How to Spot a “Controlled” Condition That Is Quietly Changing During a PSLE Science Investigation

Wait, What? Writing “Keep Temperature the Same” Does Not Make Temperature Stay the Same

A learner plans an investigation and correctly identifies temperature as a condition that should be kept comparable.

The method begins. Trial 1 starts in a cool room. By Trial 4, sunlight has warmed the table. The apparatus has also warmed from repeated use. The learner still writes “temperature controlled” because that was the plan.

But a planned control is not the same thing as an observed constant condition.

A controlled condition can fail quietly. If a factor that was supposed to stay comparable drifts during the investigation, the evidence may begin answering two questions at once.

This guide teaches how to notice that drift, explain why it weakens the comparison, and repair the method without accidentally changing the original scientific question.

Quick Answer

When a condition is supposed to stay comparable, ask four questions throughout the investigation:

What condition is meant to be controlled?
What could make it change as time or trial number increases?
Could that change also affect the measured outcome?
How can we keep it comparable, reset it, measure it, randomise order where appropriate, or record enough evidence to know whether it drifted?

Use this route:

IDENTIFY THE SCIENTIFIC QUESTION → NAME THE DELIBERATELY CHANGED FACTOR → NAME THE MEASURED OUTCOME → LIST IMPORTANT COMPARABLE CONDITIONS → ASK WHICH ONES CAN DRIFT → LOOK FOR CHANGE WITH TIME OR TRIAL ORDER → TEST WHETHER THAT DRIFT COULD AFFECT THE OUTCOME → REPAIR OR MONITOR THE CONDITION → REPEAT THE FAIR COMPARISON → LIMIT THE CONCLUSION TO THE EVIDENCE.

The Exact PSLE Science Learning Job This Guide Owns

This guide owns one learner job: how a Primary 5 or Primary 6 learner recognises when a condition intended to remain comparable in a PSLE Science investigation is quietly drifting during the experiment or across trials, and evaluates how that drift affects the evidence.

It does not replace the broad guide on variables and fair tests. It does not replace the guide on order-of-testing carryover, where the first test directly changes the state used in the next test. It does not replace the guide on the measuring method interfering with the system. This page owns a different failure: a supposedly controlled background condition changes with time, trial number or environmental drift and becomes an unintended second variable.

Why This Matters in the Current 2026 PSLE Science Frame

For examination from 2026, Standard PSLE Science assesses the 2023 Primary Science syllabus. Official assessment objectives include application of scientific facts, concepts and principles and scientific inquiry, including interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning.

Evaluating a method means more than naming what should be controlled. A learner should understand why comparability matters and how a method can fail to maintain it.

The drift checks in this guide are learning tools, not official SEAB terminology or a fixed marking formula.

First Principle: “Controlled” Describes a Method Job, Not a Magical State

A controlled condition is a factor the investigation tries to keep sufficiently comparable so it does not become another explanation for the measured difference.

That means the method has work to do.

If room temperature matters, the investigation may need the same environment or monitoring.
If starting water temperature matters, every trial may need to begin at the same chosen value rather than simply “use hot water”.
If battery condition matters, repeated use may require checking whether the source weakens.
If specimen moisture matters, samples may need the same starting condition rather than sitting exposed for different lengths of time.

Writing a control in the plan is only the intention. Evidence that the condition remained comparable comes from how the method was carried out.

What Is Drift?

Drift is a gradual or systematic change in a condition that was intended to stay comparable.

It can happen:

with clock time;
with trial number;
as apparatus warms or cools;
as a source weakens;
as a sample dries, settles or changes;
as room light or temperature changes;
as starting conditions are reset less completely;
as the operator’s procedure changes from fatigue or practice.

Not every small change matters. The scientific question is whether the drift is large or relevant enough to affect the measured outcome and therefore weaken the intended comparison.

Worked Example 1 — Starting Water Temperature Drifts Across Trials

Question: How does wrapping material affect the cooling of equal amounts of hot water over ten minutes?

The learner tests Material P first, then Q, then R. A kettle is boiled once. The water used for each later trial is taken from the same kettle as it cools.

The method may claim that “starting temperature is controlled”, but the starting temperature is quietly decreasing from trial to trial.

Now the final temperature can be affected by:

the wrapping material;
the different starting temperature.

