the case research strategy in studies of information systems · tive technique, the case research...

Case Research Strategies

The Case ResearchStrategy in Studies ofInformation Systems

By: Izak BenbasatProfessor & Chairman of MISFaculty of Commerce and Business

AdministrationThe University of British ColumbiaVancouver, Canada V6T IY8

By: David K. GoldsteinAssistant Professor of Business

AdministrationHarvard Business SchoolHarvard UniversityBoston, Massachusetts 02163

By: Melissa MeadAssistant Professor of Business

AdministrationHarvard Business SchoolHarvard UniversityBoston, Massachusetts 02163

AbstractThis article defines and discusses one of thesequalitative methods--the case research strat-egy. Suggestions are provided for researcherswho wish to undertake research employingthis approach. Criteria for the evaluation ofcase research are established and severalcharacteristics useful for categorizing thestudies are identified. A sample of papersdrawn from information systems journals isreviewed. The paper concludes with examplesof research areas that are particularly well-suited to investigation using the case researchapproach.

ACM categories: H.O., J.O.

Keywords:Case study research, data collec-tion, information systems, qualita-tive methods, research methods.

IntroductionThere has been a growing interest in the useof qualitative techniques in the administrativesciences. For example, a full issue of Ad-ministrative Science Quarterly (Volume 24,1979) has been devoted to qualitative meth-ods. This interest has been sparked by ageneral dissatisfaction with the type of re-search information provided by quantitativetechniques [51]. The dissatisfaction stemsfrom several sources: the complexity of mul-tivariate research methods, the distributionrestrictions inherent in the use of these meth-ods (e.g., multivariate normality), the largesample sizes these methods dictate, and thedifficulty of understanding and interpretingthe results of studies in which complex quan-titative methods are used.

Similarly, in the information systems (IS) field,Franz and Robey [14] have suggested the useof idiographic rather than nomothetic re-search strategies. Idiographic research at-tempts to understand a phenomenon in itscontext. In such research, the investigator in-tensely examines a single entity or a particu-lar event. Nomothetic methods, on the otherhand, seek general laws and draw solely onprocedures used in the exact sciences [54].

This article discusses the use of one qualita-tive technique, the case research strategy, instudies of information systems. It providessome suggestions about how to conduct andevaluate case study research. A sample ofcase-based research from selected IS jour-nals is categorized according to a set of char-acteristics developed in this paper. The arti-cles in the sample are then evaluated.

We are not advocating an exclusive use of thecase strategy. Many authors have comment-ed that each research strategy has advan-tages and disadvantages; no strategy is moreappropriate than all others for all researchpurposes. Benbasat [3] showed that the goalsof the researcher and the nature of the re-search topic influence the selection of astrategy. Case research is particularly ap-propriate for certain types of problems: thosein which research and theory are at their ear-ly, formative stages [44], and "sticky, practice-based problems where the experiences of theactors are important and the context of actionis critical" [4].

MIS Quarterly~September 1987 369


The information systems area is character-ized by constant technological change and in-novation. IS researchers, therefore, often findthemselves trailing behind practitioners inproposing changes or in evaluating methodsfor developing new systems. Researchersusually learn by studying the innovations putin place by practitioners, rather than byproviding the initial wisdom for these novelideas. For example, when companies expe-rienced a growth in end-user computing in thelate 1970s and early 1980s, academics werenot able to’offer a set of guidelines describinghow an organization could effectively managethe introduction of end-user computing tech-nology. Researchers first descriptively stud-ied how organizations were managing end-user computing. These studies then formedthe basis for the development of prescriptivemanagement guidelines (for example, Rock-art and Flannery [43]).

We believe that the case research strategy iswell-suited to capturing the knowledge ofpractitioners and developing theories from it.Christenson [9] points out that the trial-and-error process in which practitioners are en-gaged is necessary for knowledge to accumu-late. It is incumbent upon the scientists to for-malize this knowledge and proceed to a test-ing stage. Before this formalization takesplace, case studies could be employed todocument the experiences of practice.

The IS field has also seen a shift from techno-logical to managerial and organizationalquestions, and consequently more interest inhow context and innovations interact. For ex-ample, airline reservation systems were veryinnovative technical achievements in the ear-ly 1960s. However, they became a key com-petitive factor in the changing airline industrywithin the last few years. In order to under-stand this phenomenon, one must examinethe structure of the industry, the role ofderegulation, and the federal laws governingthe industry.

To summarize, there are three reasons whycase study research is a viable informationsystems research strategy. First, the re-searcher can study information systems in anatural setting, learn about the state of theart, and generate theories from practice. Sec-ond, the case method allows the researcherto answer "how" and "why" questions, that is,to understand the nature and complexity of

the processes taking place. Questions suchas, "How does a manager effectively intro-duce new information technologies?" are criti-cal ones for researchers to pursue. Third, acase approach is an appropriate way to re-search an area in which few previous studieshave been carried out. With the rapid pace ofchange in the information systems field, manynew topics emerge each year for which valua-ble insights can be gained through the use ofcase research.

Case Research: DefinitionThere is no standard definition of a casestudy. For our purposes, we will draw our de-finition from those presented by Benbasat [3],Bonoma [5], Kaplan [23], Stone [46], and Yin[56]. A case study examines a phenomenonin its natural setting, employing multiplemethods of data collection to gather informa-tion from one or a few entities (people,groups, or organizations). The boundaries ofthe phenomenon are not clearly evident at theoutset of the research and no experimentalcontrol or manipulation is used. Table 1 con-tains a list of eleven characteristics of casestudies summarized from the papers men-tioned above.

To place case studies in perspective, it isuseful to contrast this approach with othermethods commonly used by IS researchers.In laboratory experiments the researchermeasures dependent variables while manipu-lating independent variables in a controlled,environment. Similarly, field experiments in-volve the manipulation and measurement ofclearly defined variables, but in a natural set-ting. Finally, in field studies researchers mea-sure independent and dependent variables intheir natural context; however, no control ormanipulation is involved. A fundamental dif-ference between case studies and these al-ternative methods is that the case studyresearcher may have less a priori knowledgeof what the variables of interest will be andhow they will be measured.1

We must emphasize, however, that this is a matterof degree. There are instances of case studieswhere the investigators had a prior’notion of certaincritical variables, such as the type of industries andthe type of firms they wanted to examine. For exam-ple, Lawrence and Lorsch [29] chose industries that

370 MIS Quarterly~September 1987


Table 1. Key Characteristics of Case Studies

1. Phenomenon is examined in a natural setting.2. Data are collected by multiple means.3. One or few entities (person, group, or organization) are examined.4. The complexity of the unit is studied intensively.5. Case studies are more suitable for the exploration, classification and hypothesis de-

velopment stages of the knowledge building process; the investigator should have areceptive attitude towards exploration.

6. No experimental controls or manipulation are involved.7. The investigator may not specify the set of independent and dependent variables in

advance.8. The results derived depend heavily on the integrative powers of the investigator.9. Changes in site selection and data collection methods could take place as the investiga-

tor develops new hypotheses.10. Case research is useful in the study of "why" and "how" questions because these deal

with operational links to be traced over time rather than with frequency or incidence.11. The focus is on contemporary events.

In our survey of the literature, we identifiedthree categories of qualitative research thatappear to be considered as case studies. Ourfocus in this article, however, includes onlyone of these. For clarification we will definethe two excluded categories-applicationdescriptions and action research. Then wewill describe key aspects of the execution ofcase research.

