Delinquency Research: An Appraisal of Analytic Methods

I used the first (1967) printing of Delinquency Research: An Appraisal of Analytical Methods in an undergraduate sociology research methods course. The book was and remains interesting, accessible, and timely. The procedures presented by Hirschi and Selvin for making causal inferences with non-experimental quantitative data are as useful today as when the book first appeared. The fact that the authors were able to present this information without overwhelming the reader with sophisticated and difficult-to-understand statistical methods is a tribute to the thoughtful organization of the volume, as well as to the readable prose produced by the authors. A good part of the book's interest inheres in the author's decision to write much of it in response to flawed research done on delinquency by others, especially the once-famous and influential Gluecks. As the authors' presentation develops, it becomes abundantly clear that the Gluecks' book Unraveling Juvenile Delinquency is sadly mis-titled. The primary reason for this is undue emphasis on bivariate relationships presented in tabular form. Hirschi and Selvin make clear that failure to introduce controls renders the Gluecks' interpretation of their data dubious, at best, and often simply wrong. The Gluecks' approach fails to address the ubiquitous issue of confounding. I know of no recently published text that is as instructive and accessible for the beginning causal analyst as Delinquency Research. The first one hundred forty-one pages are especially good, a book in themselves. The material in subsequent chapters is also accessible, useful, and interesting, but not essential to understanding Hirschi and Selvin's causal framework. If a recent text of comparable quality were published, I'd definitely use it in my introductory course in quantitative research methods for non-experimental data. As it is, I still use the 1967 printing as a means of prompting myself to come up with useful examples of the consequences of failing to address the issue of confounding and the payoffs which come with taking confounding seriously in causal analyses. After working through Delinquency Research, the reader will never again look at two-variable relationships with an uncritical eye.