Twenty questions on photosynthesis can be an exam or a lesson. What changes is when the answer appears: after each response, when the sitting ends, or days later with the marked paper.
An immediate feedback quiz shows the right answer, and ideally the reason, before the next question arrives. Each miss becomes a short piece of teaching, and the total at the bottom stops measuring what the student could do alone.
EdProcto builds both from one set of questions: a quiz made in the quiz maker on the home page can be switched between an exam and a study quiz that checks every answer as it is given. The research on timing points both ways, and this guide reports both.
Should a quiz show the answer after each question or only at the end?
Show the answer after each question when the quiz exists to teach, and hold it back when the quiz is a rehearsal or a measurement.
| The quiz | Show the answer | Why |
|---|---|---|
| New or difficult material | After each question | A safety net, so the learner is not bogged down (Shute, 2007) |
| Young children, struggling learners | After each question | The second-grade experiment below, and Shute's guideline |
| A long practice set | After each question | Our judgement: forty corrections at the end are too many |
| A self-test on your own chapter | After each question, then the misses again | EdProcto's study mode runs this by itself |
| A mock exam, a test of stamina | At the end | The real paper gives no answers on the way |
| Questions that give each other away | At the end | One explanation would answer a later question |
| Any quiz whose score counts | After the scores are released | The total must show what the student did alone |
What did the research find about feedback timing?
It found no consistent winner. Laboratory experiments tend to favour a delay, classroom studies with real quizzes tend to favour the answer at once, and the largest classroom experiment we read found no difference.
Kulik and Kulik (1988) pooled 53 studies. Applied studies with actual classroom quizzes and real learning materials usually found immediate feedback more effective. Experiments on the acquisition of test content usually found the opposite, and list-learning studies varied. Shute (2007), reviewing the literature nearly two decades later, found no consistent main effect of timing.
After each question, or when the sitting ends
Three studies set these side by side.
- Corral et al. (2021) asked people to judge whether hypothetical studies were true experiments. The correct answer came after each response or after the last scenario. On a test with new scenarios the groups did not differ, with or without explanations.
- Fyfe and Rittle-Johnson (2016) taught second-grade children a strategy for a mathematics problem, then had them practise on a computer with no feedback, immediate feedback or summative feedback, the term for feedback once the work is done. Next day, feedback beat none for children who began with low prior knowledge, and immediate feedback was particularly effective for children with low and high prior knowledge alike.
- Dihoff et al. (2004) had students prepare for classroom examinations with practice tests whose items returned on those examinations. Practice feedback came immediately, at the end of the test or 24 hours later. Only the immediate feedback raised examination performance, most on items repeated in their original wording.
After a delay
Here the laboratory and the classroom part company.
Butler et al. (2007) had participants read 12 passages of 280 to 300 words and take a six-option multiple-choice test, with feedback at once for some items and after a delay for others. On a later recall test the delayed items were remembered better. In the third experiment of Corral et al. (2021), feedback two days later beat feedback after each response. Mullet et al. (2014) ran two experiments in an upper-level engineering course: homework feedback arrived straight after the deadline or a week later, and the delay led to better results on course exams that set new problems on the same concepts. The students reported that the immediate feedback had helped them most.
Against that, feedback delayed by 24 hours did not enhance learning in Dihoff et al. (2004). Fyfe et al. (2021) ran one experiment across 38 college classes, with feedback on assignments immediate or delayed by a few days. The effect of timing on later class assessments was 0.002, with a 95% interval from −0.05 to 0.05.
Age does not settle it. Metcalfe et al. (2009) found that Grade 6 pupils learning school vocabulary remembered more after delayed feedback than after immediate.
What seems to explain the difference
Four things, none settled.
- "Immediate" means two things. Shute (2007) notes that it can mean right after an item or right after the whole test. In the engineering course it meant after the homework deadline.
- A delay is a second, spaced meeting with the material. That is how Butler et al. (2007) explain their result. Corral et al. (2021) add that a longer gap may let the learner forget the wrong idea first.
- Some of the advantage is the calendar. With college students learning harder words, Metcalfe et al. (2009) found delayed feedback ahead only while it also fell closer to the final test.
- The learner and the task matter. Shute's guidelines recommend immediate feedback for a task that is difficult for the learner and for lower achievers, and a delay for simple tasks and high achievers. She adds that immediate feedback may make answering less careful.
What should the feedback say: right or wrong, the answer, or an explanation?
