|
Activity Number:
|
509
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 2, 2007 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Bayesian Statistical Science
|
| Abstract - #309616 |
|
Title:
|
Empirical Bayes Models of Poisson Clinical Trials
|
|
Author(s):
|
Boris Zaslavsky*+
|
|
Companies:
|
Food and Drug Administration
|
|
Address:
|
1401 Rockville Pike, Rockville, MD, 20852,
|
|
Keywords:
|
gamma distribution ; negative - binomial distribution ; maximum likelihood ; Poisson distribution ; WinBUGS
|
|
Abstract:
|
The Gamma distribution is a natural prior for Poisson models. Under the empirical Bayes approach, the parameters of this distribution are the maximum likelihood estimator of the marginal negative-binomial distribution. The straightforward numerical search for the maximum likelihood solution is impractical given available software. We propose a simplification to the maximum likelihood problem, by preliminary implementing Markov Chain Monte Carlo (MCMC) method using software WinBUGS. The results of WinBUGS calculations are used as a starting point in the maximum likelihood estimation of gamma distribution. Easily computable approximation formulae may be used to find the maximum likelihood point estimations for the gamma distribution. The solution for this approximation converges to the precise maximum likelihood solution as the sample size increases.
|
