In terms of actual implementation, the two programs differed in the timing of the youth's release from placement, the intensity of implementation, and the sanctions that could be imposed. The profiles of participating offenders also varied. In the Detroit program, juvenile offenders were confined in one of the State of Michigan's training schools for an average of 17.1 months. Early release played no role in the effort. The average age at first arrest was 14.4, and the participants averaged 2.5 prior arrests. More than half of the Detroit participants were known to be drug dealers, nearly half had drug use problems, and the current offense of slightly more than half was a crime against persons. In Pittsburgh, a privately run wilderness program with an average length of stay of 10.2 months was used for this experiment. The average age at first arrest was 14, and the participants averaged 4.6 prior arrests and 3.7 adjudications. Their current offenses were mostly property crimes. The study found no difference between experimental and control groups in the proportion of youth arrested, self-reporting of offenses, or drug use during a 12-month followup period.
Equally important, youth in the experimental programs did not participate any more frequently in educational or work activities than did control group youth. Also, most of the families viewed delinquency as the youth's personal problem and were not interested in making major changes in their own behavior or activities. Further, in neither of the two sites did the aftercare program have a significant effect on the youth's associations with delinquent peers. In the Detroit program, which was characterized by longer lengths of stay and no possibility of early release, no savings were apparent in residential placement costs. Consequently, the aftercare program simply produced an overall increase in cost per placement. In Pittsburgh, where reduced time in residential placement was an explicit part of the program, total placement costs were slightly reduced.
Given the absence of any impact on the participation of the experimental group in school and work, family involvement, and delinquent peer associations, there is little reason to expect lowered recidivism. Greenwood and colleagues (1993) took the position that a number of factors explain the results, including: