The Alliance-Outcome Association in Group Interventions: A Multilevel Meta-Analysis

Objective: The therapeutic alliance has been studied in group interventions, but its relationship to patient outcome is still controversial. This study systematically reviewed the association of both the member–leader and membergroup alliance with patient outcomes, and assessed the effects of several theoretically informed moderators. Method: A three-level meta-analysis was conducted on 57 studies to disentangle within and between-study variability in the alliance-outcome association. Results: The overall effect size of the unconditional model was r = −.259 (95% CI: −.302, -.214; p <.001), whereas the correlation between alliance and outcome based on the three-level meta-analytic model was r = −.215 (95% CI: −.252, −.178; p <.001) indicating a medium effect. Meta-regressions of moderators showed that the alliance-outcome correlation was higher when patients reported on their alliance with the group/other members compared with when they rated the alliance with the leader. Outcome type, rating perspective, theoretical orientation, treatment length, and year of publicationwere significant moderators of the relationship between alliance and outcome. Conclusions: The current meta-analysis highlighted that the therapeutic alliance, especially with other group members, is a significant predictor of outcomes in group therapies even after controlling for several moderators and adopting more conservative statistical modeling.