For the previous year, I taught a computer engineering class. When I came in, the class was heavily backloaded – that is to say, there was a disproportionate amount of work and student time investment concentrated into the last few weeks of the course schedule.
Backloading seems to be a pretty common situation in classes in general, as projects, final papers, and end-of-semester exams are all crammed together. And for students taking multiple classes, these effects tend to stack, giving them multiple major assignments and tests crammed into a few days. This should alarm instructors for at least two reasons.
- Stress, insomnia, and time constraints may cause students to underperform and reduce the reliability of our assessments just as we are assessing work that carries the greatest weight in most grading schemes (final projects, papers, and exams).
- These heavy load periods may cause otherwise talented students to abandon our discipline (or their educational career in general). In their 1997 book Talking About Leaving, Seymour and Hewitt survey students who have given up on science-based majors. Over one third of respondents cited course “overload” as the major reason for giving up their studies.
The backloading problem in my class prompted me to think about how I could eliminate some of this assessment load at the end of the semester. Among other changes, such as trying to spread projects across more of the semester, I decided to make a more radical change: I made the comprehensive final exam completely optional.
I came to the conclusion that the mandatory final exam ought to go based on 2 hypotheses:
- If a student can demonstrate mastery of a topic on a midterm exam, they do not need to be tested on it again.
- Some students will improve their mastery between the midterm exam and end of the semester and they should be retested if they wish to be.
Just to be clear, when I talk about final exams here, I’m referring to comprehensive exams that cover material that students have already been tested on at least once.
Let’s start with the first hypothesis. It’s possible that some students will retain the information until the semester regardless of whether we test them. For these students, the exam is meaningless, although it can offer some increase in reliability to see two similar scores. Other students will have forgotten information or gotten rusty with their skills. I argue that these student do not need to be tested either. Why? In many cases, all we really care about is whether a person is capable of acquiring the skills and knowledge as needed. If the student has proven they can do it once, then they most likely can do it again, whether it is for a final exam or some task they are given once they have gotten a job. Secondly, any exam, whether we give it in the middle of the semester or at the end, is just a snapshot of the student’s current ability level. When we give a student an A based on their final exam score, we also cannot guarantee that they will still have that level of mastery in a year’s time. The only real difference is that the final exam score is a month or two fresher than the midterm exams. Basically, we care about the peak mastery the student has proven they are capable of achieving, not their current level of mastery.
As for the second hypothesis, it is fairly straightforward. Unless we have a specific reason for the grade we give at the end of the semester to chart the student’s progress at early milestones, then we ought to give them a chance to be retested if they think their skills have significantly improved. The ideal, but impractical, approach would be some sort of continuous testing scheme, where students could take the test as many times as they wanted until they were able to achieve a satisfactory score. However, it should be practical in most cases to at least give them a 2nd chance.
The concept of an optional final exam adheres to both of these principles. It also has an added benefit of giving students additional incentive to perform well on early exams (front-loading their learning) without the problem of assigning unpalatably high static weightings to early progress.
In the scheme I implemented, I simply told students that they had the option to take the final exam if they thought they could do better on it than their existing exam score. If they chose to take it, it was averaged into their exam scores. If they chose not to, their exam scores stood unchanged.
Now, from my arguments above, it would seem to make sense that I should allow students to make the final exam worth 100% of their exam grade. After all, if they are at their peak mastery, why should earlier results even matter in my final assessment of their grade? Another approach would be to make everyone take both the midterm and final exams and then keep whichever grade was higher. These are both rational arguments, but I chose not to implement them for a few reasons. The idea of making the final exam worth 100% of their exam grade seems problematic because I want to make sure that the results of the early exams have consequence for formative reasons. I worry that if students knew they could completely replace their early exam score by doing well on the final that some students would not be encouraged to learn the material early and this would have a detrimental effect on their ability to keep pace. Just another example of how we sometimes sacrifice assessment accuracy for teaching ends.
Making the final exam optional met with generally enthusiastic support from my students. I’ll be collecting and analyzing data on the experiments impacts on the actual course grades over the next few months with the intention of publishing them at next year’s ASEE.