Application descriptions--Written by practi-tioners, application descriptions detail theauthor’s experiences implementing a particu-lar application, such as a database manage-ment system. The outcomes of these pub-lished projects are almost always successfuland the author concludes by providing a list of"dos" and "don’ts" for the implementation ofsimilar systems. The author does not conducta research study; instead, the objective is tosuccessfully implement a specific system fora given assignment. Application descriptionsare not included in our definition of caseresearch.

Action research--We have also excludedaction research. These are studies in whichthe author, usually a researcher, is a par-ticipant in the implementation of a system, butsimultaneously wants to evaluate a certain in-tervention technique. An example might bethe use of the sociotechnical approach for

differed in their rates of technological change. They -also studied effective and ineffective organizationswithin the same industry as well as comparing effec-tive organizations in different industries.

system development. The articles in both thisand the application descriptions categoriesare written by individuals who have an in-sider’s view of the system in question. How-ever, action research articles are authored bythose whose original intent is to conductresearch while effecting change [47]. Theaction researcher is not an independent ob-server, but becomes a participant, and theprocess of change becomes the subject of re-search. Thus, the researcher has two objec-tives: to take action to solve a problem and tocontribute to a set of system developmentconcepts [8].

The strength of these studies is the indepthand first hand understanding the researcherobtains. Conversely, a weakness is the poten-tial lack of objectivity stemming from theresearcher’s stake in effecting a successfuloutcome for the client organization. Further,generalizations to other situations where theintervention technique is applied by peopleless knowledgeable than the researcher maybe difficult. Examples of action research in ISare found in Gibson [16], Ginzberg [17], Mum-ford [35] and Morton [34].

Case study research-In case studies theclear objective is the conduct of research.These are efforts where research questionsare specified prior to the study by researcherswho are observers/investigators rather thanparticipants. We will discuss considerationsthat are important in selecting the case re-search approach and we will detail themechanics of executing case research.



Conducting Case Research

Research themes and objectives-Deciding on case research

Given a specific research question, one mustask whether the case method is a useful ap-proach. To judge the appropriateness of thecase strategy, we can ask the following ques-tions (drawn from Table 1):

1. Can the phenomenon of interest be stud-ied outside its natural setting?

2. Must the study focus on contemporaryevents?

3. Is control or manipulation of subjects orevents necessary?

4. Does the phenomenon of interest enjoyan established theoretical base?

Case methodology is clearly useful when anatural setting or a focus on contemporaryevents is needed. Similarly, research phe-nomena not supported by a strong theoreticalbase may be fruitfully pursued through caseresearch. A rich natural setting can be fertileground for generating theories. Conversely,when subjects or events must be controlled ormanipulated in the course of a research pro-ject, the case approach is not suitable.

Quite often, however, the decision to use acase approach is not clear-cut. Both Yin [56]and Bonoma [5] discuss the usefulness of thecase approach in various phases of research.Table 2 gives their terminology for the tradi-tional exploration, hypothesis generation,and testing phases of knowledge accrual. Re-search that is not strictly exploratory ordescriptive may be enhanced by using quail-

Table 2. Terminologies for Stages

tative methods. For example, Yin [56] statesthat case studies could be used to explainphenomena. He considers Allison’s [1] studyof the Cuban missile crisis an example ofsuch studies. Allison proposed different the-ories to account for the same course ofevents, identified the one that provided thebest explanation, and suggested how thistheory could be useful to understand othersituations. Bonoma suggests that the casestrategy could play a role in both hypothesisgeneration and hypothesis testing. Bauer (asreported in Towl [50]) refers to the use of thecritical case (crucial experiment) to test a well-founded theory.

When the case approach is deemed appropri-ate, researchers may be uncertain about howto proceed. The remainder of this sectionoffers practical aid to researchers for under-standing and implementing case research.

Unit of AnalysisPrior to searching for sites, the researchershould determine the unit of analysis most ap-propriate for the project. Will the study focuson individuals, groups (e.g., a task force,profit center, IS group) or an entire organiza-tion? Alternatively, the unit of analysis may bea specific project or decision. In making thisdetermination, the researcher should closelyexamine the research questions to be pur-sued. These often indicate an appropriateunit of analysis. Finally, the researcher shouldconsider what generalizations are hoped forat the project’s completion. Does the re-searcher hope to generalize to other orga-nizations, individuals, or decisions, forinstance?

of Case Research Programs

Traditional Phasesof Knowledge Yin’s [56]

Accrual FrameworkBonoma’s [5]Framework

Numberof Cases

Exploration

Hypothesis generation

Hypothesis testing

¯ Confirmation

¯ Disconfirmation

Description

Exploration

Explanation

Explanation

Drift

Design

Prediction

Disconfirmation

Single or multiplecase(s)

Multiple cases

Multiple cases

Single critical case .



When the research is highly exploratory, asingle case may be useful as a pilot study.The goal will be to determine the appropriateunit and familiarize the researcher with thephenomenon in its context.

Single-Case vs. Multiple-Case DesignsCentral to case research design is the deci-sion to include one or several cases in the pro-ject. Most research efforts require multiplecases, but single cases are useful in specificinstances. Yin suggests single-case studiesare appropriate if:

1) It is a reve/atory case, i.e., it is a situationpreviously inaccessible to scientificinvestigation.

2) It represents a critica/case for testing awell-formulated theory.2

3) It is an extreme or unique case.

As Table 2 shows, single-case study projectsare most useful at the outset of theory genera-tion and late in theory testing. A single caseused for exploration may be followed by amultiple-case study. This corresponds to Bo-noma’s drift stage in which researchers learnfirst hand the relevant jargon and context inwhich the phenomenon will be studied. A sin-gle case may also be used to test the bound-aries of well-formed theory.

Multiple-case designs are desirable when theintent of the research is description, theorybuilding, or theory testing. These three cor-respond to Bonoma’s design, prediction, anddisconfirmation stages, respectively. Multi-ple-case designs allow for cross-case analy-sis and the extension of theory. Of course,multiple cases yield more general researchresults.

Site SelectionThe factors that dictate a single-case designalso determine site selection. When multiplecases are to be included in a study, however,choices must be made. It is quite useful toconsider a multiple-case project as analogousto the replication that is possible with multipletraditional experiments [19]. Adopting thispoint of view, Yin proposes two criteria forselecting potential sites. First, sites where

According to Yin [56, p. 42] "To confirm, challenge orextend a theory, there may exist a single case, meet-ing all the conditions for testing the theory."

similar results are predicted may be used as"literal" replications. Second, sites may bechosen for "theoretical" replication. That is,chosen such that contradictory results arepredicted. With careful site selection, theresearcher can extend and revise the initialpropositions of the study.

Site selection should be carefully thought outrather than opportunistic. Researchers maybegin site selection by considering the natureof their topic. Research on organization-levelphenomena would require site selectionbased on the characteristics of firms. Thesemay include the industry, company size, or-ganizational structure, profit/not-for-profit sta-tus, public or private ownership, geographiccoverage, degree of vertical or horizontal in-tegration, and so on. Researchers interestedin specific technologies, IS methodologies ororganizational structures should considerthese characteristics when selecting sites.

Once the limiting factors are determined,specific sites may be identified and ap-proached. Regular scans of business news-papers and periodicals often turn up potentialsites. Library research using indexes of in-dustries, business literature, and marketingor financial research data may be helpful. Fi-nally, talking with friends, colleagues, or ac-quaintances is a good way to identify potentialresearch locations.

