Perceived Risk and Consumer Behavior: a Critical Review

Cox (1967a) in his initial elaboration of Bauer's conceptualization states that it is often necessary to infer the presence of perceived risk since ... LearnmoreaboutourTermsofServiceandPrivacyPolicy Dismiss PerceivedRiskandConsumerBehavior:aCriticalReview Citation: IvanRoss(1975),"PerceivedRiskandConsumerBehavior:aCriticalReview",inNA-AdvancesinConsumerResearchVolume02,eds.MaryJaneSchlinger,AnnAbor,MI:AssociationforConsumerResearch,Pages:1-20. AdvancesinConsumerResearchVolume2,1975    Pages1-20PERCEIVEDRISKANDCONSUMERBEHAVIOR:ACRITICALREVIEWIvanRoss,UniversityofMinnesotaTheempiricalresearchrelatingperceivedrisktoconsumerbehaviorissummarized.Theliteraturerevealsthatperceivedriskhasbeenstudiedinrelationshiptoinformationacquisitionandprocessingconstructssuchasword-of-mouthbehaviorandopinion-leadership,aswellastoovertconsumerbehaviorssuchasnewproductadoption,store/brandloyalty,andmodesofshopping.Recentresearchhasbeenconcentratedonthestudyofrelationshipsbetweenspecifickindsorcomponentsofperceivedriskorriskconsequencesandthespecificrelieversorreducersofthesecomponents.Therevieweroffersacritiqueofresearchonperceivedriskandsuggestsdirectionforfutureresearch.INTRODUCTIONWhenBauer(1960)firstproposedthatconsumerbehaviorcouldbeviewedasaninstanceofrisktaking,hemodestlyhopedthatthe"fad"hewasprobablyintroducingwould"atleastsurvivethroughinfancy"(p.23).Afterfourteenyearsthereisevidencethattheinfantisfastbecomingadult.Indeed,asthelistofreferencessuggests,recentyearshaveshownadramaticincreaseinpublicationfrequencyofempiricalresearchinthisarea,andcurrentmodelsortheoriesofconsumerbehaviorbroadlyincorporatetheperceivedriskconstruct.Engel,KollatandBlackwell(1973)positionperceivedriskspecificallyinthe"externalsearchandalternativeevaluation"stageofdecision-making(pp.376-380)andgenerallyattributetoitgreatimportance:"Decisionmaking(processes)...occurinordertoreduceperceivedrisktotolerablelevels(p.59)."HowardandSheth(1969)conceptuallydealwiththeconstructundertheirterm,"stimulusambiguity",viewedasa"perceptualconstruct"intheirtheoryofconsumerbehavior(P.30).Thereviewerhasnotfoundtheorganizationoftheempiricalliteratureonperceivedriskforthepurposesofthispaperaneasytasknoronewhichislikelytobeoptimally-satisfyingtosome(hopefully,notmost)readers.Inthefirstplace,perceivedriskhasbeenstudiedinrelationtoaverylargenumberofotherconsumerbehaviorvariables--toolargeanumbertoreviewindetailwithinthespacelimitationsimposed.Andsecondly,themannerinwhichtheconstructhasbeenoperationallyandevenconceptuallydefinedhasvariedsomuchacrossthestudies,thateffortsatsynthesisarehamperedbyquestionsof"arethesetwostudiesreallytalkingaboutthesamething?"Moreoftenthannot,theansweris.no.Afteradiscussionoftheconceptualizationoftheconstruct,thereviewerhaschosentoorganizehissummarybydiscussingthemajorconsumerbehaviorvariablestowhichperceivedriskhasbeenapplied.Some"problematic"areasinthisresearchtraditionandsuggestionsforfutureresearchareattheendofthereviewpaper.CONCEPTUALIZATIONOFPERCEIVEDRISKBauer'sinitialpropositionwasthat,"Consumerbehaviorinvolvesriskinthesensethatanyactionofaconsumerwillproduceconsequenceswhichhecannotanticipatewithanythingapproximatingcertainty,andsomeofwhichatleastarelikelytobeunpleasant"(1960,p.24).Thus,thetwoprimarystructuraldimensionswereuncertaintyandconsequenceswhichmuch,butnotall,subsequentresearchinperceivedriskhasusedinthemeasurementprocedure.Bauerstronglyemphasizesthatheisconcernedonlywithsubjective(perceived)riskandnot"realworld"(objective)risk.ItshouldbenotedthatBauerclearlyviewsperceivedriskasnotonlyrelatedtoconsumers'pre-decisioninformationacquisitionandprocessingactivitybuttopost-decisionprocessesaswell.Hence,hedescribesdissonancetheoryasconcernedwith"...waysinwhichpeoplereduceperceivedriskafterdecisionsaremade.Peoplewillseekoutinformationthatconfirmsthewisdomoftheirdecisions"(p.32).Itwouldhavebeenwellifsomeresearchersintheperceivedriskareahadmorecarefullynotedtheviewofdissonancereductionasriskreductionprocesseswho,asaresultoffailingtodoso,drewequivocalconclusionsfromtheirresearch(e.g.Arndt,1968a;CoxandRich,1964;andSchiffman,1972).Inallthesecasesperceivedriskmeasuresweretakenafterthepurchasehad(orhadnot)occurred,atwhichtimeitwouldbereasonabletoassumethatrisk/dissonancereductionprocesseshadbegun,andhencewouldlikelycontaminatetheirresponsetotheriskmeasure.Indeed,thesestudiesmightbetterhavebeenaddressedtopostpurchasedissonance/riskreductionactivityexplicitlysincewehavenoexamplesintheriskliteratureofsuchstudies.Cox(1967a)inhisinitialelaborationofBauer'sconceptualizationstatesthatitisoftennecessarytoinferthepresenceofperceivedrisksince"...consumersmaybeunableorunwillingtospecifythatasituationconfrontingthemisrisky(p.36)...(thus)...wewillassume,foroperationalpurposes,thatriskis,insomeway,perceivedbyoursubjectsinthosesituationsinwhichtheyactinsuchawayastohandle(e.g.reduce)risk"(p.37).Theamountofperceivedriskisconstruedtobeafunctionof(1?"Theamountthatwouldbelost(i.e.,thatwhichisatstake)iftheconsequencesoftheactwerenotfavorable,and(2)theindividual'ssubjectivefeelingofcertaintythattheconsequenceswillbe-unfavorable"(p.37).Theamountatstake"...isafunctionoftheimportanceormagnitudeofthegoalstobeattained,theseriousnessofthepenaltiesthatmightbeimposedfornonattainment,andtheamountofmeanscommittedtoachievingthegoals"(p.38).Whereascertaintyandconsequencesdeterminetheamountofperceivedrisk"Thenatureoftheriskperceivedshouldbeafunctionofthenatureofthebuyinggoalsinvolved"(p.38).Giventhis"twofactor"viewofriskstructureitthenfollowsthatriskmightbereducedtoa"tolerablelevel"byeitherorboth(1)reducingtheamountatstake(e.g.reducingthatwhichthepersonhopedtogain,reducingthepenaltiesforfailure,andreducingthemeansbywhichthegainwastobemade)and(2)increasingthedegreeofcertaintythatlosswillnotoccur;thatis,becomingmorecertainthatactionconsequenceswouldbefavorable.Whilemostsubsequentresearchhasemployedthesetwodimensionsspecifically(e.g.Cunningham,1967a),othershaveusedavarianttwo-dimensionaldefinitionsuchasuncertaintyandimportance(e.g.Schiffman,1972;Arndt,1968b),andsomeusejustonedimension(e.g.uncertaintyonly,Arndt,1968a).Insomecasesitisdifficulttodistinguishwhetheruncertaintyorconsequencesisbeingmeasured(e.g."howriskyisthepurchaseof____").Bettman(1973)specificallyconceptualizesriskdimensionalityasdifferentfromthatofCox(1967a)andCunningham(1967a)bysubstitutingimportanceforconsequences/dangers."