结构方程模型张伟雄PPT课件.ppt_第1页
结构方程模型张伟雄PPT课件.ppt_第2页
结构方程模型张伟雄PPT课件.ppt_第3页
结构方程模型张伟雄PPT课件.ppt_第4页
结构方程模型张伟雄PPT课件.ppt_第5页
已阅读5页,还剩30页未读 继续免费阅读

下载本文档

版权说明:本文档由用户提供并上传,收益归属内容提供方,若内容存在侵权,请进行举报或认领

文档简介

1 RunningLISRELMeasurementModelLISREL测量模型 张伟雄博士香港中文大学工商管理学院副院长管理学系教授GordonW Cheung Ph DProfessor DepartmentofManagementAssociateDean FacultyofBusinessAdministrationTheChineseUniversityofHongKong 2 MeasurementModel ConfirmatoryFactorAnalysisModel 5 6 7 8 82 72 62 52 12 3 ScalingofLatentVariable设定因子的度量单位 Virtuallyalllatentvariableshaveambiguousscales thescalechoiceislargelyarbitrary Aconvenientchoiceistogivethesamescaleasinthesensethatiszeroandisone Anotherchoiceistostandardizethelatentvariablessuchthat 将所有因子的方差固定为1 Bollen 1989 StructuralEquationswithLatentVariables Pp 153 Wiley 4 RunningLISRELProgram Titleline 标题 InputSpecification 输入设定 ModelSpecification 模型设定 OutputSpecification 输出设定 5 InputSpecification DANI k numberofindicators NO numberofcasesLA labelsfortheobservedvariables V1V2V3V4V5RAFI filename ofrawdata CMFI filename ofcovariancematrix KMFI filename ofcorrelationmatrix SDFI filename ofstandarddeviations MEFI filename ofobservedmeans SE SelectandSequence 54321 6 ModelSpecification I MONY 2NX 3NE 1NK 2LY FILX FIBE FIGA FRPH SY FRPS DI FRTE DI FRTD DI FRTY FITX FITX FIKA FILKksilabelsLEetalabels 7 ModelSpecification II FRlistofparameterstobeestimatedFIlistofparameterstohavespecifiedvaluesVAnumericalvalueofFIparametersCOparametermatrixelement expressionwithotherparametersIRlistofparametermatrixelements number numberEQparametervaluesconstrainedtobeequalSTstartingvalues 8 OutputSpecification PathDiagramOUAD OFFND 4MI 9 Exercise CFAModel Drawthediagram 10 Exercise CFAModel Drawthediagram 11 InputSpecification Titleline ExerciseDANI 8NO 200LAV1V2V3V4V5V6V7V8CM1 650 451 140 350 301 010 510 490 431 580 070 200 200 230 750 170 140 170 270 230 650 410 200 070 210 110 250 850 220 230 260 360 240 250 160 75 12 ModelSpecification MONX 8NK 2LX FILKStaffCustomerVA1LX11LX52FRLX21LX31LX41FRLX62LX72LX82PDOUAD OFFND 4MI 13 ReadingLISRELOutput Checkforunreasonableparameters 不正常参数 ModelfitParameterestimates 参数估值 MultiplecorrelationCFA1 ls8 14 Identification Degreetowhichthereisasufficientnumberofequationsto solvefor eachoftheparameterstobeestimated NumberofknownparametersNumberofestimatedparametersqDegreesofFreedom 自由度 df p qCFA1 GIF 15 Goodness of Fit拟合优度 16 MinimumFitFunctionValues拟合函数 17 Chi SquareTest Wheng 1 18 Relativechi square 2to1or3to1areindicativeofanacceptablefit CarminesandMcIver 1981 p 80 19 RootMeanSquareErrorofApproximation Browne Cudeck 1993 EstimatedRMSEA pClose p valuefortestingthenullhypothesisthatthepopulationRMSEAisnogreaterthan 05 20 ComparisonstoaBaselineModel 1 Bentler Bonett 1980 normedfitindex NFI Bollen s 