Treatment Programs for Juvenile Delinquents
Juvenile crime is often serious and may represent a significant proportion of the total criminal activity in a community. It is usually assumed that adolescents deserve and require special handling because they are in a formative period and criminal behavior at this stage of life will not necessarily be continued into adulthood. Therefore, rehabilitation has particular appeal for use with juveniles. Theoretically, rehabilitation is the focus of corrections programs for juveniles. In practice, however, as occurs with adult programs, juvenile rehabilitation programs may be poorly implemented. Strengthening implementation of existing rehabilitation and delinquency prevention programs could substantially reduce future criminality.
Broad assessments of the effectiveness of delinquency treatments have greatly benefited from the rise of meta-analysis, in which researchers aggregate the continuously growing research literature to examine and compare the effect sizes (magnitude of differences between groups) for comparisons of treatment and control groups. The most extensive meta-analysis examining the effectiveness of juvenile delinquency programs was conducted by Lipsey (1992), who examined 443 different research studies.8 Lipsey's analysis focused on interventions or treatments designed to reduce, prevent, or treat delinquency or antisocial behavior problems similar to delinquency. In 64.3 percent of the studies he examined, the treatment group did better (in most cases this finding refers to a reduction in recidivism) than the control group. Considering all treatment program studies combined, 45 percent of those who received treatment were expected to recidivate, in comparison with 50 percent of the nontreated control group. In more detailed analyses, Lipsey worked to identify the characteristics that were most important in determining differences between treatment and control groups. The more effective programs were predicted to reduce recidivism substantially (once the methodology effects were controlled for). For instance, as compared with a 50--percent recidivism rate for the control group, only 32 to 38 percent of the juveniles who were given employment and multimodal or behavioral programs were estimated to recidivate.
Overall, the results of Lipsey's meta-analysis indicated that more effective programs:
There was also evidence that more effective programs targeted higher risk juveniles, but this difference was small and nonsignificant. On the other hand, treatment in public facilities, custodial institutions, and the juvenile justice system was less effective than other alternatives, suggesting that treatment provided in community settings may be more effective. If this effectiveness is the result of increased linkages with agencies and individuals in the community, then Lipsey's work supports the proposed emphasis on reintegration in the Altschuler and Armstrong model. However, it is also possible that other factors may be important. Lipsey himself cautions that the conclusion that treatment in community settings is more effective cannot be separated from the differences in the intensity (number of meetings, length of time in treatment) and needs a more refined breakdown before definite conclusions can be drawn.
The programs that were effective were those that were either provided by the researcher or implemented in treatment settings where the researcher was influential. This may indicate that treatment delivered or administered by the researcher was better implemented than typical programs, supporting Altschuler and Armstrong's point that poor implementation of a sound theoretical model is unlikely to produce a positive outcome.
8 This was a more extensive analysis than previous meta-analyses, which had focused on delinquents in residential programs (Garrett, 1985) and treatment of adjudicated delinquents (Gottschalk et al., 1987; Whitehead and Lab, 1989). Although the conclusions from these analyses differed, all yielded a positive mean effect of about the same order of magnitude (one-fourth to one-third of a standard deviation superiority for the treatment group outcome compared with the control group outcome). See also the early discussion of the Andrews et al. (1990) meta-analysis in this Bulletin.