ContentslistsavailableatScienceDirect
International Journal of Information Management
j o ur n a l ho me p a g e :w w w . e l s e v i e r . c o m / l o c a t e / i j i n f o m g t
Predicting smartphone brand loyalty: Consumer value and consumer-brand identification perspectives
Ching-Hsuan Yeh
a, Yi-Shun Wang
a,∗, Kaili Yieh
baDepartmentofInformationManagement,NationalChanghuaUniversityofEducation,Taiwan
bDepartmentofBusinessAdministration,NationalChanghuaUniversityofEducation,Taiwan
a r t i c l e i n f o
Articlehistory:
Received1June2015 Receivedinrevisedform 24November2015 Accepted25November2015
Keywords:
Mobilecomputing Consumervalue
Consumer-brandidentification Brandloyalty
Smartphone
a b s t r a c t
Withthegrowthandcompetitionofthesmartphoneindustry,developingabetterunderstandingof whatdrivesconsumers’loyaltytosmartphonebrandshasbecomeanimportantissueforacademics andpractitioners.Thisstudyhypothesizesfourdeterminantsofsmartphonebrandloyaltybasedonthe perspectivesofconsumervalueandconsumer-brandidentification.Furthermore,thisstudyalsoexplores themoderatingeffectsofageandgenderdifferencesonthedeterminationprocessofsmartphonebrand loyalty.Datacollectedfrom157respondentswastestedagainsttheresearchmodelusingapartialleast squares(PLS)approach.Theresultsindicatethatfunctionalvalue,emotionalvalue,socialvalue,andbrand identificationhaveapositiveinfluenceonsmartphonebrandloyalty.Ofthetwomoderators,resultsshow thatageenhancestheemotionalvalue-brandloyaltyandsocialvalue-brandloyaltylinkagesbutweakens thebrandidentification-brandloyaltyrelationship.However,genderdoesnotplayamoderatingrole inthedeterminationofsmartphonebrandloyalty.Theresultsofthisstudyprovideseveralimportant theoreticalandpracticalimplicationsforsmartphonebrandmanagement.
©2015ElsevierLtd.Allrightsreserved.
1. Introduction
Withthe proliferation of thecompeting brands in themar- ketplace,keepingconsumersloyalisanimperativeformarketing managers (Jones & Sasser, 1995). Researchers have devoted a considerableamountofeffortinvestigatingthisissue.Theyhave advocatedthatthenotionofbrandloyaltyshouldbeextendedfrom patronagebehaviortopsychologicalcommitment(Oliver,1999), andbothattitudinalloyaltyandbehavioralloyaltycontributeto pro-brand consequences. Attitudinal loyalty may be positively associatedwithpatronageintention,word-of-mouth,acceptance of premium price, and resistance to counter-persuasion, while behavioralloyaltymayleadtogreatermarketshareandincreased profitability (Chaudhuri & Holbrook, 2001; Shankar, Smith, &
Rangaswamy,2003).
Standing on various theoretical grounds, researchers have investigated the reasonsfor brand loyalty, but two viewpoints havereceivedgreateramountofattention.Thefirstisconsumer valuetheory,which claims thatvalue perceptionis thepivotal predictorofbrandloyalty(Kim,Gupta,&Koh,2011;Sweeney&
Soutar,2001).Consumersremainloyaliftheyperceivesuperior
∗Correspondingauthor.
E-mailaddress:[email protected](Y.-S.Wang).
valuefromagivenbrand(Hansen,Beitelspacher,&Deitz,2013).
Thesecondviewpointistheidentificationapproach,whichputs consumer-brandidentification(hereafterreferredtoasbrandiden- tification)astheantecedentofbrandloyalty(Bhattacharya&Sen, 2003;Tuˇskej,Golob,&Podnar,2013).Consumersstickwithagiven brandonce theyidentifythemselves withtheattributes of the brand(Stokburger-Sauer, Ratneshwar,&Sen,2012).Somestud- ieshavefurthercontendedthatbothviewpointsmaypositively resultinbrandtrustandthenbrandloyalty(He,Li,&Harris,2012), supportingclaimsthatthesearethefoundationsofbrandloyalty.
Althoughresearchersgenerallyrecognizethepredictivepower ofconsumervalueandbrandidentification,managersmayface adilemmaaboutresourceallocationbecausethetwoviewpoints offerdifferentguidelinesforbusinesspractice.Strategiesderived fromconsumervaluetheoryencouragemanagerstoemphasize product development and to communicate the advantages of theproduct attributestoconsumers (Karjaluoto,Jayawardhena, Leppäniemi,&Pihlström,2012),whereasstrategiesderivedfrom the identification approach may drive managers to create an attractive brand identity and to organize a community for intimate consumer-brand and consumer–consumerinteractions (Stokburger-Saueretal.,2012).Asmanagersmayhavetorecon- cilethesemarketingcampaignstogeneratesynergies,itiscrucial todifferentiatetheeffectsofconsumervalueandconsumer-brand identificationonbrandloyalty.Therefore,anintegratedanalysis http://dx.doi.org/10.1016/j.ijinfomgt.2015.11.013
0268-4012/©2015ElsevierLtd.Allrightsreserved.
Downloaded from http://www.elearnica.ir
withloyaltydeterminantsshouldaidinunderstandingthedeter- minantpriorityandtheallocationofmarketingresources.
Inaddition,studiesonbrandloyaltyhavearguedfortheneces- sityoftaking individualheterogeneity intoconsideration (Floh, Zauner,Koller,&Rusch,2014).Modelvalidationand/orhypothe- sestestingwithentiresamplesmaysufferfromaggregationbias, andtheeffectivenessofmarketingcampaignsmaynotberealized asexpected.Comparedwithmassmarketing,targetedmarketing mightgeneraterevenuesandprofitsmoreefficiently.Assuch,indi- vidualheterogeneityneedsto beconsideredin the analysesof brandloyalty.Amoderationspecificationofindividualheterogene- itycanhelpmanagerstailorbetterloyalty programsand enrich existing knowledge. A review of prior studies reveals that age andgenderaretwotypicalvariablesofindividualheterogeneity (Venkatesh&Morris,2000),and thisstudyinvestigates whether andhowtheeffectsofloyaltydeterminantsdifferacrossageand gendercategories.
Tosumup,thisstudyhastwomainobjectives:
1.Tounderstandtherelativeinfluenceofdeterminants(i.e.,value andidentification)onbrandloyalty.
2.To understand the moderating effects of individual hetero- geneity (i.e., age and gender) on the relationships between value/identificationandloyalty.
Toachievethetwoobjectives,thisstudyusesthesmartphone industryinTaiwanastheresearchcontext.InTaiwan,smartphones haveoverwhelmed feature phonesin use, and hold71% ofthe mobilephonemarketin termsofsupplyinthethirdquarterof 2012(InformationDataCenter,2012).ArecentsurveyfromGoogle (2013)reportedthatthepenetrationrateofsmartphoneswas51%
in thefirst quarter of 2013,an increase of 19% over the same periodin2012.Giventhatsmartphoneshavegrowthpotentialin Taiwanandbrandsareacrucialfactorinsmartphonemarketing (Arruda-Filho,Cabusas,&Dholakia,2010),theresearchcontextis appropriate.Theresultsofthisstudymayprovidestrategicsugges- tionsforsmartphonemarketing.
Therestofthisarticleproceedsasfollows.Firstly,theliterature reviewdescribesthetwotheoreticalviewpointsandilluminates theirunderlyingconcepts.Then,Section3introducestheresearch modelandproposesthehypothesesregardingthedirectandmod- eratingeffectsonbrandloyalty.Theresearchmethodisdescribed inSection4,includingsampling,measurementdevelopment,and theexaminationofcommonmethodvariance.Section5reports theempiricalresults.Lastly,thediscussionofresults,theoretical andmanagerialimplications, limitations,anddirections forfur- therresearcharepresentedinSection6andtheconclusionsare inSection7.
