Methodology of under-coverage estimation used in Estonian PHC2011

Ene-Margit Tiit, University of Tartu
Mare Vähi, University of Tartu

Census population nowadays usually underestimates the real population size due to people’s mobility and also their privacy considerations. When all people of the population have ID-numbers, then overestimation (double-counting) is avoided and as a result the under-coverage is the main problem of the census population size. In Estonia during the recent decade a big amount of administrative register has been created: besides Population register, Taxation register, Buildings Register and Cadastre there are also different social registers, Data base of Educational Information etc. Hence each permanent resident of Estonia has its data in more than ten different register. All personal data in registers are identified by ID-codes, so it is quite easy to match the data from different register. In Estonia the main population problem is emigration, whereby not all people do register their departure from the country. Hence the Population register is somewhat over-covered containing about 4,6% more residents compared with the census population. So the task was to divide the people who were in Population Register as residents, but were not enumerated, into two parts – the residents who avoided the census and the people who had emigrated, but were not registered their departure. We used for solving the task discriminant analysis (linear and logistic) and as a result were able to assess both population groups – residents and non-residents. Hence we found that the under-coverage of Estonian Population and Household census 2011 was 2,2%.