Temporal stability of the Eating Disorder Examination Questionnaire.

Objectives: The current study examined the stability and internal consistency of the Eating Disorder Examination Questionnaire (EDE-Q) in a general population sample.

Methods: The EDE-Q was administered to a community sample of women aged 18-45 on two occasions, with a median test-retest interval of 315.0 days.

Results: Pearson correlations between items of the EDE-Q assessing attitudinal features of eating disorder psychopathology ranged from 0.57 for the Restraint subscale to 0.77 for the Eating Concern subscale. The stability of items addressing eating disorder behaviors was much lower, with phi coefficients for the occurrence of objective bulimic episodes, subjective bulimic episodes, and use of exercise as a compensatory behavior of 0.44, 0.24, and 0.31, respectively, and Kendall's tau b correlations of 0.44, 0.28, and 0.31, respectively, for the frequency of these behaviors, across occasions. The internal consistency of the EDE-Q was high, with a Cronbach alpha coefficient for the global scale of 0.93, compared with a value of 0.90 for the Eating Disorder Examination interview.

Discussion: Items of the EDE-Q assessing attitudinal features of eating disorder psychopathology demonstrate a high degree of temporal stability, whereas the stability of items addressing eating disorder behaviors is much lower. In the case of compensatory eating disorder behaviors, low stability is likely to reflect actual trait variation, whereas the low stability of binge eating behaviors, in particular subjective bulimic episodes, is likely to reflect both trait variation and measurement error. The high internal consistency of EDE-Q items supports its use as a screening instrument in two-phase epidemiologic studies.