2. Literaturereview 2.1. Consumervaluetheory
Consumervalueisthecornerstoneofasuccessfultransaction, and it motivates consumers to purchase repeatedly (Holbrook, 1994). The expectation disconfirmation paradigmsuggests that onceconsumershavesatisfactoryexperiencewithaproduct,they havebettervalueexpectationsandtendtorepurchasethesame productinthefutureinsteadofswitching(Anderson&Srinivasan, 2003).YangandPetersson(2004)indicatedthatvalueisasuper- ordinategoalandpositivelyregulatesloyaltybehavior,whichisat thesubordinatelevel.
InZeithaml(1988)’sexploratorystudy,thenotionofconsumer valuewasfoundtobeevolutionaryandmayoriginatefromutil- ity theory in economics, which assumes consumers’ purchase
decisionsarebasedonproductevaluations.Accordingtotheprin- cipleof utilitymaximization,a product/brandthat hassuperior attributesthanotheralternativesischosenbecauseproductper- formanceis expectedtobettersatisfyconsumers’needs. Needs gratificationisviewedasconsumervalue.Thus,productquality, whichrefersto“consumers’judgmentaboutaproduct’s overall excellenceorsuperiority”(Zeithaml,1988:3),isconceptuallyclose toproductbenefits(i.e.,whatconsumers getfromtheproduct) andconsumervalue.Products withhighqualityevaluationsare believedtodelivermorebenefitstoconsumers,and consumers perceivegreatvalue(Zeithaml,Berry,&Parasuraman,1996).In thisvein,notionsofproductquality,productbenefit,andconsumer valuewereinitiallyseenasequivalent(Zeithaml,1988).
However,asresearchonconsumervalueincreases,researchers haverecognizedthattherearenuanceddifferencesbetweenthese threeconstructs.Zeithaml(1988)proposedandelaboratedanew definition of consumer value; it essentially involves a give-get tradeoff.Consumersevaluatevalueaccordingtotheproduct’sben- efits, which are derived from the perceived quality of product attributes,andthemental,physical,andfinancialsacrificesgen- eratedfromproductacquisitionandusage(Cronin,Brady,&Hult, 2000).Bothbenefitsandsacrificesareindispensablepillarsofcon- sumervalueandcontributetoconsumervaluewithpositiveand negativeeffectsrespectively(Lin,Sher,&Shih,2005).Ifperceived benefitsoutweighperceivedsacrifices,consumersviewapotential transactionasbeingvaluable(YangandPetersson,2004).Basedon equitytheory,thepositive trade-offbetweenbenefitsandsacri- ficescreatesafeelingoffairnessforconsumers,whoarethenmore willingtorepurchasetheproduct.Thus,highperceivedvalueis accompaniedbyloyaltybehaviors(Croninetal.,2000).
Inaddition,manyresearchershaveexaminedwhetherornot thereareproductbenefitsotherthanfunctionalandeconomicones.
HirschmanandHolbrook’s(1982)well-knownstudyfoundthat consumersmayreceivesymbolic,hedonic,orestheticvaluefrom shoppingprocessesand/orproductusage.Theirnarrativeillustra- tionexpandedconsumervaluebeyondthefunctionalbenefitsand inspiredawholenewstreamofresearch.Next,Sheth,Newman,and Gross(1991)suggestedadetailedtypologyincludingfunctional, emotional,social,conditional,andepistemicvaluebysynthesiz- ingtheoriesofeconomics,sociology,psychology,andmarketing.
ExtendingShethetal.’s(1991)work,PihlströmandBrush(2008) demonstrated that conditional and epistemic values were the antecedentsofmonetary,convenience,emotional,andsocialvalue.
Babin,Darden,andGriffin(1994)developedasimplifiedbutgener- alizedvaluestructurewithtwodimensionsconsistingofutilitarian andhedoniccomponents.Similarly,SweeneyandSoutar(2001) decomposedconsumervalueintofunctional(i.e.,qualityandvalue formoney),emotional,andsocialvalueintheretailingcontext.
Based onSweeney andSoutar (2001)’s valueclassification,Kim etal.(2011)arguedthattherearesix typesof consumervalue.
Priceutilityandfunctionalqualityarerelatedtofunctionalvalue, aestheticsandplayfulnessarerelatedtoemotionalvalue,andsocial self-imageexpressionandsocialrelationshipsupportarerelatedto socialvalue.
Twointeresting findingsmaybesummarized fromthestud- iesjustdescribed.First,consumervalueisgenerallyspecifiedwith thesethreetypesofvalue(i.e.,functional,emotional,andsocial) despiteanincreasingnumberofstudiesthathaveattemptedto tapintothe natureof each valuetype (Karjaluotoet al.,2012;
Pihlström&Brush,2008).Second,thedefinitionoffunctionalvalue has expanded from physical performance/quality (Sheth et al., 1991)tophysicalperformance/qualityandvalueformoney(Kim etal.,2011;Sweeney&Soutar,2001).Sincevalueformoneyiscon- cernedwithmonetarysacrifice,thenewly-definedfunctionalvalue maybeconceptuallyequivalenttoZeithaml(1988)’sgive-getdefi- nition.Theresultisthatconsumervalue,whichcontainsfunctional,
emotional, and social components, seemsto bean appropriate frameworkbecauseitincludesallmajorbenefitssimultaneously.
2.2. Brandidentificationapproach
The brand identification approach was conceptually devel- oped from the consumer-company identification approach (Bhattacharya&Sen,2003),whichpositsthattheextenttowhich consumersidentifywitha brandrelatestoextra-rolebehaviors (e.g., recommendation or new consumer recruitment) and in- role behaviors(e.g.,productutilizationor repurchase;Ahearne, Bhattacharya, & Gruen, 2005). According toLam, Ahearne, Hu, and Schillewaert (2010:129), brand identificationis defined as
“consumers share the same self-definitional attributes with a brand”. This definition suggests that brands possess a distinct identity/personality (Donavan, Janda, &Suh, 2006; Stokburger- Saueret al.,2012).Brandspresentan extrinsiccuewithwhich consumerscaninferthequalityofaproduct.Brandsalsoproject anintrinsic identity that is manipulatedby brandmanagersto differentiateitfromcompetitors(Sung&Choi,2010;vanRekom, Jacobs,&Verlegh, 2006).Forexample,Heineken,a famousbeer brand,mayhaveconnotationsof“sober,serious,successful,anda littlealoof”(Kotler,Ang,Leong,&Tan,2003:421).Thepersonifica- tionofabrandenablesconsumerstointeractwithandestablish relationshipswiththebrand(Fournier,1998).
Priorstudieshaveexpoundedtwomechanismsthatmotivate consumerbrand identification.Oneis the needfor consistency (Kressmannetal.,2006).Consumersmaysearchforabrandwitha salientidentitythatmatchestheiractualself(Heetal.,2012).High identitysimilarity/congruence betweenconsumers and abrand facilitatestrongconsumerbelongingnessandgeneratebrandiden- tification (Lam, Ahearne,Mullins, Hayati,&Schillewaert, 2013).
Thesecond mechanism is theneed for self-esteem.Consumers canhelpformtheiridealidentity/self-imagebymeansofpurchas- inganidiosyncraticbrand(Heetal.,2012).Thecloserconsumers approachtheiridealself,thebettertheyfeel,whichhelpsraise self-esteem(Kressmannetal.,2006).Thus,abrandthatmatchesa consumer’sidealselfcanearnhis/herattachment.
2.3. Individualheterogeneity:ageandgenderdifferences
Individualheterogeneityisthevariationthatresultsfromdemo- graphics,personality,andsocio-culturalinfluences(Quesenberry&
Trauth,2012).Personaldistinctiveness,orinternalcausality,gov- ernsindividuals’presentationofconsistentattitudesandbehaviors towards specific objects or events. Of the individual hetero- geneities,demographicsaresalientvariablesthatareoftenusedin studiesexaminingtechnologyadoption(Venkatesh&Morris,2000) andconsumervalue(Deng,Lu,Wei,&Zhang,2010).Thiscurrent studylooksintothemoderatingeffectsofageandgender.
