Validation of the Dutch EDI-2 in One Clinical and Two Nonclinical Populations1

Abstract

Summary The present study establishes the validity of the Dutch translation of the EDI-2 by determining its factor structure, construct validity and its effectiveness for screening using a clinical and two nonclinical populations. An earlier study had shown that item transformation damaged the validity of the EDI-1 in a nonclinical population of female high-school students. Therefore, the issue of the use of transformed (0-3) versus untransformed (1-6) item responses was addressed first. The present study replicated the earlier finding for the EDI-2 in the same population and similar results were also found in another nonclinical population of female college students. Furthermore, untransformed responses were found to also work well in a clinical population of female patients suffering from eating disorders. Accordingly, further analyses were based on untransformed responses. The factorial integrity of the original 64-item EDI-1 was best supported in the clinical group and the nonclinical group of college students, but the three provisional scales performed poorly in terms of internal consistency and factorial integrity. In all three populations, the EDI-2 was shown to be specific for eating-disorder pathology, and to have good construct validity. It is concluded that the EDI-2 is a suitable screening tool to identify those cases in large populations that warrant more thorough examination.