RESUMO
OBJECTIVE: Ruptures and repairs in alliance and their association with treatment outcome have been studied widely. Many of these studies have used indirect methods, focused on decreases in alliance across sessions while measuring alliance at postsession. However, this approach does not establish whether observed decreases occur within (as insinuated by most theories) or between sessions. In the current study, we examined decreases of alliance measured both pre- and postsession in 3 clinical trials and explored the phenomenology and interpretation of these decreases. Additionally, we investigated the effects of rupture magnitude and the interpretation of repairs on treatment outcome, examining whether the "repairs benefit" or the "unrepaired ruptures damage" theories were supported by the data. METHOD: Presession and postsession therapeutic alliance and outcome measures were examined from patients who participated in 1 of 3 studies: cognitive-behavioral therapy (CBT) for social anxiety disorder (N = 29), CBT for panic disorder (N = 31), or short-term psychodynamic psychotherapy for depression (N = 44). Patterns of change in alliance are described. Ruptures and repairs are examined according to several criteria and are used to predict outcome using longitudinal multilevel modeling. RESULTS: In all samples, alliance systematically decreased between sessions and increased within them. Decreases in alliance as measured by traditional postsession to postsession were unreliable predictors of within-session decreases in alliance and of outcomes. CONCLUSIONS: Decreases in alliance across sessions are not valid measures of ruptures as defined by most theories. Instead, we suggest that they are indicative of an ebb-and-flow model for the development of alliance. (PsycInfo Database Record (c) 2020 APA, all rights reserved).