Full article: Celebrity vs. Influencer endorsements in advertising

Celebrity endorsements are a popular way for marketers to promote their brands, products, and services. By transferring the positive image and characteristics ... SkiptoMainContent Searchin: ThisJournal Anywhere Advancedsearch InternationalJournalofAdvertising TheReviewofMarketingCommunications Volume39,2020-Issue2 Submitanarticle Journalhomepage 143,399 Views 234 CrossRefcitationstodate 0 Altmetric Listen Articles Celebrityvs.Influencerendorsementsinadvertising:theroleofidentification,credibility,andProduct-EndorserfitAlexanderP.SchoutenDepartmentofCommunication&Cognition,TilburgUniversity,Tilburg,theNetherlandsCorrespondencea.p.schouten@uvt.nlViewfurtherauthorinformation,LoesJanssenDepartmentofCommunication&Cognition,TilburgUniversity,Tilburg,theNetherlandsViewfurtherauthorinformation&MaeganVerspagetDepartmentofCommunication&Cognition,TilburgUniversity,Tilburg,theNetherlandsViewfurtherauthorinformation Pages258-281 Received22Aug2018Accepted13Jun2019Publishedonline:02Jul2019 Downloadcitation https://doi.org/10.1080/02650487.2019.1634898 CrossMark InthisarticleInthisarticleAbstractIntroductionTheoreticalbackgroundStudyoverviewStudy1Study2DiscussionConclusionDisclosurestatementAdditionalinformationFootnotesReferencesAppendixes Articles Celebrityvs.Influencerendorsementsinadvertising:theroleofidentification,credibility,andProduct-Endorserfit FullArticle Figures&data References Supplemental Citations Metrics Licensing Reprints&Permissions PDF EPUB AbstractAbstractIntheirmarketingefforts,companiesincreasinglyabandontraditionalcelebrityendorsersinfavorofsocialmediainfluencers,suchasvloggersandInstafamouspersonalities.Theeffectivenessofusinginfluencerendorsementsascomparedtotraditionalcelebrityendorsementsisnotwellunderstood.Therefore,thepresentresearchinvestigatedtheimpactofcelebrityvs.influencerendorsementsonadvertisingeffectiveness(attitudestowardtheadvertisementandproduct,andpurchaseintention),moderatedbyproduct-endorserfit.Moreover,thisresearchinvestigatedtwopotentialmediatorsunderlyingthisrelationship:identification(perceivedsimilarityandwishfulidentification)andcredibility(trustworthinessandexpertise).Twoexperiments(N = 131,N = 446)investigatedcelebrityvs.influencerendorserswithgoodvs.poorfitwithabeautyandafitnessproduct(Study1),orafoodandafashionproduct(Study2).Overall,ourresultsshowedthatparticipantsidentifymorewithinfluencersthancelebrities,feelmoresimilartoinfluencersthancelebrities,andtrustinfluencersmorethancelebrities.Intermsofadvertisingeffectiveness,similarity,wishfulidentification,andtrustmediatetherelationshipbetweentypeofendorserandadvertisingeffectiveness.Product-endorserdidnotexplaintherelationshipbetweentypeofendorserandanyofthemediatinganddependentvariables.Inall,ourresultsshowtheaddedvalueofusinginfluencerendorsersovercelebrityendorsersandtheimportanceofsimilarity,identificationandtrustinthisprocess.Keywords:socialmediainfluencersinfluencerendorsementcelebrityendorsementidentificationcredibilityproduct-endorserfitadvertisingeffectiveness IntroductionCelebrityendorsementsareapopularwayformarketerstopromotetheirbrands,products,andservices.Bytransferringthepositiveimageandcharacteristicsofacelebrityontothebrand,marketersaimtotriggerconsumers’intenttopurchaseorusetheendorsedproductorservice(AtkinandBlock1983).Manyacademicstudieshaveconfirmedthatcelebrityendorsementssignificantlyincreaseadvertisingeffectiveness(e.g.,AtkinandBlock1983;Erdogan1999;Amos,Holmes,andStrutton2008;BergkvistandZhou2016).However,inadditiontousing‘traditional’celebritiessuchasactors,supermodels,andathletestoaddvaluetotheirbrand,companiesincreasinglyturntosocialmediainfluencers(alsocalled‘micro-celebrities’),suchasvloggersand‘Instafamous’personalities,toendorsetheirbrands(Marwick2015).Incontrasttotraditionalcelebritieswhohavegainedpublicrecognitionbecauseoftheirprofessionaltalent,socialmediainfluencers(fromnowoncalled‘influencers’),havegainedfamebysuccessfullybrandingthemselvesasexpertsonsocialmediaplatforms(Khamis,Ang,andWelling2017).Byenthusiasticallysharingself-generatedcontentontopicslikebeauty,fitness,food,andfashion,these(mostlyfemale)socialmediausershavegainedalargefollowerbase,turningtheironlinesocialpresenceintoaprimaryprofessionsuchas‘fashionblogger’,or‘fitgirl’(Lin,Bruning,andSwarna2018).Despitetheincreasingdeploymentofinfluencersforbrandendorsement,scientificknowledgeontheirmarketingvalueislimited.Whereastheeffectivenessoftraditionalcelebrityendorsementhasbeenwidelyinvestigated(BergkvistandZhou2016),theimpactofinfluencerendorsementonbrandresponsesisrelativelyunderstudied.Althoughrecentqualitativeresearchsuggeststhatinfluencers,comparedtocelebrities,haveastrongerimpactonbrandattitudesandpurchasebehaviorsofyoungconsumers(DjafarovaandRushworth2017),nostudytodatehasdirectlycomparedthetwoendorsementtypes.Scholarsaswellaspractitionerswouldbenefitfrommoreinsightintotheeffectsofcelebrityversusinfluencerendorsementsonadvertisingeffectiveness.Therefore,intwoexperiments,thepresentresearchwillinvestigatetheeffectivenessofendorsementadvertisingbycelebritiesversusinfluencersintermsofattitudetowardtheadvertisement,attitudetowardtheadvertisedproduct,andpurchaseintention.Inaddition,weinvestigatetwoimportantmediatingprocessesthatmayexplainpotentialdifferencesineffectivenessbetweenthesetwotypesofendorsements:identification(inparticular,perceivedsimilarityandwishfulidentification;cf.HoffnerandBuchanan2005)andcredibility(consistingoftrustworthinessandexpertise;cf.Sternthal,Phillips,andDholakia1978).Bothidentificationandcredibilityhavebeenshowntoplayakeyroleinexplainingcelebrityendorsementeffects,andmayplayasignificantroleininfluencerendorsementaswell(ChappleandCownie2017;DjafarovaandRushworth2017).Finally,sincetheeffectivenessofanendorsedadvertisementisinseparablytiedtothedegreeinwhichtheendorserfitstheadvertisedproduct(e.g.,KaminsandGupta1994;TillandBusler2000;Fink,Cunningham,andKensicki2004),wewilltakeproduct-endorserfitintoaccountasamoderatorintherelationshipbetweentypeofendorserandmarketingoutcomes.Wewillarguethattheeffectofproduct-endorserfitonadvertisingeffectivenessmaybeevenmorepronouncedforinfluencersthanforcelebrityendorsers.TheoreticalbackgroundInfluencersattractmillionsoffollowersbysharingcontentcuratedfromtheirdailylivesonplatformslikeInstagramandYouTube,evolvingaroundoneparticulardomainofinterest(ChappleandCownie2017;MediaKix2017).Although‘traditional’celebritieshavealsofoundtheirwaytosocialmedia,influencersbuilttheircareersonlineandwereunknowntoageneralpublicbefore.Marketershaveeagerlyembracedtheseinfluencersasspokespersonsfortheirbrands,andadvertisersareinvestinglargebudgetsoninfluencerendorsements(WFA2018).Influencersdonotonlyhavethepowertodirectlyinfluencethepurchasedecisionsofalargeaudience,buttheirfollowersalsojudgethemtobereliableinformationsources(DeVeirman,Cauberghe,andHudders2017;DjafarovaandRushworth2017).Agrowingbodyofacademicresearchisinvestigatingthemeritsofinfluencermarketing,andtheprocessesplayingaroleininfluencereffectsonbrandresponses.Forexample,LeeandWatkins(2016)showedthatvloggerspositivelyaffectconsumerpurchaseintentionsfor(luxury)brandspromotedintheirvlogs.Likewise,inaninterviewstudybyChappleandCownie(2017),consumersstatedtoregularlyfollowlifestylevloggers’productrecommendations,eitherbybuyingaproductthemselvesorrecommendingittoothers.Inthisstudy,participantsconsideredvloggersascrediblesourcesofinformation,mainlydrivenbyperceptionsoftrustandsimilarity.Inanotherstudy,CollianderandDahlén(2011)foundthatablogpostaboutafashionbrandresultedinhigherbrandattitudeandincreasedpurchaseintentcomparedtoanonlinemagazinearticleonthesametopic,becausereadersfeltclosertotheblogger.Althoughnostudyhasdirectlycomparedtheimpactofinfluencerendorsementstomoretraditionalformsofendorsementadvertising,interviewswithfemaleInstagramusers(DjafarovaandRushworth2017)suggestthatinfluencersareperceivedasmorecredibleandrelatablethantraditionalfemalecelebrities,andtheirproductreviewshaveasignificantimpactonyoungfemales’purchasingbehavior.Inanearlierstudycomparingtheeffectivenessofcelebrityendorsementswithproductreviewsofanunknown‘ordinary’customer,femaleparticipantsweremorepositiveaboutapromotedexperienceproductwhenitwaspromotedbyarelatableconsumer(WeiandLu2013).Althoughstillinitsinfancy,scientificresearchoninfluencerendorsementseemstocorroboratetheacclaimedsuccessofinfluencerendorsementsasamarketingtoolandsuggeststhatinfluencersmaynowadayshaveamoresignificantimpactonbrandattitudesandpurchasebehaviorsthantraditionalcelebrities.Wethereforeexpectthatinfluencerendorsementswillresultinhigheradvertisingeffectivenessthancelebrityendorsements.Inthisstudy,wewillmeasureadvertisingeffectivenessintermsofattitudetowardthead,attitudetowardtheadvertisedproduct,andpurchaseintention(cf.KarsonandFisher2005).Insum,wehypothesizethefollowing:H1:Comparedtocelebrityendorsements,influencerendorsementswillleadtohigherattitudetowardthead,attitudetowardtheproduct,andpurchaseintention.TheroleofidentificationandcredibilityExistingresearchonendorsementmarketinghasidentifiedtwomajorprocessesthatmayunderlietheeffectofbrandendorsementonadvertisingeffectiveness:identificationwiththeendorser(e.g.,Basil1996)andperceivedendorsercredibility(e.g.,Ohanian1991).Sincebothprocesseshavebeensuggestedtoplayasignificantroleininfluencerendorsementsaswell(ChappleandCownie2017;DjafarovaandRushworth2017),wewillinvestigatetowhatextentidentificationandcredibilitycanexplaininfluencerversuscelebrityendorsementeffects.Whenconsumersbelievethattheysharecertaininterests,values,orcharacteristicswithanendorser,theyaremorelikelytoadopttheirbeliefs,attitudes,andbehaviors(Cialdini1993;Kelman2006).Identificationderivesfrombothactualandperceivedsimilarity,orthedegreetowhichoneperceivestohavethingsincommonwithanotherperson,aswellaswishfulidentification,whichisthedesiretobeliketheotherperson(HoffnerandBuchanan2005).Inthecaseofcelebrityendorsements,identificationmostlyarisesfromwishfulidentification,oranindividual’saspirationtobelikethatcelebrity(Kaminset al.1989).Incontrast,weexpectidentificationwithinfluencerstobemorestronglydeterminedbyperceivedsimilarity(Gräve2017).Incomparisontocelebrities,influencersareperceivedasmorerelatableandapproachable,likehavingalong-distancefriend(DjafarovaandRushworth2017).Influencerstendtodirectlyaddresstheirfollowersintheirposts,whichconnotesacertaincloseness,andmakesfollowersseethemaspeers(ErzandChristensen2018;GannonandProthero2018).Theabilitytocommentoninfluencers’postsandthepossibilityforinteractionmaystrengthenthefeelingthattheinfluencerissimilartooneself(Schmidt2007).Insum,unlikecelebrities,influencerspresentthemselveslike‘ordinary’,approachable,andauthenticpersonalities(ChappleandCownie2017),whichcouldmakepeoplefeelmoresimilartothem.However,whenitcomestowishfulidentification,consumersmaybemoreattractedtotheglitterandglamourousfameoftraditionalcelebrities.Sincepeoplearemorelikelytoacceptproductclaimscommunicatedbyendorserstheycanidentifywith(Basil,1996),weexpectbothtypesofidentificationtopositivelyaffectadvertisingeffectiveness.Inthisstudy,wewillmeasureadvertisingeffectivenessintermsofattitudetowardthead,attitudetowardtheadvertisedproduct,andpurchaseintention(cf.KarsonandFisher2005).Insum,wehypothesizethefollowing:H2a:Influencerendorsementsleadtoahigherperceivedsimilaritywiththeendorserthancelebrityendorsements.H2b:Celebrityendorsementsleadtomorewishfulidentificationwiththeendorserthaninfluencerendorsements.H2c:Perceivedsimilarityandwishfulidentificationmediatetherelationshipbetweeninfluencervs.celebrityendorsementsandattitudetowardthead,attitudetowardtheproduct,andpurchaseintention.Asecondprocessthathasbeenshowntoplayaroleintherelationshipbetweenendorserandadvertisingeffectivenessisperceivedendorsercredibility(Sternthal,Phillips,andDholakia1978;Ohanian1991).Ingeneral,researchonendorsementeffectshasshownthatconsumersaremorelikelytopositivelyevaluatebrandsandproductsendorsedbypeoplethattheyperceivetobecredible(Erdogan1999;BergkvistandZhou2016).Credibilityconsistsoftwocomponents:trustworthinessandexpertise(Sternthalet