RESUMO
OBJECTIVE: The alliance influences outcomes in CBT for youth anxiety disorders. Thus, knowledge about how therapists can enhance the alliance is needed. METHOD: Seventy-three youth with anxiety diagnoses (M age = 11.5 years, SD = 2.2; range 8 to 15 years; 47.9% boys; 90.4% white-European) participated in 10-session cognitive behavioral therapy in community clinics. Therapist alliance-building behaviors in session 2 was reliably coded with the observer-rated Adolescent Alliance-Building Behavior Scale (Revised) (AABS(R)). Alliance was measured as youth- and therapist-rated alliance, and youth-therapist alliance discrepancy in session 3. Outcomes were diagnostic recovery and anxiety symptom reduction at post-treatment and one-year follow-up, and treatment dropout. We examined the direct effects of alliance-building on alliance, alliance on outcomes, and alliance-building on outcomes in multilevel mediation models, and between- versus within-therapist variance across these effects. RESULTS: The alliance-building behaviors collaborate, present treatment model, and explore motivation positively predicted alliance, whereas actively structuring the session (i.e., dominating) negatively predicted alliance. The alliance-building behaviors attend to experience, collaborate, explore motivation, praise, and support positively predicted outcomes. The alliance-building behaviors present treatment model, express positive expectations, explore cognitions, and support negatively predicted outcomes. The effect of collaborate on symptom reduction was mediated by youth-therapist alliance discrepancy. There was almost zero between-therapist variance in alliance-building, and considerable within-therapist variance. CONCLUSION: Therapist alliance-building behaviors were directly (positively and negatively) associated with alliance and/or outcomes, with only one effect mediated by alliance. Alliance-building behaviors varied far more within therapists (i.e., across clients) than between therapists.