Design-to-lure
in
the
e-shopping
environment:
A
landscape
preference
approach
Yung-Shao
Yeh,
Yung-Ming
Li
*
InstituteofInformationManagement,NationalChiaoTungUniversity,Hsinchu300,Taiwan,ROC
1. Introduction
The realization of the remarkable benefits brought by the Internetdependsonthewillingnessofe-customerstousewebsites totransactbusiness.Evidencerevealsthatwebsurfersaresatisfied with fast loadwebsite. When business takes place online, the Internet environment where the e-vendor is faceless does not allowe-shopperstoinspectproductsorinteractwiththecustomer receptionist.Thus,withoutanappealingstorefrontorstoreimage, e-shoppers find it difficult to evaluate whether e-vendors will deliver on their commitments [19]. Consequently, e-shoppers becomereluctanttoengageintransactingwithe-vendors[1,32]. The concept of website design is explored in this research becausewebsitedesignhasthepotentialofmakinge-shoppers decidewhetherornottheywillusethesiteinthefuture.When e-shoppersfirstengagewithanunfamiliarwebsiteandthewebsite itselfisnotlogicaltoshoppers,thewaythatwebsitedesignmakes e-shopperswillingtouseawebsiteandmakeapurchasewith e-vendorsis critical[10].Forthis reason,websitedesign playsan importantroleininducinge-shopperstotransactwithe-vendors and is thus critical to the successof an e-vendorin attracting
e-shoppers [13]. Essentially, e-shoppers rely on a website’s appearance, symbols, colors or whatever information it has, to arousethewillingnesstopurchase[21].
Lackofgooddesignon awebsitemakese-shoppershesitate beforetransactingwithe-vendors[22].Compoundingthisissue, there are studies examining e-stores as an environment or landscape affecting shopping behavior [11]. A ‘‘pleasant’’ envi-ronment influences shopping behavior variables, such as un-planned spending,duration ofstorevisit andsocial interaction. Thus,e-shoppersreactnegativelywhenputinanunpleasantweb environment.Consequently,theywillnotbringanybenefitssuch as sales to e-vendors [37]. Furthermore, when the website is viewedasaphysicallandscape[12],peoplewishtomakesenseof andgetinvolvedwiththeirlandscapes.Whene-storesdonotmake sense to e-shoppers, theywill not generate higher e-customer responseintheformofhighersalesvolume.
Websitedesignershaveanalyzedtheweb intermsofdesign aestheticstoensurethatthewebgivese-shoppersanenjoyable experience[28].Researchershavestresseddesignaesthetics,such asamixtureofcolors,shapes,language,musicoranimation,asan importanttoolinproducingeffectivewebsitedesign.Inanother vein,someresearchersareinvestigatingtheconceptofusability and howusabilityimpacts one-customer behavioral intentions suchasthereinforcementofe-loyalty[30].Moreover,perceived risksaswellasuncertaintywheninteractingwithe-vendors[19] arealsorelatedtowebsitedesign.Theseresearchtopicsoriginate
A R T I C L E I N F O
Articlehistory:
Received5March2014
Receivedinrevisedform4June2014
Accepted8June2014
Availableonline19June2014
Keywords: Electroniccommerce Websitedesign Multi-groupanalysis Cognitivepsychology A B S T R A C T
Withtheincreasingpopularityofonlineshopping,e-shoppershavebeenprovidedwithanewmedium
formakingpurchasesandthishasattractedincreasingattentionfromresearchersandpractitioners.
Researchers arechallengedto understandwhatconstitutesatheoreticalmodelforwebsitedesign
research.Inexploratorywork,weemployKaplanandKaplan’slandscapepreferencemodelinvolving
coherence,legibilityandcomplexity,andinvestigatetheirrelationshiptotrustandsatisfactionandtheir
impactone-shoppers’willingnesstobuy.Datafromasurveyof300shopperswereusedtovalidatethe
model.Amulti-groupanalysiswithgenderwasfurtherusedtocross-validateit.Theresultsshowthat
trustandsatisfactionaregreatinfluencesofwillingnesstobuy.Coherenceandcomplexityhavegreat
influencesontrustandsatisfaction,butlegibilityonlyhasadequateinfluencesonthesetwovariables.
Thestructuralweightsareinvariantacrossdifferentgendersubgroups.Implicationsforresearchersand
practitionersarealsodiscussed.
ß2014ElsevierB.V.Allrightsreserved.
* Correspondingauthor.Tel.:+88635712121x57414;fax:+88635723792.
E-mailaddresses:dyyeh67@gmail.com(Y.-S.Yeh),yml@mail.nctu.edu.tw
(Y.-M.Li).
ContentslistsavailableatScienceDirect
Information
&
Management
j our na l ho me pa ge : w ww . e l se v i e r . com / l oca t e / i mhttp://dx.doi.org/10.1016/j.im.2014.06.005
from the difference between e-vendors and brick-and-mortar vendors.Surprisinglyenough,theuseof atheoreticalmodel in website design to study e-customer behavior is very scarce, makingitaninterestingresearchproblemtoaddress.
A particular approachto theweb– landscapepreference – waschosenforthefollowingreasons.Landscapepreferenceisan evaluation on how people perceive the surrounding environ-mentandwhatpreferredlandscapeisforpeoplein mind[12]. Landscape preference views that environments could provide informationinmanydifferentways,suchasicons,signs,words ornon-words.Designingwebcontentisverysimilartodesigning aphysicallandscape[27].TheKaplanandKaplan’sframeworkof landscapepreference[12]providessomeguidelinesfor design-inganeffectiveweblandscape.Fromtheliteraturepointofview, veryfewstudiesfocusontheimpactoflandscapepreferencein website design on e-shopper behavioral intention and they rarely use sucha theoretical modelto empirically investigate howsuchawebsitedesignmodelaffectse-shoppers’cognitive perceptions.Then,froman understandingofmarketing strate-gies, the problem of how to convert web surfers into actual buyersis important.Thesuccessofe-vendorsdependsonthis conversion. Finally, for web developers, web design features should not be just limited to technology-driven or service quality-enhancing features, but should be extended to a cognitive approach that may also ultimately and effectively leadtocontinuedvisitsandrepeatbusiness. Awebsitewitha landscapepreferenceisreallyalensintoabetterunderstanding of what constitutes a high quality, yet interactive, website design. Inaddition, from cognitivepsychology, environmental psychologistssuchasKaplanandKaplan[12]suggestthatthe conceptoflandscapepreferencecanincreaseourunderstanding of how the interactive experience can be gained in such a setup [12]. What entails an interactive experience is often a pleasantfeeling[37]andsubsequentlypurchasebehavior[22]. Thus, an investigation from the perspective of cognitive psychologyis clearly one way of examiningthe development ofeffectivewebdesign.Weproposearesearchmodel,developed from existing research, explore and validate the path factors in landscape preferences thataffect the cognitive perceptions of e-shoppers and their impact on e-shoppers’ willingness tobuy.
Theremainderofthispaperisorganizedasfollows.Inkeeping withexistingresearch,wefeltthatthecognitiveaspectofwebsite development is important and may have implications on e-shopperbehavioralintentions.Thus,wesummarizetherelevant literatureanddevelopourhypothesesinSection2.Then,Section3 explainstheresearchmethodologyandintroducestheexperiment, followedbySection4 whichprovides theresultsofthemodel. Section5givesadiscussionoftheresultsandoftheimplicationsas well as potential limitations of the study. Finally, Section 6 concludesthispaper.
