Using Uniform Association Models - An Example From Delinquency Research (Quantitative Criminology - Innovations and Applications, P 115-128, 1982, John Hagan, ed. - See NCJ-88809)

• NCJ Number: 88814
• Author(s): M Hout
• Date Published: 1982
• Length: 14 pages
• Sale Source: Sage Publications, Inc
• Publication Type: Survey
• Language: English
• Country: United States of America

Annotation

This study uses a pairwise uniform association model to analyze data from a 3x3x3 cross classification of self-reported delinquent acts, number of delinquent friends, and attitudes toward breaking the law to demonstrate the benefits of this method over the general log-linear model.

Abstract

The data are from a 1972 Richmond Youth Study (California) and are only for nonblack males. The author initially analyzes the two-way association of friends and delinquent acts with the general log-linear method as well as with the linear-by-linear and uniform association models. He then focuses on the three-way association, obtaining parameter estimates from the process developed by Haberman called FREQ. The uniform association analysis shows that not only are the self-reported delinquent acts of white youths studied related to the number of their delinquent friends and the favorableness of their attitudes toward breaking the law, but also that these relationships are uniform. Thus, the odds of delinquency increase log-linearly with each increase in delinquent friendships and prodelinquent attitudes. The uniform association model is superior because it uses parameters which are independent of marginal shifts and are interpretable in terms of causal process rather than prediction accuracy. Moreover, the uniform association model is more parsimonious than the log-linear model because it fits a single parameter to each two-way association. Thus, it can detect a small but significant association that might be missed by the general log-linear model. Tables, graphs, and 11 references are included.