The Testing Effect: What the Research Found and How to Use It to Study

By AI Anyone · 2024-03-10 · Updated Oct 21, 2026 · 9 min read · Learning Tips

The testing effect is the finding that taking a test helps you remember more than studying again. Here is the plain definition, what Roediger and Karpicke and later research found, the research-methods meaning, and eight self-testing formats to use tonight.

The testing effect is the finding that taking a test on material helps you remember it better later than spending the same time studying it again. A test is not only a way to measure what you learned; the act of pulling an answer out of memory changes that memory, making it easier to recall next time. It is one of the most reliable results in learning research, and you can put it to work with nothing more than a blank page.

This article covers what the research found, starting with the study that made the effect famous, then turns it into practical self-testing formats you can use for any subject.

The topic draws steady interest. The term "testing effect" is searched about 1,900 times a month in the US, and "testing effect psychology definition" about 880 times (DataForSEO, October 2026), which suggests a lot of psychology students meeting it for the first time. The top Google result for "testing effect" right now is a journal article on the APA's PsycNet database (DataForSEO, October 2026), useful if you have access and want the methods, hard going if you want the idea.

What is the testing effect?

The testing effect, also called test-enhanced learning, means that recalling information strengthens long-term memory for it more than rereading or restudying does. The test can be anything that makes you produce an answer from memory: a quiz, a flash card, a practice question, or simply writing down everything you remember about a topic.

Three details define it:

  • The comparison is fair. Testing is compared with the same amount of time spent restudying, not with doing nothing.
  • The benefit shows up later. On a test a few minutes away, restudying can look as good or better. The advantage of testing appears after days or weeks.
  • It works without a grade. Low-stakes or no-stakes practice tests produce the effect. The value comes from the recall, not from the pressure.

Testing effect psychology definition, in plain words

For an exam answer: the testing effect is the enhanced long-term retention of information that has been retrieved from memory during a test, compared with information that has only been restudied. In plain words: trying to remember something helps you remember it.

You will also see it called retrieval practice. Strictly, the testing effect is the research finding and retrieval practice is the study strategy that uses it. About 2,900 people a month in the US search "retrieval practice" (DataForSEO, October 2026), more than search for the effect itself, mostly teachers and students looking for ways to apply it.

Roediger and Karpicke (2006): the study that made it famous

The best known demonstration of the testing effect is Roediger and Karpicke (2006), published in Psychological Science. Read the study.

College students read short prose passages on science topics. After a first reading, some students read the passage again, while others took a recall test: they wrote down as much as they could remember, with no feedback. Then everyone took a final test, either 5 minutes later, two days later or one week later.

What they found:

  • At 5 minutes, restudying did slightly better. If you only care about the next few minutes, rereading looks like the winner.
  • At two days and one week, testing did better. The students who had practiced recall forgot less over the delay.
  • Students misjudged it. In a second experiment, students who had studied the passage repeatedly predicted they would remember more than those who had been tested repeatedly. On the one week test, the order reversed.

That last point explains why so many students reread. Rereading makes the text feel familiar, and familiarity feels like knowing. A recall test feels harder and exposes gaps, so it seems less productive, even though it is doing more for memory.

What the wider research found

Later work has tested the effect against tougher comparisons, in classrooms, and across many types of material.

Against concept mapping. Karpicke and Blunt (2011), published in Science, compared retrieval practice with building concept maps, a study activity many teachers favor because it asks students to organize ideas. Students who practiced recalling a science text did better a week later than students who built concept maps while viewing the text, including on questions that required drawing inferences rather than repeating facts. Read the study.

Across many studies. Adesope et al. (2017) carried out a meta-analysis of practice testing, pooling results from a large body of experiments. Practice testing outperformed restudying and other comparison activities, and the benefit appeared across age groups, in both classroom and laboratory settings, and with different question formats. Read the meta-analysis.

In a rating of study techniques. Dunlosky et al. (2013) reviewed ten common techniques and rated practice testing as one of only two with high utility, alongside spreading study out over time. Read the review.

The pattern across these is consistent: if you want to remember something next week or next month, spend more of your study time answering questions and less of it reading.

Why recalling information strengthens memory

Researchers still debate the exact mechanism, but a few explanations are widely discussed, and each has a practical lesson.

  • Retrieval is practice for the real task. An exam asks you to recall. Rereading practices recognizing; testing practices the thing you will actually do.
  • Effort seems to matter. Recall that takes some struggle tends to help more than recall that is instant. A question you answer easily five seconds after reading teaches little.
  • Tests show you what you do not know. A failed recall attempt is information. It tells you exactly where to spend the next 10 minutes, which rereading never does.
  • Recall organizes knowledge. Writing everything you remember about a topic forces you to put it in some order, which builds connections between facts.

