Exploring the Relationships between Students' Engagement and Academic Performance in the Digital Textbook System

Abstract

In this paper, we analyzed the relationships between students’ engagement and academic performance in the digital textbook system. To measure students’ engagement, we first extracted features from the students’ digital textbook reading logs (click-streams) that represent their engagement with the contents. Then, we used percentile rank transformation to create normalized engagement scores and an overall engagement score. In the analysis, we first investigated the correlation between engagement scores’ and the students’ final scores. Second, we modeled students’ transition patterns from the engagement to academic performance by using Markov Chains. Third, we analyzed engagement patterns of the students with different academic performance levels. Our results showed that there is a positive moderate correlation between students’ academic performance and their engagement with digital textbooks. Our results also revealed that a single engagement score can be used to measure students’ engagement with the system, which is easy to understand by non-expert users. We also introduced our dashboard interventions that are developed based on this engagement score.