A paper comes back and one question sinks the whole class. The single mark will not show the reason. Writing effective exam questions means testing what you intend, then reading the item data the next time you set the paper. This post covers the craft and the numbers that prove a question works.
The reader is a lecturer who wrote a paper, marked it and wants the next one to be fairer and tighter. The craft part is about wording. The data part is about listening to how students actually answered. Both belong in the same habit.
What makes an exam question effective?
An effective question tests one clear skill and earns one clear mark. The stem sets the task plainly, and the options do not give the answer away. Item-writing flaws quietly break this. A clue in the phrasing, an option that is plainly implausible or a pattern in the correct position all loosen a weak question's grip.
The research on item-writing flaws shows how common they are. Flawed items make the exam easier for the test-wise and harder for the prepared. The fix is a habit of review. Write the question, set it aside and read it as a stranger would before it reaches the paper.
Why do distractors matter?
Every wrong option should look plausible to a student who does not know the answer. A distractor that nobody picks is wasted space and a hint. If the same option never gets chosen, it is not doing its job. Rewrite or replace it.
A good set of distractors separates the prepared from the guessing. Students who know the material pick the right option with reason. Students who do not should land on the wrong ones by design. That spread is what makes a question informative rather than decorative.
What does item data tell you?
Item data answers two questions about every item. How hard was it, and how well did it separate the class? Difficulty comes from the share of students who answered correctly. Discrimination comes from whether the top students got it right while the bottom did not. A question with both measures in the right range is earning its place.
A question that everyone gets right tells you little beyond the obvious. A question that nobody gets right may be mis-keyed or muddled. A question that the weak students answered better than the strong ones is actively wrong. Each case calls for a different fix, and the numbers point you to it.
The item analysis guide explains the target ranges in plain terms. Watch the discrimination figure most. When it drops low, the question is not separating ability, and you should look at the options before you reuse it.
How does Acuity read your exam?
Acuity is the assessment analytics module in Lectimax. Lectimax is an AI operating system for university lecturers. Acuity shows difficulty and discrimination for every question on your paper. You stop guessing which questions failed and start seeing them in one view.
The view turns a gut feeling into a decision. A flagged question gets a second look, and a clean one earns a place in the bank for the next term. The judgement stays with you, because Acuity reads the numbers and you read the context. For the deeper habit, the page on assessment item analysis walks the full process.
How do you write better next time?
Keep a running record of how each question performed. Write the intended difficulty next to the stem. After marking, compare the intent to the result and note the gap. Over two terms, the same bank gets sharper because every reuse carries its own history.
The comparison of cohort analytics shows why a structured view beats a spreadsheet of raw numbers. A question bank that learns is worth more than a bank that forgets. Each item carries its data, and each new paper draws from a vetted pool.
FAQ
What is the difference between difficulty and discrimination? Difficulty is how many students answered correctly. Discrimination is how well the question separates strong from weak students. Both matter, and each is measured separately.
How many options should a question have? Enough plausible options to make guessing costly, usually four. An option nobody picks is a hint and should be replaced.
Should I drop a question that everyone got wrong? Investigate first. It may be mis-keyed or badly worded. Only drop it after you understand why it failed.
How do I know if my distractors work? Check whether each one is chosen by some students. An option that never gets picked is not functioning and needs rewriting.
How often should I review item data? After every major paper, before the bank is reused. The habit keeps weak questions from resurfacing in the next term.
Where to go next
Read the guide on assessment item analysis for the full method. See how a structured view beats spreadsheet tracking for your school data. Then explore the full Lectimax feature set to find where Acuity fits.