Assessing the factor structure and measurement invariance of the eating attitude test (EAT-26) across language and BMI in young Arab women

Salma M. Khaled, Linda Kimmel, Kien Le Trung

The objective of the study was to determine the factorial structure and test the measurement invariance of the EAT-26 in a large probability sample of young female university students in Qatar (n = 2692), a Muslim country in the Middle East. The maximum number of factors was derived based on results from initial exploratory factor analysis (EFA) in the first-half of the randomly split sample (Sample 1). A subsequent EFA and Exploratory Structural Equation Models (ESEM) were conducted to identify the number of valid factors. A five-factor model with 19 items was identified as the optimal factor structure. This structure was further replicated using ESEM in the second-half of the sample (Sample 2). Multi-group Confirmatory Factor Analyses (CFAs) were conducted at this stage and their fit was evaluated with and without further sub-grouping by language (Arabic and English) and BMI (underweight, normal weight, and overweight/obese). Finally, measurement invariance tests were conducted in the entire sample assessing equivalence across language and BMI within the final five-factor model. The five-factor structure of the new EAT-19 [fear of getting fat (FGF), eating-related control (ERC), food preoccupation (FP), vomiting-purging behavior (VPB), and social pressure to gain weight (SP)] provided the best fit: CFI = 0.976, TLI = 0.952, RMSEA = 0.045 (90%CI 0.039-0.051), SRMR = 0.018, CD =1.000. CFAs supported metric invariance for language and for BMI. Language and BMI-based population heterogeneity comparisons provided modest and small-to-moderate evidence for differential factor means, respectively. Although the five-factor model of the EAT-19 demonstrated good item characteristics and reliability in this young female population, the lack of scalar invariance across language and BMI-categories pose measurement challenges for use of this scale for screening purposes. Future studies should develop culture- and BMI-specific cut-offs when using the EAT as a screening instrument for disordered eating in non-clinical populations.