# How to write multiple choice questions that diagnose

> How to write multiple choice questions whose wrong answers show what students think: distractors from real mistakes, examples and live results.

URL: https://learnworkly.com/blog/how-to-write-multiple-choice-questions · Published 2026-11-16 · Updated 2026-11-17 · Tags: Formative assessment, Online tests

A multiple choice question can do two jobs. It can tell you who got the answer right, or it can tell you what the others were thinking when they got it wrong. The second job is far more useful in a lesson, and it depends almost entirely on the wrong options. This guide shows how to write multiple choice questions where every wrong answer points to a specific mistake, and how to read the class’s answers while there is still time to act on them.

## Start from the mistake, not the answer

The biggest change in how to write multiple choice questions is the order you write them in. Most weak questions are written the wrong way round. The teacher writes the stem and the correct answer, then adds three wrong answers to fill the gaps. As a result, the fillers are silly or random, and a student who picks one tells you nothing.

Turn it around. Before you write a word, list the two or three mistakes you expect from this group on this idea. You already know them: they turn up in homework, in questions after class, in last year’s test. Each one becomes an option.

James Beadle of The Assessment Network at Cambridge suggests five questions to ask before writing an item, in [a March 2026 post on the Cambridge Assessment blog](https://www.cambridgeassessment.org.uk/blogs/writing-good-multiple-choice-questions/). Two of them carry most of the weight: “What misconceptions might learners have?” and “Could a learner with a misconception still get the key?” If a student can hold the misconception and still pick the right answer, the question cannot reveal that misconception.

## Distractors in multiple choice questions: examples that work

A distractor is a wrong option. A good one is wrong for a reason a real student finds convincing. Here are examples of distractors built that way, across a few subjects.

### Decimals

*Which number is the largest?*

- A) 0.3
- B) 0.25
- C) 0.125
- D) 0.09

The answer is A. B and C both come from reading the digits after the point as whole numbers: 125 looks bigger than 25, and 25 looks bigger than 3. A student who picks C holds that habit firmly. D is the odd one out, so a cluster on D usually means guessing, and that is worth knowing too.

### Algebra

*Which of these is a linear equation?*

- A) y = 2x + 1
- B) y = x² − 4
- C) y = 3/x

Students who think any equation with x and y in it is linear will pick B or C. Students who pick A quickly probably know that a linear graph needs x to the power of one.

*[Image: The LearnWorkly editor with a Select Answer slide, “Which number is the largest?”, and four options, 0.3, 0.25, 0.125 and 0.09, with 0.3 marked as the correct answer and the slide preview with the join code beside it]*

### Science

*A ball is thrown straight up. At the very top of its path, which forces act on it? (Ignore air resistance.)*

- A) No forces, because it has stopped
- B) An upward force from the throw and gravity
- C) Gravity only, pulling it down
- D) An upward force only

C is correct. A catches students who think “not moving” means “no force”. B, on the other hand, catches the common belief that a thrown object carries the force of the throw with it. Those two groups need quite different explanations.

### English as a second language

*She ___ to the market yesterday.*

- A) go
- B) went
- C) has gone
- D) was go

B is right. A shows a learner who has not noticed the past time marker. C is the classic present perfect error with “yesterday”. D shows someone mixing up forms of the past. If most wrong answers are C, your next ten minutes are about present perfect versus past simple.

### Workplace training

*An email that looks like it is from the IT team asks you to confirm your password through a link. What should you do?*

- A) Click the link, because it comes from IT
- B) Reply with your password so IT can check it
- C) Report the email the way your company asks, without clicking anything
- D) Delete it and tell no one

C is correct in most organisations. A and B show trust in the sender’s name. D is safer, but the security team never hears about the attack, which is a training point in itself.

## Writing good multiple choice questions: the rules that matter

Some rules are about fairness: students should succeed because of what they know, not because they spot clues. Loyola Marymount University’s guide to [selected response items](https://academics.lmu.edu/assessment/resources/assessmentresources/examsforassessment/typesofexamitems/selectedresponse/) sets out the usual ones, and they hold up in practice:

- **One problem per stem.** If a student could answer without reading the options, the stem is clear.
- **Options of similar length.** Test-wise students learn that the long, careful option is often the right one.
- **Every option fits the stem grammatically.** “An ___” followed by an option starting with a consonant gives the game away.
- **Avoid “not” in the stem.** Picking the one wrong statement does not prove you know the right ones. If you must use it, put **NOT** in bold.
- **Go easy on “all of the above” and “none of the above”.** They cut the number of real choices and reward partial knowledge.

Two more come from running questions on screens. First, if you shuffle answer options, “both A and B” stops making sense, because A and B move. Second, keep options short enough to read on a phone without scrolling.

## One correct answer, or several?

Several correct answers make students judge every option on its own, which can be a good check. However, marking is the catch: an all-or-nothing score hides the student who got three of four right. If you want part marks, split it into two questions, or use a format that checks each part, such as Match Pair or Sort It. The [Select Answer help article](/help/quiz/select-answer) covers the settings for a quiz question in LearnWorkly.

## Reading the option spread live

A percentage correct hides most of what a good question tells you. Look at where the wrong answers go.

- **Wrong answers spread evenly across options:** students are probably guessing. The idea has not landed for anyone, so reteach it.
- **Wrong answers piled on one option:** a shared misconception. Address that one mistake directly, using the option as your starting point.
- **Almost everyone right:** move on, and check the stragglers later in the lesson.
- **A distractor nobody picks:** next time, replace it with a mistake someone actually makes.

When you present a multiple choice question in LearnWorkly, the class’s answers build up as a chart on the big screen. On a **Poll** slide you choose a bar or pie chart and can turn on **Hide result**, so nobody simply follows the leading bar until you show it. On a **Select Answer** quiz slide, students answer against a countdown, and the reveal shows how many picked each option with the correct one ticked, on the screen and on phones.

This is the same idea as a hinge question: one question at a turning point in the lesson, where each wrong answer tells you what to do next. Our list of [check for understanding strategies](/blog/check-for-understanding-strategies) shows where hinge questions fit among other quick checks.

## As a poll or as a scored question

LearnWorkly has two slide types for this. A **Select Answer** slide is a scored quiz question: you mark the correct option, and fast right answers earn points for the leaderboard. A **Poll** slide is a vote with nothing scored, although you can still switch on **This slide has correct answer** to show the right option afterwards, and **Allow picking more than one option** for several answers. The setup is in the [Poll help article](/help/interactive-slides/poll).

Polls are not only for opinions. They work well for questions you do not want to grade yet:

- **Confidence checks.** “How sure are you about simplifying fractions? Very / Fairly / Not yet.” Ask it before and after the practice.
- **Prediction.** “What will happen to the balloon when we put it in the freezer?” Committing to an answer first makes students watch the demonstration closely.
- **Training openers.** “Which of these causes the most customer complaints in your team?” The spread tells the trainer where to spend the session.

Use a scored question when you need the result on record, for example an exit ticket or a test; the presentation report keeps every answer. One more option sits between the two: a **Vote · Discuss · Revote** slide runs the question in three steps. Students vote alone, discuss with someone who answered differently, then vote again for points, and the screen shows how the room changed its mind. It works best on questions with strong distractors, because a good wrong answer gives students something real to argue about.

## A quick check before you use the question

Read each item again and ask:

1. Does each wrong option match a mistake I have seen?
2. Could someone with that mistake still get the right answer?
3. Would a test-wise student spot the answer from length or grammar?
4. Do I know what I will do if most students pick each option?

If you cannot answer the last one, the question is not yet diagnostic. So write down the next step for each option, and the question becomes a plan for the rest of the lesson.