Wait, What? Sometimes the Question Is Harder to Hold Than the Science
A PSLE Science question can contain a paragraph, two labelled objects, a changed condition, a table, an arrow, a result and a final question. You may actually know the Science, yet lose the relationship while reading.
That is not always a concept problem. Sometimes too much information is competing for space in your head at the same time.
When the question feels too complicated to hold, move the important Science onto the page.
A scratch Science model is a tiny diagram you create for yourself. It is not an artwork. It is not a replacement for the question. It is a temporary scientific representation that keeps the important objects, conditions, relationships and evidence visible while you reason.
Quick Answer
When a PSLE Science question feels overloaded, reduce it to a small model:
- write the main object or system;
- mark the condition that changed;
- keep the important comparison;
- draw arrows only for scientifically meaningful relationships;
- separate what is given from what you infer;
- mark the outcome the question asks about;
- return to the original evidence before answering.
The goal is not to draw more. The goal is to carry less unnecessary information while preserving the scientific structure.
The Exact PSLE Science Learning Job This Guide Owns
This guide owns one learner job: how a Primary 5/6 learner converts a dense, unfamiliar or multi-condition PSLE Science question into a small scratch representation that makes the scientific objects, relationships, conditions and evidence easier to reason with.
It does not replace existing concept pages on heat, electricity, plants, forces, water, life cycles or any other Science topic. It also does not replace the separate guide on interpreting diagrams, tables and graphs supplied by the question. Here, the student creates a temporary representation to support thinking.
Why This Fits the Current PSLE Science Frame
For examination from 2026, SEAB states that PSLE Science assesses attainment in the 2023 Primary Science syllabus. Its assessment objectives include knowledge with understanding, application of scientific facts, concepts and principles, and scientific inquiry including prediction or hypothesis, interpretation and analysis, evaluation, and communication of explanations and reasoning.
The official syllabus also allows scientific ideas to be communicated in words, diagrams, tables and graphs. A student-created scratch representation is not an official examination requirement. It is a learning and problem-solving tool that can help the learner preserve the relationships needed for those official reasoning jobs.
Research on external representations and drawing-to-learn also supports a careful version of this idea: useful representations can make relationships more visible, but drawing works best when it is guided toward the scientific structure rather than treated as decoration.
A Scratch Model Is Not a Pretty Copy
Suppose the printed question already contains a detailed experimental setup. Redrawing every beaker, clamp, leaf and label wastes time and can create copying errors.
A useful scratch model might be only:
Setup A: warmer water
↓
? faster process
↓
measured outcome ↑
Setup B: cooler water
same other stated conditions
The model is intentionally incomplete. It keeps only the relationship that the question asks you to test.
The Five Things a Good Scratch Model Must Preserve
1. The Scientific Object
What is changing, moving, interacting, heating, cooling, growing, dissolving, conducting, evaporating, being classified or being measured?
Do not let “it” float around without an owner. Write the actual object when two or more objects appear.
2. The Condition
What is different between the relevant situations? Temperature? Light? Surface area? Circuit connection? Material? Time? Position? Amount?
Circle or box only the condition that matters to the question.
3. The Relationship
An arrow must mean something. It can mean “causes”, “leads to”, “transfers to”, “is transported to”, “is compared with”, “changes into” or another clear scientific relationship.
Do not draw arrows merely because diagrams look scientific.
4. The Evidence
Numbers, observations, labels, measured changes and stated conditions belong in the model when they control the answer.
Evidence should be marked differently from your explanation. For example:
GIVEN: water temperature fell from 70°C to 58°C INFER: less thermal energy remained in the water than before
5. The Target
What must the final answer state? A comparison? A reason? A predicted outcome? A variable? A method improvement? A conclusion?
Put a question mark beside the target. Your scratch model should help you reach that point, not wander beyond it.
The Scratch-Model Protocol
Use this when a question feels dense. Do not use it automatically for every easy question.
- Read once for the story. Do not solve yet.
- Read again for the scientific object. What thing or system matters?
- Mark the changed condition. What is different?
- Copy only essential evidence. Keep key numbers, directions, labels or observations.
- Draw the relationship. Use one or two arrows with words.
- Mark the target. What must be explained or predicted?
- Return to the question. Check that your tiny model did not accidentally omit a condition.
