Wait, What? “Repeat the Experiment Three Times” Is Not a Universal Answer
Many Primary learners memorise stock experiment phrases: keep variables the same, repeat three times, use more samples, measure carefully. These ideas can be useful. But in a PSLE Science method-evaluation question, a useful improvement must repair the actual weakness in the method.
If the problem is that two set-ups use different plant species, repeating the same unfair comparison three times does not make it fair. If the problem is a ruler read from the wrong position, adding more plants does not fix the measurement. If the question never tested light intensity, “keep light the same” may be irrelevant.
Method evaluation is not a list of good laboratory habits. It is diagnosis: what weakness threatens this conclusion, why, and what smallest change repairs it?
Quick Answer
Evaluate a PSLE Science experiment by following four moves: identify what the investigation is trying to find out; locate the method feature that weakens the comparison or measurement; explain how that weakness affects the evidence; then propose a specific improvement that directly repairs it.
This guide does not replace canonical pages on variables, fair comparisons, repeated trials, measurement or investigation limits. Those pages own the scientific inquiry components. This page owns the PSLE learner job: how to turn those components into a precise method-evaluation answer.
The Official 2026 PSLE Science Frame
SEAB’s 2026 PSLE Science syllabus states that scientific inquiry includes interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning. This makes method evaluation an explicit scientific-thinking job rather than an optional exam trick.
The 2026 paper can assess these abilities through words, diagrams, tables and graphs. A learner therefore needs to read the method as evidence, not only as a procedure to memorise.
Owned PSLE Science Learning Job
- Identify the investigation question or claim.
- Distinguish changed, measured and controlled conditions.
- Find a method weakness that actually threatens the comparison.
- Explain how the weakness affects the evidence.
- Choose the smallest specific improvement.
- Distinguish fairness, measurement quality, repetition and sample coverage.
- Decide what the data can and cannot support.
- Avoid generic “repeat more” or “keep everything the same” answers.
The Method-Evaluation Chain
WHAT IS BEING TESTED? → WHAT SHOULD CHANGE? → WHAT SHOULD BE MEASURED? → WHAT ELSE MUST BE COMPARABLE? → WHERE CAN THE METHOD MISLEAD US? → HOW DOES THAT AFFECT THE EVIDENCE? → WHAT SPECIFIC CHANGE REPAIRS IT? → DOES THE IMPROVEMENT MATCH THE CLAIM?
Stage 1 — Start With the Claim, Not the Apparatus
Suppose two plants are used to investigate whether the amount of water affects growth. Before inspecting pots, rulers or timing, write the intended relationship:
amount of water → plant growth
Now every method feature can be judged by one question: Does this feature help us isolate or measure that relationship?
Stage 2 — A Fair Comparison Is About Competing Explanations
If Plant A receives more water but is also a different species from Plant B, any growth difference has at least two plausible explanations. The method has not isolated the water condition.
Weak improvement: “Make the experiment fair.”
Stronger improvement: “Use plants of the same type and similar starting size, so the main intended difference between the set-ups is the amount of water given.”
The answer states both the repair and why it matters.
Stage 3 — Measurement Problems Need Measurement Repairs
Suppose a learner measures a shadow length but the ruler begins several centimetres away from the shadow’s starting point. This is not mainly a sample-size problem. The evidence is weakened because the measured value does not match the intended quantity.
- Identify the intended quantity.
- Check where measurement starts and ends.
- Check the unit.
- Check whether the instrument suits the range.
- Check whether the same method is used across set-ups.
Stage 4 — Repetition Solves Some Problems, Not All Problems
Repeating an investigation can help check whether a result is stable and can reveal unusual readings. But repetition does not repair a systematic design mistake.
If every trial uses the wrong measurement point, repeated trials repeat the same error. If every comparison changes two factors at once, repeated trials repeat the same confounding.
Repeat when the weakness is uncertainty or variability. Redesign when the weakness is fairness or validity of the comparison.
Worked Example — Light and Plant Growth
Two identical seedlings are placed for one week. Plant A receives more light than Plant B. Plant A also receives twice as much water.
- Claim being tested: effect of light on growth.
- Changed condition intended: light exposure.
- Additional changed condition: amount of water.
- Problem: a growth difference could be due to light, water, or both.
- Improvement: give both plants the same suitable amount of water while changing only the intended light condition.
Notice that “same amount of water” is not a magic marking phrase. It is relevant because water is a plausible competing condition for this investigation.
Worked Example — Cooling Containers
A learner compares how quickly hot water cools in two cups made of different materials. Cup A contains 200 mL of water at 80°C. Cup B contains 100 mL at 70°C.
If the intended claim is about cup material, both starting volume and starting temperature differ. The experiment cannot cleanly attribute the cooling difference to material. A better method uses equal volumes of water at the same starting temperature in otherwise comparable cups, changing the material while measuring temperature over the same time period.
