The Measurement of Spatial Segregation

Abstract

The index of dissimilarity has come to be the principal statistic for measuring segregation, particularly urban residential segregation by race. Recently, though, a literature has arisen which criticizes the dissimilarity index and proposes revisions or alternative statistics. Here a statistic is derived that explicitly incorporates the spatial relationships among the geographic parcels into the tabulation, a feature absent from the dissimilarity index and its competitors. This proximity statistic is compared with other indices and is found to be somewhat successful in distinguishing between single-cluster and multiple-cluster residential settlement patterns.