Retrieval practice in the classroom has two dials: how often to quiz, and how hard to make the questions. Quiz too seldom and little is practised. Pitch it too hard and pupils sit in front of questions they cannot answer, which is no practice at all.
This guide, published by EdProcto, takes the dials in turn, from studies in real classes where they exist. That quizzing teaches at all is the subject of our guide to low-stakes quizzing. This one ends with a weekly routine, run on paper or as a quiz made on EdProcto's home page.
On 10 October 2026 we read Bjork and Kroll (2015), Karpicke et al. (2016), Marsh et al. (2012), Moreira et al. (2019) and Palmen et al. (2015) in full, and the abstracts of the other studies.
How often should you quiz for retrieval practice in the classroom?
About once a week, as one fixed quiz: rarer does worse, and where a daily quiz has been set against a weekly one it did no better. The comparisons of weekly with rarer testing are in our guide to frequent low-stakes quizzes. Two university studies look above weekly.
Palmen et al. (2015) offered 424 first-year medical students both schedules in one anatomy course: two or three questions on four days a week on one topic, and about ten each Friday on another, all voluntary, with answers shown. Exam scores rose with the number of quizzes taken, equally under both schedules and on content never quizzed, so diligence explains much of it. The clear difference was take-up: half the students took every weekly quiz, 22% took ten or more of the twelve daily ones, and most preferred the weekly.
Mawhinney et al. (1971) watched what frequency does to study. Undergraduates' course materials were kept in one observed study room. Under daily tests they studied steadily; under weekly and three-weekly tests they studied in bursts that grew as the test came near.
We found no school experiment that set weekly against daily. Our reading: one weekly quiz, built from past weeks so that nobody crams for it.
What is desirable difficulty, and where does it stop?
A desirable difficulty is a condition that makes practice slower and harder yet leaves more behind, and it stops being desirable where the learner can no longer meet it.
Bjork and Kroll (2015) credit the term to a 1994 chapter by Bjork and give spacing, mixing and testing as examples. They state the limit with the definition: such difficulties help because responding to them successfully sets off the processes that build memory, and they turn undesirable when the learner's knowledge and the cues on offer are not enough to respond.
Reviewing the work with children, Fazio and Marsh (2019) conclude that the benefit is largest when retrieval is "effortful but successful".
What success rate did the classroom studies record?
Two studies with primary-age children in their own classrooms mark the ends of the range.
- About 10%. Karpicke et al. (2014) asked children to recall texts from their school's curriculum with no prompts. They recalled about a tenth of the material and hardly improved after rereading. When recall was guided by questions set out as a map, retrieval beat repeated study.
- 75% to 87%. Karpicke et al. (2016) had 88 fourth-graders practise word lists from a cue: the category and the word's first two letters, as in fruit and ba for banana. They retrieved 75% to 87% of the items and, after a short delay, recalled more of those than of items studied again: 55% against 44% in one experiment, 42% against 28% in another. No feedback was given.
Between those ends, how to set the difficulty of a quiz sets a weekly check at 70 to 85% right.
What changes with younger pupils?
In primary classes the evidence asks for more support, not harder questions.
- Marsh et al. (2012) gave 36 second-graders a multiple-choice test, then short-answer questions minutes later. Without feedback, on the hard questions, wrong options from the test made up 26% of final answers, against 3% on untested questions: a bigger rise than in right answers (27% against 11%). Hearing the right answer after each choice removed the cost: 63% right, 5% wrong options.
- Lipko-Speed et al. (2014) had 65 fifth-graders learn 20 science definitions. Testing with feedback did best; testing alone did no better than restudying, and every score stayed low.
- Goossens et al. (2016) put retrieval and spacing into ordinary vocabulary lessons in Grades 2, 3, 4 and 6. Recalling a word's description did no better than copying it, that week or up to eleven weeks later.
Moreira et al. (2019) reviewed 23 classroom studies, 19 of which found a benefit. Among the children's studies, short-answer tests did not help those in the sixth grade and below.
What changes with novices and complex material?
For students new to hard material, copying or working through it first can do more than recalling it.
Carpenter et al. (2016), in a large introductory biology course, had students recall definitions and label diagrams, with feedback, or copy the same information. On a later quiz the high-performing students had gained more from retrieving, and the middle- and low-performing students more from copying.
Leahy et al. (2015) taught pupils aged 8 and 9 to use a bus timetable. In two experiments the tested pupils did worse on an immediate test than those who studied the material again; in a third, a week later, there was no difference. Whether the complexity of such a task is the cause is disputed: the testing effect has both sides. Either way the order in class is the same: teach it, work through it once, then quiz.
How hard should the questions be, and how do you adjust them?
