生存分析中时间变量的选择

20096①、、、、2001Klein2003Aalen2008。10、。。2090S-plus/R、SAS、STATA、SPSS。“”。。。、。【】。、、、。。【】【】。①、。。88。。。20007180Klein2003。。。80。80。80。。、、①。。。。。②。、。Korn1997Klein2003Lamarca1998Li2005Thiebaut2004。。Korn1997、Thiebaut2004。①80。②Medline80%。8920096、Klein2003。。808080。。。。。ddi=wiyiδiziwiyiδi10。zi。hxzβ①。xβ。hxzβ=h0xeβz1h0x。。h0x。h0xCoxCox1972。。。di=tix0iδizitix0i。htx0zφγ=h0texpφx0+γz2φγ。h0tt。60701002φx0。①、、。。。42001。。2001。。90。。。。。。。Lβ=ni=1仪h0yieβziS0yiexp(βzi)S0wiexp(βzi)iiδiS0yiexp(βzi)S0wiexp(βzi)ii1-δi3S0wiexp(βzi)S0yiexp(βzi)。3。S0wiexp(βzi)。Lφγ=ni=1仪h0te(φx0+γz)S0texp(φx0+γz)δiS0texp(φx0+γz)1-δi=ni=1仪h0te(φx0+γz)δiS0texp(φx0+γz)4。34。Cox。。Lβ=rj=1仪eβzi∑i∈Ry(j)eβzi5ji。Ry={i|WiyYi}。。、。Lφγ=rj=1仪eφx0j+γzj∑i∈Rt(j)eφx0j+γzj69120096、56。80818185。111。。。、。8019982000、2002、20052008Zeng20022004。1998、20002002。。720480。。80200241.8%。Cox。。。。。。80100φ。。921eβeγ。2。3AICAkaike’sInformationCriterionHirotsuguAkaike。AIC=2k-21nL。kL。AIC。4*p0.05**p0.01***p0.001。51998～2002。eβ95%eγ95%80～8990～99-1.821.681.96***100～105-2.282.092.50***0.700.650.76***0.700.650.75***1.030.961.101.050.981.1201～60.980.901.070.970.891.0570.860.750.99*0.850.740.98*0.850.770.95**0.850.770.95**0.750.680.84***0.750.670.84***0.700.610.80***0.700.610.80***0～25%26%～50%0.820.750.89***0.790.730.86***51%～75%0.710.650.77***0.660.610.72***76%0.540.470.62***0.500.440.57***1.271.161.39***1.321.211.44***1.661.531.80***1.781.641.92***AIC5410869244。。。、、。。。Akaike1974。1。。2。9320096。、。、eγeβ。Korn1997Thiebaut2004。Korn1997。γβ。γβ、。zx0γβ。Thiebaut2004。。。。xh0x=aeφxa0。h0x=h0x0+t=aeφ(x0+t)=aeφx0eφth0t=aeφth0x=h0teφx0。。γβ。。Korn1997。。GompertzGompertz1825。Gompertzμx=aebxa0。。。Korn1997。Thiebaut2004。。Thiebaut2004。βln1ln5050%～90%。β0βThiebaut2004。。50%～60%β。。94、。、。Korn1997、Thiebaut2004。。。。。。1.2001《》《》1。2.2004《》。3.AalenO.O.Borgan覫.andGjessingH.K.2008SurvivalandEventHistoryAnalysisAProcessPointofView.SpringerNewYorkInc.4.AkaikeHirotugu1974ANewLookattheStatisticalModelIdentification.IEEETransactionsonAutomaticControl.196716～723.5.CoxD.R.1972RegressionModelsandLifeTables.JournalofRoyalStatisticalSociety.B34187～220.6.GompertzB1825OntheNatureoftheFunctionExpressiveoftheLawofHumanMortalityetc.Phil.Thans.Roy.Soc.115513～585.7.KleinP.J.andM.L.Moeschberger2003SurvivalAnalysisTechniquesforCensoredandTruncatedDataSecondedition.Springer-VerlagNewYork.Inc.8.KornE.L.B.L.GraubardandD.Midthune1997Time-to-eventAnalysisofLongitudinalFollow-upofaSurveyChoiceoftheTime-scale.AmericanJournalofEpidemiology.1451pp.72～80.9.LamarcaR.J.AlonsoG.GomezandA.Munoz1998Left-truncatedDatawithAgeasTimeScaleAnAlternativeforSurvivalAnalysisintheElderlyPopulation.TheJournalofGerontologySeriesABiologicalSciencesandMe-dicalSciences.535pp.337～343.10.LiQ.2005SubjectiveWell-beingandMortalityinChineseOldestOld.WorkingPaperinMaxPlanckInstituteforDemographicResearch.11.ThiébautA.C.M.andBénichouJ.2004ChoiceofTime-scaleinCox’sModelAnalysisofEpidemiologicCohortDataASimulationStudy.StatisticsinMedicine.2324pp.3803～3820.12.ZengYiVaupelJ.W.XiaoZhenyuZhangChunyuan&LiuYuzhi2002SociodemographicandHealthProfilesoftheOldestOldinChina.PopulationandDevelopmentReview.282251～273.9520096TheChangeofWagePaymentStructureofUrbanChinaduringEconomicTransitionAQuantileRegressionalAnalysisXiaQingjieSongLinaSimonAppleton·59·Inthispaper,weexaminethedeterminantsofChina’surbanwagesfrom1988to2002byapplyingquantileregressionmethodoverthewholerangeofwagedistribution.Wefindincreasedreturnstoeducationbutadecreaseinthereturnstoexperience.The2002dataimplythatthewideningpuregendergapmayhavecometoanend.Thereformofthestate-ownedsectorisshowntoimpactwagesofworkerswithinthosesectors.Attheturnofthecentury,theChineseeconomyquietlystarteditschangefromlabourintensivenesstoknowledgeintensiveness;besides,itseemsthatunskilledlabourbecomescarce.GenderWageDiscriminationatQuantilesEvidencefromNortheastUrbanLaborMarketinChinaZhangShiweiGuoFengming·69·BasedonthemicrodataintheNortheasturbanlabormarketinChina,thepaperregressesthewageequationsofmaleandfemaleatdifferentquantilesanddecomposethegenderwagegapwhichcomefromquantileregression.TheresultsshowthatthereisagenderwagegapintheNortheasturbanlabormarket.Inlow-incomegroups,thegenderwagegapiscausedbydifferencesinpersonalcharacteristicsandgenderdiscrimination.Inhigh-incomegroups,thegenderwagegapisentirelycausedbygenderdiscrimination.Withthemaleandfemalewagesrise,thegenderwagegapisshrinking,buttheextentofgenderwagediscriminationhascontinuedtoexpand.Therefore,whiletheGovernmentinenhancinghumancapitaloflow-incomewomen,itshouldrunfairemploymentsystemsandwagesystemstomakediscriminationlower.TheImpactofDifferentTypesofHumanCapitalontheQualityofFDIinChinaFuYuanhaiWangZhanxiang·80·Byusingprincipalcomponentregression,thispaperinvestigatestheeffectsofthreelevelsofeducationqualification,elementaryschoolorjuniorhighschool,seniorhighschool,anduniversityonthequalityofFDI.Theresultsshowthatinlabor-intensiveindustrieswidelyutilizingFDI,increasingtheproportionofhumancapitalwithjuniorhighorelementaryschoolqualificationandwithseniorhighschoolqualificationcanimprovethequalityofFDI.Comparingt