Methodological Issues in Delinquency Research: Some Alternative Analyses of Geographically Distributed Data

Abstract

To assess the substantive importance of methodological issues identified in a series of critiques of delinquency area studies, a carefully focused reanalysis was undertaken using data analyzed in an earlier study. The modification of factor analytic procedures to avoid mistakes identified in these critiques produced a clearer picture of the factors reflected in the variables used to describe 100 Indianapolis census tracts but also necessitated alternative treatments of census tract delinquency rates. An examination of results produced by these treatments, which included (1) a modified social area analysis, (2) a modified predictive attribute analysis, and (3) an analysis of covariance, clearly indicates that correction of the methodological shortcomings requires little or no modification of the substantive conclusions about the relative importance of economic, family, and racial factors presented by studies published prior to 1967. Studies of the residential distribution of children charged with delinquency in urban areas have been criticized on methodological grounds almost from their inception.1 In recent years, the most impressive methodological critiques of such studies appeared in 1967 when Hirschi and Selvin published their appraisal of analytic methods used in delinquency research, Gordon assessed studies by Lander (1954), Bordua (1958-59), and Chilton (1964), and Wilk's review of ecological studies of the correlates of crime and delinquency was published by the National Crime Commission.2 These critiques either identified or reemphasized a number of problems in correlational studies of geographically distributed dataproblems they did not always resolve. An examination of the nature of a few of these critiques indicates why this was sometimes the