|
Activity Number:
|
202
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 7, 2006 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Biometrics Section
|
| Abstract - #306776 |
|
Title:
|
Bayesian Analysis of Repeated Data with Many Zeros: Application to the Longitudinal Adolescent Substance Abuse Study
|
|
Author(s):
|
Hyonggin An*+
|
|
Companies:
|
The University of Iowa
|
|
Address:
|
200 Hawkins Drive, C22B GH, Iowa City, IA, 52242,
|
|
Keywords:
|
zero inflated Poisson ; Bayesian analysis ; random effects ; two-parts model
|
|
Abstract:
|
Repeated data with too many zeros are quite common in many fields of study. For these data, Tooze et. al. (2002) proposed a two-parts model with random effect for continuous outcomes while Hall (2000) proposed a zero-inflated Poisson regression with random effects for counting outcomes. The parameters in these models were estimated by maximum likelihood. In this paper, as alternatives to maximum likelihood approaches, we propose Bayesian estimation methods for both the two-part model and the zero-inflated Poisson regression with random effects. The Bayesian methods are more flexible and can be easily implemented in the existing software, R and WinBUGS. The results from real data analysis comparing the standard methods and the proposed methods in the longitudinal adolescent substance abuse study are also provided.
|
- The address information is for the authors that have a + after their name.
- Authors who are presenting talks have a * after their name.
Back to the full JSM 2006 program |
