Abstract
This study examines the factors associated with successful completion of the Orange County Juvenile Drug Court (OCJDC) for 232 (164 male and 68 female) adolescents between the ages of 12 and 17 years. No significant associations were found between graduation status and participants’ age, most serious prior offense, entry offense, primary drug of choice, or secondary drug of choice. While bivariate analyses revealed that Latino ethnicity (especially for females whose parents’ primary language was Spanish) and a lack of family involvement in the drug court programs were both negatively associated with successful program completion, these variables were not significantly associated with the likelihood of graduation in logistic regression models predictive of successful program completion. Rather, a shorter history of substance abuse, a history of prior juvenile offenses, and the presence of co-occurring psychiatric disorders all predicted poor completion rates from the OCJDC program, regardless of gender, ethnicity, or levels of family participation. Implications for policy and practices are discussed.
We thank the Juvenile Drug Court of Orange County, California, for their participation in this evaluation project. The court and the Orange County Department of Probation have authorized the use of their data for this study. We also thank Elizabeth H. Armstrong, MSW, for her help with the collection of the data used in this evaluation project. Finally, the authors lovingly acknowledge the late Dr. Elizabeth Deschenes who planned this study and assisted with its execution while courageously battling cancer. The authors of this manuscript are indebted to Dr. Deschenes for her contributions to this project, the field, and our own professional development as a result of her mentoring.
Notes
1. The number of days spent in the program variable was excluded from the multivariate analysis due to its overwhelmingly strong association with graduation status, such that it could practically serve as a proxy for successful graduation.
2. Another logistic regression model was run in order to assess best fit. This model only included the independent variables from that had p values less than .10. However, the chi-square value for the omnibus test of model coefficients, the –2 log-likelihood value, the Nagelkerke Pseudo R 2 value, and the Hosmer and Lemeshow goodness of fit test did not show improvement. Therefore, the model presented in was chosen as the final model.