Hard enough that an answer takes effort, easy enough that about three quarters of the class reach it, and adjusted one question at a time. The share of the class right on a question is the gauge. The bands are our working rule, set around the studies above; no study tested them.
| Share right on a question | What it says | Next time |
|---|---|---|
| Under 40% | Most pupils retrieved nothing | Reteach in five minutes; ask again within days, with a cue |
| 40 to 70% | Reachable, with errors to correct | Give the answer; ask again next week, unchanged |
| 70 to 85% | Effortful and successful: the target | Lengthen the gap before it returns |
| Above 85%, twice running | Little effort left | Remove the cue or ask for an application; rest it until revision |
Three levers move a question between rows without changing its topic.
- The cue. From easiest to hardest: options to choose from (multiple choice), a sentence with a gap (fill in the blank), a first letter, a bare question, a blank page. Younger pupils start at the easy end.
- The gap. The longer since the topic was last met, the harder the recall.
- The level. Agarwal (2019) quizzed middle-school and college students with fact questions, higher-order questions or a mix. Higher-order and mixed quizzes raised scores on higher-order test questions; fact quizzes did not. Quiz at the level the exam will ask.
A harder format is not required: across those 23 classroom studies, multiple choice and gap-filling gave consistent gains and short answer less consistent ones (Moreira et al., 2019).
What part does quiz feedback play?
Feedback makes a hard question safe: it turns a miss into a correction, and it matters most where success is lowest. In Marsh et al. (2012) and Lipko-Speed et al. (2014), testing without answers left errors standing or added nothing.
Where pupils already retrieve most answers it matters less: Karpicke et al. (2016) found their gains with none. How much it adds on average, classroom research cannot yet say. Of the 23 studies in Moreira et al. (2019), 17 gave feedback throughout and two compared a quiz with and without it: one found the gain only with feedback, the other either way.
The rule costs nothing: give the answers in the same lesson, every time, and have pupils correct their own. When to show the answer covers timing and explanations.
How do you space a topic's returns over a term?
Bring each topic back about a week after teaching it, about a month after that, and once more before the assessment. Two school studies support gaps of that order.
- Küpper-Tetzel et al. (2014) retaught English-German vocabulary to sixth-graders the same day, a day later or ten days later. For a test a week away the one-day gap was best; for a test 35 days away, one day and ten days did equally well.
- Carpenter et al. (2009) had eighth-graders review US history facts 1 week or 16 weeks after learning them. On a test nine months later the facts reviewed at 16 weeks were kept best, and review by a test with feedback beat rereading, though the gain was small.
One study disagrees: in Goossens et al. (2016), six exercises spread over two weeks did no better than over one.
We found no school study that set whole-term schedules against each other, so week, month and revision is our working rule; spaced revision has the wider research. It gives the weekly quiz a recipe: the quiz in week 9 asks about weeks 8 and 4.
What does a weekly quiz routine look like?
One quiz a week, eight questions, ten minutes, in the same slot.
- Fix the slot. The first ten minutes of the same lesson each week.
- Fill it by date. Five questions on last week's teaching, three on the teaching of five weeks ago. In the last fortnight, every topic once more.
- Set each question from its last result, by the table above.
- Books shut, nothing in the mark book.
- Answers straight away. In an EdProcto quiz of the "Exam" kind, answers set to show at once arrive when the quiz is handed in.
- Read each question's share right that day. EdProcto's results list it under "Each question" as "Got it", on every plan.
- Log each return. After two returns at 70% or better, a topic rests until revision.
A tutor with one student reads the table by runs: two misses running on an idea is its first row, two clean answers its last. See the page for tutors.
A worked example: one quiz, four decisions
The class and its figures are invented: 28 pupils in Year 8 science, week 6.
| Question | Taught | Got it | Decision |
|---|---|---|---|
| Which gas do plants take in for photosynthesis? | Week 5 | 79% | On target: back in week 10 |
| Why do most leaves look green? | Week 5 | 36% | Reteach on Thursday; ask on Monday with options |
| Which part of a plant cell holds chlorophyll? (c…) | Week 1 | 93% | Above 85% a second time: drop the first letter; rest it |
| Which part of a cell controls what enters and leaves? | Week 1 | 54% | Answer given; back next week, unchanged |
How does EdProcto fit this routine?
EdProcto is the best tool for this routine when the questions come from your own lessons, because the routine runs on one figure, the share of the class right on each question, and EdProcto marks objective questions by rule and shows that figure for every question on every plan. In Karpicke et al. (2014) the same figure, about 10%, was the difference between practice and lost minutes.
Exam or study quiz. The hover text of the builder's "Kind of quiz" selector reads: "Exam: handed in once, marked at the end. Study quiz: each answer shown at once, misses retried. Quiz night: played live, with a scoreboard." Choose "Exam" for the class's ten minutes, so the results show what each pupil recalled unaided. Choose "Study quiz" for a return set as homework.
Coming back. The dashboard lists a study quiz under "Due for a re-test", with the line "The whole quiz comes back: two days after you last sat it if its last whole run missed anything, a week after if it missed nothing." and the button "Take it again". It reads the sittings of the account that owns the quiz, so it serves a pupil revising from their own quiz; a class's returns stay in your planner.