Approaching the potential site is a crucialpoint in orchestrating a case research project.Here again, the topic of study is key to deter-mining whom to contact. The researcher musteventually contact the individual with enoughauthority to approve the project. Colleaguesmay be able to help with introductions. If not,prepare carefully before placing a cold call orwriting to the organization. The researchershould clearly describe the project and whowill be involved--researchers, assistants, orcompany employees. The contact should betold the amount of time, effort and expense re-quired of the organization.

Two key points to be addressed in order togain cooperation are confidentiality and be-nefits to the organization. The researchermust provide assurance that the organizationwill not be harmed by its participation. The or-ganization and its employees must know thatthe researcher will not betray their con-



fidence. On the other hand, the researchershould seek assurance that reasonable can-dor will be provided and that essential datawill be made available.

The benefits to an organization participatingin a research project are varied. They may in-clude learning more about the organization,getting feedback and new insights from theresearcher, and developing a relationshipwith the researcher. In addition, there is theopportunity to contribute to knowledge andbusiness research. The organization may ormay not wish to be identified when the re-search is published. If it does, there is the ad-ditional benefit of recognition and publicity.

Data Collection MethodsMultiple data collection methods are typicallyemployed in case research studies. Ideally,evidence from two or more sources will con-verge to support the research findings. Yinidentifies several Sources of evidence thatwork well in case research [56, p. 78].

1. Documentation--Written material rang-ing from memoranda to newspaper clip-pings to formal reports.

2. Archival records--Organization charts;service, personnel or financial records.

3. Interviews--These may be open-en.dedor focused (See Bouchard [6]; Cook andCampbell [10]).

4. Direct observation--Absorbing and not-ing details, actions, or subtleties of thefield environment (see Webb, et al., [53]on unobtrusive measures).

5. Physical artifacts--Devices, outputs,tools.

The goal is to obtain a rich set of data sur-rounding the specific research issue, as wellas capturing the contextual complexity.

Specific data to be collected will depend onthe research questions and the unit of analy-sis. Prior to site visits, the researcher shouldoutline, in detail, the data to be gathered. Thismay include a list of materials to be collected(documentation, archival records and physi-cal artifacts) as well as questions for inter-views and plans for direct observation. Thisformalization helps coordination when multi-ple investigators work together. It also pro-vides some separation of data collection fromdata analysis. The goals of this planningshould be to ensure good coverage of the re-

search questions and excellent use of timespent on-site.

This planning stage helps to structureprojects that are inherently flexible. It givesthe researcher a guide from which to work. Asthe project unfolds, the plan will be revisedaccording to the researcher’s judgement, un-expected observations, or limitations andopportunities.

Finally, the researcher should be meticulousin record-keeping. Precious data may be lostwhen entrusted to memory or not organizedas soon after collection as possible. This isparticularly important in multiple case de-signs where, as time passes, the details ofvarious sites tend to run together. Theresearcher’s goal should be to collect data insuch a way that another researcher could pickit up and immediately understand it and workwith it.

Data analysis and exposition-The analysis of case data depends heavily onthe integrative powers of the researcher. Us-ing multiple methods of data collection, how-ever, offers the opportunity for triangulationand lends greater support to the researcher’sconclusions. Working with a research partnermay also provide invaluable assistance. Tworesearchers can capture greater richness ofdata and rely more confidently on the accura-cy of the data.

The key elements of data analysis are alsocritical to the written results of case research.As much as possible, the contextual and datarichness of the study should be presented,and a clear chain of evidence should be es-tablished. The researcher’s reasoning inestablishing cause and effect or drawing outhypotheses should be clearly stated anddefended. The research should move fromobjectives and questions, to assumptions anddesign choices, to specific data uncovered,and finally, to results and conclusions. Read-ers should be able to follow this path readily.

A Critique of Case-BasedResearchTo gain an understanding of the nature andquality of case research on IS, we surveyed

374 MIS Quarter/y/September 1987


the following journals and conference pro-’ceedings3 for the period January 1981 to De-cember 1985: Communications for the ACM,The Proceedings of the International Confer-ence on Information Systems, Information andManagement, Quarterly, and Systems, Objec-tives, Solutions.

After excluding the two categories of applica-tion descriptions and action research articles,we identified only five case papers in theCommunications of the ACM, an average ofone per year. There were few case papers inInformation and Management; even if we in-clude the action research and applicationdescription articles in this journal, case re-search makes up only 10% of all articles. TheProceedings of the International Conferenceon Information Systems included seven casepapers, again about one per year. Finally,about 10% of the articles in the MIS Quarterlywere case studies.

Compared to these journals, Systems, Objec-tives, Solutions had a substantial proportionof case research articles (about 25% of the ar-ticles published). This journal published caseresearch articles in each issue and en-couraged researchers to submit case studies.After four years as an independent publica-tion, the journal merged with Information andManagement in 1985.4

We approached the critique of the articles inour sample in two parts. First, we looked atfour case studies in detail and evaluated theirstrengths and weaknesses. Then, we ratedthe whole sample of case studies based onour guidelines for conducting case studies.

Four case research studiesWe chose the following four case studies be-cause each took a different approach to inves-tigation and they illustrate both the strengthsand weaknesses in case study research.

3 In our estimation, these were the IS journals mostlikely to publish case-type articles. Our intention wasto provide examples of case studies, not to do an ex-haustive search.

4 Hamilton and Ives [18], based on a sample of IS arti-cles from 15 journals published between 1970-79,observed that 14% of these were case studies.Based on a similar sample between 1977-83, Vogel

¯ and Wetherbe [52] report that about 17% of the pub-lished IS articles were case st,udies.

Markus: IS Implementation-Markus [32]examined the use of a "production planningand profit analysis system" in two manufactur-ing plants within the same division of a com-pany. The system was readily accepted inone plant, but was at first strongly rejected inthe other (eventually it was put into use). Hercase study attempted to find the reasons forthe contradictory outcomes of implementa-tion. Markus stated that an explanation basedon user participation could only partially ac-count for these findings. She observed thatthe plant that accepted the system exhibiteda higher degree of user participation than theone that rejected the system initially. But howdoes one account for the later reversal?

Markus proposed a "distribution of power"model to explain the different reactions to thesystem. That is, the lack of consonance be-tween the distribution of power implied by thesystem and the distribution of power withinthe organization caused the failure of oneplant to adopt the system. Based on this samemodel, Markus also explained why one sys-tem was rejected at first, but was eventually,accepted when the organizational conditionsthat led to the initial rejection were changed.

Markus’ research had to be carried out in anatural setting since it traced the evolution ofthe system’s implementation from rejection toacceptance and the reasons for the switch.Markus was interested in answering a "why"question: Why was the system used in oneplant and not in another? The fact that bothunits were within a single company increasedthe internal validity of the case study. A largenumber of possible causes for the accep-tance in one plant and not in the other couldbe eliminated since the sites shared the sameorganizational setting and company history.Thus, Markus took advantage of a unique op-portunity to study an implementation issue.Since the opportunity was unique, Markusadded to our knowledge about the implemen-tation of systems, even though some previousresearch on the same topic existed.

The history of the implementation process,which spanned a period of six years, was de-scribed in detail. One flaw in the studythough, was the total lack of detail about thedata collection methodology. The readercould only infer that interviews took place,since direct quotations from the people in-



volved were used to support the author’s ar-guments. Aside from this, it is not clear what,if any, additional data sources were used fortriangulation and validation. This is of particu-lar concern because the history of the projectcovered six years and the memories of theparticipants may have been inconsistent.

