SHELDON GLUECK (Ed.). The Problem of Delinquency. Pp. xvi, 1183. Bos ton: Houghton Mifflin Company, 1959. $10.50

Abstract

-are by 106 authors, including 24 psychiatrists or psychoanalysts, 19 sociologists, 19 social workers, 11 lawyers, 9 psychologists, and 24 of other professions or not classified by profession. In this diversity almost everyone can find much interest and some treasure. My prize was Eisenstadt’s functional analysis of dDelinquent Group Formation Among Children of Immigrants.d If generalized to include internal migrants, this cogently explains most of the highest delinquency groups in American urban history, including today’s apparent leaders-urban-reared children of Southern-rural and PuertoRican-reared parents. Methodological deficiencies which have been apparent in some of the Gluecks’ other writings persist in the contributions here. Reliability tests for subjective data classifications are not reported, nor are classification procedures specified sufficiently to permit replication. Antecedent probability in validations of prediction tables is neglected. For instance, a table which predicts delinquency in most institutionalized delinquents might also predict it in nondelinquents, and even if it identifies 90 per cent of nondelinquents accurately, it will still, by Bayes’ theorem, make more erroneous than correct predictions of delinquency in most schools where children are less than 10 per cent delinquent. Methodological critics, such as Meehl and Rosen in psychology and Reiss and Kuhlen in sociology, are also ignored. The Glueck and Podolsky articles stress biological correlates of delinquency despite weak evidence, but neglect more markedly evident social correlates in Glueck data. The firmest biological finding, essentially, is that in high delinquency areas, average institutionalized delinquents are somewhat huskier than nondelinquents. This may reflect mainly selectivity by delinquent groups, and Detroit evidence suggests that police arrest the huskiest first. Glueck dismisses differential association theory, preferring lists of zero-order correlates of delinquency, which he calls dmultifactor theory.d Such lists achieve few of the things which Phillip Frank’s Philosophy of Science pointed out as distinguishing scientific theory: the quality of being simpler than the observations it represents, and the quality of promoting expansion of observations into unknown ter-