Worked Example 1 — Two Containers Cooling
Imagine two identical cups contain equal amounts of hot water at the same starting temperature. Cup P is wrapped in Material P. Cup Q is wrapped in Material Q. After ten minutes, the water in Cup P is warmer than the water in Cup Q.
You could keep the whole paragraph in your head. Or you could write:
same cup + same water + same start T + same time P: Material P → water ends warmer Q: Material Q → water ends cooler question target: which material reduces heat transfer better? WHY?
The model immediately protects the comparison. It reminds you that the cups started comparably and that the measured difference after the same time is the evidence.
Your reasoning chain can then be rebuilt:
READ GIVEN INFORMATION → IDENTIFY THE OBJECT → KEEP THE COMPARISON → SELECT THE RELEVANT CONCEPT → EXPLAIN THE MECHANISM → CONNECT TO THE MEASURED DIFFERENCE → STATE THE OUTCOME → CHECK AGAINST THE DATA.
The page is not teaching the full canonical topic of heat transfer. It is showing how a representation protects the reasoning job.
Worked Example 2 — A Circuit That Changes Between Two Diagrams
Suppose a question shows a circuit in State 1 and then shows a switch changed in State 2. Several bulbs and wires make the diagram visually busy.
Your scratch model might be:
STATE 1: path closed → Bulb X lights STATE 2: switch opens one path → ? effect on X Do not copy every wire. Trace only the path that includes X.
The tiny model directs attention to the relevant conducting path. It prevents the number of drawn wires from becoming the problem.
Worked Example 3 — A Plant Investigation With Several Conditions
Suppose two similar plants are placed in different conditions. The question gives light exposure, watering schedule, time and a final observation.
Do not write a plant essay. Build a condition table:
| Condition | Plant A | Plant B |
|---|---|---|
| Light | same stated condition | same stated condition |
| Water | more | less |
| Time | same | same |
| Measured outcome | given result | given result |
Now the changed condition is visible. The learner can connect the appropriate scientific relationship to the evidence without accidentally using a condition that did not differ.
Do Not Put Your Inference Into the “Given” Box
This is one of the most important rules.
If the question says:
Droplets appeared on the outside of a cold container.
that is given information.
If you write:
The container leaked.
you have added an explanation that was not observed.
A scratch model can make the boundary visible:
OBSERVED: droplets outside POSSIBLE EXPLANATIONS: ______ / ______ EVIDENCE NEEDED: ______
This helps prevent an early guess from becoming an invisible “fact” in the rest of your reasoning.
When a Table Is Better Than a Diagram
Use a table when the question contains several conditions or several setups. Tables are good for:
- same versus different conditions;
- before versus after;
- Setup A versus Setup B;
- changed versus measured variables;
- evidence versus inference.
When an Arrow Chain Is Better
Use arrows when the question depends on sequence or causality:
condition changes
↓
scientific process changes
↓
intermediate effect
↓
observed outcome
If you cannot label the arrow in words, your scientific relationship may still be missing.
When a Before/After Sketch Is Better
Use two simple boxes when an object changes over time:
BEFORE AFTER state / amount / position state / amount / position condition observed change
This is especially useful when students confuse final amount with amount of change.
When You Should Not Draw Anything
A scratch model is useful only when it reduces mental clutter or makes a hidden relationship visible.
Do not draw when:
- the relationship is already obvious;
- the question is a direct recall item;
- the printed diagram already makes the needed path perfectly clear;
- you are copying decoration instead of reasoning;
- drawing would take longer than solving;
- you are using the sketch to avoid committing to an explanation.
The scratch model serves the Science. The Science does not serve the scratch model.
The Earliest Weak-Link Diagnosis
If the scratch model is wrong, ask where it first broke.
| Visible problem | Likely first weak link | Repair |
|---|---|---|
| You copy every detail | Cannot distinguish relevant from background information | Write only object, changed condition, evidence, target |
| Your arrows have no labels | Relationship not yet understood | Write “causes / transfers / changes / compares” beside arrow |
| You model the wrong object | Question target not secured | Underline exactly what must be predicted/explained |
| You add facts not in question | Inference has leaked into evidence | Create separate GIVEN and INFER boxes |
| Your model changes halfway | Conditions not tracked | Use a comparison table or state labels |
Misconception Repair — A Diagram Does Not Have to Look Like the Real Object
A useful scientific representation may contain rectangles, arrows and labels instead of realistic drawings. The purpose is to preserve relationships.
