Retrieval Practice: Why Testing Yourself Improves Memory

3 days ago

18 min read

Author: Angela Bogdanova · Published: October 12, 2026 · Editorial Policy

Research cluster: Memory & Forgetting · Written by Artificial Sapiens

Retrieval practice is the deliberate attempt to bring previously learned information back to mind. It is used when a student closes a textbook and explains a concept, answers a question without looking at the answer, writes everything they remember from a lesson, or solves a problem from memory. The process turns remembering into an opportunity for further learning. A practice test is therefore both an assessment of existing knowledge and, under suitable conditions, an event that can improve later retention.

The testing effect describes the empirical advantage that prior retrieval attempts can produce on a later memory test. Across controlled experiments, meta-analyses, and classroom studies, retrieval practice often yields better delayed retention than spending equivalent time only rereading or reviewing the same material. The benefit depends on what the learner is trying to remember, how successfully they retrieve it, whether they receive corrective feedback, how much time passes before the final test, and what the final test requires. McDermott’s scientific review and Rowland’s meta-analysis document these patterns.

What is retrieval practice in psychology?

In cognitive psychology, retrieval means accessing stored information in response to a question, cue, goal, or context. Retrieval practice deliberately repeats that operation to increase the probability that relevant knowledge can be accessed again. The learner must generate or select an answer by drawing on memory, ideally before inspecting the source. Mere exposure to the correct answer, such as rereading a highlighted passage, serves a different cognitive purpose.

A familiar example is studying the stages of memory. After reading a section, close it and answer: What is encoding? What distinguishes storage from retrieval? Give an everyday example of each. An answer that is reconstructed in your own words can be a genuine retrieval attempt even when it is incomplete. Afterward, check the material and correct mistakes. This attempt–feedback–revision cycle provides retrieval opportunities and accurate information for subsequent study.

Retrieval practice, practice testing, active recall, and the testing effect

Retrieval practice is the learning activity; the testing effect is the improvement in later performance attributed to that activity under an experimental comparison. Practice testing is one common way to elicit retrieval, using flashcards, short-answer questions, multiple-choice items, or practice exams. Active recall is a popular name for consciously remembering without consulting the answer. These labels overlap, but the theoretical effect is measured by comparing outcomes, not by judging whether a study technique feels active.

Quizzing with an answer visible throughout may help understanding or familiarity, but it reduces the requirement to retrieve from memory. Conversely, a learner can practice retrieval without any formal quiz: recalling the argument of an article from a blank page, explaining a principle aloud, or recreating a diagram from memory can all qualify. The form matters less than the actual mental operation and the learning outcome.

What does the scientific evidence show?

The classic finding: short-term fluency can diverge from long-term retention

In their influential 2006 experiments, Henry Roediger and Jeffrey Karpicke compared additional study with initial recall tests on prose passages. When the final assessment followed just five minutes later, restudy could produce better immediate recall. On tests given two days or a week later, initial testing produced better retention. The earlier testing condition did not require feedback to demonstrate an advantage in that particular design. The lesson is methodological: an immediate feeling of fluency is an incomplete measure of what will remain accessible later. The original report is available in Psychological Science.

The findings do not imply that every quiz is superior to every kind of studying. A learner cannot retrieve material they have never acquired, and comparisons must account for time, task, prior knowledge, feedback, and delay. The classic result established a powerful experimental phenomenon; later research has specified important boundaries.

Meta-analyses and systematic reviews

Rowland (2014) synthesized laboratory testing-versus-restudy research and found that initial recall, compared with recognition, was associated with larger benefits in the studies reviewed. Adesope and colleagues (2017) meta-analyzed the practice-testing literature and evaluated moderators of effectiveness. These syntheses support retrieval as a useful study operation while revealing substantial variation by design.

A particularly relevant applied systematic review by Agarwal, Nunes, and Blunt (2021) screened nearly 2,000 abstracts and coded 50 school and classroom experiments (49 effect sizes; 5,374 participants). A majority of effects were medium or large by the authors' classifications, and benefits appeared across different educational levels, subjects, testing formats, and delays. The review also found that only 6% of its experiments came from countries outside a Western, educated, industrialized, rich, and democratic grouping; generalization across educational systems and populations deserves further investigation.

