Objective: Although without a doubt alliance has a consistent impact on outcome, to date it remains unclear whether alliance directly accounts for symptom change.
Method: Using data from a randomized clinical trial, the alliance-outcome timeline was analyzed. The Relationship factor of the Helping Alliance Questionnaire I (measured twice, middle and late during treatment) was used.
Results: Regression and mixed model analyses showed that alliance did not predict subsequent symptom change as measured by the Hamilton Depression Rating Scale beyond prior symptom change and the alliance course was not predicted by early symptom change.
Conclusions: These results lead to the question whether alliance directly accounts for subsequent change or moderates change via technique, transference phenomena, therapist competences and patient characteristics.