution arises from utility. If existing artifactsare adequate, then design-science research thatcreates a new artifact is unnecessary (it isirrelevant). If the new artifact does not map ade-quately to the real world (rigor), it cannot provideutility. If the artifact does not solve the problem(search, implementability), it has no utility. If utilityis not demonstrated (evaluation), then there is nobasis upon which to accept the claims that itprovides any contribution (contribution). Further-more, if the problem, the artifact, and its utility arenot presented in a manner such that the implica-tions for research and practice are clear, thenpublication in the IS literature is not appropriate(communication).

The Design and Implementationof Anonymity in GDSS:Gavish and Gerdes

The study of group decision support systems(GDSS) has been and remains one of the mostvisible and successful research streams in the ISfield. The use of information technology to effec-tively support meetings of groups of different sizesover time and space is a real problem thatchallenges all business organizations. RecentGDSS literature surveys demonstrate the largenumbers of GDSS research papers published inthe IS field and, more importantly, the wide varietyof research paradigms applied to GDSS research(e.g., Dennis and Wixom 2001; Fjermestad andHiltz 1998; Nunamaker et al. 1996). However,only a small number of GDSS papers can beconsidered to make true design-science researchcontributions. Most assume the introduction of anew information technology or process in the

GDSS environment and then study the individual,group, or organizational implications using abehavioral-science research paradigm. Severalsuch GDSS papers have appeared in MISQuarterly (e.g., Dickson et al. 1993; Gallupe et al.1988; Jarvenpaa et al. 1988; Sengupta and Te�eni1993).

The central role of design science in GDSS isclearly recognized in the early foundation papersof the field. The University of Arizona ElectronicMeeting System group, for example, states theneed for both developmental and empiricalresearch agendas (Dennis et al. 1988; Nuna-maker et al. 1991b). Developmental, or design-science, research is called for in the areas ofprocess structures and support and task struc-tures and support. Process structure and supporttechnologies and methods are generic to allGDSS environments and tasks. Technologiesand methods for distributed communications,group memory, decision-making methods, andanonymity are a few of the critical design issuesfor GDSS process support needed in any taskdomain. Task structure and support are specificto the problem domain under consideration by thegroup (e.g., medical decision making, softwaredevelopment). Task support includes the designof new technologies and methods for managingand analyzing task-related information and usingthat information to make specific, task-relateddecisions.

The issue of anonymity has been studiedextensively in GDSS environments. Behavioralresearch studies have shown both positive andnegative impacts on group interactions. On thepositive side, GDSS participants can express theirviews freely without fear of embarrassment orreprisal. However, anonymity can encourage free-riding and antisocial behaviors. While the prosand cons of anonymity in GDSS are muchresearched, there has been a noticeable lack ofresearch on the design of techniques for imple-menting anonymity in GDSS environments.Gavish and Gerdes (1998) address this issue bydesigning five basic mechanisms to provideGDSS procedural anonymity.



Problem Relevance

The amount of interest and research on anonymityissues in GDSS testifies to its relevance. Fieldstudies and surveys clearly indicate that partici-pants rank anonymity as a highly desired attributein the GDSS system. Many individuals state thatthey would refuse to participate in or trust theresults of a GDSS meeting without a satisfactorylevel of assured anonymity (Fjermestad and Hiltz1998).

Research Rigor

Gavish and Gerdes base their GDSS anonymitydesigns on past research in the fields of crypto-graphy and secure network communication proto-cols (e.g., Chaum 1981; Schneier 1996). Theseresearch areas have a long history of formal,rigorous results that have been applied to thedesign of many practical security and privacymechanisms. Appendix A of the exemplar paperprovides a set of formal proofs that the claimsmade by the authors for the anonymity designsare correct and draw their validity from theknowledge base of this past research.

Design as a Search Process

The authors motivate their design scienceresearch by identifying three basic types of anony-mity in a GDSS system: environmental, content,and procedural. After a definition and brief dis-cussion of each type, they focus on the design ofmechanisms for procedural anonymity; the abilityof the GDSS system to hide the source of anymessage. This is a very difficult requirementbecause standard network protocols typicallyattach source information in headers to supportreliable transmission protocols. Thus, GDSS sys-tems must modify standard communication proto-cols and include additional transmission proce-dures to ensure required levels of anonymity.

The design-science process employed by theauthors is to state the desired procedural anony-mity attributes of the GDSS system and then to

design mechanisms to satisfy the systemrequirements for anonymity. Proposed designsare presented and anonymity claims are proved tobe correct. A thorough discussion of the costsand benefits of the proposed anonymitymechanisms is provided in Section 4 of the paper.

Design as an Artifact

The authors design a GDSS system architecturethat provides a rigorous level of proceduralanonymity. Five mechanisms are employed toensure participant anonymity:

� All messages are encrypted with a uniquesession key

� The sender�s header information is removedfrom all messages

� All messages are re-encrypted upon retrans-mission from any GDSS server

� Transmission order of messages is ran-domized

� Artificial messages are introduced to thwarttraffic analysis

The procedures and communication protocols thatimplement these mechanisms in a GDSS systemare the artifacts of this research.

Design Evaluation

The evaluation consists of two reported activities.First, in Appendix A, each mechanism is proved tocorrectly provide the claimed anonymity benefits.Formal proof methods are used to validate theeffectiveness of the designed mechanisms.Second, Section 4 presents a thorough cost-benefit analysis. It is shown that the operationalcosts of supporting the proposed anonymitymechanisms can be quite significant. In addition,the communication protocols to implement themechanisms add considerable complexity to thesystem. Thus, the authors recommend that a



cost-benefit justification be performed beforedetermining the level of anonymity to implementfor a GDSS meeting.

The authors do not claim to have implemented theproposed anonymity mechanisms in a prototypeor actual GDSS system. Thus, an instantiation ofthe designed artifact remains to be evaluated inan operational GDSS environment.

Research Contributions

The design-science contributions of this researchare the proposed anonymity mechanisms as thedesign artifacts and the evaluation results in theform of formal proofs and cost-benefit analyses. These contributions advance our understanding ofhow best to provide participant anonymity inGDSS meetings.

Research Communication

Although the presentation of this research isaimed at an audience familiar with network systemconcepts such as encryption and communicationprotocols, the paper also contains important,useful information for a managerial audience.Managers should have a good understanding ofthe implications of anonymity in GDSS meetings.This understanding must include an appreciationof the costs of providing desired levels ofparticipant anonymity. While the authors providea thorough discussion of cost-benefit tradeoffstoward the end of the paper, the paper would bemore accessible to a managerial audience if itincluded a stronger motivation up front on theimportant implications of anonymity in GDSSsystem development and operations.

