RESUMO
The characteristically large general factor found in measures of the therapeutic alliance was analyzed by use of confirmatory factor analysis (CFA) in a nested design. Ratings by 38 therapists and their 144 patients on the California Psychotherapy Alliance Scales (CALPAS), the Revised Penn Helping Alliance Questionnaire (HAQ-R), and the Working Alliance Inventory (WAI) were adjusted for therapist effects. A set of models for patient and therapist ratings was tested with CFA, and a 3-factor model was confirmed, chi 2(4) = 7.19, p > .13; GFI = .98; RMSR = .02; CFI = 1.0. A shared-view factor (best represented by HAQ-R) accounted for 44% of patients' and 27% of therapists' variance. Unique factors (best represented by WAI) accounted for 56% of therapists' and 43% of patients' variance. Patient views split between HAQ (helpfulness) and WAI (goals, tasks) factors; The WAI factor was most expressive of therapist views. Patients and therapists tended to agree on helpfulness and on therapist clarity about goals and tasks. A multimeasure approach to alliance assessment is recommended.
