Capture-recapture methods for assessing the completeness of case ascertainment when using multiple information sources

IN THIS paper we consider the problem of estimating the total size of a target population from which a study sample has been obtained by merging names from several routinely collected lists. Previously, we obtained an estimate of this size and showed its relation to traditional capture-recapture models [l, 21. Section 2 of this paper summarizes the notation and assumptions, reviews the method, and corrects our previous expression for the variance of that estimate. In Section 3, we discuss some of the cases in which the underlying assumptions may fail and demonstrate that this may lead to an estimate of population size with a distressingly large bias. Section 4 proposes some methods for dealing with failures of the assumptions and Section 5 presents an example of the use of these methods. Finally, Section 6 is a critical discussion of the techniques.