A Workflow Language for Inter-organizational Processes: Aalst and Kumar

Workflow models are an effective means for de-scribing, analyzing, implementing, and managing

business processes. Workflow managementsystems are becoming integral components ofmany commercial enterprise-wide informationsystems (Leymann and Roller 2000). Standardsfor workflow semantics and syntax (i.e., workflowlanguages) and workflow architectures arepromulgated by the Workflow ManagementCoalition (WfMC 2000). While workflow modelshave been used for many years to manage intra-organizational business processes, there is now agreat demand for effective tools to model inter-organization processes across heterogeneousand distributed environments, such as those foundin electronic commerce and complex supplychains (Kumar and Zhao 2002).

Aalst and Kumar (2003) investigate the problem ofexchanging business process information acrossmultiple organizations in an automated manner.They design an eXchangable Routing Language(XRL) to capture workflow models that are thenembedded in eXtensible Markup Language (XML)for electronic transmission to all participants in aninterorganizational business process. The designof XRL is based upon Petri nets, which provide aformal basis for analyzing the correctness andperformance of the workflows, as well assupporting the extensibility of the language. Theauthors develop a workflow management archi-tecture and a prototype implementation toevaluate XRL in a proof of concept.

Problem Relevance

Interorganizational electronic commerce isgrowing rapidly and is projected to soon exceedone trillion dollars annually (eMarketer 2002). Amultitude of electronic commerce solutions arebeing proposed (e.g., ebXML, UDDI, RosettaNet)to enable businesses to execute transactions instandardized, open environments. While XML hasbeen widely accepted as a protocol for ex-changing business data, there is still no clearstandard for exchanging business process infor-mation (e.g., workflow models). This is the veryrelevant problem addressed by this research.



Research Rigor

Research on workflow modeling has long beenbased on rigorous mathematical techniques suchas Markov chains, queueing networks, and Petrinets (Aalst and Hee 2002). In this paper, Petrinets provide the underlying semantics for XRL.These formal semantics allow for powerful analy-sis techniques (e.g., correctness, performance) tobe applied to the designed workflow models.Such formalisms also enable the development ofautomated tools to manipulate and analyze com-plex workflow designs. Each language constructin XRL has an equivalent Petri-net representationpresented in the paper. The language is exten-sible in that adding a new construct simplyrequires defining its Petri-net representation andadding its syntax to the XRL. Thus, this researchdraws from a clearly defined and tested base ofmodeling literature and knowledge.


XRL is designed in the paper by performing athorough analysis of business process require-ments and identifying features provided by leadingcommercial workflow management systems.Using the terminology from the paper, workflowstraverse routes through available tasks (i.e.,business services) in the electronic businessenvironment. The basic routing constructs of XRLdefine the specific control flow of the businessprocess. The authors build 13 basic constructsinto XRL: Task, Sequence, Any_sequence,Choice, Condition, Parallel_sync, Parallel_no_sync, Parallel_part_sync, Wait_all, Wait_any,While_do, Stop, and Terminate. They show thePetri-net representation of each construct. Thus,the fundamental control flow structures ofsequence, decision, iteration, and concurrency aresupported in XRL.

The authors demonstrate the capabilities of XRLin several examples. However, they are carefulnot to claim that XRL is complete in the formalsense that all possible business processes can bemodeled in XRL. The search for a complete set ofXRL constructs is left for future research.


There are two clearly identifiable artifacts pro-duced in this research. First, the workflow lan-guage XRL is designed. XRL is based on Petri-net formalisms and described in XML syntax.Interorganizational business processes arespecified via XRL for execution in a distributed,heterogeneous environment.

The second research artifact is the XRL/flowerworkflow management architecture in which XRL-described processes are executed. The XRLrouting scheme is parsed by an XML parser andstored as an XML data structure. This structure isread into a Petri-net engine which determines thenext step of the business process and informs thenext task provider via an e-mail message. Resultsof each task are sent back to the engine whichthen executes the next step in the process untilcompletion. The paper presents a prototypeimplementation of the XRL/flower architecture asa proof of concept (Aalst and Kumar 2003).

Another artifact of this research is a workflowverification tool named Wolfan that verifies thesoundness of business process workflows.Soundness of a workflow requires that theworkflow terminates, no Petri-net tokens are leftbehind upon termination, and there are no deadtasks in the workflow. This verification tool isdescribed more completely in a different paper(Aalst 1999).

Design Evaluation

The authors evaluate the XRL and XRL/flowerdesigns in several important ways:

� XRL is compared and contrasted with lan-guages in existing commercial workflowsystems and research prototypes. Themajority of these languages are proprietaryand difficult to adapt to ad hoc businessprocess design.

� The fit of XRL with proposed standards isstudied. In particular, the Interoperability Wf-



XML Binding standard (WfMC 2000) does notat this time include the specification of controlflow and, thus, is not suitable for inter-organizational workflows. Electronic com-merce standards (e.g., RosettaNet) providesome level of control flow specification forpredefined business activities, but do notreadily allow the ad hoc specification ofbusiness processes.

� A research prototype of XRL/flower has beenimplemented and several of the user interfacescreens are presented. The screens demon-strate a mail-order routing schema casestudy.

� The Petri-net foundation of XRL allows theauthors to claim the XRL workflows can beverified for correctness and performance.XRL is extensible since new constructs canbe added to the language based on theirtranslation to underlying Petri-net repre-sentations. However, as discussed above,the authors do not make a formal claim forthe representational completeness of XRL.


The clear contributions of this research are thedesign artifacts�XRL (a workflow language),XRL/flower (a workflow architecture and itsimplemented prototype system), and Wolfan (aPetri-net verification engine). Another interestingcontribution is the extension of XML in its ability todescribe and transmit routing schemas (e.g.,control flow information) to support interorgani-zational electronic commerce.


This paper provides clear information to bothtechnical and managerial audiences. The presen-tation, while primarily technical with XML codingand Petri-net diagrams throughout, motivates amanagerial audience with a strong introduction onrisks and benefits of applying interorganizationalworkflows to electronic commerce applications.

