How to Write Original Exam-Style Practice Questions That Target the Right Skills

When people search for ways to write exam style questions, they often mean “make questions that feel realistic.” That is a useful aim, but realism is not achieved by copying a past paper, swapping a few names, or making every question harder. A strong practice question mirrors the skills, decisions, format, and level of reasoning learners need to practise while remaining an original piece of teaching material.

For teachers, tutors, parents, course creators, and independent students, the practical goal is simple: create questions that produce useful evidence about what a learner can do next. The question should test the intended knowledge or skill, rather than reward guesswork, familiarity with an unrelated context, or the ability to spot a writing flaw.

A reliable workflow is: blueprint first, define one skill target, draft the question, design answer options or marking criteria, set difficulty intentionally, review, then improve from learner responses. This approach reflects established assessment guidance that connects content specifications, the intended domain, and the cognitive demand of items. The Standards for Educational and Psychological Testing describes content specifications as a detailed account of content, often including classifications of content areas and item types.

1. Start with a question blueprint, not a blank page

A blueprint is a compact plan for the set of questions you need. It prevents a common problem in homemade practice packs: ten questions that all test the same small fact or routine because that was easiest to write.

Your blueprint does not need to be complicated. For each question or question set, record:

For a larger worksheet, add a target number of items in each category. For example, a 12-question maths set might include four foundational items, five standard application items, and three questions requiring multi-step reasoning. That distribution is your teaching decision; it should follow the learner’s needs and the published specification for the relevant assessment where one is available.

Do not label a resource “aligned to” a particular exam solely because it looks similar. Check the current official assessment specification, publicly released sample materials, and permitted question formats before making alignment claims. Rules, blueprints, and content coverage can change.

2. Give every question one clear skill target

Before drafting, complete this sentence: “A successful learner will be able to…” Use a verb that describes observable thinking: identify, calculate, infer, compare, justify, revise, evaluate, or construct.

For instance, “understand persuasive language” is too broad to guide a single question. A sharper target is: “identify which sentence provides evidence for a stated claim.” In science, replace “know photosynthesis” with “predict how reducing light intensity is likely to affect the rate of photosynthesis, using the information given.”

A single target does not mean that all questions must be simple. It means the question has a defensible primary purpose. If a learner misses an item, you should be able to say what to revisit. If the item simultaneously depends on obscure vocabulary, a complicated chart, unfamiliar cultural knowledge, and a new calculation method, an incorrect answer may be difficult to interpret.

Use a skill-and-content formula

A practical item-writing formula is:

Prompt context + task verb + content constraint + response requirement

For example: “A student solves 3(x − 4) = 18 and states that x = 2. Which step contains the error?” This targets error analysis in equation solving. It is more diagnostic than simply asking for a value of x because the response can reveal a likely misconception.

3. Reproduce the demand, not the wording

“Exam-style” should mean that the question has a comparable structure and intellectual demand, not that it reproduces an official item’s language, source passage, figures, answer options, or distinctive scenario.

To create an original parallel question, change the underlying situation and construct the item afresh while retaining the skill target. If an official sample asks learners to infer an author’s purpose from a short article, write a new passage yourself on a different subject and ask a new inference question. If an official item assesses proportional reasoning, create new quantities, a different context, and independently calculated answers.

Copyright protects original expression, while ideas, facts, systems, and methods are not protected in the same way. The U.S. Copyright Office explains that copyright protects original works of authorship, but not facts, ideas, systems, or methods of operation. Its guidance on what copyright protects is a useful starting point. In practice, treat published questions, passages, diagrams, mark schemes, and branded preparation materials as material you should not copy or closely paraphrase without permission or a clearly applicable licence or exception.

Keep a simple source record for any text, image, data table, or stimulus you use. Prefer material you wrote, openly licensed resources used according to their terms, public-domain material where status has been checked, or content for which you have permission. If you are creating commercial resources, distributing work widely, or working across jurisdictions, obtain appropriate copyright advice rather than relying on a general checklist.

4. Build multiple-choice distractors from real misconceptions

In a multiple-choice item, the incorrect options are distractors. Their job is not to trick learners; their job is to make the correct answer meaningful by representing plausible but incorrect thinking.

