Jamie-Lee Pennesi, Laura Edney, Yuan Zhou, David L Hegarty, Benjamin Buchanan, Tracey D Wade
Confirmatory factor analysis supported both the hypothesized two-factor and a bi-factor model (with a general factor), with the two-factor model retained as the preferred structure, consistent with theory and prior research. The ED-15 also demonstrated strong internal consistency and test-retest reliability. In subsamples, there was preliminary evidence of measurement invariance across gender, time, country, and to a lesser extent age group. Correlations supported good concurrent validity with eating disorder pathology and eating-related impairment, while convergent validity was mixed, with moderate associations with psychological distress and maladaptive perfectionism, but weak or negligible associations with personal standards perfectionism and autism-related traits. Receiver operating characteristic analyses deriving an optimal diagnostic cut-off indicated good discriminative validity, with the ED-15 discriminating probable eating disorder caseness.
OBJECTIVE: This study examines the factor structure, measurement invariance, and discriminative ability of the Eating Disorder-15 (ED-15) questionnaire in a large transdiagnostic clinical population of treatment-seeking adults.
METHOD: The sample consisted of 2015 participants aged 17-79 (Mage = 30.73 years; 77.0% women; 71.1% in Australia) seeking mental health treatment across different settings of routine clinical practice. After initial completion of the ED-15, subsamples completed additional measures within ±365 days of the initial ED-15 and/or repeated the ED-15 at least 7 days later.
RESULTS: Confirmatory factor analysis supported both the hypothesized two-factor and a bi-factor model (with a general factor), with the two-factor model retained as the preferred structure, consistent with theory and prior research. The ED-15 also demonstrated strong internal consistency and test-retest reliability. In subsamples, there was preliminary evidence of measurement invariance across gender, time, country, and to a lesser extent age group. Correlations supported good concurrent validity with eating disorder pathology and eating-related impairment, while convergent validity was mixed, with moderate associations with psychological distress and maladaptive perfectionism, but weak or negligible associations with personal standards perfectionism and autism-related traits. Receiver operating characteristic analyses deriving an optimal diagnostic cut-off indicated good discriminative validity, with the ED-15 discriminating probable eating disorder caseness.
DISCUSSION: Findings add to the growing evidence on the ED-15's latent structure and provide preliminary support for measurement invariance, supporting its use as a brief, psychometrically sound tool in routine clinical practice. Future research should evaluate its intended use for session-by-session monitoring of eating disorder symptom change.