Researchbackground
Morethanthreedecadesago,QuelchandTakeuchi(1981)pre dictedthatvendingmachinewouldbecomeoneofthemostimportantno n- storemarketingchannels.Reality,theirpredictionhassurelymaterial ized.Mostofcountriesallovertheworldusedvendingmachineasreta ilbusiness.Therearemorethansevenmillionsvendingmachinesino p e r a t i o n inU.S(Leaner2002),Japanhasbecomeavending-machine- countryandVietnama l s o hasthefirstvendingmachinesin2000s.
Vendingmachineoffersconsumeravarietyofproductsincludi ngfood,snack,b e v e r a g e s , newspaper,cosmetic,CDs,cigarettesan dsoon.Lee(2013)arguedthatasmorep e o p le hasjoinedworkforcedu ringthepastseveraldecadesandtheirbusysociallifeincreasinglyplace smorevalueontimeandconvenience,vendingmachinehadbecome anindispensablepartofmanypeopledaily’slives.
InVietnam,vendingmachinewasdeployedinearly2000swh entheVietnam’sG o v e r n m e n t issuedthecoinsandexpertsbelie vethattheywillbepotentialmarketinthef utur e(KieuGiang,2004).H owever,vendingmachinesectordevelopedslowlyduringpasttime.Th ereweremanyinvestorsfailedinapplyingVendingmachineforretai lbusiness.
Thefirstmainreasoncanexplainforthisistechnology.Mostofvendin gmachineatthattimeintegratedused- cointechnology,butVietnampeopledonothavehabitkeepingthec o i n availableordonotliketousecoinintransaction.Thisleadsused- coinvendingmachinetobeinconvenientforcustomertoreachandpur chasebyvendingmachine;secondly,retailersusingvendingmachines cannotfindthesuitablelocationstoinstallv e n d i n g machine,especi allytwobiggestcitiesinVietnam:HoChiMinhCityandHanoiCityfor sidewalkandpublicareasarequitenarrow.Asaresult,vendingmachin
2 ecannoto p e r a t e atthebe st;finally,mostofthecomp aniesinvestedvendingmac hinesectoratthattime,wer esmallandnotenoughabili tytoinvestnewmodernven dingmachines,whilethec ompanieshavingenoughp otentialdidnotwanttojointi ntothisindustrytopushthev e n d i n g machinesectori nVietnam(HongPhuc,20
Fortunately,thebusinessenvironmenthaschangedinrecentyearsleadingtomanyg o o d signs fromvendingmachinesector.Firstofall,technologyisimproved,currentv e n d i n g machinecanacce ptboththecashandcoin,evenconsumercanusecreditcardtop u r c h a s e someproductsoldbyvending machine.Thismakesvendingmachinetobemorec o n v e n i e n t forcustomer.Ontheotherhand,so mebigbrandnamesinVietnam:Pepsi,CocaC o l a , begintousevendingmachinetoselltheirproducta tpublicareassuchashospital,p a r k , railwaystation,busstationandairport.Allabovethingshavepromp tedvendingmachinesectorgrowquickerandquicker.
ResearchProblem
Giventheeverincreasingpresenceandthepervasivenessofvendingmachinesinlivesofconsu mers,itissurprisingthattherehavebeentodatenostudiesinmarketingandconsumerbehaviorliteratur ethatshedlightonconsumers’usagebehaviorwithvendingmachinesservicesinVietnam.Furthermore, mostresearchhasfocusedonpurchaseintention,whichiscustomeracquisitionoriented,littleresearc hhasexaminedconsumers’p a t r o n a g e behavior,whichiscustomerretentionoriented.Howeve r,customerretentionismoreimportantthancustomeracquisition,andbuildingcustomervalueisthekeyt oretaining customers(Weinstein,2002).
Customer value has emerged as a vital tool for assessing service quality, customer satisfaction, and consumer behavior (Yieh, Chen, and Wei, 2012) It not only serves as a precursor to attitudes, intentions, and behaviors (Gounaris et al., 2007; Parasuraman and Grewal, 2000; Yang and Peterson, 2004) but also plays a crucial role in influencing customer retention and purchase intentions (Chang and Wildt, 1994) Additionally, customer value acts as a strategic asset for attracting and retaining customers, becoming one of the most significant factors in the success of service providers (Gale, 1994; Parasuraman, 1997) While many studies have explored the relationship between customer value and consumer purchasing intentions or satisfaction, there is a lack of research on the impact of customer value on patronage behavior.
Thus,inordertofulfillthisgap,aninvestigationofCustomerValueandPatronageb e h a v i o r inretailingvendingmachinesectorisexamined,andinanefforttoindentifythek e y factorsimpacti ngoncustomervalueinvendingmachinesectorinVietnam,thisstudyconsiderstheconceptsofFunc tionality,Customization,andPerceivedRiskasantecedentso f Customervalue.
ResearchObjective
Toexaminecustomervalueinvendingmachinesituation,itisnecessarytofocusonidentifying keyfactorsaffectedcustomervalueinusingvendingmachine.Thesefactorsarep e r c e i v e d risk,f unctionalityandcustomization.Eachofthesecaneitherpositivelyorn e g a t i v e l y influencecustomer valueinvendingmachinesector.Knowledgeabouttheeffecto f customervalueonconsumers’vending machinepatronagebehaviorprovideshelpfulimplicationstomarketersinexecutingmarketingandb usinessstrategyandcustomerr e l a t i o n s h i p management.Therefore,theoverallpurposeofth iscurrentstudyistoexamineh o w customervalueaffectsconsumers’retailingvendingmachinepatrona geinVietnam.
