The Regional Interpretation of Manhattan Population and Housing Patterns through Factor Analysis

Abstract

IN VIEW of the primacy of New York City among American cities, it is perhaps surprising that relatively few geographical studies have been made which attempt to interpret the distribution of population groups and neighborhoods in the core of this urban area. One obstacle to such an interpretation may well be the very amount and complexity of the accumulated census-tract and planning-department data dealing with Manhattan alone. It is bewildering to the geographer to confront the hundreds of categories under which information is provided for each of the more than two hundred Manhattan census tracts, and to attempt to regionalize them without some simplifying statistical technique. The present article attempts to point the way to such an understanding of Manhattan, by adapting factor-analysis techniques to the purpose of descriptive simplification. Factor analysis is becoming increasingly helpful to the geographer in reducing a wide range of interrelated yet complexly distributed variables to a simpler pattern of factors, or analytically created surrogate variables. Simplicity is achieved by virtue of the fact that the variables are many in number and the factors are few. Recently Hadden and Borgatta' in an excellent study have utilized factor analysis to establish a classification scheme for American cities through the reduction of an intercorrelation matrix of 65 variables to 16 factors. They use a principal-axis procedure coupled with a varimax rotation on a universe of individuals comprising 644 cities with populations in excess of 25,000. They also discuss similar studies in detail.2 In addition to the studies cited by Hadden