How to Keep a PSLE Science Investigation Consistent From Question to Variables to Results to Conclusion

Wait, What? An Investigation Can Quietly Change Its Question Halfway Through

A student begins with a sensible question: “How does the amount of light affect the rate of water loss from a plant?”

Then the method changes both light and temperature. The results table records only leaf colour. The conclusion says, “Plants grow better with more light.”

Every sentence sounds scientific. The experiment is still broken.

A strong investigation keeps one scientific relationship alive from the first question to the final conclusion.

The challenge is not merely to know the names “changed variable”, “measured variable” and “controlled variable”. It is to keep those roles aligned so the method produces the evidence the original question actually needs.

Quick Answer

Before planning an investigation, rewrite the question as one relationship: How does X affect Y under stated conditions? Let X determine the deliberately changed condition. Let Y determine what you measure or observe. Keep other relevant conditions comparable. Build the results table around X and Y. Then write a conclusion that describes only the relationship the data can support.

The consistency chain is:

QUESTION → RELATIONSHIP → CHANGED CONDITION → MEASURED OUTCOME → CONTROLLED CONDITIONS → PROCEDURE → RESULTS → COMPARISON → CONCLUSION → CHECK BACK AGAINST QUESTION.

Owned PSLE Science Learning Job

This guide owns how a Primary 5/6 learner keeps an investigation scientifically consistent from beginning to end. It does not replace the guide on decoding variables and fair tests, the guide on planning an investigation, or the guide on improving a method without changing the question.

The distinctive job here is alignment: proving that the question, variables, measurements, table and conclusion all refer to the same scientific relationship.

The Current PSLE Science Frame

For examination from 2026, SEAB states that PSLE Science assesses attainment in the 2023 Primary Science syllabus. The assessment objectives include applying scientific inquiry, interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning.

That makes consistency an inquiry skill rather than an exam trick. A method is useful only if the evidence it produces answers the question that was asked.

The official assessment specification is available from the Singapore Examinations and Assessment Board. This guide does not invent an official marking phrase or one compulsory investigation template.

Start by Writing the Relationship, Not the Apparatus

Suppose the question is: “How does the distance between a lamp and a screen affect the size of a shadow when the object remains in the same position?”

The relationship can be written as:

distance between lamp and screen → shadow size

That one line tells you what must remain alive through the investigation. If the method later changes object size instead of lamp-screen distance, the question has drifted. If the table records only brightness, the measurement has drifted. If the conclusion discusses shadow darkness, the conclusion has drifted.

Writing apparatus first often hides these errors because equipment can look impressive while the scientific relationship remains unclear.

The Alignment Map

Investigation part	Question it must answer	Alignment receipt
Scientific question	What relationship are we investigating?	X affects Y
Changed condition	What part of X will we vary deliberately?	Values or categories of X
Measured outcome	What evidence will represent Y?	A measurable or observable Y
Controlled conditions	What other relevant influences must stay comparable?	Named competing conditions
Procedure	Does each step create a fair test of X and Y?	Same method except intended difference
Results	Were X and Y recorded clearly?	Condition paired with outcome
Conclusion	What relationship did the data support?	Return to X and Y only

If a row cannot be linked back to the first row, inspect it. It may be useful background information, but it should not quietly become the new owner of the investigation.

Worked Example 1 — A Clean Investigation

Question: How does exposed water surface area affect the amount of water lost by evaporation in 30 minutes?

Two containers hold the same starting mass of water. Container P exposes 80 cm² of water surface; Q exposes 40 cm². They are placed beside each other for the same time under comparable surroundings. Remaining water mass is measured.

Link	Decision
Question relationship	Exposed surface area → water lost by evaporation
Changed condition	Exposed surface area
Measured outcome	Change in water mass after 30 min
Important controls	Starting water mass, time, surrounding conditions
Results structure	Surface area paired with starting/final mass and calculated loss
Conclusion	Compare water loss at the two tested surface areas

If P loses 12 g and Q loses 6 g, the conclusion can state that the larger exposed surface area was associated with greater water loss by evaporation under the tested conditions. It should not suddenly become “larger containers always cause faster evaporation” unless container size itself was the tested variable and the evidence supports that claim.

Worked Example 2 — The Method Quietly Changes the Question

Question: How does the number of layers of cloth affect the time taken for ice to melt?

A learner wraps one ice cube in one layer of cotton cloth and another cube in three layers of wool cloth.

