客户满意度研究中的PLS方法（英文）

State-of-artonPLSPathModelingthroughtheavailablesoftwarePreparedbyYvesMarieChatelin,IdEFranceVincenzoEspositoVinzi,DMSItalyMichelTenenhaus,HECFranceJune2002State-of-artonPLSPathModeling2TableofContentsI.TheECSImodel3II.ThePLSPathmodel51.Relationbetweenthemanifestvariablesandthelatentvariables(outermodel)52.Normalizationofthelatentvariables63.Relationbetweenthelatentvariables(innermodel)6III.PartialLeastSquares(PLS)EstimationoftheECSImodel71.Manifestvariablesstandardization72.Latentvariablesestimation82.1.ExternalestimationYjofthestandardizedlatentvariable(
j–mj)82.2.InternalestimationZjofthestandardizedlatentvariable(
j–mj)82.3.Estimationoftheweightswjh’s92.4.Estimationofstructuralequations10IV.ModelValidation101.BlindfoldingapproachinLVPLS102.ResamplinginPLS-Graph12V.MissingDataTreatment14VI.UseofLVPLS1.8andPLS-Graph3.0ThecaseoftheECSImodel(withmissingdata)foramobilephoneprovider151.Optionsselection152.TheECSImodelforaMobilePhoneProvider153.TheresultsofResamplingProceduresonCustomerSatisfaction174.TheresultsofResamplingProceduresonPathCoefficientsofthemodel19REFERENCES19ANNEX1TheprogramcodefortheLohmöllerLVPLS1.8program(whensomedataaremissing)21ANNEX2TheresultsoftheLohmöllerLVPLS1.8program(ECSImodelwithmissingdata)22TheresultswithBlindfoldingonCustomerSatisfaction(RedundancyandCommunalitycriterion,30groups)23ModelValidation25ANNEX3ResamplinginPLS-Graph(ECSIModel:LatentVariableCustomerSatisfaction)26ANNEX4Glossary29State-of-artonPLSPathModeling3State-of-artonPLSPathModelingthroughtheavailablesoftwareYves-MarieChatelin,VincenzoEspositoVinzi&MichelTenenhausAbstractThepurposeofthispaperistopresentPLSPathModeling,todescribethevariousoptionsofLVPLS1.8andPLS-Graph3.0forcarryingoutapathmodel,andtocommenttheoutputofbothsoftware.PLS-Graph3.0isactuallybasedonLVPLS1.8.Asanaddedvalue,PLS-Graphhasaveryfriendlygraphicalinterfacefordrawingthemodelandaresamplingmodule(jackknifeandbootstrap).ThepresentationisillustratedbydatawhichhavebeenusedtoconstructtheEuropeanConsumerSatisfactionIndex(ECSI)foramobilephoneprovider.I.TheECSImodelTheEuropeanConsumerSatisfactionIndex(ECSI)isaneconomicindicatorthatmeasurescustomersatisfaction.AmodelhasbeenderivedspecificallyfortheECSI.Inthismodel,seveninterrelatedlatentvariablesareintroduced.Itisbasedonwell-establishedtheoriesandapproachesincustomerbehavioranditistobeapplicableforanumberofdifferentindustries.TheECSImodelisdescribedinfigure1.Asetofmanifestvariablesisassociatedwitheachofthelatentvariables.ThisstructureiscalledtheECSImodel.Theentiremodelisimportantfordeterminingthemaingoalvariable,beingCSI..ImagePerceivedvalueCustomerExpectationPerceivedqualityLoyaltyCustomersatisfactionComplaintsFigure1:CausalitymodeldescribingcausesandconsequencesofCustomerSatisfactionIntable1themanifestvariablesVjhdescribingthelatentvariables
jaregivenfortheMobilePhoneIndustry.State-of-artonPLSPathModeling4Table1:MeasurementInstrumentfortheMobilePhoneIndustryAlltheitemsarescaledfrom1to10.Scale1expressesaverynegativepointofviewontheproductandscale10averypositiveopinion.LatentvariablesManifestvariablesImage(
1)a)Itcanbetrustedinwhatitsaysanddoesb)Itisstableandfirmlyestablishedc)Ithasasocialcontributionforthesocietyd)Itisconcernedwithcustomerse)ItisinnovativeandforwardlookingCustomerExpectationsoftheoverallquality(
2)a)Expectationsfortheoverallqualityof“yourmobilephoneprovider”atthemomentyoubecamecustomerofthisproviderb)Expectationsfor“yourmobilephoneprovider”toprovideproductsandservicestomeetyourpersonalneedc)Howoftendidyouexpectthatthingscouldgowrongat“yourmobilephoneprovider”PerceivedQuality(
3)a)Overallperceivedqualityb)Technicalqualityofthenetworkc)Customerserviceandpersonaladviceofferedd)Qualityoftheservicesyouusee)Rangeofservicesandproductsofferedf)Reliabilityandaccuracyoftheproductsandservicesprovidedg)ClarityandtransparencyofinformationprovidedPerceivedValue(
4)a)Giventhequalityoftheproductsandservicesofferedby“yourmobilephoneprovider”howwouldyouratethefeesandpricesthatyoupayforthem?b)Giventhefeesandpricesthatyoupayfor“yourmobilephoneprovider”howwouldyouratethequalityoftheproductsandservicesofferedby“yourmobilephoneprovider”?CustomerSatisfaction(
5)a)Overallsatisfactionb)Fulfillmentofexpectationsc)Howwelldoyouthink“yourmobilephoneprovider”compareswithyouridealmobilephoneprovider?CustomerComplaints(
6)a)Youcomplainedabout“yourmobilephoneprovider”lastyear.Howwell,orpoorly,wasyourmostrecentcomplainthandledorb)Youdidnotcomplainabout“yourmobilephoneprovider”lastyear.Imagineyouhavetocomplainto“yourmobilephoneprovider”becauseofabadqualityofserviceorproduct.Towhatextentdoyouthinkthat“yourmobilephoneprovider”willcareaboutyourcomplaint?CustomerLoyalty(
7)a)Ifyouwouldneedtochooseanewmobilephoneproviderhowlikelyisitthatyouwouldchoose“yourprovider”again?b)Letusnowsupposethatothermobilephoneprovidersdecidetolowertheirfeesandprices,but“yourmobilephoneprovider”staysatthesamelevelastoday.Atwhichlevelofdifference(in%)wouldyouchooseanothermobilephoneprovider?c)Ifafriendorcolleagueasksyouforadvice,howlikelyisitthatyouwoul