The learner can no longer attribute every difference confidently to material type.

Repair: bring each trial to the same specified starting temperature before timing begins, while keeping the amount of water and other relevant conditions comparable.

Worked Example 2 — Apparatus Warms Up

Question: How does one stated condition affect the time taken for a response?

The same lamp or motor runs continuously across repeated trials. Its body warms as trial number increases.

If temperature can affect the measured response, the later trials begin with a different apparatus state.

This is not necessarily a direct “first trial damaged the second trial” carryover. It can be a gradual background drift caused by continued operation.

Repair options: allow the apparatus to return to a comparable starting temperature, monitor temperature, use separate equivalent apparatus where suitable, or redesign the trial order so time-linked warming does not line up perfectly with one test condition.

Worked Example 3 — Battery Condition Changes

An electrical investigation compares three conditions using the same cell or battery over many trials.

If the source weakens with use, later trials may receive a different electrical supply from earlier trials.

Suppose the learner always tests Condition A first and Condition C last. If the battery weakens throughout the session, condition and trial order become entangled.

Scientific problem: a difference between A and C might partly reflect source drift rather than the intended condition.

Repair: keep the source condition comparable, replace or recharge as appropriate, monitor relevant output if the method allows, or design the order so a systematic source drift is less likely to mimic the tested relationship.

Worked Example 4 — Samples Dry While Waiting

Several similar wet cloth samples are prepared at the beginning. They are tested one after another over an hour.

If all samples sit exposed while waiting, later samples may begin drier than earlier ones.

The learner may have written “same starting amount of water” in the plan, but the actual starting amount is drifting before testing even begins.

Repair: prepare each sample immediately before its trial, store samples so starting conditions remain comparable, or measure the relevant starting quantity directly before each test.

Worked Example 5 — Sunlight Changes During a Plant Investigation

A short investigation compares two plant-related conditions near a window. One setup is observed in the morning, another later in the afternoon.

If light level is meant to be comparable but sunlight changes strongly with time, the environment is drifting.

The important insight is not “natural light is bad”. The issue is whether the relevant environmental condition differs systematically between the compared trials.

Repair: test under comparable lighting conditions, measure or standardise the light source where appropriate, or run comparable set-ups at the same time.

Worked Example 6 — Room Temperature Changes Across the Day

An evaporation investigation runs some trials before air-conditioning begins and some after.

Temperature and possibly airflow may change. If those conditions affect evaporation, the measured difference may not belong only to the intended changed variable.

Again, the learner should ask whether the background change is scientifically relevant, not merely whether “the room changed”.

Worked Example 7 — A Control Condition Changes Because the Setup Is Not Reset Fully

A toy car is released down a ramp repeatedly. The starting mark is the same, but the ramp surface becomes dusty or worn during testing.

The learner thinks the surface is controlled because it is the same physical ramp. But its relevant property may not remain the same.

Same object does not always mean same condition.

If surface condition changes enough to affect motion, it has drifted scientifically even though the apparatus identity has not changed.

Same Object Versus Same State

This distinction is central.

What stays the same?	What may still change?
Same battery	charge/output can change
Same lamp	temperature can rise
Same cloth	moisture content can fall
Same room	temperature, light or airflow can change
Same ramp	surface condition can change
Same plant	water status, orientation or other state can change

Control reasoning must follow the relevant state, not just object identity.

How Drift Hides Inside Trial Order

Suppose Condition A is always tested first, B second and C third.

Now imagine a background factor steadily increases with time.

Because A, B and C are always tied to first, second and third position, the background drift can imitate a condition effect.

This is why trial order is not merely administrative. Order can become a hidden structure in the evidence.

How to Tell Random Variation From Systematic Drift

Repeated measurements can vary a little even when the method is sound.

Compare:

Trial sequence	First interpretation
20, 21, 20, 19, 21	small variation around a similar level
20, 23, 26, 29, 32	systematic upward drift deserves investigation
30, 29, 30, 30, 29	stable within a narrow range
10, 10, 14, 14, 18, 18	step-like change may indicate changed conditions or procedure

You do not need advanced statistics at Primary level. Look for whether the change follows time or trial order rather than scattering around one level.

Monitor a Control When “Keep It the Same” Is Hard

Sometimes a condition cannot be made perfectly constant. In real investigations, the useful response may be to monitor it.

