Component-Based PLS-SEM for Complex Modeling: Validation of a Third-Order Emotional Intelligence Scale

This study develops and validates a third-order Emotional Intelligence (EI) scale within the broader context of Social and Emotional Learning (SEL) and evaluates its stability using component-based Partial Least Squares Structural Equation Modeling (PLS-SEM). Using systematic random sampling, data were collected from 1,300 Taiwanese elementary students in grades 3 to 6, all of whom completed a 55-item EI questionnaire. PLS-SEM was employed for its suitability in analyzing complex, multidimensional constructs and in contexts where the emphasis is on theory development rather than confirmation. The model incorporated four second-order constructs, each comprising two first-order constructs, and its stability was assessed using heterogeneity tests and the Measurement Invariance of Composite Models (MICOM) procedure, with permutation testing applied to establish invariance. Results confirm that a statistical approach suited to complex questionnaires is a robust method for estimating parameters in complex models. The third-order EI scale demonstrated strong reliability and validity, and measurement invariance testing confirmed its stability across groups. The findings contribute to both EI and SEL research, as well as to the broader methodological literature on measurement invariance in complex model structures. Future studies are recommended to examine the long-term impact of EI and SEL across diverse educational settings, including those serving students with emotional and behavioral problems.