Dear Education Committee, I’d like to know more about alternatives to high-stakes exams in STEM courses. What are the pros and cons? Do they serve students? Does shifting away from exams create more work for faculty? Can they be used in the online format or only once classes are back in person?
—Asking About Assessment Alternatives
Ideally, exams help educators assess student learning and also help students consolidate their knowledge of new course material as they study. In this perfect world, exams are objective and questions are unbiased. Despite mounting critiques of their usefulness and fairness, multiple-choice exams remain the norm in large enrollment courses because they are perceived as objective and they are easy to score.
So are we stuck with multiple-choice high-stakes exams as a necessary evil? After all, students need to excel at high-stakes exams they may take in the future, right? In this column I consider three widely used alternatives to high-stakes exams: collaborative testing, mastery-based testing, and project-based assessment. I chose these methods because they can be implemented in classes with a range of enrollment sizes and are adaptable for different teaching modalities.
What’s the idea? A conventional test assesses if the student already knows something. Collaborative testing aims to make the test part of the learning experience. As the name suggests, groups of students work together to complete an assessment. The entire assessment may be collaborative or part may be completed individually.1 Most studies have focused on the latter model, also known as “two-stage” or “two-step” exams, which include both individual and collaborative components. Students first complete an exam solo and then are immediately retested on the same topics, working collaboratively. The grade each student receives is a weighted combination of group and individual scores. A student may answer a question incorrectly on an exam but then answer correctly as they work in groups.
Does it work? All the studies consulted for this article reported increased scores on group tests compared with the individual tests. However, it’s less clear if collaborative testing improves student learning over the long term. If a student worked through a question both individually and as a group, would they do better than students who only took an individual test if they saw the question again later (e.g., on the final exam)? The cohort of “yes, collaboration helps retention” studies is matched by an equal number of reports reaching the opposite conclusion. Why this lack of consensus? Ultimately, concept retention in collaborative exams may depend on the question format, and latency to retesting, group size, and group longevity.2 Although experimental design, cohort size, and level and type of course (undergraduate, nursing, medical) vary considerably among studies, almost all accounts report that this method is well received by students.
A concern about the “free rider problem,” in which students do not contribute to the group and still get the benefit of the correct answer, is a significant barrier to adopting collaborative testing. However, a recent study by Jang and colleagues (2017) provides some reassurance on this front.3 Following a cohort of 67 students in a calculus-based physics class, they found that students were more likely to arrive at a correct answer in collaborative groups, even if none of the students had gotten the correct answer in the individual round. Another potential concern for educators might be that this approach would benefit weaker students, but not stronger students. Jang et al. also found that collaboration improved scores at all performance levels (although low-performing students showed the greatest increase), including the highest performing member of each team.
Ease of implementation. How much work? It depends. Question format and the type of collaborative testing will each affect the work required to implement this approach. For instance, if the exact same exam is given, first solo and then collaboratively, and the exam is multiple choice, and each group submits one test for the collaborative part of the session, then additional workload is minimal but also there is an increased probability of free riders who simply record the correct answer but do not learn the concept being tested. Workload increases if choices are made to use open-ended questions or give a similar but not the same exam for both individuals and groups.
Modality. Collaborative testing can work for in-person, hybrid, or synchronous online courses. Coordinating collaborative exams in asynchronous online courses presents a substantial challenge.
Mastery-Based Testing (MBT)
What’s the idea? Collaborative testing, while great, still mimics the one-and-done structure of a traditional test. Without some future accountability (test corrections or a final exam), students don’t have an incentive to revisit topics they didn’t fully understand. In contrast, in a mastery-based learning paradigm, students earn points only after demonstrating that they have learned a concept or skill, rather than simply completing an exam.4 A student may take an exam on the same topic as many times as they like until they have mastered the corresponding material. This strategy seems well suited for STEM courses, particularly foundational STEM courses, because we all want our students to master foundational material before they move to upper-division courses. In addition, students come into foundational STEM courses with widely varying preparation, and the mastery-based approach allows students who start at different places to have different amounts of time and practice to be successful and master the material. Other benefits include developing students’ growth mindset (i.e., you can get better with practice!). Since students can control when they undertake a (re-)assessment, MBT seems especially suited to pandemic learning—generally minimizing the harm of compromised memory and attention that have been observed during the pandemic and allowing students some flexibility as they continue learning through trauma or illness.
Does it work? Yes. Students cannot pass the class unless they have met the learning objectives, and flexibility for students is built into the structure of the grading scheme.
