[Validation of the French version of the 10-item Big Five Inventory]

Our primary objective was to validate the French version of the BFI-10, an ultra-short ten-item version of the Big Five Inventory (BFI; John et al., 1991), which allows for a reasonably accurate assessment of personality in circumstances in which more in-depth assessment is not possible. In order to reach a thorough evaluation of the external validity, we also aimed to examine the bandwidth of the BFI-10 scales with reference to the study by De Young, Quilty and Peterson (2007) who distinguished between two aspects in each of the Big Five: Assertiveness and Enthusiasm for Extraversion; Compassion and Politeness for Agreeability; Orderliness and Productiveness for Conscientiousness; Withdrawal and Volatility for Negative Emotionality, and finally Openness to Aesthetics and Openness to Ideas for Open-Mindedness. Our concern with regard to bandwidth was to examine whether the BFI-10 scales have strong enough correlations with both aspects of each domain.

Participants. Data from four samples were analysed: Sample 1 comprised 2499 undergraduate students (1654 women) who completed the full BFI in university classes; Sample 2 comprised 13,306 participants (8471 women) who filled out the BFI-10 ten items online via Internet; Sample 3 comprised 143 undergraduate students (115 women) who completed the full BFI twice with a two-week interval; Sample 4 comprised 360 undergraduate students (183 women) who filled out the BFI and NEO PI-R.

The French version of the Big Five Inventory is a 45-item inventory, which measures the five broader domains of personality. The ultrashort Ten-item Big Five Inventory (BFI-10) was developed simultaneously in German and English by Rammstedt and John (2007); it comprises five two-item scales measuring the big five domains. The Neuroticism-Extraversion-Openness Personality Inventory Revised (NEO PI-R; Costa et McCrae, 1992) is a 240-item questionnaire which assesses the big five domains and 30 lower-order facets, i.e. six facets per domain.

Factor structure and reliability of the five two-item scales were first investigated on samples 1 and 3. Confirmatory factor analyses (CFAs) were conducted on samples 1 and 2, and discriminant validity was assessed via comparison with the NEO PI-R (sample 4). In order to examine the bandwidth of the BFI-10 two-item scales, we studied their correlation not only with the NEO PI-R domains but also with the 30 facets.

The CFAs showed the good fit of the five-factor structure, with RMSEA=.077 (.072), CFI=.974 (.956), and SRMR=.029 (.027) in samples 1 and 2 respectively. Multigroup CFA conducted in groups 1 and 2 showed invariance across gender of factor loadings and item intercepts. Test-retest reliability was satisfactory with rs ranging from .68 (Open-Mindedness) to .86 (Extraversion and Negative Emotionality). The comparison of the two-item scales with the NEO PI-R scales showed high correlations not only with the NEO domain scales, but also with several facets: Four BFI-10 two-item scales (Extraversion, Conscientiousness, Negative Emotionality, and Open-Mindedness) were highly correlated with at least three of the six NEO facet scales in each domain. For Agreeableness, the magnitude of correlations was smaller, but the pattern of correlations was the same. All BFI-10 scales had at least moderate correlations with both aspects of each domain, with the exception of Negative Emotionality, which measured the NEO Withdrawal aspect better than Volatility.

The French version of the BFI-10 demonstrated the expected five-factor structure, satisfactory reliability, and broad bandwidth. It could be a valuable tool for the assessment of personality in circumstances in which it is not possible to use a longer and more in-depth instrument, especially when personality is not the main focus of research but one of the variables to be controlled.