Skip to content

Pretesting (Prequestions Before Study)

Principle · All principles
Evidence · 28 claims (6 for, 19 mixed, 3 against) · 27 studies (13 causal, 5 quant-synthesis, 4 review, 4 theoretical, 1 associational), q1–q4 · 12 of 27 report an effect size · 22 claims rest on one study

Conditional relationship

This page owns the decision of whether to test learners on material before they study it. When learners are about to study material they do not yet know (a text, a video, a lecture, a set of facts or terms), asking them to attempt a few questions on its central content first, and then letting the study that follows supply the answers, is expected to improve later memory of the questioned content (immediate-performance and delayed-retention on items about the same content) compared with studying for the same time without the questions. The attempt does not have to succeed: in the recorded experiments almost every pretest item was answered wrongly or left blank, and the benefit was measured on exactly those items.

Claims in the wiki test this relationship, in its narrow form: two multi-experiment laboratory studies with undergraduates (Pretesting enhances learning [+S]; Experiment 1 of the text study, 75% against 56% correct, d = 1.1, counting only items failed at pretest), and a 2023 review whose accounts of text, video and classroom studies are recorded as separate claims. Both laboratory studies share an author, and the review's findings reach the wiki second-hand, one claim per study it describes.

The relationship is conditional on six things, each from the evidence below:

How this sits beside its neighbours. Retrieval Practice is testing after study, on material already met, where failure without feedback removes the benefit; here failure is the expected case and the study that follows is the feedback, so the two should not be assumed to share moderators. Productive Failure is an attempt before instruction too, but at a problem whose solution methods learners must generate, judged on conceptual understanding and transfer; a pretest is a short question attempt judged on memory of the answer. Pre-reading Questioning is this page narrowed to texts, with the attempt–read–check routine and a progression to learner-written questions; Activation is the wider family of making prior knowledge visible before instruction, of which a pretest is the best-evidenced form; Immediate Feedback holds the general timing question. This page is not one of the multimedia learning principles: it is about the order of testing and study, whatever the medium.

Default design, while the dose is untested

The direction and the conditions are from the evidence; no claim in the wiki tests a number of questions, a length of attempt or a schedule across a course, so those are labelled proposals.

  1. Pick the content the later assessment will ask about. Write the pretest on the central ideas, terms or facts of the coming material that the course's own assessment covers, not on peripheral details. The gain is on questioned content (Pretesting enhances learning [+S]); untested content may not benefit (Effects on untested materials are inconsistent [~M]).
  2. Three to six questions per segment of study, two to five minutes. The recorded studies used five questions before a text passage (Pre-instruction testing enhances memory for directly tested text content [+M]) and six questions among two-minute segments of a seven-minute video (Pre-instruction testing enhances memory for directly tested video content [+M]); the count for other material is an untested proposal.
  3. Ask for an answer or a best guess on every item, privately and unscored. Short answer or multiple choice; multiple-choice pretests at the start of lectures worked in a ten-week course (Classroom pretesting improves high-stakes exam performance [+M]). Say that most guesses will be wrong and that this is the point (framing: untested proposal, against the persistence cost on Pre-tests reduced adult learners' persistence in a MOOC [-M]).
  4. Make sure the study answers every question, soon. The text, video segment or lecture must contain the answer. For vocabulary or other paired items, show the correct answer straight after each item (Feedback timing moderates pretesting [~M]); for richer material, the lesson can supply it.
  5. Study at once, with no extra time. The benefit was against equal extra study time (Pretesting enhances learning [+S]), so the pretest minutes can come out of the study time.
  6. Across a course: open each unit or lecture with a pretest on that unit's content (early sessions: teacher-written items, answers given within the session); in middle sessions, return the pretest items later as retrieval practice after study, which is a different principle with its own evidence; in late sessions, when learners already answer most pretest items correctly before study, the pretest has become a review, so drop it or replace it with harder items on the coming material. Moving on: when a learner gets most of a unit's pretest right, skip the pretest for that learner. The arc and the threshold are untested proposals; the ten-week classroom study is the only multi-week course recorded (Classroom pretesting improves high-stakes exam performance [+M]).
  7. Judge it at a delay, on the questioned content. The effect was largest at one week in the experiment that compared delays (The pretesting effect grows with retention interval [+M]); judge it with the course's own end assessment rather than adding a test.

