HOW MASTERY EVIDENCE WORKS · TARGET → EVIDENCE → STABILITY → TRANSFER → READINESS DECISION · eduKateSG

Proving a Skill Is Ready for the Next Layer

A learner gets 9/10 immediately after a lesson. Is the skill mastered? Maybe. The result could reflect real understanding, fresh memory, repeated question form, heavy cueing or a narrow task that did not test transfer.

Mastery evidence is the collection of performance signals used to decide whether a learning target is stable enough, independent enough and transferable enough to support what comes next.

This page is an adjacent cloud of How Mastery Learning Works. Mastery Learning owns the broad instructional system: define targets, teach, check, correct and allow additional opportunities before dependent learning moves too far ahead. Mastery Evidence owns the narrower decision problem: what counts as enough proof?

The 50-Second Read

• One high score is not automatically mastery. Ask what conditions produced it.
• Accuracy matters, but so do independence and transfer.
• Delay strengthens the claim. Knowledge that returns later is more useful than knowledge available only immediately after teaching.
• Changed examples test structure. If performance survives a new surface, the learner is less dependent on memorised appearance.
• Error recovery matters. A learner who can detect and repair a mistake may be more ready than one who succeeds only when the route stays clean.
• Readiness is thresholded, not magical. Different downstream tasks require different evidence standards.
• The decision must be reversible. If later evidence shows instability, return and repair rather than defending the old mastery label.

1. Mastery Is a Decision About Future Risk

We do not need a philosophical proof that a learner knows something forever. We need enough evidence to decide whether it is safe and efficient to build the next layer.

The more dependent the future learning is on the current skill, the stronger the evidence should be. Weak signed-number control before algebra deserves a higher readiness standard than an isolated enrichment fact.

2. Accuracy Is Necessary but Incomplete

Accuracy is usually the first gate. A learner who is repeatedly wrong is not ready. But accuracy alone can be inflated by cues, recent exposure or repetitive task forms.

Ask: accurate under what conditions? With notes? With a worked example beside the question? With the topic already labelled? With a tutor prompting the first step?

3. Independence Strengthens the Evidence

A skill becomes more trustworthy when support can be removed without performance collapsing. Record the support condition alongside the score: modelled, prompted, lightly cued, independent.

The same 8/10 has different meaning if one learner needed five prompts and another needed none.

4. Delayed Retrieval Tests Durability

Immediate post-lesson performance can exaggerate stability because the explanation is still active in working memory. Retest after delay.

Durable mastery evidence asks whether the learner can reconstruct the method or retrieve the concept after the original learning episode has cooled.

5. Mixed Selection Tests Recognition of Structure

Topic worksheets answer part of the question for the learner by naming the method family in advance. Mixed work removes that cue.

If the student can identify which method applies among plausible alternatives, the evidence is stronger because selection itself has become part of the capability.

6. Changed Context Tests Transfer

A mastered idea should not depend entirely on one wording or diagram. Change the surface while preserving the underlying structure.

• change the numbers;
• change the representation;
• change the story context;
• reverse the question direction;
• combine the idea with another topic.

Successful transfer provides evidence that the learner owns more than a memorised template.

7. Explanation Tests the Model Underneath the Procedure

For some targets, the learner should be able to explain why the method works, what condition matters or why an alternative fails. This is especially useful where procedural imitation can hide conceptual weakness.

Explanation is not required for every automatic micro-skill, but where understanding supports transfer, it strengthens the mastery claim.

8. Error Recovery Tests Robustness

Real performance is not always clean. A learner may make a sign error, choose an unhelpful representation or forget a step. Can they notice the problem and return to a viable route?

Recovery matters because future learning will introduce noise. A capability that survives small disruptions is more load-bearing than one that works only under perfect conditions.

9. Fluency Can Become Part of Readiness

Some skills must become fast enough to free working memory. Slow but correct arithmetic may still overload later algebra. Slow decoding can consume capacity needed for comprehension.

Where downstream tasks require automaticity, mastery evidence should include a reasonable fluency condition, not accuracy alone.

10. Use Multiple Evidence Layers

1. Correct: can produce the right result.
2. Independent: can do so without unnecessary prompting.
3. Delayed: can retrieve after time passes.
4. Mixed: can select the method among alternatives.
5. Transferred: can use the idea when the surface changes.
6. Robust: can detect and recover from ordinary errors.
7. Fluent where needed: can perform fast enough to support the next layer.

Not every target needs all seven layers, but high-dependency targets usually deserve more than one.

11. Mastery Thresholds Should Match Consequence

A threshold is a decision rule, not a claim of perfection. Ninety percent might be appropriate in one context and insufficient in another. A learner can also pass a numerical threshold while still failing an essential conceptual condition.

Use thresholds as evidence gates, not as substitutes for judgement. Assessment Decision Thresholds owns the wider problem of converting continuous performance into categories such as pass, mastered or needs intervention.

12. Mathematics Example

A student appears secure with factorisation. Stronger mastery evidence would include accurate routine questions, independent selection inside mixed algebra, delayed retrieval one week later and use of factorisation inside equation solving.

The learner is ready when the skill functions as a tool inside later mathematics, not merely as a labelled worksheet topic.

13. English Example

A learner understands how to write a topic sentence with a model. Stronger evidence requires using the control independently across different prompts, maintaining relevance under time and recovering when the paragraph drifts.

14. Science Example

A student can recite a process. Stronger evidence asks for explanation, prediction under a changed condition, interpretation of related data and a response that uses the concept without the familiar diagram.

15. Readiness Is Not Permanent

Previously secure knowledge can decay. A mastery label should therefore remain open to revision. If later performance shows instability, return to retrieval, repair and re-verification.

This is not failure of the system. It is the system responding to new evidence.

16. The Mastery-Evidence Audit

1. What exact learning target are we judging?
2. Which future learning depends on it?
3. What accuracy evidence exists?
4. How much support was present?
5. Has retrieval survived delay?
6. Can the learner select the method in mixed work?
7. Can the capability transfer to changed contexts?
8. Can the learner explain or recover where appropriate?
9. Does fluency matter for the next layer?
10. What evidence would make us reverse the readiness decision?

Canonical Owner Boundaries

This page owns the evidence required to justify a mastery/readiness decision for a specific learning target. It does not own mastery learning broadly; that remains with How Mastery Learning Works. It connects to Readiness, Verification, Learning Dependencies and Mastery Decay.

Return: Ready Enough for What Comes Next

The useful question is not “Did the student get 90% once?” It is “What evidence tells us this skill can now carry the next load?”

Mastery evidence turns a feeling of readiness into a decision that can be inspected, tested and revised.

When the evidence is strong enough, move forward. When it is not, repair the floor before asking it to hold another storey.