Summary: A training model in the use of role play by community faculty was developed, evaluated, and tested to determine its effectiveness. The study was conducted in an Improvement of College Instruction Class at Florida Atlantic University Fall term of 1983. Prior to training, eighteen students were asked to prepare role plays which were videotaped and evaluated for the thirteen skills required for an effective role play, as established in the training model. Evaluations were made by students who prepared and observed the activities, and instructors who designed the training model. In four subsequent classes the students were instructed in the use of role play through the training model. They were assigned to prepare role plays for the sixth class which were again videotaped and evaluated. Scores from the pre- and post-evaluation instruments were analyzed to determine the effectiveness of the training model. Four hypotheses were established. A t-test measured hypotheses one through three to determine if significant differences existed between pre- and post-training scores as measured by three evaluation instruments (Self, Peer, and Instructor). The three hypotheses were rejected, using an .05 level of significance. The training model appeared to significantly improve an instructor's ability to prepare and deliver a role play. A Pearson product-moment correlation was used to test hypothesis four, the correlation of self and peer post-training scores. The study failed to reject hypothesis four, using an .05 level of significance. They were not predictors of each other's ability to conduct a role play. Four questions addressing the relationships among demographic variables were also answered. Prior teaching experience and the higher the degree held appeared to have a positive effect on the use of the training model. Teaching level supported the advantage of previous teaching experience, and subject area provided no information due to the limited number of students in, Thesis (Educat.D.)--Florida Atlantic University, 1985.