Using an ecological perspective, data integration, and longitudinal modeling as a framework for quantitative analysis in evaluation of juvenile diversion programs
This research presents a framework for quantitative analysis in evaluation of juvenile justice diversion programs using an example diversion program provided by the City of Mobile Police Department in Mobile, Alabama. The framework presented herein combines data integration from multiple sources with longitudinal analysis methods to assess ecologically-based individual-, family-, and school-level outcomes in relation to program participation in the diversion program. This study, demonstrating innovative quantitative methods for enhancing evaluation strategies, offers possibly the first ever example of a repeated-measures longitudinal evaluation design for assessing ecologically-based outcomes in the context of juvenile diversion programing. Combining youth survey data from a larger longitudinal study (the Mobile Youth Survey Project) with official record data from the Mobile County Public School System and the Mobile Police Department provided a rich source of measures used to assess program impact on eight ecologically-based outcomes. Outcome measures included individual behavior and attitude factors, family functioning factors, and school-related outcomes. Using growth curve modeling to examine group differences in developmental trajectories between program-referred youths and similar peers, significant positive effects were found for attitudes about personal violence and school related outcomes. Using discontinuous (piecewise) growth curve models to examine development for program participants prior to and after program participation, significant positive effects were also found for behavioral self-worth, attitudes about personal violence, parental monitoring, and school related outcomes. Gender differences were also found for all outcomes, except school absences. A growth curve analysis comparing program participants who received only an initial contact or less with those who received more than the initial contact indicated no significant differences in outcome measures based on level of services. Results are somewhat consistent with prior research, however, gender differences are rarely reported, and no other published diversion evaluation studies have used growth curve modeling strategies to assess program effects on change over time. Further use of these methods and the application of this framework will tremendously advance the diversion evaluation field and the understanding of what works best on what factors and for whom in juvenile diversion programming.