Dutton: Adoption of a Fiscal ImpactModeI--Dutton’s [12] study of the City of Tul-sa’s adoption of a fiscal impact model and itseventual rejection was outstanding amongcase studies we looked at for the rich level ofdetail it provided about the implementationprocess. According to Dutton, the sequenceof events was reconstructed based on news-paper accounts, government reports, docu-ments, memos, over 20 lengthy unstructuredinterviews with participants having differentperspectives on implementation, and tele-phone calls that preceded and followed sitevisits. As a final check for accuracy, a draft ofthe paper was sent to several respondentsand important participants who had not previ-ously been interviewed.

The purpose of the study was to examine thelimitations of both the technical and organiza-tional perspectives of the implementationprocess. Dutton stated that his study con-tributed to our understanding of the processof innovation rejection, a topic we know littleabout. Following a 16-page description of theevents, the organizational and political en-vironment, and the actors associated with theimplementation process, Dutton explainedhow technical, organizational, and interper-sonal factors influenced the implementationprocess. However, he concluded that thegreatest determinant was the political en-vironment; he described in detail how the or-ganizational and technical factors were con-tingent on the political ones.

Dutton described events in such detail thatreaders could make their own, different inter-pretations if they wished. It is an exploratorypaper that reveals an important factor in theimplementation process.

One criticism of the study is that Dutton didnot clearly define his original research objec-tives. Further, we do not know why he chosethis particular site. Did he know about the out-come (failure of system) before he analyzedretrospectively the events that took place? Ifthis is the case, then his original objective

might have been to explore why a failure hadoccurred. Alternatively, was his intention tostudy the implementation of computer-basedsystems in city governments and describe thefactors influencing the outcomes? In the lattercase, the investigator was in the discoverystage; in the former case he was attempting tofind the factors that caused a particular out-come. Since these concerns might affect theway the researcher approached data collec-tion and thought about the issues in the case,we believe that a clear statement of the initialobjectives of the researcher should havebeen provided in the study description.

Pyburn: Strategic IS Planning-Pyburn [39]investigated IS strategic planning processesthat were underway in several companies, ex-amined the business and technical context inwhich these plans were developed, and gen-erated preliminary conclusions about the suc-cess or failure of the planning activity. Hewanted to understand why a particular meth-odology worked well in some settings but notin others. The impetus for the study was thegrowing importance of such plans in the suc-cess of the overall IS effort, coupled with thelack of effectiveness of existing IS planningmethodologies.

The method used was a comparative casestudy which, as described by Pyburn, "relieson the fact that outcomes in the different siteswere the results of identified differences inthose factors measured for its conclusions"[39, pp. 4-5]. Therefore, he chose eight sitesthat were as similar as possible based on anumber of characteristics, i.e., a literal repli-cation. Some of these characteristics, suchas "the company had formal business plan-ning for five years" and "top managers werewilling to commit time to assist the research,"made obvious sense. However, the reasonsfor other characteristics of the site selection,such as "the companies were dominated by afounding family" and "had corporate head-quarters in relatively small cities," were notimmediately evident. It appeared that Pyburnwas more concerned with choosing highlysimilar sites, rather than choosing a modelgroup that was representative of companiesinvolved in strategic IS planning.

In each site, Pyburn conducted indepth inter-views with the senior IS executive and the topmanagement team (four to six individuals).He used a series of questions to gather data



concerning the nature of the business, thefactors critical to its success, the company’sIS planning practices, and the extent to whichIS was addressing the critical needs of thefirm. In order to increase the reliability of thedata collection process and to reduce inter-viewee bias, prior to the first interview, he ad-ministered a questionnaire that containeditems similar to the questions subsequentlyposed in the interviews. The questionnairedata were analyzed using nonparametricstatistical methods. For the interview data, acase description was written for each organi-zation. For a given topic, a side-by-sidepresentation of all comments made by eachinterviewee for each topic was provided.(However, the case descriptions and com-ments were not included in the published pa-per and are presumably part of Pyburn’s doc-toral thesis on which the article was based.)

Based on the interview data and partially sup-ported by questionnaire results, Pyburn iden-tified three IS planning styles: personal-informal, personal-formal, and written-formal.He then determined the degree of success ofIS strategic planning in each company. Sincenone of these planning approaches appearedto be uniformly successful in all the com-panies in which they were implemented, Py-burn pursued a contingency analysis. A num-ber of factors, such as the perceived status ofthe IS manager and the complexity of the ISenvironment, were identified as important in-fluences on the success or failure of the plan-ning process. Finally, all of this informationwas tied together and the critical factors in thesuccess of a particular planning style wereidentified. For example, a personal-informalstyle depended on both an informal generalmanagement style and high IS managerstatus.

Pyburn set forth clear research objectives pri-or to data collection. This led to a focused ap-proach to the interviews and site selection,and the use of triangulation in data collection.He carefully explained each step of a logicalprocess that eventually culminated in a pre-liminary typology of strategic MIS planning.The use of multiple sites allowed him to at-tempt a contingency analysis. This is a goodexample of an exploratory case study thatproposed a classification scheme for MISplanning that could be further refined andtested in other studies.

A drawback of this work was that the type ofdetailed descriptions found in the Dutton andthe Markus papers were not provided in thispaper. Of course, the journal may have im-posed length limitations that made it impossi-ble to present the data collected from each ofthe eight sites. Books or monographs mightbe better vehicles to publish case study re-search. Nevertheless, the reader has to relyon the author’s interpretation and cannot in-terpret the data independently.

Olson: Centralization of the System De-velopment Function--Olson [37] examinedthe issue of whether the systems develop-ment function should be centralized or locat-ed in the user organization. Based on an anal-ysis of the literature, Olson contended thatthere is no best way to organize the systemdevelopment function; the decision dependson factors outside the IS function. She there-fore conducted an in-depth study of two or-ganizations in order to identify the factors thatinfluence the organization of the system de-velopment function and the quality of the de-velopment process. She conducted extensiveinterviews with IS managers and key partici-pants in the system design process, and ad-ministered questionnaires to users as well asthose who were part of the design effort. Twocomputer-based systems (one an accountsreceivable system, the other a benefits sys-tem) were selected from two large multidivi-sional companies. In one case, the systemsdevelopment function was centralized; in theother it was decentralized to business divi-sions.

Olson’s results showed that in the decentral-ized development group users had a higherlevel of information satisfaction, users partici-pated less in the design of the system, andanalysts were less satisfied with their jobsthan in the centralized group. These resultswere based upon very small samples of usersand analysts in each organization, a fact ac-knowledged by Olson. She suggested severalconclusions about the effectiveness of userparticipation in the development process.

Olson chose her sample to bring out thedifferences associated with alternative or-ganizations of the system development func-tion. The views of the various participants inthe development process were sought anddata were collected by multiple means. Fur-thermore, recently completed systems were

MIS Quarter/y/September 1987 377


chosen to minimize the problems associatedwith retrospective data collection. A strengthof the paper was the author’s attempt to pro-vide explanations for the outcomes based onwhat was observed from the cases. The ISdepartments of the two companies and theiractivities were described in reasonable detail,but little was offered to the reader about thecompanies themselves. Finally, since the out-comes were based only on a sample of twocompanies, this study could be consideredonly a basic exploratory case study.