al.1978).Trustworthinessreferstoperceptionsofhonesty,integrity,andbelievabilityofanendorser,whereasexpertisereferstotherelevantknowledge,skills,orexperiencetheendorserisperceivedtobepossessing(Erdogan1999).Fortraditionalcelebrities,trustworthyendorsersseemtohavemorepersuasivepowerthanuntrustworthyendorsers(e.g.,PriesterandPetty2003),andperceivedendorserexpertisehasbeenfoundtopositivelyaffectproductattitudesandpurchaseintentions(e.g.,Ohanian,1991;EisendandLangner2010).Regardinginfluencers,studiesinthemoregeneraldomainofelectronicword-of-mouth(eWOM)haveshownthattheeffectivenessofeWOMonconsumers’productattitudesandpurchaseintentionsisdeterminedbyendorsercredibility(Reichelt,SievertandJacob2014;ErkanandEvans2016).Forinfluencersinparticular,findingsfromChappleandCownie(2017)andDjafarovaandRushworth(2017)suggestthatinfluencercredibilityplaysanimportantroleinaffectingpurchasebehavior.Inthepresentstudy,wearguethatinfluencersmaybeperceivedasmorecredibleproductendorsersthancelebrities.First,influencersareknowntoshareuser-orientedproductreviews,recommendations,andpersonalexperiencesontheirsocialmediachannels,suchasbeautyinfluencersdemonstratingmake-uparticlesintheirvlogs1.Althoughagrowingamountofthiscontentiscompany-sponsoredanddesignedtopersuade,themajorityofinfluencer-generatedcontentreflectshonestopinionsanddoesnothavepromotionalgoals(Evanset al.2017).Incontrast,consumersaregenerallywellawarethatcelebritiesgetpaidfortheirendorsements(FriestadandWright,1994).Second,influencersgenerallypromoteproductsinauthentic,real-lifesettings,whichmayincreaseperceptionsoftrustworthinessascomparedtocelebrities.UzunoğluandKip(2014)foundthatbloggersseemtrustworthybecausebypostingaboutcertainbrandstheydemonstratethattheyhavetriedtheproductsthemselves.RussellandRasolofoarison(2017)foundthatwhencelebritiesdoendorseaproductinamoreauthenticway(beingassociatedwiththeproductinareal-lifesetting),theyareperceivedasmorecrediblecomparedtomorecommercialformsofendorsement.Hence,wheninfluencersendorseaproduct,consumersmaybemorelikelytoattributethistotheendorsergenuinelybelievinginthepositivecharacteristicsoftheproductthanwhenacelebrityendorserisused(cf.ZhuandTan2007).Third,aninherentpartofinfluencers’successisthattheyhavebeenabletoestablishacareerbydevotingthemselvestoaparticulardomainofinterestandcreatetheirownexpertprofession(Balog,Rijke,andWeerkamp2008;ErzandChristensen,2018).Theirself-acclaimedexpertisecouldmakeitmorelikelyforinfluencerstobeperceivedasknowledgeableonproductsandservicesintheirdomainofinterestthantraditionalcelebrities.Basedonthesepropositions,westatethefollowinghypotheses:H3a:Influencerendorsementsleadtoahigherperceivedtrustworthinessandexpertisethancelebrityendorsements.H3b:Perceivedtrustworthinessandexpertisemediatetherelationshipbetweeninfluencervs.celebrityendorsementsandattitudetowardthead,attitudetowardtheproduct,andpurchaseintention.Product-EndorserfitInadditiontoidentificationandcredibilityaspotentialexplanatoryprocessesofinfluencervs.celebrityendorsementeffects,weinvestigatethemoderatingroleofproduct-endorserfit.Usinganendorserinanadvertisementdoesnotautomaticallyguaranteeasuccessfuladvertisement.Previousresearchhasdemonstratedthattheeffectivenessofanendorsedadvertisementisinseparablytiedtothedegreeinwhichtheimage,personality,orexpertiseoftheendorserfitstheadvertisedproduct(e.g.,Kamins1990;KaminsandGupta1994).Multiplestudieshaveconfirmedthatwhenanendorser’sperceivedexpertisematchestheproductendorsed,thisenhancesproductevaluationandpurchaseintention(e.g.,TillandBusler2000;Finket al.2004).Moreover,endorsersthatadvertiseproductsthatdonotfittheirexpertisearedeemedlesscredible(DwivediandJohnson2013;LeeandKoo2015).Forinfluencers,endorserrelevancetotheproductwasmentionedbyconsumersasimportanttoendorsementsuccessaswell(DjafarovaandRushworth2017).Indeed,theeffectofproduct-endorserfitoncredibilitymaybeevenmorepronouncedforinfluencersthanforcelebrityendorsers.Sinceinfluencershavesuccessfullybrandedthemselvesasrepresentativeofaparticulardomainofinterest,suchas‘beautyvlogger’,andregularlyshareproductinformationwiththeirfollowers(Baloget