"When in doubt, pick C" is the best-known rule in test taking and the least examined. Is C really the most common answer on a multiple-choice test? Sometimes, where a person placed the keys by feel. Hardly ever where a test maker balanced them. And where software shuffled the options, the question stops making sense.
This guide sets out what has been counted: where right answers sit, why people put them there, what "always C" is worth in marks, and what to do instead. We read every source on 10 October 2026; "abstract read" marks those where no more was open.
The publisher is EdProcto, whose multiple choice template can give each student the options in an order of their own. That is the short answer to the whole matter.
Is C really the most common answer on a multiple-choice test?
On tests keyed by hand it sometimes is; on tests built by professionals it is not, by design.
A count from teachers' own tests. Shakurnia and Elhampour (2015) took the 214 examinations logged with the automated service centre of one Iranian medical university in 2013 and counted where the right answer sat in 14,359 questions with options A to D. It was C in 4,683 (32.6%), B in 3,893 (27.1%), D in 3,523 and A in 2,260 (15.7%). The two middle letters held 59.7% of the keys, against an even 50% (abstract read). It is a plain count, at one university.
The wider record. Lions et al. (2022) reviewed the studies of where test developers put their options, in teacher-made and standardised tests alike. They conclude that the spread of keys across tests is often uneven enough to hand candidates a clue (abstract read; a companion paper puts the review at more than 50 studies).
Professional keys. Attali and Bar-Hillel (2003), in findings our guide on shuffling leaves out, report that the makers of operational tests balance their keys so thoroughly that the lean to the middle almost disappears from them, though not quite (abstract read). Chile's university admission test shows a professional key from the outside. On the four forms of its 2020 language paper, 80 five-option questions each, every letter was the key between 11 and 22 times, where 16 would be an even share, and each form was uneven in its own way (Lions et al., 2023, read in full). No letter was a safe bet.
On five options the middle is one letter, and it is a trap
With five options, C is the only true middle, and it is where the best wrong answer lives. Lions et al. (2021) sorted the wrong options of 110 Chilean admission tests, each of 80 five-option questions, set over five years, by how many candidates chose them. The strongest wrong answer sat most often at C and the weakest at E, in every subject and year. A second set of 421 questions from 11 studies in other countries showed a similar pattern (read in full). A guesser who heads for C on such a paper walks into the most tempting mistake.
Where does "when in doubt, pick C" come from?
Nobody we read has traced the saying to its first user; what the studies show is a human habit that would produce it. People hide things in the middle, and look for them there. Researchers call it middle bias.
- Writers. Asked to write one question, people put the right answer in a middle place far more often than at an edge. Our guide to shuffling gives the ratio from Attali and Bar-Hillel (2003).
- Test takers. The same paper reports that in a typical four-option test about 55% of wrong answers fall in the two middle places, where an even spread would give 50%.
- Shoppers. Christenfeld (1995) ran six studies of choices among identical things. Where a supermarket stocked four rows of one product, 71% of purchases came from the middle two, and people marking a box on a questionnaire avoided the ends as well (abstract read).
So the rule has a grain of truth on four options, where B and C are both middle. Why C took the fame and B did not, no study we found says.
What does "always C" earn?
Nothing on a balanced key, and less than one mark in ten guesses on the most middle-heavy key we could find. The table is our arithmetic for ten blind guesses on four-option questions.
| The key | C is right | Ten guesses of C earn | Against ten letters picked at random |
|---|---|---|---|
| Balanced by the test maker | 25% | 2.5 marks | No gain |
| The Iranian count, always C | 32.6% | 3.3 marks | 0.8 of a mark more |
| The same key, always A | 15.7% | 1.6 marks | 0.9 of a mark less |
| Options shuffled for each student, as in EdProcto | C names a different option for different students | Not fixed | No key letter to learn |
A student who knows 30 of 40 questions and guesses C on the rest moves from 81% to 83% on the middle-heavy key. That is the whole prize, and it needs a teacher who has never heard of the rule.
Do test-wise students gain from position rules?
A little, where keys are made by hand, and the best-evidenced gain comes from a rule other than C. Researchers file such rules under test-wiseness.
- On a balanced key. Bar-Hillel and Attali (2002) showed that the SAT of the time balanced its keys instead of randomising them, which leaves traces: a candidate who used them could add as much as 16 points to a score (abstract read). The logic, as we read it: when every letter must come up equally often, the letters your sure answers have used least are owed some.
- By simulation. Ortega et al. (2025) modelled the C rule and balanced guessing against keys placed by balancing and by randomising, for tests of several lengths and candidates of several levels. Only randomising came through with no problem of validity or fairness (abstract read; it prints no sizes).
