Ask the class a question. Three hands go up, the same three as always; after a second of silence, you either call on one of them or answer it yourself. Everyone else has learned that waiting works.
Think-pair-share breaks that pattern with one simple move: before anyone hears an answer, everyone has to have one. This guide covers what it is, how to run it with the right timings, which questions work, what the research says, the main variations, and how to turn the pair talk into data you can use.
What is think-pair-share?
Think-pair-share is an active-learning strategy in three steps. Think: the teacher asks an open question and every student works on it alone, in silence. Pair: students turn to a partner and each explains their answer. Share: a few pairs report to the class what they worked out together.
It was first described by Frank Lyman of the University of Maryland in a 1981 paper on classroom discussion that includes every student. It is usually counted among the simplest cooperative learning structures: pairs instead of groups, no formal roles, no materials to prepare.
Turn and talk is the same move without the silent first step — quicker, but the faster partner usually speaks first and the other one agrees. Think-pair-share adds the minute that makes both partners arrive with an answer of their own, and a share step that brings the best reasoning back to the whole class.
How do you conduct a think-pair-share?
The three steps, with who does what and for how long:
| Step | What the teacher does | What students do | Time |
|---|---|---|---|
| Think | Asks one open question, keeps it visible on the board, asks for silence | Think alone and jot down their answer | 1–2 minutes |
| Pair | Names the pairs and walks around listening | Take turns explaining their answer; agree, or pin down where they disagree | 2–3 minutes |
| Share | Calls on a few pairs at random, not volunteers | Report the pair's answer or their partner's idea | 2–5 minutes |
5 to 10 minutes in all.
- Set the pairs in advance. Shoulder partners are fine; what matters is that nobody spends the clock looking for someone to talk to.
- Decide who speaks first. A simple rule stops the same student from always leading — for example, the partner whose birthday comes first in the year.
- Give sentence starters. For younger students and English learners: "I think… because…", "I agree, and I'd add…", "I see it differently: …".
What makes a good think-pair-share question?
A good question has more than one defensible answer or asks for reasoning — predict, explain why, compare, choose and defend. A recall question with a one-word answer doesn't work: that's a quiz question, and a pair has nothing to discuss. To write questions that go beyond remembering, the verbs of the higher levels of Bloom's taxonomy are a good place to start. A practical test: if you can already point to the sentence in the textbook that holds the answer, the question is too closed.
12 think-pair-share prompts by subject
| Subject | Prompt | Type |
|---|---|---|
| Math | "A shirt costs $40 after a 20% discount. Was the original price $48 or $50? Convince your partner." | Spot the misconception |
| Math | "Which is larger, 3/4 or 5/8? Explain how you know without converting them to decimals." | Explain |
| Science | "A heavy ball and a light ball are dropped from the same height. Ignoring air resistance, which lands first? Why?" | Predict |
| Science | "Why does the Moon have phases? Sketch your explanation before you talk." | Explain |
| History | "The alliance system or the assassination in Sarajevo: which mattered more in starting World War I? Pick one and defend it." | Choose and defend |
| History | "If you could ask one question to someone living in the period we're studying, what would it be — and why that one?" | Change perspective |
| Reading | "What does the main character want, and what stands in the way?" | Analyze |
| Reading | "Which sentence in the passage changes the mood? Why that one?" | Find evidence |
| World languages | "Translate this sentence in two different ways. Which sounds more natural, and why?" | Compare |
| Civics | "Should the school day start an hour later? Give the best argument for the side you disagree with." | Argue the other side |
| Any subject | "What was the hardest part of today's lesson? Try to explain it to your partner." | Review |
| Workplace training | "A customer asks for a refund one week after the deadline in the policy. What do you do first, and why?" | Apply |
The two math prompts and the first science prompt hide a very common error ($50, not $48; 3/4, which is 6/8; they land together) — they make the best discussions, because half the class usually gets them wrong with confidence.
