ABSTRACT
The revised Helping Alliance Questionnaire (HAq-II) is among the most used instruments that measure therapeutic alliance. Despite its use in research, this instrument is not validated for the Brazilian population. The aim of this study was to explore the evidence of validity of the HAq-II based on the internal structure in a sample of Brazilian psychiatric patients. An ambulatory convenience sample of 204 patients with major depressive disorder (MDD) and 81 patients with obsessive-compulsive disorder (OCD) was randomized between two different types of treatment. The HAq-II was completed by patients (patient version) and by 33 therapists who performed the interventions (therapist version) between the second and third sessions. We used confirmatory factor analysis (CFA) to investigate two models: (1) a one-dimensional therapeutic alliance model and (2) a two-dimensional model considering the factors 'positive alliance' and 'negative alliance'. The internal consistency of the HAq-II was measured by Cronbach's alpha. To investigate discriminant validity, we used the t-test for independent samples, ANOVA and Pearson's correlation coefficient. The analysis showed that the best model of the instrument was one-dimensional. Cronbach's alpha for both versions displayed values above 0.89. The Brazilian version of the HAq-II presented properties similar to those found in the original version. However, studies related to the one-dimensional model with other samples are necessary.