Knew it at home, gone in the exam — why testing beats reviewing

The problem

The night before the exam, you’re ready. You’ve re-read your notes, everything is clear, everything is there. The next day, in front of the paper, nothing. The words you “knew” yesterday won’t come back, and you walk out telling yourself you worked for nothing.

You didn’t work for nothing. You worked in the way that gives the best result the night before and the worst a week later.

What the research found

Roediger and Karpicke (2006) asked students to learn a text. One group studied it twice. The other studied it once, then tried to write it out from memory without seeing it again. Then they measured what remained, at two moments.

Five minutes later: the “study twice” group is ahead, 81% against 75%. Reviewing works, in the moment.

One week later: the ranking flips. The “study twice” group drops to 42%; the “study then test” group holds at 56%. Same time spent. The only difference is what they did with the second session: re-read, or recall.

That is exactly your night before the exam. Re-reading gives you the five-minute score — and makes you believe you’re ready. Self-testing gives you the one-week score — and makes you doubt the night before. Soderstrom and Bjork (2015) have a name for this trap: we confuse performance during study with the learning that remains afterwards. The two don’t always go together, and sometimes they go in opposite directions.

One reassuring detail: self-testing works even when you fail. Kornell, Hays and Bjork (2009) showed that trying to answer, getting it wrong, then seeing the right answer is remembered better than reading the right answer straight away.

Do this today

Close your notebook. On a blank sheet, write everything you remember from the last lesson: words, sentences, rules. Then compare with the notebook and note what’s missing. What you found unaided, you’ll keep; what was missing, you now know what to revisit.

Do it today, not the night before the exam: by the day before, the doubt needs to have already happened.

How our lessons use it

Every lesson makes you answer before you re-read. “Je m’appelle Nadia” asks you to hide the text, recall Nadia’s name, her city and her Saturday, then check. That moment when you see the gap between what you thought you knew and what you actually recalled is the one that stays.

Do it in Je m’appelle Nadia (A1) today, not the night before the exam.

The honest limit

Roediger and Karpicke’s experiment used native-language texts with university students. Second-language studies point the same way, but the size of the effect varies with level and with what is measured. And self-testing doesn’t replace understanding: if you don’t understand the question, no amount of recall will save you — that’s the job of the corrections, not the test.

Your questions