The intended changed condition is number of layers. But material type has also changed. The comparison now contains at least two relevant differences.

If the learner concludes “three layers make ice melt more slowly”, the conclusion pretends that layer number was isolated even though cloth material may also affect heat transfer.

The repair is not “write a better conclusion” first. The earliest weak link is the method. Use the same cloth material and vary only the layer number if the purpose is to isolate the effect of layers.

Worked Example 3 — The Measurement Quietly Changes the Question

Question: How does the amount of water given affect the increase in height of seedlings over seven days?

The learner carefully changes water amount and controls the other relevant conditions, but records only the final height of each seedling.

That is not always enough to determine increase in height. If the seedlings began at different heights, final height and amount of growth are different quantities.

The measurement job must match the wording of the question. To compare increase, record a starting height and a later height so the change can be derived.

This is why “measure height” can be too vague. Ask: final height, height increase, height at each time point, or rate of increase?

Worked Example 4 — The Results Table Quietly Changes the Relationship

Question: How does temperature affect the time taken for a tablet to dissolve in the same volume of water?

The procedure is fair, but the table is organised like this:

Trial	Water temperature	Time taken to dissolve	Final water temperature
1	20°C	180 s	20°C
2	40°C	95 s	38°C
3	60°C	54 s	55°C

The extra final-temperature column might be useful for another purpose, but the original relationship is starting/test temperature versus dissolving time. A learner who becomes fascinated by the final-temperature changes can accidentally write a conclusion about cooling instead of dissolving.

Extra data are not automatically bad. The discipline is to keep the decisive comparison attached to the original question.

Worked Example 5 — The Conclusion Becomes Larger Than the Investigation

A learner tests three lengths of a pendulum-like classroom model and records the time for a fixed number of swings. The result shows longer strings took more time under the tested set-up.

A bounded conclusion would describe the relationship across the tested lengths. An oversized conclusion such as “length controls all motion” leaves the investigation completely.

Conclusions should return to the question, not graduate into a universal law merely because the pattern looks neat.

Question Words That Change the Measurement Job

Question wording	What the result may need	Common drift
How much remains?	Final amount	Calculating change when final amount is requested
How much changed?	Starting and final values	Comparing final values only
How fast?	Change attached to time	Comparing raw change over unequal times
When does it begin?	Repeated observations with a defined onset criterion	One final measurement
Which condition gives the greatest outcome?	Comparable outcomes across tested conditions	Claiming an untested true optimum
What is the effect of X?	Comparison isolating X as far as practical	Changing X and another relevant condition

Use the Question as a Checksum

A checksum is a simple way of checking that information has not quietly changed. In investigation work, the original scientific question can play that role.

At each stage, ask:

Does this variable still belong to the original relationship?
Does this measurement answer the exact outcome named?
Does this controlled condition protect the comparison from a competing explanation?
Does this table make the tested relationship visible?
Does this conclusion answer the original question rather than a nearby one?

If the answer becomes “not really”, stop there. The investigation has drifted.

The Difference Between a Useful Extra Measurement and Ownership Drift

Sometimes an investigation records more than one useful outcome. That can be scientifically valuable. The problem begins when the extra measurement silently replaces the original outcome.

For example, a plant investigation might record height, number of leaves and mass. If the question asks about height increase, leaf number can be supporting context but should not become the final answer to a different question.

Keep the dominant measured outcome visible. If a second outcome becomes important enough to deserve its own conclusion, it may represent a second scientific question rather than an extension of the first.

Observation and Conclusion Must Not Swap Jobs

“P lost 10 g” is a measured result. “Greater exposed surface area led to greater water loss under the tested conditions” is a conclusion about the relationship.

If you write the conclusion directly into the results table, you hide the evidence. If you copy the table values into the conclusion without stating the relationship, you never finish the reasoning.

Keep the chain visible: result → comparison → conclusion.

Find the Earliest Weak Link

Failure signature	Earliest weak link	Repair
You cannot state X and Y.	Question interpretation	Rewrite the question as one relationship.
You change two relevant conditions.	Method design	Protect the intended changed variable.
You measure an easy quantity rather than the requested outcome.	Measurement alignment	Ask what observation would answer Y directly or defensibly.
You compare final values when the question asks about change.	Quantity alignment	Add or use starting values.
Your table hides which result belongs to which condition.	Recording structure	Pair each X condition with its Y result and unit.
You explain why before describing what happened.	Result/conclusion separation	State the evidence first.
Your conclusion introduces a new factor.	Conclusion drift	Return to the original X–Y relationship.

