Psychometric evaluation of the eating disorder inventory (edi) in a clinical group

Abstract

This study examined the psychometric properties of the Eating Disorder Inventory (EDI) with a clinical sample of 79 female clients seen at an outpatient eating disorder program. The means and standard deviations for the bulimia nervosa subgroup in this study (n = 48) compared favorably with the bulimia group in the EDI Manual Supplement (1986) on all eight subscales, except Drive for Thinness. The present non-student sample had significantly lower scores on this particular subscale than did the norm group (mostly students). Another difference was that the local bulimic subgroup had significantly higher present and minimum weights (as a percentage of the expected average weight according to height). The EDI subscales also displayed respectable internal consistency reliability with this sample. In contrast to previous research, however, only five factors were found in a factor analysis rather than eight corresponding to the original subscales for the EDI. Differences between bulimic and a not-otherwise-specified (NOS) diagnostic groups were also found on both the EDI Ineffectiveness and Bulimia subscales (NOS had lower scores). The results of this study are discussed in terms of the importance of undertaking local standardization of tests in clinical settings.