Sankhya: The Indian Journal of Statistics

2003, Volume 65, Pt. 1, 56--71

Estimation Of Covariate Effects In Cox's Regression Models With Frailties


PEIDE SHI, Peking University, Beijing, China and WING-KAM FUNG, The University Of Hong Kong, China

SUMMARY. A simple approach is proposed for estimating the parameters of Cox's semiparametric regression model with frailty using clustered censored data. Unlike the usual methodologies proposed in the literature, the underlying method does not need to assume a specific frailty distribution. It can also deal with censoring and unequal cluster sizes. Inference procedures derived from the estimating equations are proposed to examine the covariate effects with censored observations. Numerical studies are conducted to investigate the properties of our proposals.

AMS (1991) subject classification. Primary 62N02, secondary 62N01.

Key words and phrases. Cox's regression model with frailty, censoring, generalized estimating equation, martingale, U-statistics.

Full paper (PDF)