Dose, as a proposal: two to five minutes of pretest per study segment of 10–30 minutes (videos of 5 to 30 minutes are the range on the video claim), one pretest per unit, for as many units as the course has.

Fitting the design to a situation

Three facts change the decision most. Ask for them when the brief omits them:

  • Will the study that follows answer the questions, and when? If unstated, ask: which part of the material answers each question, and how soon after the attempt does the learner meet it?
  • What does the end assessment cover? If unstated, ask: will the assessment ask about the same ideas the pretest can ask about, or about everything in the material?
  • Can learners leave, and how will they take failing? If unstated, ask: is the course optional or self-paced, is anything scored, and how confident are these learners?
If the brief says… Then change… Basis
Learners are novice on the topic and will know almost none of the answers Use the pretest as designed; say that wrong guesses are expected Pretesting enhances learning [+S] (undergraduates, items answered at floor, laboratory)
Learners are advanced or already answer most items Drop the pretest; use retrieval practice after study instead, or pretest only the genuinely new content untested; the recorded studies used new material
Learners are children Two or three questions read aloud, answers by pointing or choosing; answer each straight after untested; every pretesting study whose learners the claims name used adults
online-self-paced course or MOOC Keep the pretest to two or three unscored items, framed as a puzzle; show answers in the next screen; watch drop-off at the pretest screen and remove it if completion falls Pre-tests reduced adult learners' persistence in a MOOC [-M]
System-paced video Put two or three questions before each segment of a few minutes, on that segment's content Pre-instruction testing enhances memory for directly tested video content [+M] (videos of 5 to 30 minutes)
Print or a reading assignment Questions before the passage on its central concepts; then read; then check guesses against the text (see Pre-reading Questioning) Pre-instruction testing enhances memory for directly tested text content [+M]
Lecture with slides, one teacher for many learners Multiple-choice pretest at the start of the lecture on paper or clickers; no individual feedback needed when the lecture answers each item Classroom pretesting improves high-stakes exam performance [+M] (ten-week course, pen-and-paper pretests, no feedback)
Phone only One question per screen, multiple choice, the answer shown on the next screen untested proposal
VR or a simulation Ask the questions before entering, not inside the scene, and make sure the scene answers them untested proposal
Vocabulary, terms or other paired items Show the correct answer immediately after each pretest item, not at the end of the set Feedback timing moderates pretesting [~M]
Paired items with no meaningful link (arbitrary codes, unrelated word pairs) to be recalled from a cue Do not count on a pretest; teach a link or a mnemonic first, or use recognition practice Semantic association moderates pretesting [~M]
The end assessment covers everything in the material, not a few central ideas Spread questions across the material, prefer why and how questions, and add a whole-material summary after study Effects on untested materials are inconsistent [~M]; Question prompts improve learning [~M] (factual pre-questions depressed learning of unquestioned content)
A single session Pretest, study, done; expect the effect on a later test of the same content, not a lasting course effect Pre-instruction testing enhances memory for directly tested text content [+M] (immediate and one-week posttests)
A fixed short time budget Take the pretest minutes out of study time rather than adding them; attempting beat equal extra study Pretesting enhances learning [+S]
The answer cannot follow soon (no feedback, the material arrives weeks later) Do not pretest; an unresolved guess is a guess left standing Retrieval Failure Reduces Benefit [-M] (testing after study, carried here: mostly failed attempts without feedback gave no testing effect)
Learners have low literacy or are reading in a second language Use simpler words in the questions than in the material, read them aloud, accept answers by choice untested proposal
Learners with a visual or hearing impairment Make items screen-reader readable and captioned like the rest of the course; give the answer in the same accessible form untested proposal
The brief brings its own materials, routine and assessment or rubric Keep them as given; write pretest items on the content the brief's assessment covers, not copies of its items, and add no test the brief did not ask for untested proposal; the classroom gain was on exam questions about pretested content (Classroom pretesting improves high-stakes exam performance [+M])
The material is private or the learners' answers are sensitive Collect guesses privately and do not display or store them untested proposal
The task must be kept whole (a complex problem, a case) Use an attempt at the whole problem before instruction instead of item pretests (see Productive Failure) Productive failure improves conceptual learning [+M] (a different design, judged on conceptual outcomes)
High stakes, or every activity is graded Never grade the pretest; score only the later assessment Classroom pretesting improves high-stakes exam performance [+M] (the gain showed on a high-stakes final exam; whether the pretests carried marks is not recorded); scoring the pretest is untested
Learners or teachers doubt that testing before teaching helps Explain the purpose; do not rely on learners choosing it themselves Learners fail to recognize the benefits of pre-instruction testing [~M]

