How to Update a PSLE Science Explanation When New Evidence Is Added

Wait, What? A good Science explanation is allowed to change when the evidence changes. That is not a weakness. It is one of the strongest signs that you are actually reasoning from evidence instead of protecting the first answer you wrote.

In PSLE Science, a question may give you an initial diagram, observation, table or description, then add another result later. The new information might support your first explanation, weaken it, show that it was too broad, or leave it untouched. The difficult part is not merely noticing the new fact. The difficult part is deciding what the new evidence changes in your scientific model—and what it does not change.

This guide owns one exact Primary 5/6 learner job: update a provisional PSLE Science explanation when additional evidence is introduced. It does not own the underlying Physics, Chemistry, Biology, plant, animal, materials or systems concepts. Those concept pages remain canonical. Here, the scientific examples exist only to train evidence-sensitive reasoning.

Quick Answer

Do not erase your first explanation immediately. First write what the earlier evidence supported. Then isolate the new evidence and ask: Does this strengthen the explanation, weaken it, narrow it, contradict one part of it, or leave it unchanged? Revise only the affected link. Preserve every part that is still supported.

A useful reasoning chain is: READ THE EARLIER EVIDENCE → STATE THE PROVISIONAL EXPLANATION → READ THE NEW EVIDENCE → IDENTIFY THE SCIENTIFIC OBJECT AND RELATIONSHIP → COMPARE PREDICTION WITH NEW OBSERVATION → LOCATE THE AFFECTED LINK → REVISE ONLY THAT LINK → STATE THE UPDATED OUTCOME → CHECK THE WHOLE EXPLANATION AGAINST ALL THE EVIDENCE.

The Owned PSLE Science Learning Job

The current 2026 PSLE Science syllabus assesses the 2023 Primary Science syllabus. Besides knowledge with understanding, learners are expected to apply scientific facts, concepts and principles and use scientific inquiry, including interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning. Updating an explanation when another observation appears sits directly inside that broad inquiry habit: the explanation must remain answerable to the information supplied.

This is not an official answer template or marking formula. It is a learning protocol for keeping reasoning correct when evidence arrives in stages.

Why Learners Often Keep the First Explanation Too Long

The first explanation feels special because you produced it first. Once a learner has invested effort in an answer, later information can be treated as an inconvenience. The learner may squeeze the new result into the old explanation, ignore it, or rewrite the earlier evidence as though the final answer had been obvious from the start.

Science reasoning works differently. An explanation is provisional: it should fit the evidence available at that point. When the evidence changes, the explanation earns the right to stay only if it still fits.

New evidence does this	What you should do	What you should not do
Supports the same predicted pattern	Strengthen confidence cautiously	Claim absolute proof
Fits only a narrower version of the explanation	Limit the explanation to the supported conditions	Keep the original over-broad claim
Conflicts with one causal link	Repair that specific link	Throw away every part of the model
Conflicts with the predicted outcome	Recheck assumptions, conditions and mechanism	Pretend the observation did not occur
Concerns a different object or quantity	Keep the original explanation unless a connection is justified	Treat every new fact as relevant
Adds no independent information	Recognise it as a restatement or new representation	Count it as automatically stronger evidence

The Four Main Update Outcomes

1. Strengthen

The new evidence is consistent with what the explanation predicted and gives additional relevant support. You can keep the mechanism, while remaining careful not to turn support into certainty beyond the evidence.

2. Weaken

The new evidence does not fit an important prediction of the explanation. The explanation may still contain useful parts, but one or more links now need rechecking.

3. Limit

The explanation works under some of the tested conditions but not all. Instead of saying “this always happens,” the scientifically better move is often to say “this explanation is supported under these conditions.”

4. Leave unchanged

The new evidence may be true but irrelevant to the explanatory claim being tested. Scientific updating includes the discipline not to change a model merely because more information appeared on the page.

Do Not Rewrite History

Suppose your first prediction was reasonable from the evidence available, but a later result shows that it was incomplete. Do not pretend you “knew all along” what the later evidence would show. Keep the time order clear:

Earlier evidence: what was known first.
Provisional explanation: what that evidence reasonably supported.
New evidence: what became known later.
Updated explanation: what should now be changed, preserved or limited.

