# Why self-testing beats re-reading 3 times
The study by Roediger & Karpicke (2006) changed what we know about memory. A single recall test produces more rétention than three re-readings. This result, replicated dozens of times since, has a name: the testing effect. The most powerful way to anchor information in your brain is not to review it, but to try to retrieve it. For a university or preparatory school student, this distinction is the difference between a failing and a passing grade.
The foundational experiment: Roediger & Karpicke (2006)
In 2006, Henry Roediger III and Jeffrey Karpicke (Washington University in St. Louis) published a study that would become one of the most cited in educational psychology.
The protocol is simple. All participants read a scientific text. A first group re-reads it three additional times (SSSS condition). The second group re-reads it once, then takes a free recall test -- retrieving from memory everything they remember, without the text (SSST condition). Both groups are tested at 5 minutes, 2 days, and 1 week.
At 5 minutes, the re-reading group scores slightly higher. The information is fresh, the text has just been read four times. Nothing surprising.
At 2 days, reversal. The tested group remembers significantly more than the re-reading group. The gap is approximately 10 to 15 percentage points.
At 1 week, the gap widens further. The re-reading group has lost a massive portion of what they read. The tested group has retained a much higher proportion.
The conclusion is clear: in the short term, re-reading creates the illusion of mastery. In the medium and long term, testing yourself just once produces greater retention than three re-readings.
Référence: Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term rétention. Psychological Science, 17(3), 249-255. DOI: 10.1111/j.1467-9280.2006.01693.x
Why the testing effect works
The central mechanism fits in one sentence: retrieving information from memory strengthens the memory trace of that information.
When you re-read, your brain receives information from the outside. The processing is shallow -- you recognize the content without having to reconstruct it.
When you test yourself, your brain must search for the information internally. It activates retrieval cues, traverses associative networks, reconstructs the answer. This process -- retrieval practice -- physically modifies the memory trace. Each successful retrieval consolidates the neural pathway, making it more accessible next time.
The analogy: re-reading is looking at a road map. Testing yourself is driving without GPS. Harder, but it is what actually teaches you the route.
There is also an indirect effect: testing reveals your gaps. After a quiz, you know what you do not know. After three re-readings, you believe you know everything -- this is the illusion of competence, documented by Kornell and Bjork (2007). Testing calibrates your confidence; re-reading inflates it artificially.
The paradox: it works BECAUSE it is hard
Robert Bjork, professor at UCLA, formalized in 1994 a concept that illuminates the testing effect: désirable difficulties.
The principle is counter-intuitive: learning conditions that slow immediate performance are often those that maximize long-term rétention. Smooth, pleasant conditions (like re-reading) produce fragile rétention.
Testing yourself is hard. You hesitate, get stuck, make errors. But it is precisely this effort that creates deep learning. The brain only strengthens what it is forced to reconstruct.
Bjork distinguishes learning performance (what you display during the session) from actual learning (what you retain long-term). Re-reading maximizes the former. Testing maximizes the latter. If your study sessions feel easy and smooth, there is a strong chance they are ineffective. Discomfort is a positive signal.
Référence: Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In J. Metcalfe & A. Shimamura (Eds.), Métacognition: Knowing about knowing (pp. 185-205). MIT Press.
4 forms of testing
The testing effect is not limited to multiple-choice questions. Research identifies four major forms of testing, each with its advantages.
1. Free recall. You close your notes and write down everything you remember, without any cues. This is the most demanding form and the one that produces the strongest effect. Ideal for verifying your global understanding of a chapter.
2. Cued recall. You are given a cue -- a keyword, the beginning of a définition, an open question -- and you must complete the answer. Classic flashcards work on this principle: the front is the cue, the back is the answer.
3. Recognition (multiple choice). You choose the correct answer from several options. This is easier than recall because you do not have to generate the answer, only recognize it. The effect on memory is real but less powerful than recall. Beware of the risk of encoding wrong answers (distractors) if feedback is not immediate.
4. Spaced repetition flashcards. A combination of cued recall and temporal spacing. You are tested on a card, then it returns at an interval calculated based on your success. This is the most optimized form for long-term retention -- and it is the core of Wizidoo's approach.
The meta-analysis by Rowland (2014), covering 159 studies, confirms that all thèse forms produce a significant testing effect, with an advantage for formats that require generating the answer (recall) rather than simply recognizing it (multiple choice).
Feedback amplifies the effect
A fréquent question: should you provide the correction after the test?
