GLOBAL KNOWLEDGE ARK

Learning Science · Practising retrieval to support learning

Matching a retrieval question to the learning goal

Write retrieval questions that rehearse the kind of knowledge and use the learner actually needs.

Use the examples to connect this lesson with the earlier topics. No specialist equipment is needed.

Back to the chapter

After reading, you can

  • Translate a learning goal into an observable retrieval task.
  • Identify a mismatch between a prompt and the intended outcome.
01

A prompt selects what is practised

If the goal is to explain why metal feels colder than wood at the same room temperature, asking for the word "conduction" rehearses a label. The goal also requires a relationship involving heat transfer from the hand. A more aligned question asks the learner to explain the comparison while keeping the equal-temperature condition explicit.

A learning goal names what the learner should be able to do with knowledge: recall, distinguish, explain, solve, evaluate or perform. Convert that goal into a task whose response makes the relevant thinking visible. The same content can support several goals, and a single short question rarely checks all of them.

02

Retain the essential conditions

Include information needed to interpret the problem, and avoid supplying the relationship you want retrieved. To test a method choice, do not put the method’s name in the heading. To test a causal explanation, do not ask only for a chronological list. Remove irrelevant reading burdens while retaining the conditions that determine the answer.

Build useful variation. A fact question might be followed by a contrast, an example, a nonexample or an application. The variation should exercise the intended connection. Merely changing names while preserving every cue may allow pattern copying; changing everything at once can introduce new knowledge demands that confuse interpretation of the result.

03

Use answers to improve the question

Write an expected answer and decide which elements are essential. Accept accurate alternative wording. If several answers are defensible because a condition is missing, repair the prompt rather than penalising a learner for noticing the ambiguity. Good question design makes the criterion clearer without revealing the solution.

Experiments on retrieval and transfer have tested inferential questions as well as repeated factual questions. Their results support investigating more than word-for-word repetition, while leaving transfer dependent on knowledge and task design. A flashcard collection consisting only of labels cannot stand in for practice of reasoning, judgment or physical performance simply because it uses retrieval.

Authored design choices, not universal chart mandates. Data collection and the reader’s question still matter.
Data and reader questionA useful representationWhy it fits
Counts across separate categoriesBar chartAlign category amounts
Measurements ordered through time; trend soughtLine graph when linking is meaningfulShow change across time
Exact values soughtTableRead the values directly

Worked example

Questions for choosing a graph

How should a question change when the goal is selecting a graph?

  1. The goal is to choose an appropriate representation for data. A weak question asks, "What is a bar chart?" A better task gives counts in several categories and asks for a representation with a reason. Another gives measurements over time and asks what a line graph would make easier to inspect.
  2. The expected answer explains the relationship between data structure and the reader’s question. It does not demand one universal chart for every dataset. A follow-up changes the goal from observing a trend to comparing exact values; a table may now be appropriate. The retrieval task practises selection under conditions.

Supply a data situation and ask for a representation with a reason. The best choice depends on data structure and what the reader needs to inspect.

Try it yourself

Think through each question before opening the answer.

  1. How would you improve a flashcard asking only "What is a control variable?"

    Show the answer and reasoning

    Add a scenario and ask which variable should be held constant and why.

    The definition can remain one step, but the new task practises using it.

  2. Why hide a method name when assessing method selection?

    Show the answer and reasoning

    The name supplies the choice that the learner should make from the problem.

    Removing it reveals whether the relevant relationship is recognised.

  3. A question has two defensible answers. What should happen?

    Show the answer and reasoning

    Inspect missing conditions and the intended criterion, then revise or accept justified alternatives.

    Ambiguity is a design issue, not automatically a learner error.

What to carry forward

  • Make the target thinking necessary to answer.
  • Practise conditions and relationships as well as labels.

References

Sources supporting the explanations and learning checks in this lesson.

  1. Repeated Testing Produces Superior Transfer of Learning Relative to Repeated Studying (2010)Andrew C. ButlerResearch abstractdoi.orgRead source