|
Activity Number:
|
337
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 31, 2007 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #308354 |
|
Title:
|
Bayesian Sample Size Determination for the Misclassified Binary Response Variable with One Error-Free Covariate
|
|
Author(s):
|
Dunlei Cheng*+ and James Stamey
|
|
Companies:
|
Baylor University and Baylor University
|
|
Address:
|
1501 S. 9th Street, Waco, TX, 76706,
|
|
Keywords:
|
sample size calculation ; Bayesian inference ; sampling and fitting priors ; Misclassification ; logistic regression
|
|
Abstract:
|
The paper studies Bayesian simple size determination for a misclassified binary response variable when one error-free covariate is included. The performance criterion employed in this study is Bayesian power. The parameter of interest is the covariate direction in the logistic regression model. Unlike many occasions where priors for simulating data and posterior distribution are identical, this study uses more informative sampling priors in the data collection whereas vague fitting priors in the posterior simulation. Simulations demonstrate that the choice of sample priors have a great impact on sample size calculation. Moreover, multiple diagnostic tests result in a smaller sample size.
|