SignificanceoftheStudy
Researchmethodologyandscope
TheresearchconductedonconsumerinHoChiMinhCityisoneofthebiggestcitiesinVietnam, whereitcentralizesalltradingactivities,peoplecanreachnewtechnologiesinb u s i n e s s activitiessoo nest,andtheauthorcancollectdatafrompeoplefromdifferentp r o v i n c e s andsociallevelsinourcount ry.
Themainpurposeofthisresearchisperceivedcustomervalueofconsumerinusingv e n d i n g machine.Therefore,it’snecessarytoonlyfocusontherespondentshadusedthe vendingmachineforpurchasingprior.Therearethreekindsofvendingmachinewhichr e s p o n d e n t s canuseinHoChiMinhareVendingmassagechair,vendingsoftdrinkmachine,andvendingexchange coinmachine.Thus,onlyrespondentshadusedoneofthreeskindsofvendingmachinementionedabo ve,arechosentoconductthequestionnairesf o r mainsurvey.
TheAuthorappliesdataanalysistoolincludingMicrosoftExcelandSPSS,tohandlethecollect eddataandtoconducttheresearch.TheAuthoralsousesAmos22softwaretor u n ConfirmationFacto rAnalysis(CFA)andStructuralEquationModeling(SEM),andtoc h e c k relationshipsbetweenexoge nousvariablesandendogenousvariablesintheproposedr e s e a r ch model.
Thestructureofthestudy
Inthispart,researchbackground,researchproblem,researchobjective,significantofstudy,resea rchscopeandmethodology,andstructureofstudywillbepresentedsubsequently.
Inthischapter,thedefinitionsofeachconceptandanextensiveliteraturereviewwillbepresented Thenhypothesesarealsodevelopedthroughoutthepreviousliteraturesandfinally,theresearchmodelw illbeproposed.
Thischaptermentionsabouttheresearchdesign,measurementofconstructs,illustratetheproc essofconductingtheresearch,andthen,dataanalysismethodisalsopresentedtolaythefoundationof chapter4.
ThischapterincludesSamplecharacteristic,descriptivestatistic,CFA,SEMandd iscussi on o fresult.
Inthischapter,authorpresentedanextensiveliteraturereview,proposedhypothesisa n d resea rchmodel.Anliteraturereviewfocusedonknowledgeofvendingmachine,theoryr e l a t e d tocusto mervalue,retailpatronageandrelationshipofconstructsbementionedbyo t h e r s researches.Then,h ypotheseswerealsoarguedanddevelopedparticularlybyp r e v i o u s studiestobuildaresearchmodeli nthischapter.
Vendingmachine
AccordingtoMeuteretal(2000),Self- serviceTechnology(SST)wastheinnovativec h a n n e l ofmarketplacetransactionsthatwascomplet elydifferentfromtraditionalchannel.Particularly,interpersonalcontactwasnotrequiredbetweencustom erandserviceproviders,customerhavetoperformentireservicebythemselveswithoutdirectsupportofe mployees( B i t n e r etal.,2002).ThetypesofSSTsdeployedbyserviceprovidersincludeself- serviceg a s stations,launderettes,andvendingmachines(Bateson,1985);self- healthdiagnosis,A T M s , OnlineBankingandbuffetrestaurant(Meuteretal.,2000);andinthisstudy, authoronlyfocusononeoftypesofSSTsisvendingmachine.
Vendingmachineisamachinewhichdispensesitemssuchassnacks,beverages,a l c o h o l , cig arettes,consumerproducts,afterthecustomerinsertthecurrencyorcreditcardintothemachine.Formar keters,vendingmachinesupportedtoincreasethereachandintensityoftheirretaildistributornetwor k.Theyusuallywereplacedoutdoorsandinu n a t t e n d e d environmentssuchaslargeretailstores,ho spital,gasolinestation,offices,r a i l w a y station,airport,shoppingmalls(Lee,2003).Forcustomer,vend ingmachineso f f e r e d theconvenience,timesavingbenefitand24- hoursavailabilityandreasonablyfresha n d readytoserveproducts(QuelchandTekeuchi,1981).
Vending machines, a type of self-service technology (SST), exhibit characteristics identified in past research Globerson and Maggard (1991) proposed a self-service model predicting customer acceptance based on factors such as convenience, time-saving, self-control, money saved, self-image, perceived risk, and self-fulfillment Meuter et al (2000) highlighted ease of use, avoidance of service personnel interference, time-saving, convenience, and financial savings as key factors influencing SST adoption, with ease of use being the most critical The service quality of SSTs encompasses seven dimensions: functionality, convenience, enjoyment, privacy, assurance, design, and customization, where functionality includes responsiveness, reliability, and ease of use, while customization refers to the extent an SST can be tailored to individual customer preferences and transaction histories (Lin and Hsieh, 2011) Additionally, Ho and Ko (2008) demonstrated that SST characteristics positively impact customer value Thus, this study investigates perceived risk, functionality, and customization to assess customer value in the vending machine sector.