This makes learning more useful. If you hide the original reasoning, you cannot diagnose which assumption failed.

The EVIDENCE UPDATE Protocol

Step 1: Freeze the earlier state

Before reading the new evidence into the old explanation, write a short scratch note: BEFORE: Evidence ___ supported explanation ___ because ___. This prevents hindsight from quietly changing what you thought earlier.

Step 2: Read the new evidence literally first

Separate observation from inference. If the new table shows a lower value, record the lower value or relationship before deciding why it happened. If a diagram shows a changed position, record the position before naming a mechanism.

Step 3: Attach the evidence to the correct object, quantity and condition

Ask: Which set-up is this? Which specimen? Which part? Which measured quantity? At what time? Under which condition? A new observation cannot update the correct explanation if it has been attached to the wrong object.

Step 4: Ask what the old explanation predicted

A scientific explanation becomes testable when it implies what should happen under stated conditions. Say the prediction explicitly. The new observation can then be compared with something concrete instead of with a vague feeling that the answer “looks right.”

Step 5: Locate the first mismatch

Do not discard the whole explanation at once. Walk through the chain from condition to mechanism to outcome. Which exact link no longer fits? Perhaps the starting condition was misunderstood. Perhaps the mechanism was correct but the explanation claimed it applied everywhere. Perhaps the outcome belongs to a different quantity.

Step 6: Revise the smallest necessary part

If one link fails, repair one link. If the evidence shows that a relationship holds only up to a tested condition, narrow the range rather than replacing the entire concept. If the new observation concerns a different object, do not rewrite the original object’s mechanism without a scientific connection.

Step 7: Rerun the full chain

After revising, read all the evidence again. The updated explanation should fit both the earlier and new information, or clearly state which earlier interpretation has been replaced.

Worked Reasoning Example 1: New Data Strengthens the Same Relationship

Consider an original practice situation. A learner first sees three tested conditions. As condition X increases across those three cases, measured outcome Y also increases. The learner writes a provisional relationship for the tested range. A fourth measured point is then added, and it continues the same pattern.

The new point gives additional relevant support for the relationship within the now-expanded measured range. But it does not prove that the relationship continues forever. The correct update is not “Now I know Y always increases whenever X increases.” It is closer to: the additional result supports the same relationship across the tested conditions now shown.

The mechanism, if the question asks for one, still needs scientific knowledge appropriate to the context. A data pattern and a causal mechanism are different jobs.

Worked Reasoning Example 2: New Evidence Limits an Over-Broad Explanation

In another original example, the first three conditions suggest that increasing X is associated with a larger Y. The learner writes, “The larger X is, the larger Y will be.” A new result at a higher tested X shows that Y remains the same instead of increasing further.

Do not force the final result into the first statement. The new evidence tells you that the original wording was too broad. The update might be: Y increased as X increased over the earlier tested conditions, but the new result shows that this increase did not continue at the latest tested condition.

Whether the reason is a limit, competing process or some other mechanism depends on the scientific context. Do not invent an advanced explanation that the question does not support. First repair the evidence claim.

Worked Reasoning Example 3: New Evidence Weakens One Causal Link

Suppose two original set-ups differ in several visible ways. The first result seems consistent with the learner’s idea that feature A causes the difference in outcome. A new comparison is then introduced in which feature A is present but the predicted outcome does not occur.

The new comparison does not automatically prove that A is irrelevant in every situation. It does show that the simple claim “A alone causes the outcome” is no longer supported. The learner should now ask whether another condition is required, whether A was only associated with the first result, or whether the original comparison changed more than one factor.

The repair is evidence-first: the new case shows that A by itself is not enough to account for the outcome under the conditions given. Only then should the learner select the appropriate scientific concept to rebuild the explanation.

Worked Reasoning Example 4: New Information Is True but Does Not Change the Explanation

Imagine a question about the measured output of one part of a system. After the learner builds an explanation, a later sentence provides an additional property of a different component. The property is scientifically true, but no relationship is given that connects it to the measured output.

The disciplined update may be no update. More information is not automatically more relevant information. Ask whether the new fact changes a condition, mechanism, evidence link or outcome in the explanation. If not, keep it outside the causal chain.

