The impact of real relationship and working alliance on therapy outcome: a study with an Italian sample

According to Greenson (1965), Gelso & Hayes (1998) developed a tripartite model of the therapeutic relationship, including a working alliance (WA), a transference-countertransference configuration, and a real relationship (RR). Up to date, only two studies (conducted in the US) investigated the associations between the real relationship and the working alliance in psychotherapy. AIMS: This study aims to explore the association between the RR and the WA both at the beginning and late phases of therapy, and to investigate whether the real relationship predicted therapy outcome beyond the predictive power of working alliance. METHOD: Sixty-five consecutive clients at the University Counseling Center of Palermo (Italy) were included in the present study. Of the 65, 50 clients remained in treatment to complete the session packets. Clients consisted of 65% women & 35% men, attending the University courses, and ranging in age from 19-24 years (M= 22.3 , SD = 3.12). Four counselors (all women with a psychodynamic orientation, and a mean age of 36 years) provided counseling to the participants: each counselor treated on average 12 clients (range from 9 to 17). Lenght of counseling was an averaged 5.6 months, with 8.6 (SD= 4.5) sessions per client. Outcome measures: Outcome Questionnaire-45 (OQ-45, Lambert et al., 1996) ; Client Satisfaction Questionnaire-8 (Larson et al., 1979). Process measures: Real relationship Inventory – Therapist (RRI-T, Gelso et al., 2005), and Client Form (RRI-C, Kelley et al., 2004); Working Alliance Inventory-Short- Therapist (T-WAI), and Client Forms (C-WAI ; Tracey & Kokotovic, 1989). The outcome measures were administered to clients at the intake and at the end of treatment. Process measures (both client & therapist form) were filled out at the 3th and 8th session. RESULTS: Results showed that clients improved at the end of treatment, with a reduction on the OQ-45 scores of 26.8 points (p < .01). The initial symptom severity of clients was not associated with the early RR and WA scores. Two regression equations were intended to examine the role of the working alliance and real relationship in predicting outcome (OQ-45). RRI scores added to the prediction of outcome resulted in a significant increment in R squared, (Adjusted R2 from .09 to .38; CL change .029), revealing that client RR accounts for a significant amount of variance in post-symptom report, above and beyond the variance accounted for by pre-symptoms client and WA. Moreover, we analysed the differences between early (3th session) and late (8th session) perception of both the RR and the WA in clients and therapists. Clients reported no differences in both the RRI and WAI overall scores. Therapists showed significant increasing values in the WAI-Bond (t=-2.030; p = .05) and in the two RRI scales (t = -2,676; p = .01).