The Psychometric Properties of the Eating Attitudes Test Short Form (EAT-26) in a College Sample

Erguney-Okumus F. E., Sertel-Berk H. Ö.



Determining risk groups is an essential element of preventive studies in eating disorders. In this regard, the Eating Attitudes Test Short Form developed by Garner, Olmstad, Bohr, & Garfinkel (1982) is the most commonly used scale all around the world. Hence, the aim of this study is to investigate the psychometric properties of the Turkish version of Eating Attitudes Test-26 (EAT-26). A total of fifteen hundred voluntary college students (1000 in the first phase for exploratory factor analyses and 500 in the second phase for confirmatory factor analyses and other validity and reliability analyses) enrolled in the study. The Eating Attitudes Test-40 (EAT-40), Brief Symptom Inventory, and Eating Disorders Examination Questionnaire were used for analyzing the validity of EAT-26. Concurrent validity, discriminant validity, and factor analyses were calculated. In terms of reliability, Cronbach's Alpha Coefficient Analyses and test-retest methods were used. Exploratory factor analyses showed that the items of EAT-26 loaded on three factors that explained 38.5% of total variance. These factors are Preoccupation with eating, Restriction, and Social Pressure. Confirmatory factor analyses of these three factors yielded close to acceptable goodness of fit values. EAT-26 was significantly correlated with Eating Attitudes Test-40, EAT-26, and Brief Symptom Inventory in a positive direction. Participants with a higher score in EAT-26 reported significantly higher scores in the Eating Disorder Examination Questionnaire and Brief Symptom Inventory. The Turkish version of Eating Attitudes Test-26 demonstrated good internal consistency (Cronbach's Alpha = .84), and the test re-test reliability was .78. The study provides initial support for the reliability and validity of the Turkish version of EAT-26. Nevertheless, future research is needed for the cross-validation of Eating Attitudes Test-26 in clinical samples.