Search

The most recent version of the widely used NEO Inventories, the NEO-PI-3, has been translated in several languages worldwide, including Canadian French. Probably the main criticism of the validity NEO-PI-3 has to do with the confirmation of its factor structure in confirmatory factor analysis (CFA). The present study aimed to assess the structural validity of the NEO-PI-3 at the full model level, testing the original model as well as alternative more complex models. Single-factor models at the trait level, and at the basic measurement (facet) level, were also tested. French-Canadian data from real-life settings were analyzed (n = 1313). None of the three full models tested with CFA reached acceptable fit. However, acceptable fit for the oblique full model was reached in the unrestricted framework of partial confirmatory factor analysis (PCFA). Yet, the PCFA revealed substantive crossloadings between Extraversion and Agreeableness facets. Only single-factor models of Neuroticism and Conscientiousness evidenced good model fit, and thirteen of the thirty facets showed acceptable fit. Further analyses indicated that poor model fit for the facets could be explained by high covariances between item measurement errors. The inadequate model fit of the NEO-PI-3 seems to be due to the complexity of the model and to model misspecifications. [ABSTRACT FROM AUTHOR]