Statistics and Its Interface

Volume 14 (2021)

Number 3

A test against the stratified additive hazards model

Pages: 267 – 277

DOI: https://dx.doi.org/10.4310/20-SII635

Authors

Yanqin Feng (School of Mathematics and Statistics, Wuhan University, Wuhan, China; and Computational Science Hubei Key Laboratory, Wuhan University, Wuhan, Hubei, China)

Xin Yuan (School of Mathematics and Statistics, Wuhan University, Wuhan, Hubei, China)

Shishun Zhao (Center for Applied Statistical Research and College of Mathematics, Jilin University, Changchun, China)

Abstract

Stratified data are commonly encountered in practice. An interesting problem for this type of data is to test the existence of stratum effect. This paper discusses hypothesis testing of stratum effect for interval-censored data with informative observation times under the stratified additive hazards model. We construct a test statistic for this test problem, and show that the test statistic is asymptotically distributed as a chi-squared distribution. Finite sample performance of the proposed method is assessed through an extensive simulation study, which indicates the procedure works well. A real data set from a hemophilia study is analyzed to illustrate the proposed method.

Keywords

interval-censored, Informative censoring, partial likelihood, hypothesis test

Zhao’s research was partially supported by NSFC Grant 11671168.

Received 29 December 2019

Accepted 2 September 2020

Published 9 February 2021