How to Distinguish Evidence of a Difference From Evidence of a Cause in PSLE Science

Wait, what? Two set-ups can give different results without proving that the one obvious difference between them caused the result.

That sentence is one of the most useful safeguards in PSLE Science. A learner must often move from observations to explanations, but scientific reasoning does not allow us to jump from “these outcomes are different” straight to “this factor caused the difference” unless the comparison actually supports that move.

Quick Answer

Use two separate questions:

What difference does the evidence show? State only what was observed or measured.
Does the design justify a causal explanation? Check whether the relevant comparison isolates the proposed factor and whether the evidence is consistent with the scientific mechanism.

A difference in results is an observation. A cause is an explanation. The second requires more reasoning than the first.

Owned PSLE Science Learning Job

This guide owns one PSLE-specific learner job: how a Primary 5/6 learner decides whether given results support only a difference between outcomes or also support a causal explanation. It applies fair-comparison logic, concept knowledge and evidence limits to PSLE Science questions. It does not replace the broader science owners for causality, variables, experimental design or uncertainty.

Why This Distinction Matters

Science questions often contain a tempting story. Set-up A has more light and a larger measured change. Set-up B has less light and a smaller measured change. It is easy to say, “More light caused the larger change.” But before writing that, a scientific learner checks whether the set-ups were comparable in the other relevant conditions and whether the claimed mechanism fits the concept.

If A and B also began at different temperatures, used different amounts, lasted for different times or differed in another relevant factor, then the observed difference may be real while the proposed cause remains uncertain. The data has not disappeared. What changes is the strength of the conclusion we are allowed to draw.

The Core Reasoning Law

OBSERVE / READ GIVEN INFORMATION → IDENTIFY THE SCIENTIFIC OBJECT OR RELATIONSHIP → DISTINGUISH OBSERVATION FROM INFERENCE → SELECT THE RELEVANT CONCEPT → EXPLAIN THE CAUSAL MECHANISM → CONNECT TO THE QUESTION’S CONDITION → STATE THE OUTCOME → CHECK AGAINST THE EVIDENCE.

For this particular job, place a gate between observation and causal explanation:

RESULT DIFFERENCE → VALID COMPARISON? → RELEVANT CONCEPT? → MECHANISM FITS? → CAUSAL CLAIM WITHIN EVIDENCE LIMIT.

Four Levels of What the Evidence May Support

Level	What you can say	What you should not add yet
1. Observation	The measured outcome in A is greater than in B.	Why it happened.
2. Relationship	Under the tested conditions, the outcome changes with the tested factor.	A universal rule outside those conditions.
3. Causal explanation	The tested factor can explain the outcome because the comparison isolates it and the concept provides a mechanism.	Other untested causes or broader claims.
4. Generalisation	A broader claim may be reasonable only if the evidence and curriculum concept justify it.	“Always”, “all”, or “only” statements unsupported by the evidence.

Worked Example 1: Different Results, Clean Comparison

Original example: Two identical dark-coloured containers each hold 200 mL of water at the same starting temperature. They are placed for the same amount of time at different distances from the same lamp. All other relevant conditions are kept comparable. The nearer container shows a greater temperature increase.

Observation: The nearer container has a greater increase in water temperature.

Comparison check: Starting amount, starting temperature, container type, duration and lamp are comparable; distance is the planned changed condition.

Concept and mechanism: The relevant energy relationship can be used to explain why the condition associated with greater energy transfer produces the larger temperature change.

Evidence-bounded conclusion: Within this investigation, the distance condition is a defensible cause of the difference because it is the intended changed factor and the mechanism is scientifically relevant.

Notice that we still avoid an unlimited claim such as “objects nearer any lamp always become hotter”. The question’s actual conditions matter.

Worked Example 2: Different Results, Confounded Comparison

Original example: Set-up P uses 100 mL of warm water in a metal cup. Set-up Q uses 250 mL of cooler water in a plastic cup. Both are placed near a lamp. After ten minutes, their temperatures differ.

Can the learner say the cup material caused the temperature difference?

No clean causal attribution is available from this comparison because several relevant conditions differ at once: starting temperature, amount of water and cup material. The observation is still valid: the final temperatures are different. But the evidence does not isolate cup material as the cause of that difference.

A scientifically careful learner therefore separates:

What is known: P and Q produced different measured outcomes.
What is not established: which single changed condition caused the difference.
What would improve the investigation: make relevant conditions comparable and vary the factor being tested.

Worked Example 3: A Pattern Is Not Automatically a Mechanism

Original example: Four similar seedlings receive increasing amounts of water over a week. Their final heights form an increasing pattern.

The pattern may support a relationship under the tested conditions. But an explanation still requires the learner to consider whether other relevant factors were comparable and then connect the pattern to the appropriate plant science concept. “The graph goes up, so water causes growth” skips the scientific mechanism and may claim more than the evidence shows.

PSLE Science reasoning should instead ask: What did the investigation vary? What did it measure? What conditions were controlled? What concept links the changed condition to the measured outcome? Does the data support that link across the tested range?

Worked Example 4: Same Result Does Not Mean Same Cause

Suppose two set-ups produce the same temperature reading. That does not prove that the same process or amount of energy transfer occurred in both. Different starting states or different processes can sometimes lead to the same observed value. A result is evidence about the measured outcome; it does not automatically reveal the entire hidden history that produced it.

This is why a learner must keep the scientific object, condition and measured quantity clear instead of treating one number as a complete explanation.

A Difference Test and a Cause Test

Use two passes on the same question.

Pass 1 — Difference Test

What exactly was observed or measured?
Which set-up, object or time point had more, less, faster, slower, earlier or later?
Are you comparing the same quantity and units?
Are you describing the data without adding a cause?

Pass 2 — Cause Test

What factor is proposed as the cause?
Was that factor isolated in a valid comparison?
Were other relevant conditions kept comparable?
Does the relevant science concept provide a mechanism?
Does the mechanism predict the direction of the observed result?
Is the claim limited to what the evidence actually tested?

Observation, Association and Cause Are Different Jobs

Sentence	Reasoning type
“As X increased in the tested set-ups, Y also increased.”	Pattern / relationship in the data
“X caused Y because X was the isolated changed condition and the concept explains how X affects Y.”	Causal explanation
“Y was greater in set-up A than B.”	Observation / comparison
“X always causes Y.”	Broad generalisation requiring stronger support

The Earliest Weak-Link Diagnosis

If a learner overclaims a cause, find the first failed step:

Data reading: Did the learner correctly identify what actually changed?
Object tracking: Are the correct set-ups, quantities and time points being compared?
Variable structure: Can the learner identify the changed factor and measured outcome?
Fair comparison: Can the learner see other relevant differences between set-ups?
Concept selection: Is the correct scientific concept being used?
Mechanism: Can the learner explain how the proposed cause produces the outcome?
Evidence limit: Can the learner stop the claim at the boundary of the data?

Repairing the last sentence is not enough if the student failed at step 3. The earliest weak link determines the teaching move.

Misconception Repairs

Misconception: “If two results are different, the changed variable caused it.”

Repair: Only if the comparison meaningfully isolates that variable and the scientific mechanism fits. Multiple changed conditions weaken single-cause attribution.

Misconception: “A fair test proves the cause completely.”

Repair: A well-designed comparison can strengthen causal interpretation under the tested conditions. It does not make every broader statement automatically true.

Misconception: “A graph trend is the explanation.”

Repair: A trend describes a relationship. An explanation connects that relationship to a relevant concept and mechanism.

Misconception: “If my science fact is true, my cause is correct.”

Repair: A true fact can still be irrelevant to the particular condition or evidence in the question.

