How to use quizzes for retrieval practice in the classroom

Most teachers have used a quiz in class. The interesting question is not whether quizzes work, but what makes them actually move learning rather than just fill the first five minutes of a lesson. The research base behind retrieval practice is strong, but quiz formats vary a lot in real classrooms, and many common ones are doing far less than teachers assume.

This guide is aimed at the teacher's chair. How do you bake retrieval into a lesson sequence so it shifts long-term retention? How do you design questions that diagnose problems rather than just produce a number? And how do you keep the marking load low enough that you will still be running quizzes in week six, not just week one?

The short version: Low stakes, high frequency, varied question types, and a clear plan for what to do with the results.

Retrieval practice, sometimes called the testing effect, is the finding that pulling information out of memory strengthens it more than putting more information in. Karpicke and Roediger's 2008 study is the canonical reference: Students tested on word lists remembered them substantially better a week later than students who simply re-studied. Hundreds of replications have followed, including in real secondary classrooms.

The mechanism: When students try to retrieve something, the act of searching (whether they succeed or not) strengthens the memory trace. Effort is the point. The sweet spot is what Bjork calls desirable difficulty: Hard enough to require genuine effort, easy enough that most students get most questions right.

A few caveats. Retrieval practice works best on material students have already encountered in some depth. It is not a shortcut to learning content from scratch. It also works better for factual recall and well-defined procedures than for novel problem-solving. And the format shapes what gets strengthened: Multiple choice trains recognition, free recall trains production.

There are three useful slots for a quiz in a typical secondary lesson, and each does a different job.

The starter quiz is the workhorse. Five questions on the previous lesson, the lesson before that, and something from last week or last term. Sometimes called a do now, it does two things at once: Strengthens older material that would otherwise fade, and tells you whether the class is ready for today. If most students missed a question on prior knowledge today's lesson depends on, you have learned something useful before teaching starts.

The mid-lesson check is shorter, usually two or three questions tightly tied to material you have just modelled. The goal is diagnostic. Are students with you? If not, you adjust the next ten minutes rather than ploughing on. Mini-whiteboards work well here because you see everyone's answer at once.

The exit quiz does double duty. It gives students one more retrieval rep on the day's material, which boosts retention, and tells you who got it and who did not, feeding into the next lesson. Two to four questions on a slip of paper or a digital form usually does the job.

The quality of a quiz lives in the questions. A starter with five vague prompts gives you five vague answers. A starter with five well-designed questions can be the most informative two minutes of the lesson.

Mix question types deliberately. Multiple choice is fast to mark, and well-designed distractors can surface specific misconceptions. Short-answer free recall forces students to produce the answer rather than recognise it, which is closer to what most exams demand. Application questions ask students to use the idea in a new context. A good five-question starter often includes one or two of each.

Write questions at the level of detail you want students to remember. 'What is photosynthesis?' gives you a paragraph of varied quality. 'Write the word equation for photosynthesis' gives a clear yes-or-no answer to a specific question. The second is far more useful diagnostic information.

A hinge question is a multiple-choice question where each wrong answer maps to a specific, predictable misconception. Get the pattern of wrong answers and you know exactly what to address next.

A classic example from GCSE chemistry: When you electrolyse molten lead bromide, what is produced at the cathode? Options: (a) lead, (b) bromine, (c) hydrogen, (d) oxygen. Picking bromine means the student has confused cathode and anode. Picking hydrogen means they have applied the rules for aqueous solutions to a molten one. You learn something specific from each wrong pattern.

Hinge questions take longer to write than ordinary quiz questions. The pay-off is that two or three can replace half a lesson of guesswork about where to direct re-teaching. Many departments build a shared bank over time, which spreads the cost.

If quizzes feel like assessments, students treat them like assessments. They get anxious, they cheat, they avoid risk, and diagnostic value drops to near zero. The whole point of retrieval practice in lessons is that it is safe to get things wrong, because getting things wrong (and then fixing them) is where the learning happens.

A few practical moves to keep the temperature low. Do not put quiz scores in the markbook by default; the act of recording them changes how students approach the task. Mark together as a class wherever possible, or have students self-mark with answers projected. Celebrate corrections rather than scores: 'Who got something wrong and now knows the right answer?' is more useful than 'Who got full marks?'

Low-stakes framing tends to reduce anxiety and can help students engage more freely, though the direct effect on long-term recall is less settled. Tone matters.

The biggest difference between a quiz that boosts learning and one that just fills time is what happens in the ninety seconds after the quiz ends. If you collect the answers and move on, you have squandered the diagnostic value.

A simple routine: After the quiz, scan responses or ask for a show of hands per question. Identify the one or two questions with the most wrong answers. Re-teach those points immediately, or flag as priority for next lesson. The whole loop takes around five minutes and converts the quiz from ritual into feedback mechanism.

Keep a running log of which questions keep tripping students up. After three or four weeks, patterns emerge. If a misconception keeps surfacing, the issue is probably with your original explanation rather than with the students.

The basic principles of retrieval practice apply to every subject, but the question formats that work best depend on what the subject actually demands. The table below maps useful quiz formats to common secondary subjects.

A quiz routine you cannot sustain is worse than no routine. The aim is a rhythm you can keep up across a full term, which usually means cutting marking time per quiz to near zero.

A few approaches that scale. Self-marking against projected answers saves the entire marking step and pulls students into the feedback loop. Peer-marking with a clear rubric works similarly, though it needs a calm classroom culture. Whole-class marking, where you collect the slips and skim for patterns rather than scoring each one, often gives the most useful information for the least effort. You are looking for the shape of the misunderstanding, not a percentage.

Digital tools help if your school supports them. Google Forms, Microsoft Forms, or a dedicated platform like Cognito's quiz library can mark instantly and surface patterns across a class. The trade-off is screen time and login friction, which is not always worth it for a thirty-second starter. Many teachers use paper for daily starters and digital for weekly exit quizzes.

Retrieval practice gets more powerful when spaced across weeks and topics. The principle is the same as spaced repetition for individual revision: Material retrieved at spaced intervals sticks far better than material retrieved repeatedly in one sitting.

A workable pattern for a six-week unit. Every starter quiz includes one or two questions from earlier in the unit. Every fortnight, a longer quiz pulls from the start of the unit through to the present. At the end of the unit, the final check includes questions from earlier units too, especially the foundational ones. Students start to see their knowledge as cumulative rather than chapter-by-chapter.

Departments that build shared quiz banks across a scheme of work find this far more sustainable than individual teachers writing fresh questions each week. The first year takes investment. From year two onwards, the marginal cost per lesson drops considerably.

A few patterns turn up when retrieval practice is not landing. None are catastrophic, but each quietly drains the effect.

Quizzing only on the previous lesson gives short-term gains and almost no long-term benefit. Mix in older material from the start.

Using the same question format every time. Students adapt and the diagnostic value drops. Vary multiple choice, short answer, and application prompts within a single starter.

Grading quizzes in the markbook. The moment quizzes feel like assessments, the safety that makes retrieval practice work disappears.

Moving on without acting on the results. A quiz that does not change what you do next is just a worksheet with extra steps.

Letting the same students answer aloud every time. Hands-up rewards the confident few. Cold call or use mini-whiteboards so every student is doing retrieval, not just observing it.