Data were based on arrest and conviction records. Descriptive variables were evaluated regarding their association with probation outcomes at 6- and 12-month followup intervals. Stepwise multiple regression selected a subset of variables significantly related to probation outcomes. This subset was weighted to compute base expectancy scores which reflected expected outcomes on probation adjustment, then actual and expected outcomes were compared. The subset of predictor variables was used as covariates to determine 'treatment impact' after adjusting for initial group differences. Results indicated a high degree of success for both groups, but no significant group differences were found. A longer followup period might show more group differences. Characteristics of program participants may be the determining factor for success rather than any specific training program. (Author abstract)