Integrating Partial Least Squares Structural Equation Modelling and Artificial Intelligence in Educational Research: A Simulation-Based Methodological Framework for Resource-Constrained Contexts (Published)
The rise in the need to make decisions in the educational system based on data has led to the application of sophisticated data analysis methods like Partial Least Squares Structural Equation Modelling (PLS-SEM) and Artificial Intelligence (AI). Nevertheless, their joint use is not yet widespread, especially in resource-limited scenarios. This paper suggests and illustrates a unified methodological framework of applying PLS-SEM and AI via a simulation-based method. The explanatory modelling and machine learning (Random Forest) of synthetic data to represent key educational constructs were performed using PLS-SEM and machine learning, respectively. The results of the PLS-SEM showed that there were strong correlations between constructs (satisfaction, student engagement, and institutional support) that described 54% of the variance in academic performance. The AI model had a better predictive accuracy (R2 = 0.71), as compared to the PLS-SEM model. The integration had complementary and consistent results. The research shows that the combination of PLS-SEM and AI increases the explanatory and predictive power. The framework is specifically applicable to resource-limited settings and provides viable advice to education researchers.
Keywords: Artificial Intelligence, PLS-SEM, Simulation, education research, predictive analytics, resource-constrained contexts