2. Literature
Relevant works on utilizing the principles of landscape preferencetoconstructtheoriesaboutwebsitesarescarceinthe literature.Soastocomplementtheliterature,thepresentstudy identifiedthose components that a websiteshould incorporate especially whenit is designed along theseprinciples.Based on prior works, three components were identified: coherence, legibility,andcomplexity.Althoughthesevariableswereidentified andtheireffectivenesswassupportedbyrelevantliterature[27], the efficacies of these design components in making the site effectiveandingeneratinghighercustomerresponseintheformof abehavioral intentiontopurchasewerestillnot clearly under-stood.
2.1. E-shoppingbehaviorintentions
Theresearchone-commerceisextensiveandonlinebehavioral intentions suchaspurchasing haveremainedoneofthefastest growingareasofInternetresearch[9,15].Thisisthecasebecause theInternet environmentis enabledby informationtechnology andnumerousstudiesoftechnology acceptancehavemeasured behavioralintentionsratherthanbehaviors[15].More important-ly, Venkatesh and Davis [33] confirmed a strong correlation betweenbehavioralintentionsandactualbehaviors.Inaddition, Mcknightet al. [23] definede-shopping behaviorintentions as anyonewhoiswillingtotransactwiththevendorthroughtheweb. Topurchase,acustomerneedstosharepersonalinformation,such asname,telephonenumbers,address,andcreditcardnumbers.In this situation, the customer must have shopping behavior intentionandthenprovidepersonalinformationtothewebsite.
Formostshoppingsites,themajorobjectiveistopersuade e-shopperstomakeapurchase[32].Thate-shopperspurchasemay be a result of high trust and high satisfaction. Moreover, antecedentfactorswithinthetrustandsatisfactiondomainsare also effective at luring e-shoppers and converting visitors into buyers. Addressing technology-based factors such as social presence[9],cognition-basedfactorssuchasinformationquality [15], and institutional-based factors such as reputation [5], researchersfoundthesevariablestohave asignificanteffecton developing consumer trust [23]. This rationale taps into the traditionalonlineshoppingsequence(e.g.[15]).
Thedevelopmentoftrustandsatisfactionhasbeenshowntobe sensitivelyassociatedwithvariouswebsitedesignfactorsbyprior studies.Moreover,asconfirmedbyCyr[5],trustandsatisfaction arerecognizedasimperativeantecedentstoonlinepurchasingin differentcultures(i.e.IndianandHongKong)andcloselyrelatedto website success or effectiveness. Hence,we wish to test their relationshipstoawillingnesstoshop,butinanotherculture(i.e. Taiwan).Collectively,ourunderstandingofotherwebsitedesign factorsand theirresearchfocushasbeenenrichedasshownin AppendixA.
2.1.1. Trustine-shopping
Trustisthebasicconnectioninhumansociety,theessential factorbetweeninterpersonalcooperation,andisalsotheprimary mechanisminmanyeconomicalactivities.Evidencesuggeststhat consumersoftenhesitatetotransactwithe-retailingbecauseof uncertainty about vendor behavior [2] or theperceivedrisk of havingpersonalinformationstolenbyhackers[19].Trustplaysa centralroleinhelpingconsumersovercomeperceptionsofriskand insecurity. Trust makes consumers comfortably share personal information, make purchases, and act on web vendor advice— behaviorsessentialtowidespreadadoptionofe-commerce.Asfor antecedentfactorsoftrust,WalczuchandLundgren[34]indicated thatpsychologicalantecedentssuchasreputationand word-of-mouthcanbeusedtobuildconsumertrustine-retailing.Thusthe followinghypothesiswastested:
H1. Higherperceivedtrustwillresultinhigherwillingnesstoshop onwebsites.
2.1.2. Usersatisfactionine-shopping
In another vein, Zviran et al. [38] tested the effect of user satisfaction on theacceptance of e-shopping in thetechnology sector.It wasfoundthat satisfaction significantly influences e-shopperonlinebuyingbehavior.LiandYeh[18]foundsatisfaction tobethemainvariableinfluencingcustomerbuyingdecisionsin mobilecommerce.Thispoint ofviewwasalsoposited by[35]. Otherantecedentfactorsofsatisfactioninrelevantfieldsfromthe utilityaspect suchasconvenienceand usabilityandperception
aspectsuchassitedesignandsecuritywereidentified.Thusthe followinghypothesiswastested:
H2. Higherperceivedsatisfactionwillresultinhigherwillingness toshoponwebsites.
2.2. Cognitivelandscapeaswebsitedesign
Based on theconcept of physical landscape, theconcept of landscape preference examines physical environments in an attempttocreatepatternsforenvironmentaldesignthatinclude enduserenvironmentalcues,makingiteasierforpeopletoprocess information easily and effectively [12]. Applying cognitive psychology,KaplanandKaplan[12]foundthatthrough informa-tionstimulisuchassigns,icons,ortextembeddedinthephysical landscape,peopleformacognitivemap,amentalrepresentationof therealenvironmentata particularmomentintime.Cognitive mapsarealsoknownasmentalmaps,cognitivemodels,ormental models [36]. This concept has been widely used in web site usabilityandnavigationsystems[36–38].Cognitivemapcanbe recognizedasabasiccomprehensionofthestructureoftheweb site. A cognitive map enables people to make sense of the landscape or environment. This map influences ‘‘how the landscape feels to that person, what is noticed, and what is ignored’’[12].Moreover,KaplanandKaplan[12]suggestedthat the cognitive map must be extended to include peoples’ motivation to useand process information. Recent researchers have suggested that the cognitivemap undergoes a quick but sequentialformationduringtheprocessofextension[16].Lackofa mapextensionmaycausepeopletohavetroubleunderstanding information.
Thestructureofacognitivelandscapecanbeanalyzedina22 (making sense vs. involvement)(two-dimensional vs. three-dimensional)perceptualstructure.BasedonKaplan[12],ina two-dimensional environment, a cognitive landscape represents an immediateassessmentofsuchanenvironment.Someonemaking senseoftheenvironmentisabletocomprehendthesurroundings andtounderstandthecurrentsetup,ofteninalargersituationas well.Thatpersonbecomesinvolvedwiththeenvironmentandhas his/herinterestsmaintainediftheenvironmentissupportedwith richelements.Therefore,inasupportive environmentwithrich elements,theprocessof makingsensetakesstimulitomake it easiertocharacterizeandsummarize(coherence)andmakesthe subjectsustainhis/herinterest(complexity).However,sucha 2-dimensional setting is not sufficient to activate someone’s cognitivemap.
Themakingofanimmediateassessmentistrackedby[12].The cognitivelandscape moves from2-dimensional (coherence and complexity)to3-dimensional(legibility andmystery).To pene-tratedeeperinthecognitivelandscape,theenvironmentmusthelp the subject stay oriented (legibility) and provide perceived curiosityandfancy(mystery).
