Moving testing back to the classroom: Solo tests and group tests
As we witness the proliferation of AI, faculty are returning to in-class assessments. Faculty chatter on social media and the sudden increase in blue book sales provide the supporting data (Shirky, 2025).
I have been a long-time fan of a modified interteaching model (Frantz, 2019). Students would be given a list of essay questions based on the chapter to be read for the coming week. They would answer the questions and bring their answers to class. Students would then work in small groups to identify the questions that gave them the most difficulty, and that’s what I would lecture on. Students would then revise their answers and submit them the following week for a grade. I would choose two questions to score, but the students wouldn’t know which two questions, so all their answers had to be good.
Using this pedagogical strategy in the Age of AI would be challenging. I would imagine that I’d ask students to use AI to answer the questions, and then I’d ask students to revise AI’s answers based on their reading of the textbook. Even then, I imagine students feeding PDFs of the textbook into AI and asking AI to critique its own answers.
The next time I teach, I’ll be returning to in-class testing. Frankly, that will probably take the form of multiple-choice tests. I do this with the full acknowledgment that R. Eric Landrum at Boise State University is correct: Taking multiple-choice tests is the one skill that no employer wants (Landrum, 2016). On the other hand, writing skills were once highly valued by employers. I’m not sure that is still the case. AI may not produce the most eloquent prose, but what it does produce may be just fine for most business situations.
However, one skill that I’m sure employers do still value is the ability to work together in teams. The one thing that kept me giving multiple-choice tests a little longer than I really wanted to was the post-test group test that I used, because I found that so much learning occurred through it.
Here’s how it worked in my 2-hour-long classes. Students would first take the test solo. After everyone had submitted their completed bubble sheets, students would receive a brand-new bubble sheet. (I used ZipGrade, so I provided the bubble sheets.) Students would answer the test questions again, but this time they could work together, use their notes, and consult their textbooks. Since everyone had their own bubble sheets, they didn’t have to come to a consensus on the answers. If I were to use this technique in the Age of AI, I would not allow the use of the Internet, so students would not be able to use their e-books. I would, however, bring a few print copies of the book to class for students to use.
While I had some students who would work solo during the group test, most students would work in small groups. If a group couldn’t decide on an answer, they’d ask another group. I recall at least one instance when the entire class discussed a question. For that discussion, I left the room to make it easier for students to share their thoughts without worrying about what I might be thinking.
The greatest value of group testing is that students who understand a concept can explain it to those who don’t. That not only helps the students who didn’t understand the concept, but by teaching it, it strengthens the knowledge of that concept in those who do.
As an added benefit, during a group test, students with strong multiple-choice test-taking skills can model to other students how to think through a multiple-choice question. (There’s a question for anyone looking for a scholarship of teaching and learning research project: Does taking group tests improve a student’s multiple-choice test-taking skills?)
When I did this, my solo tests were worth 50 points and my group tests were worth 25. Because of how much students learned during the group tests, if I were to use them now, I’d probably make both worth the same number of points.
One caveat. Not all multiple-choice tests are created equal. So many of our psychology tests are vocabulary tests. In an upcoming blog post, I’ll share some thoughts on a different approach to multiple-choice tests.
References
Frantz, S. (2019, July 30). Interteaching: Shifting responsibility for learning from instructor to student. Macmillan and BFW Teaching Community. https://community.macmillanlearning.com/t5/psychology-blog/interteaching-shifting-responsibility-for-learning-from/ba-p/6899
Landrum, R. E. (2016, April 27). It’s time: Getting serious about national advocacy for undergraduate psychology majors. Western Psychological Association Terman Teaching Conference, Long Beach, CA. https://westernpsych.org/wp-content/uploads/2016/02/Terman-Schedule-2016-Long-Beach.pdf
Shirky, C. (2025, August 26). Students hate them. Universities need them. The only real solution to the A.I. cheating crisis. The New York Times. https://www.nytimes.com/2025/08/26/opinion/culture/ai-chatgpt-college-cheating-medieval.html