Résumé
Recent studies have questioned the appropriateness of the original Eating Disorders Inventory (EDI) in nonclinical samples of adolescents. The main objective of the present series of studies is to systematically test the construct validity of the EDI (i.e. content, factorial, convergent, discriminant and discriminative) in a nonclinical sample of French adolescents. A total sample of 1,323 adolescents was involved in these five studies. The factorial validity and the measurement invariance of the EDI were verified through confirmatory factorial analyses. Correlation and student t-tests were also used to test the convergent and discriminative validity of the EDI. Results from the first study confirmed the unsuitability of the French original EDI for young adolescents. Items were re-worded and an adaptation for adolescents was developed (EDI-A). The following four studies provided support for the factorial validity, measurement invariance, reliability, convergent validity and discriminant validity for a short form (i.e. 24 items) of the EDI-A. The present results thus provide preliminary evidence regarding the construct validity of the 24-item EDI-A for French nonclinical adolescents. Recommendations for future uses and research activities with this instrument in French speaking adolescents are outlined.