The characteristics of the four case studiesdiscussed above, as well as a sample ofothers identified in our survey, are outlined inTable 3. It is clear that the case studies haveboth similarities and differences. For exam-ple, the Pyburn and Dutton studies were bothexploratory in nature, but differed in terms ofsample selection and data gathering meth-ods. One common feature of all four studies,which is not evident in the larger sample out-lined in Table 3, was the detailed explana-tions that accompanied the findings. An in-vestigator conducting a field study or experi-ment mainly relies on theory or a priorireasoning to deductively arrive at the out-comes. In these four studies, the investiga-tors collected data, distilled the evidence, andinductively developed causal links to explainparticular outcomes. While on the positiveside the explanations were grounded in ob-served facts, our concern is that these factswere filtered through the subjective lenses ofthe investigators.

An overall evaluation of casestudiesIn this section, we will describe the natureand general quality of case research in IS.We’ll evaluate the case studies in our surveybased on our guidelines for conducting caseresearch.

Research ThemesThe predominant theme in the case studieswas implementation, that is, the possiblecauses of the success or failure of an informa-tion or decision support system (e.g., [11]).Since the process of implementation takesplace over time, is a complex process involv-ing multiple actors, and is influenced byevents that happen unexpectedly, a case

study methodology is well-suited to identify-ing key events and actors and to linking themin a causal chain. Examples of case studytopics are: the impact of organizational strat-egy on the IS organization’s structure [45], theimpact of IS on organizational change [41,42], the impact of technology on personnel[28], the influence of technology on organiza-tional communications [15], the factors affect-ing the success of end-user developed appli-cations [40], and the role of users in DSSdevelopment [31 ].

Research ObjectivesIn the published case research we surveyed,the objective of the study was seldom clearlyspecified. Among the exceptions were Py-burn [39] and Hirschheim [20], who statedthat their objectives were to describe andexplore a phenomenon that was not well-understood.

We would characterize most of the casestudies as exploratory in nature. They de-scribed the context in which an interventionoccurred and the intervention itself. For ex-ample, Kraemer [27] and Dutton [12] describ-ed the implementation of model-based sys-tems in the public sector. Almost all of theseexploratory studies concluded with a list ofsuggestions to improve the success of futureimplementation efforts. It was difficult to de-termine if the researchers were at the sametime attempting to generate hypotheses. Anexception is the study by Ives and Olson [21]which basically described the nature of an ISmanager’s job.

Some case studies pursued an explanatorystrategy by first describing the events thattook place and then presenting multiple com-peting theories to explain the course of events(e.g., Franz and Robey [14], Kling and lacono[26], Markus [33]). Each of these studies useda sample of one case. For example, Markus[33] evaluated three theories of resistance:peopledetermined, system-determined, andinteraction-determined, and concluded thatthe interaction theory did a better job of ex-plaining the causes of resistance to a system.

Some other studies also followed an explana-tory strategy by testing hypotheses derivedfrom a single theory (or a priori reasoning).White [55] examined the influence of the cog-nitive style composition of a project team on


Table 3. An Illustrative Categorization of a Sample of Case Studies

OIson [37]

Markus [32]

_Pyburn [39]

Dutton [12]

Hirschheim [20]

White [55]

Ives and Olson [21]

Keen, Bronsema,and Zuboff [24]

Fulk and Dutton [15]

Franz and Robey[14]

Data CollectionTheme Research Thrust Sample Selection Units of Analysis Method

Centralization/decentralizationof system development

Exploration 2 IS groups, differing on the Groupdegree of centralization ofsystem development

Interview andquestionnaire

Implementation Explanation 2 plants within same company Organizational Unspecified, inferreddiffering on impleme_ntation subunit to be interviewedsuccess

IS strategic planning Exploration 8 companies chosen based on Organizations Interview andsimilar characteristics questionnaire

Implementation of models in Exploration andthe public sector explanation

1 U.S. city Organization Multiple sources, highdegree of triangulation

Participative systems design Exploration 20 individuals from 8 organiza-tions with experience inparticipative design

Individuals andorganizations

Interviews

Cognitive styles of MIS Explanationproject teams

2 MIS project teams in thesame company differing in thecognitive style composition oftheir members

Groups Interviews

Nature of IS manager’s job Exploration 6 IS managers chosen basedon several common criteria

Individuals Observations of eachmanager for 3-4 days

Implementation of commonsystems in an internationalbanking setting

Exploration andexplanation

Banks in 9 countries (part ofan international bank) selectedon a number of criteria

Organizational Interviewssubunits

Effect of videoconferencingon organizationalcommunication

Exploration 2 organizations, new user,one localized experienced user

Organization Interview

User-led system design Exploration andexplanation

1 IS development team Group Longitudinal andmultiple sources

Level ofDescriptionAbout Units

High

High

Low

Very High

Low

Low

High, but ataggregate level

High

High

High


its performance, and Schonberger [45] testedthe hypothesis that organization strategieswill influence the structure of the IS organiza-tion. Although Schonberger clearly statedthat he first observed two organizations withdifferent IS department structures and thengot interested in testing the influence ofstrategy on structure based on Chandler’s [7]theory, White appeared to be more ambigu-ous. It is not clear if she first observed two pro-ject teams with different cognitive styles anddecided to test a hypothesis, or if she intend-ed to test a hypothesis and then searched foran ideal sample of project teams with differentcompositions. However, we do not wish to sin-gle out her work for criticism since this wastypical of the case studies in our sample.

In only one instance (Keen, et al., [24]) werethe authors’ stated objectives both the explo-ration and testing of the explanatory power oftheories. Two organizational change para-digms (the Lewin-Schein and Kolb-Frohmanmodels) were used for theory testing. Theseprovided a priori explanations of major factorsthat influence the success or failure of im-plementation efforts. Theory testing was doneboth by goodness of fit and by counter-examples, i.e., by the prediction and discon-firmation strategies suggested by Bonoma[5].

In summary, we found that the IS caseresearchers we surveyed did not provideclear descriptions of where their topics fit inthe knowledge building process. They thusdid not justify their research purpose (drift,description, exploration and explanation) anddid not allow readers to judge their work on amore informed basis.

Unit of Analysis and Site SelectionThe unit of analysis for a case study, and con-sequently for the selection of a particular siteto study, was not provided in many of the pub-lished works. This is a problem consistentwith the lack of clear research objectives dis-cussed above, and probably an outcome of it.Several of the cases examined a single com-pany or a subunit. Since these typically didnot represent critical, unique or revelatorycases, they were presumably chosen basedon availability and evidently the researchers’goals were to conduct exploratory casestudies.

Markus’ [32] study previously described was asingle-case sample containing two embed-ded units of analysis (two plants within thesame division). It was rare because of theunique outcomes of the implementation pro-cess (one plant had a success, the other a fail-ure with the same computerized system) andthe unusual insights it offered.

Based on a large number of criteria, Keen, etal. [24] selected nine sites (countries) affiliat-ed with an international bank. The criteria in-cluded IS development strategy, pace, focus,perceived ease of implementation, geograph-ical dispersion, the size of the country, andthe complexity of the operation. Two of thesites, which differed the most in terms ofcriteria, were selected for indepth analysis.Even though this study appeared to be asingle-case, embedded unit analysis, it couldbe considered a multiple-case design, due tothe decentralized nature of the sites.