Information Systems Design forEmergent Knowledge Processes: Markus, Majchrzak, and Gasser

Despite decades of research and developmentefforts, effective methods for developing infor-mation systems that meet the information require-ments of upper management remain elusive.Early approaches used a �waterfall� approachwhere requirements were defined and validatedprior to initiating design efforts which, in turn, werecompleted prior to implementation (Royce 1998).Prototyping approaches emerged next, followedby numerous proposals including CASE tool-based approaches, rapid application development,and extreme programming (Kruchten 2000).Walls et al. (1992) propose a framework for aprescriptive information system design theoryaimed at enabling designers to construct �moreeffective information systems� (p. 36). They applythis framework to the design of vigilant executiveinformation systems. The framework establishesa class of user requirements (model of designproblems) that are most effectively addressedusing a particular type of system solution(instantiation) designed using a prescribed set ofdevelopment practices (methods). Markus et al.(2002) extend this framework to the developmentof information systems to support emergentknowledge processes (EKPs)�processes inwhich structure is �neither possible nor desirable�(p. 182) and where processes are characterizedby �highly unpredictable user types and workcontexts� (p. 183).

Problem Relevance

The relevance and importance of the problem arewell demonstrated. Markus et al. describe a classof management activities that they term emergentknowledge processes (EKPs). These include�basic research, new product development,strategic business planning, and organizationdesign� (p. 179). They are characterized by �pro-cess emergence, unpredictable user types anduse contexts, and distributed expert knowledge�(p. 186). They are crucial to many manufacturingorganizations, particularly those in high-tech



industries. Such organizations recognize theneed to integrate organizational design and infor-mation system design with manufacturing opera-tions. They recognize the potential for significantperformance improvements offered by suchintegration. Yet few have realized that potential.Markus et al. argue that this is due to a lack of anadequate design theory and lack of scientificallybased tools, noting that existing informationsystem development methodologies focus onstructured or semi-structured decision processesand are inadequate for the development of sys-tems to support EKPs. TOP Modeler, the artifactcreated in this research effort, squarely addressesthis problem. Not surprisingly, its developmentattracted the attention and active participation ofseveral large, high-tech manufacturing organi-zations including �Hewlett-Packard, GeneralMotors, Digital Equipment Corporation, and TexasInstruments� (p. 186).

Research Rigor

The presented work has theoretical foundations inboth IS design theory and organizational designtheory. It uses the basic notions of IS designtheory presented in Walls et al. (1992) and posesa prescription for designing information systems tosupport EKPs. Prior research in developingdecision support systems, executive informationsystems, and expert systems serves as a foun-dation for this work and deficiencies of theseapproaches for the examined problem type serveas motivation. The knowledge-base constructedwithin TOP Modeler was formed from a synthesisof socio-technical systems theory and theempirical literature on organizational designknowledge. It was evaluated theoretically usingstandard metrics from the expert systemsliterature and empirically using data gathered fromnumerous electronics manufacturing companiesin the United States. Development of TOPModeler used an �action research paradigm�starting with a �kernel theory� based on priordevelopment methods and theoretical results anditeratively posing and testing artifacts (prototypes)to assess progress toward the desired result.Finally, the artifact was commercialized and �used

in over two dozen �real use� situations.� (p. 187).In summary, this work effectively used theoreticalfoundations from IS and organizational theory,applied appropriate research methods indeveloping the artifact, defined and appliedappropriate performance measures, and testedthe artifact within an appropriate context.


As discussed above, implementation and iterationare central to this research. The authors studyprototypes that instantiate posed or newly learneddesign prescriptions. Their use and impacts wereobserved, problems identified, solutions posedand implemented, and the cycle was thenrepeated. These interventions occurred over aperiod of 18 months within the aforementionedcompanies as they dealt with organizationaldesign tasks. As a result, not only was the TOPModeler developed and deployed but prescrip-tions (methods) in the form of six principles fordeveloping systems to support EKPs were alsodevised. The extensive experience, creativity,intuition, and problem solving capabilities of theresearchers were involved in assessing problemsand interpreting the results of deploying variousTOP modeler iterations and in constructingimprovements to address shortcomings identified.


The TOP Modeler is an implemented softwaresystem (instantiation). It is composed of anobject-oriented user interface, an object-orientedquery generator, and an analysis module built ontop of a relational meta-knowledge base thatenables access to �pluggable� knowledge basesrepresenting different domains. It also includestools to support the design and construction ofthese knowledge bases. The TOP Modeler sup-ports a development process incorporating the sixprinciples for developing systems to supportEKPs. As mentioned above, TOP Modeler wascommercialized and used in a number of differentorganizational redesign situations.



Design Evaluation

Evaluation is in the context of organizationaldesign in manufacturing organizations, and isbased on observation during the development anddeployment of a single artifact, TOP Modeler. Noformal evaluation was attempted in the sense ofcomparison with other artifacts. This is notsurprising, nor is it a criticism of this work. Theresimply are no existing artifacts that address thesame problem. However, given that method-ologies for developing information systems tosupport semi-structured management activitiesare the closest available artifacts, it is appropriateto use them as a comparative measure. In effect,this was accomplished by using principles fromthese methodologies to inform the initial design ofTOP Modeler. The identification of deficiencies inthe resultant artifact provides evidence that theseartifacts are ill-suited to the task at hand.

Iterative development and deployment within thecontext of organizational design in manufacturingorganizations provide opportunities to observeimprovement but do not enable formal evalua-tion�at each iteration, changes are induced in theorganization that cannot be controlled. As men-tioned above, the authors have taken a creativeand innovative approach that, of necessity, tradesoff rigor for relevancy. In the initial stages of adiscipline, this approach is extremely effective.TOP Modeler demonstrates the feasibility ofdeveloping an artifact to support organizationaldesign and EKPs within high-tech manufacturingorganizations. �In short, the evidence suggeststhat TOP Modeler was successful in supportingorganizational design� (p. 187) but additionalstudy is required to assess the comparativeeffectiveness of other possible approaches in thisor other contexts. Again, this is not a criticism ofthis work; rather it is a call for further research inthe general class of problems dealing with emer-gent knowledge processes. As additional re-search builds on this foundation, formal, rigorousevaluation and comparison with alternativeapproaches in a variety of contexts becomecrucial to enable claims of generalizability. As theauthors point out, �Only the accumulated weight of

empirical evidence will establish the validity� ofsuch claims.


The design-science contributions of this researchare the TOP Modeler software and the designprinciples. TOP Modeler demonstrates the feasi-bility of using the design principles to develop anartifact to support EKPs. Because TOP Modeleris the first artifact to address this task, itsconstruction is itself a contribution to designscience. Furthermore, because the authors areable to articulate the design principles upon whichits construction was based, these serve ashypotheses to be tested by future empirical work.Their applicability to the development of othertypes of information systems can also be tested.An agenda for addressing such issues is pre-sented. This focuses on validation, evaluation,and the challenges of improvement inherent in theevaluation process.


