Measuring the effect of on-line homework procedures on student exam performance

Abstract

One of the consistent challenges in education is finding the right mix of assessment tools to both encourage learning and provide an accurate evaluation of students. This challenge becomes even more acute in quantitative courses where students often are tasked to complete a significant number of problem sets to develop their skills. The use of course management software has provided a new method to address some of these challenges. On-line assessments can provide learning and assessment tools that are less labor intensive for the instructors and provide quicker feedback for the students. Using on-line assessments as direct replacements for traditional pencil and paper homework assignments may not, however, take full advantage of the technology. The purpose of this research effort is to explore whether on-line homework procedures and other background data about the students had a measurable effect on student exam performance. To improve the sampling for the study, multiple sections of an undergraduate Engineering Economy course were studied over multiple semesters. This paper discusses preliminary results obtained from analysis through Exam #1 for 140 students across three sections of Engineering Economy from a single semester. The variables studied include the number of attempts at an assignment, time between attempts, time between first attempt and the deadline, performance on individual homework attempts, first attempt score, maximum score achieved, and average score achieved. Student demographic data, such as total credit loads, number of semesters at the university, transfer student status, current college major, and prior GPA were also reviewed. Other confounding factors were also reviewed, such as attendance at problem solving and recitation sessions. These variables are then compared with student performance on individual exams. The goal of this research is to determine which, if any on-line homework variables have a positive impact on student exam performance. It is hoped that the results of this research, along with parallel efforts to evaluate the impact of other technologies, such as clickers and financial calculators, can be combined to provide a more effective educational experience to prepare students to become practicing engineers. The preliminary results using logistic regression found that the probability of a student earning an Exam #1 score of 80% or higher was negatively affected by transfer student status and positively affected by recitation attendance and the first attempt score for the fourth homework assignment. © 2012 American Society for Engineering Education.