STUDENT SUBMISSION PATTERNS IN ONLINE HOMEWORK AND RELATIONSHIPS TO LEARNING OUTCOMES: A PILOT STUDY

Abstract

Online homework systems are being increasingly used for auto-graded, instant feedback homework and practice for students in math, science and engineering. Students may use these systems, which often allow multiple or unlimited tries, in ways that are different from completing traditional paper-based homework, however research relating online homework system patterns of usage and learning outcomes is limited. This study explores online homework submission patterns and their links to student learning outcomes (weighted individual grades) by analyzing the submission patterns of two second-year engineering courses (~130 students each) from our institution over the 2017-2018 academic year using WeBWorK, an open online homework platform. Students in each of the two courses were clustered into three groups using a K-means algorithm based on when during the homework period they tended to submit attempts. Clusters were used to approximately represent a submission pattern, meaning groups of students that submit attempts mostly early, mostly late, or more evenly over the period. Conducting one-way ANOVAs for each course, we found that there is a significant difference between clusters (submission patterns) in terms of mean individual weighted grades on tests and exams (p < 1.07e-08, p < 2.68e-5). Post-hoc analyses revealed that the best performing cluster (students who submit attempts mostly early) had a mean tests/exams grades that were about 10% higher than worst performing cluster (students who submit attempts mostly late) (p < 2.6e-06, p < 9.9e-05).