Psychosocial Determinants of University Students’ Intention to Avoid Drug Abuse: A Structural Equation Modeling Approach with DWLS Estimation
Structural Equation Modeling (SEM) is widely employed to investigate complex relationships among latent variables. However, the Maximum Likelihood estimator commonly used in covariance-based SEM relies on the assumption of multivariate normality, which is often violated when indicators are measured using ordinal Likert scales. To address this issue, this study applies the Diagonally Weighted Least Squares (DWLS) estimator to model the relationships between psychosocial factors and University of Lampung students’ intention to avoid drug abuse. The measurement model was evaluated using validity and reliability criteria, while the structural model was assessed through path coefficients, coefficient of determination, and goodness-of-fit indices. The results show that the proposed model explains 58.8% of the variance in intention to avoid drug abuse ( ). Emotional distress ( ), self-development activities ( ), and social relationships ( ) significantly affect students’ intention to avoid drug abuse. Furthermore, the model exhibits an excellent fit to the observed data, with , and . These findings indicate that SEM with DWLS estimation provides a robust and reliable approach for modeling latent-variable relationships involving ordinal data and offers an effective alternative to conventional estimation methods in covariance-based SEM.
Keywords: Structural Equation Modeling; Diagonally Weighted Least Squares; Covariance-Based SEM; Ordinal Data; Latent Variables; Psychosocial Factors; Intention to Avoid Drug Abuse.
