Ecological Fallacies and the Analysis of Areal Census Data

Publisher Website
Google Scholar
Abstract
In many countries census data are only reported for areal units and not at the individual level. This custom raises the spectre of ecological fallacy problems. In this paper, a 10% sample census (from the United Kingdom) and individual census data (from Italy) are used to provide an empirical demonstration of the nature and magnitude of these problems. It is concluded that ecological fallacy effects are endemic to areal census data, although their magnitude is perhaps not as large as might have been expected. The principal difficulty is that there is at present no way of predicting in advance the degree of severity likely to be associated with particular variables and particular techniques. Finally, a suggestion is made concerning how the potentially serious practical consequences can be reduced.

This publication has 10 references indexed in Scilit: