Build a Question Bank for a Teaching Unit: Tags, Variants and Review

Build a question bank for a teaching unit around learning objectives, checked keys and a record of the quiz copies that use each item. A useful bank records what each item tests, which wrong idea it exposes, where its answer comes from and when it was last checked. That information lets a teacher assemble a quick diagnostic quiz today and a retrieval check three weeks later without starting again or copying an outdated answer key.

This guide builds a compact bank for a four-week science unit on energy transfer. The same structure works for vocabulary, history or staff training. It uses a planning sheet outside Skolina, then turns selected items into quiz masters. Skolina can duplicate a quiz, but it does not offer a separate, automatically synchronised library of individual questions. Keeping that distinction clear prevents a workflow that depends on a feature the product does not have.

Map the unit before writing more questions

Begin with three or four outcomes written as things a learner can do. For the energy unit, these might be: identify an energy store in a simple situation, trace a transfer between stores, explain why not all input becomes useful output, and interpret a basic efficiency calculation. If your current bank has twelve definitions and nothing about transfers, adding another definition is low value even if it is easy to write.

Make a small coverage grid. Put outcomes in rows and teaching moments in columns: before teaching, after the worked example, one week later and at the end of the unit. Mark an item as recall, application or explanation. You do not need a question in every cell. The grid exposes gaps and repetition, and gives each item a job. A question written to diagnose prior knowledge might be deliberately simple; a later question should require transfer to a new context.

The Institute of Education Sciences practice guide recommends using quizzes to revisit important content and spacing learning over time. Its evidence supports retrieval as a learning strategy, but it does not prescribe a magic number of items per unit. Your coverage map is a planning tool, not a psychometric score. For a ready-to-use starting point, see our guide to an online diagnostic quiz for prior knowledge.

Give every item a small record

A spreadsheet can handle the first version. Use one row per question and a stable ID such as ENERGY-TRANSFER-04. Do not let filenames become the only way to identify an item; copies and edits quickly blur those. The record should be short enough that you will actually maintain it.

Add a tag for the required reading or calculation load if that matters in your class. An item that tests energy transfer but requires a dense paragraph may disadvantage students whose reading is still developing. Separate the knowledge target from barriers introduced by the question itself. If your school shares this sheet, apply its access rules; a question bank usually needs no named student results.

Write one best answer and the reason it is best before inventing alternatives. If you cannot explain the key in a sentence, the stem may be too vague. Record why each wrong option is wrong as well. Our multiple-choice writing guide covers the wording and cue checks. Your bank adds the versioning and selection layer that makes those good items reusable.

Create useful variants, not cosmetic duplicates

Suppose the first item shows a torch. A later variant could use a bicycle dynamo or an insulated cup. Keep the learning target—tracing energy transfer—while changing the surface story. A cosmetic rewrite that merely swaps a name can test memory for the original answer, not flexible understanding. A substantially different scenario also lets you check whether a learner can apply the idea elsewhere.

Do not vary every detail at once. If the first question involved no calculation, adding algebra, new vocabulary and an unfamiliar diagram in the variant makes a wrong response hard to interpret. Change the context while holding the intended reasoning demand roughly steady. Record the relationship between variants in your sheet, for example ENERGY-TRANSFER-04A and 04B.

This is also where a colleague’s review is valuable. Ask them to solve the item without your key and describe why a distractor seems plausible. If they find two defensible answers, revise before classroom use. A research study that developed distractors from learner misconceptions illustrates the value of grounding wrong options in real responses. It comes from a specific assessment context; adapt the method cautiously rather than importing its findings as a universal classroom rule.

Turn selected items into short quizzes

For week 1, select four or five items that reveal prerequisites. For week 2, assemble a short exit ticket about the worked examples. For week 3, return to a few core ideas with new contexts. The exit-ticket guide shows how to connect answers to tomorrow’s lesson. Resist publishing the entire bank as one enormous quiz: long sets create fatigue and make a single misconception harder to locate.

In Skolina, create a quiz master for one moment of the unit, check the answer key and feedback, then test the student link or QR code. When another class needs a variant, duplicate the quiz and edit the copy. The product copies the quiz, not your external item record. If you change a question in the copy, update the sheet so colleagues know which wording is current. The plans and active quiz limits are worth checking before maintaining many simultaneous versions.

Be careful with AI-assisted drafts. They can produce plausible sounding distractors or a polished explanation that is still factually wrong. Check every item against your teaching source and mark the draft as unreviewed until that work is done. Do not upload identifiable student answers into a drafting prompt just to generate alternatives; write an anonymised description of the misconception instead.

Correct a bad key across copies without overwriting useful variants

Here is a constructed maintenance case for the efficiency objective. The source handout defines efficiency as useful output divided by total input, multiplied by 100. Item EFF-04 gives 40 J input and 12 J useful output. A draft key of 3% is wrong: 12 ÷ 40 × 100 = 30%. Discovering that mistake in the planning sheet does not update any quiz copy automatically.

Use a copy register alongside the item record. Distinguish a correction to an existing question from a deliberately different numerical variant. Otherwise a bulk manual edit can replace a correct variant with the wrong numbers.

The last row matters: a variant is related to the source item, but its key is not necessarily the same. For the first numerical example, state “enter a percentage as a number, without the % sign” if that matches the response field you actually test. The mathematical key 30 and the interface’s accepted representation must agree; inspect units and feedback rather than assuming a text input accepts every equivalent form.

Close the revision with a short acceptance test

Before reusing the master, solve it without looking at the saved key. Then submit one correct response and one plausible error, such as 70 from calculating the non-useful share. Check that the feedback distinguishes efficiency from the remaining 28 J: 40 − 12 = 28 J, which is 70% of input. Energy has not disappeared; the model classifies only 12 J as the intended useful output.

Record the correction date, the quiz copies checked and whether earlier feedback needs repair. Keep named student responses out of the shared bank itself. If you use advanced per-question statistics to investigate the error, check Starter, Pro or eligible institution access on the plans page. The external copy register remains your responsibility; Skolina duplication is not automatic synchronisation of individual questions.

Maintain the bank after each use

After a class, look at results by item and by option. A distractor nobody selects may add little information. A distractor selected by nearly everyone could reveal a real misconception, but it could also expose an ambiguous stem or a wrong key. Review the question before attributing the pattern to the learners. Short open explanations can help distinguish those cases. Ask: did the question produce evidence that changed my next teaching move?

Give each item one of four decisions: keep, edit, make a variant or retire. Note the reason and date. If the science content, curriculum or source changes, flag affected items for review. If you discover a wrong answer key, correct any live quiz copies and tell the students who received the earlier feedback. A visible revision history lets colleagues check the current key and find copies that need correction.

At the end of the four-week unit, reserve fifteen minutes to choose two items worth asking again next term. That modest habit creates a bank matched to your classes and curriculum. To try the workflow, create a Skolina teacher account, build one five-question master and test a duplicate with a new context. Keep the planning sheet beside it so each copy remains connected to a clear learning goal.