This is the program for the 2010 Joint Statistical Meetings in Vancouver, British Columbia.

Abstract Details

Activity Number: 302
Type: Contributed
Date/Time: Tuesday, August 3, 2010 : 8:30 AM to 10:20 AM
Sponsor: Section on Bayesian Statistical Science
Abstract - #307587
Title: MCMC Algorithms for Bayesian Models Involving Binary Data
Author(s): Santanu Pramanik*+ and Edward Mulrow
Companies: NORC and NORC
Address: , , ,
Keywords: Mixture model ; logit-normal model ; Metropolis Hastings ; Holmes and Held ; Gibbs sampler ; Bayesian computation
Abstract:

In the context of accounting reconciliation data, where estimation of the total error amount of an accounting system and numerous subsystems is of interest, a hierarchical Bayesian mixture model can be used. In this paper, we compare and contrast two different MCMC methods for a logit-normal model, one component of the mixture model. The Metropolis Hastings algorithm requires tuning of the covariance structure for the candidate distribution to achieve satisfactory performance. In some multi-parameter situations, as is ours, this can be a daunting task. The Holmes and Held algorithm uses a Gibbs sampler, after introducing latent variables, to avoid an accept-reject step and hence any tuning, but it may be a slow-running algorithm. We will illustrate the pros and cons of the two algorithms and look at a hybrid approach that might reduce the computation time considerably.


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 2010 program




2010 JSM Online Program Home

For information, contact jsm@amstat.org or phone (888) 231-3473.

If you have questions about the Continuing Education program, please contact the Education Department.