Ageaffectsindividuals’attitudesand behaviors.Thesediffer- encesoriginatefromthebio-physicalandpsychologicalchanges thatoccurasageincreases(Deng,Mo,&Liu,2014).Forexample, Carstensen,Isaacowitz,andCharles(1999)suggestedthatindivid- ualsinvariouslifestageshaveuniqueawarenessofthepassage oftime.Youngerpeoplearemorefuture-oriented,andtheyper- ceivetime astimesince birthand open-ended,while moreelder counterpartsarepresent-orientedandperceivetimeastimeleftin lifeandlimited.Differentperspectivesofthepassageoftimecause olderpeopletoemphasizesocioemotionalexperienceandyounger peopletofocusonskillsandknowledge.Erikson(1959)’s8-stage psychosocialdevelopmentelucidatedthatindividualsineachstage confrontdifferentidentitycrisesandsignificantrelationships,and thustheyhavedifferentpsychosocialneeds.
In a similarvein, gender alsoproduces distinctive attitudes andbehaviorsinmenandwomen.Accordingtogendersocializa-
tiontheory,individualsarenurturedundergenderrolesthatdrive individualstoacquiremasculine/feminineconceptsandrelevant skills,andthusdevelopvariedvaluesystems(Mason&Mudrack, 1998).Self-construaltheoryclaimsthatsex-specificselfconstrual causesindividualstoprocessinformationdifferently(Meyers-Levy
& Loken, 2015; Okazaki & Mendez, 2003a). For instance, men areperceivedasindependentandself-orientedwhilewomenare viewedas dependentand relationship-oriented.Venkatesh and Morris(2000)claimedthatmenandwomenprocessinformation usingdifferentsocially-constructedcognitivestructures,andthey demonstratedthatbehavioralpatternsarelinkedtogender.They foundthattheeffectofperceivedusefulnessonbehavioralinten- tionwasgreaterformenthanforwomenbecausemenaremore task-oriented.WomenlikelysufferfromITanxietyandconformto referencegroups;thus,theeffectsofperceivedeaseofuseandsub- jectivenormsonbehavioralintentionswerestrongerforwomen thanformen.
Takentogether,thiscurrent studyrecognizesthedifferences associatedwithageandgender,anditexploreshowageandgen- derdifferencesmightmoderatetheeffectsofconsumervalueand brandidentificationonbrandloyalty.
3. Hypotheses
3.1. Thespecificationofresearchmodel
The research model is illustrated in Fig. 1. Brand loyalty is defined as consumers’ favorable attitude toward a brand that resultsinintentionstorepurchaseandrecommend(Anderson&
Srinivasan,2003).Tounderstandtherelativeimportanceof the determinantsofbrandloyalty,thisstudyspecifiesthatconsumer valueandbrandidentificationdirectlyinfluencebrandloyalty.Con- sumervalueistheconsumer’soverallevaluationoftheutilityof aproduct/brand(Zeithaml,1988),includingfunctional,emotional, andsocialvalues(Sweeney&Soutar,2001).Functionalvalue,which isanalogoustoutilitarianvalue,isthebenefitsgainedfromaprod- uct/brandbasedonitsfunctionalperformanceandvalueformoney.
Specifically,thedefinitionalscopeoffunctionalvaluecoversthe get-givetrade-offideaofperceivedvalue.Emotionalvalue,which isequivalenttohedonicvalue,referstothefeelingsortheaffective statusarousedbyaproduct/brand(Kimetal.,2011).Ingeneral, emotionalvalueisgeneratedfromproductusage/explorationand productappearance.Finally,socialvalueistheextenttowhicha product/brandenhancesconsumers’socialwell-beingandinter- personalrelationships,anditisrootedinthesymbolicmeaningsof theproduct/brand(Rintamäki,Kanto,Kuusela,&Spence,2006).All threevaluetypesareuniqueandinterrelated(Sweeney&Soutar, 2001).
Viewingbrandidentificationas“apsychologicalstateofper- ceiving,feeling,andvaluinghisorherbelongingnesswithabrand,”
Lametal.(2010:130)conceptualizeditasasecond-orderforma- tiveconstructwiththreereflectivesub-dimensions.Becausethat affectivebrandidentificationmayblendwithotherconstructssuch asbrandlove,andthatevaluativebrandidentificationissimilarto brandattitudeanditislikelytobetheresultofidentification,this studyfollowsStokburger-Saueretal.(2012)andconsidersbrand identificationasconsumers’perceptionofbothentities’identities atthecognitivelevel.Thismeansthatconsumers’brandidentifica- tionisapsychologicalstateratherthanaprocess,anditpositively determinesbrandloyalty(Rocereto&Mosca,2012;Stokburger- Saueretal.,2012).Lastly,individualheterogeneityisconsideredto altertheeffectsofthefourdeterminantsonbrandloyalty;there- fore,themoderatingeffectsofageandgenderareexamined.
Somestudiesareinterestedintherelationshipsbetweencon- sumervalueandbrandidentification.InLamandShankar’s(2014)
Fig.1.Researchmodel.
study,brandattachment,whichwasdefinedasabrand’sresonance withaconsumer’sself-conceptandprovidesconsumersasenseof securitysimilartothenotionofbrandidentification,wasfound tobeanoutcomeofconsumervalue.However,Heetal.(2012) andSo,King,Sparks,andWang(2013)foundevidencethatbrand identificationwastheantecedentof consumervalue.Consider- ingthedeterminantpriorityofbrandloyaltyisourmainresearch inquiryandtherelationshipbetweenconsumervalueandbrand identificationappearstobecontroversial,thisstudyfocusesonhow muchconsumervalueandbrandidentificationimpactbrandloy- altyinsteadoftheirinterrelationships,andhowtheeffectsofloyalty determinantschangeinvariedageandgender.
3.2. Theeffectsofconsumervalueonbrandloyaltyinthecontext ofsmartphoneconsumption
Studiesontechnologymarketinghaverecognizedthatatech- nology product has a combination of tangible and intangible attributes.Bygroupingalltheattributesintoperformance,appear- ance,andcommunicationattributes,Lee,Ha,andWiddows(2011) suggestedthattechnologyproductsmight delivervaluetocon- sumersviathesethreeattributetypes.Consumersgainfunctional valueiftheattributeperformancesofa technologyproductare useful,easy to use, and innovative.A technology productwith attractiveappearance,novelmaterial, and atypicaldesign posi- tivelyelicitsconsumers’feelings,andoffersconsumersemotional value.Furthermore,atechnologyproductmaybeasymbolitself and/orcarrysymbolicmeanings,enablingconsumerstocommu- nicatetotheirlifestyleandbeliefstoothers,thusacquiringsocial value. Exploring the attributes of mobile phones,Horváth and Sajtos(2002)identifiedutility/usefulness,experience/enjoyment of use, and communicative power/expression as three main product-relatedconsumer responses.Thesethree responsesare similartothosementionedbyLeeetal.(2011).
Asmartphoneisastate-of-the-arttechnologyproduct.Byinte- grating components such as processor, camera, display panel, battery,andmemory/storagecapacityintoa handhelddevice,a smartphoneisatelecommunicationdeviceandatoolthatcanbe usedtolistentomusic,editdocuments,takeandrecordpictures, andstudy(Liao&Hsieh,2013;Park&Han,2013).Also,asmart- phonehasaconnectiontotheInternetforaccesstoonlineservices likee-mail,maps,informationsearches,andlocation-based ser- vices(Okazaki&Mendez,2013a).Thus,asmartphone isableto delivervariousfunctionalbenefitstoconsumers.
Priorstudieshavefoundevidenceofrelationshipbetweenfunc- tionalbenefitandbrandloyalty,and theyreportedthatabrand thatoffershighfunctionalvalueearnsconsumers’preferenceand
loyalty.Thus,therelationshipbetweenfunctionalvalueandbrand loyaltyishypothesizedasfollows:
H1. Functionalvaluepositivelyrelatestobrandloyalty.
