How to Tell When a PSLE Science Investigation Has No Valid Comparison Yet

Wait, What? An investigation can have real observations and still have no evidence for the claim it is trying to make.

A learner measures the temperature of one cup of water after ten minutes and says, “This material keeps water warm.” The number may be accurate. The observation may be real. But what is it being compared with?

Science becomes persuasive through comparison. A single value can describe a state. It cannot automatically show an effect, a difference or a cause.

Quick Answer

Before writing a conclusion, ask: Compared with what? Then check whether that comparison is scientifically valid for the question. A useful comparison may be before-versus-after for the same object, one matched set-up versus another, one condition versus a reference condition, or several values across a tested range.

If no such comparison exists yet, the correct scientific move is not to invent one. State what the observation shows, explain what cannot yet be concluded, and identify the smallest additional comparison that would answer the question.

The PSLE Science Learning Job This Guide Owns

This guide owns one learner job: recognising when a PSLE Science investigation has observations but still lacks the comparison needed to support its intended conclusion. It does not own the underlying concept being investigated. It teaches how to match scientific question, comparison, evidence and conclusion.

The current 2026 PSLE Science syllabus assesses the 2023 Primary Science syllabus and includes scientific inquiry through interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning. This is exactly why “there is a result” and “the result answers the question” are different ideas.

A Result Is Not Automatically a Comparison

Suppose a plant grows 6 cm in one week. That tells you something about the plant. It does not by itself show that a particular amount of light caused the growth. To test the effect of light, you need a comparison in which the relevant light condition differs while other important conditions are kept appropriately comparable.

Similarly, if a toy car travels 80 cm on one surface, that measurement does not yet show that the surface made it travel farther or shorter. You need a reference: another surface, a previous state, or another scientifically matched condition.

ONE OBSERVATION CAN DESCRIBE. A COMPARISON CAN TEST A DIFFERENCE.

Four Common Comparison Structures

Comparison structure	When it is useful	What it can support
Before versus after	The same object or system is measured before and after a change	Whether the measured state changed over that interval
Set-up A versus Set-up B	Two matched set-ups differ in one intended condition	Whether the outcome differs under those conditions
Reference or control comparison	A tested condition is compared with a meaningful reference	Whether the tested condition produces a difference relative to that reference
Several tested values	A variable is tested across a range	Whether a pattern or relationship appears across the tested range

Worked Example 1: One Measurement, Big Conclusion

A learner puts a metal spoon in warm water for five minutes. The spoon is warm at the end. The learner concludes, “Metal conducts heat better than plastic.”

The final warmth of the metal spoon is an observation. But the conclusion compares metal with plastic even though no plastic spoon was tested in the described investigation. The learner has imported a comparison that the evidence did not make.

A stronger response is: “The spoon became warm under these conditions. This result alone does not compare metal with plastic. A matched plastic spoon tested under comparable conditions would provide the needed comparison.”

Worked Example 2: Before-After Can Be Enough for Some Questions

A spring is measured before a load is added and again after the load is added. If the question is simply whether its length changes after the load is applied, the before-after measurements may provide the relevant comparison.

But if the question asks whether one load produces more extension than another load, the investigation needs more than one load condition. The correct comparison depends on the exact question.

Worked Example 3: A Comparison Exists, but It Is Not Valid Yet

Two cups of water are compared. Cup A contains 100 mL of water in a metal cup. Cup B contains 300 mL in a plastic cup. After ten minutes, their temperatures differ.

There is a numerical comparison. But if the scientific question is about cup material, the comparison is not clean because water amount also differs. The learner must distinguish having two results from having a valid comparison for the intended causal claim.

Worked Example 4: A Reference Is Missing

A seedling is given a treatment and later grows 4 cm. The learner says the treatment increased growth. But no untreated seedling and no pre-treatment growth rate are provided.

The observation shows growth occurred. It does not by itself show the treatment increased growth. The smallest useful next comparison might be a similar untreated seedling under otherwise comparable conditions.

The Comparison-First Protocol

State the scientific question. What relationship or difference is being investigated?
Name the measured outcome. What evidence is actually collected?
Ask “compared with what?” Identify the reference required by the claim.
Check comparability. Are the compared cases aligned in object, time, quantity and important conditions?
Check variable isolation. If the claim is causal, did other relevant differences change too?
State what the existing data show. Do not jump beyond them.
Identify the missing comparison if needed. What smallest additional case would answer the question?