Start with the correct answer, then ask: “What might a learner who has partly understood this skill do?” Useful distractors can arise from:

• a common procedural error, such as reversing an operation;
• using a relevant fact incorrectly;
• answering only one part of a two-step task;
• confusing a closely related concept or term;
• selecting evidence that is present but does not answer the exact question.

Avoid distractors that are absurd, grammatically mismatched, noticeably longer than the correct answer, or obviously different in precision. Also avoid “all of the above” and “none of the above” when the purpose is to identify a particular piece of knowledge or reasoning; they can shift the task toward test-taking strategy rather than the target skill.

Distractor analysis is routinely used in educational assessment research, and ETS notes that analysis of incorrect-option selection is part of multiple-choice assessment work. ETS research on distractor analysis highlights the role of distractors in understanding item responses. In a classroom, you do not need advanced statistical modelling to learn from options: tally which wrong answers learners choose, then ask what misconception each option may represent.

Example: revise a weak item

Weak: “What is 15% of 80?” Options: 12, 98, triangle, 6.

The unrelated option makes elimination easy, and 98 has no clear relationship to the task.

Improved: “A jacket costs $80 before a 15% discount. What is the amount of the discount?” Options: $12, $15, $68, $92.

Here, $12 is correct; $15 reflects confusing the percentage with a dollar amount; $68 is the sale price rather than the discount; and $92 reflects adding rather than subtracting. The teacher should still check that the context and currency suit the learners.

5. Set difficulty deliberately

Difficulty is not the same as wordiness. A question becomes harder when the learner must make more demanding decisions, coordinate more information, select an appropriate method independently, or explain reasoning under meaningful constraints.

Increase difficulty in controlled ways:

• move from a familiar to a less familiar context while keeping the same skill;
• remove an unnecessary prompt about which method to use;
• add one relevant step, not several unrelated complications;
• ask learners to identify an error, justify a choice, or compare methods;
• use carefully selected data that require interpretation rather than direct lookup.

Do not increase difficulty by using needlessly dense prose, ambiguous wording, inaccessible visual design, or cultural references unrelated to the construct. ETS describes fairness review as a process intended to minimise bias and other characteristics that can hinder test takers from demonstrating relevant knowledge and skills. Its fairness and validity guidance offers a useful principle for classroom writers too: remove barriers that are not part of what you intend to measure.

6. Review before learners see the question

A practical review process has three passes.

1. Content review: Is the answer correct? Is the skill target accurate? Are units, calculations, quotations, data, and explanations checked?
2. Item review: Is there one best answer where the format requires one? Does the stem contain all needed information? Are distractors plausible and parallel? Is the mark scheme clear?
3. Fairness and usability review: Could wording, context, layout, prior knowledge, or an image create an irrelevant barrier? Can learners understand what they are being asked to do?

If possible, ask another subject-knowledgeable educator to review the item. Large assessment organisations use structured writing, editorial, fairness, and review processes rather than treating a first draft as finished; ETS describes its K–12 item-development work as relying on writing guidelines, experienced writers, editorial review, and fairness guidance. See ETS’s overview of item development.

7. Treat learner responses as feedback on the question

After use, sort responses into four practical categories: learners who were correct, learners who chose each distractor, blank or incomplete responses, and comments that show confusion about wording. If nearly everyone misses a supposedly foundational item, investigate before concluding that everyone lacks the skill. The item may have an error, an unclear phrase, or an unintended prerequisite.

Maintain a small item bank with the blueprint fields, version history, answer rationale, and notes from use. Over time, this gives you a more dependable set of original practice questions and reduces repetitive writing.

Use AI as a drafting assistant, not the assessment authority

AI can help generate alternative contexts, question stems, preliminary distractors, and differentiated versions from a teacher-defined blueprint. It can also produce errors, vague wording, unsupported facts, or answer options that do not work. The educational judgement remains human: verify the content, decide whether the question fits your learners, check originality and permissions, and approve the final version.

If you want a faster first draft, try SubSchool’s AI Homework Generator with a clear skill target, desired format, difficulty intention, and constraints. Then apply the same content, item, fairness, and copyright review before sharing any question with learners.

Final checklist: Can you name the skill? Does the question require that skill? Is the wording original? Is the correct answer defensible? Do wrong answers reveal plausible misunderstandings? Is the difficulty intentional? Has a human reviewer checked the final item? If the answer is yes, you have a practice question that is more likely to be useful than merely exam-looking.