Flexible Assessment Practices in Duke STEM Courses

The last two years of pandemic teaching, and in some cases experiences even before that, have resulted in many Duke faculty reconsidering their assessment practices and policies with the intent of being more flexible, more equitable, and more focused on assessment for learning. Thanks to Victoria (Tori) Akin, Owen Astrachan, Charlie Cox and Mark Spaller for sharing their experiences here; feel free to send us your example

Tori Akin, Assistant Professor of the Practice, Mathematics

Dr. Tori Akin headshot

Historically, I’ve preferred my tests to be timed, in-class, closed book, with a page of handwritten notes. During remote instruction, in Math 230 my tests were a combination of timed and untimed. They also were open resource. I like open resource tests, because it de-emphasizes memorization and encourages me to write questions that involve problem solving. Students also seem to perform better on tests with more time and more resources, even when I ask challenging questions. I think this type of thinking better prepares students for future jobs/challenges in which they will be able to access a wealth of resources, and they will need to interpret and apply information correctly. I do believe that a certain amount of memorization is helpful for problem solving because it frees up space in working memory. This is why I have included timed sections on remote exams, asking students to do some quick recall/processing.

Owen Astrachan, Professor of the Practice, Computer Science

Dr. Owen Astrachan headshot

In Compsci 201 for Fall 2020, we adopted an approach of having several smaller midterm exams rather than two midterm exams as we had done for the past 20 years or so. We termed these “distributed midterms” (DM) at that time. We continued this approach in Spring 21, again in Fall 21, and [are doing] so again in Spring 22. This is not a departmental initiative, but rather one adopted by those teaching Compsci 201, and also by professors in Compsci 216. We have used from three to six exams in general, each takes place during class time, but is not in person, and there’s a time limit of 35-40 minutes. The total time on these midterms is roughly the same as when we used two full-class midterm exams.

We also give a retake of each exam, same topics, but different questions. These “redux” exams do not take place during class time. All exams are completed online via Gradescope. Students get the best score of the two attempts as the score that counts for that exam, and exam retakes are the norm. This approach has lessened stress for students since the exam is shorter and since there is a retake. For each exam we also provide a practice exam that illustrates the topics about which questions will be asked. Comments in course evaluations indicate that this approach absolutely lessens stress for students. Our undergraduate TAs have grading to do nearly every week, but we’ve adopted an approach that distributes this workload and the TAs have been great at getting exams graded in a timely way so that students have a grade that helps them decide if they should take the redux exam.

With three semesters of these now done, the previous exams work generally as practice exams, so creating those is straightforward. We changed the kinds of questions we use since these are taken via Gradescope with open Internet/notes.

Charlie Cox, Associate Professor of the Practice, Chemistry

Dr. Charlie Cox headshot

I use two weighting methods for assignments in my chemistry courses. The first method weights the mid-term exams and final similarly, with the final having a slightly higher weight. The second method weights the mid-term exams less and the final exam more significantly. The goal of the two weighting structures is to give students different timelines for demonstrating mastery. The first timeline rewards students for grasping content more quickly. The second time rewards students for grasping content by the end of the term. Because the timeline for mastery is important, every mid-term counts toward the final average, but the weighting is different.

Mark Spaller, Associate Professor of Chemistry and Biochemistry, Duke Kunshan University

Dr. Mark Spaller headshot

Having used a traditional grading scheme for years, for the first time I adopted a specifications grading (SG) approach, implementing it in my “Introduction to Biochemistry” course last term. There were several motivations, but a major one was to eliminate the ‘race to collect points’ mentality of the students—and the dysfunctions that can lead to—and to create an environment where the focus was on achieving course objectives by mastering the assignments to which they were linked.

SG resists quick summarization, but here’s a synopsis of my inaugural experience. All assignments were pass/no pass, and each was based on defined course objectives. The outcomes (passing or not) made it clear to the students, but also to me (regarding the efficacy of my instruction), where improvement was needed. I used weekly quizzes drawn from a larger question pool as my primary assessment vehicle, and students were allowed multiple retakes in order to pass each one, thus demonstrating their mastery of specific learning objectives.

Final course grades were linked to successfully passing a defined number of quizzes (which I divided into basic, intermediate, and advanced levels), as well as two other assignments: a special topics report and a final exam. Students were informed from Day 1 what combination of passed assignments (known as “bundles” in SG speak) was required to achieve their desired grade, and could plan accordingly. Consequently, I could teach the course without sacrificing rigor, knowing that the bar for each letter grade could be adjusted by appropriate design of the required bundles. Each bundle (grade) posed a different level of challenge, and students could select their path based on the amount of time and effort they were willing to dedicate. Grades were uncoupled from class medians, high scores, or standard deviations, so students knew they were not in competition with one another, which helped reduced distraction from the main goal of focusing on learning.


These are just a few examples of the thoughtful ways Duke and DKU faculty are implementing flexible assessment practices in their courses. If you would like to share an idea or practice from your course to be highlighted in a future DLI blog post, please contact us