al.2008),anassociativelinkbetweenproductandendorsermaybemoreeasilyestablished.Therefore,ascomparedtocelebrities,influencersmaybemorelikelytobefrowneduponwhentheyendorseproductsthatdonotfittheirspecific‘niche’specializationsandmayonlybeperceivedascredibleinformationsourcesinendorsementsthatfitwellwiththeirdomainofinterest.Agoodfitbetweentheendorserandtheproductcouldthereforebeofhigherimportanceforinfluencersthanforcelebrities,whichresultsinthefollowinghypothesis:H4:Theeffectofproduct-endorserfitontrustworthinessandexpertiseismorepronouncedforinfluencerendorsersthanforcelebrityendorsers.StudyoverviewWeconductedtwostudiestotestourhypotheses,usingdifferentendorsersandproductsineachstudy.AllhypothesesarevisualizedintheconceptualmodelinFigure1.Inbothstudies,wemanipulatedtypeofendorsementbycreatingadvertisementswitheitheracelebrityorinfluencerendorser.Fitwasmanipulatedbypairingtheendorserwithaproductthatfitordidnotfitwiththeendorser’sprofession.Theproductsinbothstudieswereselectedtobeexperiencegoodsastheirproductcharacteristics(suchasquality)aredifficulttodeterminepre-purchase,sopeopletendtorelymoreontheevaluationsofothers(ParkandLee2009).Celebrityvs.Influencerendorsementsinadvertising:theroleofidentification,credibility,andProduct-EndorserfitAllauthorsAlexanderP.Schouten,LoesJanssen&MaeganVerspagethttps://doi.org/10.1080/02650487.2019.1634898Publishedonline:02July2019Figure1.ConceptualModel.DisplayfullsizeFigure1.ConceptualModel.Celebrityvs.Influencerendorsementsinadvertising:theroleofidentification,credibility,andProduct-EndorserfitAllauthorsAlexanderP.Schouten,LoesJanssen&MaeganVerspagethttps://doi.org/10.1080/02650487.2019.1634898Publishedonline:02July2019Figure2.Exampleofaninfluencerendorsementwithpoorfit:Aproteinshakeendorsedbyabeautyblogger(blurredforlegal&copyrightreasons).DisplayfullsizeFigure2.Exampleofaninfluencerendorsementwithpoorfit:Aproteinshakeendorsedbyabeautyblogger(blurredforlegal&copyrightreasons).Thefirststudywasbasedonaconveniencesampleof131Dutchadults,consistingmostlyofstudents.Thesecondstudyusedarandomsampleof446 USadultscollectedthroughAmazonMTurk.Bothstudiesfocusedonwomenonlyasthemajorityofinfluencerstailortoafemaleaudience(GannonandProthero2018;IndaHashLabs2017).Moreover,bothstudiesfocusedonwomenbetweentheagesof18and40asthisconstitutesthemostcommonInstagramdemographic,with64.1%ofwomenbetweentheagesof18and40usingInstagramcomparedtoonly22.6%ofwomenover40(PewResearchCenter2018).DescriptionsofbothsamplescanbefoundinTable1.Celebrityvs.Influencerendorsementsinadvertising:theroleofidentification,credibility,andProduct-EndorserfitAllauthorsAlexanderP.Schouten,LoesJanssen&MaeganVerspagethttps://doi.org/10.1080/02650487.2019.1634898Publishedonline:02July2019Table1.SampledemographicsforStudy1andStudy2. DownloadCSVDisplayTableStudy1MethodDesign&manipulationStudy1employeda2(typeofendorsement:celebrityvs.influencerendorsement)X2(product-endorserfit:goodvs.poorfit)between-subjectsdesign.Moreover,asawithin-subjectfactor,participantswereshownbothanadvertisementforabeautyproduct(tintedmoisturizer)andanadvertisementforafitnessproduct(proteinshake).Thetwomanipulatedadvertisementsusedinthemainexperimentconsistedofexistingadvertisements(butunknowntoourparticipants)ofthebeautyandfitnessproducts,combinedwiththepictureoftheendorser.Forbothproductcategories,twoinfluencersandtwocelebritieswereselectedbasedonapretest(seebelow).Thecelebritygood-fitconditionsconsistedofeitheranadvertisementofatintedmoisturizer(beautyproduct)combinedwiththepictureofKateMoss,afamoussupermodel,oranadvertisementofaproteinshake(fitnessproduct)combinedwiththepictureofFatimaMoreiradeMelo,afamousDutchfieldhockeyplayer.Intheinfluencergood-fitconditions,theadvertisementofthetintedmoisturizerwasshownwithapictureofSerenaVerbon,abeautyinfluencer,andtheadvertisementoftheproteinshakewascombinedwithapictureofKaylaItsines,afitnessinfluencer.ThebeautybloggeriswellknownintheNetherlandswith119.000followersonInstagramand30.000uniquedailywebsitevisitors.Thefitnessinfluencerhasover10millionInstagramfollowersworld-wide.Inthepoor-fitconditions,weswappedtheproductsthatthecelebritiesorinfluencersadvertised,sothat,forexample,thebeautyinfluencerwasshownadvertisingthefitnessproduct(seeFigure2foranexampleandseetheappendixforallstimuli).PretestApretest(n = 33)wasconductedtoexaminewhichinfluencersandcelebritiescouldserveasgoodandpoorfitswithbeautyandfitnessproducts.Participantswerepresentedwithnamesandpicturesofsixinfluencerendorsersandsixcelebrityendorsers(threeforeachproductcategory),whichwereselectedbasedonourownknowledgeoffamouscelebritiesandwell-knowninfluencersinthebeautyandfitnessindustryandaGooglesearchforfamousinfluencersandcelebritiesinbothindustries.Foreachendorser,participantsindicatedtheperceivedfitofthepresentedcelebrityorinfluencerwithbeautyproductsandtheperceivedfitwithfitnessproductson7-pointscales(1 = notatall,7 = verywell).Forthemainexperiment,foreachproductcategorywechosethecelebrityandinfluencerwiththehighestfitscore.ProcedureAllparticipantswerepresentedwithtwostimuli,aftereachofwhichtheyfilledinaquestionnaire.First,participantswererandomlypresentedwitheitherthemanipulatedadvertisementforthefitnessproductortheadvertisementforthebeautyproduct.Thereafter,theywereshowntheadvertisementfortheotherproductcategory.Foreachproduct,participantswererandomlyassignedtooneofthefourconditions.Bothnameandprofessionoftheendorserwereshownbeforepresentingtherespectiveadvertisement.Thisway,participantscouldstillevaluatetheadvertisementsevenwhentheydidnotrecognizetheendorser.MeasuresWishfulidentificationandperceivedsimilarityweremeasuredwithtwosubscalesfromHoffnerandBuchanan(2005),onascalefrom1(totallydisagree)to7(totallyagree).Wishfulidentificationwasmeasuredwithfouritems:‘[nameendorser]isthetypeofpersonIwanttobelikemyself,’‘sometimesIwishIcouldbemorelike[nameendorser],’‘[nameendorser]issomeoneIwouldliketoemulate,’and‘I’dliketodothekindofthings[nameendorser]does’.Perceivedsimilaritywasalsomeasuredwithfouritems:‘[nameendorser]thinkslikeme,’‘[nameendorser]behaveslikeme,’‘[nameendorser]islikeme,’and‘[nameendorser]issimilartome’.Factoranalysesshowedthatallitemsloadedontheirrespectivefactors,forboththefitnessad(similarity:EV=3.33,R2=41.57%,α=.93;identification:EV=2.68,R2=33.47%,α=.85)andthebeautyad(similarity:EV=3.29,R2=41.11%,α=.93;identification:EV=2.76,R2=34.32%,α=.84).EndorsercredibilitywasassessedbyusingthetrustworthinessandexpertisesubscalesofthecredibilityscalebyOhanian(1990).Participantsratedtheendorser’strustworthinessonfive7-pointsemanticdifferentialscales:undependable–dependable,dishonest–honest,unreliable–reliable,insincere–sincere,anduntrustworthy–trustworthy.Expertisewasalsomeasuredwithfive7-pointsemanticdifferentialscales:notanexpert–expert,inexperienced–experienced,unknowledgeable–knowledgeable,unqualified–qualified,andunskilled–skilled.Allitemsloadedontheirrespectivefactors,forboththefitnessad(trust:EV=3.74,R2=37.38%,α=.91;expertise:EV=3.62,R2=36.24%,α=.90)andthebeautyad(trust:EV=3.92,R2=39.24%,α=.93;expertise:EV=3.92,R2=39.17%,α=.92).AttitudestowardtheadvertisementandproductwereassessedwithascalebasedonSpearsandSingh(2004).Bothwereratedwithfiveitemson7-pointsemanticdifferentialscales:unappealing–appealing,bad–good,unpleasant–pleasant,unfavorable–favorable,andunlikable–likable.Factoranalysesyieldedtwofactorscorrespondingwithbothsubscales,forboththefitnessad(Attitudead:EV=4.07,R2=40.69%,α=.94;Attitudeproduct:EV=4.32,R2=43.16%,α=.96)andbeautyad(Attitudead:EV=4.34,R2=43.42%,α=.95;Attitudeproduct:EV=4.38,R2=43.78%,α=.97).Finally,purchaseintentionwasmeasuredwithoneitemasking‘Thenexttimeyouarelookingtopurchaseatintedmoisturizer/proteinshake,howlikelyareyoutopurchasethisproduct?’,ratedona7-pointscalefrom1(veryunlikely)to7(verylikely).Toensurethevalidityofourmanipulationofproduct-endorserfit,weaskedparticipantstoindicateperceivedfitbetweenthepresentedproductandendorserona7-pointscalerangingfrom1(totallydisagree)to7(totallyagree).Inaddition,familiaritywiththeendorserwasmeasuredonasimilarscale:‘Iknow[nameendorser].’ResultsManipulationchecksOfourparticipants,54.7%ofrespondentswereatleastsomewhatfamiliar(4orhigheronthescale)withtheDutchhockeyplayer,and68.0%werefamiliarwiththesupermodel.Incontrast,37.5%ofrespondentswerefamiliarwiththefitnessinfluencer,and39.3%werefamiliarwiththebeautyinfluencer.Forboththefitnessproduct(Mgoodfit=4.65,SD = 1.54;Mpoorfit=3.18,SD = 1.64)andthebeautyproduct(Mgoodfit=5.86,SD = 1.24;Mpoorfit=4.29,SD = 1.61),thedataconfirmedthattheadvertisementswithgoodproduct-endorserfitwereindeedconsideredabetterfitthantheadvertisementswithpoorproduct-endorserfit,tfitnessproduct(129)=-5.30,p<.001 downloadcsvdisplaytablecelebrityvs.influencerendorsementsinadvertising:theroleofidentification downloadcsvdisplaytableinordertotestthemediationhypothesesputforwardinh2candh3b>.25.EffectsofendorsertypeonmediatorsAnalysesrevealedthatinfluencerendorsementsledtohigherwishfulidentification,F(1,127)=14.99,p<.001>.07.Agoodfitbetweenendorserandproductdidnotresultinhigherratingsoftrustworthinessandexpertisethanapoorfit.However,additionalanalysesshowedthatproduct-endorserfitdiddirectlyaffectattitudetowardthead,F(1,127)=5.53,p=.020,η2=.042,andpurchaseintention,F(1,127)=14.82,p<.001>.070.Study2MethodInStudy2,wefocusedontwootherproductcategoriesthatinfluencersregularlyendorse:foodandfashion.ThedesignandprocedureforStudy2weresimilartothatofStudy1,exceptthateachparticipantreceivedonestimulusonly,resultingina2(typeofendorsement:celebrityendorsementvs.influencerendorsement)X2(product-endorserfit:goodfitvs.poorfit)X2(producttype:foodvs.fashionproduct)between-subjectsdesign.Apretest(n = 46)wasconductedusingAmazonMTurktoselecttheendorsersandproductsforourstimuli.First,participantswereaskedtolistuptofivefemaleendorsersineachofthesefourcategories:celebritiesinthefoodindustry,foodinfluencers,celebritiesinthefashionindustry,andfashioninfluencers.Thefourendorsersthatwerementionedmostofteninthesefourcategorieswereselectedforourmainexperiment,beingRachaelRay(foodcelebrity,mentioned32times),GabyDalkin(foodinfluencer,mentioned10times),KendallJenner(fashioncelebrity,mentioned9times),andDanielleBernstein(fashioninfluencer,mentioned10times).Next,participantswereaskedtoindicateforthreefoodproductsandthreefashionproductshowwelltheyfeltthateachproductwouldfitafood/fashionendorseronafive-pointscalefrom1(Terrible)to5(Excellent).Forourmainexperiment,weselectedthetwoproductswiththehighestperceivedproduct-endorserfit,whichwasastandmixerforthefoodendorsers(M = 4.59,SD = 0.62),andawatchforthefashionendorsers(M = 4.37,SD = 0.90).Inthemainexperiment,inthegoodfitconditionsthefoodcelebrityorfoodinfluencerwerepairedwiththestandmixerandthefashioncelebrityorfashioninfluencerwerepairedwiththewatch.Inthepoorfitconditions,thiswastheotherwayaround(seetheappendixforallstimuli).DependentvariablesweresimilartothoseinStudy1,exceptforincludingtheattractivenesssubscaleofOhanian’s(1990)credibilityscale.Factoranalysisrevealedthatwishfulidentificationandperceivedsimilarityformedasinglefactor,butuponforcingthefactoranalysisintwofactors,thetwosubscalesclearlyemerged,withallitemsloadingontheirrespectivefactors(similarity:EV=3.54,R2=44.29%,α=.95;identification:EV=3.30,R2=41.27%,α=.93).Forcredibility,threefactorsemerged,representingtrustworthiness(EV=4.24,R2=28.29%,α=.93),expertise(EV=4.27,R2=28.47%,α=.94),andattractiveness(EV=3.25,R2=21.65%,α=.88).Finally,attitudetowardsthead(EV=4.46,R2=44.61%,α=.97)andproduct(EV=4.41,R2=44.12%,α=.97)alsoconstitutedtworeliablescales.ResultsManipulationchecksOfourparticipants,91%ofrespondentswereatleastsomewhatfamiliar(4orhigheronthescale)withthefoodcelebrity,while84.5%werefamiliarwiththefashioncelebrity.Incontrast,only29%ofparticipantswerefamiliarwiththefitnessinfluencer,and38.1%werefamiliarwiththefashioninfluencer.Theadvertisementswithagoodproduct-endorserfit(M = 5.15,SD = 1.51)wereindeedconsideredabetterfitthanthepoor-fittingadvertisements(M = 3.84,SD = 1.86),t(444)=-8.16,p<.001.overviewofanalysissimilartostudy1 