Correcting for linkage errors in the multiple capture

In multiple - recapture models different sources are linked and when this is not done perfectly this may lead to biased population size estimates. This paper discusses how to deal with this issue.
The main source to determine the population size of the Netherlands is the Dutch person register. However, it may be that people reside in the Netherlands that are not in this register. Therefore, Statistics Netherlands uses the multiple – recapture model to estimate the size of this unobserved group. However, this method assumes that individuals in different sources can be perfectly identified, which is not always the case. This paper presents an extension on the multiple – recapture method such that it can be applied also when individuals cannot be perfectly identified.