The practical lesson from all four: test yourself before you feel ready, accept that it will feel worse than rereading, and check your answers afterwards.

What are the three R's of memory?

The three R's of memory are usually given as recording, retaining and retrieving: getting information in, keeping it over time, and getting it back out when you need it. The same three stages are often taught as encoding, storage and retrieval. The testing effect is about the third R, and its surprise is that practicing the third stage also strengthens the second.

What was the Ebbinghaus experiment?

The Ebbinghaus experiment was a series of memory tests Hermann Ebbinghaus ran on himself in the 1880s, learning lists of nonsense syllables and measuring how quickly he forgot them. It produced the forgetting curve: memory falls off fast in the first hours and days, then more slowly. Ebbinghaus (1885) is where the modern science of memory starts, and the testing effect is one of the most effective known ways to flatten that curve.

The testing effect in research design: a different meaning

If you met the phrase in a research methods class, it may mean something else. In experimental design, a testing effect is a threat to internal validity: taking a pretest changes how people score on a later posttest, because they have practiced the test, remember the questions, or have been primed to think about the topic. It is one reason studies use control groups that take the same pretest.

The two meanings are related. The research-design problem exists precisely because taking a test changes people. In a study, that is a confound to control. In your own learning, it is the point.

Retrieval practice strategies: eight self-testing formats

Here are formats that put the testing effect to work, from fastest to most demanding. Most students should use three or four of them regularly.

FormatHow it worksBest forTime
Brain dumpClose everything and write all you remember about a topic, then checkAny subject, start of a session5 to 10 minutes
Flash cardsQuestion on the front, answer on the back; answer before you flipTerms, vocabulary, formulas, dates10 to 15 minutes
End of chapter questionsAnswer the textbook's questions before rereading the chapterTextbook courses15 to 30 minutes
Cover and recall notesFold your notes so only the headings show; recall what is under eachLecture notes10 minutes
Blank page diagramRedraw a process or structure from memory, then compareBiology, anatomy, engineering10 minutes
Write your own examWrite ten questions a teacher might ask; answer them two days laterBefore a unit test20 to 30 minutes
Explain it out loudTeach the topic to a friend, a pet, or an empty room, without notesConcepts and processes5 to 10 minutes
Past paper under timeA full past exam with the real time limitThe final week before an examExam length

Rules that make any of these work:

  • Answer first, check second. Looking at the answer and nodding along is rereading.
  • Always check. Read the correct answer for anything you missed or guessed. Feedback stops you from rehearsing an error.
  • Repeat across days. A second test two or three days later is worth more than a second test the same evening. For how to schedule that, see our guide to the science of distributed practice.
  • Keep stakes low. These are practice, not judgments. A low score today is the most useful thing a practice test can give you.

Teaching something out loud has its own research line; our post on the protégé effect covers it.

A testing effect example you can run in one evening

About 90 people a month in the US search "testing effect example" (DataForSEO, October 2026). Here is one you can run on yourself with any chapter.

  1. Read the chapter section once, carefully, for 20 minutes.
  2. Split the follow-up time in half. Spend 10 minutes rereading the first half of the section, and 10 minutes doing a brain dump of the second half, then checking it.
  3. Do nothing else with it for a week.
  4. Test both halves. A week later, write down everything you remember from each half without looking.

Before step 4, guess which half you will remember better. Many people pick the half they reread, because it felt smoother at the time. Compare your guess with the result.

Try this: one self-test before tonight's reading

Before you open your notes tonight, take a blank page and spend 5 minutes writing everything you remember from your last study session on the subject. Then open your notes and mark three things: what you got right, what you got wrong, and what you forgot entirely. Spend the rest of the session on the second and third lists. In two days, repeat the brain dump on the same topic.

If you want ready-made questions, type your topic into AI Anyone Courses and it builds a short course with lessons and quick checks; do the checks before you read the lessons, not after. Or start a voice call with a character in your browser and ask it to quiz you on one topic, one question at a time, with no hints until you have answered. Because the call is spoken, you have to produce each answer rather than recognize it. After the call the transcript lands in your chat, so you can see exactly which questions you missed and test yourself on those next.

The short version

Taking a test on what you studied helps you remember it longer than studying it again, and the research behind that claim is broad and consistent. It feels worse while you do it and works better afterwards. Replace some of your rereading with recall: a brain dump, a set of flash cards, or ten questions you wrote yourself, checked every time and repeated a few days later.

Testing EffectRetrieval PracticeLearning ScienceStudy TechniquesPsychology

Keep going with this topic

Two ways to go deeper right now. Chat is free with an account, and a full course takes a few minutes to build.

Chat free with a science and tech mentor Generate a course on this
More from the AI Anyone blog →
© 2026 AI Anyone. All rights reserved.