This work presents two types of artifacts, TOPModeler (an instantiation) and a set of designprinciples (method) that address a heretoforeunsolved problem dealing with the design of aninformation system to support EKPs. Recognizingthat existing system development methods andinstantiations are aimed at structured or semi-structured activities, Markus et al. identify anopportunity to apply information technology in anew and innovative way. Their presentationaddresses each of the design guidelines posedabove. TOP Modeler exemplifies �proof by con-struction��it is feasible to construct an infor-mation system to support EKPs. Since it is thefirst such artifact, its evaluation using formalmethods is deferred until future research.Technical details of TOP Modeler are not pre-sented, making it difficult for a technicalresearcher or practitioner to replicate their work.The uniqueness of the artifacts and the innovationinherent in them are presented so that managerialresearchers and IT managers are aware of thenew capabilities.



Discussion and Conclusions

Philosophical debates on how to conduct ISresearch (e.g., positivism vs. interpretivism) havebeen the focus of much recent attention (Klein andMyers 1999; Robey 1996; Weber 2003). Themajor emphasis of such debates lies in theepistemologies of research, the underlyingassumption being that of the natural sciences.That is, somewhere some truth exists andsomehow that truth can be extracted, explicated,and codified. The behavioral-science paradigmseeks to find �what is true.� In contrast, thedesign-science paradigm seeks to create �what iseffective.� While it can be argued that utility relieson truth, the discovery of truth may lag the appli-cation of its utility. We argue that both design-science and behavioral-science paradigms areneeded to ensure the relevance and effectivenessof IS research. Given the artificial nature oforganizations and the information systems thatsupport them, the design-science paradigm canplay a significant role in resolving the fundamentaldilemmas that have plagued IS research: rigor,relevance, discipline boundaries, behavior, andtechnology (Lee 2000).

Information systems research lies at the inter-section of people, organizations, and technology(Silver et al. 1995). It relies on and contributes tocognitive science, organizational theory, manage-ment sciences, and computer science. It is bothan organizational and a technical discipline that isconcerned with the analysis, construction, deploy-ment, use, evaluation, evolution, and manage-ment of information system artifacts in organi-zational settings (Madnick 1992; Orlikowski andBarley 2001).

Within this setting, the design-science researchparadigm is proactive with respect to technology.It focuses on creating and evaluating innovative ITartifacts that enable organizations to address im-portant information-related tasks. The behavioral-science research paradigm is reactive withrespect to technology in the sense that it takestechnology as �given.� It focuses on developingand justifying theories that explain and predictphenomena related to the acquisition, implemen-

tation, management, and use of such techno-logies. The dangers of a design-science researchparadigm are an overemphasis on the technologi-cal artifacts and a failure to maintain an adequatetheory base, potentially resulting in well-designedartifacts that are useless in real organizationalsettings. The dangers of a behavioral-scienceresearch paradigm are overemphasis on contex-tual theories and failure to adequately identify andanticipate technological capabilities, potentiallyresulting in theories and principles addressingoutdated or ineffective technologies. We arguestrongly that IS research must be both proactiveand reactive with respect to technology. It needsa complete research cycle where design sciencecreates artifacts for specific information problemsbased on relevant behavioral science theory andbehavioral science anticipates and engages thecreated technology artifacts.

Hence, we reiterate the call made earlier by Marchet al. (2000) to align IS design-science researchwith real-world production experience. Resultsfrom such industrial experience can be framed inthe context of our seven guidelines. These mustbe assessed not only by IS design-scienceresearchers but also by IS behavioral-scienceresearchers who can validate the organizationalproblems as well as study and anticipate theimpacts of created artifacts. Thus, we encouragecollaborative industrial/academic research pro-jects and publications based on such experience.Markus et al. (2002) is an excellent example ofsuch collaboration. Publication of these resultswill help accelerate the development of domainindependent and scalable solutions to large-scaleinformation systems problems within organiza-tions. We recognize that a lag exists betweenacademic research and its adoption in industry.We also recognize the possible ad hoc nature oftechnology-oriented solutions developed in indus-try. The latter gap can be reduced considerablyby developing and framing the industrial solutionsbased on our proposed guidelines.

It is also important to distinguish between �systembuilding� efforts and design-science research.Guidelines addressing evaluation, contributions,and rigor are especially important in providing this



distinction. The underlying formalism required bythese guidelines helps researchers to developrepresentations of IS problems, solutions, andsolution processes that clarify the knowledgeproduced by the research effort.

As we move forward, there exist a number ofexciting challenges facing the design-scienceresearch community in IS. A few are summarizedhere.

� There is an inadequate theoretical base uponwhich to build an engineering discipline ofinformation systems design (Basili 1996).The field is still very young lacking thecumulative theory development found in otherengineering and social-science disciplines. Itis important to demonstrate the feasibility andutility of such a theoretical base to a mana-gerial audience that must make technology-adoption decisions that can have far-reachingimpacts on the organization.

� Insufficient sets of constructs, models,methods, and tools exist for accurately repre-senting the business/technology environment.Highly abstract representations (e.g., analyti-cal mathematical models) are criticized ashaving no relationship to �real-world� environ-ments. On the other hand, many informal,descriptive IS models lack an underlyingtheory base. The trade-offs between rele-vance and rigor are clearly problematic;finding representational techniques with anacceptable balance between the two is verydifficult.

� The existing knowledge base is often insuffi-cient for design purposes and designers mustrely on intuition, experience, and trial-and-error methods. A constructed artifact em-bodies the designer�s knowledge of theproblem and solution. In new and emergingapplications of technology, the artifact itselfrepresents an experiment. In its execution,we learn about the nature of the problem, theenvironment, and the possible solutions�hence, the importance of developing andimplementing prototype artifacts (Newell andSimon 1976).

� Design-science research is perishable.Rapid advances in technology can invalidatedesign-science research results before theyare implemented effectively in the businessenvironment or, just as importantly to mana-gers, before adequate payback can beachieved by committing organizationalresources to implementing those results.Two examples are the promises made by theartificial intelligence community in the 1980s(Feigenbaum and McCorduck 1983) and themore recent research on object-orienteddatabases (Chaudhri and Loomis 1998). Justas important to IS researchers, design resultscan be overtaken by technology before theyeven appear in the research literature. Howmuch research was published on the Year2000 problem before it became a non-event?