Worked Reasoning Example 5: The New Evidence Changes Only One Part of a Multi-Part Explanation

Suppose an original explanation contains three linked ideas: condition A changes process B; process B affects quantity C; quantity C changes outcome D. A new observation directly shows that B did occur, but D did not change as predicted.

You should not delete the first link, because the new evidence supports B. Instead, inspect the later connection from B to C or C to D. This is the central skill of scientific updating: preserve supported links and repair unsupported ones.

New Evidence Is Not the Same as a New Condition

This guide is different from rebuilding an explanation when the experimental condition itself changes. Here, the world described by the question may be exactly the same; what changes is what you know about it. An extra measurement, observation or comparison can make the earlier explanation stronger or weaker even when no physical condition has been altered.

Keep these two situations apart:

Situation	What changed?	Your reasoning job
Condition update	The scientific set-up or condition is different	Rebuild what should happen under the new condition
Evidence update	Your information about the same situation is richer	Re-evaluate whether the earlier explanation still fits

Observation Before Inference

When new evidence surprises you, the temptation is to explain it immediately. Slow down enough to preserve the difference between what the evidence says and why you think it happened.

For example, “the reading at 8 minutes is lower than the reading at 6 minutes” is an observation from data. “The process slowed because ___” is an explanation. If the mechanism is not supported by the question or relevant Primary Science knowledge, do not smuggle it into the observation.

Use Prediction as the Bridge

A powerful way to evaluate new evidence is to ask what your current explanation would predict. This turns a vague explanation into something checkable.

If my explanation is correct under these conditions, what should I observe?
What new observation was actually given?
Do prediction and observation match?
If not, which assumption or link produced the wrong prediction?

A mismatch is useful. It tells you where to investigate your reasoning. It does not automatically tell you the replacement explanation.

Evidence Can Strengthen Without Proving

One more matching result can increase support for an explanation, but it does not remove every alternative explanation or prove a universal rule. Evidence quality depends on what was compared, what was controlled, what was measured, the range tested and whether another explanation still fits.

For Primary learners, a safe habit is: state what the new evidence supports, then stop at the boundary of what was actually tested.

When Evidence Seems to Contradict the Explanation

Do not jump directly to “my concept is wrong.” Check the layers in order:

Reading: Did I read the new value, label, axis, time or comparison correctly?
Object: Does the evidence belong to the same object or set-up?
Condition: Are the conditions actually comparable?
Observation versus inference: Am I treating an interpretation as though it were measured?
Mechanism: Did I select the right scientific concept?
Scope: Was my original claim too broad?
Method: Could the investigation or measurement limit what the result supports?

Only after these checks should you decide what to revise.

Common Failure Signatures

First-answer protection: the learner keeps the first explanation even when later evidence no longer fits it.
Total reset: one surprising result causes the learner to throw away every supported part of the explanation.
Hindsight rewrite: the learner changes the record of what the earlier evidence appeared to support.
Evidence swallowing: the new observation is immediately converted into a mechanism without first recording what was actually observed.
Relevance failure: every new fact is forced into the explanation even when it concerns a different object or quantity.
Over-generalisation: one supporting result becomes an “always” rule.
Exception memorising: instead of repairing the mechanism or scope, the learner memorises the new result as a special exception.
Condition drift: the learner compares evidence collected under different conditions as if they were identical.
Representation double-counting: the same measurement shown in both a table and graph is treated as two independent experiments.

Earliest Weak-Link Diagnosis

If a learner struggles with an evidence update, diagnose from the beginning rather than correcting the final sentence.

Can the learner state the earlier explanation without looking at the later evidence?
Can the learner describe the new observation without explaining it?
Can the learner identify the object, quantity, time and condition belonging to the new evidence?
Can the learner say what the earlier explanation predicted?
Can the learner decide whether prediction and observation agree?
Can the learner locate one affected causal or evidential link?
Can the learner preserve the links that remain supported?
Can the learner state a narrower explanation when the evidence supports only a narrower claim?
Can the learner rerun the updated explanation against all available evidence?

The first failed operation is the repair target. If the learner cannot separate observation from explanation, teaching “better conclusion words” will not solve the problem. If the learner identifies the mismatch correctly but lacks the relevant concept, return to the concept owner before practising explanation revision.