Inadditiontofunctionalvalue,consumersmayexperienceemo- tionalvalue suchas playfulnessand pleasurefromsmartphone usageandexploration(Alba&Williams,2013;Arruda-Filhoetal., 2010).LiaoandHsieh(2013)alsopointedoutthatthefashionable andaestheticappearanceofsmartphonescontributesemotional value.PihlströmandBrush(2008)revealedwhenconsumersper- ceivegreateremotionalvalueinaproduct/brand,theyshowmore brandloyaltyasmeasuredbyrepurchaseintentions,willingnessto pay,andpositiveword-of-mouth.Thus,therelationshipbetween emotionalvalueandbrandloyaltyishypothesizedasfollows:
H2. Emotionalvaluepositivelyrelatestobrandloyalty.
Consumermayperceivesocialvaluefromsmartphones.Recruit- ing iPhone users as interviewees, Arruda-Filho et al. (2010) conductedanetnographicanalysisandfoundthatconsumersmay experiencesocialvaluefromthepossessionandusageofsmart- phones.TheymayviewthepossessionofaniPhoneasasymbolof luxuryandhighersocialstatus(Liao&Hsieh,2013).Inaddition,the sharingofsmartphoneusageexperienceaidsconsumers’interper- sonalinteractions.Whenconsumersperceivehighersocialvalue fromaproduct/brand,theyshowgreaterbrandloyaltybehaviors suchasdisseminatingpositiveinformationandacceptingpremium prices(PihlströmandBrush,2008).Thus,therelationshipbetween socialvalueandbrandloyaltyishypothesizedasfollows:
H3. Socialvaluepositivelyrelatestobrandloyalty.
3.3. Theeffectofbrandidentificationonbrandloyaltyinthe contextofsmartphoneconsumption
Somestudieshavefoundthatmobilephonesaresurrogatesfor self-identityexpression(Mannetti,Pierro,&Livi,2002;Walsh&
White,2007;Walsh,White,&Young,2010).Byusingpersonalized ringtonesanddecorations,consumersareabletoextendtheirper- sonalitytomobilephones.Fromabrandmanagementstandpoint, Lametal.(2010)similarlyobservedthatsmartphonebrandshold uniqueidentitiesandmayaccordwithorenhancetheconsumer’s identity.Furthermore,resultsfromStokburger-Saueretal.’s(2012) study,whichsurveyedfourproductcategories(i.e.,cellphones, athleticshoes,softdrinks,andgrocerystores),confirmedthathigh brandidentificationis abletoturnconsumers intobrandloyal- istsandpreventconsumersfromswitchingtootherbrands.Thus, therelationshipbetweenbrandidentificationandbrandloyaltyis hypothesizedasfollows:
H4. Brandidentificationpositivelyrelatestobrandloyalty.
3.4. Themoderatingeffectofage
Harverila(2012)and Kumarand Lim(2008)arguedthatthe motivational need for mobile phone usage is age-specific, and theeffects ofthethree valuetypesand brandidentificationon brandloyaltymayvarywithconsumers’age(Barutc¸u,2007;Coates, 2001;Park,Eisingerich,&Park,2013;Persaud&Azhar,2012).In thecontext of mobile phone consumption, younger consumers aremoreenthusiasticusersofsmartphonescomparedwithelder consumers. As indicated by Coates (2001), younger consumers usemoresmartphonephonefunctions,suchastexting,satellite navigation,andphotoediting,whereaselderconsumerstendto usethemforcommunication.Complexfunctions,user-unfriendly menus,andunclearusageinstructionsmayhinderelderconsumers fromexploringmanysmartphoneapplications,whichmightcause elderconsumerstoperceivelessfunctionalandemotionalvalue (Kurniawan, 2008). On the contrary, younger consumers may
“engageinhighlevel....mobilephoneuse”(Walshetal.,2010, p. 194), and are more likely to appreciate its fashion-designed appearance(Park etal.,2013; Srivastava,2005).Barutc¸u(2007) alsofoundthattheyoungerconsumersweremoreacceptingof mobileentertainmentservicesthanelderconsumers.Thissuggests thatyoungerconsumersmightreceivemorefunctionalandemo- tionalvaluefromtheirsmartphonesthanelderconsumers,andthe relationshipbetweenfunctionalvalue/emotionalvalueandbrand loyaltymayvaryasconsumers’ageisconsidered.Moregenerally stated,theeffectsoffunctional/emotionalvalueonbrandloyalty mayincreaseasagedecreases.
Furthermore,studieshavesuggestedthatyoungerconsumers aremore susceptibletosocial influencefrompeers andfriends thanelderconsumers(Smetana,Campione-Barr,&Metzger,2006), andsharednormsandstandardsstronglyguidetheirconsump- tion behaviors.Walsh and White (2006) stated that displaying a mobile phonein public improvedyounger consumers’ status amongpeers.Asfortheusageofsmartphones,PersaudandAzhar (2012)demonstratedthatyoungerconsumersweremoreinvolved insocialnetworkingactivitiesthanelderconsumers.Accordingto Erikson’s(1959)theoryofpsychosocialdevelopment,youngercon- sumers(especiallyattheadolescentstage)haveastrongerneed foridentity,sotheytendtoexpressthemselveswiththeirmaterial possessionsandidentifywithbrandsthatrepresenttheirvalues andbeliefs(Syed&Nurullah,2011;Walshetal.,2010).Similarly, SheldonandKasser(2001)claimedthatagewasnegativelyasso- ciatedwith thedemand for identity. Thus, younger consumers mayfavor socialvalueandbrandidentificationmorethanelder consumers,andtherelationshipbetweensocialvalue/brandidenti- ficationandbrandloyaltymayvaryasconsumers’ageisconsidered.
Theeffectsofsocialvalueandbrandidentificationonbrandloyalty mayincreaseasagedecreases.
Insummary,thisstudyproposesthefollowinghypotheses:
H5a. The positive relationship between functional value and brandloyaltyisgreaterwhenagedecreases.
H5b. The positive relationship between emotional value and brandloyaltyisgreaterwhenagedecreases.
H5c. Thepositiverelationshipbetweensocialvalueand brand loyaltyisgreaterwhenagedecreases.
H5d. Thepositiverelationshipbetweenbrandidentificationand brandloyaltyisgreaterwhenagedecreases.
3.5. Themoderatingeffectofgender
Previousstudiesandtheorieshaveshownthatgenderhaslarge influenceonconsumervalues,preferences,andbehaviors.Okazaki andMendez(2013a,b);OkazakiandMendez(2013a,b)explained thatmaleandfemaleconsumersmayhavedifferentvalueprefer- encesandidentificationneeds,andHasan(2010)recognizedthat menand women displaydiverseperceptions and attitudes.For example,Dittmar,Beattie,andFriese(1995),foundthatmenwere moreactivity-focusedandplacedhigheremphasisonfunctional value,whilewomenweremorerelationship-orientedandfocused onemotionalandsocialvalue.Dittmar(2005)alsodeclaredthat emotionalvalueandidentity-relatedfactorsweremoreimportant forwomenthanformenwhileshopping.
Inthecontextofmobilephoneconsumption,SyedandNurullah (2011)reportedthatmentendedtotreatmobilephonesastoys, whichmeansthefunctionaland emotionalvaluethat mengain fromsmartphoneusageismorecloselylinkedtoproductexplo- rationandexperiencethanwomen.Inaddition,SyedandNurullah (2011)foundthatwomenweremorelikelytousemobilephones for communication and relationship maintenance. In a related direction, Walsh and White (2007) found evidence that social influence and normative pressure might be the main drivers of women’smobilephone use. Dittmar (2005)maintainedthat identity-relatedfactorsaremore importantforwomenthanfor menwhileshopping;therefore,brandidentificationmighthavea strongerimpactonwomen’ssmartphonepurchasedecisions.More generally,womenfavorsocialvalueandbrandidentificationmore than men.Asmenmayreceive more functionaland emotional valuefromtheirsmartphonesthanwomen,andwomenmayfavor socialvalueandbrandidentificationmorethanmen,therelation- shipsbetweenthefourdeterminantsandbrandloyaltymayvary asgenderisconsidered.
Insummary,thisstudyproposesthefollowinghypotheses:
H6a. Thepositiverelationshipbetweenfunctionalvalueonbrand loyaltyisgreaterformenthanforwomen.
