Wait, What? A Fair Test Does Not Mean Everything Must Be the Same
If everything were exactly the same, there would be no test. A scientific investigation deliberately changes something. The real challenge is deciding which other conditions must stay comparable so that the measured result can be meaningfully linked to the factor being investigated.
That is why “keep everything the same” is not a complete scientific answer. It is impossible in practice, and it hides the important reasoning: which differences could change the measured outcome?
Quick Answer
First identify the condition deliberately changed and the outcome measured. Then examine the other conditions one by one. A condition needs to be controlled when a difference in that condition could plausibly affect the measured outcome and therefore create another explanation for the result.
The working chain is: QUESTION → CHANGED CONDITION → MEASURED OUTCOME → POSSIBLE OTHER CAUSES → RELEVANT CONTROLS → FAIR COMPARISON → EVIDENCE-BOUNDED CONCLUSION.
Owned PSLE Science Learning Job
This guide owns one specific learner decision: deciding which conditions are relevant enough to control in a PSLE Science fair comparison. It does not re-teach the whole idea of variables and fair tests. The broader fair-test guides remain the canonical owners of the general concept.
This page focuses on the harder step that often comes later: once you know there are controlled conditions, how do you decide which ones actually matter for the scientific question?
Why This Matters in the Current PSLE Science Frame
The 2026 PSLE Science examination assesses the 2023 Primary Science syllabus. SEAB includes interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning within scientific inquiry. Evaluating whether a comparison is fair therefore requires more than repeating a memorised phrase.
A learner must connect method to evidence: if another changing condition can also affect the measured outcome, the investigation may no longer isolate the relationship it claims to test.
The Difference-Maker Test
For every condition besides the deliberately changed one, ask:
If this condition differed between the set-ups, could it reasonably change the measured outcome?
If yes, it is a candidate controlled condition. If no, keeping it identical may be unnecessary for the stated scientific job.
This test prevents two opposite errors: controlling too little and controlling meaningless detail.
Worked Example 1: Evaporation From Two Containers
An original investigation asks how exposed surface area affects the amount of water lost over one hour.
The deliberately changed condition is exposed surface area. The measured outcome is amount of water lost.
Now consider other conditions:
- Starting amount of water: relevant. Different starting amounts could affect the comparison and the available water surface or depth.
- Observation time: relevant. A container observed for two hours has more time to lose water than one observed for one hour.
- Surrounding temperature: relevant. Temperature can affect evaporation.
- Label colour on the container: usually not relevant to this measured outcome unless the label changes a physical condition in the set-up.
The important move is not to memorise this exact list. It is to explain why each condition could or could not become another difference-maker.
Worked Example 2: Plant Growth
A learner investigates how the amount of light affects plant growth over a stated period.
If one plant also receives much more water, water becomes a competing explanation for any growth difference. Water therefore needs to be kept suitably comparable for this test.
What about pot colour? Do not automatically list it. Ask whether pot colour is expected to alter the measured outcome under the particular conditions. If pots are placed under intense sunlight and colour materially changes root-zone temperature, it could matter. In an ordinary indoor classroom set-up where that pathway is not relevant, it may not deserve attention.
Scientific control is about causal relevance, not collecting sameness for its own sake.
Worked Example 3: Comparing Materials
An investigation compares how well different materials keep warm water warm. The material type is deliberately changed. The measured outcome is water temperature after the same period.
Relevant controls may include starting water temperature, amount of water, observation time and comparable container geometry if these can affect heat transfer. Writing “same table” is not useful unless table location is connected to a meaningful environmental difference such as sunlight or airflow.
The control must protect the scientific comparison, not merely make the apparatus look symmetrical.
Controlled Does Not Mean Frozen Forever
Sometimes a controlled condition changes naturally during the investigation. The scientific requirement is often that it changes comparably across set-ups, not that its numerical value never changes.
For example, room temperature may drift slightly while several set-ups are tested together. The learner should ask whether all set-ups experience the same surrounding conditions or whether one is systematically warmer, sunnier or more exposed to moving air.
This is different from a condition that is supposed to remain controlled but quietly drifts differently across trials.
Relevant Control vs Convenient Control
| Condition | Question to ask | Control priority |
|---|---|---|
| Starting amount | Could a different amount change the measured outcome? | Often high |
| Time | Would more or less time change the result? | Often high |
| Temperature | Could temperature affect the process? | Depends on process; often high |
| Position | Does position change light, heat, airflow or another relevant condition? | Depends on set-up |
| Decorative appearance | Does this feature have a plausible pathway to the measured outcome? | Usually low unless it changes a physical condition |
The Alternative-Explanation Test
Imagine the result has already happened. Then ask:
Could someone explain the result using this other condition instead of the condition I intended to test?
If yes, the uncontrolled condition threatens the interpretation.
Example: one seedling gets more light and also more water. If it grows more, light and water are both plausible explanations. The investigation does not isolate light cleanly.
A Control Can Be Relevant Without Being Perfectly Identical
Real specimens are not exact copies. Two leaves may differ slightly. Two seeds may not contain identical stored resources. Two pieces of material may have tiny manufacturing differences.
At Primary level, the learner should aim for scientifically suitable comparability: use similar specimens where relevant, repeat trials or use more specimens when natural variation matters, and keep the conclusion within what the evidence supports.
Do not pretend biological variation disappears because a worksheet says “same type of plant”.
