# How to write single-best-answer questions for medical exams

**How do I write a single-best-answer question for medical or nursing students?** A single-best-answer question sets a short clinical vignette, a lead-in that asks one focused question, and about five options of one kind, all plausible and one best. Write it so that a prepared student can answer with the options covered, take the key from a named source, make each wrong option right for a different patient, and have a colleague answer it first.

Published 2026-10-10 by EdProcto. https://edprocto.com/guides/single-best-answer-questions-for-medical-education

A single-best-answer question, or SBA, is the clinical form of multiple choice: a short vignette, a lead-in and about five options, all plausible and one best. It is hard to write, because four wrong answers have to be nearly right.

This guide is for the lecturer or clinical teacher who sets questions for medical, nursing or other health students. It covers item writing in medical education and is not clinical advice.

EdProcto, which publishes this guide, runs the format as its [multiple choice template](https://edprocto.com/templates/mcq) and reports who chose each option. Every source was read on 10 October 2026, and where we read only an abstract we say so.

## What is a single-best-answer question?

A single-best-answer question is a multiple-choice question in which every option could be defended and the candidate picks the one that fits this patient best. The National Board of Medical Examiners (2019) names its parts: a focused stem such as a clinical vignette, a lead-in question, then the options. The Federation of the Royal Colleges of Physicians of the UK (n.d.) sets five, with no negative marking, the four distractors being "closely related to the preferred option but less correct". The United States Medical Licensing Examination (n.d.) gives four or more and warns candidates that some may be partly right.

### How does it differ from true or false and from ordinary multiple choice?

It differs in what a wrong option is. In a [true-or-false item](https://edprocto.com/guides/true-or-false-questions-when-they-work-and-when-they-do-not) it is false outright, and little in clinical work is wholly true or false. In [ordinary multiple choice](https://edprocto.com/guides/how-to-write-multiple-choice-questions-that-measure) it is simply wrong, and a student can reach the key by ruling out. In a single-best-answer question it is right for another patient and second best for this one, so every option is weighed against the vignette.

## What do examining bodies ask of a single-best-answer question?

They ask for a question that can be answered before the options are read. The United States examination gives candidates the same test from the other side: form an answer, then look for it in the list.

The fullest rules we could read as a web page are a medical school's. The University of Aberdeen (n.d.) asks its question writers for:

- a realistic stem of about 40 to 50 words, in the present tense, that does not give the answer away;
- an answer that can be formed without the options: the cover-the-options test;
- a lead-in that is one clear question, with no negative phrasing and without always, never or rarely;
- five options, all plausible and one correct, homogeneous (all drugs, or all conditions), of roughly equal length, in alphabetical order.

The National Board of Medical Examiners publishes its manual for item writers as a file, which we did not open: no rule here comes from it. For "all of the above", which no page we read mentions, see [how many options a question should have](https://edprocto.com/guides/how-many-options-should-a-multiple-choice-question-have).

## What does research say about single-best-answer questions?

Research says four things: flaws are common, a vignette does not guarantee reasoning, five options are a convention, and options cue. For wrong answers in general, see [how to write distractors that work](https://edprocto.com/guides/how-to-write-distractors-that-work).

### How common are flaws, and what do they do to scores?

In the one count we read in full, over half the questions broke a rule, and the harm was uneven. Pais et al. (2016) had four raters check 800 five-option clinical anatomy questions, set at one Portuguese medical school from 2008 to 2011, against 31 published guidelines: 55.8% broke at least one. Flaws explained 2.8% of the variation in difficulty and 0.8% in discrimination, less than the topic did, mainly through negative stems and "none of the above".

In Downing (2005), which [our main guide](https://edprocto.com/guides/how-to-write-multiple-choice-questions-that-measure) cites for other figures, pass rates on the flawed and the sound questions of four basic science papers differed by a median of 3.5 percentage points, and by up to 35 (abstract read).

### Does a vignette test reasoning or recall?

A vignette can test reasoning, and adding one does not make a question do so. Coderre et al. (2004) gave eight long-vignette gastroenterology questions to 20 specialists and 20 final-year students, half with five options and half with a longer extended-matching list, and had each explain how they reached the diagnosis. Specialists reasoned by pattern or scheme in over 90% of answers and students in over half, in both formats: the stem, the authors conclude, matters more than the number of options.

