中介效应的检验方法

1001-4918201201-0105-111B844A*11121.5106312.cc'Bootstrapabκ2、R2med。RMBESS。。Bootstrap*[e-mail]1mediatorXMYMXY2007。XY。。。MmediatoreffectMackinnon200214casualstepsapproachc－c'differenceincoefficientsabproductsofcoefficientsabcc'1MacKinnonLockwoodHoffmanWest＆Sheets20022004。Shrout＆Bolger2002Preacher＆Hayes2008Hayes2009Mackinnon＆Fairchild2009Fairchild＆McQuillin2010ZhaoLynch＆Chen2010RuckerPreacherTormala＆Petty2011Soest＆Hagtvet2011。ababMackinnon2008FairchildMackinnonTaborga＆Taylor2009Preacher＆Kelley2011。。12ab0Ⅰ100%Mackinnonetal．2002501201212004Preacher＆Hayes2008Fairchild＆McQuillin2010。。2.1BaronKenny1986。XYc1XMa2XMYbc'3。cab。c'fullmediationc'c'＜cpartialmediation。effectsizeab/cab/c'。。1cccShrout＆Bolger2002Preacher＆Hayes2008Mackinnon＆Fairchild2009Fairchild＆McQuillin2010Zhaoetal．2010Ruckeretal．2011Soest＆Hagtvet2011。ShroutBolger2002abc'c。PreacherHayes2008c。Rucker2011c50100c0.20.1648.2%44.5%。Mackinnon2002PowerⅠ。ccMackinnonFairchild＆Fritz2007Fairchild＆McQuillin2010。cjointsignificanceMackinnon2002。ab0Preacher＆Hayes2008Mackinnon＆Fairchild2009Fairchild＆McQuillin2010Cheong2011。c'Preacher＆Hayes2008Zhaoetal．2010Ruckeretal．2011。PreacherKelley2011。。。。2011Preacher＆Kelley20112011。。c'。cPreacher＆Kelley2011Soest＆Hagtvet2011。。Rucker2011c'。2.2ab0c60120121Mackinnon2002。Sobelasymmetricconfidenceinterval。2.2.1SobelSobel^a^b^a、^b1^a^b^σ^a^bzz=^a^b/^σ^a^bzzzzzz^a^b－zα/2×^σ^a^b^a^b+zα/2×^σ^a^b00MacKinnonetal．20022004。Sobel^a^bz。^a^b^a^b^a^b^a^b^a^b^a^bSobelMacKinnonetal．2002MacKinnonLockwood＆Williams2004MacKinnon2008Hayes20092011Soest＆Hagtvet2011Cheong2011。MachoLeder-mann2011Sobel^σ^a^bDelta。2.2.2。BootstrapdistributionoftheProduct2011。Bootstrap、Bootstrap。Mackinnon2004Sobel、、BootstrapBootstrapSobelBootstrapBootstrap。Bootstrap。n1nnBootstrap1Bootstrap^a^b12BB=1000BB^a^bC^a^b*^a^b*C^a^b＜^a^b*Фz0Фz0z02Z0±Zα/2Ф2Z0±Zα/2Ф2Z0±Zα/2C、1－α00Fritz＆MacKinnon2007MacKinnon20082011。、2009Bootstrap2011Bootstrap－Ardrade2011Bootstrap。3abMackinnon2008。Mackinnon20087ab/cab/c'ab/cab/c'。ab/c。cabc701abPreacher＆Hayes2008Hayes2009Preacher＆Kelley2011。ab/cabcc=ab+c'abc'ab/c1－1Shrout＆Bolger2002Mackinnon2008Hayes2009Fairchild＆McQuillin2010Preacher＆Kelley2011ab/cShroutBolger2002ab/cabc'。ab/cab/c500MackinnonWarsi＆Dwyer1995Shrout＆Bolger2002Mackinnonetal．2007Preacher＆Hayes2008Mackinnon2008Fairchildetal．2009Fairchild＆McQuillin2010Preacher＆Kelley2011。ic=c'+∑ni=1aibiab/cMackinnonetal．2007Preacher＆Kelley2011。ab/cab/c'ab/c'5000Mackinnonetal．1995Preacher＆Kelley2011。ab/c'abc'abc'ab/c'。ab/c'、c'0ab/c'Hayes2009。