Investigating behavioral patterns to facilitate performance predictions during online peer assessment through learning analytics approach

Abstract

Insights into student behavioral patterns yield benefits for both educators and learners. Nevertheless, only a few studies have investigated the behavioral patterns of high-scoring (HS) and low-scoring (LS) students to facilitate predictions during peer assessment (PA). Therefore, we performed learning analytics to explore the behavioral patterns of HS and LS students in 52 university students from affective, cognitive, and metacognitive perspectives as these students engage in online PAs. The results indicated that on the affective dimension, HS students tended to exhibit negative affection while LS students tended to display positive affection. On the cognitive dimension, HS students demonstrated more intricate transformations compared to the LS students. Regarding the metacognitive dimension, LS students more likely reflected upon and accepted the reviews provided by others than the HS. Additionally, the findings also revealed that HS students were supported by focused social networks and overall stable activity engagement, whereas the LS students were less engaged at the beginning of the activity but had evolving social networks and were more engaged in the later stages of the activity. These findings offer valuable insights for educators in effectively predicting student performance based on behavioral patterns while providing new perspectives to support instructional decision-making processes.