Customervalue
Theconceptofvaluehasbeenappliedinvariousfieldsofstudy,suchaseconomics,socialscienc e,accounting,finance,strategy,productmanagement,informationsystem,andmarketing(Huberetal ,2001;UlagaandChacour,2001).Customersacknowledgedservicev a l u e throughdesiredpurposeo rgoalachieved(Overby,2005).Thus,in- depthu n d e r s t a n d i n g ofcustomervalueconstructwashighlyimportant.
Theconceptofcustomervaluewasoftenperceivedasanambiguousandm u l ti f a ce t ed word (Parasuraman,1997).Manyauthorshadacknowledgedthedifficultiesinvolvedindefiningcustomerv alue(e.g.PiercyandMorgan,1997;Woodruff,1997).
Numerous definitions of customer value have been proposed by various authors Holbrook (1994) described customer value as “an interactive relativistic preference experience,” while Woodruff (1997) expanded this definition to include a customer’s perceived preference and evaluation of product attributes, performance, and the consequences of use that aid or hinder the achievement of their goals Additionally, Chen and Dubinsky (2003) defined customer value as “a consumer’s perception of the net benefits gained in exchange for the cost incurred in obtaining the desired benefits.” A consensus in the literature highlights that customer value is shaped by customers’ perceptions rather than suppliers’ assumptions or intentions (Anderson and Narus).
1998;WoodruffandGardial,1996;Zeithaml,1988);asDoyle(1989,p.83)putit:valueis“ n o t whatthe producerputsin,butwhatthecustomersgetout”.Basedonthose,inthisstudy,authorusedthecommond efinitionofcustomervalueproposedbyHannyandFelix( 2 0 0 9 ) , isas“tradeoffbetweentotalpercei vedbenefitandtotalperceivedscarify”(p.480).
Shethetal(1991)arguedthatcustomervaluewasdeterminedbyfivefactors:fu nction al value ,epistemicvalue,socialvalue,emotionalvalueandconditionalvalue.Functionalperformance,econo micutility,andbenefitassociatedwithpossessingtheservicee x p r e s s functionalvalue.Itisaconcepto feconomicbenefitsofthetrade- offbetweenqualityandpriceandispresumedtobetheprimarydriverofcustomerchoice(Shethetal.,1
9 9 1 ; SweeneyandSoutal,2001).Customer’sfeelings,forinstance,loveorhate,c o m f o r t a b l e orunc omfortable,happyorsad,etcwhentheyexperiencedanorganization’sp r o d u c t orservicerepresentem otionalvalue.Epistemicvalueisthecapacityofservicesorp r o d u c t s toprovidenoveltyorcuriosityan dsatisfyadesireforknowledge(Shethetal.,1 9 9 1 ) Socialvalueislimitedtospecificsocialgroups(e.g. ,cultural- ethnicgroups).Finally,conditionalvaluerelatetospecialcases,suchasanillnessorsomespecificsocial situation( S h e t h etal.,1991;SweeneyandSoutal,2001).
CustomizationandFunctionality
Customization is defined as tailoring products to meet the unique needs of individual customers, as highlighted by Bennett (1988) and Webster (1973) Marketers increasingly focus on customization to satisfy diverse customer segments Terpstra (1981) emphasized the significance of product customization, noting that the concept of "appropriateness" is essential Given the variety of cultures and traditions worldwide, understanding specific customer demands is crucial for marketers to design products and services that align with unique cultural preferences.
Inmarketingcontext,thefunctionalitywasdefinedas“whatproductdoesforthebuyersanduse rs;utilityitofferstheuser;whatheorshecandowithit”(Yellowpencil,2 0 0 6 ) Functionalityreferredt otheabilityofproductorservicetofacilitateperformancea n d theaccomplishmentofservicecusto mergoal.Italsowasconsideredtobeparticularimportancebecausephysicalsettingsofvendingmachi newerepurposefulenvironmentthate x i s t s tofulfillspecificneedsofcustomers.
Ontheotherhand,functionalityandcustomizationaretwoofdimensionsofservicequalitycon struct(Lin&Hsieh,2011).Service,byprovidingadditionalvalueforcustomers,leadstoincreasedcusto mersatisfaction,thoughtherelationshipbetweenserviceands a t i s f a c t i o n canbenon- linear(AndersonandMittal,2000).Inaddition,servicequalityisthoughttoleadtoincreasesinfirmprof itabilityandmarketvaluethroughtheservice– profitc h a i n (Heskettetal.1994).That’sthemainreasonwhyservicequalityhaslongbeenr e c o g n i z e d asakeyfactorforcreatingcustomervalue.Basedonliteraturesabove,thef o l l o w i n g hypotheses wereproposed:
Perceivedrisk
Riskwasdefinedastermofuncertainlyandconsequencesassociatedwithconsumer’saction(Ba uer,1960).Baseonthis,manystudieshadadoptedthisdefinitiontod e f i n e theconstructofperceivedris k,suchasCunningham(1967)arguedthat“perceivedriskasaconsumer’sperceptionoftheuncertaintya ndadverseconsequencesassociatedwithbuyingaproductorservice”(p.91)orDowlingandStaelin(199 4)understoodthat“ p e r c e i v e d riskisdeterminedbythedistinctionbetweeninherentriskandhandler isk”(p 1 21 ) Ingeneral,perceivedriskreferredthatconsumerpurchasebehaviormaycausesomeu n e x p e c t e d resultsandtheseresultsmaycauseunpleasantexperience(Bauer,1960).
