Building-Theories-from-Case-Study-Research

BuildingTheoriesfromCaseStudyResearchKathleenM.EisenhardtTheAcademyofManagementReview,Vol.14,No.4.(Oct.,1989),pp.532-550.StableURL:=0363-7425%28198910%2914%3A4%3C532%3ABTFCSR%3E2.0.CO%3B2-RTheAcademyofManagementReviewiscurrentlypublishedbyAcademyofManagement.YouruseoftheJSTORarchiveindicatesyouracceptanceofJSTOR'sTermsandConditionsofUse,availableat://@jstor.org.:282007FAcademyofManagementReview,1989,Vol14,No4,532-550BuildingTheoriesfromCaseStudyResearchKATHLEENM.EISENHARDTStanfordUniversityThispaperdescribestheprocessofinductingtheoryusingcasestud-ies-fromspecifyingtheresearchquestionstoreachingclosure.Somefeaturesoftheprocess,suchasproblemdefinitionandcon-structvalidation,aresimilartohypothesis-testingresearch.Others,suchaswithin-caseanalysisandreplicationlogic,areuniquetotheinductive,case-orientedprocess.Overall,theprocessdescribedhereishighlyiterativeandtightlylinkedtodata.Thisresearchapproachisespeciallyappropriateinnewtopicareas.Theresultanttheoryisoftennovel,testable,andempiricallyvalid.Finally,framebreakinginsights,thetestsofgoodtheoryfe.g.,parsimony,logicalcoherence),andconvincinggroundingintheevidencearethekeycriteriaforevaluatingthistypeofresearch.Developmentoftheoryisacentralactivityinlackofclarityabouttheprocessofactuallyorganizationalresearch.Traditionally,authorsbuildingtheoryfromcases,especiallyregard-havedevelopedtheorybycombiningobserva-ingthecentralinductiveprocessandtheroleoftionsfrompreviousliterature,commonsense,literature.GlaserandStrauss(1967)andmoreandexperience.However,thetietoactualdatarecentlyStrauss(1987)haveoutlinedpiecesofhasoftenbeentenuous(Perrow,1986;Pfeffer,theprocess,buttheirsisaprescribedformula,1982).Yet,asGlaserandStrauss(1967)argue,itandnewideashaveemergedfrommethodolo-istheintimateconnectionwithempiricalrealitygists(e.g.,Yin,1984;Miles&Huberman,1984)thatpermitsthedevelopmentofatestable,rel-andresearchersconductingthistypeofre-evant,andvalidtheory.search(e.g.,Gersick,1988;Harris&Sutton,Thispaperdescribesbuildingtheoriesfrom1986;Eisenhardt&Bourgeois,1988).Also,itap-casestudies.Severalaspectsofthisprocessarepearsthatnoonehasexplicitlyexaminedwhendiscussedintheliterature.Forexample,Glaserthistheory-buildingapproachislikelytobeandStrauss(1967)detailedacomparativefruitfulandwhatitsstrengthsandweaknessesmethodfordevelopinggroundedtheory,Yinmaybe.(1981,1984)describedthedesignofcasestudyThispaperattemptstomaketwocontributionsresearch,andMilesandHuberman(1984)codi-totheliterature.Thefirstisaroadmapforbuild-fiedaseriesofproceduresforanalyzingquali-ingtheoriesfromcasestudyresearch.Thistativedata.However,confusionsurroundstheroadmapsynthesizespreviousworkonqualita-distinctionsamongqualitativedata,inductivetivemethods(e.g.,Miles&Huberman,1984),thelogic,andcasestudyresearch.Also,thereisadesignofcasestudyresearch(e.g.,Yin,1981,1984),andgroundedtheorybuilding(e.g.,Gla-searchthanhasexistedinthepast.Thisframe-ser&Strauss,1967)andextendsthatworkinworkissummarizedinTable1.areassuchasapriorispecificationofconstructs,Thesecondcontributionispositioningtheorytriangulationofmultipleinvestigators,within-buildingfromcasestudiesintothelargercontextcaseandcross-caseanalyses,andtheroleofofsocialscienceresearch.Forexample,thepa-existingliterature.Theresultisamorenearlyperexploresstrengthsandweaknessesoftheorycompleteroadmapforexecutingthistypeofre-buildingfromcasestudies,situationsinwhichitTable1ProcessofBuildingTheoryfromCaseStudyResearchStepActivityReasonGettlngStartedDefinitionofresearchquestionFocuseseffortsPosslblyaprlorlconstructsProvldesbettergroundingofconstructmeasuresNelthertheorynorhypothesesRetainstheoreticalflexibilitySelectingCasesSpecifiedpopulationConstrainsextraneousvarlationandsharpensexternalvalidltyTheoretical,notrandom,samplingFocuseseffortsontheoreticallyusefulcases-i.e.,thosethatreplicateorextendtheorybyfilllngconceptualcategoriesCraftlngInstrumentsMultipledatacollectionmethodsStrengthensgroundingoftheorybyandProtocolstriangulationofevldenceQualitativeandquantitativedatacombinedSynerglstlcvlewofevldenceMultiplelnvestlgatorsFostersdivergentperspectivesandstrengthensgroundingEnteringtheFleldOverlapdatacollectionandanalysis,SpeedsanalysesandrevealshelpfulIncludingfieldnotesadjustmentstodatacollectionFlexibleandopportunisticdatacollectionAllowsinv