A study of the effect of graded homework in a preparatory math course for electrical engineers

Journal of Engineering Education, Apr 2003 by Trussel, H J, Dietz, E J

abstract

Many person-hours are required to grade homework assignments in large engineering classes. An experiment was conducted to determine whether this effort is beneficial to the students as measured by their test performance. Two sections of the same course were taught by the same instructor. One section was assigned homework that was graded; the other section was assigned the same homework but it was not graded. The experiment was repeated in a second semester. In the first semester, the section with the graded homework had significantly higher test scores than the section without graded homework. In the second semester, the two sections did not differ significantly in their test scores. Possible reasons for this difference are discussed.

I. introduction

It is generally believed that in order for most engineering students to learn concepts that are taught in class, they must work example problems that illustrate those concepts. Homework problems are assigned to provide students with practice solving relevant problems. Similar problems are then given on a test to determine whether the students have mastered the concepts. To give an incentive for the students to spend the necessary time studying, homework is usually graded and that grade is used in determining the grade for the course. It was the intention of this project to determine whether grading the homework is necessary to motivate the students to learn the concepts and thereby improve their test scores.

Grading homework for engineering courses that are problem oriented requires substantial resources. There are various approaches to minimizing this effort including grading only a few problems and using computer generated and graded assignments. If students can learn as much without the instructor or teaching assistants having to grade the homework at all, those resources can be redirected to other parts of the course. For example, teaching assistant (TA) office hours could be increased, which would help many students by giving them more individual assistance; TAs could develop detailed written solutions to the homework or tutorial examples, which would augment the text's explanation of concepts.

Previous research on homework effectiveness is concentrated at the elementary and secondary school levels and tends to favor the assigning of homework [2, 3, 8, 9]. The major explanation given for the positive effect of homework is the time spent on task. These studies claim that to be effective the homework should be checked. However, a small recent study found little effect on test scores of checked homework [7]. Because these studies all involve pre-university students, they do not mention the constraints on and resources available to college level students. Given the fundamental differences between university and pre-university students, we believe that few conclusions regarding homework at the university level can be drawn from these studies of younger students.

A recent study on university students compared the effectiveness of homework submitted and graded online with that of handwritten homework graded by hand in a physics course [1]. The study found no significant difference. This does not address our question on the effectiveness of grading homework, but it does indicate the interest of the university community in such questions.

It is important to note that what is in question in this study is not whether working homework problems is beneficial to students. Most instructors would agree that often more is learned by working problems than from the class itself. The question is whether students must be rewarded or coerced to do the homework that is necessary for them to learn the material.

An opportunity to conduct an experiment presented itself when the first author developed a sophomore level course, ECE220, Foundations in Electrical and Computer Engineering. This course was designed to provide the mathematics needed for our sophomore and junior level courses in circuits, signals, and systems and electromagnetics. A lecture syllabus and a course overview are available via the Web [4, 5]. The course was taught in two sections, and both sections were to be taught by the author. It was decided to grade homework in one section but not in the other. The course is required of all electrical and computer engineering (ECE) majors, so a large number of students would be taking the course each semester.

Both an advantage and a disadvantage of using this course was that it was introduced and first taught by the author. Thus, the first semester, students had no access to old tests or homework assignments. Unfortunately, a course can only be taught for the first time once. The second time the author taught the course, a history had developed. It is suspected that this history affected the outcome of the second experiment. The course was first taught spring semester 2001. It was taught by the author again in a ten-week summer session in 2001. The summer session consisted of only one section and had required homework. The second graded homework experiment was in fall semester 2001. Thus, the fall 2001 classes had the benefit of two previous offerings.