Finally, Rosenand Purinton[27]served asa catalyst toour study.Theyincorporatedtheconceptofacognitivelandscapeinto websitedesign.Basedontheargumentthathumansarecognitive creaturesandtheInternetisaninformation-ladenenvironment, theysuggestedthatguidanceindesigningaphysicallandscapeis alsoapplicabletotheonlineenvironment.RosenandPurinton[27] used a modified Preference Framework, based on a cognitive landscape [12], to provide guidance on designing cognitive landscapes.According tothePreferenceFramework,a cognitive websitemustaccommodatethreecomponents:coherence, com-plexityandlegibility.Thefocusofthisresearchistoinvestigate howthesecognitivevariableswillbeusedtofacilitatee-shopper informationprocessingand todemonstratehowe-shopperscan useinformationtosatisfytheirneedtomakesenseofandexplore
anuncertaincyberworld.Ultimately,a positiveoverallwebsite impression will in turn influence e-shoppers’ willingness to purchase.
2.2.1. Coherence
Coherencereferstotheabilityofawebsitetoprovideconsistent and meaningful informationorganizedin anorderlyand struc-turedway[12].Whilethesefeaturesallowe-shopperstoeasilyfeel andtocorrectlyunderstandthedesignofthewebsite,theyalso facilitatetheshoppingprocess.Todothis,e-vendorscommonly usesimilardesignaestheticssuchaslayout,color,fontaswellas frames across web pages to enable e-customers to have a consistent experience across web pageswhen surfing. Using a website as a storefront, e-vendors often use an encryption mechanism or guarantees of satisfaction or a refund policy to enhance e-customers’ trust in conducting transaction activity. Since itis truethat e-customersareoftenalready interestedin receivingcompanyorproductinformation,itisnotuncommonto findthatane-vendorusesitsreputation(i.e.awardsinrelevant fields)toassuree-customerconfidence.Fromane-vendor’spoint of view, its store image should resonate on the site and its messages shouldalsobeclearly communicatedtoe-customers. Thisresultinhighertrustandmoresatisfaction[38]fore-shoppers whentheyareonawebsite,whichfurthermakese-shoppersmore willingtoprovidetheirpersonalinformationsuchasdateofbirth and credit card information to e-vendors to complete the transaction[23].Wethereforeproposethefollowinghypotheses: H3a. Higherperceivedcoherenceofwebsiteswillresultinhigher trustine-vendors.
H3b. Higherperceivedcoherenceofwebsiteswillresultinhigher satisfactionwithe-vendors.
2.2.2. Legibility
Legibilityisdefinedastheeaseofnavigationaroundawebsite [17].Legibilitycanbeseenaslowdisorientationofnavigationina website. To measure the legibility, one can examine apparent efficienciesinusernavigation.Toolssuchasvalue-addedsearch engine,landmarks,appropriatehyperlinksorframesareseenas essentialtohelpawebsitebecomenavigational.Effectiveuseof thesetoolsina websitethat isdesignedon onepageorscreen resultinlow errorrates andlessdisorientation,which allow e-shopperstoefficientlyfindwhattheywant[17].Retrievingdesired informationeasilyandquicklycanbeareasonfore-shoppersto staylongeronawebsite.Theywillalsostaylongeronwebsitesthat have lightcognitiveloadand createenjoymentandsatisfaction [16].Hence,itseemsthate-shopperswillbemorewillingtoplace trustonawebsitethatcanbeeasilyandquicklynavigatedtofind thedesiredinformationwithminimumeffort.E-shopperswillalso havehighersatisfactionwhentheirneedsaresatisfiedwithless disorientationandmoreenjoyment[38].Therefore,we hypothe-sizethat:
H4a. Higherperceivedlegibilityofwebsiteswillresultinhigher trustine-vendors.
H4b. Higherperceivedlegibilityofwebsiteswillresultinhigher satisfactionwithe-vendors.
2.2.3. Visualcomplexity
Visual complexity is defined as the ability of a website to provide diverse website components capable of creating vivid interactionandarousinge-shoppers’interest[27].E-shoppersare generallyimpatient.Geissler[10]suggestedfivetothirtyseconds asaruleofthumbforwebsitedesignerstograbane-shopper’s
attention.Oncetheyfeelbored,e-shopperssimplyleavethesite. Becauseofthis,therearevariouswebsitedesigncomponentsthat help e-shoppers develop a sense of presence in the Internet environment.Forexample,someinformationorimagesaboutthe productandbestsellerareprovidedtoincreasesociability.Insome popularyoungadultclotheswebsites,suchasAbercrombie&Fitch (A&F) or ASOS, theyintroduce their products by vivid product photos.Insocial presenceperspective,usingimage communica-tion hasbetter effect than using text-only communication [7]. Moreover,videoclippingenabledbyYoutubehavingarealsales personintroducetheproductcanbefoundtodeepenengagement. Bycombiningthese,e-shoppershaveamorepleasantexperience and havea positive attitudetoward e-vendors. Thus, it canbe reasonedthate-shopperswillhavemoretrustina websitethat allowsthemtogathermoreusefulproductinformationorrelevant images.Similarly,e-shopperswillhavemoresatisfactionforthe samereasons.Hence,wehypothesizethat:
H5a. Higherperceivedvisualcomplexityofwebsiteswillresultin moretrustine-vendors.
H5b. Higherperceivedvisualcomplexityofwebsiteswillresultin highersatisfactionwithe-vendors.
3. Researchmethodology
3.1. Instrumentdevelopmentandsubjects
Todeveloptheinstruments,weconductedaliteraturereviewto ensurethatthequestionsposedarerepresentative.However,since cognitive landscape applied to IS research is still scare, we employedsomeconstructs fromgenerallyacceptedconcepts of landscapepreferenceandmadesomemodificationstoreflectthe researchcontext.Asa result,constructsofcoherence, legibility, visualcomplexity,trust,satisfactionandwillingnesstobuywere usedand18itemsweregenerated.TheyarelistedinAppendixA alongwithcorrespondingliteraturesources.
Bothapre-testandapilottestwereconductedtoensurethe clarityof contentand appropriateness of aspects. The pre-test involvedthreerespondentswho wereexperts in thefieldof e-commerce.Respondentswereaskedtocommentonthelengthof the instrument, wording, and item order. Then, a pilot test involving 50 subjectsselected from one of the two previously selecteduniversitiesinnorthernTaiwanwasconductedtoensure theclarityofcontent.Allitemsthatwereincludedweremeasured onafive-pointLikertscale,rangingfrom‘‘stronglydisagree’’(1)to ‘‘stronglyagree’’(5).
AccordingtoaTaiwannationalsurvey[26],70.2%ofInternet usersareconcentratedinnorthernTaiwanand90.2%ofInternet usersarecollegestudents.Since[5]advocatedtheuseofstudents for e-commerce research, in order to exploit our research resources,wechose twouniversitiesinnorthernTaiwanasour research samples. These two universities are located in two differentcities, which have 75.9% and 75.5% Internet adoption rates,ranked1stand2ndamongcitiesinTaiwan.Afterobtaining permission from class instructors, information such as the backgroundandpurposeoftheresearchaswellasaquestionnaire hyperlinkwereemailedto450studentsenrollingat undergradu-atelevelandgraduatelevelonISclassesbetweenOctober2009 and November 2009. To increase response rate, subjects were incented with extra credit for their course. For those who volunteered and reported having at least one online shopping experience are based on five popular categories (i.e. clothing/ accessories,books/magazines,bookingoftravelticketsandhotel, familyitems,and3Cproducts),wecollectedtheir demographic andquestionnairedataonlineanddataweresavedontheweb
server.Atotalof300(responserateof66.7%)subjectssuccessfully completedthesurvey.Comparedwithpreviousstudies,thissizeof samplewassuitableforfurtherstatisticalanalysis[5].
