Most teaching asks students to sit still and absorb. Peer instruction asks them to turn to the person next to them and defend an answer out loud. That one change, from listening to explaining, is what makes it one of the best-evidenced teaching methods in higher education.
This guide covers what peer instruction is, the Harvard data that shows how well it works, and five methods you can run this week, whether you teach in a lecture hall or build online courses.
What is peer instruction?
Peer instruction is a teaching method where students learn a concept by explaining it to each other, not just by hearing it from an instructor. The teacher poses a conceptual question, students commit to an individual answer, then they discuss it in pairs or small groups and answer again. The act of arguing a position and hearing a classmate argue a different one is where the learning happens.
The method comes from Eric Mazur, a physics professor at Harvard University. In the early 1990s he noticed that his students could pass exams yet still hold basic misconceptions about how physics works. So he stopped lecturing straight through and started breaking his classes into short cycles of question, discussion and re-vote. The approach spread from physics into biology, chemistry, engineering, economics and beyond.
Does peer instruction work?
Yes, and the evidence is unusually clear. Mazur and Catherine Crouch published ten years of results from his introductory physics courses in the American Journal of Physics in 2001. They measured conceptual understanding with the Force Concept Inventory, a standard test given before and after a course, and reported the normalised gain: how much of the available improvement students captured.
The semester before Mazur switched to peer instruction, his normalised gain was 0.25, in line with a traditional lecture. The first semester he used peer instruction, it jumped to 0.49, close to double. As he added more reforms over the following years, gains climbed as high as 0.74.
That result is not a one-off. Richard Hake's 1998 study of more than 6,000 introductory physics students found that traditional lecture courses averaged a normalised gain of about 0.23, while courses using interactive engagement, peer instruction among them, averaged about 0.48. Two different datasets, the same finding: getting students to reason and argue roughly doubles how much of a concept they retain, compared with watching someone explain it well.
Why peer instruction works so well
The mechanism is simpler than the results suggest.
The social side matters too. It is easier to admit confusion to the person beside you than to raise a hand in front of two hundred people, so quieter students take part who would otherwise stay silent.
When to use peer instruction
Peer instruction earns its time on material students routinely get wrong, not on simple recall.
5 peer instruction methods you can use
Each of these gets students explaining rather than just listening. The first four need nothing but your voice and a way for students to respond. The fifth is Mazur's original cycle.
Think, pair, share
The simplest place to start. Pose a question, give students a minute to think and write an answer alone, then pair them to compare, then take a few answers for the whole group. In a secondary maths class this looks like: “Is 0.999... equal to 1? Decide on your own, then convince your partner.” The disagreements in the pairs are the lesson.
Reciprocal teaching
Give each student a section of a text or a step in a process to teach back to their group. In a nursing course, one student walks the group through hand-hygiene steps while the others question anything unclear, then they swap roles for the next procedure. Teaching a slice forces a depth of preparation that reading rarely does.
Peer mentoring
Pair a student who has grasped a topic with one who has not, and make it a two-way exchange rather than a rescue. The mentor sharpens their own understanding by explaining it; the mentee gets an explanation pitched at their level. This works well across year groups, where a second-year student mentors a first-year.
Peer assessment
Give students a clear marking rubric and have them grade each other's work against it. In a writing class, students swap essays and mark them on the same criteria you would use, then use the feedback to revise. Judging work against a standard teaches the standard faster than being told it.
Conceptual questioning: the ConcepTest cycle
This is Mazur's original loop, and the one with the data behind it.
The re-vote is the point. When a class moves from 40 per cent correct on the first vote to 80 per cent on the second, students see that the discussion, not the reveal, is what changed their minds.
Running the vote and re-vote cycle with AhaSlides
The ConcepTest cycle needs a fast way to collect answers, show the split, and re-vote. A live poll or quiz does exactly that. With AhaSlides, you pose the question as a poll, let the room answer from their phones, show the distribution on screen, send students off to argue, then reopen the same poll for the second vote. The two bar charts side by side make the shift visible to the whole class.
Building peer instruction into an online course
The ConcepTest loop does not have to end when the live session does. The AhaSlides Online Course platform lets you rebuild the commit, discuss and re-vote cycle as a self-paced lesson a whole cohort works through on their own time. You lay out each concept in a block-based builder that mixes rich text, video and interactive slides, then drop in auto-graded knowledge checks so a learner commits to an answer, reads a peer's reasoning in a discussion step, and answers again, exactly as they would in the lecture hall. Progress tracking and completion dashboards flag the learners still holding a misconception, so you know who needs a nudge without marking a single script by hand. It connects to systems like Canvas, Moodle and Blackboard if you already run one, learners need no app install, and it is free to deliver. Peer instruction stops being something that only works when everyone is in the room at once.
Common mistakes to avoid
Key takeaways
Peer instruction replaces one-way delivery with a cycle of commit, discuss and reconsider, and the evidence for it is strong: Mazur's Harvard data shows conceptual gains close to double those of a traditional lecture, a result Hake's 6,000-student study backs up. Start with think, pair, share, build up to the full ConcepTest cycle, and use live polls to make the shift in student thinking visible. For online courses, the Course Builder lets you keep that same loop for learners who are never in the room at the same time.
Frequently asked questions
Who created peer instruction?
Eric Mazur, a physics professor at Harvard University, developed and popularised peer instruction in the early 1990s.
How effective is peer instruction?
Crouch and Mazur's ten-year study found normalised gains on the Force Concept Inventory rose from 0.25 under traditional lecturing to 0.49 in the first semester of peer instruction, reaching 0.74 over time. Hake's larger study found interactive methods roughly double the gains of traditional lectures.
Can peer instruction work online?
Yes. The core commit, discuss, re-vote cycle transfers to live virtual sessions through polls and breakout rooms, and to asynchronous courses by building the same steps into a self-paced lesson.