Worked Example 4: One Irrelevant Difference
Two identical transparent containers hold equal amounts of water at the same starting temperature. One is wrapped in Material P and the other in Material Q. Both are placed side by side for the same time. One container has a blue number sticker and the other a red number sticker.
Should sticker colour be listed as a controlled variable?
Not automatically. If the tiny sticker has no meaningful effect on heat transfer in the stated set-up, its colour is not part of the scientific comparison. Listing it can distract from the controls that actually protect the evidence.
Worked Example 5: A Quiet Confounder
Two containers are used to compare exposed surface area and evaporation. The wide container is placed beside a fan while the narrow container is sheltered from moving air.
Air movement can affect how quickly water vapour is carried away. Now exposed surface area and airflow differ together. Even if the learner carefully controls starting water amount and time, the comparison is weakened.
The earliest weak link is not “forgot to keep everything same”. It is: another condition with a plausible effect on the measured outcome changed systematically between the set-ups.
What About Conditions the Question Does Not Mention?
Do not invent unlimited hidden problems. School questions give a bounded set of information. Reason from the conditions shown or stated and from relevant Primary Science knowledge.
If a method explicitly states that all other relevant conditions were kept the same, use that information. If the question shows a clear difference that can affect the outcome, evaluate it. Do not turn every investigation into an endless hunt for imaginary confounders.
Failure Signatures and Earliest Weak-Link Diagnosis
| Failure signature | Weak link | Repair |
|---|---|---|
| “Keep everything the same.” | No causal selection | Name the measured outcome and justify each relevant control. |
| You list ten decorative details. | Relevance discrimination | Ask whether each detail has a plausible pathway to the outcome. |
| You miss water, temperature or time when they differ. | Alternative-explanation check | Ask what else could produce the observed result. |
| You think controlled means numerically unchanging. | Comparability concept | Distinguish “kept comparable across set-ups” from “never changes with time”. |
| You demand identical living specimens. | Model limit | Use similar specimens and acknowledge natural variation. |
Misconception Repair: The Changed Variable Is Supposed to Be Different
Students sometimes write that the deliberately changed factor should also be kept the same “to make it fair”. That removes the investigation itself.
A fair comparison deliberately changes the factor of interest while protecting against other plausible difference-makers.
Misconception Repair: More Controls Are Not Automatically Better
Trying to control everything can make an investigation impractical without improving the evidence. Scientific method is not a contest to produce the longest list. It is a reasoning problem about what matters.
The Five-Step Control-Selection Protocol
- 1. State the scientific question.
- 2. Name the deliberately changed condition and measured outcome.
- 3. List only realistic other conditions that differ or could drift.
- 4. For each, ask whether it could change the measured outcome.
- 5. Control the relevant ones and keep the conclusion limited to the resulting fair comparison.
Practice Sequence
- Identify the changed and measured variables in a simple investigation.
- Choose one obviously relevant control and explain its pathway to the outcome.
- Reject one irrelevant decorative difference.
- Find one hidden competing cause in a flawed investigation.
- Repair the method without changing the original scientific question.
- Repeat the reasoning in an unfamiliar topic so the control choice is not memorised by chapter.
Unfamiliar Transfer Challenge
A fictional material changes colour depending on humidity. A learner wants to test whether temperature affects the time taken for the colour change. Two samples are placed at different temperatures, but one is also exposed to much higher humidity.
You do not need prior knowledge of the fictional material because the question itself tells you humidity affects the colour change. Humidity therefore becomes a relevant controlled condition. The learner can identify the confound from the supplied rule.
Delayed Independent Return Test
Several days later, give a new investigation from a different theme. Ask the learner to identify two relevant controls, one irrelevant difference and the reason for each decision.
A strong answer should not merely name controls. It should connect each selected control to the measured outcome.
Fair-Test Control Receipt
- What is deliberately changed?
- What is measured?
- Which other conditions could affect that measurement?
- Which conditions actually differ between the set-ups?
- Could any of them provide another explanation?
- Have I confused an irrelevant difference with a relevant control?
- Does the repaired comparison still answer the original question?
Parent and Tutor Teaching Guide
When a child says “same amount, same time, same place” by habit, ask, “Why does that one matter for the result we are measuring?” The explanation reveals whether the control has scientific meaning.
Use bad-investigation contrasts. Include one obvious confound, one irrelevant visual difference and one condition that is already controlled. Ask the learner to distinguish them rather than simply list everything visible.
Do not reward an impossibly long control list. Reward a small set of scientifically justified controls that protects the comparison.
Useful Internal Routes
- How to Decode Variables and Fair Tests in PSLE Science Questions
- How to Tell a Controlled Variable From a Control Set-Up in PSLE Science
- How to Spot a Controlled Condition That Is Quietly Changing During a PSLE Science Investigation
- How to Improve a PSLE Science Investigation Without Changing the Scientific Question
Authoritative References and Evidence Boundary
- SEAB — PSLE Science syllabus for examination from 2026
- MOE — Primary Science Teaching & Learning Syllabus 2023
- EEF — Improving Primary Science
The relevance test in this guide is an educational scaffold, not an official PSLE marking formula. Real experimental design can require deeper statistical, measurement and causal controls. At Primary level, the essential job is to identify plausible competing conditions and keep the conclusion within the evidence.
The Quiet Return
A fair test is not a room in which every detail has been frozen.
It is a comparison in which the important alternative explanations have been controlled well enough for the evidence to answer the question. Keep the conditions that matter steady. Let the condition you are testing change. Then let the result speak only as far as the comparison allows.