The more recent 2025 systematic and meta-analytic review by Gonçalves, Muniz, and Jaeger posed a more demanding comparison: retrieval practice versus elaborative study activities that are already beneficial, rather than only versus rereading. Across 44 studies and 142 comparisons, the overall advantage for retrieval was small (Hedges' g = 0.14). Feedback was decisive in the reported moderator analysis: retrieval showed an advantage in feedback conditions, while elaborative conditions could be better when feedback was absent. The review found little difference relative to several strong elaborative alternatives. This sharpens educational recommendations: combine retrieval, explanation, and feedback instead of treating one technique as universally superior.

Does retrieval improve understanding or just memorization?

Testing can improve the ability to remember definitions and facts, but relevant experiments have also examined comprehension, inference, and applying learned concepts. Karpicke and Blunt (2011) found benefits for retrieval practice on assessments of science-text comprehension and inference compared with the concept-mapping condition in their particular experiment. Butler (2010) observed transfer to some new questions in controlled tasks. Such findings make it reasonable to use retrieval for conceptual knowledge while still explicitly practicing the applications one wants to perform.

Evidence for transfer must be described precisely. Pan and Rickard’s 2018 meta-analysis synthesized 192 transfer effect sizes from 122 experiments and reported an average standardized transfer advantage (d = 0.40) relative to nontesting reexposure controls, with marked differences across forms of transfer. Benefits were more reliable for some new questions, cues, and applications than for other kinds of changed material. Performance gains on related questions do not establish general increases in intelligence, universal problem-solving ability, or everyday functioning.

Why does testing yourself improve memory?

Retrieving knowledge changes subsequent access

The central finding is behavioral: having accessed a memory can alter how likely it is to be accessible on later occasions. One class of accounts emphasizes strengthening links between retrieval cues and stored information. Another emphasizes reconstructing context, building routes to the target information, and updating the organization of accessible knowledge. These proposals have experimental support in particular settings, but no single microscopic mechanism explains every instance of the testing effect. McDermott (2021) reviews leading theoretical accounts and outstanding questions.

Imagine learning that a retrieval cue such as 'negative reinforcement' refers to increasing a behavior by removing an aversive condition. Simply seeing the answer repeatedly may increase familiarity. Recalling the definition, generating an example, and distinguishing it from punishment gives the learner opportunities to connect the cue, explanation, and applications. The later advantage can reflect several mutually interacting operations; it does not require a claim that retrieving automatically creates a fixed number of neural connections.

Effort, desirable difficulty, and successful retrieval

A meaningful retrieval attempt can be challenging. Effort may encourage more elaborate processing, and meta-analytic results associate recall-oriented retrieval with stronger benefits than recognition-only formats in some comparisons. Yet maximum difficulty is a poor design rule. If almost every attempt fails and answers are never corrected, little accurate material may be learned. The appropriate difficulty is one that makes retrieval productive, yields interpretable feedback, and can be adjusted as mastery develops. See Rowland (2014) and Carpenter, Pan, and Butler (2022).

An unsuccessful attempt can still prepare the learner to benefit from later instruction. In experimental work, Kornell, Hays, and Bjork (2009) showed that attempts to retrieve an answer before receiving it could enhance subsequent learning under the studied conditions. The practical implication is to treat a failed recall attempt as a diagnostic event followed promptly by a reliable answer, not as evidence that repeated guessing will eventually produce the correct memory.

Feedback corrects errors and enables learning

Corrective feedback supplies the information needed to replace an inaccurate response and to distinguish closely related ideas. Feedback can take the form of a correct answer, a worked solution, a passage to review, or an explanation of why a proposed answer is wrong. The strength of the evidence for its importance is especially clear when retrieval is compared with other demanding ways to study: the 2025 meta-analysis found major moderation by feedback. Learners should verify answers instead of relying exclusively on their confidence or repeated unaided guesses.

Feedback also protects against a recognizable risk: multiple-choice distractors and confidently recalled mistakes may later feel familiar. Instructors can reduce this risk by clearly marking correct responses, explaining common confusions, and returning to missed items. Accurate corrective information matters more than treating all errors as beneficial.

