Identifying Factors of In-Video Dropout in Middle School Online Math Learning Using Survival Analysis

• In Event: Data-Driven Insights in Online Learning

Abstract

This research paper explores student dropout behavior during video watching within an online math learning platform called Math Nation. Analyzing 34,666,481 log entries from Math Nation, covering 1,313 instructional videos and 14,251 secondary school students, we identified factors influencing students' in-video dropout behaviors. Using survival analysis, we examined 27 variables, including demographics, video interaction behaviors, and video characteristics like length and category. Our findings indicate that each additional minute of video length increases dropout rates by 1.26\%. Higher dropout rates correlate with frequent pauses, jumps, and rewatches. The quality of video content, creators, and student interactions are also crucial factors. This study provides insights to enhance engagement in video-based learning.

Authors

• Zifeng Liu, University of Florida

• Yukyeong Song, University of Tennessee

• Wanli Xing, University of Florida