Department of Statistics Seminar
North Carolina State University

presents

Dr. Sujit Ghosh

North Carolina State University

"Bayesian Analysis of Capture-Recapture Models"

ABSTRACT

We consider the capture-recapture problem of estimating the size of a closed population. Several applications of this problem include estimating the size of a closed biological population, software review and census undercount. For instance, in census undercount several lists of individuals are available based on different administrative records. The goal is to estimate the total number of individuals including those remaining unlisted. In this talk, we present Bayesian analysis of capture-recapture models which allow heterogeneity of capture probabilities between individuals. We utilize a nonparametric mixture model which does not force a parametric structure on the heterogeneity. We use Markov Chain Monte Carlo (MCMC) methods to obtain samples from the joint posterior distributions of the parameters. Through illustrations with several real-world data sets, we demonstrate how Bayesian analysis can be applied in this setting and some major benefits which result. In addition to allowing heterogeneity of capture probabilities between individuals, we also examine models which allow trap/bait effects and we develop rigorous Bayesian model selection criteria to assist in choosing the correct model.

Friday, April, 26, 2002

3:35 - 4:35 pm

206 Cox Hall

Refreshments will be served on the second floor of Dabney Hall (left of Room 222) at 3:00 pm.