Dimensionality, reliability, invariance, and item analysis of the depression, anxiety, and stress scale-21 (DASS-21) in Honduran and Colombian university students.

backgroundThe 21-item Depression, Anxiety, and Stress Scale (DASS-21) is a widely used instrument for assessing psychological distress in diverse populations worldwide. Nonetheless, in the Honduran and Colombian contexts there is a paucity of studies evaluating the scale's factorial structure, examining item-level parameters, and investigating measurement invariance at both the dimension and item levels. The present study analyzes the psychometric properties-structural validity, reliability, measurement invariance, item-level characteristics, and differential item functioning-of the DASS-21 among university students in Honduras and Colombia.

methodsA cross-sectional design was employed, involving 2,271 university students (1,112 from Honduras and 1,159 from Colombia). The full sample comprised 68.10% women, with a mean age of 22.21 ± 5.03 years. The participants completed the DASS-21 through an online form. As part of the factorial modeling, several structures were tested via confirmatory factor analysis; configural, metric, and scalar invariance across countries were examined; and the reliability of the best-fitting model was estimated. In addition, an item response theory evaluation was carried out with a multidimensional graded response model (MGRM) that estimated two parameters-discrimination and thresholds-and differential item functioning (DIF) was assessed.

resultsThe correlated three-factor model of the DASS-21 was identified as the best-fitting structure, as indicated by satisfactory fit indices (comparative fit index [CFI] and Tucker-Lewis Index [TLI] > .95; root mean square error of approximation [RMSEA] < .08) and factor loadings (λ > .40; p < .01). Acceptable reliability coefficients were obtained for each dimension (Ω > .80). This model demonstrated configural, metric, and scalar invariance across countries. Findings from the multidimensional graded response model showed acceptable discrimination and threshold estimates for all the items. Uniform differential item functioning was detected in four items of the scale (p < .05).

conclusionsThe results provide favorable psychometric evidence supporting the use of the DASS-21 as a screening tool for assessing depression, anxiety, and stress among Honduran and Colombian university students. Evidence of measurement invariance allows cross-cultural comparisons of latent means. Nevertheless, caution is advised when making country-specific item-level comparisons for the items that exhibited DIF.