PROBLEMThe recent and accelerating use of paraprofessional psychotherapists has given rise to increased concern over the training of such 6 , 7 ) . A wide variety of training procedures, both traditional and innovative, are currently being implemented though only very rarely with accompanying evaluation of their efficacy.Earlier efforts''. 4 . l o ) towards these training ends sought to be responsive to laboratory research which very strongly demonstrates the efficacy of modeling procedures for bringing about a diverse array of behavior change. Working with a potential paraprofessional therapist sample of psychiatric nurses, Goldstein, el d ( 5 )found that modeling procedures, when combined with information about the patient's purported social class, yielded a significant increase in nurse attraction to traditionally unattractive patients, as well as similarly significant increases in the level of empathy and warmth displayed by the nurses immediately following the observation of an attracted model. I n response to J o u r a r d )~(~) findings that self-disclosure by one member of a dyad increases the self-disclosure level of the other member, we next sought to increase the level of paraprofessional therapist self-disclosure by means of modeling procedures (ll). Once again, this time utilizing a psychiatric attendant sample, exposure to a high self-disclosing model yielded the predicted effect. The findings of Carkhuff ( 2 ) , Truax and Carkhuff ( 1 4 ) , Hart and Tomlinson(8) and others led us t o empathy as a third training target. Their investigative findings consistently point to therapist empathy as a potent psychotherapeutic ingredient. Both of our subsequent l 3 ) seeking t o enhance the level of empathy displayed, by use of modeling procedures, yielded significant increases in psychiatric attendant and minister samples.I n spite of the consistency of these beginning demonstrations of the efficacy of modeling for paraprofessional skill training purposes, a highly important proviso must be considered. One of the most frequently noted outcomes of attitude and behavior change research is that most research SS (or therapy patients) will do almost anything asked of them by the experimenter (or therapist) while in the experimental setting, but a distressingly few will be noted to be sustaining the behavior change out of the laboratory (or consulting room) context. This may mean our manipulations are weak; that Ss very often respond as we wish (predict) because of experimenter bias, evaluation apprehension or experimental demand characteristics; or simply, that we have given grossly insufficient attention to implementing transfer training principles. For whatever reason, follow-up results most often yield evidence of non-enduring effects, and such proved to be the case in our studies noted above aimed at self-disclosure or empathy-enhancement.Our typical paradigm involved pre-exposure subject measurement on the particular skill, exposure for approximately one hour to a model displaying high levels of the skill, ...