Multivariate identification of delinquent behavior in males

Three hundred and thirty-seven males (ages 11 to 18 were surveyed for individual, school, and family functioning, socioeconomic status, self-reported delinquent behavior, and official legal status. Subjects were divided by levels of frequency, variety, and seriousness of reported delinquent behavior. Results indicated that several individual, school, and, family variables have statistically significant, but low to moderate level correlations, with age of first delinquent behavior showing the highest correlation with each categorization method. Variables that had significant correlations with frequency, variety, and seriousness categorizations of delinquent behavior were entered into stepwise multiple regression equation. In combination, as a multivariate model, 15 variables were able to account for 36 to 45 percent of the variance in delinquent behavior. Age of onset of delinquent behavior was entered first for all three categorizations, but other variables entered in different orders, depending on categorization method. Discriminant analysis, applying a function developed on older adolescents, to younger adolescents revealed that there are at least three groups useful for identification research; a "chronic' delinquent group, who begin delinquency early and exhibit frequent, indiscriminant, and serious delinquent behavior, a group of "transient" delinquents, who begin delinquency around age 15, and appear to be involved in social experimentation, and a group of nondelinquents who exhibit little or no delinquent behaviors. Differences between the three groups on measured variables are discussed. Additionally, issues related to methods of categorizing and measuring delinquent behavior are discussed, with particular attention to the efficacy of multivariate models for identification of delinquents and predelinquents. Implications for research, prevention, and policy are presented.