How to write multiple-choice questions that reveal understanding

Start with an action: identify a definition, choose a suitable method, apply a rule, or distinguish two explanations. Then decide whether selecting an option can provide the evidence you need. If your goal is to hear someone justify a decision, add a discussion or another task. Choosing the best justification and producing a justification are related activities, but they are not the same assessment.^[2]

For a workplace knowledge check, “know the process” is too broad. A more usable target is “choose the next step when a delivery count differs from the order.” For a maths lesson, try “divide an available quantity into equal groups.” Write the target beside your draft so a reviewer can tell what the question is intended to test.

The stem is the question or problem before the choices. Waterloo’s guidance recommends expressing the full problem there, using familiar language, and avoiding unnecessarily negative or misleading wording. A reader should be able to start thinking about the answer without first assembling the problem from four different options.^[1]

Include context only when it helps solve the problem. An extended story, unfamiliar idiom, or unnecessary character detail adds reading work without necessarily adding useful difficulty. For this example, a lengthy description of the workshop would not change the division learners need to perform.

Solve the question before writing the choices, and record a short explanation of the answer. In the kit example, 24 divided by 3 gives 8 complete kits. Make sure the conditions do not allow a second answer: if a kit needs other limited supplies, or “complete” is undefined, a reader might reasonably interpret the problem differently.

OpenQuiz questions have 2–6 choices and exactly one correct option. Design the item for that format. Avoid choices whose meaning overlaps, and do not rely on an unstated rule to make your preferred option “more correct.” If two options are defensible, revise the stem or the options rather than expecting participants to infer your intention. Waterloo likewise recommends keeping alternatives mutually exclusive.^[1]

A distractor is an incorrect alternative that could attract someone with a specific misunderstanding. Silly choices make the item easier without helping you understand a response. Waterloo recommends plausible distractors and cautions against patterns or wording that reveal the key. Similar grammar, detail, and length help prevent accidental clues.^[1]

These are possible interpretations, not diagnoses of individual learners. Ask how someone reasoned before concluding why they selected an option. If you cannot write a plausible additional distractor, use fewer choices within the supported range instead of padding the question with an obviously wrong answer.

Ask a colleague to solve the draft without seeing your marked answer. Give them the learning target afterward and ask whether the item actually addresses it. Carnegie Mellon recommends colleague review and allowing enough time; an item should not accidentally become a test of navigating convoluted wording under pressure.^[2]

• Can the reviewer explain why the correct answer follows from the information given?
• Can they defend any other option? If so, what needs to change?
• Are units, dates, rounding instructions, and relevant assumptions explicit?
• Does every option finish the stem grammatically without revealing the answer?
• Can the audience read, think, and respond within the planned time?
• Does the item depend on an image, colour, cultural reference, or background knowledge that is irrelevant to the target?

Try the item with the device and presentation setup you will use. OpenQuiz supports a 5–300 second timer for each question. Choose a duration based on the task and participants; that range does not remove the need to plan for access requirements or alternative participation.

Prepare a spoken explanation before hosting, not only a correct-answer marker. After the question, explain the deciding step and why a tempting alternative does not fit. Invite questions about the wording. If participants point out a genuine ambiguity, acknowledge it instead of treating the leaderboard as proof that the item was fair.

• Keep a good item when it addresses the target and prompts useful reasoning.
• Revise an item when a distractor exposes a different interpretation of the wording.
• Replace an item when it mainly rewards a clue, obscure detail, or trick.
• Recheck the answer after any change to a quantity, condition, or option.

A polished question bank grows through this review loop. The goal is not to make every item difficult. It is to make the selected answer meaningful enough to support the conversation or teaching decision that follows.