Bishan Primary Science tuition should help students learn what science does when reality disagrees with expectation. A wrong prediction is not automatically a failure. It can be the most useful moment in the lesson because it exposes the learner’s current model.
This rebuilt legacy page now owns one distinct job inside the eduKateSG Science estate: prediction → result → model revision. It does not duplicate our experiment/variable-control or observation/inference pages. Its focus is what happens after a result arrives: does the student defend the old idea, guess a new answer, or revise the model in a disciplined way?
This URL serves families searching from Bishan. eduKateSG’s current centres are at 83 Punggol Central, Singapore 828761 and 8 Fourth Avenue, Singapore 268674; we do not claim a current Bishan branch. Selected Primary Science classes run in focused 3-pax small groups, typically 1.5 hours.
The Direct Answer
A strong Science learner should be able to run this loop:
- Predict: what do I expect and why?
- Observe: what actually happened?
- Compare: where do prediction and result differ?
- Diagnose: was the issue concept, setup, measurement or interpretation?
- Revise: what should change in my explanation?
- Retest: does the revised model predict a new case better?
Science is not “guess correctly first”. It is “update well when evidence arrives”.
Why Students Hide Wrong Predictions
School culture can make students fear being wrong, so they sometimes change an answer after seeing the result and pretend they “knew it”. That destroys a useful comparison.
We preserve the original prediction. The prediction tells us what model the student was using before feedback. The result tells us where that model succeeds or fails.
The learning happens in the gap.
Prediction Needs a Reason
“I think A” is not enough. We ask: what relationship makes you expect A?
Example: a student predicts that a plant kept without light will grow normally because “it still has water”. The reasoning reveals an incomplete model that overweights water and underweights light.
Now the later result can repair a specific dependency rather than merely replacing A with B.
Unexpected Result 1: Concept Model Is Incomplete
A student predicts that increasing one factor will always increase an outcome. The result levels off.
The revised model may need a limit, competing factor or necessary condition.
This is more valuable than memorising “the correct graph”. The student learns that scientific relationships often have boundaries.
Unexpected Result 2: The Setup Did Not Test the Intended Idea
Sometimes the concept is fine but the investigation changed more than one variable or measured the wrong outcome.
We ask whether the evidence is actually capable of answering the original question.
A result should not force a conceptual revision when the test itself is invalid.
Unexpected Result 3: Measurement or Observation Error
A reading may be inconsistent because of timing, instrument use, human observation or recording mistakes.
Students should not automatically discard an inconvenient result, but neither should they treat every single reading as perfect truth.
We check method, repeat where appropriate and compare the pattern.
Unexpected Result 4: The Student Misread the Evidence
A table may show a difference, but the student may read the wrong row or ignore units. A graph may rise overall but include local variation.
Before revising the concept, verify the representation.
This protects students from “learning” a false correction.
Worked Example: Dissolving
Prediction: “More stirring makes more sugar dissolve forever.”
Result: stirring helps the sugar dissolve faster, but eventually no more dissolves under the same conditions.
Revision: distinguish rate from maximum amount under current conditions.
The key learning is not a sentence to memorise. It is the repair of a category error: faster is not the same as more without limit.
Worked Example: Shadows
Prediction: moving an object farther from a light source always makes the shadow larger.
Result: under a given arrangement, the shadow becomes smaller.
Revision: the student re-examines source–object–screen geometry instead of memorising a directional slogan.
A changed arrangement becomes the retest.
Worked Example: Heat
Prediction: two cups left in the same room will remain at their starting temperatures because nothing is “heating” them.
Result: temperatures move toward the surrounding condition over time.
Revision: heat transfer is about temperature difference and energy transfer, not only obvious sources such as flames.
The Revision Must Be Smaller Than “Everything I Knew Was Wrong”
Good model revision preserves what still works. If a student’s explanation was correct except for one missing condition, add the condition instead of replacing the whole topic.
This teaches scientific continuity: update the failing part while keeping supported structure.
Counterexamples Are Useful Boundaries
If a student says “metal always feels colder because metal is naturally cold”, compare objects left in the same room. The counterexample helps separate sensation from temperature.
A good counterexample does not merely say “wrong”. It shows where the rule breaks.
Retest the Revised Model
After correction, give a new surface. If the child can predict the changed case and explain why, the model is becoming portable.
Redoing the original item proves memory of the correction. A changed case tests learning.
P3–P4: Make Prediction Visible
Middle-primary students benefit from short “predict and explain” moments before demonstrations, diagrams or data. The tutor keeps the language simple and focuses on one relationship at a time.
P5–P6: Add Boundaries and Rival Explanations
Upper-primary students can compare two plausible models, inspect data and revise claim strength. This strengthens both MCQ discrimination and open-ended explanations because the student understands why one model survives the evidence better.
Common Model-Revision Failure Modes
| Symptom | Underlying issue | Repair |
|---|---|---|
| Changes prediction after seeing result | Original model hidden | Record prediction first |
| Wrong result = wrong concept | Method not checked | Inspect setup/measurement before conceptual revision |
| Correction memorised only | No transfer | Retest on changed case |
| One exception destroys whole topic | Revision too global | Identify failing condition only |
| Student defends first answer despite evidence | Identity attached to prediction | Treat updating as scientific success |
| Data read incorrectly | Representation failure | Verify axes, units and comparison first |
Why Three Students Helps
Three students may produce three different predictions. The tutor can compare their underlying models before showing the result.
Afterwards, each student explains what changed in their model and what stayed the same. This makes revision visible instead of turning the lesson into answer revelation.
A Typical 1.5-Hour Model-Revision Lesson
- Retrieve: state prior concept.
- Predict: commit to outcome and reason.
- Observe/read data: record result accurately.
- Compare: locate prediction-result gap.
- Diagnose: concept, method, measurement or interpretation?
- Revise: update smallest necessary part.
- Retest: changed case without tutor prompting.
When Bishan Families May Consider This Support
- The child memorises corrections but repeats the same misconception.
- Unexpected data causes guessing.
- The student changes answers after seeing solutions without understanding why.
- Science rules are treated as absolute slogans.
- Open-ended explanations struggle when the question surface changes.
What Progress Should Look Like
A stronger Science learner can say, “My prediction failed because this part of my model was incomplete,” then use the revised idea on a different question.
Current class enquiries: selected eduKateSG Primary Science 3-pax small groups, typically 1.5 hours. Current centres: 83 Punggol Central, Singapore 828761 and 8 Fourth Avenue, Singapore 268674. Contact +65 8823 1234.