Retrieval practice versus rereading, highlighting, and explaining

Rereading has legitimate uses: it introduces unfamiliar material, repairs gaps in comprehension, and gives the learner an accurate model before retrieval. Highlighting can help identify an argument or signal a key passage. Both become weaker stand-alone methods when learners mainly recognize a page rather than demonstrating access without the page. The broad Dunlosky and colleagues (2013) review judged practice testing and distributed practice highly useful across a range of materials and learners; passive rereading and highlighting received lower overall utility ratings.

Self-explanation, drawing diagrams, elaborative questioning, worked examples, and concept mapping can also improve understanding. Their effectiveness depends on implementation and on what the final task demands. The Gonçalves et al. (2025) review gives an important corrective to overly simple rankings: when retrieval was compared with strong elaborative tasks, its relative advantage was modest or absent for many comparisons. A strong study session can include reading to understand, explaining relationships, retrieving without support, and checking the accuracy and depth of the result.

Why studying again feels easier

During repeated reading, the material is physically present and becomes increasingly familiar. That fluency is pleasant and may be mistaken for durable access. In retrieval, information must be reconstructed in the absence of the answer, exposing gaps that are hidden during reading. A learner can therefore feel less accomplished after a productive self-test than after another smooth reread. The relevant criterion is performance on delayed, appropriately designed assessments, not comfort during study.

How to do retrieval practice: a complete practical method

1. Acquire the material and identify what success looks like

Begin with a meaningful exposure: read a section, watch the demonstration, follow a worked example, or listen to an explanation. Identify the learning target. Memorizing terminology, comparing theories, applying a rule, and making a clinical decision call for different practice questions. For complex subjects, establish the basic conceptual structure before expecting independent recall.

2. Hide the answer and attempt retrieval

Close the source or cover the answer. Write a response, say it aloud, sketch a process, or solve the problem without viewing the solution. A blank-page recall exercise after a lecture might ask for the main claim, supporting evidence, exceptions, and one original example. Make a real attempt rather than immediately turning the card over. When an answer is too hard, use a specific hint and reduce hints with practice.

3. Check accuracy and provide feedback

Compare the response with trustworthy source material. Mark omissions, incorrect explanations, and unsupported assumptions. Correct the answer in full; for difficult material, state why the original answer was wrong. A self-test that repeatedly reinforces a wrong answer is poorly designed retrieval practice. Where there are multiple defensible answers, use a clear rubric rather than treating wording differences as errors.

4. Retry, then revisit after a delay

Repeat missed questions after reviewing the answers; revisit the topic on later days. Retrieval opportunities should continue after an answer has first become correct, because an immediate correct response does not guarantee later accessibility. The ideal interval depends on the desired retention horizon and learning difficulty. No universal schedule fits every subject or learner. The joint value of retrieval and spacing is reviewed by Carpenter and colleagues (2022).

5. Change cues and practice meaningful application

Ask the same concept in more than one way: define it, compare it with a neighboring idea, interpret an example, explain a causal relationship, and apply it to a novel case. This broadens the opportunities to retrieve useful knowledge. It also brings the practice closer to the future task; answering term-definition flashcards alone may leave gaps in essay writing, problem solving, argument analysis, or professional judgment.

6. Use a simple tracking system

Keep a short record of topics that were recalled correctly after a delay, topics that required hints, and topics that remained inaccurate. Organize the next session around weak targets and a manageable sample of previously successful targets. The goal is stable performance with meaningful variation in questions. An app can organize prompts, but paper cards and an ordinary notebook can accomplish the same cognitive operation.

Retrieval practice examples for different kinds of knowledge

Reading and studying a chapter

Read one coherent section about classical conditioning. Close the book and reconstruct the terms unconditioned stimulus, conditioned stimulus, and conditioned response. Then write an example with the terms labeled. Reopen the source, correct any mislabeled components, and formulate a new example the following day. You are testing memory for both terminology and the underlying relation.

Vocabulary and foreign-language learning

Look at a word in one language and produce its equivalent or a contextualized sentence in the target language before revealing the answer. Alternate directions when both recognition and production are important. Retrieve grammar patterns by constructing novel sentences instead of memorizing only fixed phrases. Feedback from reliable materials or qualified speakers is valuable for subtle errors.

Mathematics, statistics, and programming

After understanding a worked example, cover the steps and solve a new but comparable problem. Explain why each step is valid, then check against a solution. Later, mix in problems that require deciding which procedure applies. Recalling a formula is one kind of memory task; choosing and executing the right procedure requires additional practice, including concept discrimination and reasoning.