H6b. Thepositiverelationshipbetweenemotionalvalueonbrand loyaltyisgreaterformenthanforwomen.
H6c. Thepositiverelationshipbetweensocialvalueonbrandloy- altyisgreaterforwomenthanformen.
H6d. Thepositiverelationshipbetweenbrandidentificationon brandloyaltyisgreaterforwomenthanformen.
4. Methods 4.1. Measures
Therewerefivesetsofmeasuresdevelopedforthemajorcon- structsinthisstudy.Themeasuresforthethreevaluetypeswere adoptedfromKimetal.’s(2011)comprehensivework,andeach value had four items extracted fromcorresponding types (e.g., socialvaluewasmeasuredwithtwoitemsfromsocialself-image expressionandtwofromsocialrelationshipsupport)tomakethe measuresasbalancedaspossible.Themeasureforbrandidentifica- tionwastakenfromStokburger-Saueretal.(2012)andcontained threeitems.Thismeasureprovidesaricheroperationalizationof cognitivebrandidentificationthanthatofLametal.(2010),which directlyassessesidentitysimilarityusinga Venndiagramanda verbalitem.Lastly,themeasureforbrandloyaltywasdeveloped fromthestudiesbyAndersonandSrinivasan(2003)andZeithaml, Berry,andParasuraman(1996).Thefourchosenitemstookboth thecommitmentelementandthecomparisonelementintocon- sideration (Dick& Basu, 1994; Oliver, 1999).To ensure better measurementqualityandmitigatethenegativeeffectofresponse
Table1
Measuresandreliability.
Construcutsanditems Loading ˛ CR
Functionalvalue(Kimetal.,2011) 0.91 0.94
FV1 Xsmartphoneshaveanacceptablestandardofquality. 0.93
FV2 Xsmartphonesarereliableintheirperformance. 0.92
FV3 Xsmartphonespossessadegreeofqualitythatissatisfactory. 0.95
FV4 Xsmartphonesoffervalueformoney. 0.75
Emotionalvalue(Kimetal.,2011) 0.86 0.90
EV1 IlikethewayXsmartphoneslook. 0.84
EV2 Xsmartphonesarenotcatching.(R) 0.81
EV3 UsingXsmartphonesisinterestingtome. 0.83
EV4 UsingXsmartphonesgivesfuntome. 0.86
Socialvalue(Kimetal.,2011) 0.82 0.88
SV1 UsingXsmartphonesenhancesmyself-imagetoothers. 0.88
SV2 UsingXsmartphonesimprovesthewayIamperceived. 0.84
SV3 UsingXsmartphonesdoesnothelpmemaintainmysocialrelationshipswithothers.(R) 0.69
SV4 UsingXsmartphonesenhancesmysocialrelationshipswithothers. 0.81
Brandidentification(Stokburger-Saueretal.,2012) 0.83 0.90
BI1 IfeelastrongsenseofbelongingtoXsmartphones. 0.87
BI2 Xsmartphonesarelikeapartofme. 0.88
BI3 Xsmartphoneshaveagreatdealofpersonalmeaningforme. 0.85
Brandloyalty(AndersonandSrinivasan,2003;Zeithamletal.,1996) 0.91 0.94
BL1 IbelievethatXsmartphonesaremyfavorites. 0.88
BL2 IsaypositivethingsaboutXsmartphonestootherpeople. 0.86
BL3 IrecommendXsmartphonestosomeonewhoseeksmyadvice. 0.89
BL4 WhenIneedtomakeapurchase,Xsmartphonesaremyfirstchoice. 0.92
Note:1.(R)denotesthatanitemisinareverseform.
2.Theloadingswerederivedfromthedirecteffectmodel.
fatigue,thisstudyinterlacedtheitemsandtwoitemswerecon- structedinreverseform.Alltheconstructtermswereconcealedin ordertoreducesocialdesirabilitybias.Table1listsallthemeasure- mentitems.Alltheitemswerereflectivelyspecifiedandresponses utilized7-pointLikertscales(Viswanathan,Sudman,&Johnson, 2004).Respondentsansweredtheseitemsintermsoftheirexperi- encewiththeirmost-usedsmartphone.
Basedontheauthors’reviewsanddiscussions,themeasures weretranslatedfrom Englishinto Chinesetocollectresponses.
Alltheauthorshave richuser/consumerresearchexperiencein ISandmarketingfields.Takingresearchcontextintoconsidera- tioninsteadofliteraltranslation,thetranslationequivalencewas achievedwithgreaterpossibilities(Douglas&Craig,2007).Thelan- guageofthetranslatedquestionnairewasmadewiththeconsensus ofalltheauthors.Apretestofquestionnairewasconductedtocheck whetherrespondentswouldhaveanydifficultiesincomprehension orfinditemsambiguous(Boudreau,Gefen,&Straub,2001).After feedbackfromthepretest(aconveniencesampleof6smartphone users),refinementsweremadetothelanguageofSV3andSV4and thequestionnairewasthenfinalized.
4.2. Controlvariable
Similartotheeffectoflengthofpatronageonstoreloyaltyin serviceandretailing contextsrevealedbyJones, Mothersbaugh, andBeatty(2000),theremaybeapositiverelationshipbetween lengthofbrandrelationshipandbrandloyaltyintheproductcon- text(Kressmannetal.,2006).Hence,lengthofbrandrelationship wasincludedintheanalyticmodelasacontrolvariableforeffect purification.Datawasgatheredfromrespondents’answertothe question,“Howlonghave you beenusing your most-usedsmart- phone?”(Jonesetal.,2000).
4.3. Datacollectionandconsumerprofile
Anonlinesurveywascreatedtocollectdata.Respondentswere voluntaryrecruitsfromthelargestbulletinboardsysteminTaiwan
(telnet://ptt.cc).OnlyInternetsurferswhoweresmartphoneusers werequalified toparticipateinthis study.Accesstotheonline questionnaire was via a link embeddedin the post.The web- site(http://www.mysurvey.tw/)whichhostedthequestionnaire restrictseverycomputertooneresponse. Toencouragepartici- pation,respondentswhoprovidedusableresponseswereentered intoalottery;respondentshadaoneinthreechanceofwinninga giftvoucherasareward.Thesurveywasopenforonemonthand atotalof179responseswereobtained.Ofthese,22sampleswere invalidbecauseoflogicalinconsistenciesbetweentheirresponses tothereverseitemsandotheritems.Thus,thetotal numberof validresponseswas157.Thisnumbersatisfiedtheruleofthumb forusingthepartialleastsquares(PLS)method,whichsuggeststhat samplesizeshouldbeatleasttentimesthenumberofitemsinthe mostcomplexconstructorshouldbelargerthan150(Gefen,Straub,
&Boudreau,2000;Haenlein&Kaplan,2004;Urbach&Ahlemann, 2010).Thepoweranalysisalsosustainedoursamplesizesmetthe minimumsizerequirement(n=137)todetectminimumR2values of0.10,under5%significancelevel,80%statisticalpower,andthe maximumnumberofitems/independentvariablesinthemeasure- mentandstructuralmodelsarefour(Hairetal.,2014).
ThedemographicsoftherespondentsaredisplayedinTable2.
About55%oftherespondentsweremale.Theiragesrangedfrom 17to58,andtheaverageageofthesamplewasabout27years old.Theagedistributionshowedthat68%respondentswereinthe 21–30group.Theaveragemonthlydisposableincomewasaround US$664.Morethan90%oftherespondentsheldbachelor’sdegrees orhigher.Table2showsrespondents’mobilephoneconsumption.
Onaverage,theyboughtanewmobilephoneevery2.48years,and had1.20smartphonesand0.68featurephonessimultaneously.The numberofthesmartphonestheyhadwastwiceasmuchasthat ofthefeaturephones,whichwassimilartothemarketsharesug- gestedbyInformationDataCenter(2012)mentionedearlier.Lastly, 39%(25/64)oftherespondentswhohadpurchasedtwoormore smartphonesboughttheirphonesfromthesamebrands,whileonly 19%(29/156)oftherespondentswhohadpurchasedtwoormore
Table2
Demographicsofrespondents(n=157.)