Does moving the key change how hard a question is?
Slightly: the same question tends to be a little easier with its key first and a little harder with its key last.
- Hohensinn and Baghaei (2017) modelled the four positions in a high-stakes test and found very slight differences, a question growing slightly harder as its key moved towards the end (abstract read).
- Holzknecht et al. (2021) tracked the eyes of 30 people answering listening questions from the Aptis test, then analysed 200 live four-option questions answered by about 6,000 candidates each. Where the key sat changed how much attention it got and how hard the question was, so a candidate who by chance drew many keys in last place could lose out (abstract read).
These effects are small beside knowledge. Lions et al. (2023) suggest why their candidates showed no pull to the middle: it appears mainly when people guess, and theirs had trained for months.
What do examining bodies say they do about key position?
Very little in public. Our searches of several boards' sites on 10 October 2026 found no page in which an examining body describes how it places its keys. That is a limit of our search, not proof that none exists. What can be read are research papers.
- The SAT. Bar-Hillel and Attali (2002) call its key balancing "a well-kept trade secret".
- Chile. In the 2020 forms studied by Lions et al. (2023), an experiment run inside the test with the state body that sets it, no letter was the key more than twice running, and a key repeated the letter before it in only 5.1% to 7.6% of cases. Chance would repeat it 20% of the time on five options, by our arithmetic, so those keys look arranged, not drawn by lot.
- The advice. The same authors say to randomise the order by computer, since people cannot produce random sequences; they rate alphabetical or numerical order a good second, and rule out balancing by hand, which invites balanced guessing.
How should a teacher place the key?
Let chance place it, and where the options have a natural order, let that order place it.
- Let software shuffle. In EdProcto, "Shuffle the options" is on in the starting settings, so each student meets the options in an order of their own. Type them as they come to mind.
- Give numbers, dates and scales their natural order. 10, 20, 30, 40 reads fastest in order, and the key then falls where its value falls. Switch the option shuffle off for that quiz, and take each wrong number from a real slip, so that the key is not always the middle value (how to write distractors).
- On paper, draw lots. Roll a die or use a random number for each key's place. Do not trust your own sense of "mixed up".
- Leave the draw alone. A key drawn at random for 20 four-option questions has exactly five of each letter about once in 100 draws, and holds a run of three or more of one letter 62 times in 100, by our arithmetic. Evening it out by hand rebuilds the balanced key a test-wise guesser can use. The price is a little luck: one paper draws more last-place keys than another, the risk Holzknecht et al. (2021) name.
- Check afterwards. EdProcto's results count who chose each option, in the full analysis on the Educator plan.
For a student the advice is one line: drop the letter and work the question, as our guide to multiple-choice strategy sets out.
What follows for true or false keys?
The same rule with two options: keep true and false near half each, in no pattern, because any lean is a rule a guesser can use.
The usual claim is that teachers write more true statements than false ones. We could open no count of teacher-made true or false keys today, so we give no figure for it. What we did read is a reason to expect a lean. Bar-Hillel et al. (2014) found that in a choice between two options the one presented first is favoured: about 80% of people asked to imagine coin tosses began with heads, and the bias reversed when tails was named first. The authors apply this to true or false tests themselves (abstract read).
The arithmetic is quick. On 20 statements of which 12 are true, "always true" scores 60% and "always false" 40%; on a key of ten and ten both score 50%. Our guide to true or false questions has the rest. EdProcto's True or false keeps its two options where they are, so the balance is yours to set, statement by statement.
Which other position myths hold up?
Few, and each depends on a key arranged by a person.
| The rule | What is known |
|---|---|
| "The same letter never comes three times running" | True of the Chilean forms above; false of a random key, where runs are normal. In Paul et al. (2014) students tolerated at most three repeats, and performance fell as answer patterns grew more obvious (abstract read) |
| "Pick the letter you have used least" | Pays only on a key balanced by hand (Bar-Hillel and Attali, 2002) |
| "A is never right" | A was the rarest key in the Iranian count, yet questions keyed first run slightly easier, not harder |
| "The last option is a throwaway" | Among wrong answers, often: the weakest sat most at E in Lions et al. (2021). That says nothing about questions keyed E |
How does EdProcto place a key?
EdProcto is the best tool for this job because the place a student meets the key in is drawn on its server for that student, not left where the writer typed it, and typing habits are what made C right in a third of one university's questions.
Typing. The editor of a multiple choice question is headed "Options · mark the right one". The options are lettered in the order you type them, and the key is whichever one you mark.
