Early Identification of Underperforming Students via Reading Patterns

• In Event: Division C Section 1a Literacy Paper Session

Abstract

Reading is an important learning process that contributes to academic success, yet it’s hard to track. By collecting detailed user activity logs, online textbook platforms provide insights into students’ authentic reading behaviors. In this paper, we examined college students’ reading patterns of a statistics online textbook using the K-means algorithm. The clustering results showed two reading patterns that differed in their number of sessions and mean session durations across chapters. Students who read more often and maintained a relatively stable reading time over time performed better in the course. An exploratory logistic regression also suggested that motivation measures and prior academic achievement could be used to predict these reading pattern clusters.

Authors

• Alice Xu, University of California - Los Angeles
  Presenting Author

• Adam B. Blake, University of California - Los Angeles
  Presenting Author

• Icy Yunyi Zhang, University of California - Los Angeles
  Non-Presenting Author

• Yiling Zhao
  Non-Presenting Author

• Rebecca Epner
  Non-Presenting Author