Does Quizzing Yourself Actually Work? What the Testing Effect Data Shows
Updated: 1 day ago
Most people review the way they were taught to in school: read it again, highlight it, read it once more. It feels productive. The material looks familiar, and familiar feels like learned.
One of the best known experiments in memory research tested that feeling directly, and the result is a little uncomfortable. Here is what it found, with the arithmetic worked through.
The experiment
In 2006 Henry Roediger and Jeffrey Karpicke of Washington University in St. Louis published Test-Enhanced Learning in Psychological Science. Students read short science passages, about 250 to 275 words each, on topics like "The Sun" and "Sea Otters." Each passage was split into 30 idea units for scoring.
After the first reading, students either read the passage again or took a recall test: a blank sheet, write down everything you remember. Nobody got feedback on their tests. Then they took a final test after 5 minutes, 2 days or 1 week.
Experiment 1: 120 students, one extra read versus one test
Share of the passage recalled on the final test:
After 5 minutes: rereading 81 percent, testing 75 percent
After 2 days: rereading 54 percent, testing 68 percent
After 1 week: rereading 42 percent, testing 56 percent
At five minutes, rereading wins. That is the trap: if you check yourself right after reviewing, rereading looks like the better strategy. But by two days the result has flipped, and it stays flipped at a week.
Two calculations make the difference concrete.
Relative advantage. At one week, 56 divided by 42 is 1.33. The tested group remembered about a third more.
Time bought. The tested group recalled 56 percent after a week, slightly more than the rereading group managed after just 2 days (54 percent). In the authors' words, one recall test "prevented forgetting of information for an additional 5 days."
Experiment 2: 180 students, four reads versus three tests
The second experiment pushed further. Students got four periods of 5 minutes each with one passage, arranged three ways: four study periods (SSSS), three study periods and one test (SSST), or one study period and three tests (STTT). Thirty students were in each condition at each delay.
Final recall:
After 5 minutes: SSSS 83 percent, SSST 78 percent, STTT 71 percent
After 1 week: SSSS 40 percent, SSST 56 percent, STTT 61 percent
Students also tallied how many times they read the whole passage. The four-study group averaged 14.2 readings, the three-study group 10.3, and the one-study group just 3.4.
Here is the step by step comparison at one week:
STTT recalled 61 percent after reading the passage 3.4 times.
SSSS recalled 40 percent after reading it 14.2 times.
61 divided by 40 is about 1.53, so the quizzed group remembered about 53 percent more.
Per reading, that is 61 divided by 3.4, roughly 18 points of recall per read, against 40 divided by 14.2, roughly 2.8 points per read. That ratio is crude, but it shows how little the extra rereading bought.
The paper also measured forgetting as a share of what each group knew at the end of the session. The four-study group forgot 52 percent of it in a week. The one-test group forgot 28 percent. The three-test group forgot 14 percent.
The confidence problem
At the end of the learning session, students rated on a 1 to 7 scale how well they expected to remember the passage in a week. The four-study group gave the highest rating, 4.8. The three-test group gave the lowest, 4.0. The group that felt most prepared remembered the least, and the group that felt least prepared remembered the most.
This is why the testing effect is so underused. Rereading feels fluent, and fluency feels like learning. Struggling to recall feels like failing, even while it is doing the work.
How to use it without a study plan
The practical takeaways are simple:
After you read something you want to keep, close it and try to recall the main points before rereading.
Quiz yourself again a day or a few days later. The benefit showed up on delayed tests, not immediate ones.
Don't trust the feeling of familiarity. Trust what you can produce from memory.
One caution: these were college students recalling short prose passages in a lab. Real life is messier, and no single study settles a question. But the pattern held across both experiments and on every delayed test, and the core idea costs almost nothing to try.
Where Study Textapp fits
Study Textapp texts you one short fact a day about topics you pick in your own words, like black holes, wine or Roman history. Whenever you want, ask it to quiz you on your recent facts. It asks a question, you answer from memory, and it grades your answer. It also keeps a quiz score you can check along with how many facts you have learned per topic.
That is retrieval practice in a few seconds, on material you already met earlier in the week, which is exactly the delayed test that showed the biggest gains. You can also ask follow-up questions on any fact or ask for another fact right away. It is an AI, not a person or a teacher, and it can get things wrong.
Try it by text
Text Hi to (424) 999-9536 and tell it what you want to learn. You get a fact right away and one a day after that, at 10am or whenever you choose. A few days in, ask it to quiz you. It works on any phone that can text, with no app and no account needed. The first 20 messages are free, then $9.99/month covers every Textapp. Text STOP to end messages any time. Message and data rates may apply. See textapp.co/apps/study.
Comments