Information Technology Diffusion A Review of Empir

InformationTechnologyDiffusion:AReviewofEmpiricalResearchRobertG.FichmanMITSloanSchoolofManagement50MemorialDrive,E53-314Cambridge,MA02139Phone:[phone]Fax:[phone]June1992Keywords:InformationTechnology;InnovationDiffusion;TechnologyAdoptionAcknowledgements:TheauthorwishestothankProfs.ChrisKemerer,WandaOrlikowski,andPeterWeill,andanonymousreviewersformanyhelpfulcommentsonpriorversionsofthismanuscript.AbstractInnovationdiffusiontheoryprovidesausefulperspectiveononeofthemostpersistentlychallengingtopicsintheITfield,namely,howtoimprovetechnologyassessment,adoptionandimplementation.Forthisreason,diffusionisgrowinginpopularityasareferencetheoryforempiricalstudiesofinformationtechnologyadoptionanddiffusion,althoughnocomprehensivereviewofthisbodyofworkhasbeenpublishedtodate.Thispaperpresentstheresultsofacriticalreviewofeighteenempiricalstudiespublishedduringtheperiod1981-1991.Conclusiveresultsweremostlikelywhentheadoptioncontextcloselymatchedthecontextsinwhichclassicaldiffusiontheorywasdeveloped(forexample,individualadoptionofpersonal-usetechnologies),orwhenresearchersextendeddiffusiontheorytoaccountfornewfactorsspecifictotheITadoptioncontextunderstudy.Basedonclassicaldiffusiontheoryandotherrecentconceptualwork,aframeworkisdevelopedtoguidefutureresearchinITdiffusion.Theframeworkmapstwoclassesoftechnology(onesthatconformcloselytoclassicaldiffusionassumptionsversusonesthatdonot)againstlocusofadoption(individualversusorganizational),resultinginfourITadoptioncontexts.Foreachadoptioncontext,variablesimpactingadoptionanddiffusionareidentified.Additionally,directionsforfutureresearcharediscussed.11.IntroductionInnovationdiffusiontheoryprovideswell-developedconceptsandalargebodyofempiricalresultsapplicabletothestudyoftechnologyevaluation,adoptionandimplementation.Diffusiontheoryprovidestools,bothquantitativeandqualitative,forassessingthelikelyrateofdiffusionofatechnology,andadditionally,identifiesnumerousfactorsthatfacilitateorhindertechnologyadoptionandimplementation.Thesefactorsincludecharacteristicsofthetechnology,characteristicsofadopters,andthemeansbywhichadopterslearnaboutandarepersuadedtoadoptthetechnology(Rogers1983).Itisnotsurprisingthen,thatinnovationdiffusionisbecominganincreasinglypopularreferencetheoryforempiricalstudiesofinformationtechnologies(IT).Asaborrowedtheory,innovationdiffusionprovidestheadvantageofarichcumulativetradition.Yet,whenborrowingtheory,researchersmusttakecaretoensurethatthecontexttowhichthetheoryisbeingappliedmatcheswellwiththecontextinwhichthetheorywasdeveloped,oralternatively,totailorthetheorytoaccountforcontextualdifferences.Muchofdiffusiontheorywasdevelopedinthecontextofadoptersmakingvoluntarydecisionstoacceptorrejectaninnovationbasedonthebenefitstheyexpecttoaccruefromtheirownindependentuseofthetechnology.Yet,adoptionofITmaybeencouragedbymanagement(Leonard-BartonandDeschamps1988)orevenmandated(MooreandBenbasat1991).Adopters,ratherthanmakingabinarydecisiontoadoptorreject,maychoosedifferinglevelsofITuse(BayerandMelone1989).Inaddition,theadoptiondecisionofindividualsororganizationsmaydependonthedynamicsofcommunity-widelevelsofadoption(i.e.,whethercriticalmasshasbeenestablished)becauseofnetworkexternalities(KatzandShapiro1986;Markus1987).ThesesortsofcomplicatingfactorsarequitecommoninthecontextofITadoption;hence,theopportunitiestoapplyclassicaldiffusionasismayberareindeed.Nocriticalreviewexiststhatfocusesspecificallyontheapplicationofdiffusiontheorytotheadoptionofinformationtechnologies.ThispaperpresentstheresultsofareviewandanalysisofeighteenpublishedempiricalstudiesofITadoptionanddiffusionfromtheperiod1981to1991withafocusonidentifyinginstanceswherethe2adoptioncontextcloselymatchesthecontextinwhichclassicaldiffusiontheorywasdeveloped.Toassistinthistask,aframeworkisprovidedthatdefinesfouradoptioncontexts,oneofwhichcloselyagreeswiththeassumptionsofclassicaldiffusionandthreeofwhichreflectoneormoreimportantdivergencesfromclassicaldiffusionassumptions.Aswouldbeexpected,strongresultsweremostlikelytobefoundininstanceswheretheadoptioncontextwasagoodmatchwithclassicaldiffusionassumptions,orwhenadditionalvariablessuggestedbytheadoptioncontextwereincorporatedintotheanalysis.Theremainingsectionsareorganizedasfollows.Section2brieflydescribesthemainpointsofclassicaldiffusiontheory.Section3describesrecentconceptualworkrelevanttoITdiffusion,andestablishesthefoundationfortheITDiffusionFrameworkpresentedinSection4.Section5presentstheresultsoftheempiricalreviewandsuggestsdirectionsforfutureresearch.Finally,Section6offerssomeconcludingremarks.2.ClassicalDiffusionAninnovationisanyidea,practiceorobjectthatisperceivedasnewbytheadopter.EverettRogers,inawidelycitedwork(1983),providesasynthesisofover3000previousstudiesofadoptionanddiffusion.Theresultsofthissynthesisincludenumerousgeneralizationsaboutinnovationdiffusion,i.e.,theprocessbywhichinnovationsspreadthroughpopulationsofpotentialadopters.Amongthemorewell-establishedgeneralizationsare:1)Innovationspossesscertaincharacteristics(i.e.,relativeadvantag