|
Activity Number:
|
178
|
|
Type:
|
Invited
|
|
Date/Time:
|
Monday, July 30, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #307764 |
|
Title:
|
Bayesian Inference for a Stratified Categorical Variable Allowing All Possible Category Choices
|
|
Author(s):
|
Balgobin Nandram and Myron Katzoff*+ and Ma Criselda S. Toto
|
|
Companies:
|
Worcester Polytechnic Institute and National Center for Health Statistics and Worcester Polytechnic Institute
|
|
Address:
|
Office of Research and Methodology, Hyattsville, MD, 20782,
|
|
Keywords:
|
Bayes factor ; Monte Carlo integration ; Multinomial-Dirichlet model ; Sparse table ; Business survey ; Random samples
|
|
Abstract:
|
We consider problems of inference from survey data for proportions when sample individuals have been asked to mark all responses that apply to them, the table with mutually exclusive categories is sparse, the number of individuals to whom none of the listed categories apply is missing and the category proportions are to be compared across population strata. We consider an example from the Kansas Farm Survey. We use a Bayesian product multinomial-Dirichlet model to fit the count data both within and across education levels. We estimate the proportions of individuals with each choice; show how to estimate the most frequently indicated choice; and show, using the Bayes factor, how to test that these proportions are the same over different levels of farmers' education. Our Bayesian procedure uses a sampling based method with independent samples.
|
