Aim/Purpose: The study proposes and evaluates a theory-driven analytics methodology for inferring learners’ cognitive load and performance from Moodle log data in online learning environments. Background: Although Moodle captures rich learner interaction data, existing analytics largely focus on descriptive engagement metrics and provide limited insight into learners’ cognitive processes. This gap restricts the ability of learning analytics to support cognitively informed instructional design and adaptive learning. Methodology: An experimental design was employed, with 324 undergraduate students from three faculties in a public University, randomly assigned to treatment and control groups. The treatment group received cognitive load-informed interface scaffolds, while the control group accessed the same content without interventions. Moodle log data collected over four weeks were analyzed using statistical tests, clustering, classification, and association rule mining within a nine-step analytics workflow implemented in Python. Contribution: The study introduces a reproducible analytics methodology that explicitly embeds extraneous, intrinsic, and germane cognitive load constructs into Moodle log analysis, extending existing LMS analytics frameworks from descriptive engagement monitoring to theory-grounded explanatory and predictive analysis. Findings: Results indicate that treatment learners demonstrated significantly higher engagement and performance than control learners, with large effects for course activity completion (d = 0.77) and module views (d = 0.65), and moderate effects for quiz grades (d = 0.38). Machine learning analyses further revealed more differentiated behavioral patterns and stable associations between cognitive load-related behaviors and learning outcomes among treatment learners. Recommendations for Practitioners: Educators and instructional designers can apply the proposed methodology to unobtrusively monitor learners’ cognitive load and implement targeted interface scaffolds that enhance engagement and performance without altering core instructional content. Recommendation for Researchers: Researchers are encouraged to adopt and extend the proposed methodology to examine cognitive load dynamics across disciplines, platforms, and instructional designs, and use it as a foundation for theory-driven learning analytics studies that move beyond purely predictive models. Impact on Society: By enabling cognitively informed learning analytics at scale, this methodology supports more effective and equitable online education, particularly in resource-constrained contexts. Future Research: Future work should validate log-based cognitive load indicators using multimodal data, explore real-time adaptive learning systems responsive to inferred cognitive load, and assess the long-term effects of cognitive load-informed interventions on learner retention and achievement.
Ogwoka et al. (Thu,) studied this question.