Collaborative Testing

April 24, 2025

April is grueling. Maybe not for everyone, but certainly for me, and probably for many of our students, too. In deference to our limited resources as we wind down another semester, I will keep this Toolbox (relatively) short! 

Collaborative testing is an umbrella term that refers to any method of testing in which students work together to complete all or part of a test. You may have heard the terms dyad testing, paired testing, group testing, double testing, team-based testing, or two-stage testing. The procedure for all is similar: students complete all or some of an exam with a group of 2-6 people, often after first completing all or some of it individually. Instructors who use a collaborative method are generally more likely to think of tests as formative opportunities rather than merely summative.

I use collaborative testing for every exam in my 300-student psychology class. Students have 40 minutes to work in class, individually on a 40-question, multiple choice exam on Canvas quizzes (with lockdown browser), and then, before they submit, they have 10 minutes to discuss questions with as many others as they want (a small handful choose not to participate). Every student has a unique exam because I set Canvas quizzes to draw randomly from a large question bank. This means that it is difficult for students to “answer monger” (“What did you get for question 1?  What about question 2?”). Instead, students who collaborate typically get exposed to about 5 questions they did not have on their own exam, which creates additional learning opportunities through discussion. But, because they only have a short window for discussion, students remain motivated to come to the exam well prepared (anecdotally, I think students also realize that reciprocation is key to having willing collaborators). On average, students are typically able to discuss about 3-5 of their own questions and 3-5 of their collaborator’s questions. Before students begin, I remind them that this time is meant for discussion and learning from each other, so if the TAs and I, as we wander, hear something else, we may ask them to resume independent work. I also give a 3 minute and 1 minute warning so they can wrap up conversations. At that point, I ask anyone who is done to submit their work and exit, so that I can resettle those who need more time. They typically have about 15 minutes to resume independent work at the end.  

With any approach to group assessments, care should be taken to off-set freeloading and motivate students to come prepared. How grading occurs can help with these challenges. One strategy is to average individual scores with the group score. Because group scores are typically higher than individual scores, everyone benefits from the hive, but preparedness is rewarded. Another strategy that further motivates students to come prepared is to weigh the individual score more than the group score (e.g., 60% vs 40%). Another variation is something like a ladder: the better the group score is, the bigger boost to your individual score (e.g., a group grade of 9/10 gets a 5-point boost to the individual score, whereas a group score of 7/10 may only get a 2-point boost; to get a boost at all, an individual score must be above a certain threshold). 

You could imagine other grading schemes that might work for your own class structure. I grade students only once (rather than some combination of individual + group), after they have turned in their completed exam. Students get to choose which questions they work collaboratively on, depending on their own needs (rather than assigning certain questions to the group), and with whom they collaborate. Ultimately, each student can decide if they want to go with the group’s answer or their own, and that’s what is graded for that student. If answers conflict, students have to decide how confident they are in their own knowledge compared to their peers’ (practicing metacognitive skills is an indirect benefit, in my view). 

Below is a curated list of benefits associated with collaborative testing represented in the literature on this topic. If any of these are goals for your courses, collaborative assessments may be a new tool to explore for your own teaching toolbox! 

• Test Anxiety: Collaborative testing reduces test anxiety and especially improves performance for students with the greatest test anxiety. Why? Anxiety inhibits the brain’s ability to learn new information.  
• Repetition: Repeat testing enhances learning (e.g., an individual test repeated with the group). 
• Equity: Collaborative testing especially benefits performance of students from underrepresented groups, but both high- and low-performing students benefit. Why? One reason is because well-prepared students’ learning is reinforced by being the “teacher”; under-prepared students learn something too (also see The Effects of Collaborative testing, Collaborative Group Testing Benefits High and Low Preforming Students, and Collaborative Exams: Cheating? or Learning?). 
• Retention: Collaboration improves retention of course material even after a delay. 
• Test performance: Collaboration enhances test outcomes. Students are much more likely to change an incorrect response to a correct one than the reverse in light of group discussion. In other words, those with correct answers are able to explain, convincingly, why their answer is correct. Students gain practice with evaluating the strength of evidence in this process - another indirect benefit!
• Rapport: Group discussion under highly motivating circumstances like testing enhances peer-to-peer relations and an appreciation for diversity; plus, students generally just like this process!
• Grading efficiency: Instructors can reduce grading workload by administering individual multiple choice tests, and have a single paper turned in for the group portion of short-answer questions. 
• Test-enhanced learning bias: A “group test environment” enhances learning more than a “group discussion” environment. 
• Cheating: Peer accountability and support leads to a reduction in temptation to cheat along with enhanced confidence in testing performance.