Predicting Young Students' Self-Regulated Learning Deficits Through Their Activity and Self-Evaluation Traces
Abstract: Being well self-regulated is important for students of all ages in order to maximize their learning. To raise awareness to some self-regulation deficits we have introduced two mandatory self-regulation statements asked randomly after 1 out of 15 exercises into a literacy web-application for primary school students, to evaluate perceived difficulty [Too easy, Good, Too difficult] and desired difficulty [easier, same level, harder]. Students are assisted through various scaffolding and feedback, to help them improve their self-regulation over time.Comparing students' actual performance with their responses to self-regulatory statements can provide information about their ability to self-regulate their learning, and thus detect possible deficits. We collected above 1,000,000 responses from 300,000 students for our experiments.Using these data as well as performance data on each question of each exercise, we try to predict a student's response to self-regulation statements, thereby estimating his or her susceptibility to having a self-regulated learning deficit. The results show (a) that a student's past responses to self-regulatory statements have a significant impact on the quality of future predictions (b) that the impact of past responses - vs their current performance - is greater when the student is characterized as having a low capacity for self-regulation (c) the more past data we have on a student, the higher the accuracy of the prediction.These results pave the way (1) for adaptive polling by identifying when the model is unreliable, giving them the statements then instead of randomly, (2) for adaptive feedback, by knowing that students that would be more likely to show a deficit, so as to give them remediation.