� Rigorous evaluation methods are extremelydifficult to apply in design-science research(Tichy 1998; Zelkowitz and Wallace 1998).For example, the use of a design artifact on asingle project may not generalize to differentenvironments (Markus et al. 2002).

We believe that design science will play anincreasingly important role in the IS profession. ISmanagers in particular are actively engaged indesign activities�the creation, deployment, eval-uation, and improvement of purposeful IT artifactsthat enable organizations to achieve their goals.The challenge for design-science researchers inIS is to inform managers of the capabilities andimpacts of new IT artifacts.

Much of the research published in MIS Quarterlyemploys the behavioral-science paradigm. It ispassive with respect to technology, often ignoringor �under-theorizing� the artifact itself (Orlikowskiand Iacono 2001). Its focus is on describing theimplications of technology�its impact on indivi-duals, groups, and organizations. It regularlyincludes studies that examine how people employa technology, report on the benefits and difficultiesencountered when a technology is implementedwithin an organization, or discuss how managersmight facilitate the use of a technology. Orman(2002) argues that many of the equivocal results



in IS behavioral-science studies can be explainedby a failure to differentiate the capabilities andpurposes of the studied technology.

Design science is active with respect to tech-nology, engaging in the creation of technologicalartifacts that impact people and organizations. Itsfocus is on problem solving but often takes asimplistic view of the people and the organiza-tional contexts in which designed artifacts mustfunction. As stated earlier, the design of an arti-fact, its formal specification, and an assessmentof its utility, often by comparison with competingartifacts, are integral to design-science research.These must be combined with behavioral andorganizational theories to develop an under-standing of business problems, contexts, solu-tions, and evaluation approaches adequate toservicing the IS research and practitioner com-munities. The effective presentation of design-science research in major IS journals, such asMIS Quarterly, will be an important step towardintegrating the design-science and behavioral-science communities in IS.

Acknowledgements

We would like to thank Allen Lee, Ron Weber,and Gordon Davis who in different ways eachcontributed to our thinking about design science inthe Information Systems profession and en-couraged us to pursue this line of research. Wewould also like to acknowledge the efforts ofRosann Collins who provided insightful commentsand perspectives on the nature of the relationshipbetween behavioral-science and design-scienceresearch. This work has also benefited fromseminars and discussions at Arizona State Univer-sity, Florida International University, Georgia StateUniversity, Michigan State University, Notre DameUniversity, and The University of Utah. We wouldparticularly like to thank Brian Pentland and SteveAlter for feedback and suggestions they providedon an earlier version of this paper. The commentsprovided by several anonymous editors andreviewers greatly enhanced the content andpresentation of the paper.

References

Aalst, W. �Wolfan: A Petri-Net-Based WorkflowAnalyzer,� Systems Analysis-Modeling-Simu-lation (34:3), 1999, pp. 345-357.

Aalst, W., and Hee, K. Workflow Management:Models, Methods, and Systems, The MITPress, Cambridge, MA, 2002.

Aalst, W., and Kumar, A. �XML-Based SchemaDefinition for Support of InterorganizationalWorkflow,� Information Systems Research(14:1), March 2003, pp. 23-46.

Aboulafia, M. (ed.). Philosophy, Social Theory,and the Thought of George Herbert Mead(SUNY Series in Philosophy of the SocialSciences), State University of New YorkPress, Albany, NY, 1991.

Aiken, M. W., Sheng, O. R. L., and Vogel, D. R.�Integrating Expert Systems with GroupDecision Support Systems,� ACM Trans-actions on Information Systems (9:1),January 1991, pp. 75-95.

Alter, S. �18 Reasons Why IT-Reliant WorkSystems Should Replace �The IT Artifact� asthe Core Subject Matter of the IS Field,� Com-munications of the AIS (12), October 2003,pp. 365-394.

Applegate, L. M. �Rigor and Relevance in MISResearch�Introduction,� MIS Quarterly(23:1), March 1999, pp. 1-2.

Basili, V. �The Role of Experimentation in Soft-ware Engineering: Past, Current, and Fu-ture,� in Proceedings of the 18thInternationalConference on Software Engineering, T.Maibaum and M. Zelkowitz (eds.), Berlin,Germany, March 25-29,1996, pp. 442-449.

Basu, A., and Blanning, R. W. �A Formal Ap-proach to Workflow Analysis,� InformationSystems Research (11:1), March 2000, pp.17-36.

Basu, A., and Blanning, R. W. �Metagraphs: ATool for Modeling Decision Support Systems,�Management Science (40:12), December1994, pp. 1579-1600.

Batra, D., Hoffer, J. A., and Bostrom, R. P. �AComparison of User Performance betweenthe Relational and the Extended Entity Rela-tionship Models in the Discovery Phase ofDatabase Design,� Communications of theACM (33:2), February 1990, pp. 126-139.



Bell, D. A. �From Data Properties to Evidence,�IEEE Transactions on Knowledge and DataEngineering (5:6), December 1993, pp. 965-969.

Benbasat, I., and Zmud, R. W. �EmpiricalResearch in Information Systems: The Prac-tice of Relevance,� MIS Quarterly (23:1),March 1999, pp. 3-16.

Bhargava, H. K., Krishnan, R., and Piela, P. �OnFormal Semantics and Analysis of TypedModeling Languages: An Analysis ofAscend,� INFORMS Journal on Computing(10:2), Spring 1998, pp. 189-208.

Bodart, F., Patel, A., Sim, M., and Weber, R.�Should the Optional Property Construct beUsed in Conceptual Modeling? A Theory andThree Empirical Tests,� Information SystemsResearch (12:4), December 2001, pp 384-405.

Boland, R. J. �Design in the Punctuation of Man-agement Action� in Managing as Designing:Creating a Vocabulary for Management Edu-cation and Research, R. Boland (ed.), Fron-tiers of Management Workshop, Weather-head School of Management, June 14-15,2002 (available online at http://design.cwru.edu).

Brancheau, J., Janz, B., and Wetherbe, J. �KeyIssues in Information Systems Management:1994-95 SIM Delphi Results,� MIS Quarterly(20:2), June 1996, pp. 225-242.

Brooks, F. P., Jr. �The Computer Scientist asToolsmith II,� Communications of the ACM(39:3), March 1996, pp. 61-68.

Brooks, F. P., Jr. �No Silver Bullet: Essence andAccidents of Software Engineering,� IEEEComputer (20:4), April 1987, pp 10-19.

