Quick Read
Science is not strong because explanations never change. It is strong because explanations are expected to remain answerable to evidence.
When new observations appear, students should ask whether the current explanation still fits, needs a small adjustment, should lose confidence, or must be replaced by a better account.
- Current explanation: What does it claim is happening?
- Prediction: What should we observe if it is correct?
- New evidence: What has changed in the evidence base?
- Fit: Does the evidence support, weaken or contradict the explanation?
- Revision: Can the explanation be refined without becoming ad hoc?
- Replacement: Does another explanation account for more evidence with fewer unsupported assumptions?
This article explains evidence-driven revision inside our wider Science Tuition Sengkang learning system.
The One-Sentence Answer
Scientific explanations change when new evidence alters how well they predict and account for observations, so confidence and models should be revised in proportion to the strength and relevance of that evidence.
An Explanation Is Not the Same as a Fact
Students often learn conclusions in finished form and therefore assume scientific explanations are fixed statements.
A stronger view separates observation from explanation.
Observations describe what was measured or seen. Explanations propose why those observations occur and what mechanism connects them.
Good Explanations Make Predictions
If an explanation says a certain mechanism is operating, it should imply what we expect to observe under related conditions.
Predictions give future evidence something concrete to test.
See How Scientific Predictions Grow From Patterns, Evidence and Mechanisms.
Supporting Evidence Can Increase Confidence
If repeated observations match the explanation’s predictions, confidence can rise.
This does not make the explanation untouchable. It means the current evidence gives us more reason to rely on it than before.
Contradictory Evidence Should Lower Confidence
If a well-designed test produces a result the explanation strongly said should not happen, the explanation has a problem.
Students should not protect the explanation by automatically dismissing inconvenient evidence.
They should investigate both the evidence and the explanation.
One Anomaly Does Not Always Destroy a Strong Explanation
An unusual result may come from measurement error, hidden variables, natural variation or an overlooked mechanism.
The first response should be diagnosis rather than automatic rejection.
See How Unexpected Results Reveal Hidden Variables in Science.
Repeated Contradiction Matters More
If the same conflict appears across repeated, well-controlled and independently reproduced investigations, it becomes harder to explain away as noise.
Evidence accumulates against the current explanation.
Replication therefore affects not only confidence in measurements, but confidence in explanations. See How Replication and Reproducibility Strengthen Scientific Evidence.
Revision Can Be Small
New evidence does not always require a completely new theory.
The current explanation may still work if a missing condition, threshold, variable or boundary is added.
The key is whether the revision improves predictive accuracy rather than merely protecting the old claim after every failure.
Revision Should Not Become an Excuse
If every contradictory result is answered with a new unsupported exception, the explanation can become unfalsifiable.
A useful revision should generate clearer expectations and remain open to further testing.
Alternative Explanations Should Be Compared Against the Same Evidence
When one explanation weakens, students should not automatically adopt the first alternative.
Each candidate should be tested against the same observations, controls and predictions.
See How Students Compare Competing Scientific Explanations Against Evidence.
A Better Explanation Usually Accounts for More
An improved explanation may explain the original evidence and the newer anomaly within one coherent mechanism.
That is stronger than simply fitting the newest result while losing the earlier successes.
Revision should increase explanatory coverage, not move the problem somewhere else.
Mechanisms Matter When Explanations Change
A revised explanation should say what process produces the observed pattern.
Without a mechanism, students may simply rename the result instead of explaining it.
Mechanism helps the revised explanation make new predictions.
Scientific Models Are Designed to Be Revised
Models are simplified representations.
As evidence improves, a model may gain variables, lose assumptions, change relationships or become limited to a narrower domain.
See How Scientific Models Help Students Explain Things They Cannot See Directly.
Negative Results Can Force Revision
If an explanation predicts a detectable effect and repeated sensitive tests find none, the explanation should lose confidence under those conditions.
The absence may suggest a threshold, limiting factor or incorrect mechanism.
See How Negative Results and Missing Effects Shape Scientific Conclusions.
Converging Evidence Can Stabilise a Revised Explanation
If several independent lines of evidence support the same revised mechanism, confidence grows more strongly than it would from one repeating measurement alone.
This connects with How Multiple Pieces of Evidence Build a Strong Scientific Explanation.
Confidence Should Move Gradually
Science does not need only two states: completely true or completely false.
Students can learn to say that evidence strengthens, weakens, limits or leaves an explanation unresolved.
This calibrated language is more faithful to evidence than dramatic certainty shifts.
The Quality of New Evidence Matters
A vague observation should not outweigh many precise, reproducible measurements automatically.
Students should judge relevance, control, measurement quality, sample representativeness and independence before deciding how much the explanation should move.
Direct and Indirect Evidence Can Both Drive Revision
A direct measurement can contradict a prediction.
A pattern of indirect consequences can also reveal that an unseen mechanism needs revision.
See How Students Distinguish Direct and Indirect Evidence in Science.
Primary 3: Learn That Explanations Can Be Updated
Young students can begin with simple language: “I thought X, but now I observed Y, so I need to change my explanation.”
The goal is to make revision normal rather than embarrassing.
Primary 4: Connect New Evidence to the Exact Part That Changes
Students can identify whether new data changes the claim, the mechanism, the boundary conditions or only the confidence level.
This prevents whole explanations from being discarded unnecessarily.
Primary 5: Compare Revised Explanations
Students can evaluate whether a revised explanation accounts for more evidence than the original and whether it makes useful new predictions.
Primary 6: Evidence Revision Must Survive PSLE Novelty
At Primary 6, unfamiliar data may conflict with an initial conclusion or reveal a missing condition.
Students should be able to update the explanation specifically rather than repeat the memorised model regardless of the evidence.
Diagnose First: Where Does Evidence-Driven Revision Break?
- The original explanation is protected regardless of contradictory evidence.
- One anomaly causes the entire explanation to be abandoned immediately.
- New evidence is not judged for quality.
- Confidence remains all-or-nothing.
- Revisions add unsupported exceptions only to save the original claim.
- Alternative explanations are not compared fairly.
- The revised model fits new evidence but loses earlier evidence.
- Mechanism is replaced by relabelling.
- Predictions are not updated after the explanation changes.
- Students treat changing an explanation as failure rather than scientific progress.
Catch Up | Keep Up | Move Ahead
Catch Up: after each new result, ask whether it supports, weakens or leaves the current explanation unchanged.
Keep Up: identify which part of the explanation needs revision and what new prediction follows.
Move Ahead: compare competing revised models across multiple evidence streams and rank them by predictive success, mechanism and unsupported assumptions.
Why 3-Pax Helps Evidence Revision
Three students may respond differently to the same new evidence.
One keeps the original model, another modifies one condition, and another proposes a replacement explanation.
Comparing the three makes revision criteria explicit: which account explains more, predicts better and requires fewer unsupported fixes?
What Parents Can Look For
- The child changes confidence when evidence changes.
- One anomaly does not trigger panic.
- Repeated contradiction is taken seriously.
- Evidence quality is considered.
- Revisions remain testable.
- Alternative explanations are compared.
- Earlier successful evidence is not forgotten.
- The child can explain exactly what changed in the model and why.
Frequently Asked Questions
Why do scientific explanations change?
They change when new evidence shows that the current explanation is incomplete, too broad, too narrow or less successful than another explanation.
Does changing an explanation mean the earlier science was useless?
No. Earlier explanations may remain useful within a limited range or may provide the foundation that later evidence refines.
Should one unusual result overturn a strong explanation?
Not automatically. The unusual result should be investigated, repeated and compared with the wider evidence base before confidence is changed substantially.
How does this help PSLE Science?
It helps students respond to unfamiliar data, revise conclusions accurately and avoid forcing memorised explanations onto evidence that does not fit.
A Final Reflection: Strong Explanations Are Correctable
A scientific explanation earns trust not by being protected from change, but by surviving testing and changing when the evidence requires it.
Students who learn this become less attached to being immediately right and more committed to keeping their understanding aligned with what the evidence actually shows.
That is a deeper form of scientific confidence: confidence that remains correctable.
For the wider Primary Science journey, return to Science Tuition Sengkang.
eduKate Sengkang 