|
Activity Number:
|
454
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 7, 2003 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Biopharmaceutical Section
|
| Abstract - #301062 |
|
Title:
|
Conditional Estimation for Generalized Linear Models When Covariates Are Subject-specific Parameters in a Mixed Model with Longitudinal Measurements
|
|
Author(s):
|
Erning Li*+ and Daowen Zhang and Marie Davidian
|
|
Companies:
|
North Carolina State University and North Carolina State University and North Carolina State University
|
|
Address:
|
Department of Statistics, Raleigh, NC, 27695-8203,
|
|
Keywords:
|
conditional score ; sufficiency score ; generalized linear model ; mixed-effects model ; longitudinal data ; random effects
|
|
Abstract:
|
This paper studies estimation in generalized linear models in canonical form when unobserved covariate variables are underlying subject-specific parameters of a mixed effects model for observed longitudinal measurements. Although several methods for fitting such joint primary endpoint models have been proposed, a routine assumption that the random effects in the mixed model follow a parametric family (normality) is usually required, which may be unrealistic or too restrictive to represent the key features in the data. Without imposing any parametric distributional assumption on the random effects, we develop a sufficiency estimator and a conditional estimator for the model parameters through conditional likelihood inference approach. Normal model and logistic model are studied in detail as examples. The benefits in terms of relative bias and coverage probability are illustrated by comparing to competing methods via simulation under various distributions of the random effects and by application to real data.
|