Imagingthat,customersalwayshadtheirpurchasegoaleachtimewhentheywantedtobuysomething( productorservice).However,insomecases,customersdidnotknowwhichp u r c h a s e decisionwasth ebestchoicecouldmeettheirgoalatthebest.Asaresult,itwouldc a u s e unfavorableresultsandshapeari skconsciousness(Cox,1967).
Perceivedriskwasamulti- dimensionconstructincludinglossesandriskfactors,w h i c h together,explainoverallriskrelatedtopur chaseoruseaproductorservice.Theid en ti fi ed dimensionswerefinancial,function,psychological,so cial,time,andphysical(Roselius,1971).Inthisstudy,fourdimensionschosentoexplaintheriskinusing vendingmachinewerepsychological,social,privacy,andtimewasting.
Customervalueandretailpatronage
Retailpatronagebehaviorwasstudiedinmanyempiricalresearches(Lee,H.Y.,F ai r h u r s t , A. E.,&Lee,M.Y.,2009).Leeetal,
(2009)alsoarguedthatretailpatronagebehavior canexplainthemechanismofstorechoice.Asaresul t,ithasbeenacriticalissueforacademiciansandretailmanagers.However,becauseofthedynamicnatur eofpatronageb e h a v i o r , acomprehensivepictureofretailpatronagebehaviorisacomplicatedmodel. Laaksomen(1993)definespatronageas“allthepossibleinnerfeaturesofdynamismaroundtheshoppin gbehaviorphenomenonintermsofstorechoice”(p.9).Thus,retailpatronagew a s focusedontheiden tificationofrelevantattributesbyonekeyresearchstream
Understanding consumer patronage motives is essential for reinforcing the key determinants of buying behavior, as outlined by Hartley (1980) He defined these motives as the reasons customers choose to shop at various stores, highlighting several patronage factors These factors include convenience related to location and store hours, the assortment and variety of merchandise, the quality and fashion level of goods, competitive pricing, and the level of service provided, such as credit options, delivery, and knowledgeable sales staff Additionally, excitement generated through promotional efforts and the use of celebrities also plays a significant role in influencing customer choices.
Morgenstein and Strongin (1992) define patronage motives as the reasons customers choose one shopping location over another, which include brand preference, appealing facilities, personal service, convenience, product value, attentive sales staff, and a positive store image Additionally, Pan and Zinkhan (2006) categorized various determinants of retail patronage into three main factors: product-relevant factors, market-relevant factors, and personal factors Product-relevant factors focus on the product's functionality and attributes, such as quality, assortment, and price Market-relevant factors pertain to the services provided by the store, including convenience, service quality, store image, atmosphere, and sales personnel Lastly, personal factors encompass consumer characteristics like age, gender, and income.
AccordingtoLeeetal(2009),threegroupsofdeterminantsincludingproductquality,serviceq uality,andassortmenthadthestrongestinfluencetoconsumers’decisiontopatronizeaparticularstore Thisfindingpresentedclearlytheimportanceofservice qualityasakeytoachieveretailpatronage.Furthermore,Yieh&Wei(2012)alsostatedthatcustomerva luewasacrucialinstrumentforanalyzingservicequality.Therefore,thef o l l o w i n g hypothesisissugg ested:
Researchmodel
Theoreticalbackgroundanddefinitionsofeachconceptintheresearchmodelispr esen ted int hischapter.Basedontheliteraturereview,therelationshipsofconceptsareargued,thus,functionality,Customizationandperceivedriskareantecedentsimpactingcustomervalue,andCustomervalueinfluen cetoretailpatronage.Therearefourhypothesesinresearchmodel.
Thischaptermentionsaboutthewaytoconducttheresearchtoachievetheresearcho b j e c t i v e s Afterward,researchdesign,measurementsofconstructs,researchprocess,samplesize,datacollect ionanddataanalysiswillbepresentedmoredetail.
Researchprocess
Qualitativeresearch
Afterauthorproposedtheresearchmodelbasedonliteraturesreviewabove,themeasur emen t scalesofeachconstructalsowasbuildbyre- usedoradaptingwithpreviousr esear ch Thus,apreliminaryquestionnaireofstudywasdevelopedtoin vestigater e s p o n d en t s inHoChiMinhCity.Afterthepreliminaryquestionnairehadbeentranslatedin toVietnamese,theresearcherintervieweddeeplywith8peoplewhousedthevendingmachinepriort oanswerthequestionnairetoobtaincorrectitemsinthecontextofVietnameseconsumers,checktheme aningofwordsinoriginalmeasurementscalesandm o d i f y themtobemoresuitableandeasiertounders tand.ThisstepisveryimportantduetotherearemanydifferencesbetweenmeaninginEnglishandVie tnamese,aswellasEnglishleveloftranslator,mademisunderstandingornotconveyenoughthemea ningoftheoriginalscales.Afterfinishthisstep,mostofmeasurementscaleswouldberefinedtobeeas ytounderstandandtobemeaningfulincontextofthestudy.
Quantitativeresearch
Afterqualitativeresearch,arefinedquestionnairewascompletedtobemoreappropriatewithVi etnammarket,especiallyHoChiMinhCitymarket.ConsumersinHoChiMinhCitywouldbechosento interviewformainsurveyinthisstudyforusingc o n v e n i e n t sample.Theprocessofquantitativer esearchwasconductedbythefollowingsteps:
(2010)thesamplesizeshouldbe100orgreaterandm i n i m u m sampleequalfiveobservationsforeachi tem.Theproposedresearchmodelhasf i v e factorswith48scales.Therefore,theminimumsamplesize shouldbe:48*5$0o bserv ati on s.