Ease of Implementation. This approach means that students complete exams as many times as they need to, which in my view seems like a grading nightmare if not automated. One workaround could be limiting retesting to particular weeks to minimize the grading workload as well as simply grading pass/fail. In the long run, automatically graded exams seem more tractable, although there would be a substantial investment to create multiple assessments for retesting on the same material. One consideration is whether students can earn any partial credit toward mastery. Another question is whether particular course material may be more or less appropriate for re-testing for mastery. For instance, if some material is critical for future courses, a mastery approach may be more justified than for material that is unique to that particular course.
Modality. MBT could work well for in-person or online courses.
Projects as Alternatives to Exams
What’s the idea? Problem-based learning allows students or student teams to research and solve complex problems. Projects can be implemented throughout curricula and in both laboratory and non-lab STEM learning contexts, and are typically assessed through a finished body of work (e.g., report, presentation, poster) and milestones along the way (e.g., drafts of project components, notebook or journal entries documenting progress). Some benefits of this approach include increased student engagement, sustained learning, and communication, and the approach fosters collaboration as well.
Does it work? With so much flexibility in project-based learning, efficacy depends a lot on the design and implementation on the instructor side. Clear criteria about what makes a good project can help clarify an instructor’s goals in bringing project-based learning into their courses.
Ease of Implementation. Two challenges associated with projects for assessment are the workload of giving feedback and grading and consistency and fairness in grading.5 If students work in groups to complete projects, this can present an additional challenge of equitable participation.6 For instance, Wright and Boggs implemented projects in an advanced, elective cell biology course, in which student teams completed a semester-long project on human disease that was assessed by instructor and peer review.7 Equitable participation was structured by assigning students to roles of administrator, artist, communicator, and expediter (who can assume any of the previously listed roles based on the needs of the group) based on their own perceived strengths. Were the students learning as much? The students who participated in the project course performed similarly in a follow-on biochemistry course as a previous cohort of cell biology students who were assessed with exams instead of projects.
Modality. Projects and assessments can be adapted for distance learning with some effort to make sure students are able to work together and share their progress and products remotely.
The pandemic provides a catalyst for educators to change course structure to be more equitable and supportive of students, including shifting away from high-stakes exams. Many instructors are trying new formats of assessment because of the need to meet students where they are and support their learning. This need is highly visible during a pandemic, but only highlights the longstanding reality that even during a typical semester students often encounter situations which prevent them from doing their best at exam time. For faculty who have pivoted in-person courses to an online format and now look forward to meeting with students again face-to-face, these methods present a way to improve student learning without substantially increasing grading burden and to recognize that students will continue to need flexibility in the coming months and beyond.
Note The author thanks Shannon Seidel and Erin Dolan for generous feedback and contributions to this article.
References 1Efu SI (2019). Exams as learning tools: A comparison of traditional and collaborative assessment in higher education. College Teaching 67, 73–83. doi: 10.1080/87567555.2018.1531282. 2Cooke JE, Weir L, Clarkston B (2019). Retention following two-stage collaborative exams depends on timing and student performance. CBE—Life Sciences Education 18, ar12. doi: 10.1187/cbe.17-07-0137. 3Jang H, Lasry N, Miller K, Mazur E (2017). Collaborative exams: Cheating? Or Learning? American Journal of Physics 85, 223–227. doi: 10.1119/1.4974744. 4Collins B, Harsy A, Hart J, Haymaker KA, (Armstrong) Hoofnagle AM, Janssen MK, Kelly JS, Mohr AT, and Jessica Oshaughnessy J (2019). Mastery-based testing in undergraduate mathematics courses. PRIMUS 29, 441–460. doi: 10.1080/10511970.2018.1488317. 5Tierney RD, Simon M, Charland J (2011). Being fair: Teachers’ interpretations of principles for standards-based grading. The Educational Forum 75, 210–227. doi: 10.1080/00131725.2011.577669. 6Wilson KJ, Brickman P, Brame CJ (2018). Group work. CBE—Life Sciences Education 17, fe1. doi: 10.1187/cbe.17-12-0258. 7Wright R, Boggs J (2002). Learning cell biology as a team: A project-based approach to upper-division cell biology. Cell Biology Education 1, 145–S27. doi: 10.1187/cbe.02-03-0006.
About the Author:
Alison Dell is an associate professor of Biology at St. Francis College in Brooklyn, NY and member of ASCB’s Education Committee. Twitter @dell_alison