How to Keep the Science Consistent Across a Multi-Part PSLE Science Question

HOW TO LEARN PSLE SCIENCE — Student Guide

Wait, What? Part (b) May Still Be Living in the World Built in Part (a)

A multi-part PSLE Science question can look like several small questions. Scientifically, it may be one continuing investigation, one system changing over time, or one set of evidence being examined from several angles.

A learner answers Part (a), turns the page in their mind, and treats Part (b) as a brand-new world. The result is a strange contradiction: the switch that was opened is suddenly treated as closed again, the plant condition that changed is forgotten, or an observation from the first part disappears from the explanation.

In a linked Science question, do not remember only your previous answer. Remember the scientific state of the setup.

Quick Answer

For a multi-part PSLE Science question, keep a tiny state ledger:

Shared givens: information that continues to apply.
Current change: what has just been altered.
Current evidence: what is observed or measured now.
Scientific relationship: the concept or mechanism relevant to this part.
Current target: what this sub-question asks you to state, predict, compare or explain.

Before each new part, ask: What still holds? What changed? What new evidence was added? What is merely my earlier inference?

The Exact PSLE Science Learning Job This Guide Owns

This guide owns one learner job: how a Primary 5/6 learner preserves a coherent scientific model across linked sub-questions, including shared setup information, state changes, evidence and causal relationships, without accidentally resetting the problem or treating an earlier personal inference as a new given fact.

It does not create a new owner for circuits, plants, heat, matter, forces, ecosystems or experimental design. Those scientific concepts remain with their canonical pages. This page teaches the cross-question reasoning job.

What the Official PSLE Science Frame Supports

For examination from 2026, PSLE Science assesses the 2023 Primary Science syllabus. SEAB’s published assessment objectives include application of scientific knowledge, interpretation and analysis of information, evaluation, prediction or hypothesis, and communication of explanations and reasoning.

Those skills require learners to preserve the meaning of evidence and conditions as a problem develops. This guide’s state-ledger method is a learning tool, not an official SEAB answering format. There is also no claim here about a universal “error carried forward” marking rule. Read and answer each part according to what is actually stated and asked.

Why Multi-Part Questions Break Otherwise Good Science

A single part may ask only one small thing, but the learner has to manage several layers at once:

the original setup;
the labels of objects A, B, P or Q;
conditions that remain unchanged;
a condition altered in a later part;
measurements from earlier parts;
a previous conclusion;
a new question target.

If these layers are blended together, the learner can know each concept correctly and still answer the wrong scientific world.

The State Ledger

Ledger field	Question to ask
Objects	Which object, organism, setup or measured quantity are we still talking about?
Shared givens	Which stated facts continue to apply?
Changed condition	What is different now?
Evidence	What observation or measurement has been added?
Inference	What did I conclude, rather than observe?
Target	What does this part now want?

You rarely need to write the full table in an exam. During practice, use it until the habit becomes fast enough to compress into a few marks beside the question.

The Multi-Part Reasoning Law

READ SHARED SETUP → MARK CURRENT STATE → SEPARATE OLD EVIDENCE FROM NEW EVIDENCE → IDENTIFY WHAT CHANGED → SELECT THE RELEVANT CONCEPT → EXPLAIN THE CURRENT MECHANISM → ANSWER THE CURRENT TARGET → CHECK CONSISTENCY WITH THE WHOLE SETUP.

Rule 1 — Shared Givens Continue Until the Question Changes Them

Suppose a question begins by stating that two containers are identical, contain equal amounts of water and begin at the same temperature. Later sub-parts may refer back to those containers without repeating every condition.

If no later instruction changes those shared givens, do not quietly replace them with new assumptions.

At the same time, do not assume every condition continues forever. A later line may explicitly change the amount of water, material, temperature, switch position or exposure time. The current state must reflect the newest stated condition.

Rule 2 — Your Earlier Answer Is Not Automatically a Given

This distinction is essential.

If Part (a) gives a graph and you infer that Factor X increased the measured response, that is your interpretation. If Part (b) asks for another prediction from the original data, do not treat your own Part (a) wording as if the question itself had stated it.