A rectangle labelled “cup” can be more useful than a beautiful cup if the important question is heat transfer. Two boxes labelled “before” and “after” can be more useful than a realistic plant if the question is about a change in condition.
Misconception Repair — More Detail Is Not Always More Accurate
Extra drawing can create false precision. If the printed diagram is not drawn to scale, your copied distances should not become evidence. If the question does not specify an unseen internal process, do not invent one merely to make the picture complete.
A good scratch model is selectively accurate: accurate about what matters, silent about what is unknown.
Model Limits
- A scratch model can simplify away an important condition if you build it carelessly.
- It does not replace scientific knowledge.
- It does not prove an inference is correct.
- It is not required by SEAB as a fixed answering method.
- Different students may use different useful representations.
- A model that helped in one question may be wrong for another.
A PSLE Science Scratch-Model Toolkit
- Two-column comparison: A versus B.
- Before/after boxes: track state change.
- Arrow chain: show cause → process → outcome.
- Path trace: follow matter, energy or connection through a system.
- GIVEN / INFER split: protect evidence boundaries.
- Condition table: control several variables.
- Question-mark target: keep the required outcome visible.
Unfamiliar Transfer Test — The Mystery Device
A sealed device contains a source, a switch, two connected components and an output wheel. The printed diagram is unfamiliar. When the switch is changed, the wheel stops moving.
You do not need to know the device name. Build:
source → connection → component → wheel motion
↑
switch changes path
observed: wheel stops
question: which relationship was interrupted?
If you can reason through the unfamiliar device using the structure rather than recognition, the representation skill is transferring.
Delayed Independent Return Test
Two or three days later, choose one difficult question you have not memorised. Give yourself ninety seconds to create the smallest useful scratch model.
- Did you identify the correct object?
- Did you preserve the changed condition?
- Did you mark evidence separately from inference?
- Did your arrows have scientific meaning?
- Did the model help you answer without re-reading the whole question repeatedly?
- Did you return to the original question before finalising?
If the model itself becomes a large task, simplify it again.
The Answer-Checking Receipt
Before submitting an answer that came from a scratch model, check:
- My model used the same objects as the question.
- I did not accidentally swap A and B.
- I preserved the relevant condition.
- I did not turn an inference into a given fact.
- The causal direction is correct.
- The final answer returns to the question target.
- The evidence in the original question still supports my conclusion.
Useful Internal Routes
- How to Read a PSLE Science Question Before You Answer
- How to Turn PSLE Science Diagrams, Tables and Graphs Into Evidence for an Answer
- How to Recover When You Cannot Start a PSLE Science Question
- How to Combine Two Science Concepts in One PSLE Question
Parent and Tutor Teaching Guide
Do not begin by drawing the model for the child. Ask the child to externalise only the structure that is currently being lost.
Useful prompts are:
- “Which object are we actually tracking?”
- “What changed?”
- “What stayed the same?”
- “What does this arrow mean?”
- “Which part is given, and which part are you inferring?”
- “Can you remove one detail without damaging the Science?”
- “What does the question finally want?”
The best teaching move is often subtraction. If the learner has drawn a complicated copy, ask which marks can be erased while preserving the scientific relationship.
Over time, some learners will need fewer scratch models because the relationship becomes easier to hold internally. That is not failure of the method. It is evidence that the scaffold may be fading appropriately.
Authoritative References and Further Learning
- Singapore Examinations and Assessment Board — PSLE Science syllabus, for examination from 2026.
- Singapore Ministry of Education — Science Teaching & Learning Syllabus, Primary Three to Six, 2023 syllabus.
- Education Endowment Foundation — A systematic review of approaches to primary science teaching.
- Educational Psychology Review — meta-analysis on instructional support for drawing-to-learn.
- Hellenbrand, Mayer, Opfermann, Schmeck & Leutner — research on generative drawing and learning processes.
The Quiet Ending
The printed question may be large.
The Science underneath it may be much smaller.
A good scratch model does not solve the question for you. It gives the important relationships somewhere to stand while you solve it.
Keep the object. Keep the condition. Keep the evidence. Keep the relationship. Let the rest wait.
