An "Earlier" Warning System Based on Student Behavior During Class
Tuesday, November 15 11:15-11:45 AM Location: Edison C
Presider: Christothea Herodotou, United States
Abstract: Early warning systems attempt to identify students at risk based on student background and past and current performance. It is possible however that systems based on current performance are too late. Analytics performed over multiple semesters has revealed differences in in-class behaviors that are related to student outcomes. In-class student behavior data fwere combined with background data, student surveys and LMS data and analyzed against student exam scores. The Echo360 data included attendance, participation in activities, correctness in activities, volume of note taking and questions asked. It was found that student exam scores were related most strongly to student participation during class for the courses studied. This has led to the creation of an “earlier” warning system that predicts which students will get a score of 70 or less on the first exam. This presentation discusses the predictability of student failure based on two semesters of independent data.