Confidence Intervals for a Population Size Based on Capture-Recapture Data

Abstract Capture-recapture is a popular sampling method to estimate the total number of individuals in a population. This method is also used to estimate the size of a target population based on several incomplete records/databases of individuals. In this context, a simple approximate confidence interval (CI) based on the hypergeometric distribution is proposed. The proposed CI is compared with a popular approximate CI, likelihood CI and an exact admissible CI in terms of coverage probability and precision. Our numerical study indicates that the proposed CI is very satisfactory in terms of coverage probability, better than the popular approximate CI, and much shorter than the admissible CI. The interval estimation method is illustrated using a few examples with epidemiological data.