Of these 22 rows, 15 cite a claim; the children, advanced, phone, VR, literacy, impairment and private-material rows are untested proposals, and two rows carry a neighbouring finding (retrieval after study; productive failure) and say so.

Observation, state and explanation

What a designer can observe: each pretest answer, whether it was blank or a guess, how long the learner spent, whether they reached the study that followed, and later performance on questioned and unquestioned items. "Is curious about the answer" or "has a search set for it" are inferred states. A wrong or blank pretest answer is the expected case and shows nothing about effort. These rows are proposals, untested with learners as diagnostics.

Response Plausible explanations Observation that could change the interpretation
Most pretest items answered correctly The material is not new to these learners; items are too easy or guessable Ask how they knew; if they knew, drop the pretest for them and use retrieval practice after study.
Later test good on questioned items, poor on the rest Attention narrowed to the questioned content; the questions sampled too little Compare questioned and unquestioned items of similar difficulty; add a whole-material task and see whether the gap closes.
No gain on questioned items The study did not answer them; the answer came too late for paired items; the test came at once and the effect shows later Check that each item's answer is in the material and when it was met; re-test at a week.
Learners leave or skip at the pretest screen (online) Discouraged by failing; the step looks like a graded test; time cost Compare completion with and without the step, or with an unscored "guesses expected" version.
Learners copy answers forward or look them up They read the pretest as a graded test Restate that it is unscored; remove anything that looks like a score.

Evidence and qualifications

  • Pretesting enhances learning [+S]: two multi-experiment laboratory studies with undergraduates. Text: five experiments with an essay on human vision; attempting questions before reading beat equal extra reading time on the same content in every experiment (Experiment 1: 75% against 56%, d = 1.1), analysing only items failed at pretest; emphasis by italics or bold did not explain it, and reading the questions without attempting helped less. Trivia and word pairs: six experiments, failed attempts followed by the answer beat reading question and answer together (Experiment 1: 41% against 31%, d = 0.58), and items first answered with a wrong guess were later recalled better than items left blank. Every trial ended with the answer, so the studies show the attempt enhancing the study that followed it, not learning on its own. The two studies share an author and use short delays.
  • Pre-instruction testing enhances memory for directly tested text content [+M], video content [+M] and classroom exam performance [+M]: the 2023 review's accounts of Richland et al. (2009; immediate and one-week posttests), Carpenter and Toftness (2017; six questions among two-minute segments of a seven-minute video) and Soderstrom and Bjork (2023; ten-week research-methods course, pen-and-paper pretests without feedback, 8–9% better on final-exam questions about pretested content). No effect sizes are recorded on these claims.
  • The pretesting effect grows with retention interval [+M]: one experiment (Kliegl et al. 2022) with three delays, largest at one week.
  • Effects of pre-instruction testing on untested but studied materials are highly inconsistent [~M]: a benefit for untested content in one of five similar experiments. This is the limiting claim: the page promises nothing beyond the questioned content.
  • Question prompts improve learning [~M]: a review of adjunct questions in prose (abstract only): questions help on the same content, higher-order questions generalise further, and factual pre-questions can hurt unquestioned material.
  • Feedback timing moderates pretesting [~M] and semantic association moderates pretesting [~M]: the review's synthesis of paired-associate studies; immediate correct-answer feedback was needed for word pairs, delayed feedback sufficed for trivia, and unrelated pairs showed no cued-recall benefit.
  • Pre-tests reduced adult learners' persistence in a MOOC [-M]: 399 adults randomly assigned to pre-tests (with none, basic or elaborate feedback) or none; pre-tests reduced persistence, did not change post-test learning across everyone, and improved it among completers. Read from the abstract; motivation itself was not measured.
  • Learners fail to recognize the benefits of pre-instruction testing [~M]: one experiment in the review (Huelser and Metcalfe 2012): pretested related word pairs were remembered best and pretesting was ranked least effective.
  • Retrieval Failure Reduces Benefit [-M]: a 61-study meta-analysis of testing after study, where mostly failed attempts without feedback gave no testing effect (g = 0.03). It is the boundary for a pretest whose answer never arrives, carried from post-study testing.

