Psychometric Properties of The 26-Item Eating Attitudes Test (EAT-26): An Application of Rasch Analysis
Abstract Background: The 26-item Eating Attitudes Test (EAT-26) is a commonly used tool to assess eating disorder risk. The purpose of this study was to examine the psychometric properties of the EAT-26 with a sample of adults (n=469; age=36.17±17.83 years; female =72.5%; white=66.3%; obese BMI category=58%).Methods: Rasch analysis of the EAT-26 assessed model-data fit, an item-person map to evaluate relative distribution items and persons, item difficulty, and person’s eating disorder (ED) risk level, differential item functioning (DIF), and rating scale functioning. Results: A total of 7 misfit items were removed from the final analysis due to unacceptable Infit and Outfit mean square residual values. The item-person map showed that the items were biased toward participants with moderate to high levels of ED risk and did not cover those who had low risk for having an ED (< -1 logits). The DIF analyses results showed that none of the items functioned differently across sex, but 5 items were flagged based on obesity status. The six-category Likert-type rating scale did not function well indicating a different response format may be needed.Conclusion: Several concerns were identified with the psychometric evaluation of the EAT-26 that may question its utility in assessing ED risk. Because the EAT-26 is a frequently used screening tool for nonclinical populations, future work should focus on developing screening tools that are more effective at assessing ED risk in people with overweight and obesity.