Example: room temperature changes slightly across a long investigation. If temperature matters, recording it can reveal whether it drifted enough to threaten the comparison.

The Primary-level principle is:

If an important control might change, either keep it comparable or gather evidence that tells you whether it changed.

Reset Versus Monitor Versus Redesign

Problem	Possible response
Starting temperature differs	Reset to the same chosen starting temperature
Background temperature may drift	Keep environment comparable or monitor it
Battery weakens with use	Maintain comparable source condition or monitor/replace appropriately
Order always aligns with one condition	Change or balance order where suitable so time drift does not mimic condition effect
Sample state changes while waiting	Prepare/reset just before each trial or verify the starting state
Drift cannot be controlled enough	Limit the conclusion or redesign the investigation

When Randomising or Changing Trial Order Can Help

At Primary level, do not turn this into advanced experimental-design jargon. The intuitive idea is enough.

If every A trial happens early and every B trial happens late, a time-related drift can be mistaken for an A-versus-B effect.

Testing in a more balanced order can help separate the intended condition from a background change, provided the method remains fair and each trial can start comparably.

This does not replace resetting important conditions. Order management is one tool, not a magic cure.

Drift Is Different From a Deliberately Changed Variable

Suppose temperature is the variable intentionally changed. Then a temperature difference is not a control failure—it is the scientific test.

The learner must first know the investigation question. Only then can they decide which changes are intended and which are unwanted drift.

Drift Is Different From Natural Variation

Different living specimens may show different outcomes even under comparable conditions. That is not automatically drift.

Drift refers to a condition moving systematically with time or trial sequence. Natural specimen variation is a different evidence problem and may require suitable sampling or repeated specimens.

Drift Is Different From Measurement Interference

If opening a container to measure it changes humidity or temperature, the measurement method itself is altering the system. That has its own owner.

In this guide, the focus is a background condition that changes even though the method intended it to remain comparable.

Drift Is Different From One Unusual Result

One odd measurement may be an unusual result. Drift is more often a pattern across time or trial order.

Do not call one strange value “drift” without evidence of systematic change.

The Control-Drift Audit

What scientific question is being tested?
What factor is deliberately changed?
What outcome is measured?
Which other factors could affect that outcome?
Which of those are intended to stay comparable?
Could any of them change with time or trial number?
What physical process could cause that change?
Is the changed background condition aligned with the test order?
What evidence would reveal the drift?
Can the condition be reset, controlled, monitored or balanced?
After repair, does the method still test the original question?

The Earliest-Weak-Link Diagnostic

Failure signature	Earliest weak link	Repair
“We wrote temperature under controlled variables, so it was controlled.”	Plan confused with actual method state	Ask how temperature was kept or checked
Later trials steadily differ	Trial-order drift not considered	Check conditions against time/trial number
Same apparatus assumed to mean same conditions	Identity confused with state	Track relevant state such as heat, charge or moisture
Every condition tested at one fixed time position	Condition and order confounded	Reset or balance order as appropriate
One odd value called drift	Systematic pattern not established	Look for repeated directional change
All small fluctuations treated as fatal	Perfect constancy demanded	Judge whether variation is relevant to the outcome
Repair changes the variable being tested	Method improvement drifts into new question	Preserve the original changed factor and outcome

Misconception Repair — “Controlled Means Exactly Constant”

In real measurements, perfect constancy may be impossible. The scientific goal is to keep relevant alternative causes comparable enough that the intended comparison remains interpretable.

Do not turn small irrelevant fluctuation into an imaginary failure. Focus on whether the condition could materially affect the outcome.

Misconception Repair — “If the Same Equipment Is Used, the Control Is Fine”

The same equipment can change state. A motor warms, a battery weakens, a wet material dries and a lamp heats.

Scientific comparability follows the relevant property, not merely object identity.

Misconception Repair — “Repeating More Times Fixes Drift”

If every repeat is carried out under a progressively drifting condition, more repeats can reveal the problem—but they do not automatically remove it.

Repair the source of drift or design the comparison so drift does not systematically favour one condition.

Misconception Repair — “Average the Results and the Drift Disappears”

An average can hide systematic change. If results rise steadily across trials, averaging them into one number erases evidence that the condition may be drifting.

