Therapist and Client Perceptions of the Working Alliance: Codevelopment, Linear Growth, Variability, and Client Functioning

Using longitudinal client and therapist working alliance ratings, previous research examined how alliance: average value, linear growth, variability, stability (autocorrelation), and partner responsiveness were associated with client outcome. However, no research simultaneously examined all of these dimensions. Omitting important variables in analyses could lead to overestimation of related effects. Therefore, the purpose of this study was to examine these effects simultaneously. Therapists (N = 45) and clients (N = 236) rated alliance after every session (N = 10,720) and clients completed a measure of psychological distress after every eighth session. We used dynamic structure equation modeling to model longitudinal ratings using the longitudinal actor-partner interaction and location-scale models. Across sessions, there were significant linear growth and significant variability in client and therapist alliance ratings. The variability indicates multiple "V" patterns, which have been associated with alliance ruptures. Both actor effects were significant, showing session-to-session stability for client and therapist alliance. In addition, client-partner effect was significant, indicating higher-than-usual client alliance in a session predicting an increase in therapist alliance in the subsequent session. Growth in neither client-rated nor therapist-rated working alliances was significantly associated with client improvement. Lower variabilities (fewer fluctuations) in both client- and therapist-rated working alliances were associated with better outcomes. Higher therapist-actor and partner effects were associated with client improvement, but client-actor and partner effects were not associated with client improvement. Average working alliances were not associated with client improvement. Implications for practice and research were discussed.