|
Activity Number:
|
381
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Wednesday, August 1, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308900 |
|
Title:
|
Markov Chain Monte Carlo--Based Mixture Modeling Applications in Psychometrics
|
|
Author(s):
|
Dipendra Subedi*+
|
|
Companies:
|
Michigan State University
|
|
Address:
|
1422 Spartan Village, East Lansing, MI, 48823,
|
|
Keywords:
|
Bayesian Methods ; Mixture Modeling ; Psychometrics ; MCMC
|
|
Abstract:
|
There has been growing interest in the modeling of complex data using Bayesian methods over the past few years. Psychometric applications comprise many situations in which data come from two or more distinct populations. Bayesian mixture modeling is a powerful technique to model such data using a mixture of distributions. Although Markov Chain Monte Carlo based mixture modeling has been used extensively in other disciplines (e.g. Biostatistics, Engineering), it has received limited attention in social science. This paper reviews prior work in Bayesian mixture modeling and documents the challenges and potential solutions. With the examples from psychometrics, this study illustrates the use of WinBUGS in such modeling. This study has a potential to disseminate Bayesian mixture modeling approach in psychometrics and facilitates the modeling of complex latent variables of natural phenomenon.
|
