Changing an answer is not the mistake it is often made out to be. Studies of multiple-choice exams have found, again and again, that answer changes more often fix a wrong answer than break a right one. The common advice to “always go with your gut” is built on memory, not evidence: the times a change cost you a mark are vivid, and the times it saved one are forgotten. The useful rule is narrower. Change an answer when you can name the detail that changed your mind, and leave it when all you have is a feeling.
What the evidence says about first instinct
Research on answer changing goes back decades and covers many kinds of multiple-choice exam, including medical ones. The broad finding is consistent: when students change answers, the changes are more often from wrong to right than from right to wrong, and students who change answers tend, on average, to gain rather than lose marks from doing so.
The same research also finds that students strongly believe the opposite. Psychologists have suggested a reason. A right answer changed to a wrong one feels like a painful, avoidable error, so it is remembered. A wrong answer changed to a right one feels like the exam working as expected, so it is not. Over time, the memorable cases build a rule that the data do not support.
None of this means more changing is better. It means the decision should rest on a reason, not on a superstition in either direction.
Good reasons to change an answer
You find a detail you missed. A lab value, a time course, an age or a medication in the stem that you skimmed on first reading. If it points away from your answer, take it seriously. This is the most common kind of helpful change.
You realise you answered a different question. The lead-in asked for the next step and you chose the diagnosis, or it asked for the mechanism and you chose the drug. Fixing that is not second-guessing; it is correcting a reading error.
A later question jogs a specific memory. Occasionally a fact in one question reminds you of the relevant principle for another. If the recall is specific, use it.
You can now rule out your original choice. If you can say why your first answer is wrong, not just why another looks attractive, the change is well founded.
Poor reasons to change an answer
General unease. A vague sense that an answer was “too easy” is not evidence. Easy questions exist, and the exam is not built around trick questions.
A pattern in your answer letters. Several questions in a row with the same letter tells you nothing about the next one.
The answer you did not pick sounds more sophisticated. Rare diagnoses and unusual mechanisms appear less often than common ones, and the stem usually gives you the reason to choose them when they are right.
Time pressure at the end of a block. Rushed changes in the final minutes, made without rereading the stem, are the ones most likely to go wrong.
How to handle flagged questions in a block
The practical routine is simple. Answer every question on first pass, because a wrong answer costs nothing extra and a blank earns nothing. Flag the ones you are unsure about. If there is time left, return to the flagged questions and reread each stem in full, not just the answer options.
On the second read, ask one question: is there anything here I did not use the first time? If there is, and it points elsewhere, change. If there is not, leave your answer and move on. Do not revisit questions you were confident about simply because you have time.
With the shorter blocks introduced alongside the newer test software, the time left at the end of each block may be modest, so practise a review routine that fits in a few minutes rather than one that assumes a long tail of spare time.
Train it during question bank practice
Your own tendencies matter more than the averages. Some students change too readily; a few almost never change, even when they should. You can find out which you are.
For two or three weeks of timed blocks, note every answer you change and whether the change helped. Many question banks let you see your original choice after a block, or you can keep a short tally on paper. Then look at the pattern.
Most changes helped. Your second reads are catching real errors. Keep reviewing flagged questions and trust changes that come with a reason.
Most changes hurt. Look at why. If they were driven by doubt rather than a new detail, tighten the rule: change only when you can name the detail.
You almost never change. Check whether you are rereading flagged questions at all. A deliberate second read of the stem may recover marks you are leaving behind.
The rule to take into the exam
Answer everything, flag what you are unsure of, and reread flagged stems if time allows. Change an answer when a specific detail tells you to, and leave it when the only thing that has changed is how nervous you feel. That rule is simple enough to hold on exam day, and it matches what the evidence says.
Frequently asked questions
Is your first answer usually right on multiple-choice exams?
Not as reliably as most students believe. Studies of answer changing on multiple-choice tests have generally found that more changes go from wrong to right than the reverse. The changes that hurt are the ones made without a specific reason.
Should I go back and review flagged questions on the USMLE?
Yes, if time allows within the block, because a second read sometimes catches a detail you missed. Answer every question first, so that no question is left blank, then return to the flagged ones. Change an answer only when you can name what in the stem changed your mind.
How do I stop second-guessing myself on practice questions?
Track it. For a few weeks, note each time you change an answer and whether the change helped. The pattern is usually clearer than your memory of the painful cases suggests, and it tells you whether to trust your changes or rein them in.