The multiple-choice question writing guide (with before/after examples)
A well-built multiple-choice item has three parts: a stem that poses one complete problem, a single unambiguously correct answer, and 2–3 distractors that each represent a plausible mistake. Most flawed items break a small set of known rules — unfocused stems, giveaway cues, implausible distractors, and "all of the above" — and item-writing research has documented fixes for every one of them.
The anatomy of a multiple-choice item
Item writers use three terms worth knowing. The stem is the question or incomplete statement. The key is the correct answer. Distractors are the wrong options — and they're where the craft lives, because a multiple-choice question is only as hard as its distractors are believable. Together the rules below are a condensed version of guidelines used in classroom assessment and professional exam development.
Writing the stem
State a complete problem
Bad stem: "World War I…" — the taker can't begin thinking until they've read all four options, which quietly turns one question into four true/false items.
Good stem: "What event is generally cited as the immediate trigger of World War I?" The cover-the-options test: if a knowledgeable taker can answer with the options hidden, the stem is complete.
Cut the filler
Bad: "In the field of biology, which is a fascinating area of study, scientists who study cells have discovered that the organelle responsible for producing energy is which of the following?"
Good: "Which organelle produces the cell's energy?" Window dressing slows readers and punishes weaker readers more than weaker biologists — which means you're partly grading reading speed.
Use negatives sparingly and loudly
Bad: "Which of the following is not an example of a vertebrate that is not cold-blooded?" Good: "Which of these animals is NOT a vertebrate?" — one negative maximum, always capitalized or bolded. Better still, flip it positive: most NOT-questions can be rewritten as "Which of these IS…" with the option set inverted.
Writing the key
The correct answer must be correct beyond argument — "best answer" items are fine, but only when one option is clearly more right than the rest to any expert. Two traps to check: a second option that's technically defensible (the top source of grade disputes), and a key that's noticeably longer or more carefully qualified than the distractors. Item writers' habit: write the key, then deliberately pad or trim distractors until all options match in length, detail, and grammar.
Writing distractors — the real craft
Source distractors from real errors
Bad: "What is 15% of 200? (a) 30 (b) 7 (c) 1,000 (d) 42." Options b–d are random numbers no one would compute.
Good: "(a) 30 (b) 3 (c) 13.3 (d) 300." Each wrong option is now the result of a specific predictable error (decimal slip, dividing instead, magnitude error). When a taker picks one, the analytics tell you which mistake they made.
Keep distractors homogeneous
All options should come from the same category. If the question asks for a French king and three options are kings while one is a philosopher, the philosopher is dead on arrival. Same category, same era, same level of specificity.
Retire "all of the above" and "none of the above"
"All of the above": recognizing any two correct options gives away the answer; recognizing one wrong option eliminates it. "None of the above": when it's the key, you learn nothing about whether the taker knows the actual answer. If several options are genuinely correct, that content belongs in a multiple-select ("choose all that apply") question, which most quiz tools — Formkii included — support alongside single-answer multiple choice.
Before/after: one item, four fixes applied
| Version | Item | Flaw |
|---|---|---|
| Before | "Plants, which are autotrophs, perform photosynthesis, and this process, which occurs in the chloroplast, produces: (a) oxygen and glucose, which the plant uses for energy and growth (b) salt (c) all of the above (d) an apple" | Bloated stem, giveaway-long key, joke distractor, all-of-the-above |
| After | "What does photosynthesis produce? (a) Oxygen and glucose (b) Carbon dioxide and water (c) Nitrogen and starch (d) Oxygen and protein" | Clean stem; the key's same length as distractors; (b) is the classic inputs/outputs reversal; (c) and (d) recombine plausible terms |
Common cue leaks to proofread for
- Grammar agreement: "an ___" eliminating consonant-starting options; singular stems with one plural option.
- Keyword echo: a word from the stem appearing only in the key.
- Length cue: the key as the longest, most qualified option.
- Position habit: keys clustering at B and C — shuffle answer positions, or use a tool that randomizes for you.
- Absolutes: distractors using "always/never" while the key says "usually" — test-wise takers know absolutes are usually wrong.
Write three options, not five. Decades of item-writing research find three well-crafted options measure about as well as four or five, because most writers' fourth and fifth distractors are filler that no one picks. Two strong distractors beat four weak ones — and the quiz gets faster to take.
The 10-point item review checklist
- 1The stem poses one complete problem (passes the cover-the-options test)
- 2No filler phrases or irrelevant context in the stem
- 3At most one negative, capitalized, and only when unavoidable
- 4Exactly one defensible key — verified by a second reader
- 5All options parallel in length, grammar, and detail
- 6Every distractor is a plausible, category-consistent error
- 7No "all of the above" or "none of the above"
- 8No grammar, keyword, or absolutes cues
- 9Answer positions vary across the quiz
- 10Reading level matches the takers, not the writer
Frequently asked questions
What are the three parts of a multiple-choice question?
The stem (the question or problem), the key (the correct answer), and the distractors (the plausible wrong options). Item quality usually comes down to distractor quality — the question is only as hard as its wrong answers are believable.
How many options should a multiple-choice item have?
Three or four. Research finds three strong options measure about as well as five, because extra distractors are usually weak filler. Prioritize distractor quality over quantity.
Why is 'all of the above' considered bad practice?
It lets takers answer through deduction rather than knowledge: spotting two correct options confirms it, and spotting one wrong option eliminates it. A select-all-that-apply question tests the same content honestly.
What is the cover-the-options test?
Hide the answer options and check whether a knowledgeable person could still answer the question from the stem alone. If they can't, the stem is incomplete and the item is really several true/false questions in disguise.
How do I find good distractors?
Mine real mistakes: wrong answers from past quizzes, common misconceptions, and predictable procedural errors like reversed steps or decimal slips. Distractors built from actual errors make per-question analytics diagnostic instead of decorative.