Ontheotherhand,totestthetheoreticalandmodelandhypothesesbyStructureEquationMo del(SEM)method,theminimumrequiredsampleshouldbe200observations.
Thecurrentstudyusedthesamplesizeat247observations,soit’sappropriatedforC F A andSE Manalysiscomparetomentionedgeneralrulesabove.Afterthat,authorusedconveniencesamplin gmethodtoconductsampling.Allrespondentswereaskedwhethertheyhaveeverusedvendingmachine beforeansweringthequestionnaire.
Thequestionnaireissenttointervieweesastwoways.First,therespondentswouldb e receivedh ardcopiestoanswerthequestionsbyauthoratUniversities,CompaniesandO r g a n i z a t i o n s inHCM City.Second,tobemoreconvenientforrespondents,theq u e st i o n n a i r e s werealsobroadcastedviai nternetbyGoogleDocs.Thus,asurveylinkwassenttorespondentsviaemailaddress,Facebookpagea ndYahoomessenger.Toanswertheq u e s t i o n n a i r e , respondentsclickonthelink,typetheiransw erandchoosetheanswertheya g r e e ordisagreeasthe5- pointsLinkerstypeandsubmitthelinktoresearcher.Tomakesurethat,allofrespondentscanundersta ndclearlyaboutvendingmachineatbeginningofinterviews,ashortdescriptionofvendingmachinew aspresentedforreferencepurpose.
Over the course of one month, a total of 334 responses were collected from a sample of 250 hard-copy questionnaires and 200 emails sent to various respondents The hard-copy questionnaires were distributed among the author's colleagues, friends in Ho Chi Minh City, students at the University of Economics Ho Chi Minh City, and customers who had recently used vending machines at locations such as trade centers, hospitals, game zones, and cinemas The author received 176 completed hard-copy questionnaires Additionally, 158 online feedback responses were gathered from partners, current customers, classmates from the Master of Business Administration program at ISB School, and friends via Facebook and Yahoo Messenger Ultimately, the total responses included 176 hard-copy answers and 158 online feedbacks.
Aftercheckingtheanswer,theerroranswerswhichweremissedanswersoranswerwith onlyon evaluesuchas“1”,“2”,“3”,“4”and“5”,orwerecheckedaszigzagline,werer em o v e d outoffinaldata toanalyze.Besides,authoralsoeliminatedtheanswerswhichr e s p o n d en t s choseanswerwas“no”w hentheywereasked:“HaveyoueverusedVendingmachine?”Asaresult,87responsesoftotalof334o neswereeliminated;totalincluded247answersinfinal.Thefinalsampleissuitabletotherequirement ofminimumsamplesize:2 4 0 observations.Thefollowingtablesummarizedcollecteddatafromthesur vey.
The survey results indicate that the percentage of returned hardcopy questionnaires exceeds that of the online channel, with a larger sample size for hardcopy responses This discrepancy stems from the sampling method employed by the author For hardcopy questionnaires, respondents were specifically asked if they had used a vending machine, and only those who answered "yes" received the questionnaire, ensuring a higher quality of responses In contrast, online respondents could prematurely terminate the survey by claiming they had never used a vending machine, despite possibly having done so Additionally, there was a tendency for online participants to provide random answers, often selecting the same option for multiple questions, which led to the exclusion of these responses from data analysis.
DatafromhardcopyquestionnairesweretypedandtheanswersweresubmittedviaG o o g l eDocstoolwerealsocopiedandpastedinExcelFiletogether.Afterthat,datawererev i e wed forcompl etion,codedandinputrawdatainIBMSPSSStatisticVersion16withthescalesasmentionedinmeasu rementscalespart.
Researchdesign
Thisstudyaimedtoinvestigatehowcustomervalueinfluencesoncustomer’sretailingpatronage behaviorandtheantecedents(customization,functionalityandperceivedr i s k ) ofcustomervalueinven dingmachinesectorinVietnam.Therefore,aresearchmodelw a s proposedasFigure1.Thisresearchus edprimarydatathatcollectedfromparticipantsw h o wererequiredtohaveusedVendingmachineservic espriortocompletingthesurvey.
Particularly,datawascollectedfromcustomerswhousedVendingmassagechairs,vendingc o i n exch angemachineandsoft- drinkvendingmachinesatTradecenter,Supermarket,Mall,Hospital,Cineplex,andGameZoneinHoChiMinhCity.
Measurementsofconstructs
Establishedscaleswereused,oradaptedforuse,wherepossibletomeasureeachofinvestigated constructs.Allitemsweremeasuredon5-pointLinkers- type.Wherenecessary,thesurveyquestionswereslightlyadaptedtoreflecttheindustriesinvestigate d.
Fourfirstorderconstructs(functionality,customization,perceivedriskandretailingv e n d i n g machinepatronage)andonesecondorderconstruct(customervalue)wereusedinthisstudy
CustomervalueofeachrespondentwasmeasuredbyLin&Huang(2012)andS w e e n y &Sou tar(2001).Customervaluewasasecondorderconstructcomprisingfivec o m p o n e n t s : functionalval ue,emotionalvalue,socialvalue,andepistemicvalue.
