Abstract
BACKGROUND: Empirical literature indicates that the therapeutic alliance explains a modest but reliable proportion of variance in predicting alcohol-related outcomes among individuals in treatment for alcohol use disorders (AUDs). Hartzler and colleagues (2011) showed in the COMBINE data set that alcohol abstinence self-efficacy is a potentially important statistical mediator of the relationship between the alliance and client outcomes. METHODS: The purpose of this study was to replicate this finding in the Project MATCH data set. We used total alliance ratings on the Working Alliance Inventory and tested both client and therapist ratings in mediation analyses. RESULTS: We found that posttreatment self-efficacy accounted for the effect of therapist and client ratings of alliance (measured at session 2) on posttreatment drinking outcomes (drinks per drinking day and alcohol-related problems). In addition, we found a moderation effect of treatment, such that the association between the client's rating of the alliance and self-efficacy changes was positive for individuals in the cognitive behavioral treatment group but negative for those receiving motivation enhancement or Twelve-Step Facilitation. CONCLUSIONS: This study reaffirms the importance of the therapeutic alliance and self-efficacy in predicting AUD outcomes. Future research should examine changes in the therapeutic alliance throughout treatment and how these changes are related to self-efficacy and AUD treatment outcomes over time.