3.2. Experimentalmanipulations
Theexperimentconsistsoftwosteps.Thefirstispresentinga purchasingscenarioofanonlinetravellingpackagetoparticipants and next is asking them to make a purchasing of an online travellingpackagebybrowsingtheselectede-travellingwebsites for at least 5min before completing the questionnaire. The experiment has two questions for participants: (1) what was notincludedintoday’spromoteditemsonthefrontpage?(2)What wasthecurrenttargetedcountryfortravelling?
All tasks were completed online to increase the realism of onlineshopping.AppendixCshowsthescenario.Thegoalofthe scenario was to stimulate participants’ memory of navigating around these websites. We selected three online travel sites (http://www.eztravel.com.tw/, http://www.settour.com.tw/, and http://www.liontravel.com/) as targetsfor participants to navi-gate.Thesesiteswerewell-knownbytheselectedstudentsubjects and were top-ranked in the field of travel. By selecting well-recognized websites, participants experienced less stress. Also, since participants are well familiar with these websites, the trainingtimecanbereduced.
The questionnaire comprised two parts. The first part was intended to measure potential e-shopper perception of each construct in the model. The second part recorded e-shopper demographic data. Demographics of the subjectsare shown in Table1.
3.3. Instrumentvalidityandreliability
Itemsselectedforconstructswereprimarilyadaptedfromprior studiestoensurethevalidityofcontent[4].Inourstudy,website satisfaction is considered as a measure of the utility of an e-commercewebsiteanddefinedastheattitudetowardthewebsite [24].InKaplanandKaplan’smodel,coherenceistomeasurethe abilityof awebsite toprovideconsistentand orderlycontents, structures, and multimedia components.E-shoppers can easily grasptheorganizationofawebsitebyprovidingacommonlook and feel toeach page, and this positively affects their attitude towardthewebsite[16].Ontheotherhand,legibilityistomeasure theeaseofnavigatingasite.Soe-shopperswillbepleasedwhen theycanbenavigatedthewebsiteeasilyandquicklytofindthe informationtheywant[16].Complexityimpliesthewebsitedesign containsavarietyofimagesthatsatisfythedesiretoexplorethe environment[27].Usingproductimagesthatfitthesitecontent ties the complexity of the site to its content enhancing comprehensibility and encouraging rather than discouraging exploration. So e-shoppers wanted more visual effects and graphicsasthiswouldimprovetheonlineshoppingenvironment [27].Trustisbasedontheexpectationthatatrusteewillactinthe interestsofthetrustorwithoutaguaranteeandthatitisinvolved withminimizingtheriskofharm[37].Finally,willingnesstobuyis basedontrustandsatisfaction.Thisincludeswillingnesstopayto accessinformationonthewebsiteandhaspossibilitytobuyinthis website[23].
Inthisresearch,constructsofcoherence,legibility, complexi-ty,trust,satisfactionandwillingnesstobuycamefromexisting literature, which demonstrated strong content validity. AppendixBshowsthe18surveyitemsusedinthequestionnaire. ConstructreliabilitywasexaminedusingCronbach
a
-values. [25]recommends thata
-values shouldbe greaterthan 0.7 for itemstobeusedtogetherasaconstruct.AsshowninTable3,a
-values ranged from 0.906 (willingness to buy) to 0.816(complexity).Therefore,allconstructspassedthetestofconstruct reliability.
Constructvalidity examinestheextenttowhich a construct measures the variable of interest and ensures that there are relatively high correlations between measures of the same construct (convergent validity) and low correlations between measuresofconstructsthatareexpectedtodiffer(discriminant validity) [29]. An exploratory factor analysis with a varimax rotationwasperformedon6latentvariablesthatcorrespondto8 hypotheses[25].Factorextractionwasbasedontheexistenceof eigenvalueshigherthan1.Asaruleofthumb,ameasurementitem ishighlyloadedifitsloadingcoefficientisabove0.6andisnot highlyloadedifthecoefficientisbelow0.4[14].Table2showsthat allloadingswerewellabove0.6(convergentvalidity).Thereare someotherwaystoexamineconvergentvalidity.[14]proposed assessingthreecriteria:(1)allfactorloadingsshouldbesignificant atP<0.05 andexceed0.7,(2)thecompositereliabilityofeach constructshouldexceed0.7,and(3)theaveragevarianceextracted (AVE)byeach constructshouldexceedthevariancedue tothe measurementerrorforthatconstruct[14].Table3showsallfactor loadings that surpass 0.7 and are significant at P=0.001. Composite reliabilities ranged between 0.86 to 0.92 and AVE valueswerewellabove0.5. Asshown inTable3,allconstructs demonstratedconvergentvalidityonallthreemeasuresproposed by[14].
Discriminantvaliditywasbasedontheprinciplethatconstructs differed from each other. By satisfying the criteria of [14] for discriminantvalidity,thesquaredcorrelationsbetweenitemsin any two constructs shouldbe lowerthan theaverage variance sharedbyitemswithinaconstruct.AsshowninTable4,theresults showedthatthesquaredcorrelationsofeachconstructwereless than the average variance extracted (AVE) by the indicators measuringthatconstruct.
4. Results
To test the proposed hypotheses, the Structural Equation Modeling(SEM)approachwasadopted,supportedbyAMOS5.0 withamaximumlikelihoodestimation.SEMcansimultaneously test structural and measurement models and provides a more completeanalysisoftheinter-relationshipsinamodel[14].SEM has received popularity in several fields including marketing, psychology,socialscienceandinformationsystems.
Results of SEM include two components: the measurement modelandthestructuralmodel.Themeasurementmodel,giving relationships between latent variables and observed variables, aimstoprovidereliabilityandvaliditybasedonthesevariables. Thestructuralmodel,ontheotherhand,studiesthepathstrengths anddirectionofrelationshipsamonglatentvariables[14].
Aconfirmatoryfactoranalysiswasconductedtovalidatethe measurement model. It was measured by examining if the measurement model has acceptable goodness-of-fit measures. Seven common model-fit measures were used to assess the model’s overall goodness of fit: the ratio of
x
2 todegrees-of-freedom(d.f.),goodness-of-fitindex(GFI),adjusted goodness-of-fitindex(AGFI),comparativefitindex(CFI),normfitindex(NFI),
Table2
Exploratoryfactoranalysis.
Items Coherence Legibility Complexity Trust Satisfaction Willingness
tobuy C-1 0.862 0.037 0.056 0.148 0.131 0.005 C-2 0.842 0.074 0.044 0.119 0.190 0.151 C-3 0.815 0.82 0.120 0.098 0.043 0.146 L-1 0.055 0.851 0.052 0.013 0.041 0.099 L-2 0.017 0.855 0.092 0.112 0.160 0.026 L-3 0.069 0.822 0.064 0.139 0.092 0.110 CX-1 0.065 0.123 0.799 0.159 0.109 0.151 CX-2 0.045 0.051 0.835 0.219 0.083 0.120 CX-3 0.006 0.049 0.818 0.024 0.181 0.171 T-1 0.124 0.086 0.174 0.835 0.060 0.217 T-2 0.201 0.097 0.128 0.861 0.094 0.173 T-3 0.100 0.096 0.138 0.794 0.305 0.170 S-1 0.085 0.196 0.160 0.050 0.844 0.128 S-2 0.101 0.088 0.121 0.194 0.810 0.350 S-3 0.161 0.075 0.175 0.256 0.776 0.318 WTB-1 0.135 0.016 0.155 0.272 0.221 0.800 WTB-2 0.073 0.105 0.212 0.157 0.235 0.847 WTB-3 0.139 0.115 0.171 0.180 0.254 0.842 Mean 3.54 3.66 3.44 3.55 3.48 3.46 SD 0.71 0.69 0.73 0.83 0.89 0.81 Table1 Demographicsofsubjects.
