|
Activity Number:
|
248
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
IMS
|
| Abstract - #309355 |
|
Title:
|
Inference for the Cox Regression Models with Time-Dependent Coefficients via Empirical Likelihood
|
|
Author(s):
|
Yanqing Sun*+ and Rajeshwari Sundaram and Yichuan Zhao
|
|
Companies:
|
The University of North Carolina at Charlotte and National Institute of Child Health and Human Development and Georgia State University
|
|
Address:
|
Department of Mathematics and Statistics, Charlotte, NC, 28223,
|
|
Keywords:
|
counting processes ; empirical likelihood approach ; local partial likelihood ; proportional hazards model ; pointwise and simultaneous confidence bands
|
|
Abstract:
|
The Cox regression models with time dependent coefficients have been studied by a number of authors recently. In this paper, we develop pointwise and simultaneous confidence bands for the time dependent regression coefficients via the empirical likelihood approach. The empirical likelihood function is formulated through the local partial likelihood for the regression coefficient functions. We show that the proposed pointwise/simultaneous confidence bands have asymptotic correct levels. Our simulation results indicate that the empirical likelihood based pointwise/simultaneous confidence bands yield better coverage probabilities than those centered at the estimators of the regression coefficient functions. The proposed method is demonstrated through a real data application.
|
