Effect of list errors on the estimation of population size

In many situations, it is possible to estimate the size of a closed population if some members of the population are recorded on one or more administrative lists. An important example is estimating the prevalence of a disease, where some members of the disease population may be recorded on lists such as disease registries, hospital admission records, and general practitioner records. An incomplete contingency table is formed by matching the lists and the missing cell count estimated by prediction based on a fitted model, which assumes that the matching is done without error. In practice, matching errors do occur, and in this article, we examine the effect of these errors on the estimation process and show how the standard models may be modified to allow for matching errors.