1986 relativefitindex RFI 21 ComparisonstoaBaselineModel 2 Bollen s 1989 incrementalfitindex IFI Tucker LewisNon normedfitindex NNFI 22 ComparisonstoaBaselineModel 3 McDonaldandMarsh s 1990 relativenoncentralityindex RNI Bentler 1990 comparativefitindex CFI 23 OtherFitIndices Goodnessoffitindex GFI AdjustedGoodnessoffitindex AGFI 24 FitIndicesforcomparisonacrossnon nestedmodels Akaike 1973 AkaikeInformationCriteria AIC Browne Cudeck 1989 BCCBayesInformationCriterion Schwarz 1978 Raftery 1993 25 FitIndicesforcomparisonacrossnon nestedmodels Bozdogan s 1987 CAICECVI 26 Hoelter sCriticalN Thelargestsamplesizeforwhichonewouldnotrejectthehypothesisthatamodeliscorrectatalpha 05 27 RootMeanSquareResidual RMR Thesquarerootoftheaveragesquaredamountbywhichthesamplevariancesandcovariancesdifferfromtheirestimatesobtainedundertheassumptionyourmodeliscorrect ThesmallertheRMRis thebetter 28 CFAExercise2 1 Thedatasetcontains204observationsof12continuousvariables The12variablesareindicatorsofthelatentvariablesSelf esteem DepressionandImpulsiveness The12variablesareobservedona5 pointLikertscale ThedatasetislistedinthefileCFA2 PSF 29 CFAExercise2 2 SELF1toSELF5areindicatorsofthelatentvariableSelf esteem DEPRES1toDEPRES4areindicatorsofthelatentvariableDepression IMPULS1toIMPULS3areindicatorsofthelatentvariableImpulsiveness PathDiagramCFA2 GIF 30 CFAExercise2 Output ProgramFileCFA2 ls8 31 NestedModels Whenoneormorefreeparametersofamodelareconstrained themodelthusconstrainedissaidtobenestedintheonefromwhichitwasderived Theconstrainedmodelusuallyhaveaworsefit Doestheimprovementinfitprovidedbyamorecomprehensivemodelwarrantpreferringittoamoreparsimoniousmodelnestedinit 32 LikelihoodRatioTest Problem sensitivetosamplesizeSolution CFIchangesinCFIlessthan 01Cheung Rensvold 2002 StructuralEquationModeling 9 233 255 33 ModificationIndices修正指数 Calculatedforeachconstrainedrelationshippossibleinaspecifiedmodel TheModificationIndex MI valueforaspecificconstrainedparameterindicatestheimprovementinoverallmodelfit reductioninChi squarestatistic thatispossibleifthatparameterisnotconstrained CFA2 ls8 Hair etal 1995 MultivariateDataAnalysiswithReadings Pp 620PrenticeHall 34 Second OrderFactorAnalysis高阶因子分析 Amoregeneralandabstractvariablemaydeterminethe first order latentvariables CFA3 GIF 2ord CFA 1 GIFThe fi

温馨提示

  • 1. 本站所有资源如无特殊说明,都需要本地电脑安装OFFICE2007和PDF阅读器。图纸软件为CAD,CAXA,PROE,UG,SolidWorks等.压缩文件请下载最新的WinRAR软件解压。
  • 2. 本站的文档不包含任何第三方提供的附件图纸等,如果需要附件,请联系上传者。文件的所有权益归上传用户所有。
  • 3. 本站RAR压缩包中若带图纸,网页内容里面会有图纸预览,若没有图纸预览就没有图纸。
  • 4. 未经权益所有人同意不得将文件中的内容挪作商业或盈利用途。
  • 5. 人人文库网仅提供信息存储空间,仅对用户上传内容的表现方式做保护处理,对用户上传分享的文档内容本身不做任何修改或编辑,并不能对任何下载内容负责。
  • 6. 下载文件中如有侵权或不适当内容,请与我们联系,我们立即纠正。
  • 7. 本站不保证下载资源的准确性、安全性和完整性, 同时也不承担用户因使用这些下载资源对自己和他人造成任何形式的伤害或损失。

评论

0/150

提交评论