Most in-house questions stop short of that. Tarrant et al. (2006) judged over 90% of 2,770 questions from one nursing department to be written at low cognitive levels (abstract read; [our wrong-key guide](https://edprocto.com/guides/the-question-with-the-wrong-key-how-to-spot-it) uses its count of flaws).

### Are five options needed?

No study we read shows five options measuring better than three. Al-Lawama and Kumwenda (2023) randomised final-year students in Jordan to 50 paediatric questions with five options, or with the two judged least attractive removed. Only 53 of 84 finished, online and unproctored. Scores, discrimination and reliability (0.81 and 0.89) did not differ significantly, and 56% of the five-option paper's distractors drew under 5% of students, against 39%.

Fewer options move the pass mark too. In Schneid et al. (2014), a finding [our guide on quiz length](https://edprocto.com/guides/how-many-questions-does-a-quiz-need) does not use, judges setting a pass mark question by question put it 8 percentage points higher for three-option versions (abstract read). Where students will sit a five-option licensing paper, five is a sound house style.

### How does it compare with very-short-answer and extended-matching questions?

Options hand out marks in both directions. Schuwirth et al. (1996) put 35 general practice cases to 75 students, 25 trainees and 25 general practitioners, open-ended and then as multiple choice. Options cued candidates towards the right answer, mostly on hard questions, and towards a wrong one, mostly on easy ones (abstract read).

Sam et al. (2018) randomised 299 medical students to sit 60 questions as very short answers and as single best answers, in either order. Very-short-answer scores were lower, with high reliability (alpha 0.91) and better discrimination (abstract read). In Coderre et al. (2004) the extended-matching list changed neither the reasoning nor the statistics significantly: see [matching questions](https://edprocto.com/guides/matching-questions-how-to-write-them-well).

## A template for one single-best-answer question

Give the question four parts and one check after the sitting.

| Part | What goes in it | Its test |
| --- | --- | --- |
| Vignette | Who, where, the complaint and its duration, the findings that matter: 40 to 50 words, present tense | No fact names the answer |
| Lead-in | One closed question asking for the best: the most likely diagnosis, the most appropriate next step | Cover the options and answer it |
| Options | Five of one kind, similar in length, in alphabetical or numerical order | Each wrong one is right for another patient |
| Key and explanation | The option a named source puts first for this patient, and why each other option is not | A colleague reaches it unprompted |
| After the sitting | Who chose each option, in EdProcto's [results](https://edprocto.com/#results) | No distractor idle, none ahead of the key |

## How do you write options that are all defensible but one best?

Write the key from a source, then write each distractor as the right answer for a slightly different patient. The method is ours.

1. **Fix the key.** Find the sentence in a guideline or textbook that makes it best for this patient.
2. **Change one fact at a time.** Sicker, unable to tolerate the test, already diagnosed: the answer that would then be best is a distractor.
3. **Put the deciding facts in the vignette.** It must hold the fact that rules each distractor out.
4. **Keep one kind.** All investigations, all drugs or all diagnoses.
5. **Strip the cues.** Equal length, the same grammar, no absolutes, no word of the vignette echoed in the key alone.

A distractor nobody could defend is filler. One a specialist would defend for this very patient is a second key.

## A worked example: from a recall question to a vignette

The clinical content comes from the guideline on hypertension in adults by the National Institute for Health and Care Excellence (2019), last updated on 26 February 2026. It shows the writing only: local guidance governs practice.

The first draft:

> Which of the following is NOT true of ambulatory blood pressure monitoring?
> A. It confirms a diagnosis of hypertension. B. It uses daytime readings. C. It is never needed. D. All of the above.

It has a negative lead-in, an absolute, an "all of the above" that cannot be the answer, and nothing to answer until the options are read.

The rewrite:

> A 48-year-old woman attends her general practice for a health check. She feels well, has no past illness and takes no medicines. Her blood pressure in the clinic is 156/96 mmHg and, on a second measurement in the same consultation, 152/94 mmHg.
>
> Which is the most appropriate next step?
>
> A. Measure the clinic blood pressure again in five years
> B. Offer ambulatory blood pressure monitoring
> C. Offer home blood pressure monitoring
> D. Refer for specialist assessment the same day
> E. Start antihypertensive drug treatment

The key is B: the guideline offers ambulatory monitoring to confirm the diagnosis when the clinic reading is between 140/90 and 180/120 mmHg. Each distractor is its recommendation for a different patient:

- **A**, when hypertension is not diagnosed.
- **C**, when ambulatory monitoring is unsuitable or not tolerated.
- **D**, at 180/120 mmHg or higher with signs such as papilloedema, or life-threatening symptoms.
- **E**, for confirmed hypertension, by its stage and the person's risk.