Failure Signatures

“The object became warm, so this material is the best conductor.”
“The plant grew, so the fertiliser increased growth.”
“The bulb lit, so this circuit is brighter than the other circuit,” when no other circuit was tested.
Two set-ups are compared even though they differ in several important conditions.
A final value is treated as evidence of change without a starting value or reference.
A learner adds an imaginary control group that is not in the question.
A conclusion uses the word “caused” when the investigation only records one unpaired observation.

Earliest Weak-Link Diagnosis

When a learner overclaims from one result, ask three questions:

What exactly did we observe?
What are you comparing that observation with?
Was that comparison actually measured or are you imagining it?

If the learner cannot answer Question 2, the missing idea is comparison structure. If a comparison exists but is scientifically unfair, the problem is variable control. If both are correct but the causal explanation is wrong, then repair the concept or mechanism.

Misconception Repair

“Every investigation needs a control set-up.” No. The needed comparison depends on the question. Some questions use before-after measurements, some compare matched set-ups, and some examine a trend across several values.

“Two numbers automatically form a valid comparison.” No. They must describe comparable scientific quantities under relevant conditions.

“If a result is surprising, compare it with what I expected.” Your prediction is useful, but expectation is not a replacement for measured reference evidence.

“No valid comparison means the experiment is useless.” Not necessarily. The observation may still describe the system and may help design the next investigation. It simply cannot support the stronger comparative claim yet.

Comparison and Causal Reasoning

To argue that Condition X affected Outcome Y, you need evidence that distinguishes what happens under X from an appropriate alternative or reference. This is why control-of-variables reasoning matters. A controlled comparison reduces the number of competing explanations.

Research on elementary scientific reasoning shows that identifying and evaluating controlled versus confounded comparisons is a real learning challenge. It is not enough to know the vocabulary “changed variable” and “controlled variable”. The learner has to recognise what evidence a comparison actually creates.

Original Practice: Does the Comparison Exist?

Case A: One ball is dropped from a height and takes 1.2 seconds to reach the ground. Can you conclude it falls faster than a second ball? No second ball or earlier reference has been measured.

Case B: The same ice cube is measured before and after ten minutes. Can you conclude its mass changed over the interval? Yes, if mass was measured at both times and the question is about change in that same cube.

Case C: Two plants receive different light levels but also different amounts of water. Can you conclude light caused the growth difference? The comparison is confounded because another relevant condition differs.

Case D: Four identical set-ups test increasing amounts of one condition while the measured outcome is recorded. Can you look for a trend? Yes, if the set-ups are otherwise scientifically comparable and the same outcome is measured.

Retrieval and Transfer Sequence

Round 1: sort ten short scenarios into “description only”, “valid before-after comparison”, “valid set-up comparison”, “confounded comparison”, or “needs another reference”.
Round 2: for each weak investigation, add the smallest missing comparison without changing the scientific question.
Round 3: change topic and repeat—from heat to plants, forces, electricity or materials.
Round 4: take one graph and explain what each comparison actually supports.
Delayed return: several days later, diagnose a new investigation without the labels.

Unfamiliar Transfer Challenge

A new coating is placed on one container. After fifteen minutes, the water inside is 42°C. The learner says, “The coating reduces heat loss.” What is missing?

You need a relevant comparison—perhaps an otherwise matched uncoated container starting under the same relevant conditions. The 42°C result is real, but without a reference it cannot tell you whether the coating changed the outcome.

Answer-Checking Receipt

I can state the exact scientific question.
I know what outcome was measured.
I can name the two states, set-ups or conditions being compared.
I know whether that comparison was actually measured.
I have checked that the compared cases are scientifically comparable.
I have not turned one result into a claim about an unmeasured alternative.
If the comparison is missing, I can identify the smallest useful next comparison.
My conclusion stays within what the current evidence supports.

Parent and Tutor Teaching Guide

When a child makes a strong claim from one result, use one question: “Compared with what?” Do not answer it for them. Let them search the method, diagram or data for the reference.

Then ask whether the comparison is fair for the scientific job. This two-step routine—find the comparison, then judge the comparison—prevents children from memorising “control variables” without understanding why they matter.

A useful teaching contrast is to show three investigations: one with no comparison, one with a comparison that changes too many conditions, and one with a clean comparison. Ask the child to explain why they support different strengths of conclusion.

Useful Internal Routes

Authoritative and Research References

Quiet Return

Science does not become stronger merely because a number has been recorded. The number becomes evidence for a claim only when the comparison answers the question. Before you conclude, ask the simplest scientific question of all: Compared with what?

Small Group Tutorials