Return to the evidence whenever possible.

Carry forward the question’s world. Re-check your own conclusions.

Rule 3 — A Later Change Can Replace an Earlier State

A circuit begins with the switch closed. Part (a) asks what happens. Then the question says, “The switch is opened.” From that sentence onward, the system state has changed.

A learner who keeps using “switch closed” in the next explanation is solving the old state, not the current one.

Worked Example 1 — Circuit State Across Three Parts

Imagine an original circuit containing a cell, two lamps and a switch. The diagram shows a closed conducting path.

Part (a): predict whether Lamp X lights.

Your state ledger:

STATE A
source present
path to X closed
X connected in path
TARGET: light / not light

Now the question says the switch is opened.

Part (b): explain the effect on Lamp X.

STATE B
same source
switch now OPEN
relevant path interrupted
TARGET: effect on X + reason

Later, a wire is added creating a different path.

Part (c): predict again.

STATE C
switch remains open unless changed
NEW wire adds path
trace path to X again
TARGET: new outcome

The important skill is not memorising “closed switch means light”. It is updating the system after each stated change.

Worked Example 2 — A Plant Investigation That Changes Phase

Two similar plants are first kept under comparable conditions except for one factor. Their responses are measured. In a later part, both are moved to a new light condition and observed again.

A common error is to answer the later part using the original light conditions because those were important earlier.

Use phase labels:

	Phase 1	Phase 2
Plant A	original conditions	new stated condition
Plant B	original conditions	new stated condition
Evidence	first measurement	later observation
Question	interpret Phase 1	reason from Phase 2

Now you can ask what evidence belongs to which phase instead of mixing both.

Worked Example 3 — Cooling Water Across Linked Questions

A container of warm water is measured at regular times. Part (a) asks you to describe the trend. Part (b) asks for a scientific explanation. Part (c) says the same investigation is repeated with insulation added.

Each part has a different job:

Part (a): report what the data show without inventing a mechanism.
Part (b): connect the observed temperature change to the relevant thermal process.
Part (c): update the system because a new material condition has been introduced, then predict or explain the changed pattern.

The same dataset may support several reasoning jobs. Do not give the Part (b) explanation when Part (a) asks only for the trend, and do not forget the new insulation condition in Part (c).

Rule 4 — Keep Object Labels Stable

Multi-part questions often use P/Q, A/B, X/Y or Plant 1/Plant 2. A surprisingly large error source is switching identities halfway through.

Write a tiny identity key:

P = thicker material
Q = thinner material

or

A = more light
B = less light

Do not rely on visual position alone because diagrams can change between parts.

Rule 5 — Separate “Still True” From “No Longer Relevant”

An earlier fact may still be true but irrelevant to the current target.

For example, Part (a) may establish that a material is flexible. Part (b) may ask about electrical conductivity in a new use. Flexibility may remain true but may not help answer the conductivity question.

Consistency does not mean repeating everything from earlier parts. It means preserving the correct scientific state while selecting the evidence relevant now.

Rule 6 — Do Not Let a Wrong Earlier Inference Poison Every Later Part

If you become uncertain about an earlier answer, go back to the original diagram, table, observation or stated condition. Reconstruct from evidence rather than building a tower on top of an unverified guess.

This does not imply any promise about how an examination marker handles linked errors. It is simply the best scientific reasoning habit: the primary source is the question evidence, not your memory of what you wrote three minutes ago.

The “Shared / Changed / New” Margin Code

During practice, mark each sub-part with three letters:

S — Shared: which original givens still apply?
C — Changed: what condition has been altered?
N — New: what new evidence or target appears?

A ten-second note can prevent a full-state reset.

When a Scratch Model Helps

For a complicated sequence, use a small state model:

STATE 1 ──change A──▶ STATE 2 ──change B──▶ STATE 3
  data 1                 data 2                ?

This is especially useful when the system changes physically between sub-parts. It keeps the order of changes visible.

