Congruence and Discrepancy Between Working Alliance and Real Relationship: Variance Decomposition and Response Surface Analyses
We examined how congruence and discrepancy in clients’ and therapists’ ratings of the working alliance (WA) and real relationship (RR) were related to client-rated session quality (SES; Session Evaluation Scale). Ratings for 2517 sessions of 144 clients and 23 therapists were partitioned into therapist-level, client-level, and session-level components and then analyzed using multilevel, polynomial regression and response surface analysis. For both clients and therapists, at all levels of analysis (except the therapist level for therapist ratings), SES was highest when combined WA and RR ratings were high, and lowest when combined ratings were low. For client ratings, discrepancy between WA and RR, at the client and session levels, was associated with greater session quality. Some clients perceived greater session quality when, across all sessions, WA was stronger than RR and other clients perceived greater session quality when RR was stronger than WA. Within clients, session quality was highest when some sessions had a stronger WA than RR whereas other sessions had a stronger RR than WA. These findings are compatible with a responsiveness framework, therapists varied the balance of WA and RR to suit situational demands or needs of different clients. When therapists rated WA and RR the opposite pattern of results emerged; clients perceived greater session quality when therapists’ WA and RR ratings, for a session were high and consistent (i.e., no discrepancy between WA and RR). In addition, across all sessions, clients perceived greater session quality when WA and RR ratings were high and consistent.