Exploring differential item functioning on eating disorder measures by food security status

Area of Science:

• Psychiatry
• Psychology
• Public Health

Background:

• Food insecurity is linked to increased eating disorder (ED) pathology.
• Existing ED measures may not accurately reflect pathology in food-insecure individuals.
• Differential item functioning (DIF) can reveal item biases in questionnaires.

Purpose of the Study:

• To investigate if item endorsement on two common ED questionnaires differs between food-secure and food-insecure individuals.
• To identify potential biases in ED measures within food-insecure populations using DIF analysis.

Main Methods:

• 634 cisgender women recruited via Amazon Mechanical Turk.
• Differential item functioning (DIF) analysis applied to the Short Eating Disorder Examination Questionnaire (S-EDE-Q) and the Eating Disorder Diagnostic Scale for DSM-5 (EDDS-5).
• Hybrid ordinal logistic regression/item response theory approach used, with statistical (p < .01) and clinical significance (pseudo ΔR² ≥ .035) defining DIF.

Main Results:

• No clinically significant DIF found for the S-EDE-Q.
• Two items on the EDDS-5 demonstrated statistically and clinically significant DIF.
• Food-insecure participants were more likely to report eating large amounts when not hungry and experiencing negative emotions about overeating at lower ED pathology levels.

Conclusions:

• Findings suggest that ED measures may require adaptation for food-insecure populations.
• The EDDS-5 items showing DIF may not fully capture ED pathology in this demographic.
• Further research is needed to refine ED assessment tools for diverse socioeconomic contexts.