Evaluation of the factorial structure of the Body Shape Questionnaire: exploratory or confirmatory factorial analysis?
ABSTRACT Objective: To conduct a discussion about the strategies to assess the factorial structure of psychometric instruments using like an example the Body Shape Questionnaire (BSQ). Methods: We evaluated seven different factorial models of the BSQ that were obtained through exploratory factorial analysis (EFA) and are presented in the literature. The confirmatory factor analysis of these models was performed for study sample using the indices chi-square per degree of freedom ratio (χ2/df), Comparative Fit Index (CFI), Tucker-Lewis index (TLI), Root Mean Square Error of Approximation (RMSEA), and Weighted Root Mean Square Residual (WRMR). Convergent and discriminant validities were assessed by the average variance extracted and coefficient of determination among BSQ factors, respectively. Reliability was assessed using the composite reliability and the Cronbach's alpha coefficient. Results: 739 female college students with a mean age of 20.4 (standard deviation = 2.4) participated. All models assessed showed adequate fit for the study sample, however, the discriminant validity was inadequate in the all models. Furthermore, composite reliability of the models was adequate. Conclusion: Although all tested models of BSQ showed good psychometric indicators, it is worth noting that they were obtained in different samples using the EFA and without theoretical plausible justification for the construction of the factors, which may make it difficult to choose a model for clinical use. Therefore, considering the importance of preserving the theoretical concept during the development of an instrument, we suggest to researchers and clinicians who are cautious about the use of models without theoretical support.