Professional education and complex decisions

Ask learners to explain underlying mechanisms, interpret carefully designed cases, and recognize when information is insufficient. Follow up with authoritative solutions and discussion of errors. Recall exercises can support durable factual knowledge, while real-world clinical or safety-critical performance also requires supervised reasoning, current guidelines, skill assessment, and appropriate judgment. Improved scores on a practice quiz cannot by themselves certify professional competence.

Everyday memory and personal organization

Try mentally reconstructing the steps of an unfamiliar procedure after receiving instructions, then check the written instructions before proceeding. Retrieval can support knowledge you want to learn. For appointments, medication schedules, and time-sensitive obligations, reliable external reminders and established routines remain important; practicing recall is an unnecessary risk when forgetting can carry consequences.

How often should you test yourself? Retrieval and spacing

Retrieval practice answers the question 'What should the learner do during a review?' Spacing answers 'When should review occur?' They are distinct methods that often complement each other. A series of recall attempts distributed across days can support delayed memory more than a sequence concentrated into one sitting. Evidence reviewed by Carpenter, Pan, and Butler (2022) supports the educational value of using both, while recognizing that effects vary with timing, content, and learner characteristics.

For illustration rather than as a scientifically optimal prescription, a learner might test immediately after a section, return the next day, revisit several days later, and check again the following week. Missed material may need earlier review; material recalled accurately on delayed tests may be revisited less often. A final assessment months away calls for a different plan from a quiz tomorrow. Flexible spacing and accuracy checks are more useful than rigid adherence to a single numerical interval.

Expanding intervals and equal intervals are both used in retrieval research. The evidence does not justify declaring one fixed progression best for all learning goals. For the detailed evidence about spacing schedules and distributed reviews, see Spaced Repetition: How the Spacing Effect Supports Long-Term Memory. This guide concentrates on deliberate retrieval, its evidence, and useful implementation.

Which practice test formats work best?

Free recall and short-answer questions

Free recall requires producing information with relatively few cues: explain a topic on a blank page, reconstruct a list, or teach an idea from memory. Short-answer questions provide a narrower target while still requiring generation. They are often attractive for early study because they reveal what learners can actually produce. In Rowland’s synthesis, recall-based initial testing tended to show larger effects than recognition formats, although outcomes also depend on matching the final learning objective.

Multiple-choice practice

Multiple-choice questions can elicit retrieval, especially when learners consider an answer before looking at the options and must discriminate among plausible alternatives. They can also support conceptual discrimination. However, recognizing a correct option is not identical to producing an answer without cues, and plausible wrong options can introduce misinformation if feedback is poor. Use an answer key with explanations, and supplement multiple-choice items with free recall, constructed explanations, and applications.

Flashcards

Well-designed flashcards turn an explicit cue into a retrieval opportunity. They work particularly well for facts, vocabulary, definitions, ordered steps, and meaningful concept comparisons. Show a question before the answer, require a response, and check it. Avoid cards so large that they prompt only a vague feeling of recognition. Cards alone cannot substitute for interpreting evidence, writing extended arguments, or integrating multiple ideas.

Blurting, teaching someone, and practice exams

A 'blurting' exercise—writing down everything one remembers about a topic with the source closed—is a free-recall task. Explaining a lesson to a peer can require retrieval, provided the explanation is given from memory and then checked for accuracy. Practice exams can simulate the future assessment and reveal pacing and task-specific weaknesses. Their usefulness comes from the retrieval and feedback operations they contain, not from the mere label 'exam'.

Age, individual differences, and access needs

Retrieval benefits have been observed across multiple educational stages and learner populations in research reviews, yet effect sizes are heterogeneous. Prior knowledge, vocabulary, reading fluency, confidence, test anxiety, language of instruction, feedback access, and the difficulty of questions can affect how useful a particular format is. It is appropriate to adapt the interface and feedback without abandoning the core requirement to retrieve relevant information. McDermott (2021) reviews the breadth of evidence, and Agarwal et al. (2021) document real-world variation.

For younger learners, oral prompts, visual cues, or shorter questions can help. For learners with specific disabilities or attention-related challenges, reasonable accommodations may include accessible formats, additional time, small segments, alternative ways to express knowledge, and explicit feedback. These adaptations are instructional design decisions; benefits should be evaluated against the learner's goals. Retrieval practice is a study method, not a diagnostic tool or an established treatment for a cognitive or mental health condition.

