Residential Segregation in United States Cities: A Causal Analysis

Abstract

This study investigates causes of black residential segregation. The units of analysis are cities and a causal model is specified and evaluated using the technique of path analysis. The principal hypotheses are supported, although there is some interaction with regional location. The data suggest that the relative status of blacks and black population size are important determinants of segregation; these findings are especially interesting given the general belief that black socioeconomic progress is unrelated to changes in segregation, as well as the tendency to ignore absolute size of the black population in analyses of segregation. Also important are percent black and the relative growth rates of the white and black populations; however, the impact of the latter variable was smaller than anticipated in view of the emphasis often given to it. This study is concerned with the causes of residential segregation in large United States cities. The approach is comparative in that we analyze the determinants of intercity variation in residential segregation. A causal model is specified and three basic hypotheses are examined: (1) black-white status differentials affect residential segregation, (2) both the absolute and relative size of black populations have important and independent effects upon segregation, and (3) the rate at which the black population is growing relative to the white rate affects segregation. Despite the attention devoted to the causes and consequences of black segregation, there have been surprisingly few comparative studies using cities as units of analysis. Instead, explanations by social scientists have tended to treat segregation as a qualitative rather than as a continuous variable, and concern has been with specification of the conditions under which it emerges. For example, Warren (1969) argues that black segregation can best be understood as the outcome of internal colonialism, in which white domination of ghetto institutions creates a sense of helplessness and dependency on the part of ghetto residents. Moreover, in those studies where segregation is taken as a variable, it has been treated as a causal factor, and its effects on such dependent variables as assimilation, infant mortality, and fertility differentials analyzed (e.g., Duncan and Lieberson, 1959; Jiobu, 1972; Jiobu and Marshall, 1971; Lieberson, 1963; Marshall and Sinnott, 1971). A major exception is Taeuber and Taeuber (1965a) who examined factors producing changes in segregation levels during the 1940s and 1950s. Also important is Schnore and Evensen's (1966) analysis of city age and segregation levels in southern cities. THEORETICAL FRAMEWORK AND HYPOTHESES The ecological approach to residential segregation tends to conceptualize it as the spatial expression of intergroup variations in status. Park (1967:68) provides the classic statement of this view: It is because social relations are so frequently and so inevitably correlated with spatial relations; because physical distances, so frequently are, or seem to be, the indexes of social distances, that statistics have any significance whatever for sociology. In a closely reasoned statement of this hypothesis, Lieberson (1963) argues that a group's status is an important determinant of its competitive position. Low-status groups tend to be spatially isolated from higher-status groups, partly because high-status persons avoid locating their residences in the same areas, and partly because low-status groups are less able to compete for the more attractive residential sites occupied by high-status groups. Consistent with this hypothesis Lieberson (1963) shows that intergroup segregation levels in 1930 varied closely with such indicators of group status * We wish to express our appreciation to the anonymous referees who offered numerous and extremely helpful comments. Of course, we alone are responsible for the interpretations as well as any remaining ambiguities or errors.