Interventions for Noninstitutionalized Juveniles
The database was divided into studies of interventions with noninstitutionalized juveniles and studies of interventions with institutionalized juveniles because the circumstances of treatment are different and because the nature and response of the juveniles receiving treatment may differ. This section examines the effects of noninstitutional treatment using the method-adjusted effect size values (discussed previously) in relationship to four clusters of variables. These clusters, which were associated with more than half of the variation among effect sizes across the studies, are listed in decreasing order of magnitude:
This model was further reduced to include only the variables in each category that were most closely related to intervention effects on recidivism among noninstitutionalized serious juvenile offenders. Intervention effectiveness was associated with the characteristics of the juveniles who received treatment. The effects were larger for more serious offenders (indicated by the types of prior offenses that included both person and property offenses) than for less serious offenders. Type of treatment was important and is discussed in the next section. Longer treatment was positively associated with effectiveness, whereas the mean number of hours per week was negatively correlated due to the small effects realized for low-intensity programs that operate continuously or meet frequently, such as wilderness/challenge and group counseling programs. Among general program characteristics, only the level and nature of the researcher's participation made a significant, independent contribution to effect size. Effects were larger when the researcher was more involved in the design and delivery of treatment.
Type of Treatment and Effects on Recidivism
To compare differences in treatment, observed effect sizes (the original effect size computed in each study), equated effect sizes (the effectiveness after controlling for all common variables), and method-adjusted effect sizes (the effectiveness after controlling for differences in study methods and procedures) were examined to consider the magnitude of the mean effect, the variance around each of those means, and the extent of agreement across the three different effect size estimates. Three types of treatment showed the strongest and most consistent evidence of reducing recidivism in noninstitutionalized serious offenders:
It should be noted that there are only a small number of studies for each type of treatment because these treatments have not often been studied in reference to serious noninstitutionalized offenders. Following these treatment types in effectiveness were multiple services and restitution programs for juveniles on probation or parole.
The types of treatment that showed the clearest evidence that they were not effective included wilderness/challenge programs, early release from probation or parole (based on only two programs), deterrence programs, and vocational programs.
One group of treatment types, including employment-related programs; academic programs; and advocacy/social casework, group counseling, and family counseling programs, presented mixed or ambiguous evidence. This group showed inconsistent effect size estimates. On the other hand, their equated effect sizes (which account for method and procedure, juvenile characteristics, and other differences) were favorable. However, without understanding the variables contributing to these differences, it is difficult to know whether the effectiveness calculated for this group of treatments was due to the treatment or to another variable.