Characteristics Frequency(%) Mean(SD)
Gender
Male 86(54.8%)
Female 71(45.2%)
Age(years) 26.98(6.60)
Below20 15(9.6%)
21–30 107(68.2%)
31–40 29(18.5%)
41–50 4(2.5%)
Above51 2(1.3%)
Educationlevel
Highschooldiploma 8(5.1%)
Juniorcollege 4(2.5%)
Bachelor’sdegree 109(69.4%)
Master’sdegree 36(22.9%)
Occupation
Publicemployee 12(7.6%)
Manufacturing 21(13.4%)
Service 33(21.0%)
Student 72(45.9%)
Others 19(12.1%)
Disposableincome(USD) 664.01(569.15)
Howoftendoyoubuyanewmobilephone?(years) 2.48(0.77)
Howmanysmartphonesdoyouhave? 1.20(0.45)
Howmanyfeaturephonesdoyouhave? 0.68(0.82)
Arethelastfeaturephoneandthelastsmartphoneyouboughtthesamebrand?
Yes,theyarethesamebrand. 29(18.5%)
No,theyarenotthesamebrand. 127(80.9%)
Ineverboughtafeaturephone. 1(0.6%)
Ineverboughtasmartphone. 0(0.0%)
Arethelasttwosmartphonesyouboughtthesamebrand?
Yes,theyarethesamebrand. 25(15.9%)
No,theyarenotthesamebrand. 39(24.8%)
Iboughtasmartphoneonce. 93(59.2%)
Howlonghaveyoubeenusingyourmost-usedsmartphone?(years) 1.47(0.95)
Note:(1)USD30ND.
mobilephonesboughtthesamebrandsfortheirlastsmartphones andlastfeaturephones.
4.4. Commonmethodvariance(CMV)
Common methodvariance (CMV), which resultsin spurious relationshipsbetweenvariables,isaconcernforallself-reported questionnaires.AsPodsakoff,MacKenzie,Lee,&Podsakoff(2003) suggested, this study attempted to control for the CMV effect in thedevelopment and design of thesurvey questionnaire by counterbalancingitemorder,improvingitemcomprehension(i.e., throughpretesting),protectingrespondentanonymity,andreduc- ing evaluation apprehension. Also, Harman’s single-factor test was used to examine whether CMV had occurred statistically (Podsakoff et al., 2003).The results of exploratory factor anal- ysisrevealed that all theitems did not converge into a single factor,but thefirst factoraccounted for53.43% variance ofthe items.Furtherconfirmatoryfactoranalysiswhich isan alterna- tiveofexploratoryfactoranalysis(Malhotra,Kim,&Patil,2006;
Podsakoffet al., 2003), wasperformed and theresultsshowed thatthefive-factormodel(2(142)=485.52,GFI=0.76,AGFI=0.68, CFI=0.95, IFI=0.95, RMSEA=0.12, RMR=0.074, NFI=0.93, and NNFI=0.94) yielded better goodness of fit than the one-factor model(2(152)=884.49,GFI=0.58,AGFI=0.47,CFI=0.90,IFI=0.90, RMSEA=0.20,RMR=0.098,NFI=0.88,andNNFI=0.88).Thisindi- catesthattheCMV effectmightbeproperlycontrolledandthe surveydatawasacceptableforPLSanalyses.
5. Results
The PLS method, which is variance-based structural equa- tionmodelingandisadistribution-freetechnique(Hair,Sarstedt, Ringle,&Mena,2012),wasusedtoanalyzethedata.Theresults wereproducedusingtheSmartPLS2.0M3package(Ringle,Wende,
&Will,2005).
5.1. Measurementmodel
Table1presentsthepsychometricpropertiesofthemeasures.
ThevaluesofCronbach’s˛andcompositereliability(CR)ranged from0.82 to0.94, indicatingthat thefive setsofmeasureshad stronginternalconsistency(Henseler,Ringle,&Sinkovics,2009).
Thefactorloadingswereallabove0.69,suggestingthataroundor morethanhalfofthevarianceofanitemwasattributedtoitscor- respondingconstruct(Chin,1998).Thus,indicatorreliabilitywas adequate.Averagevarianceextracted(AVE)valueswereallhigher thanthethresholdvalue(0.5)withtheminimumvaluebeing0.65 (seeTable3),therefore,convergentvaliditywasevident.
DiscriminantvaliditywasassessedusingtheFornell-Larckercri- terionandcross-loadings(Urbach&Ahlemann,2010).Asshownin Table3,theAVEvalueofeachconstructwassuperiortoitscorre- spondingsquaredcorrelations,demonstratingthatconstructswere discriminable.Theresultsalsoshowedthateachitemloadedonits designatedconstructwithoutcross-loadings.Accordingly,discrim- inantvaliditywassatisfactory.
Table3
Descriptivestatisticsanddiscriminantvalidity.
M SD FV EV SV BI BL
FV 5.07 1.03 0.79
EV 5.24 0.94 0.55 0.70
SV 4.27 1.07 0.27 0.30 0.65
BI 4.50 1.27 0.31 0.34 0.56 0.75
BL 4.77 1.27 0.53 0.58 0.45 0.52 0.79
Note:Diagonal elementsandoff-diagonalelementsrepresentaveragevariance extractedandsharedvariance,respectively.
5.2. Structuralmodel
Table4showstheresultsofthemodel evaluation.Theesti- mationof standardizedpathcoefficientswasbasedonthepath weightingscheme. Thesignificance testing wasbased on1000 bootstrappedsamplesandtheoptionofconstructlevelchanges (Tenenhaus, Vinzi, Chatelin, & Lauro, 2005). First the effect of control variable was checked. Model 1 shows that consumers’
length of brand relationship had a significant positive impact on brand loyalty (=0.24, p<0.01). Next, Model 2 examines theeffects of the fourdeterminants and found that functional value (=0.26, p<0.001), emotional value (=0.32, p<0.001), socialvalue(=0.14,p<0.05),andbrandidentification(=0.28, p<0.001)predictedbrandloyaltywithstatisticalsignificanceand explained73.17%ofthevariance.TheseresultssupportH1,H2,H3, andH4.Furthermore,Models3and4taketheproduct-indicator approachtotestmoderatingeffects(Chin,Marcolin,&Newsted, 2003)andfoundthatonlyagestrengthenedtheemotionalvalue- loyalty and social value-loyalty linkages while weakening the relationshipbetweenbrandidentificationandloyalty.Thus,H5d issupported,but H5a,H5b,and H5carenot supported.Gender differencesdidnot exertanymoderating effectacrossthefour hypothesizedrelationships.Thus,H6a,H6b,H6c,andH6darenot supported.
Moreover, this study assessed the strength of the moder- ating effects by evaluating effect sizes (Urbach & Ahlemann, 2010). The definition of effect size is “the degree to which the phenomenon is present in the population” (Cohen, 1988), and 0.02, 0.15, and 0.35 are the respective values to identify whether the effect is small, medium, or large. Based on the formulaoff2=(R2modelwithmoderator−R2modelwithoutmoderator)/(1−R2
modelwithmoderator),agehadamediumeffectsize(0.23),whilegender (0.05)hadasmallone.
Tofurtherunderstandhowthemoderatingroleofagefunctions, respondentsweredividedintotwogroupsusinglatentscoresof age.Consideringtherangeofagesinthesamplewas17–58andthe averageagewas27,amiddle-agedgroup(positivelatentscores) andyoungergroup(negativelatentscores)werefoundandshowed distinctpatterns,whichareshowninFigs.2–4.Fig.2showsthat theslopeforthemiddle-agedgroupwassteeperthantheyounger group,indicatingthattheimpactofemotionalvalueonbrandloy- altywasstronger formiddle-aged consumersthan for younger consumers.Anincreaseinemotionalvaluewouldleadtogreater brandloyaltyofmiddle-agedconsumers,comparedwithyounger consumers.Itwasalsothecaseintherelationshipbetweensocial valueandbrandloyalty,whichispresentedinFig.3.