Bunge, M. A. Treatise on Basic Philosophy:Volume 7�Epistemology & Methodology III:Philosophy of Science and Technology�PartII: Life Science, Social Science and Tech-nology, D. Reidel Publishing Company,Boston, MA, 1985.

Chaudhri, A., and Loomis, M. Object Databasesin Practice, Prentice-Hall, Upper SaddleRiver, NJ, 1998.

Chaum, D. �Untraceable Electronic Mail, ReturnAddresses, and Digital Pseudonyms,� Com-munications of the ACM (24:2), February1981, pp. 84-87.

Chen, P. P. S. �The Entity-Relationship Model:Toward a Unified View,� ACM Transactionson Database Systems (1:1), 1976, pp. 9-36.

Davis, G. B., and Olson, M. H. ManagementInformation Systems: Conceptual Foun-dations, Structure and Development (2nd ed.),McGraw-Hill, New York, 1985.

DeLone, W. H., and McLean, E. R. �The DeLoneand McLean Model of Information SystemsSuccess: A Ten-Year Update,� Journal ofManagement Information Systems (19:4),Spring 2003, pp. 9-30.

DeLone, W. H., and McLean, E. R. �InformationSystems Success: The Quest for theDependent Variable,� Information SystemsResearch (3:1), March 1992, pp. 60-95.

Denning, P. J. �A New Social Contract forResearch,� Communications of the ACM(40:2), February 1997, pp. 132-134.

Dennis, A., George, J., Jessup, L., Nunamaker, J.,and Vogel, D. �Information Technology toSupport Electronic Meetings,� MIS Quarterly(12:4), December 1988, pp. 591-624.

Dennis, A., and Wixom, B. �Investigating theModerators of the Group Support SystemsUse with Meta-Analysis,� Journal of Manage-ment Information Systems (18:3), Winter2001-02, pp. 235-257.

Dey, D., Sarkar, S., and De, P. �A ProbabilisticDecision Model for Entity Matching in Hetero-geneous Databases,� Management Science(44:10), October 1998, pp. 1379-1395.

Dey, D., Storey, V. C., and Barron, T. M.�Improving Database Design through theAnalysis of Relationships,� ACM Transactionson Database Systems (24:4), December1999, pp. 453-486.

Dickson, G., Partridge, J., and Robinson, L.�Exploring Modes of Facilitative Support forGDSS Technology,� MIS Quarterly (17:2),June 1993, pp. 173-194.

Drucker, P. F. �The Coming of the New Organi-zation,� Harvard Business Review (66:1),January-February 1988, pp. 45-53.

Drucker, P. F. �The New Productivity Challenge,�Harvard Business Review (69:6), November-December 1991, pp. 45-53.

Dym, C. L. Engineering Design, Cambridge Uni-versity Press, New York, 1994.



eMarketer. E-Commerce Trade and B2B Ex-changes, March 2002 (available online athttp://www.emarketer.com/products/report.php?ecommerce_trade).

Feigenbaum, E., and McCorduck, P. The FifthGeneration: Artificial Intelligence and Japan�sComputer Challenge to the World, Addison-Wesley, Inc., Reading, MA, 1983.

Fjermestad, J., and Hiltz, S. R. �An Assessmentof Group Support Systems Experimental Re-search: Methodology and Results,� Journalof Management Information Systems (15:3),Winter 1998-99, pp. 7-149.

Gallupe, R., DeSanctis, G., and Dickson, G.�Computer-Based Support for Group Prob-lem-Finding: An Experimental Investigation,�MIS Quarterly, (12:2), June 1988, pp. 277-298.

Gavish, B., and Gerdes, J. �Anonymous Mecha-nisms in Group Decision Support SystemsCommunication,� Decision Support Systems(23:4), October 1998, pp. 297-328.

Geerts, G., and McCarthy, W. E. �An OntologicalAnalysis of the Primitives of the Extended-REA Enterprise Information Architecture,�The International Journal of AccountingInformation Systems (3:1), 2002, pp. 1-16.

Gelernter, D. Machine Beauty: Elegance and theHeart of Technology, Basic Books, New York,1998.

Glass, R. �On Design,� IEEE Software (16:2),March/April 1999, pp. 103-104.

Halpin, T. A. Information Modeling and RelationalDatabases, Morgan Kaufmann Publishers,New York, 2001.

Henderson, J., and Venkatraman, N. �StrategicAlignment: Leveraging Information Techno-logy for Transforming Organizations,� IBMSystems Journal (32:1), 1993.

ISR. Editorial Statement and Policy, InformationSystems Research (13:4), December 2002.

Jarvenpaa, S., Rao, V., and Huber, G. �ComputerSupport for Meetings of Groups Working onUnstructured Problems: A Field Experiment,�MIS Quarterly (12:4), December 1988, pp.645-666.

Johansson, J. M. �On the Impact of NetworkLatency on Distributed Systems Design,�Information Technology Management (1),2000, pp. 183-194.

Johansson, J. M., March, S. T., and Naumann, J.D. �Modeling Network Latency and ParallelProcessing in Distributed Database Design,�Decision Sciences Journal (34:4), Fall 2003.

Kalakota, R., and Robinson, M. E-Business 2.0:Roadmap for Success, Addison-WesleyPearson Education, Boston, MA, 2001.

Keil, M. �Pulling the Plug: Software ProjectManagement and the Problem of ProjectEscalation,� MIS Quarterly (19:4) December1995, pp. 421-447.

Keil, M., Cule, P. E., Lyytinen, K., and Schmidt, R.C. �A Framework for Identifying SoftwareProject Risks,� Communications of the ACM(41:11), November 1998, pp. 76-83.

Keil, M., and Robey, D. �Turning Around TroubledSoftware Projects: An Exploratory Study ofthe Deescalation of Commitment to FailingCourses of Action,� Journal of ManagementInformation Systems, (15:4) December 1999,pp. 63-87.

Kernighan, B., and Plauger, P. J. The Elementsof Programming Style (2nd ed.), McGraw-Hill,New York, 1978.

Kim, Y. G., and March, S. T. �Comparing DataModeling Formalisms,� Communications ofthe ACM (38:6), June 1995, pp. 103-115.

Klein, H. K., and Myers, M. D. �A Set of Principlesfor Conducting and Evaluating InterpretiveField Studies in Information Systems,� MISQuarterly (23:1), March 1999, pp. 67-94.

Kleindorfer, G., O�Neill, L., and Ganeshan, R.�Validation in Simulation: Various Positionsin the Philosophy of Science,� ManagementScience (44:8), August 1998, pp. 1087-1099.

