Le modèle des « Big Five » rencontre aujourd’hui un large consensus pour l’évaluation de la personnalité. Le Big Five Inventory (BFI), développé par John et al., a été traduit et validé en France. Le but de cette étude est de présenter la poursuite du travail de validation du BFI français (BFI-Fr) sur un large échantillon d’étudiants, pour vérifier (1) ses qualités psychométriques internes et (2) assurer une validation convergente avec le NEO Personality Inventory, Revised (NEO-PI-R). Une première étude de validation interne et de comparaison interculturelle a été réalisée auprès d’étudiants (n=2499). Elle a mis en évidence une réplication satisfaisante de la structure factorielle américaine et une confirmation des premiers travaux de validation en langue française. Elle atteste de la solidité de la structure factorielle du BFI-Fr et de ses bonnes qualités psychométriques. La comparaison des scores et écarts-types entre les étudiants français, américains et espagnols montre des résultats très similaires, témoignant indirectement de la comparabilité des outils. Une seconde étude a mis en évidence une validation convergente et discriminante satisfaisante par rapport au NEO-PI-R. Le BFI-Fr est désormais un outil valide, permettant aux cliniciens et chercheurs de disposer d’un inventaire des cinq grands facteurs de la personnalité en langue française. Il est simple et robuste, fiable et économique.

After five decades of research, an initial consensus on a general taxonomy of personality traits, the “Big Five” personality dimensions, is nowadays largely accepted. These dimensions do not represent a particular theoretical perspective but were derived from factor analyses of the natural-language terms people use to describe themselves and others. The Big Five Inventory (BFI) does not use single adjectives as items because such items are answered less consistently than when they are accompanied by definitions or elaboration. It uses 44 short phrases based on the trait adjectives known to be prototypical markers of the Big Five. The Big Five have been most typically labeled E (Extraversion, Energy, Enthusiasm), A (Agreeableness, Altruism, Affection), C (Conscientiousness, Constraint, Control of impulse), N (Neuroticism, Negative affectivity, Nervousness), and O (Openness, Originality, Open-mindedness). The BFI has been translated and validated in different languages. The development was done in such a way that they resembled as closely as possible the original English version, both in psychological meanings and psychometric properties.

The goal of this paper is to present the validation process of the French BFI (BFI-Fr) on a large student sample to verify psychometric properties, including factor structure and internal reliability and to show that the scales possess the necessary convergent and discriminant validity with the NEO personality inventory, revised (NEO-PI-R).

Study 1: Internal consistency and intercultural comparison.

Two thousand four hundred and ninety-nine students were included (women 69%; mean age 20.2 years old, S.D.=2.21, between 15 and 46 years). The 45 items of the BFI-Fr were filled out anonymously by the students at the university.

A factorial analysis using principal components was performed on the student answers (raw data) and resulted in a five-factor varimax-rotated solution that was easily verified as the expected five dimensions E, A, C, N, and O, which explained 42% of the total variance. Cronbach’s alpha coefficients which measure the internal coherence were respectively: 0.82, 0.75, 0.80, 0.82, and 0.74. This factorial analysis represents a very good replication of the American BFI. The mean internal consistency (0.79) is excellent, providing clear evidence of the psychometric qualities of the tool (internal validity). Normality of the distribution factors was verified before comparing the scores of French students with those of American and Spanish students. Mean scores and standard deviations were very similar in the three countries. As in previous research, gender differences in personality were found: females had higher scores (p<0.001) for N, A, and C.

Study 2: convergent and discriminant validation with the NEO-PI-R.

The goal of study 2 was to compare the BFI-Fr with the NEO-PI-R.

Three hundred and sixty students (women 55%, mean age 21.1 years, S.D.=2.30, between 18.3 and 45.5) were included. In the same session, they completed both the BFI-Fr and the French NEO-PI-R.

Internal consistencies of the five personality dimensions were comparable for the BFI and the NEO-PI-R. Correlations between the corresponding pairs were all high (mean=0.74) and significant (p<0.001). These results provide evidence of the convergent validity of the BFI-Fr. Discriminant validity was excellent, with correlations between the other scales much lower than the convergent correlations, averaging only 0.14.

All three studies demonstrate that the BFI-Fr is a valid, powerful yet very efficient tool, as are the original English version and the other translations. The much longer NEO-PI-R remains the instrument of choice. Psychiatrists, psychologists, and researchers can now make use of another inventory in French to measure the Big Five which has the advantages of being simple, robust, reliable, and economical (5 to 10min to complete).