Preacher2011ab/cab/c'κ2kappa-squared。R2medMackinnon2008Fairchildetal．2009。。3.1κ2κ2κ2=ab/abmaxababmaxabκ201。abmax。、bcaacbabababab。R。3.2R2medR2medR2med=r2MY－R2YMX－r2XYr2MYYMYM2X、MY、r2MYMY12R2YMX3YXM21、23r2XYYXYX213R2medYXMXM21。Fairchild200950R2med。R2medSPSS、SASRR2med。R2medHayes2009Fairchild＆McQuillin2010Preacher＆Kelley20112Venn4YuanMackinnon2009N=354145。RMBESSBootstrap80120121。。ccc=0.164t=1.414p=0.158。c23aa=0.40t=3.34p=0.001bb=0.14t=2.75p=0.006。cⅡ。Bootstrapab0.0140.1290Yuan。。c'c'=0.108t=0.924p=0.356。1。ab/cc34.2%ab/c'c'52.1%κ22.6%R2med0.3%YXM。95%ab/c、ab/c'R2med95%0ab/c、ab/c'R2med0。1ab/c0.342－0.56111.254ab/c'0.521－0.39275.579κ20.0260.0070.057R2med0.003－0.0010.0141Bootstrap1000095%RBootstrap95%。。5cc'Bootstrapabc'κ2、R2med。。YuanMackinnon2009MarkovchainMonteCarloMCMC。WangZhang2011MCMCcensoreddata。MCMCYuanMackinnon2009MCMC1MCMCcredibleintervals。Muthén2010MCMCBootstrap。cross-sectionaldatalongitudinaldataMackinnonetal．2007Maxwell＆Cole2007Mackinnon2008Mackinnon＆Fairchild2009Selig＆preacher2009Soest＆Hagtvet2011Cheong2011。Cheong2011、Sobel、SobelSobelCheongSobel901Sobel。Bootstrap、MCMC。Mackinnon2008Fairchild＆McQuillin2010Preacher＆Kelley2011。、。MaxwellCole2007ab/c。Cheong2011ab/cab/cCheongab/c。、、。、、。、、。BaronR．M．＆KennyD．A．1986．Themoderator-mediatorvariabledistinctioninsocialpsychologicalresearchConceptualstrategicandstatisticalconsiderations．JournalofPersonalityandSocialPsychology5161173－1182．CheongJ．W．2011．Accuracyofestimatesandstatisticalpowerfortestingmediationinlatentgrowthcurvemodeling．StructuralEquationModeling182195－211．FairchildA．J．MacKinnonD．P．TaborgaM．P．＆TaylorA．B．2009．R2effect-sizemeasuresformediationanalysis．BehaviorResearchMethods412486－498．FairchildA．J．＆McQuillinS．D．2010．EvaluatingmediationandmoderationeffectsinschoolpsychologyApresentationofmethodsandreviewofcurrentpractice．JournalofSchoolPsychology48153－84．FritzM．S．＆MacKinnonD．P．2007．Requiredsamplesizetodetectthemediatedeffect．PsychologicalScience183233－239．HayesA．F．2009．BeyondBaronandKennystatisticalmediationanalysisinthenewmillennium．CommunicationMonographs764408－420．MachoS．＆LedermannT．2011．EstimatingtestingandcomparingspecificeffectsinstructuralequationmodelsThephantommodelapproach．PsychologicalMethods16131－43．MacKinnonD．P．2008．Introductiontostatisticalmediationanalysis．MahwahNJEarlbaum．MackinnonD．P．＆FairchildA．J．2009．Currentdirectionsinmediationanalysis．CurrentDirectionsinPsychologicalScience18116－19．MacKinnonD．P．FairchildA．J．＆FritzM．S．2007．Mediationanalysis．AnnualReviewofPsychology5859