Measure Items All Male Female
Age Below20 2 1 1
20–25 154 55 99
26–30 128 70 58
Over30 16 5 11
Total 300 131 169
No.yearsshoppingonline Years<1 2 1 1
1Years< 3 197 75 122
3Years<5 90 50 40
Years5 11 5 6
Total 300 131 169
No.purchaseslastyear 1Times<3 2 1 1
3Times<5 152 58 94
Times5 146 72 74
Total 300 131 169
Table3
Constructvalidity.
Construct Items Loadings Cronbach’s
alpha Composite reliabilities Average variance extracted Coherence C-1 0.74 0.826 0.87 0.70 C-2 0.81 C-3 0.79 Legibility L-1 0.74 0.818 0.87 0.70 L-2 0.83 L-3 0.77 Complexity CX-1 0.79 0.816 0.86 0.67 CX-2 0.79 CX-3 0.73 Trust T-1 0.82 0.877 0.87 0.71 T-2 0.89 T-3 0.81 Satisfaction S-1 0.74 0.884 0.90 0.76 S-2 0.91 S-3 0.89 Willingness tobuy WTB-1 0.90 0.906 0.92 0.80 WTB-2 0.86 WTB-3 0.85 Table4 Discriminantvalidity. Discriminantvalidity 1 2 3 4 5 6 (1)Coherence 0.70 0.12* 0.14* 0.34** 0.29** 0.29** (2)Legibility 0.12* 0.70 0.21** 0.24** 0.29** 0.22** (3)Complexity 0.14* 0.21** 0.67 0.38** 0.39** 0.43** (4)Trust 0.34** 0.24** 0.38** 0.71 0.44** 0.50** (5)Satisfaction 0.29** 0.29** 0.39** 0.44** 0.76 0.59** (6)Willingnesstobuy 0.29** 0.22** 0.43** 0.50** 0.59** 0.80 *
Denotessignificanceatthe.05level.
*
Denotessignificanceatthe.01level.
***
rootmeansquareerrorofapproximation(RMSEA)androotmean squareresidual(RMR).Theseindiceswereemployedtotest the goodness-of-fitnotonlyofthemeasurementmodelbutalsoofthe structural model. Overall, seven goodness-of-fit measures, as showninTable5,suggestedthatbothmodelsadequatelyfitthe data,followingthesuggestedcut-offvalueguideline(Fig.1).
Fig.2showstheresultsofthestructuralmodel.Theresultsin Table6showthatallpathcoefficientsofhypothesizedcausallinks weresignificant.Moreprecisely, coherence, legibilityand com-plexitysignificantly influenced e-shoppertrust and satisfaction and explained the variance of 35.7% in trust and 33.5% in satisfaction. Moreover, e-shopper trust and satisfaction were foundtosignificantlyinfluencee-shopperwillingnesstobuyfrom ane-vendor.Approximately51%ofthevarianceinthewillingness tobuy online wasaccounted for by the variables of trust and
satisfaction in the model (R2=0.509). Thus, hypotheses H1
through H8 were supported. All the R2 of the endogenous
constructsinthemodelexceedthe10%benchmarkrecommended by[14]. Coherence Legibility Visual Complexity Trust Satisfaction Willingness to buy R2 =0.357 R2 =0.335 R2 =0.509 0.29*** 0.59*** 0.33*** 0.27*** 0.14* 0.18** 0.37*** 0.37***
Fig.2.Standardizedstructuralmodelanalysis.
Coherence Legibility Visual Complexity Trust Satisfaction Willingness to buy H1 H2 H3a H3b H4a H4b H5a H5b
Fig.1.Proposedresearchmodel.
Table5
Fitindicesforthemeasurement.
Recommendedcriteria Measurementmodel Structuralmodel
x2 /d.f. <3.0 1.52 1.68 GFI >0.9 0.93 0.93 AGFI >0.8 0.91 0.90 CFI >0.9 0.98 0.97 NFI >0.9 0.94 0.93 RMSEA <0.08 0.042 0.048 RMR <0.05 0.034 0.046
5. Discussion
This study deals with cognitive psychology factors in e-commerce and further enriches its content by developing the relationshipswithe-shoppers’ trust andsatisfaction. Thesetwo constructs, trust and satisfaction, are often referred to as precursors for a website’s success [5,9]. On the other hand, coherence indicates that e-shoppers feel easily and correctly understand thedesignof thewebsite.Thiscanfacilitate the e-shopping process and e-shoppers would have good shopping experience.Inthiswebsite,e-shoppersarewillingtoprovidetheir personalinformation tocomplete thetransaction.Theprevious situationmust be based on highly trust and satisfaction for e-vendors.Thisresulthasalsobeenconfirmedbypreviousstudies [23,38]. Legibility indicates the ease of navigation around a website.Foraclearstructureandlayoutofawebsite,e-shoppers can easily and quickly retrieve information their desired and higherpossibilitytostaylongeronawebsite.Inotherwords,this comfortable e-shoppingenvironment canreduce cognitiveload andcreatesatisfactionfore-customers.Soe-shopperswillplace trustinaneasilynavigationalwebsiteandbuytheproductsthey desire.
By providing clear and vivid productphotos, a websitecan attracte-customersandhaschancestoletthemstaylongerfor further shopping.Visual complexity plays animportant role to demonstrate the specific characteristics of the products. If e-vendorsoffermorevisualdesignson theirproducts,e-shoppers canunderstandtheproductseasilyandplacehighertrustforthis e-vendors.Thusthis can betruethat vivid productphotoscan satisfye-shoppers’ desiresfor productinformation, which text-onlydescriptionforproductcannotmake[36].
Theproposedmodelhadrobustpsychometricpropertiesand wascapableof explaining muchof thevariance in e-shoppers’ willingnesstobuy.AccordingtothepathmodelofFig.2,wefound thate-shopperwillingnesstobuycanbetriggeredbytrustand satisfaction, bothof which weresignificantly influencedby the three proposed cognitive constructs: coherence, legibility and visual complexity. Altogether, these factors explained approxi-mately 51% of the variance in e-shopper willingness to buy. Consistent with Lee and Kozar [16], what were identified as theoreticalusabilityfactorsin websitedesignexerted power to influenceane-shopper’swillingnesstobuy.Theirfindingsindicate thatablendoflegibility,coherence,andcomponentsandcontent designedbyvisualcomplexitycansuccessfullycommunicatewith onlineconsumersandinvokepositiveappraisals.
