Measurement invariance of the Eating Attitudes Test-26 in Caucasian and Hispanic women

To determine whether the EAT-26 functions similarly in Caucasian and Hispanic samples, the current study investigated the factor structure of the Eating Attitudes Test (EAT-26) in 235 undergraduate Caucasian (53.6%) and Hispanic (46.4%) women, and tested for measurement invariance across the two samples. A Confirmatory Factor Analysis (CFA) of the original 3-factor structure of the EAT resulted in a poor fit in both the Caucasian and Hispanic samples. We then performed a CFA using a previously discovered 4-factor, 16-item structure. This abbreviated measure was a good fit in both the Caucasian and Hispanic samples, and the model was invariant across all dimensions tested. The 16-item EAT is a better-fitting measure in Caucasian and Hispanic women than the commonly used EAT-26. This replicates an earlier finding and generalizes those conclusions to a Hispanic sample.

► We examined the factor structure of the Eating Attitudes Test in Caucasian and Hispanic women.
► The original factor structure was a poor fit in both samples.
► We replicated a more recent 4-factor, 16-item factor structure.
► This factor structure was invariant across ethnicities.