A study by Ein-Dor and Segev [13] measuredthe relationship between the success of an ISand the perceived importance and invest-ment in that system. They examined 10 sub-systems supported by a logistics IS in a largeorganization in Israel. This was an embeddedcase study design in which the researchersconducted quantitative analysis of a largenumber of subunits. However, a problem ofembedded designs that Yin [56] mentionedand this study appeared to have, is that itmakes the subunits the sole focus of the studyand ignores the context--the characteristicsof the organization as a whole. There was lessfocus on the context and more focus on a fewaspects of the subunits. Thus, this studyseemed to fall somewhere between a caseand a field study, but had the strengths ofneither.

The sites for some of the multiple-case re-search studies were chosen based on the .study’s research theme and objective. For ex-ample, to measure the effects of centraliza-tion, Olson [37] chose two sites--one in whichthe system development function was cen-tralized and one in which it was decentralized.To examine the influence of strategy on struc-ture, Schonberger [45] observed two organi-zations that had different IS department struc-tures. The sites were chosen partly becausethey differed in terms of the outcome variableof interest, and partly on an opportunistic ba-sis. Similarly, in their paper on factors affect-



ing user involvement in DSS development,Mann and Watson [31] presented three casesthat demonstrated substantially differentdegrees of user involvement. However, thereader is not told whether .the three caseswere chosen from a larger sample because oftheir unique characteristics.

In other instances, the researcher sought ahomogeneous sample with as many similarcharacteristics as possible [39]. However, aswe mentioned in our discussion of Pyburn’sstudy [39], it was not obvious why these par-ticular characteristics were chosen. In anycase, the investigator did not know in ad-vance what independent variables wouldprove to be important. The idea was to mini-mize the extraneous variables so that if signi-ficant factors were indeed found, there wouldbe a high degree of confidence that only thosefactors caused the observed differences. Incontrast to Pyburn’s approach, Fulk and Dut-ton [15], in their study of the impact ofvideoconferencing on organizational commu-nications, chose two companies with differentcharacteristics. One was a large aerospacecompany with offices on the east and westcoasts that had recently started usingvideoconferencing. The second was a largewestern public utility with several years of rou-tine utilization.

None of the multiple-case studies clearly stat-ed the site selection objectives, i.e., whetherthe investigator pursued a literal or a theoreti-cal replication. Nevertheless, in most in-stances, the rationale for choosing a particu-lar site combination was stated.

Sometimes the reasons for site selectionwere not easy to infer. For example, Robey(1983) studied eight companies located different countries. Even though the samplesappeared to be from differing organizations,the researcher did not explicitly state whethera literal or theoretical replication was the goal.In one study [28] the three cases were chosenfor a variety of disparate reasons. One com-pany was chosen because one of the authorswas involved in the implementation of a sys-tem there. In the second company, 2 out of 17branches were selected because the imple-mentation of a system went smoothly in oneand was "less successful" in the other. Thethird company was included in the sample be-cause it had a recent, major shift from a man-ual system to a computerized system and it

was self-contained. It thus appeared that thefirst case was chosen opportunistically, thesecond due to the observed outcomes, andthe third based on a possible causal factor. Inthis particular study, it was difficult to inferwhether the authors’ goal was a literal or theo-retical replication, or whether the cases werechosen for exploratory or illustrative reasons.

Authors in our sample did not indicate if theircase studies were part of systematic/pro-grammatic research plans. Most seemed tobe stand-alone, one-shot studies. Only onecase study was a triangulation in that it was afollow-up to a survey [15]. Another case studywas part of a large-scale effort to study vari-ous aspects of IS and their organizational im-pact, but the linkage to the larger study wasnot described in the paper [41].

Data CollectionIn about half of the case studies, the datawere collected by multiple means; the otherhalf relied solely on interviews. In two casestudies, though, the data collection methodwas not specified at all. We believe that aclear description of data sources and the waythey contribute to the findings of the researchis an important aspect of the reliability and va-lidity of the findings. Yin’s [56] suggestionsabout describing the case study protocol andhaving a case study database can serve asimportant guidelines.

Almost all of the studies used interviews fordata collection. The interview questions wererarely specified and, when they were, it was ina very general form. Hirschheim [20], a not-able exception, included the questions usedin his semistructured interview in an appen-dix. In some studies, researchers interviewedindividuals who had different perspectives ona given process, e.g., managers, users, anddesigners. Sometimes the researchers men-tioned that they used documents and obser-vations, but they did not provide any more de-tail about them.

A few studies of implementation in the publicsector collected data from a large number ofsources [25, 27]. These studies were similarto the Dutton [12] study in the richness of theirdescriptions and data sources. The onlystudy that used a longitudinal methodology isalso an exemplary effort of data collection[14]. To study user involvement in IS design,



the investigators collected data over a two-year period. In order to achieve triangulation,data were gathered through questionnaires,critical incident files, unstructured interviews,documents and memoranda, observations atmeetings and tape recordings. In this way,they attempted to get both an objective viewof events and the subjective interpretations ofparticipants.

Keen, et al. [24] used an outside expert withno knowledge of the system being implement-ed to interview participants with whom the re-search team (authors) had already talked andwho were in conflict. This was done to reduceany bias arising from the expectations of theresearchers. If other studies used such tech-niques to increase the objectivity of data col-lection, no mention was made of them.

Case data in the research surveyed wasmostly qualitative. A few studies includedquantitative observations, mostly in the formof questionnaire data [12, 22, 37]. In general,however, the degree of detail about data col-lection methods was not very revealing, asubstantial problem with most of the casestudies observed.

Concluding CommentsOur intent in this article was to clarify thenature of the case research method, explainwhy it might be utilized in IS studies, sur-vey its uses in IS research within the lastfive years, and offer some suggestions forimprovement.

The case research strategy has mostly beenused for exploration and hypothesis genera-tion. This is a legitimate way of adding to thebody of knowledge in the IS field. Explorationis, however, not the only reason for applyingthe case method. As discussed earlier, vari-ous authors have suggested the use of casesfor providing explanation and for testinghypotheses.

No research strategy is better than all others.Unlike some of the contributors to Mumford,et al. [36], we do not advocate exclusive reli-ance on case/action research methods. Theselection of a research strategy depends onthe current knowledge of a topic and the na-ture of the topic, among other factors. Thecase strategy is particularly well-suited to ISresearch because the technology is relatively

new and interest has shifted to organizationalrather than technical issues. For example,case studies have been helpful in identifyingthe causal chain that led to the success or fail-ure of an information system by revealing inchronological fashion the various actors andevents that influenced the final outcome.

Several current topics within IS research areamenable to the case study approach. Theuse of an expert system for management sup-port isone such area. Since expert systemsare just beginning to be introduced into or-ganizations, a case study of companies thatare rather far along in the use of such a sys-tem would provide valuable insights. Sviokla[48] has examined how an expert system thatassists financial planners affects their jobsand the quality of the plans they produce. Thisrevelatory use of the case method can pro-vide hypotheses about the impact of expertsystems technology on organizations. Thesehypotheses can then be tested using anotherresearch method, like a field experiment.

The relationship between information tech-nology and corporate strategy is another areathat could be explored further using a struc-tured program of multiple case studies. Fromthe case studies conducted to date, there isevidence that some companies use informa-tion technology more effectively as a strategicweapon than others [38]. A systematic studyof several companies within one industrycould provide important insights into whysome companies use information technologymore successfully than others.

One of the more difficult decisions thatresearchers must make is to determine whenfurther case studies are needed in an emerg-ing research area. For example, in the studyof information technology and corporatestrategy, Bakos and Treacy [2, p. 107] believethat:

As this area of research matures, thereis an increasing need to move beyondframeworks and toward explanatorymodels of the underlying phenomena.This type of research will allow us tobuild a cumulative tradition and to makenormative statements to guidemanagerial actions.