Theresultsofthestudyalsorevealthatvisualcomplexityis equally important to an e-shopper’s cognitive perceptions (i.e. trust and satisfaction). This is not surprising because visual complexity was defined in an operational sense here as the capabilityofawebsitetocreatevividinteractionsandtoarouse
e-shoppers’ interest instaying.Thus, seeing somethingvisually appealing(visualcomplexity)thatalsoprovidesexpectedcuriosity allowsthee-shoppertodrawinferencesoftrustworthiness and satisfactionaboutthee-vendor.Te’eni[31]andLeeandKozar[16] indicatedtheimportanceofthishedonicfeatureofacommercial websiteasthiscanentailacognitiveappraisalwhichdirectlyleads toe-shopperbuyingbehavior.Moreover,theresultsofourstudy indicatethatcoherenceisthestrongestfactor.Bothlegibilityand visualcomplexityhavesignificantimpactontrustandsatisfaction, suggesting that e-vendors should providerelevant content and product photos that satisfy consumers’ desires for website or productinformation.
A multi-group analysis was conductedin order to examine whether the proposed model and instruments were invariant acrossdifferentsubgroups(e.g.gender)[8].Ontheonehand,the analysisextendsthegeneralizabilityofmeasurementitems;onthe otherhand,itdirectlycomparesthestructuralweightsbyusing equivalentmeasurementswhereobservedscoresfromdifferent groups areon thesamescale. Thisanalysisgainssupportfrom Dengetal.[8]whosuggestedthatamulti-groupanalysismaybea moreappropriatetestofdifferencesinstructuralweightsthanan analysisofcovariancewhenthesamplesineachgroupexceedthe minimumof100.AppendixDprovidesthedetailedproceduresof this analysis. Nevertheless, some interesting differences were foundfromthemulti-groupanalysis.Forgender,allhypothesized relationships were significant except legibility–trust (Lamb-da=.06, P>0.05) and legibility–satisfaction (Lambda=0.13, P>0.05)forthemalegroup.However,noneofthehypothesized relationshipswereinsignificantforthefemalegroup.Satisfaction (Lambda=0.58, P<0.05) showeda significantly stronger influ-ence on e-shopper willingness to buy than trust in males (Lambda=0.30, P<0.05). This is also true for females (Lamb-da=0.53,P<0.05andLambda=0.27,P<0.05forsatisfactionand trust, respectively). Since satisfaction is primarily an affective evaluative response [20], thesignificant resultsin oursamples suggest that females may be more task-oriented and more emotional than males. Finally, complexity was the strongest factorinfluencingsatisfactionformales,whilecoherencewasthe strongestforfemales.
From a theoretical perspective, this study proposed and validateda newmodelfore-commerce,especially ine-shopper buyingintention.Althoughtheliteraturehasprovideda compre-hensiveunderstandingofwhypeoplebuyonlineandtherationales behindthisbuyingintention,asawaytoinnovateandcontribute totheliterature,ourmodelincorporatesconceptsfromcognitive psychologyandvalidatesitsrelationshipwithe-shopperbuying intention. The significantresultssuggest that a scrutinyof the relationship between the proposed constructs may provide anothereffectiveway ofexplaining e-shopperbuyingbehavior. Another academic implication is that the multi-group analysis extends the generalizability of results across different gender groups.Thefindingsimplythatthemodelisrobustandthatthe resultscanbeappliedtobothmalesandfemales.Femalesmaycare aboutthewebsitefunctionsandcomponentsthatcanfacilitatethe e-shoppingprocess.ThismaybetrueinTaiwanbecausefemales prefer a grounded and secure environment to complete the transaction.Forinstance,onefamouscosmeticsproductwebsitein Taiwanemphasizesproductinformation,especiallyiningredients. Thisisveryusefulforfemalestoconsiderwhetherornottobuythe product.Ontheotherhand,malescareaboutthevisualeffectofa website that includes vivid photos or video clips (visual complexity). This may be true because males prefer highly interactiveenvironment,suchastheonlinegamewebsite.
Fromapracticalperspective,theresultsofthisstudycanhave directimplicationsforwebsitedevelopers.Repeatpurchasesare deemed essential to differentiate a successful from a failed
Table6
Resultsofhypothesistesting.
Hypothesis Causalpath Pathcoefficient t-Values Supported
H1 T!WTB 0.28 5.34 Yes H2 S!WTB 0.50 10.18 Yes H3 C!T 0.41 5.30 Yes H4 C!S 0.36 4.53 Yes H5 L!T 0.20 2.33 Yes H6 L!S 0.26 2.95 Yes H7 CX!T 0.48 5.53 Yes H8 CX!S 0.51 5.73 Yes
Notes:T,trust;S,satisfaction;C,coherence;L,legibility;CX,complexity;WTB,
willingnesstobuy.
*
Denotessignificanceatthe.05level.
*
Denotessignificanceatthe.01level.
***
endeavor for an e-vendor.Website developers should consider featuresthatfacilitateorientation,increasecuriosityandmaintain informationandlayoutconsistency,astheycanevokee-shoppers’ willingnessto buy.Website developers arealso encouraged to exploreothermeansofincreasingenjoymentandfunthatcanbe experiencedonane-vendorsite.
6. Conclusion
Despitethesevaluableimplications,likeallotherstudies,this studyhadsomelimitations.Atfirst,theresultsderivedfromthe three websites built in Taiwan cannot be overlooked. Other demographicinfluencesarestillinneedoffurtherinvestigation. Priorstudiessuggestthatculturaldifferences,subjects’agesand attitudes also exert influence in explaining e-shopper buying intentions. Second, by selecting well-recognized online travel agenciesthatareequippedwithsimilardesigns,ourstudydidnot takeanyaccountofwebsitevariation.Asaresult,furtherinsights acrossdifferent websites in different service domains must be carefully interpreted and compared. Future work may employ othertypesofcommercialwebsites.Differentperceptions deliv-eredbydifferentwebsitelayoutanddesignmayresultindifferent results.Third,becauseoursamplesare universitystudents, the generalizationofourresearchtoelderpeoplemayhavelimitation. Finally, becauseof thechosen methodology, the data is cross-sectional.Futureworkshoulddeterminetheextenttowhichthe findingspresentedinthispapercanbeappliedtoothertimes.
In ‘‘IT doesn’t matter,’’ Carr [3] argued the role of IT and suggestedthat advances in ITexceed userrequirements. Com-pounding this issue, Cyr et al. [6] employed the concept of aestheticsin IT,which redefined theexperience of using IT as experienceenhancinginsteadoffunctionalityimproving.Thecore of aesthetics is to make the use of IT a visually appealing experience.Ourstudysuggeststhatwebsitedesignmustnotonly helpe-shopperscompletetheirshoppingprocess,butalsocreate an interactive, yet vivid, experience to hold onto existing customersand,at thesametime, toattractnewones. Froman e-shopper’spointofview,suchanexperiencemaydeterminehis/ herwillingnesstopayareturnvisitandthereforetomakeanew purchase.Thus,thesuccessofe-vendorsinthemarketbeginswith effectivewebsitedesign.
AppendixA. Previouswebsitedesigndimensionsidentifiedby researchers
Dimensionsof
websitedesign
Researchers Research
focus
Websitedesignfactors
Functionality factor
[5] Usability Convenience,Sitenavigation,
[16] Informationarchitecture,
Sitespeed,Sitedesign,
Searchability/accessibility,
Transactionefficiency
[10] Interactivity Onlinesupport,Customization,
[11] Customerservice/aftersales
Psychological factor
[5] Loyalty Sitedesign,Trust,
[6] Privacypolicies,Reputation
Socialpresence,Familiarity,
Word-of-mouth,Control,
Usefulness,Ease-of-use
Content factors
[6] Aesthetics Design,Presentationquality,
[15] Style/atmosphere,Design elements [23] [15] Marketing mix Communication,Product display, [20] Productinvolvement, Promotion,Price
AppendixB. Instrumentitems Coherence[16]
C-1.Allcomponentsofwebsitearewellrelatedtoeachother. C-2.Componentsofwebsiteworkwelltogether.
C-3.Eachcomponentofwebsiteseemstohangtogether. Legibility[16]
L-1.ItisclearwhereIcangointhewebsite. L-2.Itiseasytogetaroundthewholewebsite.
L-3.Itdoesnottakemuchtimetofigureoutawayofmoving aroundthewebsite.
Complexity[27]
CX-1.Thewebsitecontainsmanyvisualimages.
CX-2. The graphics and pictures in the websitefit withthe content.
CX-3.Thewebsiteusesdifferenttypesofvisualimages. Trust[37]
T-1.Thiswebsitegivesmeafeelingoftrust.
T-2.Thiswebsitegivesmeatrustworthyimpression. T-3.Ihavetrustinthiswebsite.
Satisfaction[37]
S-1.Iamsatisfiedwiththiswebsite.
S-2.ThiswebsiteofferswhatIexpectfromagoodwebsite. S-3.Thiswebsitegivesmeafeelingofsatisfaction. Willingnesstobuy[23]
WTB-1.Thelikelihoodofpurchasingonlineis:
WTB-2.TheprobabilitythatIwouldconsiderbuyingonlineis: WTB-3.Mywillingnesstobuytheproductonlineis:
AppendixC.Onlinetravelpackagepurchasescenario
Torefreshyourmemorywithpastonlinepurchases,weaskyou to navigate the following e-travel sites (http://www.eztravel. com.tw/,http://www.settour.com.tw/,andhttp://www.liontravel. com/)foratleast5min.Beforeyouareaskedtomakeapurchase, makeyourselffamiliarwiththecontents,searchengineandlayout ofthewebsites. Then,assumethat youwish totraveltoeither Vancouver,CanadaorLasVegas,U.S.inthenearfuture,butyouare not sure about the possible travel costs. To facilitate decision making, you wish to make a price comparison. Follow the guidelineswrittenbelowandusethefeaturesprovidedbyeach websitetosearchfortheinformationyouneedforthistrip,butdo notactuallymakeapurchase.
1.Takenomorethan5mintoselectthetravelinformationabout VancouverandLasVegas.Youmaycomparetravelplans,flight informationandspecialoffers.
2.Onceyou havemadeyour choice,clicktogatherinformation suchastraveldescriptionandcustomerreviews.
3.ClickHelptoreviewavailablecustomersupportservice. 4.Completecheckoutprocessbyenteringallinformationrequired
exceptcreditcardinformation. 5.Returntohomepage.
1.Reviewthe‘‘VisaInformation’’and‘‘TravelInsurance’’. 2.Usetheremainingtimetoevaluatethefeatures(e.g.content,
layout,color,hyperlinks,etc.)ofeachwebsiteandseeiftheyare conciseandreadable.
AppendixD. Multi-groupanalysis
Multi-groupanalysishasrecentlybeenfavoredbyresearchers asawayofvalidatingthegeneralizabilityofconstructsincluding
end-userperceptionsof websiteusability[16],andtechnology adoption model [8]. The significant results may indicate the invariance of factor loadings or structuralinvariance. Prior to evaluating the structural invariance, we at first conducted a comprehensiveassessmentofitem-factorloadings and model-datafitfor eachsubgroup.Thesignificantresults maysuggest thatdataforeachgroupfitthemodelandcanbeprocessedfor further statistical analysis. Then, we conducted a structural weights invariance test. The criterion for examining the invariancetestisbasedonthestatisticalmagnitudeassociated with the change in chi-squares/degrees of freedom (X2/d.f.).
In addition, we relied on several fit indices (CFI, TLI, and RMSEA)whichsurpassacceptablethresholdstoreflectadequate model-fit.
AsshowninTableC1andTableC2,theitem-factorloadings betweenmalesandfemaleswerestillhigh,rangingfrom0.70to 0.94.Fitindicesforbothmaleandfemalegroupsdemonstratean adequatemodel-fit.Basedon thesefindings,weproceededtoa multi-groupanalysis.
TableC1.
Construct Items Gender
Male(n=131) Female(n=169) Coherence C-1 0.72 0.75 C-2 0.83 0.81 C-3 0.79 0.80 Legibility L-1 0.81 0.70 L-2 0.82 0.83 L-3 0.71 0.82 Complexity CX-1 0.74 0.77 CX-2 0.82 0.81 CX-3 0.70 0.78 Trust T-1 0.84 0.80 T-2 0.90 0.87 T-3 0.80 0.84 Satisfaction S-1 0.71 0.77 S-2 0.94 0.87 S-3 0.89 0.91 Willingnesstobuy WTB-1 0.84 0.83 WTB-2 0.88 0.88 WTB-3 0.90 0.91 TableC2.
Fitindex Gender
Male Female X2 /d.f. 1.29 1.47 CFI 0.97 0.96 NFI 0.89 0.91 RMSEA 0.05 0.06
Twomodelswereusedinthemulti-groupanalysis,asshowninTableC3.Model
1istheequalpatternsmodelwhichservesasthebaselinemodel.Witha
chi-square/degree of freedom ratio of 1.36 and relative fit indices (CFI=0.97,
TLI=0.96, RMSEA=0.036), this baseline model demonstrates an adequate
model-fit.Model2isafactorloadingsinvariantmodel,whichwasmodeledby
forcingeachitem-factorloadingtobeequalacrossdifferentgroups.Model2
also demonstrates adequate model fit (chi-square/d.f.=1.36, CFI=0.97,
TLI=0.96, RMSEA=0.035). The non-significant change in the chi-squares
(6.91 with5d.f., p<0.2274)indicates that the model was invariantacross
subgroups.
TableC3.
No. Model x2 d.f. pValue CFI TLI RMSEA
Model1 Equalpatterns 371.243 273 0 0.97 0.96 0.036
Model2 Factorloadings
invariant
378.153 278 0 0.97 0.96 0.035
ThestandardizedstructuralweightswereshowninTableC4forbothmaleand
femalegroups.Theseweightswereestimatedwithitem-factorloadingsheldequal
acrossgroups.Thus,theywerethebestestimatesofthetruestructuralweights
[16]. TableC4. D.V. I.V. Gender Males Females Trust Coherence 0.28 0.37 Legibility 0.06 0.23 Complexity 0.46 0.30 Satisfaction Coherence 0.21 0.34 Legibility 0.13 0.24 Complexity 0.53 0.29
Willingnesstobuy Trust 0.30 0.27
Satisfaction 0.58 0.53
References
[1]E.P.Becerra,P.K.Korgaonkar,Effectsoftrustbeliefsonconsumers’online inten-tions,Eur.J.Mark.45(6),2011,pp.936–962.
[2]D.Belanche,L.V.Casalo,M.Guinaliu,Websiteusabilityconsumersatisfactionand theintentiontouseawebsite:themoderatingeffectofperceivedrisk,J.Retail. Consum.Serv.19,2012,pp.124–132.
[3]N.Carr,ITdoesn’tmatter,Harv.Bus.Rev.81(5),2003,pp.41–49.
[4]L.J.Cronbach,Testvalidation,in:R.L.Thorndike(Ed.),Educational Measure-ment,seconded.,AmericanCouncilonEducation,Washington,1971,pp.443– 507.
