Premium
A Bayesian Model for Estimating Census Undercount, Taking Emigration Data from Foreign Censuses *
Author(s) -
Redfern Philip
Publication year - 2001
Publication title -
international statistical review
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.051
H-Index - 54
eISSN - 1751-5823
pISSN - 0306-7734
DOI - 10.1111/j.1751-5823.2001.tb00459.x
Subject(s) - census , emigration , net migration rate , geography , statistics , population , demography , demographic analysis , bayesian probability , econometrics , mathematics , population growth , sociology , archaeology
Summary The new method of measuring net undercount in a census of population is a variant of demographic analysis and is founded in the belief that the most reliable counts of people are the counts of births and deaths. It has been applied to the 1991 census of England and Wales (E&W). The method has two distinctive features. The first is that it bases its estimates of numbers of emigrants on figures of persons born here and recorded as residents in the 1990–91 censuses of other countries. The corresponding data on immigrants are taken from our own census. The method does not therefore require the data on migratory flows that are an essential component of conventional demographic analysis. The second feature is the Bayesian approach in which (1) each of 30 uncertain elements in the calculations is given an a priori error distribution and (2) three empirical constraints are imposed on the sex‐age profile of percentage net undercount. This, in conjunction with a Monte Carlo process, generates an error distribution for net undercount. These merits of the method are offset by the demerit that the calculations must await results from the censuses of the other countries in which substantial numbers of our emigrants reside.