In two studies, clients with alcohol problems were randomly assigned to counselors who varied in their level of skill in client-centered counseling. One study involved nine counselors, all of whom were delivering the same manual-guided behavior therapy for problem drinkers (Miller, Taylor & West, 1980). Before collecting outcome data, their supervisors independently rank-ordered the nine counselors on their skill in accurate empathy, based on observation (via one-way mirror) of treatment sessions. Inter-rater agreement was high, and when follow-up data became available, client outcomes were examined.

The proportion of poor outcomes varied from zero (for Counselors 1 and 3) to 75 percent (Counselor 9). As is apparent, although the clients were all ostensibly receiving the same behavior therapy, the single best predictor of their post-treatment drinking was the counselor to whom they had been randomly assigned. Specifically, the more empathic their counselor, the less clients were drinking at six, 12 and 24 months (Miller & Baca, 1983; Miller et al., 1980). A larger multi-site trial similarly found significant outcome differences across the case loads of counselors delivering standardized manual-guided treatments, despite intensive training and monitoring.

In a second study randomly assigning cases the counselors, Valle studied counselors’ level of skilfulness in providing the conditions of client-centered counseling (empathy, honesty, acceptance). Figure 2 shows relapse rates in the caseloads of counselors with low, medium and high levels of client-centered skills. Once again, a primary determinant of clients’ outcomes was the counselor to whom they had been randomly assigned. At six months the likelihood of relapse was four times higher with low relative to high-empathy counselors; and even at 24 months the ratio remained over two to one.