Distribution-Free Methods of Estimating Location Difference With Censored Paired Data
针对配对数据存在任意右删失的情况，基于广义Gehan-Gilbert检验，提出了位置差异的非参数点估计和区间估计方法，并证明了点估计的渐近正态性和强相合性。
Abstract Nonparametric point and interval estimators of location difference in matched pairs data with arbitrary right censorship are proposed, based on a generalized Gehan—Gilbert test (Wei 1980). The point estimator is shown to be asymptotically normal and strongly consistent. The limits of the corresponding interval estimator are readily computable. An example is also provided to illustrate the techniques to problems in survival data analysis.