A Two-Column Scratch Method

Before new evidence	After new evidence
What did I know?	What new observation or information do I know now?
What did my explanation predict?	Does the new evidence match that prediction?
Which links were supported?	Which exact link is strengthened, weakened, limited or untouched?
What was uncertain?	Has that uncertainty changed?

This is a practice tool, not a compulsory examination layout. Its job is to make evidence updating visible until the learner can perform it mentally.

How to Repair an Explanation Without Patching It With Random Facts

A common correction habit is to add more Science words until the answer seems sophisticated. Instead, use a surgical repair:

Cross out only the unsupported link.
Write the new evidence beside that link.
Ask which relevant concept explains the evidence under the stated condition.
Insert the smallest replacement mechanism.
Read the sentence before and after it to make sure the causal direction still works.
Remove any added fact that does not help the chain.

Retrieval and Practice Sequence

Start with one short explanation and one new supporting observation. Ask the learner to identify that the explanation can stay.
Use a second example where the new evidence limits an “always” claim. Ask the learner to rewrite only the scope.
Use a third example where one causal link fails but earlier links remain supported.
Add a true but irrelevant new fact and ask the learner to leave the explanation unchanged.
Show the same evidence in a different representation and ask whether it is genuinely new information or merely another view of the same information.
Use an unfamiliar Science context and ask for BEFORE, NEW EVIDENCE, AFFECTED LINK and UPDATED EXPLANATION.
Return after a delay without the scratch template and ask the learner to update independently.

Unfamiliar Transfer: The Topic Should Be Able to Change

Real mastery means the learner can update explanations in different scientific contexts. The surface topic may change from an investigation to a system diagram, a graph, a classification observation or an energy-related situation. The stable learner job is evidence revision.

Ask: What was believed before? What has now been observed? Which link changes? If the learner can answer those three questions without relying on a familiar chapter example, transfer is improving.

Delayed Independent Return Test

Several days later, present an original problem in two stages. Show Stage 1 only and ask for a provisional explanation. Keep the answer. Then reveal Stage 2 with an additional observation. Ask the learner to update without prompts and to mark the exact phrase that changed.

A strong return has four receipts: the learner preserves the earlier evidence accurately, reads the new evidence correctly, changes only the affected reasoning, and can explain why the rest of the answer stayed the same.

Answer-Checking Receipt

I can state what the earlier evidence supported before looking at the new information.
I described the new evidence before explaining it.
I attached the new evidence to the correct object, quantity, time and condition.
I know what the old explanation predicted.
I compared prediction with observation.
I identified whether the new evidence strengthens, weakens, limits or leaves the explanation unchanged.
I changed only the affected link.
I preserved evidence and reasoning that remain valid.
I did not turn one supporting result into a universal rule.
I checked the updated explanation against all the evidence, not only the newest piece.

Parent and Tutor Teaching Guide

When a child changes an answer after seeing new evidence, do not begin by asking whether the final answer is correct. Ask, “What exactly changed in the evidence?” Then ask, “Which sentence in your explanation should change because of that?” This reveals whether the learner is updating scientifically or merely chasing the latest clue.

If the learner deletes the whole answer, ask them to underline the parts the new evidence still supports. If the learner refuses to change anything, ask what their explanation predicted and compare that prediction with the new observation. If the learner adds many facts, ask which one actually repairs the broken link.

Use staged examples during teaching. Hide the second observation at first. Let the learner commit to a reasoned provisional explanation, then reveal the new result. This makes evidence updating observable. It also teaches that changing a scientific explanation for a good reason is not the same as guessing again.

Fade the prompts. The long-term goal is a learner who can independently say: “This is what the evidence supported before; this is the new observation; this is the one part of my explanation that now needs to change.”

Useful Internal Routes

Authoritative and Learning-Evidence References

The official sources above govern syllabus and examination-specific claims. The learning-science sources support broader teaching and practice choices; they do not create PSLE marking rules. All worked situations on this page are original learning examples and do not reproduce national examination questions.

Quiet Return

Scientific reasoning is not loyalty to your first explanation. It is loyalty to the best explanation the evidence currently supports. Keep what survives the new evidence. Change what does not. And always be able to say exactly why.

Small Group Tutorials