Sure and wrong: why your confident mistakes are the easiest to fix

You get the mock back. The answer you'd have bet anything on has a red cross through it. It's the worst feeling in revision. It might also be the most useful one, because of a finding psychologists call the hypercorrection effect.

What the research found

In 2001, Brady Butterfield and Janet Metcalfe at Columbia University asked people general-knowledge questions and had them rate how confident they were in each answer. Then they showed the right answers and tested them again. The surprise: the wrong answers people had been most confident about were the most likely to be corrected. Low-confidence guesses that were wrong tended to stay wrong.

Metcalfe's 2017 review of research on learning from errors found the effect in children and young adults, on tests straight away and later. Making errors and then getting corrective feedback, the review argues, can help learning rather than hurt it. The likely reason is attention: when you're sure and wrong, the gap between what you expected and what's true grabs you, so the right answer gets noticed and stored.

The catch: you have to check

None of this works without feedback. A confident mistake you never check is just a confident mistake, and it will be there in the exam. The effect is a reason to test yourself and check straight away, not a reason to guess and move on.

How to revise with it

• Answer before you look. No peeking at notes. Commit to an answer, even a guess.
• Mark your confidence. Next to each answer, write S for sure or G for guess. It takes a second.
• Check straight away. Mark every answer as soon as you finish the set.
• Start with sure-and-wrong. Those are the beliefs that were quietly wrong. Write the right answer and one line on why your first answer was wrong.
• Then the guesses. A wrong guess needs more work than a confident error, because it didn't grab your attention. Turn it into a flashcard.
• Retest in a few days. Same questions, mixed up. See which corrections stuck.

A worked example

Say you do ten biology questions on enzymes. You mark seven as S and three as G. When you check, you got two of the S answers wrong: you were sure enzymes are "killed" by heat, and sure that the active site changes shape to fit any substrate.

Those two go to the top of your list. For each, write the correct version in one sentence ("Heat denatures an enzyme: the active site changes shape, so the substrate no longer fits") and one line on what tricked you ("I'd remembered the word 'killed' from Year 8"). That second line matters. It names the old belief, so the next time it pops up you notice it.

Then the three guesses. Whether you got them right or wrong, they're shaky, so each becomes a flashcard. A lucky guess today is a wrong answer on exam day.

What not to do

• Don't skip the confidence rating because it feels pointless. It's what tells you which mistakes to fix first.
• Don't only reread the mark scheme. Reading the right answer once is not the same as being tested on it again.
• Don't avoid hard questions to protect your score. A practice test is the cheapest place to be wrong.

Where this fits in your week

It works best with questions you can mark quickly: past paper questions, flashcards, or a short daily quiz. If you're working through O-Level or GCSE mocks now, go through the marked paper with this lens before you file it away: find every question you were sure about and got wrong. That list is your best revision for the week.

For more on why testing yourself beats rereading, see active recall vs passive studying, and for making the cards, how to make flashcards that work. For a quick daily round, Daily 5 on Versed Play gives you five questions and shows the right answer straight after each one, free and with no sign-up.

Sources

• Butterfield, B., & Metcalfe, J. (2001). Errors committed with high confidence are hypercorrected. Journal of Experimental Psychology: Learning, Memory, and Cognition, 27(6), 1491-1494.
• Metcalfe, J. (2017). Learning from errors. Annual Review of Psychology, 68, 465-489.