The Michigan State Diversion Project is a behavioral treatment program for arrested juveniles that uses college students as the principal caseworkers. The program is based on three recurring themes in research and program experience with juvenile offenders: 1) youths respond better if treated outside the juvenile justice system, 2) a youth’s community and family are the natural context for intervention, 3) and service delivery by nonprofessionals may be both more effective and less costly than relying on credentialed professionals. The caseworkers, mainly juniors and seniors, are given 8 weeks of training in behavioral intervention and advocacy followed by 18 weeks of intensive supervision. During the 18-week intervention the caseworkers spend 6 to 8 hours per week with the juvenile in their home, school, and community.

The goal of this program is to provide effective and affordable treatment for juvenile offenders.

The program was evaluated using an experimental design. The evaluation found that the active strategies that occurred outside the juvenile system (action and family focus), along with the relational strategy, tended to work better than those used for the placebo group. Moreover, the placebo group worked better than the control group and the court context group, but the court context group did worse than the control group. For example, 67 percent of the court context youth and 62 percent of the control group had one or more court petitions during the 2 years following the intervention, compared with 38 percent of the action group, 46 percent of the family focus group, and 33 percent of the relational group. The placebo group fell in the middle (52 percent).

Thus, the evidence suggests that active, hands-on intervention of several kinds works better than normal court processing of juvenile offenders, but only if they were thoroughly separated from the system. The researchers note that these findings should be interpreted with caution because the samples were small, and no significant effects were found for measures of self-reported delinquency.