An empirical Bayes approach to analyzing recurring animal surveys

Recurring estimates of the size of animal populations are often required by biologists of wildlife managers. Because of cost or other constraints, estimates frequently lack the accuracy desired but cannot readily be improved by additional sampling. This report proposes a statistical method employing empirical Bayes (EB) estimators as alternatives to those customarily used to estimate population size, and evaluates them by a subsampling experiment on waterfowl surveys. EB estimates, especially a simple limited—translation version, were more accurate and provided shorter confidence intervals with greater coverage probabilities than customary estimates.