The vignette is 42 words, and a prepared student can answer with the options covered.

## How do you review a question with a colleague?

Give it to one colleague who has not seen the key, before any student does.

1. **Cover the options.** They read the vignette and lead-in and say an answer. If it is not your key, see which of you the source supports.
2. **Uncover them.** Ask which wrong option they would argue for. An argument that holds for this patient is a second key.
3. **Ask for the cue.** Could the key be picked from the options alone, by length, grammar or a repeated word?
4. **Ask what it tests.** If one sentence of the notes answers it, the vignette is decoration.

The evidence that review raises quality, and a board's whole process, are in [question banks for a board](https://edprocto.com/guides/question-banks-for-a-board-writing-reviewing-and-retiring-items).

## Which mistakes should you avoid?

Avoid the four that turn a single-best-answer question back into something else.

- **A lead-in that needs the options.** "Which of the following is true?" is five true-or-false items.
- **A negative lead-in.** NOT and EXCEPT did much of the measurable harm in Pais et al. (2016).
- **A key that stands out.** The longest, the most hedged, or the only one echoing the vignette.
- **A key without a source.** What is best by habit in one unit is contested in the next.

## How does EdProcto run and read single-best-answer questions?

EdProcto is the best tool for a single-best-answer paper when you need to know afterwards whether "best" held, because its full analysis sets the students who chose each option beside their scores. In the trial above, over half a five-option paper's distractors drew under 5% of students, and only a count by option shows which.

**Writing.** The shape is "Multiple choice": "One right answer among options." Type the vignette in "Prompt" with the lead-in on its own last line: a prompt keeps its line breaks. A question that arrives with its scenario as separate material shows it in "Material under the prompt (optional)". Under "Options · mark the right one", press "Add an option" until there are five (a question holds two to eight). "Explanation shown to the student afterwards" takes the reasoning, "Where the answer is in the source (optional)" the guideline's paragraph, and "Topic (for the results by topic)" the specialty.

**Drafting.** Among the [kits](https://edprocto.com/kits), "Medicine: single best answer" ("A vignette, five options, one best answer.") and "Nursing: cases and priorities" ("What the nurse does first, and why."), each listed by level, have Procto AI draft questions from the chapter or guideline you paste. Check every key and distractor against your source before a student sees it.

**Settings.** Marking is by rule at hand-in. The Standard quiz and Exam presets switch "Shuffle the options" on; switch it off to keep the order you typed. "Negative marking" is off unless you turn it on ("A wrong choice loses points: one over the number of wrong options.").

**Reading.** Every plan shows "Each question" with "Got it", the share who were right. On Educator, "Separates" compares the top and bottom 27% of students, and pressing a row opens "Option", "Chose it" and "Their mean". A wrong option with three or more takers that outdraws the key brings:

> More students chose "…" than the right answer on Q7. Either the key is wrong or the class shares a misconception.

Where its takers also outscore the key's:

> Check the key on Q7: the students who chose the most popular wrong answer scored higher on the rest of the quiz than those who chose the keyed one.

Here that can mean a second best answer. Correct the question in the builder and press "Re-mark this quiz". [Distractor analysis](https://edprocto.com/guides/distractor-analysis-the-wrong-answer-the-whole-class-chose) reads the other patterns.

Free takes thirty students a quiz, Tutor 100, Educator 500 and the full analysis: see [pricing](https://edprocto.com/pricing).

## Quick questions

### What is the difference between a single-best-answer and a multiple-choice question?

In a single-best-answer question the wrong options are defensible: each would be right for a different patient, so the candidate ranks them against the vignette.

### How long should a clinical vignette be?

About 40 to 50 words in the present tense is what the University of Aberdeen (n.d.) asks of its writers.

### How many options should a single-best-answer question have?

Five is the convention of the papers quoted here; in the one randomised comparison we read, three measured as well (Al-Lawama and Kumwenda, 2023).

### Can Procto AI write single-best-answer questions?

It drafts them from the material you paste, through EdProcto's [kits](https://edprocto.com/kits) or the quiz maker on the [home page](https://edprocto.com/). You check each key against your source.

