Investigating behavior patterns of students during online self-directed learning through process mining

One of the recognized ways to enhance teaching and learning is having insights into the behavior patterns of students. Studies that explore behavior patterns in online self-directed learning (OSDL) are scant though. In addition, the focus is lacking on how high-achieving (HA) students’ behavior patterns affect the academic performance of low-achieving (LA) students. To fill these research gaps, this study investigates (1) how the behavior patterns in OSDL vary between HA and LA students and (2) how HA students’ behavior patterns affect LA students’ academic performance. We used three perspectives of learning achievement, engagement, and cognitive load to examine academic performance. By utilizing process mining, we reviewed the log data of 71 college students on the Moodle platform and designed a pretest–posttest test without a control group. Results show obvious variances in the behavior patterns between HA and LA students. In particular, HA students performed more OSDL behaviors; their behavior patterns were more in line with self-directed logic. By contrast, LA students exhibited unmethodical behavior patterns; they were unable to process course content in depth. An instructional intervention was created with HA students’ behavior patterns as basis. The engagement of LA students increased, and their cognitive load was reduced after the instructional intervention. However, their learning achievement did not increase substantially. The interview results were consistent with the quantitative data. These findings indicate that the behavior patterns of HA students can shed light on how to guide the OSDL of LA students. This study also provides fresh methodological perspectives for assessing OSDL. Copyright © 2023 The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature.