Measurement errors in multiple systems estimation

Dual and multiple system estimation use the presence (‘capture’) of peoplein different data sources as the basis for estimation of the population size. Where further characteristics are also available, these can be used to provide estimates of the population size classified by these characteristics. We consider the situation that there are measurement errors in these classifying variables, but not in the linkage of people between data sources. We consider strategies to produce estimates of the population size and breakdown using a consistent, adjusted definition taking account of all the evidence in the collected data sources