How to Write Better Multiple-Choice Questions (with Examples)
Multiple-choice questions get a bad reputation, but they're powerful when written well. Here's how to write items that actually measure understanding — with classroom-tested examples.
Multiple-choice gets a bad reputation in teacher circles, and a lot of that is deserved — most multiple-choice items are written to be easy to grade, not to actually measure thinking. But a well-written multiple-choice question can target a specific misconception, reveal a student's reasoning, and grade itself in two seconds. Here's how to write items that do real work.
The anatomy of a good item
Every multiple-choice question has three parts: the stem (the question), the key (the correct answer), and the distractors (the wrong answers). The biggest mistake teachers make is treating distractors as filler. They aren't — they're where the assessment happens.
Rule 1: Put the work in the stem, not the answers
If a student can answer the question without reading the choices, the stem is doing its job. If they have to scan the answers to figure out what you're asking, the stem is too vague.
Strong: During photosynthesis, plants convert carbon dioxide and water into glucose and what other product?
Rule 2: Make every distractor plausible
A distractor is plausible when it represents a real misconception a student might have. "Banana" is not a plausible distractor for a chemistry question; "oxygen gas" and "nitrogen gas" are. The best distractors come from your own student work — write down the wrong answers students actually gave you on past tests and turn those into distractors.
Rule 3: Keep the choices parallel
Mixed lengths, mixed grammar, and mixed specificity all leak the answer. If three choices are short and one is long with a qualifier, students will guess the long one. Match length, structure, and tone across all four options.
- Same number of words (within ~30%)
- Same grammatical form (all noun phrases, all sentences, etc.)
- Same level of specificity
Rule 4: Avoid "all of the above" and "none of the above"
"All of the above" rewards students who can spot two correct choices and skip the rest. "None of the above" tests recognition of incorrect answers, not knowledge of the correct one. Both reduce the question to a trick. Skip them.
Rule 5: Test one thing per item
If a question combines two skills (e.g. "convert this fraction to a decimal and then add it to..."), you can't tell which skill the student missed. Split it into two items so the diagnostic is clean.
How AI can help (and how it can hurt)
AI generators are great at producing the first draft of a multiple-choice set — they handle the busywork of writing four parallel options. But they tend to write generic distractors that don't reflect real misconceptions. The fix: generate the draft, then rewrite the distractors yourself using the wrong answers from your own students.
In FoxFish's exam generator you can regenerate any single item or distractor without re-rolling the whole quiz, which makes this hybrid workflow fast.
A quick checklist
- Can a student answer the stem without seeing the choices?
- Is every distractor a real misconception?
- Are the choices parallel in length and grammar?
- Is the item testing exactly one skill?
- Have you avoided "all/none of the above"?
Run every item through that list before printing. The 30 seconds per question pays back the first time you see a class-wide misconception heatmap that actually means something.