부산대학교 도서관

We consider the problem of estimating the closed population size for two-sample capture-recapture experiments. For this problem, the Lincoln-Petersen method has been routinely used. However, this approach may result in a biased estimator of the population size for models with some of the major sources of variation in the capture probabilities such as behavioral response and heterogeneity. In this paper, we theoretically examine the bias of the Lincoln-Petersen estimator when the assumptions of equal catchability and independence between samples do not hold. A method to evaluate the magnitude of the bias is developed, based on the cross-product ratio of the capture probabilities in the two samples. The performance of this method is demonstrated through simulation study.