Estimation of current population size and distribution using symptomatic indicators

Abstract

This text seeks to learn from past experience about practical ways to estimate current population size and distribution at the municipal level using symptomatic indicators. In 1968, in order to analyze patterns of migration and settlement along the Belém-Brasília highway crossing the northern part of the state of Goiás 8 years after its completion in 1960 but before the 1970 demographic census, it became necessary to develop a new method of estimating post-censal population using available indirect indicators. It was possible to use non-demographic data available at the municipal level such as school enrollment and agricultural and livestock production. 12 indicators were ranked according to their correlations with municipal population size in previous years, when the different kinds of data were available. It was found that the best estimates could be made by using the 8 indicators that were most highly correlated, attributing equal weights to each indicator, but not more than 8. The error increased when including the second and third best indicators, then dropped to the lowest point as smaller errors were mutually compensated, before increasing again after 8 indicators, as errors for additional indicators grew. The inverted S-shaped curve, rising to a maximum before falling to a minimum then rising again, indicates that selected symptomatic indicators can be useful for analysis of population size and distribution in post-census years, including years for which the decennial census is delayed. Use of such a curve may also be useful for estimation of other variables at various subnational levels.