Moreover, both Figs. 2 and 3 reveal that as emotional value/social value increased, the differences in brand loyalty betweenthemiddle-agedconsumergroupandtheyoungercon- sumergroupalsoincreased.AcomparisonbetweenFigs.2and3 indicatethattheintersectionoccurredatalowersocialvalue(see Fig.3)thanemotionalvalue(seeFig.2).Atagivenlevelofemotional value/socialvalue,thedifferencegapinbrandloyaltybetweenthe twoconsumergroupswaswiderinthecaseofsocialvalue,com- paredwiththeemotionalvaluecase.
Fig.2.Themoderatingeffectofageontherelationshipbetweenemotionalvalue andbrandloyalty.
Fig.3.Themoderatingeffectofageontherelationshipbetweensocialvalueand brandloyalty.
Fig.4. Themoderatingeffectofageontherelationshipbetweenbrandidentification andbrandloyalty.
Table4 Modelevaluation.
Variable Expectedsign Model1 Model2:directeffect Model3:age Model4:gender
Lengthofbrandrelationship C.V.(+) 0.24**(3.19) 0.02(0.54) 0.01(0.35) 0.03(0.58)
Age M.V. – – 0.13*(2.45) –
Gender M.V. – – – −0.08(1.81)
Functionalvalue H1(+) – 0.26***(3.66) 0.25**(3.05) 0.25**(3.18)
Emotionalvalue H2(+) – 0.32***(4.91) 0.30***(4.15) 0.31***(4.48)
Socialvalue H3(+) – 0.14*(2.08) 0.14*(2.29) 0.15*(1.99)
Brandidentification H4(+) – 0.28***(4.09) 0.31***(4.53) 0.29***(4.05)
Functionalvalue×age H5a(¡Ð) – – −0.12(1.35) –
Emotionalvalue×age H5b(−) – – 0.22*(2.29) –
Socialvalue×age H5c(−) – – 0.29*(2.54) –
Brandidentification×age H5d(−) – – −0.22*(2.00) –
Functionalvalue×gender H6a(+) – – – 0.09(1.11)
Emotionalvalue×gender H6b(+) – – – −0.05(0.67)
Socialvalue×gender H6c(−) – – – −0.02(0.27)
Brandidentification×gender H6d(−) – – – 0.01(0.11)
R2brandloyalty=effect size
(f2)
5.92% 73.17% 78.25% 74.52%
0.2336 0.0530
Note:(1)Thenumbersinparenthesesrefertot-valuesderivedfrom1000bootstrappedsamples.
(2)*Denotesp<0.05;**denotesp<0.01;***Denotesp<0.001.Two-tailedtest.
(3)C.V.istheabbreviationfor“controlvariable”;M.V.istheabbreviationfor“moderatorvariable”;n.s.referstonosignificantdifference.
(4)TheformulaforcalculatingtheeffectsizeisbasedonCohen(1988).f2=(R2modelwithmoderator–R2modelwithoutmoderator)/(1–R2modelwithmoderator).
Incontrast,Fig.4showsthattheyoungerconsumergrouphad asteeperslopethanthemiddle-agedconsumergroup.Itindicates thattheyoungerweremoresensitivein brandloyaltythanthe middle-agedasthebrandidentificationincreased.Interestingly, eventhoughtheimpactofbrandidentificationonbrandloyaltywas strongerfortheyoungerconsumers,themiddle-agedconsumers consistentlymaintainedahigherbrandloyaltythantheyounger consumersuntilthebrandidentificationreachedhigherlevels.The gapbetweenthetwogroupsbecamesmallerasbrandidentification increased.
6. Discussion
6.1. Discussionoftheresults
Whileconsumervalue andbrandidentificationmightdomi- natetheformationofbrandloyalty,thereislittleunderstanding regardingtheirrelativeimportance.Therolesofageandgenderdif- ferencesininfluencingtheserelationshipsarelessexamined.For thesereasons,thisstudyintendstoknowthedeterminantpriority ofbrandloyaltyandtakesageandgenderdifferencesintoaccount.
Thetwotheoriesareexpectedtobeelaboratedthroughconsumer segmentation,andtheresultsmayprovidemarketingmanagers withmoreexquisiteandfeasiblestrategicdirectionstoguidethe deploymentoftheirresources.
Theempiricalresultsofthisstudyrevealedthatconsumervalue (i.e.,functionalvalue,emotionalvalue,andsocialvalue)andbrand identificationpositivelypredictedbrandloyalty.Emotionalvalue wasevaluatedtobemostinfluentialtobrandloyalty,andwasfol- lowedbybrandidentification,functionalvalue,andsocialvalue.
Therelativeimportanceofthetopthreedeterminantswereclose, andwasnearlytwiceasmuchasthatofsocialvalue.Asexpected, theresultsfurthershowedthatagemoderatedtheeffectofbrand identificationonbrandloyalty,andtheeffectofbrandidentifica- tionwasgettinggreaterasagedecreased.Surprisingly,againstour expectations,theeffectoffunctionalvalueonbrandloyaltydidnot changeasageincreased.Apossiblereasonmayexplaintheinsignif- icanteffectofagedifferences.Mostrespondentsinthisstudywere middle-agedandyoungerconsumers(17–58yearsold),andthey hadallexperiencedrapidtechnologicalchangesintheirlifetimes andmighthavehadhighacceptanceofsmartphones(Dengetal., 2014).Persaudand Azhar (2012)’s studyreportedthat younger consumers(13–24yearsold)usedsmartphonesmainlyfortex-
ting,photo-taking,andvideo-viewing,andmiddle-agedconsumers (35–54yearsold)engagedinsmartphoneactivitiesfortheuseofe- mail,maps,newsandinformation,andbanking.SimilartoPersaud and Azhar (2012),theeaseof useof thehandheldsmartphone maycontributemiddle-agedandyoungerconsumerstoexperience multiplebenefitsfromsmartphoneusage.Bothconsumergroups emphasize functionalvalue equally,andfunctional valuemight thuscontributetobrandloyaltywithoutagedifferences.
In addition,thisstudyfailed topredicttheeffect ofagedif- ferencesontherelationshipbetweenemotionalvalueandbrand loyalty.Thatis,agesignificantlystrengthenedtheeffectofemo- tionalvalueonbrandloyalty,andtheeffectofemotionalvalue onbrandloyaltywasgreaterasageincreased.Ourfindingswere differentfromthestudiesofBarutc¸u (2007)and Coates(2001), whichevidencedthatyoungerconsumersweremorelikelytoben- efit from theexploration of new mobile technologyand enjoy mobile entertainment services. As with the explanation men- tioned earlier, both middle-aged and younger consumers may experiencefunctionalvalueequally.Comparedwithyoungercon- sumers,middle-agedconsumersmayconfrontmorecomplicated challenges and tasks from work and daily life. The functional valueexperiencedfromsmartphoneusagemayhelpmiddle-aged consumerstomanagethesechallengesandtasksefficientlyand effectively and gain much more pleasure.This result may also beexplainedbythesocioemotionalselectivitytheorywhichsug- geststhatolderpeopleareinsearchofemotion-relatedgoalsand material(Charles,Mather,&Carstensen,2003).Thus,theeffectof emotionalvalueonbrandloyaltywasgreaterasageincreased.
Opposite to the results of Persaud and Azhar (2012), this studyshowedthemoderatinginfluenceofageontherelationship betweensocialvalueand brandloyalty waspositive. Thesocial valuemiddle-agedconsumersperceiveledtogreaterbrandloyalty thanyoungerconsumers.Theresultofthisstudymaybeinlinewith ChurchillandMoschis’s(1979)argumentthatsocialconsumption motivationincreaseswithageandhumandevelopmentmaturity.
Thus,thesymbolicrepresentationsofbrandsmightbemoreimpor- tantformiddle-agedconsumerstoearnsocialstatusandtofoster interpersonalrelationships.Ourresultsmayalsoechothefindingof Heckhausen(1997)thatmiddle-agedconsumerswere,compared withyoungerconsumers,inpursuitofcommunitygoals,andthus socialvalueismorecrucialforthem.Byandlarge,theeffectsof emotionalvalue,socialvalue,andbrandidentificationonbrand loyaltyvariedwithage.
