|
Activity Number:
|
100
|
|
Type:
|
Topic Contributed
|
|
Date/Time:
|
Monday, August 3, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #304903 |
|
Title:
|
A Full Gibbs Sampler for a Multinomial Probit Model with Endogeneity
|
|
Author(s):
|
Lane Burgette*+ and Erik Nordheim
|
|
Companies:
|
University of Wisconsin-Madison and University of Wisconsin-Madison
|
|
Address:
|
1220 Medical Sciences Center, Madison, WI, 53706,
|
|
Keywords:
|
Bayesian ; multinomial probit ; endogeneity ; marginal data augmentation
|
|
Abstract:
|
We introduce methods for estimating a Bayesian multinomial probit switching model for unordered selection and response categories via a fully Gibbsian estimation strategy. We achieve this through the use of marginal data augmentation and a modified Imai/van Dyk-type prior for the covariance matrix. Compared to related work that requires a Metropolis-Hastings step, this method improves computational efficiency and overall simplicity. Additionally, we modify the Chib method to estimate Bayes factors. The estimation strategy is applied to simulated data and is used to model retirement outcomes for groups with differing retirement preferences in the Wisconsin Longitudinal Study.
|