...theriskinherentinabrandchoicesituationwithinaproductclasswilldependuponthedegreetowhichabuyerbelieveshecanconstructareasonabledecisionruleformakingabrandchoice,andtheimportancetohimofmakingasatisfactorychoicewithinthatproductclass"(Bettman,1973,pp.184-185).Andratherthanratingtheuncertaintydirectly,Bettman'sprocedureistocomputethepercentageofbrandsfallingaboveanacceptablelevelofqualitytotheconsumer(Bettman,1973,1974).Hereportsresearch(1973,1974)whichsupportshisconceptualizationasopposedtoCunningham's(1967a)buthebaseshisargumentslargelyuponhisfindingthatwhenuncertaintyandimportancearemeasuredasheproposes,boththesecomponentscontributevariancetotheoverallriskratings,whereasusingCunningham's(1967a)uncertaintyanddangercomponents,byfarthemostvarianceisexplainedbythedangercomponentalone.Thatis,intheCunninghamprocedure,whenthereisagreatdealofdanger,certaintydoesn'tmatterandeffectsofcertaintyarefeltonlyatlowlevelsofdanger.IntheBettmanprocedure,theeffectsofcertaintyaremostpronouncedathighlevelsofimportance,whichiswhatBettmanargues"should"bethecase.Bettman(1972)didfindthatwhenuncertaintyanddangeraremeasuredasdefinedbyCunningham,thetwocomponentsarenotindependent,andthatthedangercomponentisclearlymoreimportantthantheuncertaintycomponent(alsoseeSlovicandLichtenstein,1968).Theissueraisedhereisnotmoot,butisverydifficulttoaddressempirically,sincetheremaywellbedifferencesofopinioninwhattheconceptualdefinitionofriskisthusleadingtodifferentviewsofitsfundamentaldimensionalstructure.Whetherornottherelationshipbetweenthetwodimensions,uncertaintyandconsequences(orimportance),isadditiveormultiplicative(mosthaveassumeditmultiplicative)wastestedbyBettman(1972;1974)asacombinationruleor"cognitivealgebra"question,andthroughbothgraphicalandstatisticaltestshefoundsupportfortheadditiveratherthanthemultiplicativeprocedure.ImplicittothequestionsraisedbyBettmanishisdistinctionbetween"inherentrisk"and"handledrisk.""Inherentriskisthelatentriskaproductclassholdsforaconsumer,theinnatedegreeofconflicttheproductclassarousesintheconsumer.Handledriskistheamountofconflictaproductclassengenderswhenthebuyerchoosesabrandfromthatproductclassinhisusualbuyingsituation.Thus,handledriskincludestheeffectsofinformation-->_riskreductionprocessesastheyhaveactedoninherentrisk"(Bettman,1972,p.394).Bettmannotesthatthesetwodifferenttypesofriskhavebeenconfusedintheresearchliterature;Cunningham(1967a)usinginherentriskandCoxandRich(1966)andSpence,EngelandBlackwell(1970)usinghandledrisk.Subsequently,onlyafewresearchers(e.g.LutzandReilly,1973)haveexplicitlynotedthetypeofrisktheyaremeasuringinthesenseofBettman'sdistinction.ThatthedistinctionisimportantisdemonstratedbyBettman's(1972)researchwhichfoundthatofnineproductsstudied,toothpasteandmargarinehadthehighestratingsforrelativeinherentriskwhilebeerandinstantcoffeehadthehighestrelativeratingsforhandledrisk.PERCEIVEDRISK,WORD-OF-MOUTH,ANDOPINIONLEADERSHIPWord-of-mouthandopinionleadershipweretheconceptsfirstresearchedinrelationshiptoperceivedrisk,perhapsbecauseBauerhadassertedthat"...oneoftheveryimportantfunctionsofopinionleadersistoreducetheperceivedriskofthebehaviorinquestion"(Bauer,1960,p.26).Researchonthemannerinwhichphysiciansadoptednewdrugsundoubtedlyinfluencedhisview(Coleman,Katz,andMenzel,1957;Coleman,Menzel,andKatz,1959).Thisresearchsuggestedthatdoctorstendedtorelyontheircolleagues,especiallythemore"respected"ones,earlyinthediffusionprocess,ratherthanonnon-professionalmedicalsources.Oncethedrughadbecomereasonablywellestablished,personalinfluenceseemedtoplayalessimportantrole.Astheseverityofthediseaseforwhichthedrugwastobeusedincreasedsodidthepropensityfordoctorstorelyuponprofessionalascomparedtocommercialsources(Bursk,1960;BauerandWortzel.1966).Cunningham(1964,1966,1967a,1967b,and1967c)measuredtheuncertaintyanddanger(consequences)housewivesperceivedinthefabricsoftener,dryspaghetti,andheadacheremediesproductcategories(uncertainty--wouldanuntriedbrandworkaswell;consequences--howmuchdangerwouldsheseeintryingabrandshehadneverusedbefore.)Inaddition,brandpurchasebehavior,word-of-mouthactivity,andvariousdescriptivemeasureswereobtained.Cunninghamhypothesizedthat"...thoseusersofaproductwhowerehighinperceivedriskwouldreduceriskthroughconversationandthusagreaterproportionofthehighriskperceiverswouldbeclassifiedas'talkers'(arespondentwhodiscussedtheproductcategorywithinthelastsixmonths)thanwouldthelowriskperceivers"(1967b,p.271).Thedatasupportedthishypothesisforheadacheremediesandfabricsoftenersbutnotfordryspaghetti.Regardingthedirectionoflowofword-of-mouthasafunctionofperceivedrisk,therewereproductdifferences.Thosewhoperceivedhighriskregardingheadacheremediesweremorelikelythanlowriskperceiverstohaveinitiatedtheirlastconversationabouttheproduct,buttherelationshipwasreversedforfabricsofteners.Also,thosehigherinperceivedriskforfabricsoftenersanddryspaghetti(butnotheadacheremedies)weremorelikelytohaverequestedinformationthanthoselowerinperceivedriskandweremorelikelytoclaimtheyhadmadearecommendationintheirlastconversation,afindingatvariancewiththeresearcher'shypothesis.Nevertheless,Cunninghamviewsthisfindingasonewhich"...stronglysupportsthenotionadvanced...thathighperceivedriskconsumersaresoughtoutbyotherswhopresumablyvaluetheirexpertopinion"(1967b,pp.282-283).Thus,inhisviewofthedata,thehighriskperceiverismoreapttobeanopinionleader.Aprobleminthisresearchdesignisthatsubjectswereretrospectivelyreportingtheirroleintheword-of-mouthprocess.Thequestionregardingopinionleadershipwas,"Whenyoubringupthesubjectofproductsandbrands,doyouusuallyasksomeoneelseforinformationordoyoujustsuggesthelpfulinformationfromyourownexperience?"(1967b,p.279).Thosewhoarehighriskperceiversmightreportthattheysuggestedinformationtoothers(morethanlowriskperceivers)aspartofarisk(dissonance)reductionprocess.Cunninghamrecognizesthisissuebutdiscountsthisinterpretationinsustainingtheconclusionreportedabove.Arndt(1967b,1967c,1968b,1968c)studiedword-of-mouthflawwithinamarriedstudenthousingcomplexconcerningtheadoptionprocessforanewbrandofcoffee,PERKY,andthedatasupporthishypothesisthatopinionleaderswouldbelowerinperceivedrisk,contrarytoCunningham'sresearchfindingsabove.Arndtbelievesthatthedifferencesbetweenthetwostudiesmightbeexplainedbydifferencesinmethodologyorinproductschosenforstudy.Indeed,thereweremethodologicaldifferences;Arndtmeasuredimportance(Howimportantisittoyouthatanewbrandofcoffeeyouhavenevertriedbeforeisasgoodasyourpresentbrand:notimportant,fairlyimportant,important,orveryimportant?")ratherthanconsequences.Arndtalsofoundthatword-of-mouthhadmoreeffectonhigh-thanonlow-riskperceivers.Thatis,thosewhowerehighriskperceiversappearedtopaymoreattentiontowhattheyhadheard,particularlytounfavorablecomments.Ingeneral,Arndtviewshisdataassupportiveoftheconclusionthat,"word-of-mouthseemedtoflowfromthelowtothehigh-riskperceivers"(1967c,p.294),primarilybecausethelow-riskperceiversweremorelikelytoreporthavinggivenadviceaboutPERKYthanhigh-riskperceivers.Inmostallotherregards,however,thehigh-riskperceiversseemedtobemore"active"thanthelow-riskperceivers:ininitiatingpre-purchaseconversation,inoverhearingcomments,andinseekinginformation.Variousotherstudieshaveaskedsubjectstoevaluatetheimportanceofalternativeinformationsources,andpersonalsources(i.e.,word-of-mouth)areinvariablyratedhighinimportance,andthereisevidencethatitisinparticularly"high-risk"situationswherepersonalinfluenceismostimportant,supportingArndt'sview(1967a).Forexample,Roselius(1971)foundthatthosewhoperceivedhighriskfor"time","ego",or"money"lossratedword-of-mouthmorehelpful(asa"reliever")thandidsubjectsingeneral.AndPerryandHamm(1969)foundthat"...thehighertheriskinvolvedinaparticularpurchasedecision,thegreatertheimportanceofpersonalinfluence"(p.354)intheirstudyofsocialandeconomicrisksperceivedbysubjectsacross25productcategories.ThesameconclusionwasreachedbyLutzandReilly(1973)intheirstudyoftheeffectsofsocialandperformanceriskonconsumerinformationacquisition;word-of-mouthwasthemostimportantofthefoursourcesofinformationavailableexternallytosubjects(word-of-mouth,massmediaadvertising,ratingmagazine,andsalesclerks).However,theydidnotfindthatvariationsinlevelsofsocialriskinfluencedinformationsearchbehaviorastheyhadhypothesized.Finally,ShethandVenkatesan(1968)alsofoundthattheexperimentallycreatedhigh-risk(regardingthehair-sprayproductcategory)groupsoughtpersonalsourcesofinformationsignificantlymorethandidthelow-riskgroup.Theresearchonperceivedrisk,word-of-mouth,andopinionleadershipwouldseemtosupportthegeneralizationthatword-of-mouthfunctionsasanimportant(butnotnecessarilythemostimportant,e.g.Roselius,1969)riskrelieveracrossmostoralltypesofrisks.Thenatureanddirectionofword-of-mouthflow,andspecificallyasthisrelatestoopinionleadership,islessclear.Certainlythisisaverycomplexissueandonenoteasilyamenabletoinvestigation,especiallythroughaself-reportmode.PERCEIVEDRISK,NEWPRODUCTADOPTIONANDBRAND/STORELOYALTYThedrugadoptionstudiespreviouslyreferredto(e.g.Coleman,et.al.,1957;andColeman,et.al.,1959)suggestthehypothesisthatthosehighinperceivedriskforaproductcategorywouldbelesslikelytoadoptatall,ortoadoptquickly,ifatall,anewbrandintroducedwithinthatcategory,andvice-versa.Theresearchsubsequentlywouldseemtostronglysupportthishypothesis.BothArndt(1967b)andCunningham(1967b)inthestudiespreviouslyreferredto,foundevidencetothiseffect(althoughnotnecessarilyaclearrelationshipforallproductsstudied),asdidSchiffman(1972)inhisstudyoftheadoptionofasaltsubstituteproductamongelderlyconsumers.IfoneacceptsSheth's(1968)assumptionthattheadoptionofastainlesssteelrazorbladeisalow-riskdecision,thenonemighttakehisresultsasevidenceforthesamehypothesis:"...ashighas89percentoftotalrespondentsadoptedthestainlesssteelbladesinslightlymorethanayear'stimefromthe"fluxofthreemajorbrandsinthemarketin1963"...and..."morethan90percentoftotalrespondentsadoptedwithinoneyearafterbecomingaware.Similarly,ashighas49percentadoptedthestainlesssteelbladesimmediatelyaftertheybecameaware"(pp.180-181).Hecontinues,"...itcanbeeasilyseenthatforalowriskinnovationalsopossessingstrongrelativeadvantage,thediffusionisfasterbothintermsoftimeofadoptionandthementalprocessofadoption"(p.181).Thehypotheticalrelationshipbetweenperceivedriskandbrand/storeloyaltyiscloselyrelatedtoitsrelationshipwithnewproductadoption.Loyaltyshouldbestrongeramongthoseperceivinghigh-riskintheproductcategoryandforbasicallythesamereason:"Muchbrandloyaltyisadeviceforreducingtherisksofconsumerdecisions"(Bauer,1960,p.25).Arndt(1967b)foundthatthosehighinperceivedriskforcoffeeweremorelikelytobebrandloyalandhencelesslikelytoadoptthenewcoffeeunderstudy.Cunningham(1967c)similarlyfoundsupportiveevidenceforthisrelationshipbutitwaslessstrongfordryspaghettithanforfabricsoftenersorheadacheremedies,thussuggestingthatwhereriskisgenerallylowfor.theproductcategory(e.g.dryspaghetti),brandloyaltyplaysasmallerroleasariskreductionprocess.ShethandVenkatesan(1968)studiedthedevelopmentofbrandloyaltyasarisk-reductionprocessinrepetitive(overtime)consumerbehaviorandfoundsupportfortheirhypothesisthatbrandloyaltyincreasedovertime.Itshouldbenoted,however,thatthedevelopmentofbrandloyalty(repeatselectionofbrands)wasquickerforthelow-riskthanforthehigh-riskgroups.Theauthorsconcludethat"...perceivedrisk"isanecessaryconditiononlyforthedevelopmentofbrandloyalty.Thesufficientconditionistheexistenceofwell-knownmarketbrand(s)onwhichtheconsumercanrely."(p.310)Hisrich,Dornoff,andKernan(1972)hypothesizedthat,"Iftheproductisintolerablyambiguous,perhapsthestore,whichmightbefarlessambiguous,canserveasasurrogate"(p.435).Theychosedraperies,furniture,andcarpetingastheir"ambiguous"products,butfoundtheirdatarejectedthehypothesis.Forallthreeproducts,andforbothmaleandfemalesubjects,andateverylevelofperceivedrisk,thenumberofstore-loyalbuyerswaslessthanthenumberofnon-loyalbuyers.Theyconclude,"Ataminimum,thissuggeststhatthesebuyersdidnotconsiderrepeatpatronageasaviablerisk-handlingstrategy.Indeed,dependingonpriorresults,notshoppingatapreviously-patronizedstoremighthaveservedasaformofriskreduction"(pp.438-439)PERCEIVEDRISKANDMODEOFSHOPPINGNotingthatmanywomendonotorderanymerchandisebyphone,CoxandRich(1964)hypothesizedthattelephoneshoppingcreatesperceivedriskofsufficientmagnitudetodetermanywomenfromshoppingbythismode.Althoughintheirmeasurementprocedure,inthereviewer'sopinion,therewascontaminationbetweenthecriterionmeasure(whetherornottheitemwaspurchasedbyphone)andthedependentvariablemeasure(whatitemscouldbeboughtbytelephonewithoutworryandwhichwouldbeworriedabout),theauthorsconcludethat"...highperceivedriskislikelytobeastrongdeterrenttopurchasinganitembytelephone"(p.499).Whenrespondentswereaskedwhytheyhadnotshopped,