The Relative Efficacy of Comparatively Weak and Strong Reinforcement Contingencies During Self-Evaluation
The problem. To assess the effectiveness of comparat1vely weak and strong reinforcement contingencies during self-evaluation training in producing maintenance of room cleaning behavior of youth in a shelter environment was the problem. Procedure. Youth were trained to evaluate their own rooms and to determine the number of room points they would earn in a token economy based on that evaluation. Training procedures included, modeling, feedback following practice evaluation, and either token reinforcement for accurate evaluation or monetary reinforcement for accuracy. After training to criterion youth were allowed to assess their own room and reinforcement was based on this evaluation. The effect of varying the conditions under which training occurred was measured by monitoring performance when all of the obtrusive contingencies were withdrawn. Findings. In three successive maintenance conditions dramatic reductions in the percentages of items cleaned according to the defined criteria occurred across seven of the eight subjects. Percentages of items cleaned appropriately were generally high for the group during all of the conditions in which token reinforcement was present. This included the two baseline/tokens conditions as well as the two self-evaluation conditions. Conclusions. Token reinforcement in the form of points that could be exchanged for reinforcers consistently produced high cleaning percentages whether the condition existed alone or was combined with additional reinforcers for accurate self-evaluation. Self-evaluation did not significantly increase the youths cleaning behavior above the rates obtained during the baseline/tokens phase. Self-evaluation added desirable features to the token economy but did not contribute to any long term maintenance of the behavior when removed. Recommendations. Researchers should continue to examine self-control strategies as potential techniques that may lead to generalization of treatment effects. Research studies may be most profitable if they experimentally verify the variables that operate when generalization is achieved.
34 leaves. Advisor: W. Scott Wood.