Exam prep

Negative marking in multiple choice: when to guess and when to leave it blank

Negative marking turns every multiple-choice question into a small bet. Once you know the odds, deciding when to answer and when to leave a blank stops being a matter of nerve.

By the Ace team · November 18, 2026 · 7 min read

Negative marking in multiple choice: when to guess and when to leave it blank — Ace blog

Short answer: under negative marking, answer a question when your chance of being right is higher than the break-even point, which depends on the penalty. With four options and a penalty of one third of a mark, a blind guess is worth zero on average, but ruling out even one option makes guessing worth it. With a +4/−1 scheme, even a blind guess is slightly positive on average. The real skill is judging how sure you are, and that can be practised.

What is negative marking?

Negative marking means a wrong answer loses marks instead of just scoring zero. It exists to stop random guessing from inflating scores. On a four-option test with no penalty, a student who knows nothing can expect to get about a quarter of the questions right by luck.

The most common version is sometimes called formula scoring or correction for guessing. The University of Southampton's physics department describes the rule it uses: with N possible answers, a right answer gains one mark and a wrong answer loses 1/(N−1) marks. On a four-option question that is a penalty of one third of a mark. The stated aim is that zero knowledge should score roughly zero, and full knowledge should score 100%.

Other exams use different numbers. Some award four marks for a right answer and take one away for a wrong one. Some take away a full mark. Before you decide anything, read the rules for your own exam, because the right strategy depends on them.

When is guessing worth it? The maths

Every answer is a small bet. The question is whether the bet is worth more than the zero you get for a blank. Here are the numbers for a four-option question.

Scheme 1: +1 for right, −1/3 for wrong.

  • Blind guess (1 in 4 chance): ¼ × 1 − ¾ × ⅓ = 0.25 − 0.25 = 0. On average, it neither helps nor hurts.
  • One option ruled out (1 in 3): ⅓ × 1 − ⅔ × ⅓ = 0.33 − 0.22 = +0.11 marks.
  • Two options ruled out (1 in 2): ½ × 1 − ½ × ⅓ = 0.5 − 0.17 = +0.33 marks.
  • Break-even point: you should answer whenever you think you have better than a 25% chance of being right.

Scheme 2: +4 for right, −1 for wrong.

  • Blind guess: ¼ × 4 − ¾ × 1 = 1 − 0.75 = +0.25 marks.
  • One option ruled out: ⅓ × 4 − ⅔ × 1 = 1.33 − 0.67 = +0.67 marks.
  • Two options ruled out: ½ × 4 − ½ × 1 = 2 − 0.5 = +1.5 marks.
  • Break-even point: a 20% chance of being right.

Scheme 3: +1 for right, −1 for wrong. This is much harsher. A blind guess is worth ¼ − ¾ = −0.5 marks, and you need better than a 50% chance before answering pays. Here, leaving a blank is often the sensible choice.

Two cautions. First, these are averages. Ten blind guesses under Scheme 2 should gain you about 2.5 marks overall, but on a bad day they could lose several. If a pass mark matters more than a high score, you might accept a little less on average for a little less risk. Second, the maths only works if your sense of "I can rule this one out" is accurate. That is where most students go wrong.

What research says about how students guess

Students do not treat these bets in a purely mathematical way. Akyol, Key and Krishna (2022) modelled answering and skipping on the Turkish University Entrance Exam, which penalises wrong answers. They found that women and high-ability candidates were more risk averse, so more likely to skip. The effect of those differences on scores was small, and the authors concluded that a higher guessing penalty made the exam more precise with little effect on gender bias.

The bigger problem may be judging what you know. Higham and Arnold (2007) gave students three formula-scored exams and estimated, for each student, how many questions they should have answered to get their best score. Higher-scoring students were better at monitoring their own knowledge, but students in general did not answer the number of questions that would have given them their best score, and that pattern held even after feedback from earlier tests. Knowing the maths is not enough; you also have to know which answers you really know.

