|
Activity Number:
|
207
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 10, 2004 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
| Abstract - #300208 |
|
Title:
|
Semiparametric Likelihood for Two-stage ODS Scheme with Auxiliary Covariates and Random Effects
|
|
Author(s):
|
Jinhong You*+ and Haibo Zhou
|
|
Companies:
|
University of North Carolina and University of North Carolina, Chapel Hill
|
|
Address:
|
Dept. of Biostatistics, Chapel Hill, NC, 27599-7420,
|
|
Keywords:
|
auxiliary information ; centered effect ; two-stage sampling ; outcome-dependent ; semiparametric likelihood ; asymptotic normality
|
|
Abstract:
|
With the field of epidemiology expanding and evolving an increasing number of studies are conducted using the ODS design with a "continuous" outcome. Another complexity in practical studies often involves the cluster- or center-effects of the study subjects. In this paper we consider a two-stage outcome-dependent design in which center effects are involved, and the selection of the second-stage subsample depends on both the continuous outcome and a continuous auxiliary covariate of the true exposure variable. By nonparametrically estimating the conditional distribution of the exposure variable we propose a semiparametric likelihood method to handle the center-effects and the outcome-dependent nature of the subsampling components. The resulted estimator of the interested parameters is consistent and asymptotically normal. Simulation results show that the proposed estimator performs better than the competing methods that either use only the second-stage data or ingnore the center effects. We illustrate the proposed method with a hypothetical dataset from an environmental epidemiologic study.
|