In the IS area, there is some merit to both theview that more case research is needed and



the view that quantitative techniques shouldbe employed. Case studies can provide theorganizational context for the study of therelationship between strategy and informa-tion technology. This is important even in thisstage of the research. A formal model canprovide significant insights into the morequantitative aspects of the issue, such as theeconomic impact of investing in informationtechnology.

Recently there have been calls for a moredetailed understanding of phenomena in theIS area. There is an interesting parallel be-tween our call for higher quality case researchand the argument presented by Todd andBenbasat [49] for increased use of protocolanalysis in the study of decision support sys-tems. The use of protocol analysis in ISstudies has increased within the last fewyears because many researchers believe thatthey have to open the "black box." They wantto examine processes in order to better un-derstand the effects of information technolo-gy on the people who work with it, and their in-fluence on the technology. The intent is to car-ry out a detailed, indepth examination of asmall number of individuals. In a similar way,case research obtains detailed data aboutone or a few units.

The key difference between the case re-search method and protocol analysis is that,while both methods require little a priori speci-fication of dependent variables and their mea-surement allow the investigator to use a highdegree of discretion in structuring and inter-preting the data, the case method also offerslittle control over the antecedent condition,e.g., independent variables.

In this article we identified a number of prob-lems that were common to most of the caseresearch studies in the sample. Some ofthese might be alleviated by asking theauthors to provide more information abouttheir research objectives and research plans.However, it appears to us that, in many in-stances, the investigators had not consideredsome of the methodological issues. In gener-al, the objectives of the researchers were notclearly specified. The reasons for selectingsingle-case versus multiple-case designswere not explained and the choice of particu-lar sites was not tied to the design approach.In many cases the data collection method wasambiguous and details were not provided.

The use of triangulation to increase reliabilitywas rare.

Yin states that the reader of the case studyshould be able to follow the derivation of anyevidence from initial research questions tothe conclusions of the study. This chain of evi-dence will improve the reliability of the data.The point is that a case study should be morethan an exercise in storytelling or an opinionpiece; it should adhere to certain rules ofprocedure, as described in Yin [56] and dis-cussed here. Although these procedures arenot as detailed as one would find in a fieldstudy, survey or experiment, they are criticalto allowing readers to assess the reliabilityand validity of the study’s findings. In this way,IS researchers can better contribute to theknowledge building process and IS case re-search will come into its own.

AcknowledgementProfessor Benbasat was a Marvin Bower Fel-low at the Harvard Business School whileworking on this paper. He thanks the Fellow-ship program for its support during the1985-86 academic year.

References1. Allison, G.T. Essence of Decisionmaking:

Explaining the Cuban Missile Crisis, Little,Brown, Boston, Massachusetts 1971.

2. Bakos, J.Y. and Treacy, M.E. "Informa-tion Technology and Corporate Strategy:A Research Perspective," MIS Quarterly,Volume 10, Number 2, June 1986, pp.107-119.

3. Benbasat, I. "An Analysis of ResearchMethodologies," in The Information Sys-tems Research Challenge, F. WarrenMcFarlan (ed.), Harvard Business SchoolPress, Boston, Massachusetts, 1984, pp.47-85.

4. Bonoma, T.V. "A Case Study in Case Re-search: Marketing Implementation,"Working Paper 9-585-142, HarvardUniversity Graduate School of BusinessAdministration, Boston, Massachusetts,May 1983.

5. Bonoma, T.V."Case Research in Market-ing: Opportunities, Problems, and a Pro-cess," Journal of Marketing Research,Volume 22, Number 2, May 1985, pp.



199-208.6. Bouchard, T.J., Jr."Field Research Meth-

ods," in Industrial and Organizational Psy-chology, M.D. Dunnette (ed.), RandMcNally, Chicago, Illinois, 1976.

7. Chandler, A.D., Jr. Strategy and Struc-ture: Chapters in the History of the Ameri-can Industrial Enterprise, MIT Press,Cambridge, Massachusetts, 1962.Checkland, P. Systems Thinking, Sys-tems Practice, John Wiley and Sons, NewYork, New York, 1981.Christenson, C. "Proposals for a Programof Empirical Research into the Propertiesof Triangles," Decision Sciences, Volume7, Number 3, October 1976, pp. 631-648.

10. Cook, T.D. and Campbell, D.T. Quasi-experimentation: Design and Analysis Is-sues for Field Settings, Rand McNally,Chicago, Illinois, 1979.

11. Dickson, G.W. and Janson, M. "The Fail-ure of a DSS for Energy Conservation: ATechnical Perspective," Systems, Objec-tives, Solutions, Volume 4, Number 2,April 1984, pp. 69-79.

12. Dutton, W.H. "The Rejection of an Inno-vation: The Political Environment of aComputer-Based Model," Systems, Ob-jectives, Solutions, Volume 1, Number 4,November 1981, pp. 179-201.

13. Ein-Dor, P. and Segev, E. "Perceived Im-portance, Investment. and Success ofMIS, or the MIS Zoo," Systems, Objec-tives, Solutions, Volume 4, Number 2,April 1984, pp. 61-67.

14. Franz, C.R. and Robey, D. "An Investiga-tion of User-Led System Design: Rationaland Political Perspectives," Communica-tions of the ACM, Volume 27, Number 12,December 1984, pp. 1202-1217.

15. Fulk, J. and Dutton, W. "Videoconferenc-ing as an Organizational Information Sys-tem: Assessing. the Role of ElectronicMeetings," Systems, Objectives, Solu-tions, Volume 4, Number 2, April 1984,pp. 105-118.

16. Gibson, C.F. "A Methodology for Im-plementation Research," in ImplementingOperations Research/Management Sci-nce, R.L. Schultz and D.P. Slevin (eds.),American Elsevier, New York, New York,1975.

17. Ginzberg, M. "A Prescriptive Model forSystem Implementation," Systems, Ob-jectives, Solutions, Volume 1, Number 1,January 1981, pp. 33-46.

i8. Hamilton, S. and Ives, B. "MIS ResearchStrategies," Information and Manage-ment, Volume 5, Number 6, December1982, pp. 339-347.

19. Hersen, M. and Barlow, D.H., Single-Case Experimental Designs: Strategiesfor Studying Behavior, Pergamon Press,New York, New York, 1976.

20. Hirschheim, R.A. "User Experience Withan Assessment of Participative SystemsDesign," MIS Quarterly, Volume 9, Num-ber 4, December 1985, pp. 295-304.

21. Ives, B. and Olson, M.H. "Manager orTechnician? The Nature of the Informa-tion Systems Manager’s Job," MIS Quar-terly, Volume 5, Number 4, December1981, pp. 49-63.

22. Jick, T.D. "Mixing Qualitative and Quan-titative Methods: Triangulation in Action,"Administrative Sciences Quarterly, Vol-ume 24, Number 4, December 1979, pp.602-611.

23. Kaplan, R.S. "The Role of Empirical Re-search in Management Accounting,"Working Paper 9-785-001, Division ofResearch, Harvard Business School,Boston, Massachusetts, 1985.

24. Keen, P.W., Bronsema, G., and Zuboff,S. "Implementing Common Systems:One Organization’s Experience," Sys-tems, Objectives, Solutions, Volume 2,Number 3, August 1982, pp. 125-142.

25. King, J. "Successful Implementation ofLarge-Scale DSS," Systems, Objectives,Solutions, Volume 3, Number 4, Novem-ber 1983, pp. 183-205.