So the strategy is not just "guess when the maths says so". It is "get better at knowing when you actually know", because that is what the maths depends on.

How to practise for a negatively marked exam

  1. Rate your confidence before you check. When you do practice questions, write a confidence level next to each answer before you see whether it is right: sure, fairly sure, or guessing. Afterwards, count how often each group was correct.
  2. Calibrate your break-even point. If your "guessing" answers turn out right 40% of the time, they are worth answering under Scheme 1. If your "fairly sure" answers are only right half the time, you are overconfident, and under Scheme 3 those answers cost you.
  3. Practise elimination on purpose. For questions you are unsure of, cross out the options you can rule out and note why. The reason matters: "I've never seen this term" is weak; "this mechanism happens in the wrong organelle" is strong.
  4. Score your mocks with the real rules. Mark your practice papers with the penalty, and also work out what you would have scored if you had left your low-confidence answers blank. That comparison tells you where your threshold should be.
  5. Learn from the wrong answers you were sure about. Those are the expensive ones. Our guide to why you get multiple-choice questions wrong covers the usual causes.

You need a steady supply of practice questions on your own course for this to work. Ace (acenow.app) is a study app that turns your own lecture PDFs, slides or YouTube lectures into multiple-choice questions checked against the source. Under each answer it shows the sentence from your file that backs it, so when you were confident and wrong you can see exactly which line you misremembered. It runs on the web and on iPhone, and it is free to try.

A simple plan for the exam itself

  • Check the scheme first. Write the break-even percentage at the top of your paper: 25% for −1/3, 20% for +4/−1, 50% for −1.
  • First pass: answer what you know. Skip anything that needs long thought, and mark it.
  • Second pass: eliminate. For each marked question, cross out what you can. If your chance is now above break-even, answer it.
  • Leave the true blind spots. Under −1/3 a blind guess is worth nothing on average, so a blank costs you nothing. Under a full −1 penalty, leave them.
  • Don't change answers on a whim. Change one only when you find a concrete reason. Our guide on how to guess on multiple choice covers elimination tactics in more depth.

If your exam has no penalty at all, ignore everything above and answer every question. If it uses single-best-answer questions, see our guide to single best answer questions as well.

Frequently asked questions

Should I guess if there is negative marking?

Guess when your chance of being right is above the break-even point. With four options and a one-third penalty, that is 25%, so ruling out even one option makes guessing worthwhile. With a full one-mark penalty, you need better than 50%.

Is a blind guess ever worth it with negative marking?

It depends on the scheme. Under +4/−1 with four options, a blind guess gains 0.25 marks on average. Under a one-third penalty it is worth zero on average, and under a one-mark penalty it loses half a mark on average.

How many questions should I leave blank?

There is no fixed number. Leave blank the questions where your honest chance of being right is below the break-even point. Practise rating your confidence on mock questions to find out how accurate your sense of "fairly sure" really is.

Why do I lose marks even when I eliminate options?

Usually because the options you eliminated were not really wrong. Research on formula-scored tests found that students misjudge how many questions they should answer, so check whether your eliminations hold up when you review practice questions.

Sources

  1. Akyol, P., Key, J., & Krishna, K. (2022). Hit or miss? Test taking behavior in multiple choice exams. Annals of Economics and Statistics, 147, 3–50. IDEAS/RePEc record
  2. Higham, P. A., & Arnold, M. M. (2007). How many questions should I answer? Using bias profiles to estimate optimal bias and maximum score on formula-scored tests. European Journal of Cognitive Psychology, 19(4–5), 718–742. University of St Andrews record
  3. University of Southampton, Physics and Astronomy. Compensatory marking. phys.soton.ac.uk

Turn your notes into practice questions

Drop a PDF, slides or a YouTube lecture into Ace and get practice questions checked against your own material — then play them solo or against a friend.

Try Ace free