[Validation of the French version of the "Beliefs About Voices Questionnaire-Revised" (BAVQ-R)]

The BAVQ-R explores the beliefs and reactions that "voice hearers" have about their voices. Recently, the BAVQ-R was translated into French, but little is known regarding its validity and reliability. Thus, the aim of this study was to examine further the psychometric characteristics of the BAVQ-R French version.

Seventy-six adult outpatients participated in the study. In addition to the BAVQ-R, patients were administered a psychiatric symptoms severity interview, and self-esteem and quality of life questionnaires. Then, the BAVQ-R was again completed six weeks later after the seven sessions of brief group cognitive behavioral therapy: the Voices Group.

The BAVQ-R was well accepted and was completed in less than 10minutes. All items presented floor and ceiling effects. For 15 items, item-total correlations were inferior to 0.30. Internal consistency was 0.83 for the global score, 0.83 for the malevolence, 0.81 for the benevolence, 0.64 for the omnipotence, 0.82 for the resistance and 0.77 for the engagement. The temporal stability was superior to 0.70, except for the omnipotence (0.65). Regarding, the construct validity, on the one hand malevolence, omnipotence and resistance were correlated (0.56≤r≤0.64; P<0.001) and on the other hand benevolence and engagement (r=0.62; P<0.001). Regarding convergent validity significant correlations were found between malevolence, omnipotence resistance and severity of anxio-depressive symptoms, negative self-esteem (0.40≤r≤0.52; P<0.001). The BAVQ-R allows discriminating neither between psychotic and non-psychotic patients (Mann-Whitney, all P>0.05) nor between groups established on the basis of the voice sources (Kruskal-Wallis, all P>0.05). Moreover, the BAVQ-R does not seem to be sensitive to change: no difference was observed after Voices Group intervention (Mann-Whitney, all P>0.05). Sociodemographic variables are not related to BAVQ-R.

The French version of the BAVQ-R has sufficiently adequate psychometric characteristics to be used in a clinical setting. However, floor and ceiling effects suggest a lack of sensitivity of the items. Psychometric characteristics of the BAVQ-R could be improved by selecting items and extending the Likert scale to score them.