Using Fair Comparisons in an Investigation | Singapore Primary Science Guide

eduKate Learning Manual — Scientific Inquiry

Teaching goal: By the end of this manual, a learner should be able to design and critique a fair comparison: change the factor being investigated, observe or measure the outcome, keep other relevant conditions consistent, and explain what the evidence can — and cannot — show.

WAIT, WHAT? A Perfectly Fair Test Can Still Answer the Wrong Question

You can control every variable beautifully and still collect evidence that does not answer the question you actually asked. A flawless method for measuring temperature cannot answer a question about mass. Fairness is therefore not a ritual checklist; it is part of a larger chain linking question → changed factor → measured outcome → controlled alternatives → conclusion.

The first RFE test is simple: does the comparison still point back to the original question?

A fair comparison is one of the most useful tools in Primary Science because it helps us separate a possible cause from other competing changes. If one seedling receives more water but also receives more light, grows in different soil and begins much larger, we cannot confidently attribute a difference in growth to water alone. Too many things changed.

The purpose of a fair comparison is therefore not to make two situations identical. If they were identical, there would be nothing to investigate. The purpose is to make them similar in the relevant ways except for the factor we deliberately want to compare.

1. The Big Idea: Make the Comparison Answer the Question

A fair comparison connects four pieces of reasoning:

Question: What relationship are we investigating?
Change: What factor will deliberately differ?
Measure: What outcome will be observed or measured?
Keep consistent: What other relevant conditions should be kept the same so they do not provide competing explanations?

For older Primary learners, these ideas may be connected to the terms independent variable, dependent variable and controlled variables. The terminology is useful, but the reasoning comes first.

2. Fair Does Not Mean “Everything Is the Same”

Suppose we ask: How does the distance between a torch and an object affect the size of the shadow?

The distance between torch and object must change. That is the factor being investigated.
The shadow size must be observed or measured. That is the outcome.
The same object, light source, screen position and measuring method should be used where relevant.

If a learner says, “For a fair test, everything must stay the same,” ask: “Then what are we testing?” One meaningful factor has to change.

3. The Change–Measure–Keep Routine

Primary learners can design many investigations using three questions:

What will I CHANGE?
What will I MEASURE or OBSERVE?
What important things will I KEEP THE SAME?

Then add two more questions that make the investigation stronger:

How many times should I repeat the comparison?
What would make the conclusion too strong for the evidence?

4. Worked Example: Shadow Size

Question: How does the distance between a torch and an opaque object affect the size of the shadow formed on a screen?

Change: distance between torch and object.
Measure: a chosen dimension of the shadow, such as its height.
Keep consistent: same torch, same object, same screen, same object orientation, same measuring method and a suitable fixed arrangement apart from the distance being changed.
Repeat: collect more than one reading where practical and check unexpected results.

Now imagine changing both the distance and the object. If the shadow changes size, we no longer know whether distance, object size or both produced the difference. The comparison has lost its ability to answer the original question cleanly.

5. Worked Example: Water and Seedling Growth

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

Change: amount of water given.
Measure: increase in seedling height over the chosen time.
Keep consistent where possible: plant type, starting size, soil type and amount, container size, light conditions, location, measuring method and observation schedule.

Real living things contain natural variation, so even a carefully designed comparison does not make two seedlings biologically identical. This is why repeated observations and cautious conclusions matter.

6. Why Repeated Trials Matter

One result can be affected by accident, measurement error or natural variation. Repeating a comparison helps us see whether a pattern occurs consistently.

Repetition does not magically remove every problem. Repeating a badly designed investigation gives us the same design problem many times. First make the comparison sensible; then use repetition to check reliability.

7. Relevant Conditions: What Actually Needs to Stay the Same?

Children sometimes produce long memorised lists of “things to keep constant”. A better question is: Could changing this condition reasonably affect the outcome?

If yes, it may need to be controlled. If not, controlling it may add complexity without improving the investigation.

For a shadow investigation, the colour of the ruler used to measure the shadow is unlikely to affect shadow size. The position of the light source clearly can. Scientific control is about relevance, not ritual.

8. Correlation, Cause and What a Fair Comparison Can Support

A carefully controlled comparison can strengthen a causal explanation because it reduces alternative reasons for the observed difference. But conclusions should still match the evidence.

If three classroom trials show a pattern, a sensible conclusion is that under the conditions tested, changing the factor was associated with the observed outcome. It would be too strong to claim that the rule has been proved for every possible object, organism, temperature or environment.

This habit — making the conclusion no larger than the evidence — is one of the foundations of scientific integrity.

9. Common Investigation Mistakes — and Repairs

Changing more than one important factor. Repair: identify the question and change only the factor needed to answer it.
Measuring a different outcome from the one in the question. Repair: make the evidence directly answer the stated question.
Keeping irrelevant things the same while missing an important condition. Repair: ask which conditions could realistically affect the outcome.
Using different measuring methods for different groups. Repair: standardise the measurement method.
Running only one trial and treating it as universal proof. Repair: repeat where practical and limit the claim.
Changing the method halfway through. Repair: plan before collecting data and record any unavoidable change.
Deleting an unexpected result because it “must be wrong”. Repair: check the measurement and method; keep honest records of anomalous results.
Assuming a fair test is the only valid scientific method. Repair: use fair comparisons when the question is about the effect of a factor; use observation, field comparison, modelling or other methods when they fit the question better.

10. Teach It: A Quick Fair-Comparison Lesson

Use a torch, one opaque object, a screen or wall, a ruler and a safe working space.

