Impact of linkage errors and erroneous captures

28/09/2017 11:40
Capture-recapture methods are used to estimate the population size of usual residents in the Netherlands. The sensitivity of this method is determined for violation of the assumptions of perfect linkage and erroneous captures
Capture-recapture estimation is an often used methodology for hard-to-reach populations and relies heavily on a couple of assumptions. In this paper for two of these assumptions, sensitivity analyses are conducted: we investigate the effect of linkage error and erroneous captures on the population size estimate. These sensitivity analyses are conducted for the two and three register case, and for two different nationality groups. It was found that for the nationality group with a high implied coverage of one register over the other, the population size estimator was robust to linkage error and erroneous captures. For the nationality group with a low implied coverage, the population size estimator was not robust. It is concluded that researchers should consider the implied coverage of their registers for capture-recapture estimation, since this plays a crucial role in the robustness of the population size estimation.