Psychometric Properties of Various Factorial Models of the ORTO-15 Tested With Brazilian Respondents

Abstract

The ORTO-15 is a commonly used screening instrument for symptoms of Orthorexia Nervosa (ON), an obsessive overconcern with healthy eating. However, its limitations have been evidenced in several published factorial models. We analyzed the psychometric properties of seven different ORTO-15 factorial models using data from Brazilian adult participants (n = 1455; 71.4% women; M age = 29.5, SD = 8.9 years) from the country’s five macro-regions who answered online questions for sample characterization and completed the Portuguese version of the ORTO-15 containing 15 items. We tested the fit of each model separately, analyzing data for men and women with separate confirmatory factor analysis, and using the Weighted Least Squares Mean and Variance Adjusted estimation method and goodness-of-fit indices. We also calculated the average variance extracted and the alpha ordinal and omega coefficients to analyze the convergent validity and reliability of the factors. None of the models tested presented adequate properties of validity and reliability. Although we found some acceptable reliability coefficients, they do not guarantee the validity of the data. Future investigators should be cautious when choosing the ORTO-15 for ON screening.