Estimation of the Cox model with grouped lifetimes

This paper presents how random numbers can be used to transform grouped lifetimes into a pseudo-complete sample. The aim of the study is to investigate the Fisher consistency of the partial likelihood estimator of the regression parameters in the Cox model based on the restored sample. It has been proven that for elliptical-type distributional assumptions about explanatory variables the estimators of the regression parameters in the Cox model based on the pseudo-complete sample are consistent up to a scaling factor. A simulation study illustrates the asymptotic properties of the estimates. In addition, real data case analysis is presented.