Ask the learner to form a testable question about distance and shadow size.
Ask what should change.
Ask what should be measured.
Ask which conditions need to remain consistent.
Before collecting data, deliberately suggest an unfair change: “What if we use a much larger object for the second measurement?” Let the learner explain why that weakens the comparison.
Collect simple observations.
Ask the learner to state a conclusion that is no stronger than the evidence.

Do not use lasers or unsafe light sources. Ordinary torches are sufficient for the teaching idea.

11. Guided Practice: Diagnose the Design

For each investigation, identify the problem.

A child compares two plants. Plant A receives more water and is placed beside a sunny window. Plant B receives less water and is placed in shade. The child concludes that extra water caused Plant A to grow faster.
A child compares how quickly 50 mL and 100 mL of water evaporate from containers of different shapes and concludes that starting volume alone caused the difference.
A child tests the attraction of a magnet using a steel paper clip for one trial and an aluminium foil ball for another, while also changing the distance.
A child repeats the same well-controlled shadow measurement three times and gets one unusual value.

Discussion: In 1, water and light both differ. In 2, both volume and container geometry may affect the outcome. In 3, material and distance both change. In 4, the unusual value should be checked rather than automatically removed.

12. Independent Challenge: Build the Investigation

Choose one question and write a complete fair-comparison plan.

How does the distance between a light source and an object affect shadow size?
How does exposed surface area affect evaporation over a fixed period?
How does the number of identical cells in a suitable low-voltage circuit affect the brightness of a bulb, using school-approved equipment?

State the factor changed, outcome measured, important conditions kept consistent, equipment needed, recording method, number of trials and one limitation. Electrical investigations should use only suitable low-voltage educational equipment under appropriate adult or teacher supervision.

13. How an Adult Should Teach This

Ask “What question are we trying to answer?” whenever the method becomes complicated.
Ask “What else changed?” when a learner makes a causal claim.
Ask “Could that other change also affect the result?” instead of simply saying “wrong variable”.
Let children critique flawed investigations; diagnosis often builds deeper understanding than copying a perfect procedure.
Do not teach control variables as a memorised list. Teach relevance.
Model honest limitations. A good investigation can still have constraints.
Do not let the expected textbook answer override the observed data.

14. What Mastery Looks Like

Beginning: the learner knows that comparisons should be “fair” but cannot explain what that means.
Developing: the learner can identify what changes and what is measured in a clear example.
Secure: the learner can choose relevant conditions to keep consistent and explain why.
Strong: the learner can diagnose confounding changes, plan repeats and limit the conclusion to the evidence.
Advanced for Primary: the learner can explain why a fair test is appropriate for some questions but not all scientific questions.

15. Connection to Singapore Primary Science and PSLE

Scientific inquiry in Singapore Primary Science includes making predictions, interpreting information, evaluating observations and methods, and communicating explanations and reasoning. The current syllabus also encourages learners to collect evidence in authentic contexts and examine assumptions and uncertainty. A fair comparison therefore matters not as a formula to memorise, but as a method for making evidence more interpretable.

16. Continue the Scientific Inquiry Sequence

17. Trusted References

eduKate Learning Manual principle: Change what you mean to test. Measure what the question asks. Keep competing causes under control. Then let the evidence speak only as far as it can.

Latest-Standard Strengthening — The Competing-Cause Gate

Before interpreting a difference as the effect of the factor you changed, ask: What else could have changed that would produce the same outcome? A condition deserves control only when changing it could plausibly affect the measured result. Fairness is therefore not a memorised list of constants; it is the deliberate closing of alternative causal routes.

What Would Make Us Reject the First Explanation?

Suppose you investigate whether exposed surface area affects evaporation. If the containers also differ in starting volume, location, airflow, temperature or material, a difference in water loss cannot cleanly isolate surface area. Improve the test, predict the expected pattern, repeat it, and where practical swap the container positions. If the result follows the location rather than the surface-area condition, the first explanation needs revision.

Model Limit: Fair Comparison Reduces Confounding; It Does Not Guarantee Universal Causation

A well-controlled Primary investigation can make a causal interpretation much more defensible under the conditions tested. It cannot remove every source of natural variation, unnoticed influence or measurement error, and it does not automatically prove the same effect across every object, organism or environment. Formal causal inference and statistical experimental design belong to later study.

Changed-Problem Transfer

Two ice cubes melt at different rates. One sits on a metal plate near a window; the other is larger and sits on a plastic plate farther inside the room. A learner concludes that metal makes ice melt faster. Redesign the comparison so plate material is the intended changed factor. Name the outcome, the relevant conditions to control, one repeat or reversal check, and one result that would make you question the original explanation.

RFE Check: What Should Survive After the Page Is Closed?

The durable routine is: What is the question? What factor changes? What outcome is measured? What else could produce the same outcome? How will I control or check those alternatives? What result would make me revise the conclusion? That routine turns “fair test” from a school phrase into a working method for protecting cause-and-effect reasoning.

Teaching Guide — Use This Last

For parents, tutors and teachers: begin with a deliberately flawed comparison and let the learner find the competing cause before designing the improved test. Keep asking “What else changed?” and “Could that also affect the outcome?” Do not accept a memorised control-variable list unless the child can explain why each condition matters. Finish by asking what evidence would force revision. Stop helping when the learner can independently spot a confound, choose relevant controls, propose a repeat or reversal check, and limit the conclusion to the evidence.