Questioning Final Exams

By David Bressoud @dbressoud

In this month’s column, I discuss my own thoughts about how to handle exams. I invite my readers to share their own thoughts here. I will summarize these in a future column.

Last month I mentioned Paul Tough’s new book, The Years that Matter Most: How College Makes or Breaks Us. Chapter 8, “Getting an A,” describes the enormous attention that is required to identify and address the needs of at-risk students, especially those whose background is such that any obstacles or failures lead them to question whether they belong in calculus, a STEM program, or even college.

The story of Ivonne Martinez and how Uri Treisman’s interventions eventually led to her success is very moving and inspirational. Despite repeated failures through the first semester of calculus, she was able to pull it together before the end of the term, earning a high A on the final exam. Because of Uri’s practice of allowing a student’s grade on the final exam to supersede everything else that semester, she had earned a high A for the course. That was certainly well deserved. By the end of the course, she had mastered everything she needed to know. But the ending still left me slightly unsettled.

The problem that I have is how much depends on that final exam. Ivonne was able to pull it together. Some students do. But in my own experience, final exams are a terrible vehicle for assessing what a student has learned and will be able to carry into future courses. The difficulty is that over a short span of days every instructor is giving a high pressure, timed assessment. Students are trying to cram in as much knowledge and memorized facts and procedures as they can. A memory dump after each exam is not only natural, it is a survival mechanism so that one can focus on preparing for the next day’s exam.

A practice that I picked up from one of my colleagues when I was at Penn State is to allow students to redo problems that they have missed on an exam. I realized long ago that it is a mistake to grade on a curve. Even if 60% is a solid B, it does not feel like that to the student who receives it. In addition, the fact that the curve is based on how other students have performed sends the message that one’s grade is based on how one compares to the others in the class, not how much you have actually learned. It is known to increase competition, rather than cooperation, among students in a class. But constructing the test so that everyone doing B work can be expected to correctly answer 80% or better under time pressure means being unable to ask questions that probe understanding.

Allowing students to earn back points if they can explain in full sentences how to get to the correct solution—not just what that solution is—can be an effective learning experience. The first semester I tried this, one student complained that he did not like it because he wanted to be able to forget about that mathematics once he had been tested on it. That convinced me I had to do this. I have adopted this as standard procedure. Most students appreciate the opportunity to improve their grade (there are always a few who refuse to spend any more time on that material), and many praise this structured means to learn from their mistakes. But there are no options for allowing students to improve a grade on a final exam. Because of this, I now seldom have it count more than 10%.

Even though students are enabled to learn from mistakes, I consider timed exams to be problematic. I now build at least three major projects into each course. I find these are much better measures of what a student is learning. Each requires submission of an initial draft to which I respond followed by a final submission. Even with Macalester’s classes capped at 32, this is time-intensive, so I allow group submissions on the first, and usually most problematic, project. If someone considers their group truly dysfunctional, they may submit their own report. But I have not found a good way of assessing each individual’s understandings without requiring individual reports for the later projects. Together with homework (both online assignments testing basic mastery and a weekly set of problems that require written explanations), these make up over half the grade.

Critical but helpful comments on the first draft are essential for the effective use of project reports. Once a grade has been assigned, most students do not pay much attention to what you have written on their paper. And if students have received useful feedback on the first draft, the final submission is usually pretty easy to read and grade. A good rubric that is easy to use is essential for providing this feedback. The one I use is based on that developed by one of my colleagues at Macalester (in Sociology), available here. For instructors with larger classes, I suggest that they consider Peer Assisted Reflection (see my column from December 2016) as a promising technique when the number of students is too large to provide individualized feedback on the first draft.

At least one lesson I hope all readers will take away from this is that there are a lot of good ideas out there, which is one of the reasons why I am anxious to see what others have to say. As each person picks up a promising idea and improves it for their situation, it gets better.

Reference

Tough, P. (2019). The Years that Matter Most: How College Makes or Breaks Us. Boston, MA: Houghton Mifflin Harcourt.