Impact of School Closure during COVID-19 Emergency: A Time Series Analysis of Learning Logs

Abstract

Recent spread of the COVID-19 forces governments around the world to have temporarily closed educational institutions. Although many studies were published to announce the best practice under the school closure, we need to understand the impact of school close on students’ learning before that. In this paper, we evaluate the impact of the school closure on our online teaching-learning environment. We use CausalImpact model to infer the impact on our learning analytics system using the learning log stored in the system. The results show that the school closure increased the number of logs on LMS by 163%, but decreased the number of logs on e-book reader by 77%. However, focusing on a particular course, we found that students’ learning engagement on online system increased both in LMS and e-book reader. We discussed that it is caused by the following reasons: 1) Changes in major users on our online learning platform, and 2) Limited functions of our e-book reader which was developed for face-to-face learning, not online learning. Further, the results also suggested that CausalImpact model is useful for evaluating the effectiveness of a specific event from learning logs collected by learning analytics systems.