调节效应与中介效应的比较和应用

　　　　　2005,37(2):268～274　　ActaPsychologicaSinica268　　　:2004-06-30*(DBA010169)()。　:,E-mail:[e-mail]*温忠麟1,2　侯杰泰2　张　雷2(1,510631)　(2,)　　,。、、、、,、。,。　,,,。　B841.2　　(moderator)(mediator),。,,。,:(1),。,,(interactwith,1.2),(mediating)[1]。(2)。,[2,3]。(3)。,[4]。,。,Holmbeck[5]。　　,。,,,,,。　　,。。,。1　　　,(Y)(X)。,,“XY”。,,。1.1　　　YXM,M[6]。,YXM,1。(、、),(、、),()[7]。1　　2:269　　　　,,:,,。,(、),:,;,,。　　,(,[8])。,YXY=aX+bM+cXM+e(1)Y=bM+(a+cM)X+eM,YX。YXa+cM,M,c(moderatingeffect)。1.2　　　(1)c。c(H0∶c=0),M。(interactioneffect)(1),cXM,。,。　　,。,,;,,。(,[8],[9])。,,,,。,,,,。,,。、、,,,(1),。,。1　(M)(X)(ANOVA),。:M,YX。,。,,　　Y=aX+bM+cXM+e:　　1.YXM,R21。　　2.YX、MXMR22,R22R21,。,XM,,。,　　Y=aX+bM+cXM+e()。　　XM,(XM2,;MX2,)。1.3　　　。。(、)(observablevariable),(latentvariable)。1.3.1　　。,(categoricalvariable),,(continuousvariable),。,。1。270　　　　　　　　37　　。,,:(1)YXM,R21。(2)YX、MXMR22,R22R21,;,XM,,。　　、,[10]。、,,(dummyvariable),,。　　,XM(),。,XM,XM。1.3.2　　([11])。,。:,;。　　,[12,13]。。(LISREL8.3,AMOS4.0,EQS6.0)(subgroupmultigroup)。,,χ2。,,χ2。χ2χ2χ2,。χ2,。　　,[14]。AlginaMoulder[15](),WallAmemiya(GAPI)[16](),(constraint),。Marsh,WenHau[17],,,,。2　　　[18]。,。2.1　　　XY,XMY,M。,:——————,“”[6]。2　　　,2。,cXY,abM(mediatingeffect),c′。,c=c′+ab(2)c-c′=ab。2.2　　,,。,2。,c^c,a,b,c′^a,^b,^c′,。^a^b。[^a^b/(^c′+　2:271　　^a^b)]、(^a^b/^c′)[19]。　　YX,c(H0∶c=0),M。、、,[18],,,。3。3　　　3Sobel[20,21],z=^a^b/sab^a,^ba,b,sab=^a2s2b+^b2s2a,sa,sb^a,^b。3　　　,,2。2　MMXYXY、、XYXYY=aM+bM+cXM+eM=aX+e2　　Y=c′X+bM+e3MX,MY,MXMX、YMYX(),XYMX、YMX、Y()MX、Ycab^c^a^bcab,c(t);(F),Sobel[18]　　,,。[6,7]。,、,,,。,,。,272　　　　　　　　37(),(),。,,,,。,,,。　　Thompson,“”、“”、“”、“”、“”“”“”[22]。,“”“”,“”()“”();“”“”(、)。“”“”:“”,“”,“”“”。　　,,。,。4　　　(X)(Y)。()。,[23]。(X),9(、、),(Y),,()。(U),9(,,)。(W)(“”“”5)。N=595。,。:Y18.87W1.130.45X-9.78-2.2094.25U0.630.09-0.220.564.1　　　,(W)。,。,,。　　(W)3。WX(t=-0.98,R20.001),(W)。-0.232,(W),(W),(),()0.232,W。,(W),(W)。(W)。　　,,,。,,,,。3　(W)R2R2Y=2.262W-0.051X0.162Y=-0.051+2.249W　　　-0.051X-0.023WX0.1630.001　　(W)4。,,。(3t),。t,。0.338×0.349/0.232=50.8%,。　2:273　　　　,(W),。:,,,。,,,,,。4　(W)y=-0.232xSE=0.040,t=-5.8**w=-0.338xSE=0.039,t=-8.7**y=0.349wSE=0.040,t=8.7**　-0.114xSE=0.040,t=-2.8**　:SE。**0.01,。4.2　　　(W),(U),。,。,,。5　(U)R2R2Y=1.085U-0.101X0.089Y=0.023+0.987U　-0.107X+0.101UX0.1180.029**　　(U)5。UX(t=4.452,R23%),(U)。(U)[14]。　　,(U),。,(U)。Y=0.023+0.987U+(0.101U-0.107)X,(U),(X)(Y)。U1.06(U,0.75,U=1.061.4),(X)(Y)。,U=1.5(2),XY()。5　　　,,。。,,,,。　　,,,。,。