Kruchten, P. The Rational Unified Process: AnIntroduction (2nd ed.), Addison-Wesley, Inc.,Reading, MA, 2000.

Kuhn, T. S. The Structure of Scientific Revolu-tions (3rd ed.), University of Chicago Press,Chicago, IL, 1996.

Kumar, A., and Zhao, J. �Workflow Support forElectronic Commerce Applications,� DecisionSupport Systems (32:3), January 2002, pp.265-278.

Lee, A. �Inaugural Editor�s Comments,� MISQuarterly (23:1), March 1999, pp. v-xi.

Lee, A. �Systems Thinking, Design Science, andParadigms: Heeding Three Lessons from thePast to Resolve Three Dilemmas in the



Present to Direct a Trajectory for FutureResearch in the Information Systems Field,�Keynote Address, Eleventh International Con-ference on Information Management, Taiwan,May 2000 (available online at http://www.people.vcu.edu/~aslee/ICIM-keynote-2000).

Leymann, F., and Roller, D. Production Workflow:Concepts and Techniques, Prentice-Hall,Upper Saddle River, NJ, 2000.

Madnick, S. E. �The Challenge: To Be Part of theSolution Instead of Being Part of the Prob-lem,� in Proceedings of the Second AnnualWorkshop on Information Technology andSystems, V. Storey and A. Whinston (eds.),Dallas, TX, December 12-13, 1992, pp. 1-9.

Marakas, G. M., and Elam, J. J. �SemanticStructuring in Analyst Acquisition and Repre-sentation of Facts in Requirements Analysis,�Information Systems Research (9:1), March1998, pp. 37-63.

March, S. T., Hevner, A., and Ram, S. �ResearchCommentary: An Agenda for InformationTechnology Research in Heterogeneous andDistributed Environment,� Information Sys-tems Research (11:4), December 2000, pp.327-341.

March, S. T., and Smith, G. �Design and NaturalScience Research on Information Techno-logy,� Decision Support Systems (15:4),December 1995, pp. 251-266.

Markus, M. L., Majchrzak, A., and Gasser, L. �ADesign Theory for Systems that SupportEmergent Knowledge Processes,� MIS Quar-terly (26:3), September, 2002, pp. 179-212.

McCarthy, W. E. �The REA Accounting Model: AGeneralized Framework for Accounting Sys-tems in a Shared Data Environment,� TheAccounting Review (58:3), 1982, pp. 554-578.

Meehl, P. E. �What Social Scientists Don�tUnderstand,� in Metatheory in Social Science,D. W. Fiske and R. A. Shweder (eds.),University of Chicago Press, Chicago, IL,1986, pp. 315-338.

Newell, A., and Simon, H. �Computer Science asEmpirical Inquiry: Symbols and Search,�Communications of the ACM (19:3), March1976, pp. 113-126.

Norman, D. The Design of Everyday Things,Currency Doubleday, New York, 1988.

Nunamaker, J., Briggs, R., Mittleman, D., Vogel,D., and Balthazard, P. �Lessons from aDozen Years of Group Support SystemsResearch: A Discussion of Lab and FieldFindings,� Journal of Management Informa-tion Systems, (13:3), Winter 1996-97, pp.163-207.

Nunamaker, J., Chen, M., and Purdin, T. D. M.�Systems Development in Information Sys-tems Research,� Journal of ManagementInformation Systems (7:3), Winter 1991a, pp.89-106.

Nunamaker, J., Dennis, A., Valacich, J., Vogel, D.,and George, J. �Electronic Meeting Systemsto Support Group Work,� Communications ofthe ACM, (34:7), July 1991b, pp. 40-61.

Orlikowski, W. J. �Using Technology and Con-stituting Structures: A Practice Lens forStudying Technology in Organizations.�Organization Science (11:4), December 2000,pp. 404-428.

Orlikowski, W. J., and Barley, S. R. �Technologyand Institutions: What Can Research onInformation Technology and Research onOrganizations Learn From Each Other?,� MISQuarterly (25:2), June 2001, pp 145-165.

Orlikowski, W. J., and Iacono, C. S. �ResearchCommentary: Desperately Seeking the �IT� inIT Research�A Call to Theorizing the ITArtifact,� Information Systems Research(12:2), June 2001, pp. 121-134.

Orman, L. V. �Electronic Markets, Hierarchies,Hubs, and Intermediaries,� Journal of Infor-mation Systems Frontiers (4:2), 2002, pp.207-222.

Pahl, G., and Beitz, W. Engineering Design: ASystematic Approach, Springer-Verlag, Lon-don, 1996.

Parsons, J., and Wand, Y. �EmancipatingInstances from the Tyranny of Classes inInformation Modeling,� ACM Transactions onDatabase Systems (25:2), June 2000, pp.228-268.

Petroski, H. Invention by Design: How EngineersGet from Thought to Thing, Harvard Univer-sity Press, Cambridge, MA, 1996.



Purao, S., Storey, V. C., and Han, T. D. �Im-proving Reuse-Based System Design withLearning,� Information Systems Research(14:3), September 2003, pp. 269-290.

Rittel, H. J., and Webber, M. M. �Planning Prob-lems Are Wicked Problems,� in Develop-ments in Design Methodology, N. Cross (ed.),John Wiley & Sons, New York, 1984.

Robey, D. �Research Commentary: Diversity inInformation Systems Research: Threat,Opportunity, and Responsibility,� InformationSystems Research (7:4), 1996, pp. 400-408.

Royce, W. Software Project Management: AUnified Framework, Addison-Wesley, Inc.,Reading, MA, 1998.

Salton, G. Automatic Text Processing: TheTransformation, Analysis, and Retrieval ofInformation by Computer, Addison-Wesley,Inc., Reading, MA, 1988.

Schneier, B. Applied Cryptography: Protocols,Algorithms, and Source Code in C (2nd ed.),John Wiley and Sons, New York, January1996.

Schön, D. A. The Reflective Practitioner: HowProfessionals Think in Action, Basic Books,New York, 1983.

Seddon, P. B. �A Respecification and Extensionof the DeLone and McLean Model of ISSuccess,� Information Systems Research(8:3), September 1997, pp. 240-253.

Sengupta, K., and Te�eni, D. �Cognitive Feedbackin GDSS: Improving Control and Conver-gence,� MIS Quarterly (17:1), March 1993,pp. 87-113.

Silver, M. S., Markus, M. L., and Beath, C. M.�The Information Technology InteractionModel: A Foundation for the MBA CoreCourse,� MIS Quarterly (19:3), September1995, pp. 361-390.

