[French validation of the modified version of the SAS].

INTRODUCTION: According to research in sport psychology, anxiety is one of the most important and studied topics. Since its conceptualisation, many scales assessing sport anxiety have appeared in the English language. At present, the multidimensional conceptualisation of anxiety is dominant. Smith et al. developed a self-report questionnaire to assess the anxiety trait in sport (Sport Anxiety Scale, SAS). As a consequence, it seems important to have such an instrument in French language in order to fill in questionnaires on lack of self-report so as to assess cognitive and somatic anxiety traits.

LITERATURE FINDINGS: The SAS was developed with 30 items in two factors: somatic anxiety and cognitive anxiety. However, the final version of the SAS had 24 items in three factors. A third factor was identified during factorial analyses: concentration disruption. Sarason's researches corroborate in part this third factor with the theory of cognitive interferences. Dunn et al. underlined the theoretical incoherence of this assumption. In the same way, Neil et al. suggested the deletion of this third dimension.

AIM OF THE STUDY: Considering all these results, the aim of this study is to suggest a French translation of the SAS, taking into account the modified version suggested by Neil et al.

METHOD AND PROCEDURE: The scale was translated according to Vallerand's process. Two hundred and seven sportsmen (135 boys and 71 girls) aged 20 years (S.D.=2.5 years) filled out the scale on site. A Confirmatory Factorial Analysis (CFA) using the LISREL 8.30((c)) software was conducted. ANOVAs were also carried out to compare trait anxiety scores as a function of gender; level and type of sport.

RESULTS: The CFA results on both models are acceptable (chi(2)/ddl=1.60, Comparative fit index (CFI)=0.98, Global fit index (GFI)=0.92, Standardized root mean square residual (SRMR)=0.05, Root mean square error of approximation (RMSEA)=0.05) and Cronbach's alpha for each scale ranged from 0.86 for cognitive trait anxiety to 0.89 for somatic trait anxiety. The ANOVAs show a difference according to gender (girls were more anxious than boys); level (the higher the level, the more the sportsmen were anxious) and type of sport (those who practiced collective sports were more anxious than those who practised individual sports). To conclude, and waiting for other researches using the French version of SAS, it appears that this scale is generally acceptable, and can be a useful tool for research on the anxiety trait in sport.