Strong representations of the survival function estimator for truncated and censored data with applications

Abstract

A strong i.i.d. representation is obtained for the product-limit estimator of the survival function based on left truncated and right censored data. This extends the result of Chao and Lo (1988, Ann. Statist.16, 661-668) for truncated data. An improved rate of the approximation is also obtained on compact sets. Applications include density and hazard rate estimation. The advantage of the improved rate of the approximation is illustrated via kernel density estimation. © 1993 Academic Press, Inc.