罗胜强管理学研究方法-07结构方程模型

StructuralEquationModelingStructuralEquationModeling1CorrelationandRegressionCovarianceStructureAnalysisStructuralCovarianceStructureAnalysis,StructuralEquationModelingandLISRELWhatisSEM?Howareestimationsobtained?Theideaofgoodness-of-fitModelIdentificationTestingnestedmodelsKennethLaw@人民大学人力资源研究所2010StructrualModelStructrualModel2TransformationalLeadershipLeader-memberExchange(LMX)SubordinatePerformanceTransactionalLeadershipKennethLaw@人民大学人力资源研究所2010TheideaofCFATheideaofCFA3ConfirmatoryFactorAnalysis(ExploratoryFactorAnalysis)EFAKennethLaw@人民大学人力资源研究所2010TtdlTwomeasurementmodels4Fairnessperceptionx1=1+1Classicalmeasurementmodelx2=1+2LatentCo-genericmeasurementmodelx1=111+1x2=211+2Item2Item1x1x222112LatentKennethLaw@人民大学人力资源研究所2010Elfli(EFA)Exploratoryfactoranalysis(EFA)5FactorsFVarF1F2F3h2x1.60-.06.02.36x2.81.12-.03.67F1F2x3.17.73.08.60x4.01.65-.04.42x5.03.10.87.6512226547x1x2x6x3x4x5x6.12.22.65.47F3KennethLaw@人民大学人力资源研究所2010Cfifli(CFA)Confirmatoryfactoranalysis(CFA)6FactorsVarF1F2F3x1.6000x2.8100F3F1F2x30.730x40.650x500.870065fitx600.65x1x2x6x3x4x5KennethLaw@人民大学人力资源研究所2010ConfirmatoryFactorAnalysis(CFA)ConfirmatoryFactorAnalysis(CFA)FairnessTrustin7x1=111+1x2=211+2FairnessperceptionTrustinSupervisor22112x3=322+3x4=422+444224Item2Item3Item1Item4x1x2x3x4Item2Item3Item11.Mysupervisorisfair.2.Mysupervisortreatsuswithoutbiases.3IbelievethatmysupervisorwillprotectmeKennethLaw@人民大学人力资源研究所20103.Ibelievethatmysupervisorwillprotectme.4.Itrustmysupervisor.ConfirmatoryFactorAnalysis(CFA)ConfirmatoryFactorAnalysis(CFA)8x1=111+1x2=211+2x3=322+311+3x4=422+4Item2Item3Item1Item4x1x2x3x4Item2Item3Item1NoCrossLoadings1.Mysupervisorisfair.2.Mysupervisortreatsuswithoutbiases.3Ibelievethatmysupervisorwillprotectme(crossloadingiscommoninpersonalityinventories)KennethLaw@人民大学人力资源研究所20103.Ibelievethatmysupervisorwillprotectme.4.Itrustmysupervisor.ConfirmatoryFactorAnalysis(CFA)ConfirmatoryFactorAnalysis(CFA)+9x1=111+1x2=211+2x=+x3=322+3x4=422+4Lettheerrorsofx2&x3correlateItem2Item3Item1Item4x1x2x3x4Item2Item3Item1Nltd1.Mysupervisorisfair.2.Mysupervisortreatsuswithoutbiases.3IbelievethatmysupervisorwillprotectmeNocorrelatederrorsKennethLaw@人民大学人力资源研究所20103.Ibelievethatmysupervisorwillprotectme.4.Itrustmysupervisor.Whtltd?Whynotcorrelatederrors?10x1=111+1x2=211+2x2211+2x3=322+3x4=422+4x4422+4x1x2x3x4KennethLaw@人民大学人力资源研究所201011FromcorrelationtoFromcorrelationtoStructuralEquationqModelingKennethLaw@人民大学人力资源研究所2010StructuralModelStructuralModel12PathdiagramJobsatisfactionJobperformanceprxy=.36PredictorvariableIndependentvariableCriterionvariableDependentvariableIndependentvariablepKennethLaw@人民大学人力资源研究所2010StructuralModelStructuralModel13AbilityPersonalityPersonalityJobsatisfactionJobperformanceAgeGenderCriterionvariableGenderTenureCeovbeDependentvariableKennethLaw@人民大学人力资源研究所2010StructuralModelStructuralModel14AbilityPersonalityJobperformancePersonalityJobsatisfactionAgeGenderTurnoverGenderTenureKennethLaw@人民大学人力资源研究所2010PathModelPathModel15PathAnalysisTransformationalLeadershipLeader-memberExchange(LMX)SubordinatePerformanceTransactionalLeadershipKennethLaw@人民大学人力资源研究所2010PathmodelvsMeasurementModelPathmodelvs.MeasurementModel16Inapath(structural)model,allthevariablearemeasured,(thatisobservedandconcrete,weassumenomeasurementerrors).TransformationalTransformationalLeadershipLeader-memberExchange(LMX)SubordinatePerformanceTransactionalLeadershipInameasurementmodel,theitemsareobservedandconcrete,butthelatentvariables(constructs)areunobservedandabstract.FairnessperceptionTrustinSupervisorLatentx1x2x3x4ObservedLatentKennethLaw@人民大学人力资源研究所2010TerminologyandsymbolsTerminologyandsymbolsCircle–unobservableconstructs17CircleunobservableconstructsRectangle–observablemeasuresArrow–causaldirectionDiagramflowsfromlefttorightgowsoeogTwowaycurvedarrow–unexplainedcorrelationsx1y1y2Factorloading()x2y3g()Latentconstruct()Factorcorrelation()Measurementerror(,)x3xy4()Estimationerror()Exogenousvariables,(variableswhichcausesarenotexplained)Endogenousvariables,(variableswhichcausesareexplainedinthemodel)x4KennethLaw@人民大学人力资源研究所2010Theeffectofexogenousvariablesonendogenousvariables()Theeffectofendogenousvariablesonendogenousvariables()StructuralEquationModelingStructuralEquationModeling1811x1y1TransformationalLeadership112x11y2y1223y3LMXJobperformance23x24y3434x24TransactionalLeadershipKennethLaw@人民大学人力资源研究所2010StructuralEquationModelingStructuralEquationModelingfi19TransformationalLeadershipx1y1x2y2LMXJobperformancey3x3y4TransactionalLeadershipx4KennethLaw@人民大学人力资源研究所2010StructuralEquationModelingStructuralEquationModelingTransformational20x1TransformationalLeadershipy1x2y2LMXJobperforman