Misconception Repair — “Keep Everything the Same” Is Too Crude

A fair comparison does not literally require every property in the universe to be identical. It requires relevant competing influences to be controlled well enough for the intended comparison.

If you are testing exposed surface area, container shape may necessarily differ because it creates the different surface areas. The important question is which other features could independently affect the measured outcome and therefore need control or careful interpretation.

Memorising “same everything except one” can be useful as a starting idea, but scientific judgement is needed to identify what actually matters.

Model Limit — A Perfectly Aligned Investigation Can Still Have Weak Evidence

Alignment is necessary, but it is not sufficient for a strong investigation.

You can ask the right question, change the right variable and measure the right outcome, yet still use an unsuitable instrument, too few specimens, an unhelpful range, inconsistent timing or a method that disturbs the system.

That is why alignment should be followed by evidence-quality checks. This page protects the route; other investigation guides inspect the strength of each component.

The One-Line Investigation Receipt

Before doing or analysing an investigation, complete this sentence:

We change ______, measure ______, keep ______ comparable, and use the results to decide whether ______ affects ______ under these tested conditions.

If the blanks cannot be filled without changing the meaning of the original question, the plan is not ready.

Practice Sequence

Practice 1 — Question only: identify X and Y. Do not design the method yet.

Practice 2 — Method audit: inspect a short procedure and circle every step that changes X, measures Y or controls a competing condition.

Practice 3 — Results audit: check whether the table contains the quantities needed to answer the original question.

Practice 4 — Conclusion audit: underline the variables named in the conclusion. Do they match the original X and Y?

Practice 5 — Repair: change the smallest broken component while preserving the original question.

Unfamiliar Transfer Challenge

A fictional material produces a stronger colour signal when more of substance Z reaches its surface. A learner asks: “How does the thickness of the material affect colour signal strength?”

The learner tests different thicknesses but also changes the area of each sample. The detector records the total light reaching it rather than signal strength per comparable area. The final conclusion says, “Thicker material absorbs more Z.”

You do not need to know the fictional material. Audit the chain:

Question: thickness → colour signal strength.
Method problem: sample area also changes.
Measurement problem: total light may combine area and signal intensity.
Conclusion problem: absorption of Z was not directly established by the stated measurement.

The Science context is new. The consistency logic survives.

Delayed Independent Return Test

After a delay, take a new investigation and work from a blank page. Without hints, write:

the exact question relationship;
the changed condition;
the measured outcome;
the two most important controlled conditions;
the result comparison that would answer the question;
a conclusion sentence limited to the tested conditions.

Then inspect the supplied method and table. If you can detect where either has drifted from your map, the learner job is transferring.

Answer and Method Checking Receipt

Can I state the investigation as X → Y?
Is X the condition deliberately changed?
Does the measurement actually represent Y?
Are important competing conditions controlled or recognised?
Does the procedure change anything else that could explain the result?
Does each result remain attached to the correct X condition?
Are units and time points clear?
Did I distinguish result from explanation?
Does the conclusion return to the same X and Y?
Did I keep the claim within the tested range and conditions?

Parent and Tutor Teaching Guide

When a child gives a method, do not immediately ask whether it is “fair”. Ask for the one-line relationship first. Then inspect every step against that line.

Useful prompts include: “Which variable in your question is this step changing?” “What does this measurement tell us?” “If we remove this result column, can we still answer the original question?” “Does your conclusion mention the same relationship you started with?”

When an investigation is flawed, repair the earliest broken link. If the question is misunderstood, changing the apparatus will not help. If the method is sound but the table omits the starting measurement needed for change, repair the recording plan rather than reteaching the whole topic.

The aim is not to make children recite variable names. It is to make the logic of the investigation visible enough that they can detect when the evidence has stopped answering the question.

Useful Internal Routes

Authoritative and Research References

These references support the curriculum, inquiry and teaching frame. The alignment map and one-line receipt above are eduKate teaching scaffolds, not official SEAB marking requirements.

The Quiet Return

A scientific investigation is a promise: the evidence collected at the end should still answer the question asked at the beginning.

Keep the relationship visible. Let every variable, measurement and table column justify its place. Then let the conclusion return to that same relationship without becoming larger than the evidence.

When the route stays intact, the investigation becomes easier to trust—and much easier to repair when one link breaks.

Small Group Tutorials