The influence of peer and educational variables on arrest status among at-risk males

In this article, we examine the predictive power of selected social and academic variables regarding arrest frequency for 11th-grade boys who 7 years earlier had been judged to be at risk for developing antisocial behavior patterns. Multiple regression analyses indicated that (a) measures of academic and social skills that reflected more skilled and positive profiles were associated with participants' exhibiting less frequent and less severe forms of antisocial behavior, as shown by arrest records at 11th grade, and (b) antisocial measures on which participants scored higher were associated with more frequent arrests for more serious offenses in Grade I I. We also conducted an analysis of the residuals to study those persons and their characteristics who were misclassified by the predictive model. Implications of these results for future research and intervention, particularly within the context of schooling, are discussed.