购买意愿量表2006.“Converting Web Site Visitors into Buyers：How web Site

购买意愿量 表 关于同志近三年现实表现材料材料类招标技术评分表图表与交易pdf视力表打印pdf用图表说话 pdf 2006.“ConvertingWebSiteVisitorsintoBuyers：HowwebSite133JournalofMarketingVol.70(April2006),133–148©2006,AmericanMarketingAssociationISSN:0022-2429(print),1547-7185(electronic)AnnE.Schlosser,TiffanyBarnettWhite,&SusanM.LloydConvertingWebSiteVisitorsintoBuyers:HowWebSiteInvestmentIncreasesConsumerTrustingBeliefsandOnlinePurchaseIntentionsTheauthorsinvestigatetheimpactofWebsitedesigninvestmentsonconsumers’trustingbeliefsandonlinepur-chaseintentions.Suchinvestmentssignalthecomponentoftrustingbeliefsthatismoststronglyrelatedtoonlinepurchaseintentions:ability.Theseeffectswerestrongestwhenconsumers’goalsweretosearchratherthantobrowseandwhenpurchasesinvolvedrisk.AnnE.SchlosserisAssistantProfessorofMarketing,UniversityofWash-ingtonBusinessSchool(e-mail:aschloss@u.washington.edu).TiffanyBarnettWhiteisAssistantProfessorofBusinessAdministration(Market-ingGroup),CollegeofBusiness,UniversityofIllinois(e-mail:tbwhite@uiuc.edu).SusanM.LloydisAssistantProfessorofMarketingandEndowedFellow,KogodSchoolofBusiness,AmericanUniversity(e-mail:smlloyd@american.edu).TheauthorsthankRuthBoltonandthetwoanonymousJMreviewers,aswellasJimBettman,KatherineLemon,MichaelMazis,andKentMonroe,fortheirhelpfulcommentsonpreviousversionsofthisarticle.TheauthorsalsothankSandraM.DahneandS.RandFishkinfortheirtechnicalassistance.FundingforStudy1waspro-videdbytheCenterforInformationTechnologyandtheGlobalEconomyattheKogodSchoolofBusiness,AmericanUniversity.“Inimportantways,usingtheInternetinvolvesaleapoffaith.Wetypeinourcreditcardnumbersandotherpersonalinformationinordertomakepur-chasesovertheInternetandtrustthatthisinformationwillnotbeusedinunauthorizedorfraudulentways”(BarghandMcKenna2004,p.585).Firmshaverespondedtosuchcon-sumerconcernsbyinvestinginWebsitesecurity,whichhasbecomeamultibilliondollarindustry(eMarketer2005).Yetevenexperiencedonlinebuyersviewpurchasingonlineasrisky(eMarketer.com2005;Forrester2005).Indeed,con-sumers’perceptionsoftherisksinvolvedinprovidingper-sonalinformationonlineoftencontrasttheviewsofsecurityexperts,causingtheseconsumerstoavoidonlineactivitiesthatareactuallysafe(Dunn2004).Thisavoidancehasledsomeexpertstospeculatethattheimmediatethreattoe-commerceisconsumers’perceptions(eMarketer2005;Hoffman,Novak,andPeralta1999;Rust,Kannan,andPeng2002).Thus,althoughitisnecessary,investinginback-endtechnologytoprotectconsumers’informationandensurethate-commercetransactionsrunsmoothlyisnotsufficient.Firms,particularlythoseattemptingtoconvertvisitorstobuyers,stillfacethechallengeofestablishingconsumers’trustonline.Thus,itisimportanttounderstandhowcompa-niescancommunicatetheirtrustworthinesstoconsumers.Marketingmanagersfacethechallengeofestablishingconsumers’trustinavarietyofcontexts,butdoingsoincomputer-mediatedenvironmentssuchastheInternetmaybeparticularlydifficult(NaquinandPaulson2003).Acom-monapproachistopostexplicitstatementsthatassurecus-tomersthatpersonaldatawillbediscreetlyusedandpro-tected(i.e.,privacyandsecuritystatements).However,evidenceontheeffectivenessofsuchstatementsiscontra-dictory.Whereassomeresearchhasshownthatsuchstate-mentshelpinstillconsumerconfidenceine-commercesites(Palmer,Bailey,andFaraj2000),otherssuggestthattheyarenotnecessarilythemostimportantpredictorofonlinetrust(Montoya-Weiss,Voss,andGrewal2003;Sultanetal.2002).Findingsfromarecentlarge-scalestudysuggestthatdespiteconsumers’claimsthatprivacypoliciesareimpor-tantforestablishingcredibility,consumersreferinsteadto“surface”elements,suchasWebsitedesign(Foggetal.2002).Inthisarticle,wedevelopaconceptualframeworkforunderstandinghowmarketingsignalsinfluenceconsumers’trustinane-commercesetting.Weclarifyimportantdis-tinctionsrelatedtotrustthathavebeenlargelyoverlookedintheliteraturebutarekeytounderstandinghowonlinefirmscanbestconvertvisitorstobuyers.Specifically,wearguethatdifferentWebsitesignalscaninfluencedifferentbeliefsaboutafirm’strustworthiness,whichinturnhavedifferentialeffectsononlinepurchaseintentions.Further-more,theseeffectsmightvaryaccordingtotheconsumers’purposeforvisitingthesiteandthelevelofrisktheyper-ceiveinthepurchase.Consistentwithpriorresearch(e.g.,McKnight,Choudhury,andKacmar2002),weconceptual-izeonlinepurchaseintentionsintermsofcustomeracquisi-tion—thatis,consumers’intentionstomakeaninitialonlinepurchasefromafirm,despitetheironlinepurchasehistorywithotherfirms.Webeginwithareviewofthetrustliterature.Wethenpresentandtestourconceptualframe-workandconcludewithadiscussionofthetheoretical,managerial,andpublicpolicyimplicationsofthesefindings.TrustTrusthasbeendefinedas“awillingnesstorelyonanexchangepartnerinwhomonehasconfidence”(Moorman,Zaltman,andDeshpandé1992,p.315);“ageneralizedexpectancyheldbyanindividualthattheword,promise,oralorwrittenstatementofanotherindividualorgroupcanbereliedupon”(Rotter1980,p.1);and“abeliefinaper-son’scompetencetoperformaspecifictaskunderspecificcircumstances”(SitkinandRoth1993,p.373).Reflectedintheseandotherdefinitionsoftrustisacognitiveaspect(i.e.,trustingbeliefs)andabehavioralaspect(i.e.,trustinginten-tions)(Kimetal.2004;Moorman,Zaltman,andDeshpandé1992).Trustingbeliefsrepresenta“sentiment,orexpectationaboutanexchangepartner’strustworthiness”(Moorman,Deshpandé,andZaltman1993,p.315).Althoughvarioustrustingbeliefshavebeenstudiedintheliterature,themajoritycanbeconceptuallyclusteredintothreedimen-sions:ability,benevolence,andintegrity(McKnight,Choudhury,andKacmar2002).“Abilitybeliefs”reflectconsumers’confidencethatthefirmhastheskillsnecessarytoperformthejob(Mayer,Davis,andSchoorman1995),“benevolencebeliefs”reflectconfidencethatthefirmhasapositiveorientationtowarditsconsumersbeyondan“ego-centricprofitmotive”(Mayer,Davis,andSchoorman1995,p.717),and“integritybeliefs”reflectconfidencethatthefirmadherestoasetofmoralprinciplesorprofessionalstandardsthatguideitsinteractionswithcustomers.Thesetrustingbeliefsarerelated,yetdistinct.Forexample,con-sumersmaybelievethatthefirmcaresaboutitscustomersandthusintendstodeliverasmooth,error-freetransaction(i.e.,thefirmisbenevolent),buttheymayalsobelievethatthefirmlackstheabilitytodoso.Likewise,althoughintegrityandbenevolencebeliefsaresimilar,theformerfocusesonmeetingobjectivestandardsofcorporatecitizen-ship,andthelatterfocusesoncustomerwelfarethatgoesbeyondnormalbusinessactivity.Forexample,despitecon-sumers’beliefsthatthefirmfollowsaprofessionalcodeofconduct(i.e.,hasintegrity),theymaystillquestionthefirm’sgenuineconcernforitscustomers(i.e.,itsbenevolence).Althoughability,benevolence,andintegritybeliefsareacknowledgedasconceptuallydistinct(e.g.,Kumar,Scheer,andSteenkamp1995),theyareoftencombinedintoaglobalmeasureoftrustingbeliefs(e.g.,DoneyandCannon1997).Whereascombiningthesebeliefsintoasinglevaria-bleisaparsimoniousapproachtostudyingtrust,itcanmakeitdifficulttoidentifywhatactionshouldbetakentobuildtrust(SmithandBarclay1997).Becauseaglobalmeasurelikelyobscuresthereasoncertainsignalsaremoreeffectivethanothersinaffectingonlinepurchaseintentions,wetreateachtrustingbeliefseparately.Trustingintentionsrepresent“awillingnesstomakeoneselfvulnerabletoanotherinthepresenceofrisk”(Kimetal.2004,p.105).Whatdistinguishestrustingintentionsfromothertypesofbehavioralintentionsisthattheyinvolverisk(Moorman,Zaltman,andDeshpandé1992).Asreflectedintheopeningquotation,purchasingonlineinvolvesrisk,especiallywhenapersonlacksexperiencewiththeonlinefirm.Specifically,theconsumermustbewillingtotransferresources(e.g.,creditcardandotherper-sonalinformation)totheonlinefirm,theconsequencesofwhichcouldbedamaging.Forexample,amongtherealand/orperceivedrisksarethatthefirmmayovercharge,failtodelivertheproduct,deliveraninferiorproduct,orfailtoprotectpersonalinformation.Totheextentthatconsumersareconcernedabouttheseandotherrisksofpurchasingonline,onlinepurchaseintentionsreflecttrustingintentions.Thedistinctionbetweentrustingbeliefsandtrustingintentionshasbeenacknowledgedbysomeresearchers(e.g.,Moorman,Zaltman,andDeshpandé1992;Sirdesh-mukh,Singh,andSabol2002)butignoredbyotherswhohavestudiedonlytrustingbeliefs,implicitlyassumingthatthesebeliefsimplytrust(e.g.,DoneyandCannon1997;Ganeson1994;Kumar,Scheer,andSteenkamp1995;Mayer,Davis,andSchoorman1995;MorganandHunt1994).Forexample,MorganandHunt(1994)arguethattrustingbeliefsaresufficientformeasuringtrustbecausesuchbeliefsimplythattrustingintentionswillfollow.Incontrast,Moorman,Zaltman,andDeshpandé(1992)arguethattrustislimitedwhentrustingbeliefsdonotaccompanyacorrespondingtrustingintentionorwhentrustinginten-tionsoccurwithoutcorrespondingtrustingbeliefs(e.g.,underconditionsofcoercionorlimitedalternatives).Inotherwords,theseresearchersarguethatbothtrustingbeliefsandtrustingintentionsmustbepresentfortrusttoexist.Likewise,wearguethattrustingbeliefsareaneces-sarybutnotsufficientconditionfortrusttoexist,becauseincreasingtrustingbeliefswillnotalwayshaveacorre-spondingpositiveeffectontrustingintentions.ConceptualFrameworkDrawingfromresearchontrust(Mayer,Davis,andSchoor-man1995;Moorman,Deshpandé,andZaltman1993),con-sumergoals(e.g.,HoffmanandNovak1996),andmarket-ingsignals(e.g.,KirmaniandWright1989;PrabhuandStewart2001),wedevelopaconceptualframeworkforunderstandinghowdifferentsignalsinfluenceability,benevolence,andintegritybeliefsandthusinfluenceonlinepurchaseintentions(seeFigure1).Becauseourobjectiveistounderstandhowtoincreaseconsumers’willingnesstobuyonline,wefocusonthosewhosegoalismostconsistentwithbuyingonline,namely,consumerswhosearchforproductinformation(orsearchers;HoffmanandNovak1996;Moe2003;Schlosser2003).Indeed,searchersthinkaboutandarepersuadedmorebyproductinformation(Schlosser2003)andhavehighervisitor-to-buyerconver-sionratesthanthosewhodonotsearch(Moe2003).Webeginbyexaminingtherelationshipbetweensearchers’trustingbeliefsandintentions.Searchers’TrustingBeliefsandOnlinePurchaseIntentionsSearchingreflectspurposive,task-specificbehaviors,suchastheplannedacquisitionofinformationduringprepur-chasedeliberation(HoffmanandNovak1996;Janiszewski1998).Similartothosewhoreadatexttofindananswertoaquestion(Rosenblatt1978),searchersarelikelymotivatedtofindtherightanswerefficiently.Suchafact-gathering,134/JournalofMarketing,April2006ConvertingWebSiteVisitorsintoBuyers/135136/JournalofMarketing,April2006likelyviolatesconsumers’expectationsaboutnormsofconductfore-commercefirms,itwilllikelyraiseconcernsaboutthefirm’sbenevolenceandintegrity.willlikelyuseperceivedWebsiteinvestmenttoinferafirm’sabilitymorethanitsbenevolenceorintegrity.Ifthisisthecaseandabilitybeliefslargelyinfluencesearchers’onlinepurchaseintentions(H1),itfollowsthatH2:Searchers’onlinepurchaseintentionswillbehigheratahigh-investmentWebsitethanatalow-investmentWebsite.H3:Beliefsaboutthefirm’sabilitywillmediatetherelation-shipbetweensiteinvestmentandsearchers’onlinepur-chaseintentions.Thefirsttwostudiestestthesehypotheses,andthelasttwostudiesexaminetwopotentialmoderators(goalandperceivedrisk).Furthermore,totestthegeneralizabilityoftheseeffects,wevariedacrossstudiesthesamplesrecruited(studentsversusnonstudents),companies(afictitiousver-suswell-knownfirm),andproducts(homefurnishingsandaccessoriesversuscameras).Study1MethodSampleanddesign.Thesampleconsistedof111respondentswhoparticipatedinexchangefor$10andwererecruitedthroughanelectronicandaprintednewsletterdis-tributedtouniversityemployees.Thesamplewas68%female,withameanageof37.5andameanincomeof$35,000to$49,999.RespondentshadamedianeducationoffouryearsofcollegeandusedtheInternetanaverageoffourtosixtimesperweek.Werandomlyassignedrespon-dentstoahigh-orlow-investmentsite.Websiteinvestmentmanipulation.WemanipulatedWebsiteinvestmentthroughthepresenceofsophisticatedWebsitetechnologyandvisualdesignelements.Specifically,thehigh-investmentsitehadawhitebackground,sophisticatedfonts(imagesforthenavigationbar;Garamondfont),andanenhancedzoomfeaturecreatedwithDesignWithinReach.Thisenableduserstozoominonanypartofanimageandtochooseamongthreepreselectedzoomsthatexecutedautomaticallywithasingleclick.Incontrast,thelow-investmentsitefeaturedthedefaultbackgroundcolorandfont(gray;TimesNewRoman)andalimitedzoomfea-ture,which,whenclicked,simplyprovidedalargerviewofthefocalproduct.Thecontentandlayoutofbothsiteswereidentical.Totesttheeffectivenessofthismanipulation,43under-graduatesviewedscreencapturesofthehomepage(fontandbackgroundcolorwereusedtomanipulateinvestment)andazoompage(technologywasusedtomanipulateinvestment).Theorderwascounterbalanced,andpartici-pantsviewedonlythosepagesspecifictothehigh-orlow-investmentsite.Afterviewingeachpage,participantsreportedhowmuchtime,effort,andmoneytheybelievedthefirminvestedineachpageonascalefrom1(“verylit-tle”)to7(“agreatdeal”).Weaveragedtheresponsestotheseitems(αs>.94)andanalyzedthemwitha2(invest-ment:highversuslow)×2(page:zoomversushome)×2(order:viewedhomeorzoompagefirst)analysisofvari-ance(ANOVA).Insupportoftheinvestmentmanipulation,participantsperceivedthehigh-investmentsiteasrequiringgreaterinvestmentthanthelow-investmentsite(Ms=4.09versus3.18;F(1,39)=9.02,p<.01).Furthermore,theeffectivenessoftheinvestmentmanipulationwasunaffectedbythetypeororderofpageviewed(Fs(1,39)<3.35,notsignificant[n.s.]).Thus,itappearsthatfont,backgroundcolor,anduseoftechnologycommunicateinvestment.Par-ticipantsalsoreportedhowinformative,entertaining,andwellorganizedtheyfoundthepagestobeonthesameseven-pointscale.Asweexpected,investmentdidnotaffectthesevariables(Fs(1,39)<1.97,n.s.).Procedure.Participantssatatacomputerandreceivedapaperbooklet.ThefirstpageinstructedparticipantsthattheywouldbevisitingasiteforanewfirmcalledUrban-Furniture(UF)andtolimittheirvisittothehomepageandliving-roomsectionsofthesite.Allparticipantsweretoldtoimaginethattheywantedtopurchasecontemporaryfurni-tureandwereconsideringUF.Asinpriorresearch(Schlosser2003),toinstillasearchinggoal,participantswereaskedtowritedowntwoquestionstheyhadforUFaboutitsproductsbeforevisitingthesite.Participantsthenvisitedeitherthehigh-orthelow-investmentsite,whichwaspreloadedontheircomputer.Afterward,theycompletedthesurvey,whichcontainedthreeitemsthatmeasuredtheironlinepurchaseintentions(α=.91)andamodificationofMayerandDavis’s(1999)scaleoftrustworthiness(forthemeasuresusedinthisandtheotherstudies,seetheAppendix).ThisscalemeasuredbeliefsaboutUF’sability(α=.90),benevolence(α=.88),andintegrity(α=.71).Then,totesttheeffectivenessoftheinvestmentmanipulationinthemainexperiment,partici-pantscompletedthethree-itemWebsiteinvestmentscale(seetheAppendix;α=.95).Attheendofthesurvey,participantsreportedtheiredu-cationandincomelevelsaswellashowoftentheyusedtheInternettopurchasegoodsinthelastsixmonthsonascalerangingfrom1(“notatall”)to7(“quiteoften”).Weincludedthesemeasurestocontrolforindividualdiffer-encesinInternetexperiencebothdirectly(i.e.,throughself-reporteduseoftheWeb)andindirectly(i.e.,usingdemo-graphicvariablesassociatedwithWebuse).Tocontrolforvarianceduetomechanicalerror,weaskedparticipantstoreporttheextenttowhichtheyencounteredproblemsattheUFsite(e.g.,errormessages,serverdelays,crashing)onascalefrom1(“notatall”)to7(“quiteoften”).ResultsManipulationchecks.Weanalyzedtheinvestmentmanipulationwithaone-wayanalysisofcovariance(ANCOVA),controllingforreportedproblemswiththesite,priorWebpurchasehistory,education,andincome.Insup-portofthemanipulation,perceivedWebsiteinvestmentwashigheramongthoseinthehigh-investmentconditionthanamongthoseinthelow-investmentcondition(Ms=3.46versus1.97;F(1,108)=32.45,p<.0001).Trustingbeliefsandonlinepurchaseintentions.WeusedhierarchicalregressiontotestH1.Wefirstmodeledn.s.).WefoundfurthersupportforH3usingthecriteriathatArraySobel(1982)endorsedfortestingmediation(GoodmanIteststatistic=2.22;p<.05).Becauseinvestmentdidnotaffectbenevolenceandintegritybeliefs,thesebeliefscannotbeconsideredmediators.ConclusionsStudy1providessupportforH1–H3.Specifically,searchers’onlinepurchaseintentionswereinfluencedbytheirtrustinthefirm’sabilityratherthantheirtrustinitsbenevolenceandintegrity.Asaresult,abilitysignals(i.e.,Websiteinvestment)influencedtheironlinepurchaseinten-tions.Furthermore,theirtrustinthefirm’sabilitymediatedthiseffect.Itispossiblethatsearchers’onlinepurchaseintentionswereinfluencedbytheirabilitybeliefsratherthanbytheirbenevolenceandintegritybeliefsbecausetherewerenosig-nalsregardingthefirm’sbenevolenceandintegrity.Inthepresenceofsuchsignals,abilitybeliefsmayhavelessinflu-ence.Indeed,theimpactofagivensignalisweakenedwhenother,morerelevantsignalsarepresent(KirmaniandWright1989).Ifbenevolenceandintegritysignalsaremorerelevant,theeffectofinvestmentonsearchers’onlinepur-chaseintentionsshouldbeweakerwhensuchsignalsarepresent.However,ifabilityismorerelevant,investmentshouldaffectsearchers’onlinepurchaseintentionsregard-lessofwhetherbenevolenceandintegritysignalsarepre-sent.WedirectlytestthisinStudy2bymanipulatingthepresenceandstrengthofafirm’sprivacyandsecuritystate-ment.Becausebenevolencerepresentsthefirm’sorientationtowardcustomersandintegrityrepresentswhetherthefirmwilldowhatitpromises(Mayer,Davis,andSchoorman1995),onemethodofsignalingafirm’sbenevolenceandintegritymaybethroughformalstatementsofitsintentionstoconsumers,suchasthroughprivacyandsecuritystate-ments.Ifso,suchstatementsshouldaffectconsumers’beliefsaboutafirm’sbenevolenceandintegrityratherthanitsability.Yetifabilityisastrongerdriverofsearchers’TABLE1HierarchicalRegressionAnalysisofTrustingBeliefsonOnlinePurchaseIntentionsModel1Model2StudyGoalVariableβR2FforR2βΔR2FforΔR21SearchersAbility.27*.078.31*.19.01.38Benevolence.10Integrity.022SearchersAbility.39*.1515.57*.39*.00.08Benevolence.06Integrity–.063SearchersAbility.42*.1815.70*.43*.031.50Benevolence.15Integrity–.20BrowsersAbility.15.021.62.10.104.28*Benevolence.33*Integrity–.01*p<.05.ConvertingWebSiteVisitorsintoBuyers/137138/JournalofMarketing,April2006andintegritybeliefs(F(2,69)=4.03,p<.05)butnotforabilitybeliefs(F(2,69)=2.34,n.s.).Benevolenceandintegritybeliefsweresignificantlyhigheramongthosewhoreadastrongstatementthanamongthosewhoreadaweakstatement(benevolence:Ms=3.04versus2.65;F(2,68)=2.89,p<.05[one-tailedtest];integrity:Ms=3.19versus2.66;F(2,68)=7.12,p<.05).Thesebeliefswerealsohigherforthosewhoreceivedastrongstatementthanforthosewhoreceivednostatement(benevolence:Ms=3.04versus2.44;F(2,68)=6.55,p<.05;integrity:Ms=3.19versus2.76;F(2,68)=5.02,p<.05).However,themerepresenceofaprivacy/securitystatementdoesnotappeartosignalafirm’sbenevolenceandintegrity;thedifferencebetweenaweakstatementandnostatementwasnotsignifi-cantatp<.05.Itappearsthatboththepresenceandthestrengthofthestatementsignalthefirm’sbenevolenceandintegrity.Websiteinvestmentandonlinepurchaseintentions.Weanalyzedonlinepurchaseintentionswitha2×3ANCOVA.InsupportofH2,onlinepurchaseintentionswerehigheramongthosewhovisitedthehigh-investmentsitethanamongthosewhovisitedthelow-investmentsite(Ms=–.16versus–.79;F(1,69)=3.06,p<.05[one-tailedtest]).Aprivacy/securitystatementdidnotmoderatethiseffect(F(2,69)<1).Thus,regardlessofwhetherbenevolenceandintegritysignals(i.e.,privacy/securitystatements)werepre-sent,abilitysignals(i.e.,Websiteinvestment)significantlyinfluencedsearchers’onlinepurchaseintentions.Theonlyothersignificanteffectwasaprivacy/securityeffect(F(1,69)=4.95,p=.01).Insupportoftheargumentthatbenevolenceandintegritysignals(e.g.,astrongprivacy/securitystatement)shouldhavelittleeffectonsearchers’onlinepurchaseintentions,onlinepurchaseintentionsdidnotdifferbetweenthosewhoreceivedthestrongprivacy/securitystatementandthosewhoreceivednostatement(Ms=–.36versus.09;F(1,46)=1.10,n.s.).However,onlinepurchaseintentionswereloweramongthosewhoreceivedaweakprivacy/securitystatementthanamongthosewhoreceivedastrongstatement(Ms=–1.15versus–.36;F(1,41)=3.82,p<.05[one-tailedtest])ornostatement(Ms=–1.15versus.09;F(1,49)=8.58,p<.01).Thisfindingisconsistentwithexistingresearchthatnega-tivecuesregardingafirm’scharactertendtobeunexpected,violatingestablishednormsabout“businessasusual”(Garfinkel1963).Insuchsituations,peopleappearunwill-ingtobuyonlinefromthefirm.Totestwhetherabilitybeliefsmediatetheinvestmenteffectononlinepurchaseintentions(H3),weaddedabilitybeliefsasacovariatetothe2×3ANCOVA.Consistentwiththerequirementsformediation,abilitywasasignifi-cantcovariate(F(1,68)=8.47,p<.01),andtheinvestmenteffectbecamenonsignific