Premium
Testing the equality of two survival functions with right truncated data
Author(s) -
Chi Yunchan,
Tsai WeiYann,
Chiang ChiaLing
Publication year - 2006
Publication title -
statistics in medicine
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.996
H-Index - 183
eISSN - 1097-0258
pISSN - 0277-6715
DOI - 10.1002/sim.2556
Subject(s) - truncation (statistics) , parametric statistics , test statistic , statistics , mathematics , statistic , log rank test , contrast (vision) , rank (graph theory) , data set , parametric model , cumulative distribution function , computer science , survival analysis , statistical hypothesis testing , probability density function , combinatorics , artificial intelligence
To compare the survival functions based on right‐truncated data, Lagakos et al. proposed a weighted logrank test based on a reverse time scale. This is in contrast to Bilker and Wang, who suggested a semi‐parametric version of the Mann–Whitney test by assuming that the distribution of truncation times is known or can be estimated parametrically. The approach of Lagakos et al. is simple and elegant, but the weight function in their method depends on the underlying cumulative hazard functions even under proportional hazards models. On the other hand, a semi‐parametric test may have better efficiency, but it may be sensitive to misspecification of the distribution of truncation times. Therefore, this paper proposes a non‐parametric test statistic based on the integrated weighted difference between two estimated survival functions in forward time. The comparative results from a simulation study are presented and the implementation of these methods to a real data set is demonstrated. Copyright © 2006 John Wiley & Sons, Ltd.