How to Rule Out an Option When You Do Not Know the Answer
One elimination takes a four-option guess from 25% to 33%, which clears every break-even published here. The skill is learnable, and most of it is reading the options as claims rather than as answers.
Most advice about guessing stops at the arithmetic, and the arithmetic is the easy half. A blind guess between four options returns 25%. Rule one out and the same guess returns 33%, which clears every break-even this site publishes, including the harshest at 25%. That is a large return for a few seconds of work, and nobody teaches the few seconds. The useful question is what you are actually looking for when you look at four options you cannot answer.
- One elimination is worth more than any shortcut: 25% becomes 33%, and two eliminations make it 50%.
- You are checking whether each option answers the question that was asked, which is a different test from whether it is true.
- The patterns below come from how questions get written under time pressure, so they weaken as the paper gets more careful.
- Treat every one of them as a tiebreaker after reading the options properly, never as a substitute for reading them.
What one elimination is worth on a four-option question
| What you have ruled out | Chance of a correct answer | Commonest scheme here, break-even 20% | Harshest scheme here, break-even 25% |
|---|---|---|---|
| Nothing | 25% | Already worth guessing | Level with the line |
| One option | 33% | Clearly worth guessing | Clearly worth guessing |
| Two options | 50% | Worth guessing | Worth guessing |
Read the options as claims, not as answers
The habit that makes elimination possible is small and it changes everything. Instead of asking which option is the answer, take each one on its own and ask what it is claiming and whether that claim is the sort of thing the question asked for. An option can be perfectly true and still be wrong, because it answers a question nobody asked. An option can also be in the wrong category altogether, which is the easiest kind of wrong to spot and the kind most people read straight past while hunting for the right answer.
Five things to check, in the order they are quickest to check
- 1.The category. A question asking for a cause will sometimes offer an effect, and a question asking for a unit of pressure will offer a unit of force. Nothing about the subject is needed to rule that out, only the grammar of the question.
- 2.The magnitude. If three options sit in the same order of magnitude and one is a hundred times larger, the outlier is usually there to catch an arithmetic slip, and you can often tell which slip. The same check works on dates, populations and percentages.
- 3.The absolutes. A question about a tendency rarely has an answer containing "always", "never" or "in every case", because a single exception would make the option indefensible and the person writing it knows that. This one cuts both ways, so use it only where the stem is clearly hedged.
- 4.The pair that cannot both be wrong. Two options that are exact opposites cannot both be false, so if the other two are obviously filler, your guess is already down to a coin. Two options that say the same thing in different words are usually both wrong, because a question with two correct answers would not have survived proofreading.
- 5.The echo. An option repeating a distinctive phrase from the stem is sometimes the answer and more often the trap for a candidate skimming for familiar words. Weak on its own, useful when it points the same way as one of the checks above.
Why these work at all
None of this is about the subject. It is about the order the four options were written in. A correct answer has to survive scrutiny, so it gets qualified and it gets checked. The three wrong options only have to be wrong, and under deadline the quickest way to write a wrong option is to reach for something adjacent: the other unit, the rounder number, the stronger claim. We measured the clearest version of that on our own questions, where picking the bulkiest option without reading the question scores 44.7% against 24.9% for a shuffled control. The full measurement is here.
An option can be perfectly true and still be wrong, because it answers a question nobody asked.
Where this stops working
Being straight about the limits: these are habits of writing rather than rules of exams, and a carefully edited paper has had most of them edited out. The checks are also worth less the better you know the subject, which is the right way round, since a candidate who can answer the question has no use for them. Nobody publishes an analysis of their own distractors, so treat any claim about what your specific paper does, including ours, as a claim about how questions tend to get made and not about the paper in front of you.
The cost of reaching for them too early
Every one of these checks is a tiebreaker, and using one as a first move is how candidates talk themselves out of an answer they knew. The sequence that works is to read the stem, read all four options once, answer if you can, and only then start ruling out. Reaching for the pattern first saves four seconds and spends the knowledge you came in with, which is a bad trade on every paper.
How to practise it
Elimination is trainable, and your own wrong answers are the training set. Take the questions you got wrong in your last mock, cover the explanation, and write one sentence per option saying why the three wrong ones are wrong. Most of the value is in the two you had not considered. Doing this on twenty questions teaches more than attempting another full paper, because it works on the exact gap between recognising an answer and being able to defend one. How to build an error log covers how to keep that from becoming a pile of loose paper.
The arithmetic behind the table is in understanding negative marking, which has the break-even for every scheme published here. Speed versus accuracy in timed tests covers when the seconds this takes are seconds you do not have, and the questions you cannot guess covers the papers where there is nothing to rule out. The papers themselves are in the exam directory.
Put the strategy into practice
Choose a checked mock test and apply what you learned while the idea is fresh.
Frequently asked questions
How do I eliminate options in a multiple choice exam?
Take each option as a claim on its own and ask whether it answers the question that was asked, which is a different test from whether it is true. In order of speed: check the category, so that a question asking for a cause is not answered with an effect; check the magnitude for the outlier that catches an arithmetic slip; distrust absolutes such as "always" and "never" on a question about a tendency; look for a pair of exact opposites, which cannot both be false; and treat an option echoing the stem's wording with suspicion. One elimination takes a four-option guess from 25% to 33%.
Is it worth guessing after eliminating one option?
Yes, on every scheme published on this site. A three-way guess returns 33%, and the harshest penalty here, one-third of a correct answer, breaks even at 25%. On the commonest scheme the break-even is 20% and even a blind four-way guess is already worth taking. The real reason to leave a question blank is the time it would cost, not the deduction.
Does the "avoid absolutes" rule actually work?
Sometimes, and only where the question is clearly asking about a tendency rather than a definition. Plenty of correct answers contain "always" because plenty of statements are always true, so the check is weak on its own. Use it when it agrees with a stronger signal, such as an option that is in the wrong category for the question, and drop it when it is the only thing you have.
How can I get better at ruling out wrong answers?
Work on the questions you have already got wrong. Cover the explanation and write one sentence per option saying why each wrong one is wrong, including the two you never considered. That drill targets the gap between recognising an answer and being able to defend one, which is the gap elimination lives in, and twenty questions of it teaches more than another full mock.
