What does "check for understanding" actually mean?
It means collecting evidence, during the lesson, of what students can do with the idea you just taught, and using that evidence to decide what happens next.
In Barak Rosenshine's summary of classroom research, the more effective teachers "frequently checked to see if all the students were learning the new material." The less effective teachers "simply asked, 'Are there any questions?'" and, hearing none, handed out worksheets (Rosenshine, American Educator, 2012).
England's Department for Education builds the same idea into teacher training. The Initial Teacher Training and Early Career Framework asks new teachers to "check prior knowledge and understanding during lessons", for example "by using common misconceptions within multiple-choice questions", and to check "pupils' understanding of instructions before a task begins" (DfE ITTECF, 2024).
The test for any check, with pupils or students: afterwards, do you know something about the class that you did not know before?
Why does "any questions?" fail?
Two reasons, both described by Leahy, Lyon, Thompson and Wiliam in their study of teachers using assessment for learning (Educational Leadership, 2005):
- Students can opt out. If the rule is hands up, students "can disengage from the classroom by keeping their hands down."
- You hear one student. With traditional questioning "the teacher gets to hear only one student's thinking." To judge the whole class, you need responses from everyone, in real time.
Which techniques check the whole class at once?
Grouped by what each one tells you. Pick two or three and use them every lesson until students expect them.
Everyone answers at the same moment
1. ABCD cards. Each student has four cards labeled A to D. Ask a multiple-choice question, count down, and everyone shows a card. Leahy and colleagues describe the decision this gives you: all correct, move on; none correct, reteach; a mix, run a discussion or pair students up.
2. Mini whiteboards. Better than cards when you want working, a diagram or a short phrase. Boards stay face down until you say "show me." Scan for the two most common wrong answers.
3. Fingers on chests. One to four fingers held against the chest, so neighbors cannot copy. Useful for a quick check in the middle of an explanation.
4. Clickers. Each student presses a key on a handset (also called keypads or voting handsets), and the chart appears on your slide. The difference from cards is the record: you see exact percentages, and every answer is saved for later. With OMBEA, a ResponsePad clicker needs no Wi-Fi, no app and no log-in, and results appear live in PowerPoint (see features).
Students show how sure they are
5. Traffic-light cards. Green means "I've got it," red means "I'm lost." Leahy and colleagues report a 9th grade algebra teacher whose student waved her neighbors' red cards when the teacher moved on too fast. Pair it with a content question: students who feel sure can still be wrong.
6. Agree or disagree. After one student answers, ask everyone else to show whether they agree. Rosenshine lists this among the ways effective teachers involve all students.
You hear the reasoning
7. No hands up, then a random name. Ask, give think time, then call on a name drawn at random (lolly sticks in many UK classrooms, popsicle sticks in the US). Everyone has to prepare an answer because anyone may be asked.
8. Explain how you got it. Rosenshine found the most effective teachers asked students "to explain the process they used to answer the question." The DfE framework adds prompting pupils to elaborate "to check that a correct answer stems from secure understanding." A right answer from a lucky guess sounds different when explained.
9. Tell your neighbor. Everyone turns and gives the answer, or a one-sentence summary, to a partner. You walk and listen to five or six pairs. Rosenshine lists this as well.
Everyone answers, then talks, then answers again
10. Vote, discuss, revote. This is Peer Instruction. Students answer alone, discuss with a neighbor, then answer again. In a Science study of an introductory genetics course at the University of Colorado Boulder, students answered a concept question, discussed it, then answered a new question on the same idea on their own. Scores on the new question averaged 21 percentage points higher than on the first vote. Even among students whose group got the question wrong twice, 44% answered the new one correctly, well above chance (Smith et al., Science, 2009).
Checks you collect
11. Check the instructions. Before a task, ask one question about the task itself: "What is the first thing you do?" It catches students who misread the task before they start on it.
12. Exit ticket. One question at the end, answered by everyone, read before the next lesson. Leahy and colleagues call these "exit passes." We cover formats in exit ticket ideas.
How do you write a good check-for-understanding question?
A good question tells you something whichever answer a student picks.
- One idea per question. If a student gets it wrong, you should know why.
- Wrong answers that mean something. Build each distractor from a real misconception, as the DfE framework suggests. If 40% pick C and C is "added the denominators," you know exactly what to reteach.
- Answerable in under a minute. This is a check, not a test.
- Hard enough to be informative. If everyone gets it right, you learned little.
Example in math: "How many fractions are there between 1/6 and 1/7?" Leahy and colleagues use this question because some students think there are none, and others write things like 1 over 6½. Both errors tell you where to start.
Example in a training room: "A customer asks for a refund after 35 days. Which of these is our policy?" Each option is a mistake staff actually make.
What do you do with the answers?
Decide your rule before you ask, so the result triggers an action.
| Share correct | What it usually means | What to do |
|---|---|---|
| Above 80% | Most have it | Move on. Follow up with the few who missed it. |
| 35% to 70% | Split room | Vote, discuss, revote (technique 10). |
| Below 35% | The idea did not land, or the question is unclear | Reteach with a new example. Check the wording. |
The thresholds come from two sources. Rosenshine reports that the optimal success rate during practice "appears to be about 80 percent," which shows students are learning and still challenged. Crouch and Mazur, after ten years of Peer Instruction at Harvard, say 35% to 70% of students should answer correctly before discussion. Below 35%, too few understand the concept for a useful discussion. Above 70%, "there is little benefit from discussion" (Crouch and Mazur, American Journal of Physics, 2001).
Use these as starting points. Your own data over a term or semester will tell you which numbers fit your subject.
How often should you check for understanding?
More often than feels natural. Rosenshine describes effective teachers checking "at each point" of new material.
Attention is one reason. In a study of three general chemistry courses, students used clickers to self-report attention lapses. Lapses came in shorter and shorter cycles as a lecture went on, and they dropped after clicker questions and demonstrations, an effect that persisted into the lecture that followed (Bunce, Flens and Neiles, Journal of Chemical Education, 2010).
A practical rhythm: a question after each new idea, a hinge question before independent work, and an exit ticket at the end. In a 50-minute lesson that is often four or five checks, each under two minutes.
How do you check for understanding in a large lecture hall?
Cards and whiteboards work in a high school or secondary school classroom because you can read every one from the front. In a 200-seat lecture theatre you cannot, so you need counted answers.
A meta-analysis of 225 studies in undergraduate STEM found that students in traditional lectures were 1.5 times more likely to fail than students in classes with active learning (Freeman et al., PNAS, 2014).
What works at scale:
- Clickers or phone polls for the count. ResponsePads support up to 2,000 people in one session; phones need Wi-Fi or mobile data.
- Anonymous mode for questions where students fear being wrong in front of peers.
- Vote, discuss, revote works in any size of room: neighbors are always available.
- A background question at the start (year, program, prior course) so you can split results by group later.
The higher education and schools pages show how this looks in practice.
Try it tomorrow
Pick one idea from tomorrow's lesson, write one multiple-choice question where each wrong answer is a real misconception, and decide your 80% and 35% rules before you ask. If you want every answer counted and saved without phones, ask us for a quote for a class set of clickers.