Stage 5 — Ask Whether the Outcome Matches the Question
An investigation asks which material is the best thermal insulator, but the learner measures only the colour of each container. Even a perfectly fair comparison does not answer the scientific question because the measured outcome is wrong.
Method quality begins with alignment: the measurement must be evidence for the claim.
Stage 6 — Range and Coverage Matter
Suppose an investigation changes a factor across only two very similar values and then makes a broad claim about all possible values. The method may not support such a wide conclusion.
At Primary level, you do not need formal statistics. You do need to ask whether the tested range or examples are sufficient for the claim being made.
Stage 7 — One Anomalous Result Does Not Automatically Destroy the Investigation
If one reading is very different from the rest, first check the method, instrument, recording and conditions. Repeating the measurement may help decide whether the unusual result is stable or accidental. Do not simply erase a result because it is inconvenient.
Science improves explanations by confronting unexpected evidence, not hiding it.
Stage 8 — “More Accurate” Must Say How
Weak: “Use a more accurate method.”
Stronger: “Measure the water volume with the same measuring cylinder at eye level for every set-up.”
Even stronger when relevant: explain why the change reduces a specific source of inconsistent measurement.
Different Weaknesses Need Different Repairs
- Two factors changed: control the unintended factor.
- Outcome measured poorly: improve the measurement procedure or instrument use.
- Result may be unstable: repeat under the same conditions and compare readings.
- Sample is too narrow: include more suitable examples or a wider relevant range.
- Claim is too broad: narrow the conclusion to what was actually tested.
- Procedure differs between set-ups: standardise the relevant steps.
PSLE Answer Shape for Method Evaluation
A useful structure is:
Weakness → Why it matters → Specific improvement.
Example: “The two plants received different amounts of water, so any difference in growth may not be due only to light. Give both plants the same amount of water while changing the light condition.”
Observable Failure Signatures
“I always write ‘repeat three times’.” Ask what problem repetition would solve in this exact method.
“I write ‘keep everything the same’.” Name the specific relevant condition instead.
“I find a mistake but cannot explain why it matters.” Complete the sentence: “Because this also changes ___, the result may not show the effect of ___ alone.”
“My improvement changes the investigation question.” Re-read what relationship the experiment is supposed to test.
Misconception Repair — Fair Does Not Mean Identical
The factor being investigated must differ. If every condition were literally identical, there would be no comparison. A fair test keeps relevant competing conditions comparable while deliberately changing the chosen factor.
Model Limit — Not Every Scientific Investigation Is a Fair Test
Science also uses observations, field studies and other methods where variables cannot be controlled in the same way. At PSLE level, follow the method given and evaluate what that method can support. Do not force every investigation into the same template.
Unfamiliar Transfer Challenge
A learner wants to compare how fast two paper towels absorb water. Towel A is tested with 20 mL of water for 30 seconds. Towel B is tested with 50 mL for 10 seconds.
Without knowing any brand, identify the main problem: both water amount and contact time differ. A fair comparison of absorption rate or amount needs a consistent measurement design. The exact repair depends on what quantity the learner intends to compare.
Delayed Independent Return Test
Two days later, take a new investigation. Without hints, write the claim, intended changed factor, measured outcome, one relevant controlled condition, one possible weakness, the effect of that weakness on the evidence, and one precise improvement. If those seven moves are independent, method evaluation is becoming transferable.
How to Check Your Method-Evaluation Answer
- Did I identify what is actually being tested?
- Is the weakness real and relevant?
- Did I explain how it affects the evidence?
- Does my improvement directly repair that weakness?
- Did I avoid generic phrases?
- Did I accidentally change the investigation question?
- Did I claim more certainty than the method supports?
Parent and Tutor Teaching Guide
Ask one diagnostic question at a time: “What is the experiment trying to find out?”, “What changed?”, “What else changed that should not have?”, “Why is that a problem?” and “What single change would repair it?”
Do not teach a bag of marking phrases. Teach the learner to connect method design to the strength of the conclusion.
Authoritative Reference Basis
- Singapore Examinations and Assessment Board — PSLE Science syllabus for examination from 2026.
- Singapore Ministry of Education — Primary Science Teaching & Learning Syllabus 2023.
- Education Endowment Foundation — A Systematic Review of Approaches to Primary Science Teaching, 2023.
- Primary science inquiry assessment research on planning, analysis, interpretation and inquiry subskills.
- eduKate Singapore Science Learning Library for canonical owners on fair comparisons, repeated investigations and investigation limits.
The Quiet Ending
A strong experiment answer is not a ritual. It is a scientific argument about evidence. Find the claim, locate the weakness, explain why it matters, and repair the method. Once you can do that, you no longer need to guess which “standard improvement” the question wants. You can reason it out.