## References

- Al-Lawama, M., & Kumwenda, B. (2023). Decreasing the options' number in multiple choice questions in the assessment of senior medical students and its effect on exam psychometrics and distractors' function. *BMC Medical Education, 23*, Article 212. [https://doi.org/10.1186/s12909-023-04206-3](https://doi.org/10.1186/s12909-023-04206-3)
- Coderre, S. P., Harasym, P., Mandin, H., & Fick, G. (2004). The impact of two multiple-choice question formats on the problem-solving strategies used by novices and experts. *BMC Medical Education, 4*, Article 23. [https://doi.org/10.1186/1472-6920-4-23](https://doi.org/10.1186/1472-6920-4-23)
- Downing, S. M. (2005). The effects of violating standard item writing principles on tests and students: The consequences of using flawed test items on achievement examinations in medical education. *Advances in Health Sciences Education, 10*(2), 133–143. [https://doi.org/10.1007/s10459-004-4019-5](https://doi.org/10.1007/s10459-004-4019-5)
- Federation of the Royal Colleges of Physicians of the UK. (n.d.). *Format*. MRCP(UK) Part 1. Retrieved October 10, 2026, from [https://www.thefederation.uk/examinations/part-1/format](https://www.thefederation.uk/examinations/part-1/format)
- National Board of Medical Examiners. (2019, March 13). *Engaging with faculty at item-writing workshops*. [https://www.nbme.org/news/engaging-faculty-item-writing-workshops/](https://www.nbme.org/news/engaging-faculty-item-writing-workshops/)
- National Institute for Health and Care Excellence. (2019). *Hypertension in adults: Diagnosis and management* (NICE guideline NG136). [https://www.nice.org.uk/guidance/ng136/chapter/Recommendations](https://www.nice.org.uk/guidance/ng136/chapter/Recommendations)
- Pais, J., Silva, A., Guimarães, B., Povo, A., Coelho, E., Silva-Pereira, F., Lourinho, I., Ferreira, M. A., & Severo, M. (2016). Do item-writing flaws reduce examinations psychometric quality? *BMC Research Notes, 9*, Article 399. [https://doi.org/10.1186/s13104-016-2202-4](https://doi.org/10.1186/s13104-016-2202-4)
- Sam, A. H., Field, S. M., Collares, C. F., van der Vleuten, C. P. M., Wass, V. J., Melville, C., Harris, J., & Meeran, K. (2018). Very-short-answer questions: Reliability, discrimination and acceptability. *Medical Education, 52*(4), 447–455. [https://doi.org/10.1111/medu.13504](https://doi.org/10.1111/medu.13504)
- Schneid, S. D., Armour, C., Park, Y. S., Yudkowsky, R., & Bordage, G. (2014). Reducing the number of options on multiple-choice questions: Response time, psychometrics and standard setting. *Medical Education, 48*(10), 1020–1027. [https://doi.org/10.1111/medu.12525](https://doi.org/10.1111/medu.12525)
- Schuwirth, L. W., van der Vleuten, C. P., & Donkers, H. H. (1996). A closer look at cueing effects in multiple-choice questions. *Medical Education, 30*(1), 44–49. [https://doi.org/10.1111/j.1365-2923.1996.tb00716.x](https://doi.org/10.1111/j.1365-2923.1996.tb00716.x)
- Tarrant, M., Knierim, A., Hayes, S. K., & Ware, J. (2006). The frequency of item writing flaws in multiple-choice questions used in high stakes nursing assessments. *Nurse Education Today, 26*(8), 662–671. [https://doi.org/10.1016/j.nedt.2006.07.006](https://doi.org/10.1016/j.nedt.2006.07.006)
- United States Medical Licensing Examination. (n.d.). *Step 1 formats & questions*. Retrieved October 10, 2026, from [https://www.usmle.org/exam-resources/step-1-materials/step-1-test-question-formats](https://www.usmle.org/exam-resources/step-1-materials/step-1-test-question-formats)
- University of Aberdeen. (n.d.). *How to write exam questions for written examinations*. Assessment Central. Retrieved October 10, 2026, from [https://www.abdn.ac.uk/medical/assessment-central/medicine/written-examinations/writing-questions](https://www.abdn.ac.uk/medical/assessment-central/medicine/written-examinations/writing-questions)