Abstract

Residential segregation is part of the causal path of different health outcomes and is studied mostly linked to race. In 1988, Massey and Denton proposed 5 theoretical dimensions of residential segregation: unevenness, exposure, clustering, concentration, and centralization, which they confirmed using indices from USA race proportions. These indices are widely used, however, in Chile segregation is determined by socioeconomic and educational levels. This study aims to evaluate whether the dimensions are maintained when using different segregation variables and populations, such as educational level in Chile. We replicated Massey and Denton’s evaluation by constructing 20 educational segregation indexes for 247 Chilean municipalities. We constructed the indices correlation matrix and carried out a principal axes factor analysis and compared our results with those authors. There is no high intercorrelation between the indices associated with the 5 theoretical dimensions except for the unevenness dimension and, partially, exposure. This is reaffirmed by the exploratory factor analysis, in which it is not possible to empirically verify the same dimensions. This study demonstrates the need to locally characterize not only the indices of residential segregation but the theoretical dimensions to which they refer in order to improve the interpretations of the effects of segregation on health.