It should give the correct answer and a short explanation. Van der Kleij et al. (2015) pooled 70 effect sizes from 40 studies of feedback on single items in computer-based learning. Saying only whether the response was correct had an average effect of 0.05, giving the correct answer 0.32, and elaborated feedback such as an explanation 0.49, with its lead widest on higher-order outcomes. Effects were lower where feedback was delayed, and lower in primary and high schools. The spread was wide: single effects ran from −0.78 to 2.29.
Why an explanation must be short and about the reasoning
Because a long explanation is not taken in. Shute (2007) asks for feedback that gives the what, how or why of the problem, in pieces small enough not to be discarded, and "as simple as possible but no simpler". It should be about the task, not the learner. She reports an experiment in which the more complex the feedback on 16 multiple-choice questions, the less well undergraduates corrected their errors. Explanations win on average, and long ones can lose.
Take one question: which is larger, 3/8 or 2/5?
| Feedback | What the student reads |
|---|---|
| Right or wrong only | "Incorrect." |
| The correct answer | "2/5 is larger." |
| A short explanation | "2/5 is larger. Over 40, 3/8 is 15/40 and 2/5 is 16/40." |
| Too much | Three paragraphs on equivalent fractions, and a note to revise chapter 4 |
Write the third line. It gives the answer and the one step that settles it, and says nothing about the student. In EdProcto every question has a field for this line, and Procto AI drafts one with each question it writes, for you to edit.
Should students answer again until they get it right?
Not within the same question, on the evidence we read. In Butler et al. (2007), some participants were shown the correct answer and others kept choosing until they found it. Final recall was the same.
Shute (2007) reports an earlier argument that lower-ability students gain more from being given the correct response than from being told to try again. With multiple choice, and this point is ours, a second and a third click are elimination, not recall.
A second attempt is worth more after a gap. Retry the misses when the sitting ends and the answer has left the screen, then bring the whole quiz back days later. Two neighbouring guides cover the questions you got wrong and the spacing research behind a quiz made from a chapter.
How does EdProcto's study mode work?
EdProcto is the best tool for an immediate feedback quiz made from your own material, because its study mode does in one sitting what the research above supports: one honest attempt, the answer with its reason, then a second attempt at the misses. The reason is the part that pays: 0.49 for an explanation in the pooled studies, against 0.05 for a bare right or wrong.
- Make the quiz. Paste a chapter or a lesson handout into the quiz maker and choose "Test yourself", or open a ready-made quiz from the kits. In the builder, any quiz can be set as an exam, a study quiz or a quiz night.
- It begins by itself. On your own study screen there is no name to give and nothing to agree to.
- Answer, then press Check. The page holds no answer key. The response is marked on the server, and only then does the page show the verdict, the correct answer where yours was not right, and the explanation.
- Finish. The result lists what you missed, counted by topic, with a button that starts a retry of those questions alone.
- Come back. The dashboard lists the quiz as due for a re-test two days after an attempt with misses, and a week after a clean one.
For a class or a tutee, share the study quiz with a code: the student opens it on any device, gives a name, and has each answer checked the same way. The list of misses and its retry belong to the study screen of the account that owns the quiz, so a student who wants them needs a free account and a copy of their own, which a quiz published as a kit provides, answers included. See the pages for tutors and for students. The shape of the questions is a separate choice among the templates.
The Practice preset is not the same thing
A teacher's ordinary quiz can also show its answers at once. The Practice preset in a quiz's settings removes the time limit and the security measures, allows going back, turns off shuffling, permits retakes without limit, and sets the correct answer, the explanation and the score to show at once.
The kind of quiz is a separate choice. A quiz of the exam kind is handed in whole and marked at the end, so with the preset its answers arrive with the result, the moment the student submits. A study quiz checks each answer as it is given. Use the preset for a low-stakes class quiz taken in one go, as in our guide to retrieval practice, and the study kind when each miss should be put right before the next question.
Study mode, the re-tests and the kits are in the Free plan, with 20 questions a month written by Procto AI. The Learner plan lifts the limit on self-tests and raises that to 100. The Tutor and Educator plans add more students a quiz, all 111 templates and more of Procto AI's writing. See the pricing page.
What should an immediate feedback quiz never be used for?
A score that counts. The answers are shown as the quiz goes, so the total includes whatever the student learned from the feedback on the way, and the first student to finish holds every answer. Shute's guidelines make the point from the other side: feedback should come only after the learner has attempted an answer, because studies in which learners could see the answers first showed inconsistent benefits.
Hold the answers back as well for a mock exam, which also tests stamina and pace, and for a set in which one explanation would answer a later question.
