|
Activity Number:
|
248
|
|
Type:
|
Invited
|
|
Date/Time:
|
Tuesday, August 4, 2009 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods
|
| Abstract - #302809 |
|
Title:
|
Model and Variable Selection in Hierarchical Small Area Models
|
|
Author(s):
|
Michael D. Larsen*+ and Lu Lu
|
|
Companies:
|
Iowa State University and Iowa State University
|
|
Address:
|
Department of Statistics, Ames, IA, 50011,
|
|
Keywords:
|
Posterior predictive check ; Discrepancy measures ; Benchmarking ; Small area estimation ; Generalized linear mixed model ; Posterior mean square error
|
|
Abstract:
|
In the context of small area estimation, hierarchical Bayesian (HB) models are often proposed to produce more reliable estimators of small area quantities than direct estimates, such as design-based survey estimators. A method which benchmarks the HB estimates with respect to the higher level direct estimates and measures the relative inflation of posterior mean square error due to benchmarking in the posterior predictions is developed to evaluate the performance of hierarchical models. Both numerical and graphical summaries of posterior predictive discrepancy measures are available. The benchmarked hierarchical Bayesian posterior predictive model comparison method is shown to be able to select proper models effectively in a simulation study. The method is then applied to selecting a generalized linear mixed model for complex sample survey data.
|
