An inverse-probability-weighted estimator under dependent censoring and left-truncation

• Main Authors: Hsin-Ping Wu, 吳欣屏
• Other Authors: 沈葆聖
• Format: Others
• Language: en_US
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/33360608118251356044

碩士 === 東海大學 === 統計學系 === 91 === Satten, Datta and Robins (2001) proposed an estimator (denoted by S(t)) of survival function of failure times that is in the class of survival function estimators proposed by Robins (1993). The estimator is appropriate when data are subject to dependent censoring. In this article, we consider an alternative estimator of survival function (denoted by Sw (t)) that is represented as an inverse-probability-weighted average (Satten and Datta (2001)). Both estimators S(t) and Sw (t) can be extended to the data subject to dependent censoring and left-truncation. Simulation results show that for dependent censored data the standard deviation and the mean-squared error of S(t) are smaller than those of Sw (t), while for dependent censored and left-truncated data the situation is reversed when censoring is not heavy. Key Words: Dependent censoring;left truncation; Aalen's linear model.