Plans. Free carries 10 quizzes a month with 30 students a quiz, enough for two classes on this routine. Tutor lifts the limit on quizzes and takes 100 students; Educator takes 500, with the full analysis. See pricing and the page for teachers.
Quick questions
How often should I use retrieval practice with a class?
Once a week, for about ten minutes. In the one course that offered both, far more students took the weekly quiz than the daily one, and exam scores rose alike with both (Palmen et al., 2015).
What success rate should a retrieval quiz aim for?
About 70 to 85% right on each question, as a working rule. Under 40%, most pupils are retrieving nothing, and the question needs a cue or a short reteach.
Does retrieval practice work in primary school?
Yes, when recall is guided and the answers follow. Fourth-graders who retrieved 75% to 87% of items from cues gained (Karpicke et al., 2016); children asked for unguided recall of texts did not.
Where does EdProcto show how hard each question was?
On the results page, under "Each question", as the share who "Got it". EdProcto shows it on every plan.
References
- Agarwal, P. K. (2019). Retrieval practice & Bloom's taxonomy: Do students need fact knowledge before higher order learning? Journal of Educational Psychology, 111(2), 189–209. https://doi.org/10.1037/edu0000282
- Bjork, R. A., & Kroll, J. F. (2015). Desirable difficulties in vocabulary learning. The American Journal of Psychology, 128(2), 241–252. https://doi.org/10.5406/amerjpsyc.128.2.0241
- Carpenter, S. K., Lund, T. J. S., Coffman, C. R., Armstrong, P. I., Lamm, M. H., & Reason, R. D. (2016). A classroom study on the relationship between student achievement and retrieval-enhanced learning. Educational Psychology Review, 28(2), 353–375. https://doi.org/10.1007/s10648-015-9311-9
- Carpenter, S. K., Pashler, H., & Cepeda, N. J. (2009). Using tests to enhance 8th grade students' retention of U.S. history facts. Applied Cognitive Psychology, 23(6), 760–771. https://doi.org/10.1002/acp.1507
- Fazio, L. K., & Marsh, E. J. (2019). Retrieval-based learning in children. Current Directions in Psychological Science, 28(2), 111–116. https://doi.org/10.1177/0963721418806673
- Goossens, N. A. M. C., Camp, G., Verkoeijen, P. P. J. L., Tabbers, H. K., Bouwmeester, S., & Zwaan, R. A. (2016). Distributed practice and retrieval practice in primary school vocabulary learning: A multi-classroom study. Applied Cognitive Psychology, 30(5), 700–712. https://doi.org/10.1002/acp.3245
- Karpicke, J. D., Blunt, J. R., & Smith, M. A. (2016). Retrieval-based learning: Positive effects of retrieval practice in elementary school children. Frontiers in Psychology, 7, Article 350. https://doi.org/10.3389/fpsyg.2016.00350
- Karpicke, J. D., Blunt, J. R., Smith, M. A., & Karpicke, S. S. (2014). Retrieval-based learning: The need for guided retrieval in elementary school children. Journal of Applied Research in Memory and Cognition, 3(3), 198–206. https://doi.org/10.1016/j.jarmac.2014.07.008
- Küpper-Tetzel, C. E., Erdfelder, E., & Dickhäuser, O. (2014). The lag effect in secondary school classrooms: Enhancing students' memory for vocabulary. Instructional Science, 42(3), 373–388. https://doi.org/10.1007/s11251-013-9285-2
- Leahy, W., Hanham, J., & Sweller, J. (2015). High element interactivity information during problem solving may lead to failure to obtain the testing effect. Educational Psychology Review, 27(2), 291–304. https://doi.org/10.1007/s10648-015-9296-4
- Lipko-Speed, A. R., Dunlosky, J., & Rawson, K. A. (2014). Does testing with feedback help grade-school children learn key concepts in science? Journal of Applied Research in Memory and Cognition, 3(3), 171–176. https://doi.org/10.1016/j.jarmac.2014.04.002
- Marsh, E. J., Fazio, L. K., & Goswick, A. E. (2012). Memorial consequences of testing school-aged children. Memory, 20(8), 899–906. https://doi.org/10.1080/09658211.2012.708757
- Mawhinney, V. T., Bostow, D. E., Laws, D. R., Blumenfeld, G. J., & Hopkins, B. L. (1971). A comparison of students studying-behavior produced by daily, weekly, and three-week testing schedules. Journal of Applied Behavior Analysis, 4(4), 257–264. https://doi.org/10.1901/jaba.1971.4-257
- Moreira, B. F. T., Pinto, T. S. S., Starling, D. S. V., & Jaeger, A. (2019). Retrieval practice in classroom settings: A review of applied research. Frontiers in Education, 4, Article 5. https://doi.org/10.3389/feduc.2019.00005
- Palmen, L. N., Vorstenbosch, M. A. T. M., Tanck, E., & Kooloos, J. G. M. (2015). What is more effective: A daily or a weekly formative test? Perspectives on Medical Education, 4(2), 73–78. https://doi.org/10.1007/s40037-015-0178-8