Though Syed and Nurullah(2011) stated that the men and womenhave different mobilephone usagepatterns,this study foundthatthemoderatingeffectofgenderwasabsentinallthepre- dictedrelationshipsbetweenbrandloyaltyandthefourpredictors.
TheseresultssupportLeong,Ooi,Chong,andLin’s(2013)findingsof nosignificantdifferencesbetweenmenandwomenintheadoption ofmobileentertainmentservices,anditalsoagreeswithAlbert, Merunka,andValette-Florence’s(2013)findingsthattherewasno gendereffectontherelationshipbetweenbrandidentificationand brandcommitment.Theremaybetwopossiblereasonsforour results.First,thegenderindifferencesmaybeattributedtothecul- tureintermsofmasculinityandfeminine.Masculinity/feminineis adimensionwhichHofstede(1991)identifiedtoevaluatenational culture,andreferstothedegreetowhichgenderinequalitiesare espousedbyasociety(Srite&Karahanna,2006).Zhou,Dai,&Zhang (2007)addressedthatamasculineculturetendstoshowgreater genderinequalitiesorgenderdivide,andthesocialgenderroles aredistinctlyconstructed.Oppositely,thesocial genderrolesin a feminine culturemaybe highly overlapped.The respondents inthisstudywererecruitedinTaiwan,whichisalessmasculin- itysocietywithascoresofmasculinity/feminineevaluation(i.e., 45)comparedwiththe51sampledcountries(Mcountries=51.24) (Erumban&deJong,2006).Thus,theeffectsofgenderdifferences didnot significantin therelationshipsbetweenthefourdeter- minantsandbrandloyalty.Thesecondreason mightberelated totheresearchcontextinwhichsmartphonebrandswereunder evaluation.ThoughVenkateshandMorris(2000)findthattherea- sonstousecomputersintheworkplacemayvarybetweenmen andwomen,genderdifferencesmaynotsignificantlyexistinthe contextofsmartphoneusewhichisnotmandatory(Yol,Serenko,
&Turel,2006).Similartotheviewthatgendereffects maynot alwaysexistorfunctioninthesamedirectionacrossproducttypes (Dittmar,2005),contextualfactors,suchasvoluntaryandmanda- toryuse,mayinfluencetheoccurrenceofgenderdifferences.To sumup,manyresearchersannouncethatmenandwomenshow differentmobilephoneusagepatterns.Mostoftheirargumentsare basedonobservationsorqualitativeevidence(Lemish&Cohen, 2005; Srivastava,2005;Syed&Nurullah,2011;Walsh&White, 2007).Unlikethesestudies,theresultsofthisstudywhichwere analyzedwithsurveydatafoundnogenderdifferencesinanyof thefourrelationships.
6.2. Theoreticalimplications
Theempiricalfindingsofthis studyhavefivemaintheoreti- calimplications. First,this study validatesthat consumervalue and brand identification are two dominant grounds to predict consumers’loyalintentions.Specifyingconsumervaluewhichis classifiedintothreedistincttypes(i.e.,functional,emotional,and socialvalue)andbrandidentificationasthedeterminantsofbrand loyaltyjointly,ourresultsshowthat73%varianceofbrandloyaltyis explained.Thehigherthevalueandidentificationconsumersper- ceive,thegreatercommitmenttorepurchaseand recommenda givenbrandwillbe.Specifically,emotionalvalue,brandidentifica- tion,andfunctionalvaluehavehigherinfluenceonbrandloyalty thansocial valuewithnearlytwicethemagnitude (seeModels 2–3inTable4).Thisfindingimpliesthatsmartphonebrandloy- altymayprimarilydependonindividual-relatedfactorsratherthan interpersonalfactors.
Second, this study provides evidence for the effect of con- sumervalueonbrandloyaltywiththe3-valueframework. The significanceoffunctionalvalue,emotionalvalue,andsocialvalue indicatesthat consumersstay witha certainsmartphonebrand basedonavarietyofvalueevaluations.Theexplanatorypowerof consumervalueappearsnotonlyinaretailingcontext(Sweeney
&Soutar,2001)butalsoinatechnologyproductcontext.Leeetal.
(2011)arguethatproductattributeswhichareclassifiedasperfor- mance,appearance,andcommunicationattributesarerelatedto consumers’approachbehaviorinthecontextofhigh-technology products, and Horváth and Sajtos (2002) consider the three product-related consumer responses (utility/usefulness, experi- ence/enjoyment of use, and communicative power/expression) positivelyleadtothebuyingbehaviorofmobilephone.Basedon consumervaluetheory,thisstudymayextendthetwostudiesand elaboratethatthethreeproductattributesdelivercorresponding valuetoconsumersandpromptthemtopurchasetheproductofa brandrepeatedly.
Third,priorstudiesonself-brandrelationshipsuggestshorten- ingthedistancebetweenconsumersandbrandidentitybyfiguring outconsumeridentityandbrandidentityrespectively.Onemajor criticismofthis suggestionisabrandmayconfrontavarietyof consumeridentities,and itlimitsthegeneralizabilityofempiri- calresultsandtheirpracticalapplicability(Geuens,Weijters,&De Wulf,2009).Brandidentificationisaconsumer’ssubjectiveeval- uationof self-brandcongruence. In accordance withHe et al.’s (2012)findings,theresultsofthisstudydoimplythatbrandswill successfullyearnconsumers’loyaltyviatheirperceivedidentifica- tionwhilepurchasingsmartphones.Theestablishmentofasalient brandidentitymaybemorecontributivetogainconsumers’per- ceivedidentification.
Fourth,thisstudyverifiestheexistenceofmoderatingrolefor agedifferencesbutnotforgenderdifferences.Notalldemographics areexcellentpredictorstosegmentconsumerbehaviors.Similar toDittmar’sclaims(2005)thattheeffectofgenderdifferencesis product-specific,ourfindingsmaycorroboratethatgreatercareis necessarywhileusinggenderasamoderator.
Lastly, this study developed moderating hypotheses of age mainlybased onprior studies onmobilephone usageand age- related theories. The results revealed that three of the four hypotheseswereobservedsignificantly,andtheeffects ofemo- tional value/social value/brand identification on brand loyalty changedasageincreased.Onlytheeffectofagedifferencesinthe relationshipbetweenbrandidentificationandbrandloyaltymet ourexpectation.Itindicatesthatagingisahighlycomplexpro- cessand involvesvariousdevelopment stages.Our resultsmay implythat thedeterminantsof brandloyaltymaychangefrom identity-driventoemotional/socialvalue-driven(seeFigs.2–4and thestatementsinSection5.2).Ajointconsiderationofage-related theoriesmaypreventfromthedilemmaofagestereotypeandpro- videabetterpredictabilitybyclearly identifyingtheneedofan individualinagivenstage(Dengetal.,2014).
6.3. Managerialimplications
As our investigation in Table 2 indicates, 39% respondents revealed that the brand which they chose in their last two smartphoneshoppingexperienceswerethesamewhileonly19%
respondentsagreedthatintheirlastshoppingoffeaturephoneand smartphone.Patronageandcross-buyingofthesamebrandwere notcommonbehaviorsforconsumerswhenshoppingformobile phones,whichsuggeststhatahighermarketingeffortisneeded.
Giventhatconsumervalueandbrandidentificationcontribute tobrandloyalty,smartphonemarketers shouldnotonlydeliver functional, emotional, and social value to consumers, but also establishbrandidentityintermsofattractiveness,distinctiveness, andsaliencetoearnconsumers’identificationwiththebrand(Kim, Han,&Park,2001).Consideringtherelativeimportanceofthefour loyaltydrivers,smartphonesmanagershavetoputmoreemphasis ontheindividual-relateddrivers(i.e.,functionalvalue,emotional value,andbrandidentification).Thisstudyindicatesthatoutofthe threeindividual-relateddrivers,emotionalvalueisthemostsignif- icant,followedbybrandidentification,andlastly,functionalvalue.