| Sumario: | The article presents the author's response to an article by Ronald Roesch and Raymond R. Corrado criticizing his work on the effects of criminal recidivism of rehabilitation and diversion services. Roesch and Corrado state that the sample of 2,860 male clients may be completely unrepresentative of the eighteen projects, the fifty-three projects, and especially the criminal justice system as a whole. They argue that the possibility of unrepresentativeness follows from two factors. First, the sample was nonrandom because neither random sampling nor representative sampling techniques were employed. Second, the 2,860 clients in the sample represented less than 10 percent of possible clients. Roesch and Corrado's use of the term nonrandom sample may be misleading and is not pertinent to the question of the representativeness of the sample. Roesch and Corrado's reservations about the representativeness of the sample appear to reduce to how much confidence there can be, in the absence of statistical demonstration, that the sample is not unrepresentative. Roesch and Corrado accept the use of arrest as measure of outcome, but criticize the failure to use outcome measures of noncriminal behavior as well, stating that they were just as important and also may have been affected by project services. But one cannot use outcome measures of noncriminal behavior to assess the effectiveness of these projects because to do so assumes or implies that these measures are dependent variables. Roesch and Corrado argue that the design of the Evaluation is fundamentally flawed because limitations of the control group used. They also suggest that the most important shortcoming of the design was that there was no control group composed of clients who were continued through the usual criminal justice system and were exposed to court processing, probation or incarceration.
|
|---|