Consistency and Efficiency of Inferences with the Partial Likelihood
研究了双样本生存问题中，在固定或独立右删失下，最大偏似然估计和偏似然得分检验的一致性与有效性，发现估计一致性普遍成立，但检验完全有效性仅在回归参数与干扰参数对称进入风险函数时成立。
Under various parameterizations of the two-sample survival problem with nuisance hazards and fixed or independent right-censoring, the consistency and efficiency of maximum partial likelihood estimators and partial likelihood score tests are investigated rigorously. Consistency of estimation is shown to be typical but full efficiency of tests seems to hold generally only when regression parameters and nuisance parameters enter the hazard function symmetrically.