Classroom Response Systems and Peer Instruction

A traditional lecture has one direction of traffic. The teacher talks; students listen; the teacher cannot tell what most of the students understood until the end-of-term exam. Peer instruction breaks that pattern. The teacher pauses every few minutes to ask a hard concept question, sees how the class is doing in real time, and lets the students do the explaining when the responses split.

The method came out of a physics class at Harvard in the early 1990s, but it works in any subject where students hold predictable wrong ideas that a single question can surface.

The Mazur experience

Eric Mazur taught introductory physics at Harvard from 1984. By his own account he was a successful lecturer. His students did well on the standard problems. Course evaluations were strong.

In 1990 he read work by physics education researchers Ibrahim Halloun and David Hestenes on the common-sense mechanics ideas that students arrive with. He later used the Force Concept Inventory, a conceptual test developed by Hestenes, Wells, and Swackhamer in 1992, to check whether students who could solve textbook problems also understood the underlying ideas. The results were unsettling: students who could apply Newton’s laws to numerical exercises often failed basic conceptual questions about forces.

Mazur tried the test on his own class. The pattern held. His students could compute, but their picture of how forces worked was often a folk theory with the right formulas pasted on top. Telling them the right answer in lecture had not changed the folk theory.

He tried something different. He started posing concept questions during the lecture, asking each student to commit to an answer, and then asking pairs of students who disagreed to argue with each other before answering again. The second-attempt scores were much higher than the first. The students did the work the lecture had not done.

He wrote up the method in Peer Instruction (1997). It has since been studied across physics, chemistry, biology, computer science, statistics, and other fields, with broadly similar results.

Flashcard

Why did Eric Mazur change from lecturing to peer instruction?

Tap to reveal

Answer

Because his students could compute but could not explain.

A conceptual test showed that students who passed standard physics problems still held folk-theory ideas about forces. Lecturing more had not changed those ideas. Peer instruction did, because the explaining is done by classmates who just figured it out themselves.

The peer instruction cycle

A single round of peer instruction has five short steps.

1. Pose a concept question. The teacher puts up a multiple-choice question that targets a known difficulty. Good questions are hard but not unfair. They have one clearly correct option and at least one distractor that catches a common misconception.

2. Individual answer. Students answer privately on their device. About thirty seconds. The privacy matters; students should not see each other’s answers yet.

3. Look at the distribution. The teacher sees the result on a live histogram. Three patterns are common:

• Most students correct: the class understands; move on.
• Most students wrong on one distractor: a shared misconception; the teacher addresses it directly.
• Roughly split between correct and one wrong option: the sweet spot for peer instruction. Move to step 4.

4. Discuss with a neighbour. Students turn to a person nearby and try to convince each other. Two to three minutes. The teacher walks around and listens but does not give the answer.

5. Answer again. Students answer the same question privately. The new histogram almost always shows a much higher correct rate. The teacher confirms the answer and explains the reasoning briefly.

The full cycle takes five to eight minutes. Most classes can run three or four cycles in an hour and still cover most of the planned material.

Why peers explain better than experts

The counter-intuitive part of peer instruction is that classmates often help each other more than the teacher can. Three reasons.

A peer who just figured the question out remembers exactly where their own confusion was. The teacher figured the question out twenty years ago and no longer remembers the path. The peer’s explanation is closer to the language the learner needs.

A peer at the same level uses words and analogies that match the learner’s current mental model. An expert explanation often uses words that depend on concepts the learner has not yet built.

A peer in the next seat has lower social cost than a teacher at the front. A student who would never raise a hand will argue with a desk partner. The argument is where the learning happens.

The act of explaining also helps the explainer. Research on the protege effect suggests that students who teach a concept tend to learn it better than students who only study it. Peer instruction gives both members of the pair this benefit.

Classroom response systems

A CRS is the technology that makes peer instruction practical for a large class. Without one, a teacher in a hall of forty students cannot quickly see what most of them think.

The first generation was hardware “clickers”: small devices the students used to send a multiple-choice answer to a receiver at the front. Many universities bought clicker systems in the 2000s. They worked, but maintaining a class set of devices was expensive.

The current generation uses students’ own phones or laptops. Tools like Poll Everywhere, Mentimeter, Kahoot, Slido, and the polling features in most LMS platforms do the same job. The teacher posts a question to a code or short URL; students answer on their devices; the teacher sees the distribution live.

A good CRS for peer instruction has four features.

Anonymous student view. Students should not see each other’s answers in real time. Public answers push the herd toward popular choices.

Live aggregate for the teacher. The teacher needs to see the distribution at a glance to decide which of the three patterns above applies.

Re-poll on the same question. The second round on the same question is the whole point of peer instruction. A tool that forces a new question for each round defeats the method.

Low friction. Students should be able to join in seconds without an account. Anything more than that and the tool stops being used.

Flashcard

What are the four features of a CRS that work well for peer instruction?

Tap to reveal

Answer

Anonymous student view, live aggregate for the teacher, re-poll on the same question, low friction.

• Anonymous, so students do not just copy the popular answer.
• Live aggregate, so the teacher can pick the right next move.
• Re-poll, so the teacher can compare round-one and round-two responses.
• Low friction, so students actually use it.

Designing concept questions

Peer instruction lives or dies by the quality of the question. A weak question wastes the cycle.

Good concept questions share three properties.

They target a known misconception. The wrong options are wrong for a specific reason that students often believe. A question with three obviously absurd options and one correct option is a recognition test, not a concept test.

They cannot be solved by memory alone. A question that asks “what is Newton’s third law” tests recall. A question that gives a collision scenario and asks how the forces compare tests whether the student understands the law. Peer instruction needs the second kind.

They have one defensible answer. Two or more options that could be defended turn the discussion into a debate about the question, not the concept.

Building a bank of concept questions takes time. Most subject communities have shared collections; the Force Concept Inventory was an early example, and similar inventories now exist for genetics, statistics, computer science, and many other fields.

What this means for a teacher

Peer instruction is not a new lecture trick. It is a shift in what the lecture period is for. Less time on broadcast; more time on probing the class’s understanding and letting students do the explaining.

The trade-off is coverage. A teacher who runs three peer instruction cycles in a 50-minute class covers fewer slides than a teacher who lectures the full hour. The argument for the trade is that the students leave with more of the material actually understood. The teacher who covered more had a higher word count and a lower learning yield.

The tool is doing the easy job: collecting answers from a large class quickly. The hard job is still the teacher’s: choosing the right questions, reading the histograms, and judging when to discuss and when to explain.