The Zaichkowsky Personal Involvement Inventory

Their 20 item scale (1985) taps four facets of involvement: perceived importance, decision risk (probability of making a mistake), sign value (whether a ... LearnmoreaboutourTermsofServiceandPrivacyPolicy Dismiss TheZaichkowskyPersonalInvolvementInventory:ModificationandExtension ABSTRACT-ArevisedversionofZaichkowsky's(1985)PersonalInvolvementInventory(PII)wasdevelopedandtested.TermedtheRPII,therevisionattemptstoincorporatethemultifacetedperspectiveoninvolvementdevelopedbyLaurentandKapferer(1985),andalsotopurgethePIIofsomepotentiallyproblematicscaleitems.Findingsfrom136studentswhorated12productsshowedtheRPIItobesuccessful. Citation: EdwardF.McQuarrieandJ.MichaelMunson(1987),"TheZaichkowskyPersonalInvolvementInventory:ModificationandExtension",inNA-AdvancesinConsumerResearchVolume14,eds.MelanieWallendorfandPaulAnderson,Provo,UT:AssociationforConsumerResearch,Pages:36-40. AdvancesinConsumerResearchVolume14,1987    Pages36-40THEZAICHKOWSKYPERSONALINVOLVEMENTINVENTORY:MODIFICATIONANDEXTENSIONEdwardF.McQuarrie,SantaClaraUniversityJ.MichaelMunson,SantaClaraUniversityABSTRACT-ArevisedversionofZaichkowsky's(1985)PersonalInvolvementInventory(PII)wasdevelopedandtested.TermedtheRPII,therevisionattemptstoincorporatethemultifacetedperspectiveoninvolvementdevelopedbyLaurentandKapferer(1985),andalsotopurgethePIIofsomepotentiallyproblematicscaleitems.Findingsfrom136studentswhorated12productsshowedtheRPIItobesuccessful.INTRODUCTIONTheconstructofinvolvementhasbeenacentralconcerninconsumerresearchoverthepastdecade.Earlyworkfocusedonadichotomyofhighandlowinvolvementproducts,withthelatterdemandingadifferentmodelofhowconsumersprocessinformationandmakechoices(KassarjianandKassarjian1979;Robertson1976).Latereffortsattemptedtofurtherdifferentiatetheconceptofinvolvement.Thus,HoustonandRothschild(1978)distinguishedsituational,enduringandresponseinvolvement,andBlochandRichins(1983),writingonproductimportance,distinguishedinstrumentalfromenduringimportance.Overtime,definitionsanddistinctionsproliferated,tothedistressofsomescholars.Cohen(1983)attemptedtobringorderbyinsistingthattheantecedentsandconsequentsofinvolvementbeconsideredseparatelyfromthestateitself.Rothschild(1984)declaredthattheconceptualelaborationoftheinvolvementconstructhadreachedapointofdiminishingreturns.Hearguedthataconsensushadformedaroundadefinitionofinvolvementas"astateofarousal,interestormotivation,"andthatthenewpriorityshouldbedatacollectionandnotfurtherconceptualization.During1985,twomilestoneswerereachedintheefforttogroundtheinvolvementconstruct.Zaichkowsky(1985),intheJournalofConsumerResearch,andLaurentandKapferer(1985),intheJournalofMarketingResearch,reportedthedevelopmentofmethodologicallysoundmeasuresofinvolvement.Theseauthorswerecarefultomeasurethe"state"ofinvolvement,ratherthanrelyingonindicantsassociatedwiththeantecedentsandconsequentsofthisstate.Sheresultineachcaseisa"multi-item"scale(i.e.,inventory)whichsurvivedmultipletestsofvalidity,andwhichisclaimedtobeofgeneralapplicabilityacrossproductcategories.Thesetwoinventoriespromisetobeasignificantcontribution.Therefore,onenoteswithconsternationthattheseseparateeffortshaveproducedtwoverydifferentinventories.ThePersonalInvolvementInventory(PII)ofZaichkowskytreatsinvolvementasaunidimensionalconstruct;its20itemsaresummedtoproduceasinglescore.Whereas,LaurentandKapfererareadamantthatinvolvementismulti-faceted,andclaimthatanInvolvementProfile(IP)isrequired.Theyarguethataconsumer'sinvolvementcannotbeexpressedinasinglescore,becausethetypeofinvolvementisasimportantasitslevel.Their20itemscale(1985)tapsfourfacetsofinvolvement:perceivedimportance,decisionrisk(probabilityofmakingamistake),signvalue(whetheraproductrevealstheconsumertootherpeople),andapleasurecomponent.Onlythefirst,andtosomeextentthelastofthesefacetsisrepresentedamongtheitemscomprisingZaichkowsky'sPII.Whilethetwoinventorydevelopmenteffortsdidusedifferenttypesofitems(semanticdifferentialinthePIIandLikertintheIP),anddifferentpopulations,thehighstandardofrigoradheredtoinbotheffortsmakesitdifficulttoexplainawaytheirdivergentresultsonmethodologicalgrounds.Theproblemisconceptual:Isinvolvementwithaproductcategoryonething,ormany?WefindLaurentandKapferer's(1985)argumentsfortheirIPpersuasive.Theypointfirsttothetendencyofresearchersandmanagerstouseinvolvementinassociationwithvariousqualifiers:situationalorenduring,personaloremotional,andsoforth.Second,eachoftheirfourfacetscanbeconvincinglyrelatedtoarousal,whichCohen(1983)hasarguedisthefundamentalconstituentofthestateofinvolvement.Perceivedimportance,decisionrisk,psychosocialrisk(signvalue),andpleasureareallplausiblesourcesofagreaterorlesserdegreeofarousal.Third,theiranalysesdemonstrateboththatindividualproductswillberankeddifferentlyonthefourfacets.DespitethesegoodconceptualargumentsfortheuseoftheIPratherthanZaichkowsky'sPIIinstudiesofinvolvement,thereremaintwoproblems:(1)thefullIPhasneverbeenpublished;(2)whilethetextofthemeasurecoulddoubtlessbeobtainedfromtheauthors,thereisnoguaranteethattranslationsofthe20LikertstatementsintoEnglishwillyieldthesameitemstructureastheFrenchoriginals.GiventhatadditionalworkwouldinanycaseberequiredbeforetheInvolvementProfilecouldbewidelyusedinthiscountry,itseemsworthwhiletoaskwhetherZaichkowsky'sPersonalInvolvementInventorycouldnotinsteadbeadaptedtoreflectamoremulti-dimensionalperspective.TheprimarygoalofthecurrentstudyisthustomodifythePIIsoastoproduceameasurethatwillincorporateriskandsign-valuecomponents,aswellasperceivedimportanceandpleasure.Ourobjectiveistomaintainasmuchcontinuityaspossiblewiththeexistingmeasure,whilestillincorporatingthenecessaryadditionalmaterial.AwedidnotwantthefinalrevisedversiontoexceedthelengthofthepresentPII,itwasnecessarytodeterminecandidatesforeliminationfromamongthePII'soriginalitems.AconceptualcritiqueofthePIIwasdevelopedtoguidethesechoices;weweremindfulofthestricturesplacedbyZaichkowskyoninvestigatorsseekingtoshortenorotherwiserevisethescale(footnote1,p.344).CRITIQUEOFTHEPERSONALINVOLVEMENTINVENTORYExaminationofthePII'sscaleitemsalertedustothepossibilityofinterpretationalconfounding.Ingeneralterms,interpretationalconfounding"occursastheassignmentofempiricalmeaningtoanunobservablevariablewhichisotherthanthemeaningassignedtoitbyanindividualaprioritoestimatingunknownparameters"(Burt1976,p.4).ThePIIseemstoincludetwodistinctgroupsofadjectives.Onegroupcontainsitemsthatwouldpossesshighfacevalidityasindicantsofinvolvement,priortoanyempiricalvalidationwork(e.g.,"interesting-boring").Buttheothergroupcontainstermswithquitedifferentconnotations(e.g.,"beneficial-notbeneficial,""valuable-worthless").Inotherwords,thePIIcontainssomeadjectivesclassicallyassociatedwithastateofinvolvement,andothersnormallyassociatedwiththemeasurementofattitude.Empirically,Zaichkowsky(1985)foundallthesetermstobehighlyconsistent(coefficientalpharangedashighas.97);butconceptually,theyrepresenttwodifferentconstructs.Thus,wesuspectthatinterpretationalconfoundingmaybepresentintheformofattitudinalcontamination.Whileameasureofinvolvementshouldbedistinctfromameasureofattitude,itisnotclearthatthePIIsatisfiesthistest(PetersandChurchill1986).Ifinterpretationalconfoundingispresentintheformofattitudinalcontamination,thenthePIIcanbeexpectedtooverestimateconsumers'involvementwithcertaintypesofproducts,particularlythosethatcanbelikedorendorsedwithoutbeingexperiencedasarousingorinteresting.Accordingly,thisoverestimationmaybemostmarkedforwhatmightbetermed"humbleproducts;"i.e.,everydayitemswhich,whileindispensable,areofnogreatimportanceinthemselves.Conversely,itmaynotoccur,orbenoticeable,forthoseproductswhicharenotusedbyeveryone,norconsiderednecessities;i.e.,thespecialinterestcategoryofLastovickaandGardner(1979).Hereattitudinaltermslike"needed"or"essential"maybeexpectedtoreflectinvolvementequallywellasarousal-basedtermssuchas"interesting"or"exciting."ThereissuggestiveevidencethatinterpretationalconfoundingdoesoccurwhenthePIIisusedtoscaletheinvolvementlevelofvariousproductcategories.Zaichkowskyfoundlaundrydetergenttobethethirdmostinvolvingproductamongthe14sheexamined.Shedefendstheseresultsintermsofthepresumedcentralroleofdetergentinthelivesofa"relativelyhomogeneousgroupofmiddle-agedfemales"(p.346),contrastingthiswiththeperipheralroleplayedbyatechnicalappliancesuchasacolortelevision.Herpointhasmerit,butremainstroubling,giventheexactlyoppositeresultsreportedbyKapfererandLaurent(1984)andLaurentandKapferer(1985),intwoseparate,largestudiesofFrenchhousewives.Inthelatterone,detergentwasthelowest-scoringproductonallfourfacetsofinvolvement;intheformer,itwasamongthelowest;andinbothstudies,detergentscoredmuchlowerthantelevision.WhileourmajorcriticismsofZaichkowsky's(1985)PIIconcerntheabsenceofamulti-dimensionalapproachandthedangerofattitudinalcontamination,wewouldalsonotetwootherpotentialproblems.First,onemightquestiontheappropriatenessoftermssuchas"superfluous"and"mundane"ifthemeasureweretobeappliedoutsideaUniversitysetting.Ingeneral,thesyllablecountforthePIIseemsuncomfortablyhigh.Second,sometermsinthePIIappeartoberedundant(e.g.,"interesting-boring"and"interested-uninterested"arebothincluded).Despitethesecriticisms,theextensiveandlaudablevalidationworkperformedbyZaichkowskyarguesforarevisionofthePIIratherthanitsabandonment.Wesoughtbothtokeepthebestofheritempool,andtoreplacepotentiallyproblematicterms.METHODDevelopmentof-theInstrumentFourofthe20itempairsinthePIIwerediscardedonpriorigrounds:"superfluous-vital,""mundane-fascinating,""significant-insignificant,"and"fundamental-trivial."Allseemedinappropriateforusewithanon-collegeeducatedpopulation.Eightnewitempairsweredevised.SincethePIIalreadyhadmanyitemsreflectiveofperceivedimportanceofhedonicvalue,thenewitemsweredesignedtoreflectthefacetsofdecisionriskorsignvalue(theoneexceptionwas"fun-notfun").The16oldand8newitempairswerearrangedasshownintheExhibit.Unfortunately,preliminaryanalysesindicatedthatthefouritempairsdesignedtomeasuresignvaluetitnotcohereasasinglefactor.Theproblemlaywiththethirdandfourthpairslisted(i.e.,"heedothers'wishes"and"myownbusiness"),whichwererelativelyuncorrelatedwithanyoftheother22itempairs.Perhapstheytapajointdecision-makingdimension,ratherthansignvalue.Thetworemainingpairs("sayssomethingaboutme"and"tellsmeaboutaperson"),takendirectlyfromKapfererandLaurent(1984),werefoundtohavehighloadingsonthepleasurefactor,anddidnotconstituteaseparatefactorbytheminimumeigenvaluetest(i.e.,X>1.0).Thisisnotsurprising,sincethesignandpleasurefacetshadthehighestinter-correlationinLaurentandKapferer(1985).Theselattertwoitemswerethereforeincorporatedintothepleasuresub-scaleinthepresentanalysis.Inthediscussionbelow,"RPII"and"OPII"refertotwooverlappingsubsetsofthe22remainingscaleitems,asshownintheExhibit.SubjectsandProcedureThesemanticdifferentialinstrumentshownintheExhibitwascompletedby80undergraduateand56MBAstudents.[Althougheachstudentratedatotalof12stimulusobjects,twoseparatelistswereused,sothatinall24objectswererated.Someoftheseratings(e.g.,ofissues--cf.HupferandGardner1971)weretakenforadifferentpurposeandarenotdiscussedhere.Asaresult,thenumberofvalidcases(studentsXproductsminusmissingcases)is899fortheanalysesreportedinTables1and2andFigure1,and449forthosereportedinTable3.]Afteroneweek,67ofthe80undergraduatescompletedaformwhichmeasuredtheconsequencesofinvolvement(describedbelow).Lastly,fourweeksaftertheinitialadministrationofthesemanticdifferentialinstrument,52undergraduatescompletedthe22scaleitemsintheExhibitasecondtimeforautomobilesandtoothpaste.Thus,104caseswereavailablefortestingthestabilityoftheRPIIandOPIIovertime.CriterionmeasuresFivecriterionmeasures,modeledafterthoseusedbybothZaichkowsky(1985)andLaurentandKapferer(1985),weredevelopedtoassessvariousconsequencesofinvolvement.Asnotedbytheseauthors,ahigherdegreeofinvolvementshouldproduce:l)greaterinformationsearch;2)greatercomplexityofchoiceprocesses;3)greaterbrandcommitment;and4)greaterdifferentiationofbrands.Thefirstofthesewasassessedwiththefollowingtwoitems,eachmeasuredonafive-pointLikertscale:"Iwouldbeinterestedinreadingaboutthisproduct,"and"Iwouldpayattentiontoanadforthisproduct."Thesecondconsequencewasmeasuredusingasinglefive-pointLikertitem:"Iwouldcompareproductcharacteristicsamongbrands."Inasmuchasallthreeoftheseitemswerefoundtobehighlycorrelated(.66,.53and.55,respectively),theyweresummedtogether(coefficientalphas.81).Brandcommitmentwasmeasuredasthedegreeoflikingforthebrandnowownedorregularlyused(seven-pointsemanticdifferential,anchoredby"like"and"dislike").Branddifferentiationwasmeasuredasthenumberofacceptablebrandsintheproductcategory,withfiveresponseoptions:mineonly,acouple,several,many,all.Thismeasurewasreverse-scored:thefewertheacceptablebrands,thegreaterthebranddifferentiation.RESULTSReliabilitySablelshowsinternalconsistencyestimatesfortheinventoriesandsub-scales,andalsotheirinter-correlations.ConsistentwithZaichkowsky(1985),theOPIIwasfoundtoexhibitaveryhighdegreeofinternalconsistency(.95).But,despitebeingcomposedof3sub-scales,theRPIIalsoexhibitsmorethansatisfactoryinternalconsistency(.93).Itshouldbenoted,however,thatthishighalphacoefficientdoesnotnecessarilyimplythattheRPII(orOPIIforthatmatter)isunidimensional.Itispossibletohaveahighalphaforascalewhichhastwoormoredimensions.Therefore,asubsequentfactoranalysis(discussedbelow)wasusedtoassessdimensionality.WithintheRPII,theimportanceandpleasuresub-scalesareveryconsistent,whiletheriskscaleissomewhatlessso.ThissamepatternwasobservedbyLaurentandKapferer(1985).Additionally,theinter-correlationoftheimportanceandpleasuresub-scalesissimilartothatfoundbyLaurentandKapferer(1985);however,ourmeasureofdecisionriskismorecloselyrelatedtotheothertwosub-scalesthanwasthecaseintheirstudies.TheintercorrelationoftheOPIIandRPIIinventoriesisquitehigh(r-.87),aswouldbeexpectedgiventhesubstantialscaleoverlapbetweenthetwo.ThisdegreeofassociationshowsthatweweresuccessfulinmaintainingcomparabilitowiththeOPII.EXHIBITTEXTOFSCALEITEMSCOMPLETEDBYRESPONDENTSShetest-retestresultsweremoredisappointing.However,thecorrelationbetweenthetwoadministrationsoftheRPII(.80)washigherthanthatoftheOPII(.69).Test-retestcorrelationsforindividualitemsrangedfrom.20to.74fortheOPII,andfrom.20to.75fortheRPII.Thesenumbersarelowerthanthe.88to.93rangereportedbyZaichkowskyforthetotal20iteminventory,andthe.60to.93rangeshereportedforindividualitems.Severalplausibleexplanationsforthelowernumbersinthecurrentstudyexist:thelongertimeframebetweentestandretest;greaterheterogeneityinthecircumstancesofinventoryadministration(i.e.,completedoutsidetheclassroom);fewerpeopleandfewerproductstested;differencesinthesetofproductsthatwasexamined;and,ofcourse,thefactthatonly16ofZaichkowsky's20itemswereincludedintheOPII.Onthebasisoftheinternalpsychometriccriteriadiscussedabove,wecanconcludethattheRPIIisonaparwiththeOPII.Inaddition,itappearsthatourimportance,pleasureandrisksub-scalesareonaparwiththosedevelopedbyLaurentandKapferer.FactorStructureoftheRPIIandOPIIAprincipalcomponentsanalysiswithvarimaxrotationofallfactorswitheigenvaluesgreaterthanonewasperformedforeachinvolvementinventoryseparately(Table2).RegardingtheOPII,wefound,asdidZaichkowsky,onemajorandoneminorfactor,whichhereaccountedfor60%and15%ofthetotalexplainedvariance.However,itisclearthatthemajorfactorreflectsattitudemorethaninvolvement;thehighestloadingsareforthescaleitems"needed,""essential,"and"beneficial."Itemssuchas"interesting"loadmostlyonthesecondfactor.Totheextentthatasimplesumreflectsprimarilythefirstfactor,andtotheextentthattermstypicallyassociatedwithinvolvementdonotloadheavilyonthatfirstfactor,theOPIImaynottrulyreflecttheconstructofinvolvement.Thequestionofwhetherthisisapracticaloronlyatheoreticalproblemisdeferreduntilthesectiononscalingofproductsbelow.TABLE1INTERNALCONSISTENCYANDINTER-CORRELATIONOFINVENTORIESANDSUB-SCALESTheresultsfortheRPIIareencouraging.Threefactorswereretained,accountingfor54%,13%and10%,or77%intotal,oftheexplainedvariance.Therotatedloadingsshowthatsimplestructureobtains,andthateachitemloadsontheexpectedfactor.Theseloadingsarebothmorehomogeneous,andsomewhathigherthanthosereportedinLaurentandKapferer(1985).Totestthegeneralizabilityoftheseresults,weseparatedthesampleintoundergraduateandMBAgroups,andredidthefactoranalysisoftheRPIIforeachgroupseparately.TheMBAstudentswereenrolledinaneveningprogram,andcomparedtotheundergraduates,weregenerallyolder,employedfull-time,andmorelikelytobemarried.Acomparisonofthefactorstructuresforthetwogroupsshowedboththesamesimplestructure,andalsoverylittledifferenceintheloadingsforindividualscaleitems.Aswasthecasewiththereliabilityanalyses,thesefactoranalysesaddtoourconfidencethattheRPIIholdspromiseasareplacementfortheOPII.ScalingofProductInvolvementIftheOPIIdoessufferfromattitudinalcontamination,thenthisshouldbemanifestasanoverestimateoftheinvolvementassociatedwithcertainproductcategories.Figurelshowsthe12productsinvestigated,orderedfromlowesttohighest,basedontheirinvolvementscoreontheRPII.Thesescoresareplottedbythesolidline.ThedashedlineplotsthescoresofeachproductontheOPII.Deviationsbetweenthetwolinesindicateproductswhicharescaleddifferentlybythetwomeasures.VisualexaminationofFigurelprovidessuggestiveevidenceofinterpretationalconfoundingintheOPII.Overestimation,relativetotheRPII,occursinexactlythosecases--detergent,motoroilandtoothpaste--whereitwaspredicted:"humblebutusefulproducts"thateveryoneconsumes.However,theRPIIandOPIIareinsubstantialagreementregardingmostoftheproductsexamined.Forinstance,theyareequallyeffectiveindeterminingthatbusinesssuitsaremoreinvolvingthansoftdrinks(Figurel).WhatemergesisthattheOPIImaysystematicallyerrinestimatingthedegreeofinvolvementforcertainproductsvis-a-visothers.Thisfinding,ofcourse,holdstrueonlytotheextentthattheorderingofproductsduetotheRPIIdoesnotalso,inthereader'seyes,sufferfrominterpretationalconfounding.TABLE2ROTATEDFACTORELOADINGSFORTHEOPIIANDRPIIPredictionoftheConsequencesofInvolvementThefinalcomparisonoftheOPIIandRPIIconcernstheirabilitytopredictsomeoftheconsequenceswhicharesupposedtofollowfrominvolvement:greaterbrandcommitment,greaterbranddifferentiation,andmoreinformationsearchandchoicecomplexity.RegressionanalyseswereusedtocomparethedegreeofassociationbetweentheOPIIandtheconsequencesofinvolvementwiththatbetweenthethreesub-scalesoftheRPIIandtheconsequencesofinvolvement.Ifinvolvementismulti-faceted,asLaurentandKapfererclaim,theninclusionofeachofthesubscalesintheregressionanalysesshouldexplainmorevariancethanwhentheOPIIisenteredassolepredictor.Ontheotherhand,ifZaichkowskyiscorrectandinvolvementisunidimensional,thenthemorepsychometricallypowerfulOPIImeasure,consistingof16highlyconsistentitems,shouldbemoresuccessfulthanthethree"fragments"ofinvolvementrepresentedbytheRPIIsub-scales.Threeregressionswererunforeachconsequenceofinvolvement(Table3):l)onewheretheOPIIwastheonlypredictorvariable(leftmostcolumn);2)onewherethethreeRPIIsub-scalesweretheonlypredictors(column2);and3)onewherethethreesub-scaleswereaddedtoanequationcontainingtheOPII,andtheincrementinexplainedvariancedetermined(in-texttabulation).Lookingatthefirstandsecondcolumns,ineverycasethethreesub-scalesoftheRPIIareequaltoorsuperiortotheOPIIinexplainingvariationintheconsequencesofinvolvement.Sinceacomparisonofsingleandmultiplecorrelationsisinherentlyunfair,itisimportanttonotethatevenwhentheOPIIisenteredfirstintotheregressionequation,thesub-scalesareabletoexplainsignificantadditionalvariationwhentheyareadded(R2increments-.023,.023lp<.01 authors edwardf.mcquarrie j.michaelmunson volume na-advancesinconsumerresearchvolume14 shareproceeding featuredpapers seemore featured analyzingtheperceptionofexperientialluxuryconsumptionofmillennialsoninstagram:anewmethodologicalappr oach marinaleban matthiasplennert readmore a1.trustingandactingonchanceonline shivaunanderberg ellengarbarino toomuchofagoodthing tarjegaustad bendiksamuelsen lukwarlop gavanfitzsimons engagewithus becominganassociationforconsumerresearchmemberissimple.membershipinacrisrelativelyinexpensive joinacrnow search closemodalwindow>