An investigation of carefulness among students using an educational game for Physics

Document Type


Publication Date



This work is an investigation of the implications of outliers in a student carefulness model. The authors built and empirically validated a model of carefulness among students using the gameplay logs of an educational game for Physics. In their prior work, carefulness was found to exist but there was no relationship found between carefulness and post-test learning gain. In this work, cluster-based outlier analysis is used and found that clusters of outliers existed in the dataset. After qualitative inspection of the clusters formed, we found that outliers existed and some even formed a cluster of outliers. We also found that the learning gains of the outliers and nonoutliers were statistically different and the degrees of carefulness between the clusters that existed in the more careful group were also significantly different, ie carefulness existed in varying degrees and levels. With the findings of this work, outlier detection and removal resulted to a more robust carefulness model that had significant relationships to learning gain. We recommend that appropriate meta-cognitive interventions and scaffolding in educational software be designed such that students will be more careful and gain more learning.