Functionalvaluewasmeasuredbytenitems,expressingfunctionalperformance,economicutility,an dbenefitsassociatedwithusingvendingmachine.Emotionalvaluewasmeasuredbyfiveitems,addressi ngdegreeofcustomer’sfeelingwhentheyusevendingmachinetop ur ch ase productsorservices.Wh ileepistemicvaluewasmeasuredbyfiveitemsassessingr e s pondents’levelofcuriosityanddesirefork nowledgewithvendingmachine.Finally,S o c i a l valueandconditionalvalueweremeasuredbyfour andfiveitemstoaskrespondentsa b o u t culture- ethnicgroupandvaluerelatedtospecialcasessuchaspromotion,discount,a n d environmentpollution.
MeasurementofFunctionalityandCustomizationofvendingmachinewerebasedonLinandH sieh(2011)scales.Functionalitywasmeasuredbyfiveitemstoassessr e s p o n d e n t s ’perceptionof responsiveness,reliability,andeaseofuse.Customizationmeasuredbyaskingrespondentsaboutdegre eofunderstandingspecificneeds,p e r s o n a l i z a t i o n ofvendingmachine.
Perceivedriskwasmeasuredbyusingeight- itemperceivedriskscaleadaptedfromRaf ale etal(2012),toassessfourdimensionschosentoexplaint heriskinusingvendingmachinewerepsychological,social,privacy,andtimewasting.
Retailingvendingmachinepatronageconstructwasmeasuredbyfiveitemsadaptedf r o m Hoi zerandStem(1985)scaleembodyingdegreeofrespondents’choicebetweenpu r ch asi n g byvending machineandotherretailchannels.
Construct Scaleitem Linker poi nt
Customization Thevendingmachineunderstandsmyspecificn e e d s Strongly disagree / strongly agree
VendingmachineserviceisonethatIwoulde n j o y Stronglydis agree/stronglyVendingmachineservicemakemewant touse
Beforeusingthevendingmachine,Iwouldobtainsub stantialinformationaboutthedifferentmakesandmo delsofvendingmachine Strongly disagree / strongly agree
Iwouldacquireagreatdealofinformationa b o u t th edifferentmakesandmodelsbeforeusi ng vendin gmachine.
Iwillpayslightly moreforproductsif Ican buythemthroughvendingmachine.(never/always )
Iwillincreasemyinterestinvendingmachinewhe nmoregoods/ servicesaremadeavailablethroughthem BecauseIammorefamiliarwithvendingmachine,I prefershoppingbyvending machinet h a n otherretailchannels.
Ishop byvending machineevenwhentheselection/varietyofgoodsis poor.
Usingvending machinecanworsentheimageotherpeoplehaveof Usingvending machinemakessomepeoplewhoseopinionIvalueth inkthatIamnotactingcorrectly
Dataanalysismethod
DescriptiveStatistics
AccordingtoLoetherandMcTavish,“DescriptiveStatisticsareusedtohelptheresearchers andconsumersofresearchreportsunderstandmoreabouttheresearchdata.Theyassistwithunderstandi nghowthedataaredistributedacrossthepossiblerangeofvalu e; withknowingwhetherornottheshape ofvariableisnormal;andwithunderstandwhetherone’ssubjectstendtoclumptogetherinonespoton thedistributionoriftheyarewid el y scatteredthroughoutthepossiblerangeofvalue”(ascitedinMarr y,L.M,2003,p 3 5 ) Awidevarietyofdescriptivestatisticsisusedinresearch(Marry,L.M,2003),bu tinthisstudyweusedmeasuresofshapetocheckdistributionofvariables.Thus,Marry,L.M,
( 2 0 0 3 ) alsoarguedthat“theshapeforvariablesthatisthemostusefulformostofdata analyzingtestingisnormaldistribution,oftencalledthebell- shapecurve”(p.111).Inthisstudy,totestnormalityoftheshapeofdistribution,twodescriptivestatistics:S kewnessandK u r t o s i s wereused.
Glass&Hopkin(1996)statedthat“skewmeasureswhetherthetwohalvesofthed ist ri bu ti on aresymmetrical”and“anormaldistributionispresentedbyavaluethatrangesfrom ±2.Valuesmuchlo werthan-
2orhigherthan+2,denoteaskewedratherthannormald i st r i b u t i on”(ascitedinMary,L.M,2003,p.1 11).
Similarly,Mary,L.M(2003)alsomentionedthat“adistributionshouldbenicelyande v e n l y r ounded_neithertoopeaklikeapencilpeakuponthegraph,nortooflat,likealowhill.AndtheKurtosisstati sticinSPSSmustbebetween±2”(p.112)
ConfirmatoryfactorsAnalysis(CFA)
(2006)statedthat“Confirmatoryfactoranalysis(CFA)isaconfirmatorytechnique– itistheorydriven.Therefore,planningoftheanalysisisdrivenbytheoreticalr e l a t i o n s h i p s a mongtheobservedandunobservedvariables.WhenaCFAisconducted,ahypothesized modelisus edtoestimateapopulationcovariancematrixthatiscomparedw i t h theobservedcovariancematrix”. (p.323).CFAisusedtotestreliabilityandvalidityofm easurementpriortotestingastructuremodel(Ja mesetal,2006).
