How Practice Testing Improves Exam Scores: The Testing Effect Explained

The testing effect is the finding that taking a test on material improves your memory of it more than spending the same amount of time re-studying it. The test is not an assessment of learning that has already happened. The test IS the learning event. Retrieval is not a measurement tool. It is a memory-strengthening tool. Roediger and Karpicke (2006) proved this definitively: students who read a passage once and took three recall tests retained dramatically more at 1 week than students who read the same passage four times with no testing. The test group spent less time with the material. They retained more. The test was the better teacher.

One proposed neural mechanism is reconsolidation. Every time you retrieve a memory, the neural trace becomes temporarily unstable — open to modification and strengthening — before re-stabilizing in a more durable form. Re-reading does not trigger retrieval. It triggers recognition, which involves different brain regions and does not initiate the reconsolidation process. A student who re-reads a chapter 4 times gains little that lasts a week. Mostly, they have refreshed their familiarity with the text. A student who tests themselves 3 times has strengthened the memory traces 3 times. The difference shows up on exams.

Butler (2010) added a critical finding: the testing effect extends beyond factual recall to transfer — the ability to apply knowledge to new problems. Students who were repeatedly tested on material performed better on novel inferential questions than students who repeatedly studied. Testing builds flexible, usable knowledge. Re-studying builds narrow, recognition-based familiarity. For any exam that asks you to apply concepts rather than regurgitate definitions, testing is not just better — it is essential.

Karpicke and Roediger (2008) published in Science a finding most students do not expect. Once students had correctly recalled a foreign-language word pair, studying it again did nothing for their score a week later, while retrieving it again produced a large gain. Retrieval, not re-exposure, did the work. Feedback matters most when you get an answer wrong: Pashler, Cepeda, Wixted, and Rohrer (2005) found that showing the correct answer after an error raised final retention by 494%, while feedback after correct answers made little difference. So test yourself early, and check the answer every time you miss.

This finding dismantles the student's primary objection to self-testing: 'I do not know the material well enough to test myself yet.' The research says the opposite: test yourself as early as possible, even when you will fail most of the attempts. Searching your memory, even when the search fails, prepares you to learn the answer when you check it (Kornell, Hays, and Bjork, 2009). Testing when you feel unprepared is not a waste. It is the most efficient use of your study time.

The practical protocol: study a chapter once. Immediately close the book and test yourself. You will recall 30 to 60% of the material. That is fine. The 40 to 70% you cannot recall are exactly what you need to focus on. Open the book, review only the missed items for 5 minutes, close the book, and test yourself again. The second retrieval attempt will be dramatically more successful because the failed first attempt primed your memory for the correct information. Two retrieval attempts — one failed, one successful — produce more durable encoding than any number of re-reads.

Larsen et al. (2009) conducted a randomized controlled trial with paediatric and emergency medicine residents learning two neurology topics, status epilepticus and myasthenia gravis. Each resident took repeated tests on one topic and studied a review sheet on the other. About 6 months later, they remembered the tested topic significantly better. This is real-world evidence from a high-stakes domain where knowledge gaps have patient consequences. The testing effect is not a laboratory curiosity — it works under authentic educational conditions with complex material.

McDaniel et al. (2011) tested the effect in 8th grade science classes. Within each class, some of a unit's content was covered by low-stakes multiple-choice quizzes with feedback and some was not. Quizzing produced gains of 13% to 25% on the unit exams, review quizzes just before the exam helped most, and the advantage for quizzed content lasted to the cumulative semester and end-of-year exams. The authors attribute the effect to retrieval practice: the benefit was for the material students had been quizzed on, not a general boost to learning.

Roediger, Agarwal, McDaniel, and McDermott (2011) found the same in a sixth grade social studies course at a suburban middle school. Across three experiments, items students had been quizzed on scored better on chapter and semester exams than items that were not quizzed or that were simply shown twice for review. The effect held when the final test asked for short written answers instead of multiple choice. The testing effect is one of the most robust and replicable findings in educational psychology.

After every study session, give yourself a self-test. Do not wait until you feel ready. Do not review your notes first. Close everything. Write down every concept, definition, formula, and relationship you can recall from the session. When you run out of things to write, open your notes. Compare. The items you missed are your targets for the next session. This 5-minute post-study test will do more for your retention than 30 minutes of additional reading.

Build a question bank from your course material. For each major concept, write one open-ended question that requires explanation, not recognition. Do not write multiple-choice questions — the testing effect is strongest for free recall formats that require generating answers from memory without cues. Test yourself on the question bank at increasing intervals: after 1 day, 3 days, 7 days, and 30 days. Shuffle the order each time. A bank of 50 questions tested 3 times across 2 weeks will produce dramatically better exam performance than re-reading the same 50 concepts for the same total time.

After each self-test, spend more time analyzing your errors than you spent taking the test. For every wrong answer, identify the error type: knowledge gap (you never learned it), retrieval failure (you knew it but could not access it), reasoning error (you knew the facts but applied them wrong), or careless error (you misread or rushed). Each error type requires a different corrective strategy. A self-test you take, score, and ignore teaches far less than one where you analyze and target the errors.

Vidbyte does not space your reviews or classify your errors; you decide when to test yourself and why each answer was wrong. What it removes is the work of building the tests: generate a quiz or a sectioned exam from your material in the Sandbox.

Every quiz and exam is saved to your Review Vault, so a self-test can be retaken on each review day. Go through the misses yourself using the error types above. After a quiz, Update Quiz can generate a follow-up version: under Next Quiz, pick Review Weaknesses, Next Topic, Challenge Mode, or Similar Quiz.

Generating quizzes and exams needs Google sign-in and an active Explorer, Pioneer, or Enterprise subscription. To see how your current habits score on retrieval and spacing first, take the free Study Evaluator, which needs no account.