| Abstract: | Indirect estimation techniques for measuring emigration were first proposed by members of the IUSSP Working Group for the Study of International Migration, as a complement to methods based on more traditional types of data such as place of birth, place and duration of residence, and date of arrival. 2 approaches were suggested: 1, based on obtaining information from mothers on the place of residence of their children, was proposed by Somoza; the other, which requires all respondents to furnish information about the place of residence of their brothers and sisters, was developed by Hill. The aim of both methods is to obtain an estimate of the total number, and distribution by age and sex, of persons born in 1 country, but residing in another. The residence of children method has now been tried in at least 10 different censuses and surveys, mainly in Latin America and the Caribbean, and the analysis of these data allows assessment of the methods' strengths and weaknesses with some confidence. The residence of sibling technique has not been used as widely, so that, in spite of its theoretical attractions, its performance in the field cannot be evaluated to the same extent. This article seeks to summarize the methodology and to examine some recent theoretical developments. In particular, an assessment is presented of biases in the estimates of emigration obtained by the indirect methods. 3 main sources of error have been identified: reporting errors arising during fieldwork, estimation errors arising in the course of analysis through the use of biased inputs, and errors due to the use of age distributions of relatives derived from model stable populations. |