downloadcsvdisplaytabledirecteffectsofendorsertypeonadvertisingeffectivenessh1posedthatinfluencerend orsementswouldresultsinahighereffectivenessthancelebrityendorsements.thiswasconfirmedforproductinten tiononly>.40.Celebrityvs.Influencerendorsementsinadvertising:theroleofidentification,credibility,andProduct-EndorserfitAllauthorsAlexanderP.Schouten,LoesJanssen&MaeganVerspagethttps://doi.org/10.1080/02650487.2019.1634898Publishedonline:02July2019Table5.Study2Pearsonproduct-momentcorrelationsbetweenallmediatinganddependentvariables,forbothfitnessandbeautyproducts. DownloadCSVDisplayTableEffectsofendorsertypeonmediatorsSimilartotheresultsofStudy1,analysesrevealedthatinfluencerendorsementsledtohigherwishfulidentification,F(1,438)=9,69,p=.002,η2=.022,perceivedsimilarity,F(1,438)=27.39,p<.001>.400.DiscussionThegoalofourstudywastocomparetheeffectivenessofendorsementadvertisingbysocialmediainfluencerstotraditionalcelebrityendorsements.Wehaveinvestigatedourhypothesesintwoexperiments,usingdifferentcelebrityandinfluencerendorsersacrossfourpopularinfluencermarketingdomains(beauty,fitness,food,andfashion)usingdifferentmeasuresofadvertisingeffectiveness.Overall,ourresultsshowedthatparticipantsidentifymorewithinfluencersthanwithcelebrities,feelmoresimilartoinfluencersthancelebrities,andtrustinfluencersmorethancelebrities.Intermsofadvertisingeffectiveness,similarity,wishfulidentification,andtrustworthinessmediatetherelationshipbetweentypeofendorserandadvertisingeffectiveness,albeitnotconsistently.Incontrasttoexpectations,product-endorserfitdidnotmoderatetherelationshipbetweentypeofendorserandexpertiseortrust.Hypothesis1wasconfirmedonlyforpurchaseintention.Participantsweremorewillingtobuyaproductendorsedbyaninfluencerthanbyacelebrity.Nodirecteffectswereobservedforattitudetowardtheadvertisementorproduct.Therefore,althoughparticipantsaremorelikelytobuyproductsfromaninfluencer,theydonotparticularlylikeadvertisementsofinfluencersandtheproductsadvertisedbyinfluencersmorethanthoseofcelebrities.InlinewithH2a,influencerendorsementsledtohigherperceivedsimilaritythancelebrityendorsements.However,contrarytoH2b,influencerendorsementsalsoledtohigherwishfulidentificationwiththeendorserthancelebrityendorsements.Inhindsight,thismaynotcomeasasurprise,asmanypeoplenowaspiretobecomeasocialinfluencer(Chae2018).Becauseinfluencersareseenasmoresimilarto‘ordinarypeople’thancelebrities,perceivedsimilaritymayevenmakeiteasierforpeopletowishfullyidentifywiththem,becauseitiseasiertobelievetheycouldbelikethem.Moreover,UzunoğluandKip(2014)previouslyfoundthatamainreasonforfollowingInstagrambloggerswasacombinationofadmirationontheonehand,andthefeelingofconnectednesstotheinfluencerontheother.H2cposedthatperceivedsimilarityandwishfulidentificationwouldmediatetherelationshipbetweentypeofendorserandattitudetowardthead,product,andpurchaseintention.Overall,wefoundsupportforthishypothesis.InStudy1,perceivedsimilaritymediatedtherelationshipbetweentypeofendorserandattitudetowardtheproductandpurchaseintention,albeitonlyforthefitnessproduct.Anexplanationforwhysimilaritydidnotmediatetherelationshipbetweenendorsertypeandpurchaseintentionforthebeautyproductmaybebecausepeoplefeeltheymaybeabletoachieveasimilarfitnessorshapeastheinfluencerandarethereforemoreinclinedtotrytheproteinshake,whereasobtainingsimilarbeautybysimplyusingatintedmoisturizerismoredifficult.InStudy2,perceivedsimilaritymediatedtherelationshipbetweentypeofendorserandattitudetowardtheadandpurchaseintention.Peoplefeelmoresimilartoinfluencersthancelebritiesandasaresultliketheirendorsementsmoreandaremoreinclinedtopurchasetheendorsedproduct.Similartothefitnessproductinstudy1,theproductsendorsedinstudy2(foodandfashion)mayactuallyallowtheconsumertofeelmoresimilartotheendorser,incontracttobeautyproducts,thereforeresultinginhigherpurchaseintention.Inall,perceivedsimilarityisanimportantexplanationforthesuccessofinfluencerendorsements,butitsroledependsonthetypeofproductsendorsed.Mostlikely,whentheproductendorsedwillnotleadtheconsumertofeelmoresimilar,endorsingproductswillnotbeaseffectiveaswhentheproductendorsedmayactuallymaketheconsumerfeelmoresimilartotheinfluencer(e.g.,Imayfeelasfashionableastheinfluencerbybuyingawatch,butI’llneverbecomeasprettybywearingmake-up).Wishfulidentificationalsomediatedtherelationshipbetweenendorsertypeandattitudetowardtheadandpurchaseintention,albeitonlyinStudy2.ApossibleexplanationmaybethattheproductsweusedinStudy2(awatchandastandmixer)aremoregenerallyappealingtoalargergroupofpeoplethantheproductsinStudy1(atintedmoisturizerandproteinshake),andpurchaseintentionisthereforemorelikelytobeaffectedwhenpeopleaspiretobelikeanendorser.Thismirrorstheexplanationgivenabove:Itislikelythatwishfulidentificationexplainstheeffectivenessofinfluencersendorsementsonlywhenconsumersactuallyfeeltheymaybecomeliketheendorser(i.e.,genuineaspiration),whichismostlikelyeasierintheareasoffoodandfashion,thanitisbecomingfitorpretty.Ourthirdhypothesis(H3a)posedthatinfluencerendorsementswouldleadtohigherperceivedtrustworthinessandexpertisethancelebrityendorsements.Inbothstudies,wefoundthatinfluencerendorserswereindeedseenasmoretrustworthy,whereasnodifferenceswerefoundonexpertise.Anexplanationfornotfindingdifferencesinexpertisecouldbethatinbothstudies,weselectedinfluencersandcelebritiesthatarewell-knownexpertsintheirspecificfields.Inaddition,theexpertisesubscaleusedinourexperimentsassessedgenericexpertise(e.g.,‘thispersonisknowledgeable’)anddidnotmeasurespecificexpertiseontheproductendorsed.Futureresearchisthereforeadvisedtomeasurespecificexpertise,astogainbetterinsightintheroleofexpertiseininfluencerversuscelebrityendorsements.InH3b,weposedthattrustworthinessandexpertisewouldmediatetherelationshipbetweentypeofendorserandadvertisingeffectiveness.Expertisewasnotasignificantmediatorinanyofthetestedrelationships.However,inbothstudies,trustworthinesswasanimportantvariableexplainingwhyinfluencersaremoreeffectiveendorsersthancelebrities.Trustworthinessmediatedtherelationshipbetweentypeofendorserandattitudetowardtheadvertisement(inStudies1and2),attitudetowardtheproduct(inStudies1and2)andpurchaseintention(inStudy1).Overall,ourresultsareinlinewithearlierstudiesthatstresstheroleofauthenticityandtrustininfluencerendorsements(UzunoğluandKip2014).Areasonthattrust,notexpertisemediatedtherelationshipbetweentypeofendorserandadvertisingeffectivenessmaybethattrustisasmuchofasocial-affectiveconstructthanacognitiveconstruct,inthattrustdependsonafeelingofintegrity,believability,andmutualunderstanding(UzunoğluandKip2014).Thus,consumersaremostlikelytotrustaninfluencermoreandarethereforemorepersuadedbyinfluenceradvertisementsbecauseaninfluencerismorelikethemandhencetrustworthy.H4wasnotconfirmed.Wedidnotfindanyinteractioneffectsofproduct-endorserfitandendorsertypeontrustworthinessorexpertise.WealsodidnotfindanymaineffectsofproductfitinStudy1.Anexplanationmaybethatafitnessinfluencerendorsingabeautyproductmaynothavebeenconsideredaverypoorfit,sincethefitnessinfluencerwasanattractivewomanandbeautyproductsaregenerallyendorsedbyprettyendorsers.Instudy2,wedofindmaineffectsofproduct-endorserfitonexpertiseandtrustworthiness,asthedifferencebetweenfoodandfashionproductsisliketobemoremarkedthanthedifferencebetweenbeautyandfitness.Yet,wefindnoevidencethatfitismoreimportantforinfluencersthancelebrities.ConclusionThepresentresearchisoneofthefirsttodirectlycomparecelebrityandinfluencerendorsementsintermsoftheiradvertisingeffectiveness.Wehaveshownthatinfluencersaredeemedmoretrustworthythancelebrities,andthatpeoplefeelmoresimilartoinfluencersandidentifymorewiththemthancelebrities.Theseprocesses,inturn,affectadvertisingeffectiveness.Therefore,afirstimplicationofourstudyisthatinfluencersmaybemoreeffectiveproductendorsersthantraditionalcelebritiesandapracticalrecommendationisthereforetocontinuetouseinfluencerendorsersinmarketingcampaigns.Asecondimplicationistheimportanceofinvestigatingtheprocessesunderlyingeffectiveproductendorsements.Wishfulidentification,similarity,andtrustareimportantexplanationsforwhyproductendorsementsworkandalsoexplainwhyinfluencerendorsementsaremoreeffectivethancelebrityendorsements.However,ourresultsalsoshowthattheunderlyingprocessesexplainingadvertisingeffectivenessmaydependonspecificproduct-endorsercombinations.Specifically,foraninfluencerendorsementtobemoreeffectivethanacelebrityendorsement,anendorsedproductmustbeabletoenhancefeelingsofsimilarityandwishfulidentification.Inall,ourresultsshowthatwhenendorsingproducts,itisimportantforinfluencerstobeperceivedassimilartotheiraudienceandthatidentificationwithaninfluencerneedstobebasedontrueaspirationratherthanonlywishfulthinking.Third,findingnodirecteffectsofendorsertypeonadvertisementandproductattitude,whilewedofindseveralmediationeffectsviatrust,similarity,andidentificationmayindicatethattheremaybeseveralothermediatorsthatmayexplaintherelationshipbetweenendorsertypeandadvertisingeffectiveness.Possibly,thepositiveeffectofinfluencerovercelebrityendorsementsonadvertisementandproductattitudeviathemediatorsmaybemitigatedbyothervariables,suchaslikeabilityoftheinfluencer(DeVeirmanet al.2017;Gräve2017).Although(orbecause)ourstudyisoneofthefirsttoinvestigatetheeffectsofinfluencerendorsementsonadvertisingeffectiveness,itisnotwithoutdrawbacks.First,apotentialimprovementmaybethewaywepresentedtheendorsementstoourrespondents.Inallexperimentalconditions,ourstimuliconsistedofimagesofaproductnexttoanendorser,resemblingabasicadvertisingformat.Althoughthismakesforanexperimentallyvalidcomparisonbetweenendorsertypes,thisisnothowinfluencersonsocialmedianormallyengagewithaproduct.Usually,theproductthatisendorsedispartofalargermessageandisintegratedinasocialmediapost,suchasavlogoranInstagrampost(KapitanandSilvera2016).Theoryonproductplacementteachesusthatintegrationofaproductinastorylineinavisuallyappealingwayispositivelyrelatedtoendorsementeffectiveness(Russell1998).Wethereforerecommendinvestigatingdifferenttypesofsocialmediaendorsementswithdifferentlevelsofproductengagementinfuturestudies.Ontheotherhand,thatarelativelysimpleendorsementasoursyieldedsucheffectsshowsthepotentialpowerofinfluencerendorsementsovercelebrityendorsements.Furthermore,futureresearchcouldexamineothermoderatorsthanproduct-endorserfitinfluencingtherelationshipbetweenendorsertypeandadvertisingeffectiveness.Forexample,asthisstudyonlyincludedexperiencegoods,futureresearchcouldcomparetheeffectsofinfluencervs.celebrityendorsementsonothertypesofproducts,suchassearchgoods.FindingsofWeiandLu(2013)suggestthatincontrasttoexperienceproducts,searchproductslikeapairofbootsaremoreeffectivelyendorsedbycelebritiesthanby‘ordinary’consumers.Futureresearchcouldalsodistinguishbetweeninformationalandtransformationalorutilitarianandhedonicgoods.Finally,inourstudywecomparedendorsementsoftraditionalcelebritieswithendorsementsbysocialmediainfluencers.However,inreality,thisdistinctionisnotalwayssoclear-cut.Numerouscasesareknownofsuccessfulsocialmediainfluencerstransgressingintomore‘traditional’celebrities,pursuingacareerastalkshowpresenterorfashionmodelandmakingtheirwaytothegeneralpublicandmassmedia.Ontheotherhand,manytraditionalcelebritieshavebecomepopularinfluencersonsocialmedia.Thisraisesthequestionwhichtypeofinfluencersarethemostsuccessfulendorsers,andtowhatextentpopularityoftheendorserisanimportantvariableinexplainingendorsereffectiveness.Inourstudiesweusedwell-knowninfluencerswithalargefollowerbase,so-called‘micro-celebrities’,butinfluencerswhoarerelativelylesspopularmaybeevenmoreeffectiveendorsers.Ascomparedtomorepopularinfluencers,‘micro-influencers’maybeexpertsinarelativelysmallfieldandengagewiththeiraudiencemore,andmaythereforebeseenasmoresimilartotheirfollowers(DeVeirmanet