All articles
5 min readlearning / training

Fail the quiz first: pretesting beats extra reading time

In the AI trainings I run, the first 10 minutes go to context and definitions. Eight studies say I should spend them on questions the room cannot answer yet. Lindsey Richland ran that comparison five times in 2009: students who guessed at buried concepts before reading an essay beat students who got extra time with the same essay, in all five experiments 1. The guesses were mostly wrong. The wrong guesses were the point.

What the research shows.

Richland, Kornell and Kao handed participants an essay about vision. The test group answered questions about concepts embedded in the essay before reading it. The extended study group got longer with the passage instead. Final test scores favored the test group every time, and the advantage survived when the analysis counted only the items that group had failed on the pretest 1. Failing to retrieve an answer prepared people to absorb it.

Nate Kornell, Matthew Hays and Robert Bjork isolated the same mechanism that year with word pairs built so the guess would miss. Shown a cue like "whale", participants tried to name a weak associate like "mammal", then saw the pair. Trials where the guess landed were thrown out of the analysis, and guessing still beat studying the pair from the start 2.

Steven Pan and Faria Sana ran the scaled-up version: five experiments, 1,573 participants, expository text passages paired with either a pretest or a posttest. Both beat a no-test control. Pretesting produced the higher overall scores, in multiple-choice and cued-recall formats, with feedback and without it, on final tests given 5 minutes or 48 hours later 3.

Pretests also change what happens during the session. Pan, Sana, Schmitt and Bjork paired a 26-minute recorded lecture with either pretest questions or algebra problems, then probed for mind wandering throughout. The pretested group drifted off less and scored higher afterward, whether the questions arrived at the start or in blocks across the lecture 4.

The classroom version comes from Nicholas Soderstrom and Elizabeth Bjork, who opened lecture sessions in an undergraduate research methods course with pretests for 10 weeks. Memory and transfer performance on the high-stakes end-of-term exam improved against control questions 5.

Kathryn St. Hilaire, Jason Chan and Dahwi Ahn pooled the field in 2024. For the material covered by the prequestions, the average effect was g = 0.54 across 97 experiments 6. That is a large effect for an intervention costing 4 minutes of preparation.

The gain stops at the questions you asked.

The same meta-analysis measured what happens to everything else in the lesson: g = 0.04 across 91 experiments, sitting on zero 6. Alexander Toftness and colleagues found the same wall with real lecture recordings over 20 minutes long. Prequestioned students beat controls overall and on the prequestioned content; on the rest of the lecture they were no better in one experiment and barely better in the other 7. Soderstrom and Bjork's classroom study did find spillover to related material that was never pretested 5, so the boundary is contested. Plan for the narrow result. Whatever you leave out of the pretest, treat as unprotected.

Your audience will also fight you on this. Pan and Michelle Rivers ran five experiments where pretesting produced better memory than reading, and participants ranked it as the less effective method anyway. The belief held after they had lived through both conditions repeatedly, and only shifted once the researchers fed back performance scores or reminded people of their own predictions 8. Expect the room to look annoyed around minute three.

The protocol.

  1. Before the session, list the 6 facts you want people to still hold in a month. Turn each into one question. That list is your retention list, since the payoff lands on questioned content and nowhere else 6.
  2. Run the questions cold, before any teaching. Written answers, 30 seconds each, no skipping and no "I don't know". A wrong answer carries the full effect 1 2.
  3. Reveal the correct answer as soon as the guess is on paper. Yeray Mera's team compared immediate feedback against feedback delayed 24 and 48 hours: immediate won, though pretesting still beat read-only study under the delay 9.
  4. Past 20 minutes of material, split the pretest. Drop a fresh block of questions every 10 minutes rather than firing all of them at the opening. Both placements worked in the 26-minute lecture study 4.
  5. Close the session with the same questions. The pretest becomes retrieval practice on material people now know, and you get a before-and-after score for free.
  6. Say the score gap out loud when you show it. People rate pretesting as the weaker method until they see their own numbers 8.
  7. Apply it to your own reading. Before a paper or a long report, write 3 questions from the title and the section headings, answer them in a sentence each, then read. The reading itself changes: you scan for the answers you got wrong.

Take the deck you are presenting next week, pull the 6 facts you want to survive it, and rewrite them as questions your audience will get wrong on the first try.

Sources.

  1. 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. doi.org/10.1037/a0016496
  2. 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. doi.org/10.1037/a0015729
  3. Pan, S. C., & Sana, F. (2021). Pretesting versus posttesting: Comparing the pedagogical benefits of errorful generation and retrieval practice. Journal of Experimental Psychology: Applied, 27(2), 237-257. doi.org/10.1037/xap0000345
  4. Pan, S. C., Sana, F., Schmitt, A. G., & Bjork, E. L. (2020). Pretesting reduces mind wandering and enhances learning during online lectures. Journal of Applied Research in Memory and Cognition, 9(4), 542-554. sciencedirect.com
  5. Soderstrom, N. C., & Bjork, E. L. (2023). Pretesting enhances learning in the classroom. Educational Psychology Review, 35, 97. doi.org/10.1007/s10648-023-09805-6
  6. 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. doi.org/10.3758/s13423-023-02353-8
  7. Toftness, A. R., Carpenter, S. K., Lauber, S., & Mickes, L. (2018). The limited effects of prequestions on learning from authentic lecture videos. Journal of Applied Research in Memory and Cognition. sciencedirect.com
  8. Pan, S. C., & Rivers, M. L. (2023). Metacognitive awareness of the pretesting effect improves with self-regulation support. Memory & Cognition. pubmed.ncbi.nlm.nih.gov/36637644
  9. Mera, Y., Dianova, N., & Marin-Garcia, E. (2025). The pretesting effect: exploring the impact of feedback and final test timing. Journal of Cognition. doi.org/10.5334/joc.455

Keep reading