26. Kling, R. and lacono, S. "The Control ofInformation Systems Developments AfterImplementation," Communications of theACM, Volume 27, Number 12, December1984, pp. 1218-1226.

27. Kraemer, K.L. "The Politics of Model Im-plementation," Systems, Objectives, So-lutions, Volume 1, Number 4, November1981, pp. 161-178.

28. Land, F., Detjejarvwat, N. and Smith, C."Factors Affecting Social Controls: TheReasons and Values--Part I1," Systems,Objectives, Solutions, Volume 3, Number4, November 1983, pp. 207-226.

29. Lawrence, P.L. and Lorsch, J.W. Organi-zation and Environment, Harvard Busi-ness School Press, Boston, Mas-sachusetts, 1986.

30. MacKenzie, K.D. and Howe, R. "Para-digm Development in the Social Sci-

384 MIS Quarterly/September 1987


ences: A Proposed Research Strategy,"Academy of Management Review, Vol-ume 3, Number 1, 1978, pp. 7-23.

31. Mann, R.I. and Watson, H.J. "A Contin-gency Model for User Involvement inDSS Development," MIS Quarterly, Vol-ume 8, Number 1, March 1984, pp.27-38.

32. Markus, M.L. "Implementation Politics:Top Management Support and User In-volvement," Systems, Objectives, Solu-tions, Volume 1, Number 4, November1981, pp. 203-215.

33. Markus, M.L. "Power, Politics, and MISImplementation," Communications of theACM, Volume 26, Number 6, June 1983,pp. 430-444.

34. Morton, M.S. Management Decision Sys-tems: Computer-Based Support for Deci-sion Making, Harvard Business School,Division of Research, Boston, Mas-sachusetts, 1971.

35. Mumford, E. "Participative Systems De-sign: Structure and Method," Systems,Objectives, Solutions, Volume 1, Number1, January 1981, pp. 5-19.

36. Mumford, E., Hirschheim, R., Fitzgerald,G. and Wood-Harper, T. Research Meth-ods in Information Systems, North Hol-land, New York, New York, 1985.

37. Olson, M. "User Involvement and Decen-tralization of the Development Function:A Comparison of Two Case Studies,"Systems, Objectives, Solutions, Volume1, Number 2, April 1981, pp. 59-69.

38. Parsons, G.L. "Information Technology:A New Competitive Weapon," SloanManagement Review, Volume 25, Num-ber 1, Fall 1983, pp. 3-14.

39. Pyburn, P.J. "Linking the MIS Plan withCorporate Strategy: An ExploratoryStudy," MIS Quarterly, Volume 7, Number2, June 1983, pp. 1-14.

40. Rivard, S. and Huff, S.L. "User Devel-oped Applications: Evaluation of SuccessFrom the DP Department Perspeqtive,"MIS Quarterly, Volume 8, Number 1,March 1984, pp. 39-50.

41. Robey, D. "Computer Information Sys-tems and Organization Structure," Com-munications of the ACM, Volume 24,Number 10, October 1981, pp. 679-686.

42. Robey, D. "Information Systems and Or-ganizational Change: A ComparativeCase Study," Systems, Objectives, Solu-

tions, Volume 3, Number 3, August 1983,pp. 143-154.

43. Rockart, J.F. and Flannery, L.S. "TheManagement of End-User Computing,"Communications of the ACM, Volume 26,Number 10, October 1983, pp. 776-784.

44. Roethlisberger, F.J. The ElusivePhenomena, Harvard Business School,Division of Research, Boston, Mas-sachusetts 1977.

45. Schonberger, R.J. "Strategy and Struc-ture: A Tale of Two Information SystemsDepartments," Systems, Objectives, So-lutions, Volume 1, Number 2, April 1981,pp. 71-77.

46. Stone, E. Research Methods in Organiza-tional Behavior, Scott, Foresman andCompany, Glenview, Illinois, 1978.

47. Susman, G.L. and Evered, R.D. "An As-sessment of the Scientific Merits of Ac-tion Research," Administrative SciencesQuarterly, Volume 23, Number 4, Decem-ber 1978, pp. 582-603.

48. Sviokla, J.S. "Planpower, XCON, andMudman: An In-Depth Analysis intoThree Commercial Expert Systems inUse," unpublished doctoral dissertation,Harvard Business School, Boston, Mas-sachusetts, 1986.

49. Todd, P. and Benbasat, I. "Process Trac-ing Methods in Decision Support Sys-tems Research: Exploring the BlackBox," Working Paper No. 1140, Faculty ofCommerce and Business Administration,The University of British Columbia, Van-couver, British Columbia, Canada, Janu-ary 1986.

50. Towl, A.R., To Study Administrations byCases, Harvard Business School, Bos-ton, 1969.

51. Van Maanen, J. "Introduction," in Varie-ties of Quafitative Research, J. Van Maa-nen (ed.), Sage Publications, BeverlyHills, California 1982, pp. 11-29.

52. Vogel, D.R. and Wetherbe, J.C. "MIS Re-search: A Profile of Leading Journals andUniversities," Database, Volume 16,Number 1, Fall 1984, pp. 3-14.

53. Webb, E., Campbell, D., Schwartz, R.and Sechrest, L. Unobtrusive Measures:Nonreactive Research in the SocialSciences, Rand McNally, Chicago, Il-linois, 1966.

54. Weick, K.E. "Theoretical Assumptionsand Research Methodology Selection" in


Information Research Planning

55.

56.

The Information Systems Research Chal-lenge, F. Warren McFarlan (ed.), HarvardBusiness School Press, Boston, Mas-sachusetts, 1984, pp. 111-132.White, K.B. "MIS Project Teams: An In-

vestigation of Cognitive Style Implica-tions," MIS Quarterly, Volume 8, Number2, June 1984, pp. 95-101.Yin, R.K. Case Study Research, Design

and Methods, Sage Publications, BeverlyHills, California, 1984.

About The AuthorsIzak Benbasat is Professor and Chair of theMIS Division at the University of British Co-lumbia. He received his Ph.D. in MIS from theUniversity of Minnesota in 1974. During the1985-86 academic year he was a Marv-in Bower Fellow at the Harvard BusinessSchool. He is currently studying the topic ofknowledge acquisition for expert systems de-velopment, supported by a grant from theNatural Sciences and Engineering Council ofCanada.

David Goldstein is an Assistant Professor ofBusiness Administration at Harvard Universi-ty. He received his Ph.D. in Management

Science from Sloan School of Management atM.I.T. Goldstein arrived at the Harvard Busi-ness School in 1984 and has taught control inthe MBA program and MIS in the MBA andExecutive Education programs. His researchinterests are on the impact of informationtechnology on the effectiveness of managers,the management of the systems developmentfunction, and the use of qualitative methods inMIS research.

Melissa Mead joined the information systemsfaculty at the Harvard Business School in1985. Her teaching at Harvard includescourses in both control and management in-formation systems, in addition to knowledge-based systems and doctoral seminars on re-search methods and topics. Professor Meadcompleted a Ph.D. in Management Scienceand Information Systems at The University ofTexas at Austin in 1985. Prior to her arrival atHarvard, Professor Mead worked as a consul-tant for Analysis, Research & ComputationInc. Her research interests include the incor-poration of information technology as an ele-ment of business strategy, particularly theuse of information technology by generalmanagers during times of transition to fosterand sustain changes in strategic direction.


the case research strategy in studies of information systems · tive technique, the case research...

Documents