Turn an exam into a learning experience with two-stage exams

Turn an exam into a learning experience with two-stage exams

What's a two-stage exam?  Here's an explanation from my forthcoming paper, Physics exams that promote collaborative learning, with Georg W. Rieger and Cynthia E. Heiner:

The two-stage exam is a relatively simple way to introduce collaborative learning and formative assessment into an exam. Their use is rapidly growing in the physics department at the University of British Columbia, as both students and faculty find them rewarding.  In a two-stage exam students first complete and turn in the exam individually, and then, working in small groups, answer the exam questions again.  During the second stage, the room is filled with spirited and effective debate with nearly every student participating.  This provides students with immediate targeted feedback supplied by discussions with their peers.

In the classic exam, students are intensely engaged with the material, but they lose the opportunity for formative assessment, because the feedback is mainly right/wrong and comes a long time after the exam.  By contrast, in a two-stage exam, students receive immediate, specific feedback and increase their mastery.

In our two-stage exams, students participated strongly in the discussions, and their reactions were overwhelmingly positive.  Even those who found the discussions uncomfortable, because they saw where they had made mistakes, acknowledged that they learned what they needed to learn.

Read the paper to see how to implement two-stage exams.  Although it was written for a physics-specific journal, Physics Teacher, it outlines pretty general techniques you can adopt in your classroom.


From Jane Maxwell, Department of Chemistry, University of British Columbia: I was inspired to try a two-stage review with a larger (~120 students) third-year class, CHEM 311. Overall it went really well! Carl suggested I post a summary here in case any of you are considering this for the future.

  • CHEM 311: Instrumental Analysis is a required course for students majoring in Chemistry or in Biomedical Lab Sciences
  • Students all have a common prerequisite (CHEM 211), but the emphasis of that course is quite different in 1st and 2nd terms (depending on the instructor)
  • Not all students have studied Electricity & Magnetism due to a gap in our prerequisites (BMLS students and CHEM majors without Physics 12 will not take PHYS 102)
  • This variation in students’ background knowledge was a major concern for the instructor. Based on this thread, I suggested the two-stage review activity as a way to simultaneously address the following goals:
  1. get a snapshot of students' understanding of key concepts that can be used to tailor lectures and activities
  2. provide students with immediate feedback on their background knowledge (via the group stage) and some clarification on some topics from group discussion - the use of IF-AT cards also provides feedback to the group when it is wrong
  3. give students a low-stakes opportunity to experience and practice for the two-stage midterm

The instructor was enthusiastic about the idea, and was happy to devote time on the first day of class, which he considered to be generally unproductive time.

  • The instructor, lecture TA, and I identified the topics and key concepts that we wanted to include in the review activity - generally topics/concepts from 2nd-year prerequisite course and 1st-year physics (E&M)
  • I developed a set of 18 multiple-choice questions, with feedback from instructor and lecture TA
  • The questions were targeted at a "quiz" level rather than "final exam" level
  • The instructor and I introduced the activity after the standard discussion of the course syllabus and a brief introductory lecture
  • Groups of 5 students were assigned randomly by the instructor in advance. Students rearranged into their groups during a 5-minute break before starting the activity
  • Students answered the questions individually on a scantron sheet
  • After 15 minutes (due to time constraints) we collected the scantrons, and the groups redid the test using an IF-AT (scratch) card
  • All cards collected at the end of the class (approx 15 min for group stage)
  • We repeatedly emphasized that this was not a graded activity, but a review intended to benefit them and the flow of the class

  • I was initially worried that a scantron-based "activity" (QUIZ) on the first day would seem draconian, but the students accepted the activity with very little grumbling. Attitudes during and afterwards seemed positive (One student told me "That was fun!")
  • Most groups appeared to engage in good discussions about concepts - lots of hand gesturing, etc.
  • Students required direct prompting to sit in a formation conducive to group discussion - some groups ignored initial instructions to do this and left some group members excluded from the conversation
  • Despite the fact that we were rushed, the majority of individuals and groups completed the activity
  • Two groups stayed late to complete the activity, even though we told them it was ok if their cards were incomplete (since we ran out of time)
  • I observed one group exchanging contact information to form a study group

  • I’m very glad that we did use the scantrons in the end - the results were VERY informative, as was the degree of improvement by groups – eg. when can they self-correct?
  • Setting groups is, I believe, very important to avoid clique-y behaviour within cohorts, to help students experience the diversity of expertise among their peers, and to create new connections. We used randomized groups, but of course you could also use existing data to maximize heterogeneity, or create groups on the fly using semi-random variables such as day or month born, last digits of phone number, etc.
  • We posted the solutions on the class Connect site, along with some suggestions for resources/tutorials to help students get up to speed. In the future, I would be interested in using online gradebook features to give each student individualized feedback based on their scantron results.