The switch. Under "Sequencing", "Shuffle the options" is on in the starting settings. The Practice preset switches it off; Standard quiz and Exam switch it on. With it off, every student sees the order you typed.
Each student's order. With it on, the server draws the order of each question's options for a student's attempt and keeps it. Each student has an order of their own, and a reload shows the same one.
Shapes that hold their options. True or false and Yes, no or maybe keep the option shuffle off, greyed with the reason "too few options to shuffle".
Kits made by rule. A kit that makes its questions by rule shuffles the options of each one as it is made, every order as likely as any other, so the right answer sits in each place as often as in the next.
Results. On the Educator plan, pressing a question's row opens a table headed "Option", "Chose it" and "Their mean". It lists each option in its own words with the key marked, whatever order a student saw.
A quiz takes thirty students on Free and 100 on Tutor; Educator takes 500 and adds the full analysis (pricing).
Quick questions
Is C the most common answer on a multiple-choice test?
Only on some tests keyed by hand. In one university's 14,359 questions C was right in 32.6%; on a balanced key it has no edge.
Should I balance my answer key so each letter is used equally?
No. Draw each key's place at random and leave the result alone: a hand-balanced key can be exploited (Bar-Hillel and Attali, 2002).
Does the position of the right answer change how hard a question is?
Slightly. Questions run a little easier keyed first and a little harder keyed last (Hohensinn and Baghaei, 2017).
How does EdProcto decide where the right answer sits?
With "Shuffle the options" on, EdProcto draws an order for each student on its server and keeps it; with it off, students see the order you typed.
References
- Attali, Y., & Bar-Hillel, M. (2003). Guess where: The position of correct answers in multiple-choice test items as a psychometric variable. Journal of Educational Measurement, 40(2), 109–128. https://doi.org/10.1111/j.1745-3984.2003.tb01099.x
- Bar-Hillel, M., & Attali, Y. (2002). Seek whence: Answer sequences and their consequences in key-balanced multiple-choice tests. The American Statistician, 56(4), 299–303. https://doi.org/10.1198/000313002623
- Bar-Hillel, M., Peer, E., & Acquisti, A. (2014). "Heads or tails?": A reachability bias in binary choice. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40(6), 1656–1663. https://doi.org/10.1037/xlm0000005
- Christenfeld, N. (1995). Choices from identical options. Psychological Science, 6(1), 50–55. https://doi.org/10.1111/j.1467-9280.1995.tb00304.x
- Hohensinn, C., & Baghaei, P. (2017). Does the position of response options in multiple-choice tests matter? Psicológica, 38(1), 93–109. https://eric.ed.gov/?id=EJ1125979
- Holzknecht, F., McCray, G., Eberharter, K., Kremmel, B., Zehentner, M., Spiby, R., & Dunlea, J. (2021). The effect of response order on candidate viewing behaviour and item difficulty in a multiple-choice listening test. Language Testing, 38(1), 41–61. https://doi.org/10.1177/0265532220917316
- Lions, S., Dartnell, P., Toledo, G., Godoy, M. I., Córdova, N., Jiménez, D., & Lemarié, J. (2023). Position of correct option and distractors impacts responses to multiple-choice items: Evidence from a national test. Educational and Psychological Measurement, 83(5), 861–884. https://doi.org/10.1177/00131644221132335
- Lions, S., Monsalve, C., Dartnell, P., Blanco, M. P., Ortega, G., & Lemarié, J. (2022). Does the response options placement provide clues to the correct answers in multiple-choice tests? A systematic review. Applied Measurement in Education, 35(2), 133–152. https://doi.org/10.1080/08957347.2022.2067539
- Lions, S., Monsalve, C., Dartnell, P., Godoy, M. I., Córdova, N., Jiménez, D., Blanco, M. P., Ortega, G., & Lemarié, J. (2021). The position of distractors in multiple-choice test items: The strongest precede the weakest. Frontiers in Education, 6, Article 731763. https://doi.org/10.3389/feduc.2021.731763
- Ortega, G., Lions, S., & Dartnell, P. (2025). Toward evidence-based guidelines on answer keys' placement in multiple-choice tests: A simulation study. Applied Measurement in Education, 38(3–4), 284–303. https://doi.org/10.1080/08957347.2025.2608017
- Paul, S. T., Monda, S., Olausson, S. M., & Reed-Daley, B. (2014). Effects of apophenia on multiple-choice exam performance. SAGE Open, 4(4). https://doi.org/10.1177/2158244014556628
- Shakurnia, A., & Elhampour, H. (2015). Survey on correct response position in multiple choice tests: A descriptive study. Strides in Development of Medical Education, 12(2), 426–432. https://sdme.kmu.ac.ir/article_90391.html