Peer Instruction – Eric Mazur’s approach

Peer instruction

What is it?

Peer Instruction is an approach to teaching difficult concepts which students often misunderstand – bottlenecks in learning which students must pass through if they are to progress further. Students are assigned preparatory work in advance of the session. During the session they respond to a question (with discrete options and a correct answer) twice – first individually, and then after trying to persuade a fellow student with a different response. Peer Instruction was developed by Eric Mazur, Professor of Physics and Applied Physics at Harvard. It is closely associated with the Flipped Lecture approach and is particularly well-researched in physical sciences, life sciences, and maths (Knight and Brame, 2018).

Why do it?

Difficult concepts are challenging to teach because, as an educator, once you have yourself mastered a difficult concept your perspective changes in ways which cannot be unlearned (Cousin, 2006) and you may begin to take that concept as self-evident. This means that experts often find it difficult to return to what used to elude them in the early stages of their study, while in contrast peers who have only very recently grasped a concept often teach it well.

Mazur describes how he accidentally hit on this approach in the 1990s.

“I remember at one point trying to explain something to my students that I thought was completely trivial. I took two minutes, made some sketches on the board … I could see at once from my students’ faces that they were confused, so I asked if they had a question. They were so confused that they could not articulate a question. … I tried it a different way, I added equations, and after eight minutes of what I thought was an absolutely brilliant explanation the entire blackboard was covered with equations and drawing, I turned around and the students looked even more confused. … In a moment of despair, I said “Why don’t you discuss it with each other”. And something happened in my classroom that I’d never seen before. The whole classroom erupted … and in just two minutes they had figured out what the correct answer was.”

In their review of the research literature, Vickrey and colleagues (2015) found that when used in a research-based way, peer instruction is associated with overwhelming improvements in students’ ability to solve conceptual and quantitative problems.

How to set it up

Before the session

  1. Pick a concept that students find difficult or misunderstand, and which is key to deep learning and progression. This might be something that you have seen was poorly understood in a previous session, or in formative assessment
  2. Set preparatory work for students to complete before the session. This may be reading, watching a video, or carrying out a task.
  3. Test students on the concept. This motivates engagement with the preparatory work and activates their knowledge. It also gives you feedback in advance about where the misconceptions lie – the basis for the Peer Instruction question(s) you will ask during the session.
  4. Based on this advance feedback from students, prepare question(s) you will use during the session. Peer Instruction questions tend to have discrete options and a single correct answer – see Considerations for some possibilities. They often elicit the application of theories, evaluation of arguments, or other higher-level learning.
  5. Select a classroom response tool for students to use when you pose your question(s), and set up your question. At King’s every staff member can have an uncapped account at PollEverywhere allowing students to respond on any networked device, but there is no detriment (Vickrey et al, 2015) to using low-tech options such as coloured cards with letters, or Plickers. Prepare your question(s) there.

During the session

  1. Give a mini-lecture on the concept which doesn’t cue students about the question you’re going to ask.
  2. Pose the question and ask students to think individually about the answer for a few minutes. This wait time is important (Vickrey et al, 2015).
  3. Then ask them to respond using the tool you chose above. Rather than stipulating an absolute time to respond, Vickrey and colleagues (2015) recommend allowing 80% of students to respond and then giving final notice.
  4. Check the responses – but don’t show them to students yet since that would divert the discussion and influence subsequent responses (Galloway, 2017; Vickrey et al, 2015). If 30-70% have the right answer, then proceed to the next step (see Considerations below).
  5. Ask students to find somebody nearby who chose a different response, and try to convince each other. Prompt students to explain their reasoning.
  6. After students have justified their respective choices, ask them to respond to the question a second time.
  7. In a plenary discussion either explain the correct answer yourself or invite students to do it. At this point you can show students the change between the first and second attempt, so they can see the different that peer interaction made.
  8. Repeat the cycle with your other questions.


The in-session questions need to elicit, confront and resolve students’ difficulties and misconceptions, and require them to apply what they have learned. Life Sciences Education summarises the form these questions could take as logistical, recall, algorithmic, or conceptual. Conceptual questions may be application questions, case-study questions, or procedural questions, among others. Formerly these have been restricted to multiple choice questions, but technology has expanded the possibilities to include, for example, clicking on a region of an image.

The advance test questions need to be appropriately challenging, address the main ideas of the session, and impossible to answer by skimming. You may also want to ask students to indicate how confident they are in their response. The Quiz activity in KEATS is a good testing platform. Some educators allocate marks for this stage.

If fewer than 30% or more than 70% answer correctly first time, it is impossible to organise the Peer Instruction. So instead, educator can initiate a plenary and provide an explanation, and then move on to a question at a more appropriate level.

The reason not to display the chart of responses before the closing discussion is that it would may interfere with divergent thinking by biasing students towards the most common answer, and generally interfere with exploring the problem (Galloway, 2017).

You may need to remind students not to confer before they have answered the first question.

The Peer Instruction stage may only last a few minutes – observe the room to see whether the discussion is still productive or whether it is time to ask the question again.

Examples and resources

  • This comprehensive, evidence-informed guide from CBE Life Sciences Education engages with some of the challenges, including unproductive discussion. Summarises key research articles about Peer Instruction.
  • Ross Galloway blogs about Peer Instruction in his Physics lectures at Edinburgh University, 2017.
  • In this short report of a 2018 joint study in medical education, Nephtali Marina-Gonzalez (UCL) found that the technique  allowed him to scale up teaching to large group settings without any decline in students’ achievement in exams. Manuel Hernandez-Guerra and Enrique Quintero (Universidad del la Laguna, Tenerife, Spain) confirmed that Peer Instruction improved students achievement compared to a traditional lecture.
  • Eric Mazur tells the story of his discovery.
  • Eric Mazur demonstrates peer instruction in action:


Cousin, G., 2006. An introduction to threshold concepts. Planet.

Galloway, R., 2017. Peer instruction. Teaching Matters blog, University of Edinburgh.

Knight, J.K., Brame, C.J., 2018. Peer Instruction. LSE 17, fe5.

Vickrey, T., Rosploch, K., Rahmanian, R., Pilarz, M., & Stains, M. (2015). Research-Based Implementation of Peer Instruction: A Literature Review. CBE—Life Sciences Education, 14(1), es3.


1 Trackback / Pingback

  1. Join us to discuss peer instruction on Monday 7th December at 2pm GMT | ACM SIGCSE Journal Club

Leave a Reply

Your email address will not be published.