Date of Graduation


Document Type



The present study examined the effectiveness of a preschool peer mediator to facilitate generalization of social behavior change. Previous research has shown peers to be effective mediators in treatment programs, however, their efficacy as generalization facilitators is less clear. Preschool children observed to play aggressively with toys and speak in an aggressive manner were selected as subjects. The children were divided into two groups of two. The peer tutor did not engage in these aggressive behaviors. Peer tutor training consisted of discussion, role play, modeling, praise, and physical affection. A peer tutoring procedure in which the peer tutor delivered instructions, withdrew attention, praised, and gave stickers was used to train subjects. Teacher supervision of the peer tutor was a component of the subject training procedures, which included an intensive training phase and a maintenance training phase. Training was delivered sequentially across the two groups according to a multiple baseline design. Individual training was provided initially for one subject. Later, both children in the group participated in training followed by a reversal of experimental conditions for the group. The discriminative function of the peer tutor was assessed in the generalization or nontraining setting where his presence was held constant throughout the study. Results indicated decreases in aggressive verbalizations and aggressive use of toys following group training. In contrast, individual training did not produce such an effect. Following the reversal of conditions, the findings were less consistent. Measurement of process variables, such as praise, questions, and prompts, suggest that procedures were followed accurately by teachers and the peer tutor in the training setting and the generalization setting. Alternative explanations for the present results are discussed. These included the peer tutor having acquired a discriminative function for the subjects through training, the subjects having acquired a discriminative function for the peer tutor, the peer tutor having acquired a discriminative function independent of training, and the subjects becoming S('D)'s for each other. Process variables are also discussed as possibly being partly accountable for the findings. A critical analysis of the present study is presented; practical and research implications follow.