When a Table Helps

If several variables or setups must be compared, a table can prevent conditions from migrating between parts:

Condition	Part (a)	Part (b)	Part (c)
Object	A	A	A
Condition X	low	low	high
Condition Y	same	same	same unless stated
Evidence	measurement 1	same evidence interpreted	new measurement

Observation, Inference and Earlier Answer Must Stay Separate

Use three boxes if the question is becoming confusing:

QUESTION GIVES       I INFERRED       CURRENT PART ASKS
-------------        ----------       ----------------
measured value       mechanism?       predict outcome
switch position      likely cause?    compare setup

Only the first column is guaranteed by the question. The second column may be correct, but it should remain open to checking.

Earliest Weak-Link Diagnosis

Failure signature	Likely first weak link	Repair
You answer Part (b) using the original condition after it changed	State update failed	Mark C = changed before solving
You swap A and B	Object identity lost	Write a one-line identity key
You treat your Part (a) guess as fact	Evidence/inference boundary lost	Return to the original data
You repeat the same answer in every part	Current target not read	Circle the command word and target
You forget an earlier measurement that still matters	Shared evidence not carried	Use state ledger or S/C/N code

Misconception Repair — “Each Part Is Independent” Is Too Simple

Some sub-questions are largely independent. Others depend on a common setup or a new state created by earlier instructions. You must read the actual structure instead of adopting a universal rule.

The safest habit is: read the current part locally, then check it against the shared setup globally.

Misconception Repair — “Use My Previous Answer” Is Also Too Simple

An earlier answer can remind you of your reasoning, but it is not automatically authoritative. If Part (a) was wrong, blindly carrying it forward can multiply the error. Use the question’s stated evidence and conditions as the anchor.

Model Limits

Not every multi-part item uses one continuous system state.
A later part may explicitly introduce a fresh setup.
The same fact can remain true but become irrelevant.
Your earlier inference is not equivalent to a stated given.
This guide does not claim any universal carry-forward marking policy.
A state ledger is a learning scaffold, not an official required format.

Practice Sequence — Learn the State, Not the Letter

Take one original three-part question.
Before answering, write the initial state in one line.
After each instruction, update only what changed.
For every part, mark S, C and N.
After completing all parts, check whether the scientific state remains consistent.
Now rewrite the same structure with different objects.
Return several days later to a new multi-part question without the scaffold.

Unfamiliar Transfer Test — The Mystery System

A made-up device has a source, valve, tube and wheel. In Part (a), the valve is open and the wheel turns. In Part (b), the valve is closed. In Part (c), a second route is added around the closed valve.

You do not need to know what the machine is called. Can you preserve each state, update only the changed connection, and reason about the wheel’s outcome from the current path? If yes, the multi-part consistency skill is transferring beyond familiar textbook diagrams.

Delayed Independent Return Test

Two days later, attempt a linked question from another theme. After finishing, reconstruct your state ledger from memory:

What were the original shared givens?
What changed before each part?
Which evidence belonged to each state?
Did any of your own inferences accidentally become a “given”?
Did the final part remain consistent with the latest stated condition?

The Final Multi-Part Answer Receipt

I know which object this part is about.
I know which original givens still apply.
I updated every explicitly changed condition.
I separated earlier evidence from my earlier inference.
I used the current state, not the starting state.
I answered the current command word and target.
My explanation does not contradict the question’s latest information.
I returned to the original evidence if an earlier answer felt uncertain.

Useful Internal Routes

Parent and Tutor Teaching Guide

When a child gives a scientifically strange answer in Part (c), do not assume the concept is weak. Ask the child to reconstruct the states.

“What was true at the beginning?”
“What changed before Part (b)?”
“Did that change remain in Part (c)?”
“Which information is measured, and which part did you infer?”
“Are you still talking about A, or did you switch to B?”
“What does this part ask that the previous part did not?”

If the child’s Science becomes correct once the state is reconstructed, the repair is likely state tracking rather than another hour of chapter rereading.

Authoritative References and Further Learning

The Quiet Ending

Part (a), Part (b) and Part (c) may look like three boxes on a page.

Sometimes the Science underneath is one world changing state.

Carry the world forward. Update what changed. Re-check what you inferred. Then answer the Science that exists now.

Small Group Tutorials