Misconception Repair — “Every Change With Time Is Drift”

If time is the deliberately changed variable or the investigation is measuring a process over time, the change may be exactly what the experiment is designed to observe.

Drift is an unintended change in a condition meant to remain comparable.

How This Appears in a Method-Evaluation Question

A useful reasoning shape is:

The condition ______ was intended to stay the same, but it may ______ as the trials continue. Because ______ can also affect the measured ______, the comparison may not isolate ______. Improve the method by ______ while keeping the original tested factor unchanged.

This is a reasoning scaffold, not a compulsory PSLE sentence.

How This Appears in Data

Sometimes the question does not announce the drift. The data themselves may suggest it.

Trial	Condition being tested	Outcome
1	A	10
2	A	12
3	A	14
4	A	16

If nothing in the planned condition changed but the outcome rises steadily, ask whether a hidden background condition is changing with trial number.

Do not assume drift immediately. It is one hypothesis to investigate.

Practice Sequence — Make the Invisible Control Visible

Start with a perfectly described fair comparison and name the controls.
Add one realistic drift: starting temperature falls across trials.
Ask which measured outcome could be affected.
Add a different drift: apparatus warms with use.
Separate it from one isolated unusual result.
Use a living sample and distinguish natural variation from environmental drift.
Give a method where the test order aligns with time and ask how to reduce that problem.
Give a case where the control varies slightly but not enough to matter scientifically.
Return later with an unfamiliar investigation and no hint that drift exists.

Unfamiliar Transfer Challenge

A fictional sensor is used to test Conditions P, Q and R. The sensor is powered continuously and slowly warms. P is always tested first, Q second and R last. The measured output also rises from P to Q to R.

Can you conclude R causes the greatest output?

Not confidently yet. Condition and test order are aligned, and sensor temperature is drifting upward. If sensor temperature can affect output, it provides another possible explanation.

What would improve the evidence?

keep or reset sensor temperature comparably;
monitor the sensor temperature;
use a balanced trial order where appropriate;
repeat under a method that separates the tested condition from warming over time.

The transferable habit is to ask whether trial number itself has become a hidden variable.

Delayed Independent Return

Several days later, inspect a fresh investigation and answer without notes:

What is deliberately changed?
What should stay comparable?
Which “controlled” condition could drift?
What process could cause the drift?
Would that drift affect the measured outcome?
Does it align with time or trial order?
Can it be reset or monitored?
Would a different order help?
Does the repair preserve the original question?
What conclusion remains justified if the drift cannot be ruled out?

The Control-Drift Receipt

I know the scientific question.
I know the deliberately changed factor.
I know the measured outcome.
I can name important conditions that should remain comparable.
I distinguish a planned control from evidence that it stayed controlled.
I can identify states that may change even when the same apparatus is used.
I check for directional change with time or trial number.
I separate drift from random variation and one unusual result.
I can explain how drift would affect the measured outcome.
My repair does not change the scientific question.
I limit the conclusion if the drift cannot be ruled out.

Useful Internal Routes

Parent and Tutor Teaching Guide

Children often learn fair-test vocabulary as a static table: changed variable, measured variable, controlled variables. To move beyond labels, make a controlled condition fail during a thought experiment.

Say: “We wrote ‘same starting temperature’. But Trial 1 starts at 80°C, Trial 2 at 76°C and Trial 3 at 72°C. Is it still controlled just because the worksheet says so?”

Then ask what mechanism could make the condition drift and how the drift could affect the measured outcome.

Use several forms of drift—warming apparatus, weakening source, drying sample, changing daylight—so the learner does not memorise one example. The underlying idea is always the same: an intended control must remain scientifically comparable in practice.

Finally, show a small fluctuation that probably does not matter and ask the learner not to over-diagnose. Strong scientific evaluation is selective: it cares about alternative causes that can meaningfully affect the evidence.

Authoritative and Research References

The examples in this guide illustrate general experimental-comparison reasoning. They do not create a universal PSLE requirement to randomise trials, monitor every possible variable or use one fixed method-improvement sentence.

The Quiet Ending

A fair test is not made fair by the words written in the planning box.

It stays fair only while the important alternative causes stay under control.

So when an investigation runs for a while, ask one more question:

What was supposed to stay the same—and did it really?

Small Group Tutorials