Taken together: for new material studied soon after, a short attempted pretest beats the same time spent studying, on the questioned content, from immediate tests to a week, in laboratory studies and one multi-week classroom course; for unquestioned content, for arbitrary pairs without immediate answers and for learners who can leave, it is not established or has a cost. Where the claims name the learners, they are adults, mostly undergraduates.

The evidence by question

Further evidence, not yet read against this model

Objective and learner-valued goal

The designer's objective is that learners remember and can use the questioned content on a later assessment the course already has, at a stated delay. The pretest answers are a means, not an outcome, and should never be read as a measure of the learner. Ask separately what the learner wants from the material: a learner who wants to finish quickly experiences the pretest as a delay, and one who reads failing as a verdict may stop, as some did in the MOOC. Record where those diverge.

What would revise this model?

The expectation should weaken if classroom comparisons with comparable learners and equal time find no advantage on questioned content at a delay, or find that the gain on questioned content is matched by a loss on unquestioned content, so that learning of the whole material is unchanged. It should be narrowed if the effect holds only for fact-like content and not for concepts or application. If unscored, explained pretests still reduce persistence in courses learners can leave, the self-paced row should drop pretests. Studies with children or school-age learners would extend or limit a model that rests on adults. Do not protect the model by calling every null "the wrong questions" after the fact.

Claims

Examples

Key Sources

  • 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:10.1037/a0016496
  • 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:10.1037/a0015729
  • 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, 97. doi:10.1007/s10648-023-09814-5
  • Janelli, M., & Lipnevich, A. A. (2021). Effects of pre-tests and feedback on performance outcomes and persistence in Massive Open Online Courses. Computers & Education, 161, 104076. doi:10.1016/j.compedu.2020.104076
  • Hamaker, C. (1986). The effects of adjunct questions on prose learning. Review of Educational Research, 56(2), 212–242. doi:10.3102/00346543056002212
  • Rowland, C. A. (2014). The effect of testing versus restudy on retention: A meta-analytic review of the testing effect. Psychological Bulletin, 140(6), 1432–1463. doi:10.1037/a0037559
Page metadata
Field Value
type principle
id pretesting
title Pretesting (Prequestions Before Study)
description When learners are about to study material they do not yet know, asking them to attempt questions on it first, even though most attempts will be wrong, and then letting the study, lesson or feedback supply the answers, is expected to improve later memory of the questioned content over studying for the same time without the questions; laboratory and classroom studies recorded in the wiki test this, with the gain bounded to the questioned content, dependent on the answer arriving (immediately, for arbitrary paired items) and offset in one online course by learners who stopped.
status review
canonical True
generated by claude/unspecified on 2026-10-14
trust unverified — no one has confirmed this page's content
sources 6 — richland-kornell-kao-2009, kornell-hays-bjork-2009, pan-carpenter-2023, janelli-lipnevich-2021, hamaker-1986, rowland-2014

Frontmatter is the page's machine-readable record — see the Schema & Guide.