Skip to content

Study techniques and their evidence

The pretesting effect: guessing before you learn

How learning works · Published · Evidence: Moderate · About this guide

The pretesting effect, also called the prequestion effect, is the finding that trying to answer questions before a lesson can help you remember those answers, provided the correct answer follows [1]. In a 2024 meta-analysis the benefit was moderate for the material the questions asked about (g = 0.54) and close to zero for the rest of the lesson (g = 0.04) [2]. Most of the experiments used university students.

Most of us think of a test as something that comes after learning. A line of research asks what happens if you reverse the order. You try to answer questions on a topic before you have studied it, get most of them wrong, and then study. The short answer: the questions you tried tend to stick better than if you had spent the same time just reading, provided the correct answer comes afterwards. The rest of the lesson usually gains little.

What does "guessing before learning" mean?

Researchers call it pretesting or prequestioning. Before a reading, a lecture or a video, you see a few questions about what is coming and try to answer them. The material is new to you, so most of your answers are guesses and many are wrong. Then you study, and the answers turn up.

The research uses two versions. In a pretest, the correct answer appears straight after each guess. With prequestions, you answer first and then find the answers yourself while reading or watching [2]. Both rest on the same idea: an attempt that fails can still prepare you to learn.

What did the experiments find?

In one well-known set of five experiments, students read a science essay about vision [3]. In the first experiment, one group spent two minutes answering five questions about the essay and then eight minutes reading it. The other group read it for the full ten minutes. Almost every answer given beforehand was wrong: only 5% were right. Yet on the final test, the group that answered first got 75% of those questions right, against 56% for the group that only read. The 75% counted only questions each person had missed at the start.

Two details rule out the simplest explanations. First, in the later experiments the tested sentences, or their key words, were set in italics or bold for everyone, so the authors concluded that knowing where to look did not explain the gain. Second, trying to answer the questions helped more than memorising the same questions without answering them. In the experiment that tested a week later, the benefit was still there.

A second study found the same pattern with very different material [4]. People answered made-up trivia questions that nobody could know, such as "What peace treaty ended the Calumet War?". In other experiments they saw a cue word such as whale and guessed a loosely linked word, here mammal. Trying first and then seeing the answer led to better memory than seeing the question and the answer together. Only wrong guesses were analysed, so the gain came from attempts that failed.

Does it help with the rest of the lesson?

Mostly not, and this is the most important limit. A preregistered meta-analysis pooled the studies on prequestions [2]. For the questions people had been asked beforehand, the benefit was moderate: an effect size of g = 0.54, across 97 comparisons. Its size varied a lot from study to study, and a correction for the chance that weaker results went unpublished lowered it a little, to g = 0.45. For material in the same lesson that the questions did not cover, the average benefit was close to zero: g = 0.04, across 91 comparisons. On that other material the studies split almost evenly: about half found a gain, the rest none or a loss.

Chart of two estimates: prequestions helped the material they asked about (g = 0.54) but not the rest of the lesson (g = 0.04).

Prequestions: the benefit for the material asked about and for the rest (St. Hilaire et al. 2024)

The average benefit of prequestions over none, as Hedges' g with its 95% confidence interval, from St. Hilaire, Chan and Ahn's preregistered meta-analysis (2024): moderate for the material the questions asked about (0.54, 97 comparisons) and close to zero for material they did not ask about (0.04, 91 comparisons). Results varied widely between studies; group averages, not a prediction for any one learner.

Data: St. Hilaire, Chan & Ahn (2024), Psychonomic Bulletin & Review 31, p. 425 and Table 2. [2]

Data table
Prequestions: the benefit for the material asked about and for the rest (St. Hilaire et al. 2024)
MaterialHedges' g95% confidence intervalComparisons
Material the questions asked about0.540.42 to 0.6697
Material they did not ask about0.04-0.04 to 0.1191
  • Corrected for results that may have gone unpublished, the estimate for the material asked about was 0.45; for the other material it stayed 0.04.
  • For questions asking for understanding rather than facts, the estimate for the material asked about was 0.28 (22 comparisons), with an interval from −0.03 to 0.60.
  • On the material not asked about, 54% of the studies found a gain and the rest none or a loss.

So guessing first tends to fix particular points more firmly. It is not a general boost for everything you study afterwards, and it matters which questions you choose. The benefit has also been more consistent for facts stated directly in the material than for questions that ask you to reason beyond it [1]. For questions that ask for understanding, the average estimate was smaller (g = 0.28, from 22 comparisons), and its margin of error was wide enough to include no benefit at all [2].

Why would a wrong guess help?

Does a wrong guess hurt? In these studies it did not, as long as the correct answer came afterwards: the questions people had first answered wrongly were remembered better than when they had only read the material [3][4]. Why it helps has several explanations and no settled answer. The most common one is attention: a question you have tried to answer makes you notice its answer when it appears and take it in more carefully [2]. Attention is probably not the whole story, though. In the vision study, highlighting the tested material for everyone did not erase the advantage [3]. Curiosity about the answer is another candidate [1].

A third explanation is that searching your memory for an answer brings related ideas to mind, and the correct answer then lands among them. Word-pair studies fit this. Guessing helped when the cue and the answer were related in meaning, and it did not help when they were unrelated [5][6]. In word pairs, a sensible guess, even a wrong one, seemed to prepare the ground. Relatedness is not the whole explanation either. In another set of experiments, people learning the meanings of unfamiliar English words or the translations of foreign words remembered more when they had guessed first and were then shown the answer than when they simply read the word with its meaning, although their guesses were unlikely to have anything to do with the answer [7].

What is clear is the condition: the answer has to come afterwards. A review sums it up this way. Testing before learning can benefit learning "if there is an opportunity to study the correct answers afterwards" [1]. The authors of the vision study also warned that a test with no teaching or feedback after it is likely to be of little use for the questions people got wrong [3].

Why does it feel like a waste of time?

It means being wrong on purpose, and many people dislike that. In one study, people remembered related word pairs better when they had guessed at them first than when they had simply studied them. Even so, when asked after the final test which way of studying had given them the best memory, they named simply studying for longer, and put guessing first well below where it really stood [6]. In another set of experiments, people judged guessing and then being corrected to be a less effective way to learn than reading, when it had in fact worked better [7]. A review notes that many learners do not see the benefit even after they have experienced it, and that students prefer to avoid making errors [1].

Pretests may also have a cost. In one study run inside an open online course about climate change, learners who were given pretests were more likely to drop out as the course went on. Overall, the pretests did not improve exam results; among the learners who stayed to the end, those given pretests did better [1]. The review gives no reason for the drop-outs, and the learners who stayed had chosen to, so their better results do not show that the pretests caused them. The same review suggests that some time to get used to the approach, and an explanation of why the errors are part of it, may help.

How could you try it yourself?

These follow what the studies did. They are reasonable ways to apply the findings, not guarantees:

  • Turn what is coming into a few questions. Before a chapter or a lecture, take its headings, its learning goals or its end-of-chapter questions, and pick two or three that cover the points you most need. In the vision study, five questions took two minutes.
  • Write down a real guess. Commit to an answer rather than "no idea". The benefit was found for attempts, even wrong ones.
  • Check every guess. While you study, watch for the answers. Afterwards, compare each guess with the correct answer and note what was wrong.
  • Then practise recall as usual. The authors of the meta-analysis believe prequestions act mainly on how you take in new material, while practising recall after you have learned helps you keep it over time [2]. In that view, the two do different jobs.
  • Expect to be wrong. In the studies, most first answers were wrong. That is what the method looks like, not a sign that it has failed.

What the research does not say

  • It does not say guessing first improves everything in a lesson. On average, the material the questions did not cover gained about nothing [2].
  • It does not say wrong answers help on their own. The benefit depends on meeting the correct answer afterwards [1][3]. How soon it has to come is less clear. In some word-pair studies the answer had to follow straight away, while with trivia questions an answer shown up to a day later still helped [5][8][1].
  • It has not yet shown how well the benefit carries over to real courses. Most experiments used university students [2]. Classroom and online-course studies are fewer, and their results vary more [1].
  • It does not settle why it works. Attention, curiosity and a search through related memories are all candidates [1].
  • It does not say reasoning questions benefit as much as factual ones. The evidence there is weaker [1][2].
  • It does not give a recipe. How many questions to ask, and how long before studying, are open questions. Very few studies have put a delay between the questions and the lesson [2].

Sources

Every source the guide cites

  1. [1] Pan, S. C., & Carpenter, S. K. (2023). Prequestioning and pretesting effects: A review of empirical research, theoretical perspectives, and implications for educational practice. Educational Psychology Review, 35(4), Article 97. review · doi.org/10.1007/s10648-023-09814-5
  2. [2] St. Hilaire, K. J., Chan, J. C. K., & Ahn, D. (2024). Guessing as a learning intervention: A meta-analytic review of the prequestion effect. Psychonomic Bulletin & Review, 31(2), 411–441. meta-analysis · doi.org/10.3758/s13423-023-02353-8
  3. [3] Richland, L. E., Kornell, N., & Kao, L. S. (2009). The pretesting effect: Do unsuccessful retrieval attempts enhance learning? Journal of Experimental Psychology: Applied, 15(3), 243–257. experiment · doi.org/10.1037/a0016496
  4. [4] Kornell, N., Hays, M. J., & Bjork, R. A. (2009). Unsuccessful retrieval attempts enhance subsequent learning. Journal of Experimental Psychology: Learning, Memory, and Cognition, 35(4), 989–998. experiment · doi.org/10.1037/a0015729
  5. [5] Grimaldi, P. J., & Karpicke, J. D. (2012). When and why do retrieval attempts enhance subsequent encoding? Memory & Cognition, 40(4), 505–513. experiment · doi.org/10.3758/s13421-011-0174-0
  6. [6] Huelser, B. J., & Metcalfe, J. (2012). Making related errors facilitates learning, but learners do not know it. Memory & Cognition, 40(4), 514–527. experiment · doi.org/10.3758/s13421-011-0167-z
  7. [7] Potts, R., & Shanks, D. R. (2014). The benefit of generating errors during learning. Journal of Experimental Psychology: General, 143(2), 644–667. experiment · doi.org/10.1037/a0033194
  8. [8] Kornell, N. (2014). Attempting to answer a meaningful question enhances subsequent learning even when feedback is delayed. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40(1), 106–114. experiment · doi.org/10.1037/a0033699

This guide is general information about learning research for students, not medical advice.