Testing without feedback works. The original Roediger & Karpicke (2006) study did not include feedback -- and the effect was already massive. The mere act of attempting retrieval is enough to strengthen the memory trace.
But feedback amplifies the effect. When you discover your answer was wrong, you correct the erroneous trace and reinforce the correct one. Rowland (2014) estimates feedback adds approximately 0.15 to 0.20 standardized effect size (Cohen's d) compared to testing alone.
The type matters too: elaborated feedback (explaining why) is superior to simple feedback (right/wrong). Immediate feedback appears slightly more effective than delayed feedback.
The implication: test yourself, and if possible, check your answers immediately after. Even without correction, testing is always better than re-reading.
How many tests do you need?
Rowland's (2014) meta-analysis, 159 studies and more than 20,000 participants, provides clear answers.
A single test already produces a significant effect compared to zero tests. The barrier to entry is low. A single quiz after a reading session changes the game.
Multiple tests increase the effect. Each additional test strengthens the trace, with slightly diminishing returns. Three to five tests spread over time produce an excellent effort-to-benefit ratio.
Spacing between tests matters. Testing three times on the same day is less effective than testing once per day for three days. Testing effect and spacing effect reinforce each other. For more on spacing, see our article on spaced repetition.
Référence: Rowland, C. A. (2014). The effect of testing versus restudy on rétention: A meta-analytic review of the testing effect. Psychological Bulletin, 140(6), 1432-1463. DOI: 10.1037/a0037559
Practical application for students
Here is how to integrate the testing effect into your study sessions without overhauling your schedule.
Replace the third re-reading with a test. Read your notes once carefully. Re-read a second time if needed. Then close them and test yourself. Write down key concepts from memory. Answer questions. Use flashcards. This is more effective than re-reading a third, fourth, or fifth time.
Use the 30/70 rule. Spend 30% of your study time reading and understanding, and 70% testing yourself. Most students do the opposite -- 90% reading, 10% testing (when they bother at all). Reverse the proportion.
Start each session with a quiz. Before you begin studying a chapter, test yourself on what you already know. This pre-test activates prior knowledge and prepares the ground for new learning. Even if you answer poorly, the retrieval effort makes the subsequent reading more effective.
Vary the formats. Alternate between free recall, flashcards, multiple choice, and open questions. Variety prevents automation and maintains the level of effort -- therefore the level of learning.
Target your weaknesses. After each test, identify the questions you failed. Thèse are your priorities for the next session. Do not waste time reviewing what you already master.
This is exactly the principle Wizidoo is built on. Every interaction is a quiz -- never passive re-reading. The algorithm identifies your weak points and focuses 70% of questions on your gaps. You import your course material, and Wizidoo automatically generates adaptive quizzes. First course free.
For more active learning techniques, see our article on active vs passive learning.
Is testing without feedback useful?
Yes. The original Roediger & Karpicke (2006) study did not include feedback, and the tested group retained significantly more than the re-reading group. The mere effort of retrieval is enough to strengthen memory. Feedback amplifies the effect but is not a necessary condition.
How many quizzes per session?
There is no magic number, but research suggests that even a single test produces a significant benefit. For a typical study session (30 to 60 minutes), aim for 2 to 3 mini-quizzes spread throughout the session as a good balance. The key is that testing should occupy at least half of your study time. See our article on rapid memorization techniques for complementary strategies.
Does the testing effect work for creative subjects?
The testing effect has been documented primarily for factual memorization (vocabulary, definitions, dates, formulas). For creative subjects where the goal is original production (writing, arts, music), the direct effect is less studied. That said, every subject contains a base of factual knowledge -- and for that part, testing is superior to re-reading. A literature student who tests themselves on literary movements or key quotes will retain more than by re-reading their notes.
Multiple choice or open questions: which is more effective?
Open questions (free recall or cued recall) produce a stronger testing effect than multiple choice, because they require generating the answer rather than recognizing it. Rowland's (2014) meta-analysis confirms this advantage. However, multiple choice remains superior to re-reading. The ideal format is to combine both: multiple choice to cover a large volume of content, and open questions for the most important concepts.
Sources
- Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning. Psychological Science, 17(3), 249-255. DOI: 10.1111/j.1467-9280.2006.01693.x
- Rowland, C. A. (2014). The effect of testing versus restudy on retention. Psychological Bulletin, 140(6), 1432-1463. DOI: 10.1037/a0037559
- Bjork, R. A. (1994). Memory and metamemory considerations in the training of human beings. In Metacognition: Knowing about knowing (pp. 185-205). MIT Press.