Journal of Official Statistics, Vol.10, No.3, 1994. pp. 245–256
Analysis of Sample Based Capture-Recapture Experiments
Juha M. Alho
Abstract:The estimation of population totals in sample based capture-recapture experiments is considered. We permit heterogeneous capture probabilities and use logistic regression to model them. A Horvitz-Thompson type estimator is introduced and an estimator for its variance is given. These formulas appear to be new even when specialized to the homogeneous case. Some aspects of experiments that combine a census with a sample (such as the U.S. Post Enumeration Survey) are studied. In particular, we discuss the role of sampling weights in the estimation of capture probabilities, and we show how purely sample based estimators can be combined with the census estimates to reduce variance. Logistic catch effort models are applied to the optimal allocation of resources to the sampling part and the capture-recapture part of the experiments. We show with an analytical example that there can be a genuine trade-off between the two sources of error.
Keywords:Catch effort; Horvitz-Thompson; optimization; post enumeration survey; variance estimation.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group