|
Activity Number:
|
287
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 10:30 AM to 12:20 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308582 |
|
Title:
|
A Bayesian Multiple Comparison Procedure for Unbalanced Mixed Models
|
|
Author(s):
|
Junfeng Shang*+ and Farroll T. Wright and Joseph Cavanaugh
|
|
Companies:
|
Bowling Green State University and University of Missouri-Columbia and The University of Iowa
|
|
Address:
|
450 Math Sciences Building, Bowling Green, OH, 43403,
|
|
Keywords:
|
Bayesian method ; unbalanced mixed model ; imputation ; hierarchical model
|
|
Abstract:
|
We propose a Bayesian hierarchical model for multiple comparisons in a mixed modeling framework under a simple-order restriction. We employ Markov Chain Monte Carlo (MCMC) methods to estimate parameters and to obtain estimates of the posterior probabilities that any two of the means are equal, which allows one both to determine if these two means are significantly different and to test the homogeneity of all of the means. In unbalanced data, the behavior of the model is explored along with multiple imputations for values missing at random. Our simulation and application results exhibit that our proposed hierarchical model can effectively unify parameter estimation, tests of hypotheses, multiple imputations for unbalanced data, and multiple comparisons in one setting.
|
