Analysis of Student Learning in High Enrollment Computer-based College Courses

14 Aug
Wednesday, 08/14/2013 9:00am to 11:00am
Ph.D. Dissertation Proposal Defense

Gordon Anderson

Computer Science Building, Room 151

Computer aided learning is ubiquitous at the college and university level. Over the past ten years, instructors have turned to computer-based Learning Management Systems (LMS) to deal with high-enrollment courses of 100 to 1,000 or more students for both on-campus and on-line courses.

Historically, educators have held assumptions about the nature of successful learning behavior, so-called "good learning virtues," which, when followed, lead to the optimal learning experience. Examples of these virtues include: Timeliness, Thoroughness, Persistence, Planning, and Resiliency. A key unanswered question is if these assumptions match the reality of actual student behavior: are these learning virtues actually beneficial for learning, and if so, are they beneficial for all students?

Since an LMS can record student interactions with a course's content, we have an opportunity as never before to analyze student behavior in an attempt to learn how students are learning in these high-enrollment, computer-aided coursesĀ and what effect this behavior has on outcomes.

This thesis takes the form of a series of studies that attempt to address specific questions about student learning in large courses that use an LMS. We studied data from two high-enrollment courses in different domains: Computer Science, approximately 350 students/semester, and Chemistry, approximately 1,200 students per semester for five semesters. To study Planning, we measured the extent to which students engaged with preparatory material, such as the textbook, before working on homework problems. We have found that students who followed this pattern get more homework problems correct and score significantly higher on exams. This is especially true for students with no previous experience with the topics covered in the course. To study timeliness, we measured the amount of assigned work that was done along with its proximity to the assignment due date ("procrastination"). We found a negative effect on exam scores for students who did the majority of their work within ten hours of the due date, although higher performing students were less affected by this pattern. We have also found that students who frequently jump from one homework problem to another had lower than expected outcomes.

Based on these preliminary findings, we conclude that the "good learning virtues" seem to hold true, but not for all students. We plan further to develop our analyses to strengthen a causal argument for the effectiveness of our measures of planning and timeliness on learning. We intend to identify more precisely the students who are most vulnerable to lapses in time management and planning.

Our results are meant to be applicable to courses in any STEM discipline using an LMS with content components similar to the introductory courses we studied. These results will be of great use to educators, content authors, and system designers in evaluating student progress and implementing and targeting effective interventions for the students who can most benefit from them.

Advisor: Robert Moll