Abstract
The efficacy of social areas as depictions of urban residential structure requires within-social-area homogeneous distributions of population and housing. Using both personal and neighborhood data from the 1970 Public Use Sample of the United States Census, direct measurements of homogeneity and correlation patterns for persons and neighborhoods yield a picture of residential heterogeneity for the cities of St. Louis and Kansas City, Missouri, and Minneapolis-St. Paul, Minnesota. Results indicate that common social area depictions of residential structure grossly over-simplify the actual pattern. Although some confirmation of the sectoral hypothesis is implied, evidence indicates that concentric zonation contains little of the total residential variation and covariation.