Measuring Residential Segregation: An Application of Trend Surface Analysis

Abstract

RACIAL RESIDENTIAL SEGREGATION has been characterized by Taeuber as pervasive aspect of American urban life.' Recent studies have found that residential segregation by race does not vary with, and is more extensive than, segregation by income, occupation, or educational attainment. Thus, the study of residential segregation is an important link, both as cause and effect, in an understanding of many urban problems.2 For instance, a major implication of residential segregation is that it significantly lowers the probability of Afro-American home ownership or purchase relative to otherwise comparable whites.3 An intuitive notion of segregation is the systematic location of a phenomenon. Writing in 1947,4 Jahn, Schmid, and Schrag set forth what has become the basis for the characteristics a segregation index should have. A segregation index should be: (1) expressed as a single quantitative value so as to facilitate such statistical procedures as comparison, classification, and correlation; (2) relatively easy to compute;