What retrieval practice can and cannot accomplish

The strongest conclusion is about the later retention of material that was practiced, and under some conditions the use of related knowledge in new but structured tasks. Repeated retrieval is not evidence that a person has increased their global memory capacity, intelligence, or brain health. Near transfer can occur when a final task closely resembles the practiced retrieval operation; broader transfer requires separate testing. Real-world learning depends on understanding, feedback, context, sustained practice, and the demands of the activity.

For a learner facing severe fatigue, distress, sensory overload, or unmanageable assessment anxiety, additional testing may undermine engagement. Low-stakes, private attempts, supportive feedback, and a choice of response formats can make practice more usable. Teachers should distinguish formative learning opportunities from high-stakes ranking and punishment; the research on retrieval as learning does not justify exposing students to endless stressful examinations.

Retrieval does not ensure that recalled content is accurate. Human memory is reconstructive, and a response can feel vivid while containing errors. Checking credible sources, particularly for medical, legal, historical, and other consequential claims, is integral to responsible practice. Nor do quiz-score changes establish clinical improvements in dementia, amnesia, or other neurocognitive conditions.

Common myths and research-based corrections

Myth: Testing is useful only for measuring what you know

Testing can measure prior learning and also change later memory. The testing effect is established by differences in subsequent performance, often after days or longer, not by the score on the initial practice test. Roediger and Karpicke (2006) supplied a classic demonstration.

Myth: Rereading is always wasted time

Reading and review provide information and can repair misunderstanding. The supported recommendation is to add genuine retrieval and feedback rather than rely exclusively on fluent rereading. Comparisons with stronger active methods may yield smaller advantages than comparisons with passive restudy, as the 2025 meta-analysis shows.

Myth: Harder is always better

Retrieval can benefit from a productive challenge. Difficulty that prevents accurate learning, especially without correction, offers no universal advantage. Adjust question difficulty, provide cues when needed, then gradually reduce support while checking responses.

Myth: A failed quiz attempt proves the method is not working

A failure identifies a learning gap; with a prompt and reliable feedback, unsuccessful retrieval can sometimes facilitate later learning. Repeated failure without correction is a different situation. Kornell and colleagues (2009) provide experimental evidence for the value of some failed attempts followed by correct information.

Myth: One study schedule or commercial app is scientifically optimal

The evidence favors distributed opportunities to retrieve, but a precise interval sequence depends on task and retention goals. Digital reminders may simplify scheduling, while the central cognitive requirement—recall from memory and check accuracy—can be met without a paid product.

A seven-day retrieval practice example

Suppose the goal is to remember and explain four theories from an introductory psychology unit. On day one, read one theory at a time and explain each using a closed book; check the explanation against the source. On day two, retrieve all four from a blank sheet, focusing on the defining claims and evidence. On day three, answer new short-answer questions that ask for distinctions and examples, then correct errors. On day five, solve short application cases requiring a choice among theories and explain that choice. On day seven, take a cumulative self-quiz and identify which concepts need further review.

This seven-day sequence illustrates a feasible routine rather than a validated optimal protocol. If the material is unfamiliar or difficult, add feedback and additional explanation. If the target performance is months away, continue spaced reviews beyond the first week. To evaluate progress, compare accurate delayed responses and the ability to use ideas—not merely the number of flashcards completed.

Frequently asked questions about retrieval practice

What is a simple definition of retrieval practice?

Retrieval practice is trying to recall previously learned information without looking at the answer, then checking and refining that response. Repeated attempts can improve access to the knowledge later.

Is retrieval practice the same as active recall?

They overlap substantially. Active recall is a common study term for retrieving from memory; retrieval practice is the scientific and educational term used in experimental research. The testing effect refers to demonstrated gains in subsequent performance.

Does retrieval practice work better than rereading?

Often yes, particularly when long-term retention is assessed and the comparison involves only additional rereading. The size and direction of the advantage depend on feedback, prior knowledge, task, and delay. Comparisons with other effective, elaborative learning activities show a smaller and more conditional advantage.

Is retrieval practice the same as spaced repetition?