For a mark that matters, EdProcto's Exam preset shows nothing until you release the scores, and the integrity page lists the measures for higher stakes. No setting makes copying impossible.
What mistakes spoil an immediate feedback quiz?
The mistakes that spoil it leave the feedback on the screen and out of the student's head.
- Recording the score. The result of a study quiz is its list of misses. The percentage is not a mark.
- Praise in place of information. "Well done" says nothing about the task.
- Clicking through. Ask for a real attempt first: in EdProcto the check comes only after an answer is given.
- One sitting. Monday's quiz still has to come back on Thursday.
Quick questions
Is immediate feedback better than delayed feedback?
Neither wins consistently. Kulik and Kulik (1988) found classroom studies favouring immediate feedback and experiments on test content favouring a delay, and Fyfe et al. (2021) found no difference across 38 college classes.
Should a practice quiz show the correct answer after each question?
Yes, when the aim is to learn and nothing rides on the score. A quiz made on EdProcto's home page can be set as a study quiz, which gives the answer and its reason after each check.
Does answer-until-correct help students learn?
No more than being shown the answer, in the one direct comparison we read: Butler et al. (2007) found equal recall. A retry after a gap is the better second attempt.
Do young children need the answer straight away?
The evidence is thin but leans that way for the youngest: second-grade children practising mathematics did best with immediate feedback in Fyfe and Rittle-Johnson (2016). See our guide to questions for young learners.
References
- Butler, A. C., Karpicke, J. D., & Roediger, H. L. (2007). The effect of type and timing of feedback on learning from multiple-choice tests. Journal of Experimental Psychology: Applied, 13(4), 273–281. https://doi.org/10.1037/1076-898X.13.4.273
- Corral, D., Carpenter, S. K., & Clingan-Siverly, S. (2021). The effects of immediate versus delayed feedback on complex concept learning. Quarterly Journal of Experimental Psychology, 74(4), 786–799. https://doi.org/10.1177/1747021820977739
- Dihoff, R. E., Brosvic, G. M., Epstein, M. L., & Cook, M. J. (2004). Provision of feedback during preparation for academic testing: Learning is enhanced by immediate but not delayed feedback. The Psychological Record, 54(2), 207–231. https://doi.org/10.1007/BF03395471
- Fyfe, E. R., de Leeuw, J. R., Carvalho, P. F., Goldstone, R. L., Sherman, J., Admiraal, D., Alford, L. K., Bonner, A., Brassil, C. E., Brooks, C. A., Carbonetto, T., Chang, S. H., Cruz, L., Czymoniewicz-Klippel, M., Daniel, F., Driessen, M., Habashy, N., Hanson-Bradley, C. L., Hirt, E. R., … Motz, B. A. (2021). ManyClasses 1: Assessing the generalizable effect of immediate feedback versus delayed feedback across many college classes. Advances in Methods and Practices in Psychological Science, 4(3), Article 25152459211027575. https://doi.org/10.1177/25152459211027575
- Fyfe, E. R., & Rittle-Johnson, B. (2016). The benefits of computer-generated feedback for mathematics problem solving. Journal of Experimental Child Psychology, 147, 140–151. https://doi.org/10.1016/j.jecp.2016.03.009
- Kulik, J. A., & Kulik, C.-L. C. (1988). Timing of feedback and verbal learning. Review of Educational Research, 58(1), 79–97. https://doi.org/10.3102/00346543058001079
- Metcalfe, J., Kornell, N., & Finn, B. (2009). Delayed versus immediate feedback in children's and adults' vocabulary learning. Memory & Cognition, 37(8), 1077–1087. https://doi.org/10.3758/MC.37.8.1077
- Mullet, H. G., Butler, A. C., Verdin, B., von Borries, R., & Marsh, E. J. (2014). Delaying feedback promotes transfer of knowledge despite student preferences to receive feedback immediately. Journal of Applied Research in Memory and Cognition, 3(3), 222–229. https://doi.org/10.1016/j.jarmac.2014.05.001
- Shute, V. J. (2007). Focus on formative feedback (Research Report No. RR-07-11). Educational Testing Service. https://doi.org/10.1002/j.2333-8504.2007.tb02053.x
- Van der Kleij, F. M., Feskens, R. C. W., & Eggen, T. J. H. M. (2015). Effects of feedback in a computer-based learning environment on students' learning outcomes: A meta-analysis. Review of Educational Research, 85(4), 475–511. https://doi.org/10.3102/0034654314564881