Think-pair-share variations
- Think-ink-pair-share — students write their answer before talking: the thinking minute becomes real, and the teacher has something to read.
- Think-pair-square — after the pair talk, two pairs join into a group of four and compare conclusions before the share.
- Share your partner's idea — in the share step, each student reports what their partner said, not what they think: listening becomes part of the task.
- Timed turns — each partner speaks for a fixed time, say 45 seconds, while the other only listens; useful when one student tends to take over.
- Think-pair-vote — students answer a multiple-choice question alone, discuss it with someone who chose differently, then answer a second question on the same concept: the version that turns the discussion into data (see below).
Does think-pair-share work?
Few studies test think-pair-share as a package; the evidence is about its two ingredients, and it is strong.
Time to think. In Mary Budd Rowe's research on wait time, teachers typically waited one second or less for a student to start answering. When the pause reached three seconds or more, answers got longer, "I don't know" became rarer and more students volunteered — including those seen as weaker. The think step is guaranteed wait time for everyone.
Talking it through. In a study published in Science in 2009 (Smith and colleagues), students in an undergraduate genetics course answered a question alone, discussed it with their neighbors and answered again; then they faced a second, similar question on their own. Across sixteen pairs of questions, correct answers rose on average from 52% to 68% after discussion, and reached 73% on the new question: they weren't copying their partners, they had understood. And it happened even in groups where, at first, nobody knew the right answer.
The same mechanism is at the heart of Peer Instruction, the method physicist Eric Mazur introduced at Harvard in 1991 — the source of the rule of thumb in the PanQuiz section below.
Think-pair-share as formative assessment
Think-pair-share is also one of the simplest formative assessment examples. During the pair step, walk around with one clear goal — collect two wrong explanations to work on with the class. The share step shows reasoning, not just answers. One caution: a good discussion doesn't prove that each student understood. For that you need an individual answer at the end — one question answered alone, or an exit ticket in the last minutes.
Think-pair-vote with PanQuiz
With a live quiz, think-pair-share produces data — every student's answer before and after the discussion.
- Think. Put a multiple-choice question on the screen; every student answers alone from their phone.
- Read the split. After the question, you see how the class divides across the options; decide with the table below.
- Pair. If the class is split: "Find someone who chose a different answer and convince them." Two or three minutes.
- Share and check. Two pairs explain their reasoning; then a second question on the same concept, with different numbers or context, answered alone.
From Crouch and Mazur's rule (2001):
| Correct answers on the first question | What to do |
|---|---|
| Above 70% | Little to gain from discussion: explain briefly and move on |
| Between 35% and 70% | The right range for pair talk: discussion pays off most when the class is split down the middle |
| Below 35% | Too few students have the idea for a useful discussion: reteach first, or rephrase the question |
A "twin" question beats asking the same one again — it shows whether students understood, not whether they remember what their partner picked. For distractors that match real misconceptions, see how to write multiple-choice questions.
Both questions take minutes to prepare — generated by AI from a topic or from your own documents with the AI quiz generator, or from a PDF with the free PDF to Quiz tool — and every question stays editable. Students join with a PIN from any browser, with no account. Licenses are per teacher, with education discounts; the 15-day Premium trial needs no credit card.
PanQuiz doesn't form the pairs or record what they say — it gives you every student's own answer before and after the discussion, which is the part that's hard to get any other way.
One question, a split class, two minutes of pair talk, one twin question: in five minutes you know whether the discussion produced learning.
Start your free 15-day trial→Five mistakes that ruin think-pair-share
Skipping the think (without the silent minute, the pair starts from whoever is fastest and the other one follows); closed questions (if the answer is one word, there's nothing to talk about); pairs made on the spot (time is lost, and the same students end up left out); only volunteers in the share (you're back to the same three hands — call on pairs at random); no closing check (discussion feels like understanding; one individual question tells you whether it is).