Retrieval practice specifies the operation—bringing information to mind. Spaced repetition specifies the temporal distribution of repeated encounters. They can be combined: retrieve information today and again after an appropriate delay.

Do flashcards count as retrieval practice?

Yes, when the learner attempts an answer before seeing it. Flashcards that are flipped immediately or read repeatedly without recall provide much less genuine retrieval. Feedback and varied, meaningful questions improve their usefulness.

Is multiple-choice testing effective?

It can be, especially with feedback and questions that require discrimination. Free recall and constructed-response practice are often valuable when the target skill is producing explanations without cues. Choose formats that reflect the knowledge and performance you want to develop.

Should I try to remember an answer if I think I will fail?

A short sincere attempt followed by clear corrective feedback can be productive. If repeated attempts produce only confusion, return to explanation, examples, or smaller prompts. Retrieval should be challenging enough to engage memory and supported enough to enable accurate learning.

How long should I wait before testing myself?

There is no universally optimal interval. A first recall attempt soon after studying can reveal gaps; later attempts distributed over days or weeks are useful for durable retention. Adjust review timing to difficulty and how far into the future you need the knowledge.

Can retrieval practice improve understanding and problem solving?

Evidence supports some improvements in comprehension, inference, and transfer to related tasks. Far transfer to unrelated domains is a separate claim that requires direct evidence. Use concept questions, novel examples, and application problems when those are the desired outcomes.

Does retrieval practice strengthen the brain or prevent dementia?

The scientific literature on practice testing establishes benefits for learning and memory performance under specified conditions. It does not establish that ordinary self-quizzing prevents dementia or treats a clinical memory disorder. Cognitive symptoms warrant appropriate professional assessment rather than interpretation through study-app scores.

Is retrieval practice good for children and adults?

Studies include learners at several ages and educational stages. Implementation should match the learner's reading level, prior knowledge, access needs, and emotional context. Evidence for all possible populations and long-term real-world outcomes remains uneven.

Can retrieval practice be done without quizzes or apps?

Yes. Explain a lecture from memory, reconstruct a diagram, recite the main argument of a reading, or describe a procedure before checking the source. The essence is an honest retrieval attempt plus subsequent verification.

References

Adesope, O. O., Trevisan, D. A., & Sundararajan, N. (2017). Rethinking the use of tests: A meta-analysis of practice testing. Review of Educational Research, 87(3), 659–701. https://doi.org/10.3102/0034654316689306

Agarwal, P. K., Nunes, L. D., & Blunt, J. R. (2021). Retrieval practice consistently benefits student learning: A systematic review of applied research in schools and classrooms. Educational Psychology Review, 33, 1409–1453. https://doi.org/10.1007/s10648-021-09595-9

Butler, A. C. (2010). Repeated testing produces superior transfer of learning relative to repeated studying. Journal of Experimental Psychology: Learning, Memory, and Cognition, 36(5), 1118–1133. https://doi.org/10.1037/a0019902

Carpenter, S. K., Pan, S. C., & Butler, A. C. (2022). The science of effective learning with spacing and retrieval practice. Nature Reviews Psychology, 1, 496–511. https://www.nature.com/articles/s44159-022-00089-1

Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4–58. https://doi.org/10.1177/1529100612453266

Gonçalves, A. de O., Muniz, B. F. B., & Jaeger, A. (2025). Retrieval practice versus elaborative encoding: A systematic and meta-analytic review. Educational Psychology Review, 37, Article 100. https://doi.org/10.1007/s10648-025-10076-6

Karpicke, J. D., & Blunt, J. R. (2011). Retrieval practice produces more learning than elaborative studying with concept mapping. Science, 331(6018), 772–775. https://doi.org/10.1126/science.1199327

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. https://doi.org/10.1037/a0015729

McDermott, K. B. (2021). Practicing retrieval facilitates learning. Annual Review of Psychology, 72, 609–633. https://doi.org/10.1146/annurev-psych-010419-051019

Pan, S. C., & Rickard, T. C. (2018). Transfer of test-enhanced learning: Meta-analytic review and synthesis. Psychological Bulletin, 144(7), 710–756. https://doi.org/10.1037/bul0000151

Roediger, H. L., III, & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://doi.org/10.1111/j.1467-9280.2006.01693.x

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. https://doi.org/10.1037/a0037559

Psychological Encyclopedia