How to Read Data Before Explaining It

Before thinking about causes, make a neutral data statement. For a table, compare the correct rows and columns. For a graph, identify the axes, units, direction and relevant interval. For a diagram, identify what differs between set-ups. This protects the learner from choosing a favourite concept first and then forcing the evidence to match it.

A useful habit is to say: “The evidence shows…” before saying “This can be explained by…”. The first sentence anchors the observation. The second sentence earns the explanation.

When the Data Cannot Decide Between Causes

Sometimes more than one explanation remains possible. That is not a failure of science; it is a correct reading of limited evidence. If two proposed causes both fit the observations and the investigation did not distinguish them, the learner should not pretend the data chose one.

Instead, ask what new comparison or observation would separate the explanations. This is scientific inquiry: not merely selecting an answer, but understanding what evidence would be needed to decide.

A Practice Ladder

Describe only: Give data and forbid causal language. The learner states only the measured difference.
Check comparability: Add one hidden difference between set-ups and ask whether a single cause can still be isolated.
Add the concept: Give a clean comparison and ask for the mechanism.
Change the surface: Use a different topic while keeping the same evidence structure.
Competing causes: Give two plausible explanations and ask what additional evidence could distinguish them.
Delayed return: Repeat the decision days later without the original wording.

Unfamiliar Transfer Challenge

Imagine a new apparatus you have never seen. Two chambers produce different gas-volume readings. Chamber A differs from Chamber B in both temperature and the amount of material inside. A statement claims temperature caused the difference.

You do not need to know the apparatus yet to reject the strength of that causal claim. The comparison changes more than one relevant condition, so the result alone cannot isolate temperature as the single cause. After that reasoning gate, you can use the given scientific context to decide what further explanation is possible.

This is transfer: the learner recognises the structure of evidence even when the surface context changes.

Delayed Independent Return Test

Several days after practice, give a fresh question with a new topic and ask the learner to produce two separate statements: (1) what the evidence directly shows, and (2) what causal explanation, if any, the design supports. No hints should be given about variables or fair tests.

A successful return has four receipts: the learner identifies the correct evidence, notices relevant comparison conditions, selects the correct concept, and limits the conclusion appropriately.

Answer-Checking Receipt

Evidence receipt: Did I state what was actually observed or measured?
Comparison receipt: Did I check whether the proposed cause was isolated?
Mechanism receipt: Did I explain how the concept connects cause to outcome?
Condition receipt: Did I tie my explanation to the actual conditions in the question?
Limit receipt: Did I avoid “always”, “only” or universal claims that the evidence does not establish?

Common Traps

Explaining before reading the data.
Treating correlation in a table or graph as automatic proof of cause.
Ignoring unequal starting conditions.
Using a correct science fact that does not explain the observed relationship.
Claiming one cause when several conditions changed.
Writing a universal conclusion from a narrow tested range.
Assuming the examiner expects a fixed phrase rather than evidence-linked reasoning.

Parent and Tutor Teaching Guide

When a child jumps to a cause, do not immediately say “wrong”. Ask: “What does the evidence show before we explain it?” Then ask: “What else is different between these set-ups?” Those two questions reveal whether the weakness is data reading or causal attribution.

A strong teaching exercise uses three versions of the same investigation: one clean comparison, one with two relevant conditions changed, and one with a measurement problem. Ask the learner to rank the strength of the causal conclusion and justify the ranking. This teaches that evidence quality is not all-or-nothing.

Do not turn the lesson into a university-level discussion of causality. For Primary learners, the practical target is narrower: observe carefully, compare validly, use the relevant concept, explain the mechanism, and stop the conclusion where the evidence stops.

Useful Internal Routes

Authoritative References

Quiet Return

Good scientific reasoning is not afraid to say, “The results are different, but this comparison does not yet tell us why.” That sentence is not weaker than a confident guess. It is stronger because it respects the evidence.

Learn to separate the observation from the cause. Then, when the evidence really does support the mechanism, your explanation becomes much more powerful because you have earned it.

Small Group Tutorials