Psychometric properties of a short version of the Eating Attitudes Test (EAT-8) in a German representative sample

Disordered eating is common for all ages and sexes in the general population. However, only some individuals are known to develop clinically relevant eating disorders. There is a critical need of efficient, reliable and valid screening instruments to measure disordered eating for the general population. The Eating Attitudes Test was shortened into an 8-item-version to screen individuals at high risk of developing clinical eating disorders in a general population sample.

Psychometric properties (Cronbach's α, construct and concurrent validity and factor structure), cutoff scores (with sensitivity, specificity, positive and negative predictive value) and norms were determined in a representative sample of the German population (N=2527). Factorial validity was investigated using item response modeling.

Results confirmed reliability and validity of the questionnaire. Internal consistency and convergent validity were good. Analysis revealed different cutoff points for male and female participants. Values for sensitivity and specificity were satisfying and the positive predictive value was higher compared to other short screening instruments for disordered eating. Factorial analysis revealed a one-factor solution with an excellent model fit. The elimination of one item was discussed. Gender- and age-specific norms are reported.

Overall results indicated that the EAT-8 is an efficient instrument suitable for screening purposes in large general population samples.