The Importance of Race in Juvenile Commitment in the New Jersey Family Court

Abstract

In this study, we examine how important juvenile race and other factors are in juvenile commitment classification in the New Jersey Family Court. Data from the Family Court in New Jersey for the year 2010 comprise the population. Given the class imbalance in the dependent variable, we employ balanced random forest (RF). Variable importance plots and an information gain summary are used to assess the role of the juvenile’s race and other variables for classes of the dependent variable. The results from balanced RF indicate that the juvenile’s delinquency history and the offense seriousness make the most important contributions to commitment to juvenile state incarceration. The juvenile’s race makes a very weak contribution to commitment; in fact, when the balanced RF was rerun with the juvenile’s race omitted, the estimated misclassification error slightly dropped for commitments. Balanced RF is an attractive procedure for handling dependent variables with highly imbalanced classes. The juvenile’s adjudication history and offense seriousness emerged as the most important variables to state incarceration. The race of the juvenile was not an important variable with respect to commitment.