•  
  •  
 

ORCID

Surina He: https://orcid.org/0000-0002-9859-9749

Okan Bulut: https://orcid.org/0000-0001-5853-1267

Tarid Wongvorachan: https://orcid.org/0000-0002-9622-3780

Abstract

Log data extracted from learning management systems (LMS) have become an increasingly valuable resource for understanding students’ learning behaviors. These data are unobtrusive, cost-effective, and offer fine-grained insights. However, their complexity and large volume present substantial challenges in identifying interpretable underlying dimensions across different event types. To address this challenge, this study applied exploratory factor analysis, psychometric network analysis, and consensus hierarchical clustering to analyze 122,167 log entries from 25 event types collected in a large undergraduate course (n = 527 students). Four distinct clusters of student activities were identified in the event logs: submission, quiz-taking, discussion, and background-viewing. These clusters offer meaningful insights into different patterns of student engagement, indicating that engagement is primarily driven by course requirements and self-directed actions. A structural equation model was applied further to explore the relationship between these clusters and academic performance. Results revealed that submission, discussion, and background-viewing activities were positively associated with students' course grades, whereas quiz-taking behavior did not show a significant relationship. These findings suggest that the presence of specific behaviors may enhance learning outcomes. This study contributes to the growing body of research by showcasing the utility of interpretable methods for analyzing complex LMS data in educational settings.

DOI

https://doi.org/10.59863/FZRN7818

Download

Share

COinS