[5]D.Cyr,Modelingwebsitedesignacrosscultures:relationshipstotrust, satisfac-tion,ande-loyalty,J.Manage.Inf.Sys.24(4),2008,pp.47–72.
[6]D.Cyr,M.Head,A.Ivanov,Designaestheticsleadingtom-loyaltyinmobile commerce,Inf.Manage.43,2006,pp.950–963.
[7]D.Cyr,M.Head,H.Larios,B.Pan,Exploringhumanimagesinwebsitedesign:a multi-methodapproach,MISQ.33(3),2009,pp.539–566.
[8]X.Deng,W.J.Doll,A.R.Hendrickson,A.J.Scazzero,Amulti-groupanalysisof structuralinvariance:anillustrationusingthetechnologyacceptancemodel,Inf. Manage.42,2005,pp.745–759.
[9]D.Gefen,E.Karahanna,D.Straub,TrustTAMinonlineshopping:anintegrated model,MISQ.27(1),2003,pp.51–90.
[10]G.L.Geissler,Buildingcustomerrelationshipsonline:thewebsitedesigners’ perspective,J.Consum.Mark.18(6),2001,pp.488–502.
[11]V.Kaltcheva,B.A.Weitz,Whenshouldaretailercreateanexcitingstore envi-ronment?J.Mark.70,2006,pp.107–118.
[12]R.Kaplan,S.Kaplan,R.L.Ryan,WithPeopleinMind,IslandPress,WashingtonDC, 1998.
[13]F.P.Karimov,M.Brengman,L.V.Hove,Theeffectofwebsitedesigndimensionson initialtrust:asynthesisoftheempiricalliterature,J.Electron.Commer.Res.12 (4),2011,pp.272–297.
[14]R.B.Kline,PrinciplesandPracticeofStructuralEquationModeling,fourthed.,The GuilfordPress,NewYork,NY,2010.
[15]M.Koufaris,Applyingthetechnologyacceptancemodelandflowtheorytoonline consumerbehavior,Inf.Syst.Res.13(2),2002,pp.205–223.
[16]Y.Lee,K.A.Kozar,Designingusableonlinestores:alandscapepreference per-spective,Inf.Manage.46,2009,pp.31–41.
[17]Y.Lee,A.N.K.Chen,Usabilitydesignandpsychologicalownershipofavirtual world,J.Manage.Inf.Sys.28(3),2011,pp.269–307.
[18]Y.M.Li,Y.S.Yeh,Servicequality’simpactonmobilesatisfactionandintentionto use3Gservice,in:Proceedingsofthe39thHawaiiInternationalConferenceon SystemSciences,BigIslands,HI,2009,pp.1–10.
[19]Z.Liao,M.T.Cheung, Internet-basede-shoppingandconsumerattitudes:an empiricalstudy,Inf.Manage.38(5),2001,pp.299–306.
[20]H.H.Lin,Y.S.Wang,Anexaminationofthedeterminantsofcustomerloyaltyin mobilecommercecontexts,Inf.Manage.43(3),2006,pp.271–282.
[21]P.B.Lowry,D.W.Wilson,W.L.Haig,Apictureisworthathousandwords: sourcecredibilitytheoryappliedtologoandwebsitedesignforheightened credibility and consumer trust, Int. J. Hum.—Comput. Interac. 30, 2014, pp.63–93.
[22]J.Luo,S.Ba,H.Zhang,Theeffectivenessofonlineshoppingcharacteristicsand well-designedwebsitesonsatisfaction,MISQ.36(4),2012,pp.1131–1144.
[23]D.H.McKnight,V.Choudhury,C.Kacmar,Theimpactofinitialconsumertruston intentionstotransactwithawebsite:atrustbuildingmodel,J.Strateg.Inf.Syst. 11,2002,pp.297–323.
[24]S. Muylle, R. Moenaert, R.M. Despontin, The conceptualization and empiricalvalidationofwebsiteusersatisfaction,Inf.Manage.41(5),2004, pp.543–560.
[25]J.C.Nunnally,I.H.Bernstein,PsychometricTheory, McGraw-Hill,NewYork,NY, 1994.
[26]RDEC(ResearchDevelopment EvaluationCommission), ReportforDigital Divide,RDEC, 2010Athhttp://www.rdec.gov.tw/ct.asp?xItem=4024389&ctNode=12062&mp=100i (accessedonMarch31,2010).
[27]D.E.Rosen,E.Purinton,Websitedesign:viewingthewebasacognitivelandscape, J.Bus.Res.57,2004,pp.787–794.
[28]P.V.Schaik,J.Ling,Theroleofcontextinperceptionsoftheaestheticsofweb pagesovertime,Int.J.Hum.–Comput.Stud.67,2009,pp.79–89.
[30]E.Toufaily,L.Ricard,J.Perrien, Customerloyaltytoacommercialwebsite: descriptivemeta-analysisoftheempiricalliteratureandproposalofan integra-tivemodel,J.Bus.Res.66,2013,pp.1436–1447.
[31]D.Te’eni,Acognitive-affectivemodeloforganizationalcommunicationfor de-signingit,MISQ.25(2),2001,pp.251–312.
[32]N. Vila,I. Kuster,Consumerfeelingsand behaviourstowardswelldesigned websites,Inf.Manage.48,2011,pp.166–177.
[33]V.Venkatesh,F.D.Davis,Atheoreticalextensionofthetechnologyacceptance model:fourlongitudinalfieldstudies,Manage.Sci.46(2),2000,pp.186–204.
[34]R.Walczuch,H.Lundgren,Psychologicalantecedentsofinstitution-based con-sumertrustine-retailing,Inf.Manage.42,2004,pp.159–177.
[35]N.Wagner,K.Hassanein,M.Head,Understandingwebsitesatisfactionforolder adults,SpecialInterestGrouponHuman–ComputerInteraction(SIGHCI) Pro-ceedings,2010,pp.18–22.
[36]J. Webster,S.J.Ahuja,Enhancingthedesignofwebnavigationsystems:the influenceofuserdisorientationonengagementandperformance,MISQ.30 (3),2006,pp.661–678.
[37]K.D.Wulf,N.Schillewaert,S.Muylle,D.Rangarajan,Theroleofpleasureinweb sitesuccess,Inf.Manage.43,2006,pp.434–446.
[38]M.Zviran,C.Glezer,I.Avni,Usersatisfactionfromcommercialwebsites:the effectofdesignanduse,Inf.Manage.43,2006,pp.157–178.
Yung-ShaoYehreceivedhisBAinEconomicsatSimon
FraserUniversity inCanada andPhDinInformation
Management at National Chiao Tung University in
Taiwan.Hisresearchinterestscoveronlinecustomer
behavior and network economics. He is currently
workingforHTCandresponsibleforcustomer
segmen-tationintelligence.
Yung-Ming Li is a Professor at the Institute of
InformationManagement,NationalChiaoTung
Uni-versityinTaiwan.HereceivedhisPh.D.inInformation
Systems from the University of Washington. His
researchinterestsinclude networkscience,Internet
economics,andbusinessintelligence.Hisresearchhas
appeared inIEEE/ACM Transactions onNetworking,
INFORMS Journal on Computing, Decision Sciences,
European Journal of OperationalResearch, Decision
SupportSystems, InternationalJournal ofElectronic
