Accounting for Demography and Preferences: New Estimates of Residential Segregation with Minimum Segregation Measures

The index of dissimilarity (D) has historically been and continues to be a widely used quantitative measure of residential segregation. Conventional interpretations of D imply that normatively desirable residential patterns occur when ethnoracial compositions of lower-order geographic units (such as neighborhoods) match those of higher-order units (such as metropolitan areas). However, it is likely that average preferences for same-group contact in neighborhoods sometimes exceed group population shares in metropolitan areas. In such situations, there is mathematical tension between the capacity for group preferences for co-ethnic neighbors to be satisfied and the degree of residential segregation. In this article, I quantify this tension by calculating DΔ, or the difference between D and the minimum segregation measure D*, which returns the lower bound on segregation for a given average in-group preference level and ethnoracial share. Positive scores on DΔ indicate that a metropolitan area is more segregated than necessary to satisfy average group preferences, while negative scores indicate that observed residential patterns do not satisfy such preferences. I use data from the 2010 decennial census and 2006–2010 American Community Survey to analyze the associations between predictors of residential segregation and DΔ