Online Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 557 - Current Trends in Bayesian Survey Sampling: A Celebration of Glen Meeden's 80th Birthday
Type: Invited
Date/Time: Thursday, August 6, 2020 : 3:00 PM to 4:50 PM
Sponsor: Survey Research Methods Section
Abstract #314448
Title: On the use of estimated Bayes factors in latent-class regression modeling
Author(s): Joseph L. Schafer*
Companies: U.S. Census Bureau
Keywords: Bayesian hypothesis testing; classification; EM algorithm
Abstract:

In the Bayesian paradigm, the Bayes factor summarizes evidence from data for comparing alternative hypotheses. For latent-class analysis, a unit-level estimate of the Bayes factor allows us to separate the estimation of within-class item-response parameters from the estimation of regression coefficients for predicting class prevalences or subsequent outcomes. Bayes factors for item-response categories neatly summarize an item's contribution to diagnostic classification.


Authors who are presenting talks have a * after their name.

Back to the full JSM 2020 program