Science tuition for Yishun families should help students look not only for evidence that supports an idea, but also for evidence that could show the idea is incomplete, wrong or too broad. This is one of the strongest protections against confirmation bias.
This rebuilt legacy URL now owns one distinct Science reasoning job: claim → expected result → disconfirming result → test → revise explanation → stronger model. Old broad promotional copy, result promises and stale Yishun-centre wording have been removed.
eduKateSG does not claim a current Yishun branch. Current centres are at 83 Punggol Central, Singapore 828761 and 8 Fourth Avenue, Singapore 268674. Selected Primary Science classes run in focused 3-pax small groups, typically 1.5 hours.
The Direct Answer
For any scientific claim, ask:
“What observation would make me change my mind?”
If no possible result could ever count against the claim, the claim is difficult to test scientifically.
A strong explanation risks being wrong: it makes predictions that evidence can support or challenge.
Worked Example: “All Metals Are Magnetic”
Claim:
“All metals are magnetic.”
Disconfirming evidence:
Find one metal object that is not attracted by the magnet under a suitable test.
One counterexample is enough to show the universal claim is false.
Revised idea:
“Some metals are magnetic; being metallic alone does not guarantee attraction.”
Worked Example: Surface Area and Evaporation
Claim:
“Greater exposed surface area increases evaporation under otherwise similar conditions.”
Expected result:
wider dish loses more water.
Disconfirming result:
repeated fair trials show the wider and narrower containers lose the same amount—or the narrower one consistently loses more.
Next step:
check whether the setup truly isolated surface area, then revise the explanation if the result remains.
Worked Example: Plant Growth
Claim:
“More light caused Plant A to grow more.”
What result would weaken that claim?
- Plant A and B received different water amounts;
- the darker plant consistently grew more in repeated fair trials;
- starting plant sizes were very different;
- growth measure was recorded inconsistently.
Disconfirming evidence may attack the result, the design or the explanation.
Supporting Evidence vs Disconfirming Evidence
Supporting:
“The wider dish lost more water.”
Disconfirming:
“A repeated fair trial found no difference.”
A strong learner should be able to describe both before making a final judgement.
The Counterexample Test
Universal claim:
“Every object that moves is living.”
Counterexample:
a moving toy car.
The counterexample shows that movement alone is not sufficient evidence of life.
The Reversal Test
Claim:
“If a plant receives more water, it will always grow more.”
Test the opposite range:
What if the plant receives excessive water?
If growth decreases, the original “more is always better” claim was too broad.
Worked Example: Food Web
Claim:
“If rabbits decrease, foxes will decrease.”
Potential disconfirming evidence:
fox numbers remain stable while another prey species increases.
Revised model:
foxes may compensate with alternative food, so rabbit decline does not guarantee immediate fox decline.
Worked Example: Insulation
Claim:
“Material X is a good insulator.”
Supporting:
cup wrapped with X cools less than unwrapped cup.
Disconfirming:
under repeated fair tests, X gives the same cooling as no wrapping.
The learner should revise the claim rather than keep it because “the textbook says insulating materials exist”.
Prediction Before the Test
Write both:
If my explanation is correct, I expect ___.
If my explanation is wrong/incomplete, I might observe ___.
This makes the investigation intellectually honest before the data arrives.
Confirmation Bias
Confirmation bias means noticing evidence that agrees with an existing belief while overlooking evidence that conflicts.
Science repair:
- record all results;
- mark supporting and conflicting evidence separately;
- do not delete anomalies automatically;
- revise the model if conflict persists.
Worked Example: Magnet Strength
Prediction:
Magnet A is stronger because it attracted more paper clips once.
Disconfirming test:
repeat with same clips and distances several times.
If B performs better in most repeated trials, the first observation was not enough.
Disconfirming Evidence Can Reveal a Hidden Variable
Suppose an evaporation prediction fails only on a windy day.
The conflict may reveal airflow as a hidden factor.
The model becomes richer:
surface area matters, but airflow also affects evaporation.
Disconfirming Evidence Can Reveal a Measurement Problem
If all results conflict with expectation, check the measuring method:
- wrong scale;
- different start points;
- inconsistent timing;
- instrument resolution;
- data transcription.
Not every conflicting result means the scientific concept is wrong; sometimes the test is weak.
Disconfirming Evidence Can Reveal a Boundary
Claim:
“Increasing light increases plant growth.”
If the relationship works from low to moderate light but stops at high light, the original claim needs a boundary:
“Within the tested range, increasing light increased growth.”
Do Not Treat One Strange Result as Automatic Refutation
One anomalous trial may come from random variation or method error.
Response:
- check procedure;
- repeat;
- compare with other trials;
- decide whether the conflict persists.
The Strongest Test Is Discriminating
A good test separates competing explanations.
Hypothesis A:
plant wilted because it received less water.
Hypothesis B:
plant wilted because its roots were damaged.
Design comparisons that vary one factor while holding the other comparable.
Worked Example: Shadow
Claim:
“Shadow size depends only on object size.”
Disconfirming evidence:
same object produces different shadow sizes when moved relative to source/screen.
Revised model:
shadow size also depends on geometry and relative positions.
Worked Example: “Heavier Objects Fall Faster”
At Primary level, students should avoid oversimplified universal claims from everyday observation.
Teacher-led demonstrations and later Science learning can challenge naive rules. The important reasoning skill is to identify what evidence would test the claim rather than protect it.
The Claim-Challenge Table
| Claim | Supporting result | Disconfirming result |
|---|---|---|
| Wider surface → faster evaporation | wider loses more | repeated fair trials show no/less loss |
| All metals magnetic | several metals attracted | one metal not attracted |
| More light → more growth | growth rises with light | relationship reverses/plateaus |
Disconfirming Evidence in MCQ
Question:
Which result would best show that the student’s explanation is incomplete?
Students should choose the result that directly conflicts with the predicted relationship, not merely a different fact.
Disconfirming Evidence in Open-Ended Answers
Useful structure:
“If X were the only cause, we would expect ___. However, the data shows ___. Therefore, X alone cannot fully explain the result.”
Why This Improves Learning
Students stop seeing correction as punishment.
A wrong idea becomes a model that met evidence and needed revision.
This is a more durable scientific mindset than memorising the model answer before seeing the experiment.
Curriculum Context
MOE’s Primary Science syllabus emphasises questioning, predicting, investigating, evaluating evidence and revising explanations across P3–P6. Official reference: MOE Primary Science Syllabus.
Common Disconfirming-Evidence Failure Modes
| Symptom | Problem | Repair |
|---|---|---|
| Only supporting evidence collected | Confirmation bias | Ask what could refute idea |
| One counterexample ignored | Universal claim protected | Revise scope |
| One odd trial destroys whole model | Anomaly overread | Check/repeat |
| Conflicting result deleted | Evidence contaminated | Record and investigate |
| No possible result could change belief | Claim not genuinely tested | Define predicted/disconfirming outcomes |
Why Three Students Helps
One student defends the claim, one designs the strongest supporting test, and the third acts as challenger who must identify the observation that would force revision.
When Yishun Families May Consider This Support
- The child changes data to preserve the expected answer.
- Counterexamples are treated as annoying exceptions rather than evidence.
- Universal claims survive obvious boundary cases.
- Investigation questions need stronger critical reasoning.
What Progress Should Look Like
A stronger learner can say what result would support an idea, what result would challenge it and how the explanation should change when persistent evidence does not fit.
Independent location notice: eduKateSG does not claim a current Yishun branch. Current centres: 83 Punggol Central, Singapore 828761 and 8 Fourth Avenue, Singapore 268674. Contact +65 8823 1234.