1　DaviesPT,CummingsEM.Children'semotionsasorganizersoftheirreactionstointeradultanger:Afunctionalistperspective.DevelopmentalPsychology,1995,31:677～6842　IreysHT,Werthamer-Larsson,KolodnerKB,GrossSS.Mentalhealthofyoungadultswithchronicillness:Themediatingeffectofperceivedimpac.tJournalofPediatricPsychology,1994,19:205～2223　LazarusRS,FolkmanS.Stress,appraisal,andcoping.NewYork:Springer,1984.22～234　ThompsonRJ,GilKM,BurbachDJ,KeithBR,KinneyTR.Roleofchildandmaternalprocessesinthepsychologicaladjustmentofchildrenwithsicklecelldisease.JournalofConsultingandClinicalPsychology,1993,61:468～4745　HolmbeckGN.Towardterminological,conceptual,andstatisticalclarityinthestudyofmediatorsandmoderators:Examplesfromthechild-clinicalandpediatricpsychologyliteratures.JournalofConsultingandClinicalPsychology,1997,65:599～6106　JamesLR,BrettJM.Mediators,moderatorsandtestsformediation.JournalofAppliedPsychology,1984,69(2):307～3217　BaronRM,KennyDA.Themoderator-mediatorvariabledistinctioninsocialpsychologicalresearch:Conceptual,strategic,andstatisticalconsiderations.JournalofPersonalityandSocialPsychology,1986,51(6):1173～11828　AikenLS,WestSG.Multipleregression:Testingandinterpretinginteractions.NewburyPark,CA:Sage,19919　SchumackerRE,MarcoulidesGA.Interactionandnonlineareffectsinstructuralequationmodeling.Mahwah,NJ:LawrenceErlbaumAssociates,199810　CohenJ,CohenP,WestSG,AikenLS.Appliedmultipleregression/correlationanalysisforthebehavioralsciences(3rded.).Hillsdale,NJ:Erlbaum,200311　HauKT,WenZ,ChengZ.StructuralEquationModelandItsApplications(inChinese).Beijing:EducationalSciencePress,2004274　　　　　　　　37　　(,,..:,2004)12　JöreskogKG.Simultaneousfactoranalysisinseveralpopulations.Psychometrika,1971,36:409～42613　YangF.Modelinginteractionandnonlineareffects:astep-by-stepLISRELexample.In:RESchumacker,GAMarcoulides(Ed).Interactionandnonlineareffectsinstructuralequationmodeling.Mahwah,NJ:Erlbaum,1998.17～4214　WenZ,Hau,KT,MarshHW.Methodsandrecentresearchdevelopmentinanalysisofinteractioneffectsbetweenlatentvariables(inChinese).AdvancesinPsychologicalScience,2003,11(5):593～599　　(,,..,2003,11(5):593～599)15　AlginaJ,MoulderBC.AnoteonestimatingtheJöreskog-YangmodelforlatentvariableinteractionusingLISREL8.3.StructuralEquationModeling,2001,8(1):40～5216　WallMM,AmemiyaY.Estimationforpolynomialstructuralequationmodels.TheJournaloftheAmericanStatisticalAssociation,2000,95:929～94017　MarshHW,WenZ,HauKT.Structuralequationmodelsoflatentinteractions:evaluationofalternativeestimationstrategiesandindicatorconstruction.PsychologicalMethods,2004,9(3):275～30018　WenZ,ChangL,HauKT,LiuH.TestingandApplicationoftheMediatingEffects.ActaPsychologicalSinica,2004,36(6):614～620　　(,,,..,2004,36(5):614～620)19　MacKi