The working alliance predicts session-by-session change in problems and functioning in short-term psychodynamic psychotherapy.

Abstract

The working alliance is hypothesized to be crucial for facilitating change during psychodynamic therapy. However, few studies have examined the session-by-session effects of the alliance on outcomes within this therapeutic approach. We aimed to test the effect of the client- and therapist-reported alliance on next-session problems/functioning in short-term psychodynamic therapy conducted in a university training clinic. A secondary aim was to test whether baseline personality functioning and adverse childhood experiences moderate the association between alliance and outcome. A total of 152 clients, receiving short-term psychodynamic psychotherapy from novice therapists, completed the Clinical Outcomes in Routine Evaluation-10 before each session. After each session, clients and therapists completed the Working Alliance Inventory-Short. Data were analyzed using continuous time structural equation modeling, which allows for relatively strong causal inferences despite the nonexperimental design. Improvement in the client-reported alliance significantly predicted a reduction of problems/dysfunction in the following session with a small effect size that increased to a large effect size after 7-8 sessions. The effect of problems/functioning on the client-reported alliance was also statistically significant. No associations were found between the therapist-reported alliance and outcome. The alliance effect varied considerably between therapist-client dyads and was stronger for clients reporting more adverse childhood experiences. In conclusion, the client-reported alliance was a predictor of outcome in the following session. Its effect on outcome peaked after 7-8 sessions. While this long-term effect aligns with the principles of psychodynamic therapy, it may also stem from the statistical model used. (PsycInfo Database Record (c) 2026 APA, all rights reserved).