事件史分析方法介绍

19973(102)POPULATION&ECONOMICSNo.3,1997(Tot.No.102),,,1996,,(361005),1960,,(361005)(EventHistoryAnalysis),:;,;;,,,,,,,,,,,:(censoring)(ex2planatoryvariable):20,,,,,,,(survivalanalysis);(failuretimeanalysis);(Markov),,,185060,,,,,,,74©1995-2004TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.,,,1,,,6070,,Tuma2,,11,;:,;,,,,,,,,,,,,,,(competingrisks),,,,,(censoring)(explanatoryvairable),,,,,,(hazardrate),,,,,P(t,t+s)(t,t+s),,s=1,s,h(t),:h(t)=limP(t,t+s)/s(1)s0,,(84©1995-2004TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.)P(t)t,:h(t)=-P(t)P(t)(,)(1)21,,,;,,GompertzCox3,,,,,,,,,,,,,,,P(t)tnx1(t),x2(t),,Xn(t),P(t)x1(t),x2(t),,xn(t),:P(t)=a+b1x1(t)+b2x2(t)++bnxn(t)(2)t=1,2,5,,P(t),,1,,P(t)logit,:log(P(t)/(1-P(t)))=a+b1x1(t)+b2x2(t)++bnxn(t)(3)P(t)01,,logitbi(i=1,n)xi(i=1n),logit,a,,,:log(P(t)/(1-P(t)))=a(t)+b1x1(t)+b2x2(t)++bnxn(t)(4)a(t),b1,b2,,bn(,),(proportionalharzardsmodel):logh(t)=a(t)+b1x1+b2x2++bnxn+u(5)h(t),xi(i=1,n),a,b1,b2,,bn,uxi(i=1,n)u,(exponential),(weibull),(Gompertz),(gamma)(5)(hi(t)/hi(t)=c)c,(,)(partiallikelihoodmethod),94©1995-2004TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.,bi(i=1,,n),bi(i=1,,n)a(t),,,,,4:SASSPSSBMDP200,5060,,5112001112001055225189113231016410614131541084512141108551292008481,,,()15200,129,,,5,:(x1)(x2),,:(x3(t));(x4(t));(x5(t))1,0(,,),,200,,11,11(,,),,,,,1,12005141,,,:,,189,2525/189=01132,11,(),,,,,(6)bi(i=1,5)a(t)5log(P(t)/(1-P(t)))=a(t)+62i=1bixi+65i=3bixi(t)(6)05©1995-2004TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.a(t)5,,,:,,,3,35,5,200,8481,5,1,0848,logit(SPSSSAS)(6)22Logit848btbt10561261-196-2111321025106-1078-214733-174-116010231794-118-14210069213333(5)2135-116-31123322612530105330101,;,5,,,3333:,;,,:PL=7kk=1Lk(A1),PL,KLk,A1,10,5,5,12,4659,,15©1995-2004TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.A1itikLk121ebx1/(ebx1+ebx2++ebx10)242ebx2/(ebx1+ebx2++ebx10)353ebx3/(ebx1+ebx2++ebx10)453564ebx5/(ebx1+ebx2++ebx10)697115ebx7/(ebx1+ebx2++ebx10)8123912310123:i,tii,k3,A112,10212:L1=h1(2)h1(2)++h1(2)(A2),:hi=exp(a(t)+bxi)=exp(a(t))exp(bxi)(A3)(A2):L1=exp(bx1)exp(bx1)++exp(bx10)(A4):L2=exp(bx2)exp(bx2)++exp(bx10)(A5)L3,L4,L5A1,Liii-1i+1,i,(Ai)(i=2,3,4,5,6)A(1),,LogPL,(,):13Cox,D.R.(1972)Regressionmodelsandlifetabel.JournaloftheRoyalstatisticalSociety,SeriesB34:187202.2Tuma(1976)Rewards,resourcesandtherateofmobility:anonstationarymultivariatestochasticmodel.AmericanSociologi2calReview41:338360.4Efron,B(1977)TheefficiencyofCoxslikelihoodfunctionforcensoreddata.JournaloftheAmricanStatisticalAssociation72:557-565.5Tuma,N.B.(1982)Nonparametricandpartiallyparametricapproachestoeventhistoryanalysis,PP.160inS.Leinhardt(ed.)SociologyMethodology1982.SanFrancisco:jossey-Bass.25©1995-2004TsinghuaTongfangOpticalDiscCo.,Ltd.Allrightsreserved.