AccordingtoHairetal(2010),thecompositereliabilityestimatestheextenttowhichasetoflatentconst ructindicatorsshareintheirmeasurementofaconstruct.Compositer e l i a b i lityshouldexceed0.7(asc itedinShu-HsunandYing-Kin,2008).
Averagevarianceextracted(AVE)isusedtotestconvergentvalidityofmeasurementscales.AVEistheamountofcommonvarianceamonglatentconstructindicators(Hairetal,2010).AVEshouldb egreaterthan0.4ifthecompositereliabilityofallscalesarehigherthan0.6(ascitedinChun-Cheetal,2013).
Basedontheresultofcorrelationsofconstructs,andChi- squaredifferenttest,d i s c r i m i n a n t validityofmeasurementscalesistest.Chi- squaredifferenttest“allowsr e s e a r c h e r s tocomparetwomodels,oneinwhichtheconstructarecorr elated,andoneinwhichtheyarenot.Whenthetestissignificanttheconstructspresentsdiscriminantv alidity”(ascitedinZaitandBertea2011,p.218).
StructureEquationModeling(SEM)
(2006)likedtothinkthat“SEMasCFAandmultipleregressionsbecauseSEMismoreconfirmatoryt echnique,butitalsoisusedforexplorationpurpose”( p 3 2 5 ) Todeterminehowwellaproposedmod elfitsthesampledataanddemonstrateswhich onethemostsuperiorfit,agroupoffitindicesincluding Chi-Squaretest,RMRandSRMR,CFIandRMSEA,isusedtoassessthedegreeofmodelfit.Inwhere,
“RMSEAtellsushowwellmodel,withunknownbutoptimallychosenparametersestimatesw ouldfitthepopulations’covariancematrix”(DiamantopoulosandSiguaw,2 0 0 0 , p.85)
“CFI_Comparativefixindex,assumesthatalllatentvariablesareuncorrelated( n u l l / i n d ep en de n t model)andcomparesthesamplecovariancematrixwiththisnullmodel”( H o o p e r at al,2008,p.55).AccordingtoHuandBentler(1999),valueforthisstatisticshouldbegreaterthan0.9inor dertoensurethatmisspecifiedmodelsarenotaccepted
Bootstrap
Toevaluatereliabilityoftheestimatesinquantitativeresearchbysamplingmethod,a u t h o r s dividedbigsampleintotwoparts.Firstoneisusedtocalculateparametersofther e s e a r c h model;othe risusedtore-assesstheresultabove.Anotherway,authorshavetore- samplingtotest.Twothesewaysrequiretospendmoretimeandexpenseoncollectingdataa n d arenot effectiveinpractice,especiallyinSEMmethodrequiredalargesample.Incaseo f this,bootstrapisthem ostsuitablemethodtoreplacefortwomethodsabove.Accordingto Hilmer(2001),
The goal of bootstrap methodology is to use sampled data to replicate the overall population distribution by resampling from the data to estimate a sampling distribution This process begins with an initial sample dataset of size n, from which random samples of the same size are drawn with replacement The parameter of interest is calculated for each drawn sample, and this process is repeated multiple times The bootstrap estimator is then determined by averaging the parameter estimates from these different bootstrap samples Essentially, the distribution of parameter estimates from the bootstrap samples mimics the traditional sampling distribution of parameter estimates from samples drawn from the entire population This technique enables researchers to generate an estimated sampling distribution even when they only have access to a single sample rather than the entire population.
This chapter outlines four key components: research design, measurement scales, research process, and data analysis methods It distinguishes between qualitative and quantitative research The qualitative phase involved in-depth interviews with eight respondents to assess the clarity and contextual relevance of Vietnamese translations of original English scales The quantitative phase followed a structured four-step process: developing the questionnaire, defining sample size, distributing the questionnaire to respondents, and compiling, checking, and coding the data for analysis in SPSS software, resulting in a final sample of 247 participants The chapter concludes with a discussion of the data analysis methods employed, with the next chapter dedicated to presenting the data analysis and results of the main survey.
Theresearchresultwasreportedinthischapter.Firstly,samplecharacteristicwasp r e s e n t e d Next,descriptivestatisticsincludingSkewnessandKurtosiswereconductedtoc h e c k distributio nvariables.Furthermore,reliabilityandvalidityofscaleswerealsoc h e c k e d byrunningConfirmato ryFactorAnalysis.Finally,SEMmodelwasusedtotestr e s e a r c h model.
Samplecharacteristic
The total data sample consisted of 39.9% male and 60.7% female valid subjects Among the respondents, those aged 26-35 represented 45.7%, while 18-25 accounted for 38.9% Individuals aged 36-45 made up 12.9%, and those aged 46-60 were only 2% Notably, only 0.5% of the valid sample was aged 10-17 In terms of education, 68.8% held college or university degrees, 21.5% had MBAs, 8.9% completed high school, and just 0.8% had education below high school Regarding occupation, 66% of respondents were office staff, 11.3% were managers or businessmen, 10.6% were retired, 10.1% were students, and the remainder worked in various other roles.
Finally,intermsofincome,rankingfromlessthanfivemillionstomorethan15millions,thelargestinc omegroupwasfromfivemillionstolessthan10millions,accountedfor4 7 % , thesecondwaslessth anfivemillions(occupied19%),thethirdwasfrom10millionstolessthan15million(occupied17.4
%),andmorethan15millionsaccountedfor16.6%.A l l respondentsconfirmedthattheyhadusedthe vendingmachinebeforetheyproceedtoa n s w e r thequestionnaire.
