How to Assign Question Difficulty Levels
Set useful easy, medium and hard question tags, then review them against student results without confusing difficulty and cognitive demand.
A question’s difficulty level should describe the expected challenge for a defined group of learners, not the writer’s confidence in the question. Before use, label an item provisionally from the knowledge and steps it demands. After use, compare that judgement with student responses and revise the label if the evidence warrants it. “Easy,” “medium” and “hard” can be useful operational tags, but they are not universal properties of a question.
This matters when building a paper from a bank: if the labels are inconsistent, a requested mix of difficulty can be misleading. It also matters when interpreting results. Our item-analysis guide explains the post-exam evidence; this article focuses on deciding and maintaining the tag.
Difficulty is relative to the learner and context
“Calculate 15% of 80” may be straightforward for a class that has practiced percentage calculations, challenging before the unit and trivial in an advanced statistics course. A long question is not necessarily hard, and a short question is not necessarily easy. Reading load, time pressure, familiarity with the format and prior instruction also change performance.
Keep the target cohort in the question record, for example “Grade 7, after the percentages unit,” rather than writing an unqualified “easy.” If the same item is used across programmes, the difficulty label may need to be reviewed for each use.
A practical pre-use rubric
| Tag | Working meaning for this cohort | Illustrative sign |
|---|---|---|
| Easy | Applies a recently taught skill in a familiar form | One clear step with no irrelevant reading load |
| Medium | Transfers a taught skill to a slightly different context or combines a few steps | Requires choosing a method, then applying it correctly |
| Hard | Requires a less familiar application, deeper reasoning or several dependent decisions | A well-prepared learner must plan and justify a solution |
This is a team convention, not a psychometric scale. Decide the criteria with your teachers and calibrate with examples from your syllabus.
Worked example: one topic, three levels
Suppose the outcome is interpreting percentages in a Grade 7 class.
- Easy: “What is 10% of 50?” Direct application of a practiced fact; answer: 5.
- Medium: “A price of $80 is reduced by 15%. What is the sale price?” Student must find the discount ($12) and subtract it; answer: $68.
- Hard: “After a 20% discount, a bag costs $48. What was its original price? Explain your method.” Student must work backward from 80%; answer: $60.
The examples are original and illustrative. A teacher should not copy the tags without considering what that class has learned. Notice also that cognitive demand and difficulty are not identical: a question can require a higher-order explanation yet be familiar to a well-prepared class. Our Bloom’s taxonomy guide addresses the kind of thinking a prompt asks for.
After the exam: compare the tag with evidence
For a right/wrong item, a simple difficulty index is the proportion of students who answered correctly. If 24 of 30 answer correctly, the observed facility is 24/30 = 0.80. That does not automatically mean the item is objectively easy or good. The cohort may have been unusually well prepared; an answer clue may have inflated performance; or the item may have been intentionally easy. Conversely, low facility can reflect poor wording rather than desirable challenge.
Look at the item, the cohort and other evidence together. Cambridge International describes expert review of question and mark-scheme alignment in its assessment quality process. Classroom teams can use a smaller version of that habit: review surprising results, revise the question or tag, and record why.
A governance rule for a shared bank
Record four fields beside the item: provisional difficulty, intended cohort, latest observed result and last review date. Keep the original judgement separate from the observed statistic. When an item changes substantially, treat the previous response data as evidence about the old version, not the new wording.
In Examiar, difficulty can be used when planning a question mix. Treat the label as a planning aid and still check the complete paper for balance of content, thinking demand and timing. The question paper generation guide shows the broader workflow. After the exam, use the free item analysis calculator to calculate the proportion answering an item correctly and compare upper and lower groups. Those results inform review; they do not automatically determine a question’s quality.
FAQ
Should every exam have one-third easy, medium and hard questions?
No. A fixed split ignores the assessment’s purpose and cohort. A diagnostic quiz, practice paper and selective examination can reasonably use different distributions. Start with learning outcomes and the decision the score will support.
Can item analysis set the label automatically?
It can inform the label, but a single result should not decide it alone. Review sample size, cohort, wording and teaching context before changing a shared-bank tag.