al.2017;Gräve2017).Nevertheless,eventhe‘micro-celebrities’weusedinourstudiesappeartobemoreaspirable,relatable,andtrustworthythantraditionalcelebrities,andarethereforemoreeffectivebrandendorsers.Inconclusion,influencerendorsementsaremoreeffectivethancelebrityendorsements,whichcanbeexplainedbyprocessesofwishfulidentification,similarity,andidentification.Foraninfluencerendorsementtobemoreeffectivethanacelebrityendorsement,anendorsedproductmustbeabletoenhancefeelingsofsimilarityandwishfulidentification.Moreover,aninfluencermustelicittrustinordertobeeffective.Influencersarenotdeemedmoreknowledgeablethancelebrityendorsers,andexpertisedoesnotexplainwhyinfluencerendorsementsmaybemoreeffectivethancelebrityendorsements.Product-endorserfithasnoeffectontherelationshipbetweentypeofendorsementandtrust,expertise,oradvertisingeffectiveness,althoughtheeffectivenessofinfluencervs.celebrityendorsementshingesuponspecificinfluencer-productcombinationsthatelicitfeelingsofsimilarity,identification,andtrust.Table1.SampledemographicsforStudy1andStudy2. Study1Study2N131446MeanAge(SD)23.80(4.07)31.53(5.04)Educationlevel(%)   Collegedegree(masterorprofessional)40.517.9 Collegedegree(bachelor)31.346.6 Vocational/Associatedegree18.327.6 Highschool10.07.8UseInstagram(%)an/a81.4Instagramuseinhoursperweek(SD)bn/a3.43(3.52)Note:aWedidnotaskforInstagramuseinStudy1.bCalculatedforInstagramusersbymultiplyingdaysusedperweekanduseperactiveday.Table2.Study1meansandstandarddeviations(inparentheses)forallmediatinganddependentvariablesforthefitnessproductandthebeautyproduct,asafunctionoftypeofendorsementandproduct-endorserfit. Fitnessproduct Celebrityendorsement Influencerendorsement Poorfit(n = 39)Goodfit(n = 36) Poorfit(n = 29)Goodfit(n = 27)Wishfulidentification3.03ab(1.19)2.76a(1.17) 3.36ab(1.30)4.02b(1.29)Perceivedsimilarity2.27a(0.81)2.33a(0.94) 3.09b(1.47)3.51b(1.51)Trustworthiness4.16a(0.68)4.29a(1.24) 4.66ab(0.88)4.91b(0.85)Expertise4.66(0.96)4.76(1.19) 4.62(1.08)5.02(1.12)Attitudetowardtheadvertisement3.65a(1.12)3.80ab(1.07) 3.61a(1.21)4.40b(1.39)Attitudetowardtheproduct3.50a(1.23)3.38a(1.34) 3.30a(1.35)4.36b(1.53)Purchaseintention2.23a(1.56)2.53a(1.44) 2.66a(1.52)4.04b(1.91)Influencerfamiliarity4.13(2.47)4.42(2.38) 3.07(2.19)3.52(2.56)Product-endorserfit3.23a(1.83)4.31b(1.45) 3.10a(1.37)5.11b(1.55) Beautyproduct Celebrityendorsement Influencerendorsement Poorfit(n = 39)Goodfit(n = 36) Poorfit(n = 29)Goodfit(n = 27)Wishfulidentification3.17a(1.36)3.52ab(1.23) 4.19b(1.49)3.79ab(0.78)Perceivedsimilarity2.57a(1.22)2.44a(0.87) 3.59b(1.48)3.82b(1.16)Trustworthiness4.48(1.05)4.48(0.87) 4.83(1.05)4.97(0.92)Expertise4.58(1.22)5.12(0.90) 4.77(1.19)4.88(0.93)Attitudetowardtheadvertisement4.55(1.18)4.87(0.96) 4.68(1.25)4.87(0.89)Attitudetowardtheproduct4.66(1.35)4.99(1.00) 4.84(1.07)4.78(1.11)Purchaseintention3.31a(1.79)3.83ab(1.44) 3.52ab(1.57)4.41ab(1.25)Influencerfamiliarity4.72a(2.09)4.69a(2.33) 3.17b(2.39)3.63ab(2.17)Product-endorserfit4.56a(1.52)6.00b(1.27) 3.93a(1.69)5.67b(1.21)Note:Differentsuperscriptsindicatesignificantdifferencesbetweenconditions(acrossrows).Table3Study1Pearsonproduct-momentcorrelationsbetweenallmediatinganddependentvariables,forbothfitnessandbeautyproducts.  12345678FitnessProduct1Wishfulidentification        2Perceivedsimilarity.604*       3Trustworthiness.448*.538*      4Expertise.373*.386*.665*     5Attitudetowardtheadvertisement.329*.353*.374*.454*    6Attitudetowardtheproduct.432*.495*.402*.442*.686*   7Purchaseintention.429*.601*.415*.411*.492*.654*  8Influencerfamiliarity.219*.247*.197.307*.001.237.152* 9Product-endorserfit.236*.329*.311*.406*.567*.517*.481*.090BeautyProduct1Wishfulidentification        2Perceivedsimilarity.664*       3Trustworthiness.460*.454*      4Expertise.327*.262*.712*     5Attitudetowardtheadvertisement.230*.230*.538*.437*    6Attitudetowardtheproduct.258*.200*.611*.471*.740*   7Purchaseintention.367*.360*.417*.254*.595*.584*  8Influencerfamiliarity.225*.117.088.131.014.028.021 9Product-endorserfit.095.103.332*.422*.452*.439*.428*.124Note:*p<.01.table4.study2meansandstandarddeviations relatedresearchdata whatisinfluencermarketingandhowdoesittargetchildren frontiersmediasa linkingprovidedby disclosurestatementnopotentialconflictofinterestwasreportedbytheauthors.additionalinformationnoteson contributorsalexanderp.schoutenalexanderp.schoutenisanassistantprofessorofbusinesscommunicationanddi gitalmediaattilburguniversity.hisresearchinterestsincludesocialmediause referencesamos alternativeformats pdf epub moreshareoptions relatedresearch peoplealsoreadlistsarticlesthatotherreadersofthisarticlehaveread. recommendedarticleslistsarticlesthatwerecommendandispoweredbyouraidrivenrecommendationengine. citedbylistsallcitingarticlesbasedoncrossrefcitations.articleswiththecrossreficonwillopeninanewtab. peoplealsoread recommendedarticles citedby yourdownloadisnowinprogressandyoumayclosethiswindow didyouknowthatwithafreetaylor choosenewcontentalertstobeinformedaboutnewresearchofinteresttoyou easyremoteaccesstoyourinstitution saveyoursearchesandschedulealertstosendyounewresults exportyoursearchresultsintoa.csvfiletosupportyourresearch haveanaccount loginnow don registerforfree loginorregistertoaccessthisfeature registerafreetaylor register noworlearnmore>