|
Activity Number:
|
1
|
|
Type:
|
Invited
|
|
Date/Time:
|
Sunday, July 29, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308095 |
|
Title:
|
Discrete Hierarchical Model Search with Conjugate Priors for Log-Linear Parameters
|
|
Author(s):
|
Adrian Dobra*+ and Helene M. Massam
|
|
Companies:
|
University of Washington and York University
|
|
Address:
|
15514 Bothell Way NE, Seattle, WA, 98155,
|
|
Keywords:
|
Model selection ; log-linear models ; Bayesian statistics ; Contingency tables ; Reversible jump
|
|
Abstract:
|
We propose a novel reversible jump Markov chain Monte Carlo method (Green, 1995) for model selection in hierarchical log-linear models. Our approach is based on the conjugate priors for log-linear parameters introduced in Liu and Massam, 2006. We discuss the computation of Bayes factors as an alternative to RJMCMC and the Bayesian iterative proportional fitting algorithm for sampling model parameters. We also compare our work with similar results based on multivariate normal priors for log-linear models (Dellaportas and Forster, 1999; King and Brooks, 2001).
|