Simon, H. A. The Sciences of the Artificial (3rd

ed.), MIT Press, Cambridge, MA, 1996.Sinha, A. P., and Vessey, I. �An Empirical Inves-

tigation of Entity-Based and Object-OrientedData Modeling: A Development Life CycleApproach,� in Proceedings of the TwentiethInternational Conference on InformationSystems, P. De and J. I. DeGross (eds.),Charlotte, NC, December 13-15, 1999, pp.229-244.

Storey, V. C., Chiang, R. H. L., Dey, D., Goldstein,R. C., and Sundaresan, S. �Database Designwith Common Sense Business Reasoningand Learning,� ACM Transactions onDatabase Systems (22:4), December 1997,pp. 471-512.

Tam, K. Y. �Automated Construction of Knowl-edge-Bases from Examples,� InformationSystems Research (1:2), June 1990, pp. 144-167.

Tichy, W. �Should Computer Scientists Experi-ment More?� IEEE Computer (31:5), May1998, pp. 32-40.

Trice, A., and Davis, R. �Heuristics for Recon-ciling Independent Knowledge Bases,� Infor-mation Systems Research (4:3), September1993, pp. 262-288.

Tsichritzis, D. �The Dynamics of Innovation,� inBeyond Calculation: The Next Fifty Years ofComputing, P. J. Denning and R. M. Metcalfe(eds.), Copernicus Books, New York, 1998,pp. 259-265.

Venkatesh, V. �Determinants pf Perceived Easeof Use: Integrating Control, Intrinsic Motiva-tion, and Emotion into the Technology Accep-tance Model,� Information Systems Research(11:4), December 2000, pp. 342-365.

Vessey, I., and Glass, R. �Strong Vs. WeakApproaches to Systems Development,�Communications of the ACM (41:4), April1998, pp. 99-102.

Walls, J. G., Widmeyer, G. R., and El Sawy, O. A.�Building an Information System DesignTheory for Vigilant EIS,� Information SystemsResearch (3:1), March 1992, pp. 36-59.

Wand, Y., and Weber, R. �On the Deep Structureof Information Systems,� Information SystemsJournal (5), 1995, pp. 203-233.

Wand, Y., and Weber, R. �On the OntologicalExpressiveness of Information SystemsDesign Analysis and Design Grammars,�Journal of Information Systems (3:3), 1993,pp. 217-237.

Weber, R. �Editor�s Comments: Still DesperatelySeeking the IT Artifact,� MIS Quarterly (27:2),June 2003, pp. iii-xi.

Weber, R. Ontological Foundations of InformationSystems, Coopers & Lybrand, Brisbane,Australia, 1997.



Weber, R. �Toward a Theory of Artifacts: AParadigmatic Base for Information SystemsResearch,� Journal of Information Systems(1:2), Spring 1987, pp. 3-19.

WfMC. �Workflow Standard�Interoperability Wf-XML Binding,� Document Number WFMC-TC-1023, Version 1.0, Workflow ManagementCoalition, 2000.

Winograd, T. Bringing Design to Software,Addison-Wesley, Inc., Reading, MA, 1996.

Winograd, T. �The Design of Interaction,� inBeyond Calculation: The Next 50 Years ofComputing, P. Denning and R. Metcalfe(eds.), Copernicus Books, New York, 1998,pp. 149-162.

Zelkowitz, M., and Wallace, D. �ExperimentalModels for Validating Technology,� IEEEComputer (31:5), May 1998, pp. 23-31.

Zmud, R. �Editor�s Comments,� MIS Quarterly(21:2), June 1997, pp. xxi-xxii.

About the Authors

Alan R. Hevner is an Eminent Scholar and Pro-fessor in the College of Business Administration atthe University of South Florida. He holds theSalomon Brothers/Hidden River Corporate ParkChair of Distributed Technology. His areas ofresearch interest include information systemsdevelopment, software engineering, distributeddatabase systems, and healthcare informationsystems. He has published numerous researchpapers on these topics and has consulted forseveral Fortune 500 companies. Dr. Hevnerreceived a Ph.D. in Computer Science fromPurdue University. He has held faculty positionsat the University of Maryland at College Park andthe University of Minnesota. Dr. Hevner is amember of ACM, IEEE, AIS, and INFORMS.

Salvatore T. March is the David K. Wilson Pro-fessor of Management at the Owen GraduateSchool of Management, Vanderbilt University. Hereceived a B.S. in Industrial Engineering and M.S.and Ph.D. degrees in Operations Research fromCornell University. His research interests are ininformation system development, distributed data-

base design, and electronic commerce. Hisresearch has appeared in journals such as Com-munications of the ACM, IEEE Transactions onKnowledge and Data Engineering, and Informa-tion Systems Research. He served as the Editor-in-Chief of ACM Computing Surveys and as anassociate editor for MIS Quarterly. He is currentlya senior editor for Information Systems Researchand an associate editor for Decision SciencesJournal.

Jinsoo Park is an assistant professor of infor-mation systems in the College of Business Admin-istration at Korea University. He was formerly onthe faculty of the Carlson School of Managementat the University of Minnesota. He holds a Ph.D.in MIS from the University of Arizona. Hisresearch interests are in the areas of semanticinteroperability and metadata management ininterorganizational information systems, hetero-geneous information resource management andintegration, knowledge sharing and coordination,and data modeling. His published researcharticles appear in IEEE Computer, IEEE Transac-tions on Knowledge and Data Engineering, andInformation Systems Frontiers. He currentlyserves on the editorial board of Journal of Data-base Management. He is a member of ACM,IEEE, AIS, and INFORMS.

Sudha Ram is the Eller Professor of MIS at theUniversity of Arizona. She received a B.S. inScience from the University of Madras in 1979,PGDM from the Indian Institute of Management,Calcutta in 1981 and a Ph.D. from the Universityof Illinois at Urbana-Champaign, in 1985. Dr.Ram has published articles in such journals asCommunications of the ACM, IEEE Transactionson Knowledge and Data Engineering, InformationSystems Research, and Management Science.Her research deals with interoperability in hetero-geneous databases, semantic modeling, dataallocation, and intelligent agents for data manage-ment. Her research has been funded by IBM,National Institute of Standards and Technology(NIST), National Science Foundation (NSF),National Aeronautics and Space Administration(NASA), and the Office of Research and Devel-opment of the Central Intelligence Agency (CIA).

Documents

design science in information systems research - Temple Fox MIS