The Cross Cultural Validity of the EDI-3 Eating Disorder Risk Subscales: A Comparative Factor Analysis with Caucasian American and African American Samples

This study examined the cross-cultural validity of the eating disorder subscales of the Eating Disorder Inventory-3 (EDI-3) in Caucasian American and African American female samples. A comparative factor analysis was conducted in order to determine the fit of a three-factor model of Eating Disorder Risk, within the EDI-3, proposed by Garner (2004). The sample consisted of 197 Caucasian American and 104 African American females. Confirmatory Factor Analyses (CFA) of the three-factor model were conducted separately for each ethnic group using Lisrel 9.1. CFA results indicated the 3-factor model was an adequate fit for the Caucasian sample, but only a mediocre fit in the African American sample. Follow-up EFAs indicated that the Caucasian data produced a pattern matrix consisting of three factors (DT, B, BD) while the African American data produced a pattern matrix consisting of four factors (DT, B, BD, BS). Assessment of item composition differences under similar factors, mean differences, and correlations further elucidated differences found between the two samples. The overall results of the factor analyses suggest that the EDI-3 may be capturing different constructs for African American Americans compared to Caucasians.