The Discriminating Test | Marie Curie Series | Which Small Test Separates Competing Explanations?

Quick Read

A learner fails a question.

Several explanations may fit: the concept may be weak; retrieval may have failed; the wording may have hidden the structure; the learner may know the method but not how to select it; the method may be correct but execution may have broken under load.

Repeating the entire topic may help, but it may also treat every possibility as though it were the same problem.

The Discriminating Test asks: what is the smallest useful observation that would separate the most plausible explanations?

The One-Sentence Answer

When several explanations fit the same learning failure, choose one changed condition that should make the explanations behave differently, observe the result, and narrow the diagnosis only as far as the evidence allows.

Where the Discriminating Test Sits in Curie

Evidence Conflict preserves disagreement when learning signals do not line up.

The Discriminating Test is the next move. It does not demand more evidence indiscriminately. It asks which single observation would be most useful for telling the competing explanations apart.

This is one of the ways Curie stays calm: uncertainty becomes a question to narrow, not a reason to label the whole learner.

Start With More Than One Plausible Explanation

If we begin with one explanation only, every later observation can be forced to fit it.

A stronger habit is:

1. state what happened;
2. list two or three plausible explanations;
3. ask what each explanation predicts under a changed condition;
4. choose the smallest test that produces different predictions;
5. observe;
6. update.

This keeps the learner’s evidence correctable by reality.

Examples of Discriminating Tests

Observed problem	Competing explanations	Small discriminating test
Cannot start a Mathematics question	Concept missing vs route selection weak	Name the relevant concept but do not give the method; see whether the learner can now begin
English inference answer is weak	Meaning misunderstood vs evidence bridge weak	Ask for the passage meaning orally before asking for supporting evidence
Science explanation uses wrong mechanism	Concept confusion vs wording load	Present the same mechanism in a simpler diagram and ask for prediction
A-Math solution collapses midway	Method wrong vs algebra execution unstable	Supply the method choice but let the learner execute independently

Why the Smallest Useful Test Is Often Better

A large test changes many things at once. That can produce a dramatic result without telling us which variable mattered.

A discriminating test tries to keep most of the situation stable and change only the condition that separates the competing explanations.

The purpose is not experimental purity. Education is human and messy. The purpose is simply to avoid creating more ambiguity than we remove.

English: Meaning, Evidence or Expression?

An English learner may produce a weak written answer for very different reasons.

• They may not understand the passage.
• They may understand but choose weak evidence.
• They may have the evidence but struggle to express the relationship clearly.
• They may know all three but lose control under time.

Ask the meaning orally first. If meaning is secure, preserve it and test evidence selection. If evidence is secure, test expression. For developmental English stages, continue through the English Tutor route.

Mathematics: Concept, Representation, Route or Execution?

A wrong answer can sit at several layers. Change one layer at a time.

If a representation is supplied and the learner now selects the method, representation may have been the earlier boundary. If the method is supplied and execution remains weak, the repair lies elsewhere.

For developmental Mathematics from Year 0 through University, continue through the Mathematics Tutor route. Detailed teaching remains with the Mathematics estate.

Science: Competing Mechanisms Need Different Predictions

Science gives a literal version of discrimination: if two explanations predict different outcomes, change the condition and see which prediction survives.

For learning, the same logic applies carefully. If a learner’s problem could be vocabulary or mechanism, simplify the language without changing the underlying mechanism. If performance improves sharply, the evidence narrows the problem.

For developmental Science stages, continue through the Science Tutor route.

Additional Mathematics: Preserve the Route, Isolate the Break

In A-Math, one long solution can hide several possible weak links. If route choice is known, supply it and observe algebra. If algebra is known, change the representation and observe recognition. If both are secure, increase load or remove checking support.

The aim is to find the earliest layer that changes the result.

Repair: Narrow Before You Rebuild

1. Describe the observed failure without interpretation.
2. List the two strongest plausible explanations.
3. Ask what each would predict under one changed condition.
4. Run that smallest useful test.
5. Remove explanations the evidence no longer supports.
6. Repair the remaining boundary.
7. Retest under ordinary conditions.
8. Use transfer or robustness only after the repair stabilises.

The Result Should Narrow the Claim

A good discriminating test rarely proves an entire diagnosis.

Its value is more modest and more useful: it removes ambiguity.

We now know the learner can do X when Y is supplied, so the remaining problem sits after X and before Z.

Tutorial Is a Good Place to Run Small Discriminating Tests

A Tutorial allows the human tutor to alter one support, one representation or one condition and immediately observe what changes.

That is one advantage of a small learning room: intervention and observation can remain closely connected without turning the child into a laboratory object.

What Tutor Should Say

The Tutor dashboard can show the narrowed result:

“When the question is represented for you, you choose and execute the Mathematics method independently. The next repair should focus on turning the words into the right structure, not reteaching the method.”

What Parents Can Ask

• What are the two most plausible reasons this went wrong?
• What small change would make those explanations behave differently?
• Can we give one piece of support without giving the whole solution?
• What did the result rule out?
• What is the smallest repair now justified?

Voyage: Better Learners Eventually Run Discriminating Tests on Themselves

Across the Voyage, part of independence is learning to ask, “Do I not know the concept, or do I know it but cannot retrieve it? Do I not understand the problem, or do I understand it but not know how to begin?”

Self-directed learners become better at narrowing their own difficulty before seeking help.

Darwin: Adaptation Improves When the Constraint Is Correctly Located

The Darwin Series asks how a learning method must adapt when the environment changes. A discriminating test prevents us from changing everything when only one part has become insufficient.

Boundaries

• Do not pretend one small test proves a complex learner state.
• Do not manufacture artificial difficulty merely to separate hypotheses.
• Do not keep testing after enough evidence exists to begin a narrow repair.
• Do not use the method to label personality, intelligence or clinical conditions.
• Allow more than one explanation to remain possible when the evidence does not discriminate them.

The Core Rule of the Discriminating Test

When several explanations fit the same learning problem, choose the smallest humane change that should make those explanations behave differently, observe the result, and narrow the repair without claiming more certainty than the evidence earns.

Frequently Asked Questions

Is this the same as giving another test?

No. The point is not more testing. The point is one targeted observation chosen because competing explanations predict different outcomes.

What if the result still fits both explanations?

Keep both. The test did not discriminate them. Choose another small observation only if the distinction matters for the next educational decision.

Why is this useful for parents?

It discourages large conclusions from one result and helps support begin at the actual boundary rather than restarting the whole subject.

---

Marie Curie Series · Evidence Conflict · Evidence Convergence · Observed, Inferred, Unknown · Tutorial · Tutor

eduKate Sengkang | Small groups of up to 3 | 1.5-hour tutorials | Properly Taught Kids Shine a Bright Light Into the Future.