Good exam questions are not built to test whether you know the right answer. They are built to test whether you can tell it apart from something that looks almost exactly like it. That is what a distractor is: an option chosen precisely because a candidate with incomplete understanding will find it attractive. Which means the option you nearly picked contains more diagnostic information about your reasoning than the option that was correct, and almost every candidate throws it away by reading the explanation and moving on.
Key takeaways
- Distractors are engineered to be attractive, so the one that tempted you is telling you something specific.
- The correct answer teaches you one fact. The near-miss teaches you a rule about when things differ.
- The question to ask is not why the right answer is right, but when your option would have been right.
- That single sentence converts a lost mark into a transferable principle the next question will test.
- Audit the near-misses even when you got the question right, because a lucky discrimination is fragile.
How a distractor is actually built
Understanding how these options are constructed tells you how to mine them.
A well-written distractor is never absurd. It is typically one of a small number of things: the right answer for a slightly different patient, the right answer at a different stage of the same patient's care, a treatment that is correct in a different health system, an option that would be right if one detail in the stem were absent, or the answer to the question you expected rather than the one asked.
Notice what all of these have in common. Each is correct in some world. The question defines a narrow set of circumstances in which it is not. So the gap between the distractor and the answer is never a fact. It is a condition, a boundary, a discriminator, and boundaries are what the exam is testing and what your revision usually neglects.
The one sentence that matters
When you get a question wrong, or when you hesitated before getting it right, do not begin by reading why the correct answer is correct. Begin with a different sentence, and write it down:
"My option would have been correct if..."
Complete it precisely. "Amoxicillin would have been correct if she were not penicillin allergic." "Chest X-ray would have been correct if he were haemodynamically stable." "That dose would have been correct if his eGFR were normal." "Referral would have been correct if this were not a red flag needing same-day admission."
That sentence is the entire lesson. It contains the discriminator, and the discriminator is what generalises. The correct answer is worth one mark, once, in one question you will never see again. The discriminator is worth marks repeatedly, because the exam will test the same boundary with a different drug, a different organ and a different patient.
The four-step audit
The full protocol takes about ninety seconds per question and is worth more than another twenty questions.
One: name the two. Write down the option you chose and the option that was correct. If they were not adjacent in your mind, if you were genuinely guessing across five options, then this was a knowledge gap and not a discrimination failure, and you should treat it as such and move on to content revision. The distractor audit is for near-misses.
Two: state why yours was attractive. In one line. "It is the standard first-line treatment." "It is what we do in my department." "It is what I would do in the country I trained in." This exposes the source of the error, and the sources are limited in number: outdated knowledge, local habit, imported practice, pattern-matching, or misreading.
Three: find the single discriminator. What one thing in the stem made your option wrong? There is almost always exactly one, and it is almost always a specific clause you either skipped or read without weighting. Find it and underline it mentally, because that clause is the question.
Four: state when your option would be correct. The sentence above. This is the step that converts the error into a rule.
Audit your near-misses even when you were right
Here is the refinement that most people miss. Do this for questions you got right but hesitated on.
A question where you agonised between two options and happened to select the correct one is not a success. It is a coin flip that landed well, and it will not land well every time. On a dashboard it is indistinguishable from a question you knew cold, which is one of several reasons a dashboard misleads.
So flag hesitation as you go, not just error. Mark any question where two options felt live, whatever you eventually chose, and audit them all. That set of items, the ones where your discrimination was uncertain, is the single most productive review list you can build.
Generate the contrasting case
The final step is what proves the rule is real and transferable rather than remembered.
Having written "my option would have been correct if the patient were not pregnant", go and construct that patient. Ask yourself the same question with the discriminator flipped, and answer it. If the patient were not pregnant, what would you give? What would you monitor? What changes?
If you can answer that fluently, you have learned a boundary and you will recognise it in a different presentation. If you cannot, you have learned that in this specific question, the answer was the other one, which is worth almost nothing.
This is also the fastest way to spot a fragile discrimination. A rule you genuinely hold survives having its conditions varied. A memorised answer does not.
Why this beats reading explanations
The reason this method works is that it inverts the passivity of ordinary review.
Reading an explanation is receptive. It produces a feeling of understanding, which is uncorrelated with the ability to produce the answer next time, and it is why candidates can review hundreds of questions and improve very little. Writing down why your option was attractive, finding the discriminator, and stating the condition under which you would have been right, is generative. It requires you to produce the reasoning rather than recognise it, and produced reasoning is what survives to exam day.
It is slower per question. It is dramatically faster per mark.
Where iatroX fits
The Socratic Tutor is built around this exact inversion: it asks you to reason before it explains, so you are forced to articulate why you chose what you chose, and it names the misconception behind the error rather than simply restating the correct option. That is the distractor audit, run for you, on the question in front of you. The adaptive engine then returns the same discriminator in a different clinical context, which is the test of whether the boundary genuinely transferred or whether you memorised an item. Try it with free sample questions at iatroX. For the errors you make while feeling certain, see confidence calibration.
Frequently asked questions
What is a distractor audit? A review method in which you analyse the wrong option you nearly chose rather than the correct answer. You identify why it was attractive, find the single detail in the stem that made it wrong, and state the conditions under which it would have been right.
Why is the near-miss more useful than the correct answer? Because distractors are engineered to be attractive to candidates with incomplete understanding. The gap between the distractor and the answer is a boundary condition, and boundaries generalise across questions in a way that individual facts do not.
Should I audit questions I got right? Audit the ones where you hesitated between two options. Choosing correctly after agonising is a coin flip that landed well, not secure knowledge, and it is indistinguishable from real understanding on a dashboard.
How long does this take? About ninety seconds per question, which is slower per question and considerably faster per mark than reading explanations, because it produces transferable rules rather than the passive fluency that leaves your score stalled.