186 – Current Topics in Phone Surveys
A Comparison of the Performance of Four Confidence Intervals for Population Size Based on a Capture-recapture Design
Jianjun Hua
Dartmouth College
Paul Nelson
Kansas State University
We use simulation to study and compare the performance, in terms of coverage rate and length, of four methods of constructing confidence intervals for population size based on a two-stage Capture-Recapture (CR) experiment. Two methods are based on the asymptotic normality of point estimators and two are obtained from inverting chisquare and likelihiood ratio tests. In the scope of the settings we studied, we found that the method based on inverting a chisquare test is best and that none of the methods performs well if sampling fractions are small. As a practical matter, our conclusion is that that CR designs are most useful for relatively small populations, such as endangered species, where there may be a rough prior estimate of poulation size to guide sample size selections or in a populations where large samples are easy to obtain.