To determine what happens as a result of a training experience and why certain effects of training show up in only some participants, a series of six two-hour seminars in handling problem employees was given to front line non-academic university supervisors. The Solomen Four Group Evaluation Design and questionnaires were used. Criteria changes were small in relation to training alone, but some sharp changes were associated with the completion of the questionnaire alone. More change was produced when training was preceded by questionnaire completion. Research interviewing made a strong training impact on supervisory morale, self concept, and identification with the organization. Analysis of all tests and measures showed that individuals most likely to change after training had, in descending order of predictive potency: high tolerance of ambiguity, high self-esteem, high authoritarianism, high intelligence, female sex, younger age, and less experience. Role expectations were the single best predictor of change. (Appendixes include outlines of training, copies of handouts, listening techniques, questionnaires, tables of results, and a bibliography.) (author/jf)