Descriptivestatistics
AsauthormentionedinChapter3,thenormaldistributionispresentedwhenSk ewn ess andKu rtosisstatisticmustbebetween±2.Therefore,theSkewnessandKurtosisv a l u e ofanyitemaremuchlow erthan-2orhigherthan+2,willbeeliminatedmeasurement scalesbeforerunningCFA.TheresultofSkewnessandKurtosisweresummarizedinbelowtable.
Deviation Skewness Kurtosis Statistic Statistic Statistic Statistic Statistic Statistic
0.942to0.751.Thatmeansallscaleshadvaluesbebetween±2.Asar e s u l t , therewerenotanyitemsbe deletedandallofthemwouldbeusedinestablishingthemainsurveytotestreliabilityandvaliditybyCFAmethod.
ConfirmatoryFactorAnalysis(CFA)
Theresearchmodelhadfourfirst- orderconstructsconsistedofFunctionality,Customization,PerceivedRiskandRetailingVendingMac hinePatronage,andonesecond- orderconstructwereCustomerValuehavingfivecomponents:FunctionalValue,EmotionalV a l u e , Epi stemicValue,SocialValueandConditionalValue.CFAhadbeencarriedoutto assessthemeasurementmodelsofeachconstruct.And,thenthefinalmeasurementmodelw o u l d be conductedtotestreliabilityandvalidity,beforestructureequationmodelingwasusedtotestthetheore ticalmodelandhypotheses.
TheresultindicatedthattheCFAmodelsofCustomization(P=0.000( Socvalue 807 epivalue < > Condvalue 564
Besides,tomakesuretheyweresignificantlessthanunity,theauthorconductedthetest,inwher e,nullhypothesis:HowasT(correlation)=1,H1:T≠1.TheresultoftestindicatedthatHowasrejected (SeetheTable5).Thatmeandiscriminantvalidityofthec o m p o n e n t s ofthesecond- orderconstructwasconfirmed.
Correlations Estimate SE CR P-value
Emovalue < > Condvalue 0.751 0.042185 5.90255 0.0000 epivalue < > Socvalue 0.807 0.037729 5.115436 0.0000 epivalue < > Condvalue 0.564 0.052757 8.264335 0.0000
Finally,totestcompositereliability,convergentanddiscriminantvalidityoftheconstructs,CFAofthefinalmeasurementmodelwasconductedasthefigure5
TheCFAresultofthefinalmeasurementmodelwaspresentedanacceptableleveloff i t (Chi- square38.027,P=0.000(0.4).Thesefindingindicatedthatallscalesmeasurementwereun idimensional,andconvergentvalidity wasachieved.Table6summarizedtheresultofmeasurementr eliabilityandv a l i d a t i o n
Thediscriminantvaliditymeasuresdifferencesbetweenconstructs.FornellandLarcker(1981) arguedthatitexitsifitemssharemorecommonvarianceswiththeirrespectiveconstructthanwithothe rconstructs.Thecorrelationsbetweenthecomponentsofe ach constructindicatedthattheywerelesstha n0.9(Seetable7),exceptforCorrelationb e t w e e n RepatroandCVas0.968.However,tomakesuret hesecorrelationsweresi g n if i c a n t lessthanunityandtheywerenotthesame,theauthorconductedth etest,inw h e r e , nullhypothesis:HowasT(correlation)=1,H1:T≠1.TheresultoftestindicatedthatH owasrejected(SeetheTable7).Thesefindingssupportedwithin-constructd i s c r i m i n a n t validity.
Correlations Estimate SE CR P-value funtio < > perrisk -0.271 0.061497 20.6677 0.0000 funtio < > custom 0.805 0.037903 5.14472 0.0000 funtio < > repatro 0.868 0.031724 4.160845 0.0000 funtio < > CV 0.872 0.031273 4.092937 0.0001 perrisk < > custom -0.09 0.063628 17.13072 0.0000 perrisk < > repatro -0.306 0.060823 21.47212 0.0000 perrisk < > CV -0.249 0.061875 20.18572 0.0000 custom < > repatro 0.795 0.038755 5.28966 0.0000 custom < > CV 0.865 0.032057 4.211245 0.0000 repatro < > CV 0.968 0.016033 1.995931 0.0470
To confirm the discriminant validity of the constructs Repatro and CV, the author employed the Chi-square difference test This method, as outlined by Segar (1997), enables researchers to compare two models: one where the constructs are correlated and another where they are not A significant test result indicates the presence of discriminant validity (Zaita and Bertea, 2011) Additionally, Zaita and Bertea (2011) recommend that the measurement model should be reflective and analyzed in pairs Consequently, the author introduced two models in Amos, setting the correlation to zero for the first model while allowing free correlation in the second model, as illustrated in the accompanying figure.
Basedonthat,theauthorcomparedtwoconstructs:RepatroandCVthatweresuspectedtohave problemswithitemsdiscriminatingamongthem.Asmentionedabove,t w o modelswouldbeanaly zedthroughCFAandeachmodelpresentedavalueofChi- squareanddegreesoffreedom(Df).TheresultoftestwouldbeshowinTable8.
Thedifferencetestresultwassignificant(p=0
