The effects of instructional variations during a role playing assessment on the assertive behavior of untrained, low assertive college students were investigated. Based on the responses of 1,392 students, 80 students who indicated a moderate to severe problem in refusing requests, and expressed interest in an assertion training workshop, were randomly assigned to one of eight assessment groups in a 2x2x2 factorial design. The instructions varied on three factors--knowledge of type of assertive behavior under study x knowledge of criteria for effective behavior x instructions on how to act (normally vs. effectively). Results indicated that Factor 1 affected pulse and self-reported anxiety; Factor 2 affected response latency and duration; and Factor 3 affected a posttest self-report measure of refusal behavior. Content was not affected. Implications for subject selection, training procedures, and the assessment of effects of training are discussed. (Author)