Four options, no idea, pick one, and one time in four it is correct. The score sheet records a mark. Your revision plan records nothing to do. The topic is quietly filed as known.
Across a few thousand practice questions this compounds. If you are guessing on one question in ten, a meaningful slice of your correct answers carry no knowledge at all, and your practice percentage is systematically higher than your real position. The gap shows up on exam day as the unpleasant sensation of a paper that feels harder than anything you practised.
The three kinds of correct
- Knew it. Retrieved the fact, held the reason, could defend it in a viva. This is the only kind that is genuinely finished.
- Reasoned to it. Did not know it outright but eliminated to the answer. Legitimate and worth having, but slower than you can afford at 36 questions in 42 minutes, and it fails when the distractors are good.
- Guessed it. No better than chance. Indistinguishable from the first kind on the score sheet and completely different in reality.
A practice percentage that counts guesses is not measuring your preparation. It is measuring your preparation plus your luck.
Eliminating to an answer is not the same as guessing
It is worth separating these. Reaching the key by ruling out three options is real skill and it will earn you marks in the hall. But it consumes time you do not have in quantity, and it tends to collapse precisely on the well-set questions where every distractor is defensible. Treat it as a fallback you are glad to have, not as knowing the topic.
How to capture it
The mechanism is the same one that catches confident errors: rate your confidence before you see the outcome. Sure, fairly sure, guessing. Three levels are enough; more just slows you down and the extra granularity adds nothing you will act on.
Then treat the guessed-and-correct pile exactly as you would treat errors. That is the discipline people find hardest, because there is no sting to motivate it — the question is already marked right, and it takes deliberate effort to go back to something that did not hurt.
The arithmetic of an inflated percentage
Suppose you genuinely know 60% of a set and guess the rest at random. A quarter of that remaining 40% comes out correct by chance, which is 10 percentage points. Your practice score reads 70%. Nothing has gone wrong with the bank or with you — the number simply includes marks that carry no knowledge, and the gap is invisible unless something records it.
The figure matters less than the direction: a practice percentage always overstates, and it overstates most for the candidates guessing most, which is exactly the group least served by an encouraging number.
Guessing in practice and guessing in the exam are different decisions
In the exam, guessing is correct — the marking scheme rewards it. In practice, an unrecorded guess corrupts your data. Both can be true at once, and the resolution is simply to record that it was a guess. You still answer; you just do not let the result be filed as knowledge.
Guess freely in the hall. Never guess silently in practice.
The pile nobody reviews
Guessed-and-correct is the least revisited category in anyone's history, because there is no sting attached to it. Errors annoy you into action; a lucky mark produces nothing but a slightly better percentage. Treating that pile as though it were an error pile is a small discipline with a disproportionate payoff, and it is almost entirely a matter of deciding to.
In Recallab
The Lucky Guesses app deals you the questions you got right while unsure, and it is live for every account. The scheduler also treats them as weaker than a confident correct answer, so they return sooner rather than being filed away as settled.