Primary 5 Science Learning Guide | Variables, Fair Tests & Experimental Design
A fair test is not a ritual list of variables. It is a design that makes one scientific relationship easier to see.
Wait, What? “Keep Everything the Same” Is Not a Complete Answer
Primary 5 Science is often the point where investigation questions become more demanding. Students may know the words “changed variable”, “measured variable” and “controlled variable”, yet still struggle to identify them inside an unfamiliar setup. The difficulty is not vocabulary alone. It is deciding what job each part of the investigation is performing.
A strong learner starts with the scientific question. What relationship is the investigation trying to test? Once that is clear, the roles become easier to assign. One condition is deliberately changed. One outcome is observed or measured. Other relevant conditions are kept sufficiently similar so they do not provide competing explanations for the result.
Quick Answer
In a fair test, the changed variable is the condition deliberately altered by the investigator. The measured variable is the outcome used to judge what happened. Controlled variables are other relevant conditions kept sufficiently similar so the comparison isolates the relationship of interest. Reliability improves when observations are repeated or more specimens are used where appropriate. Accuracy depends on suitable methods and instruments. Validity asks whether the investigation actually tests the intended scientific relationship.
Start With the Question, Not the Apparatus
Consider the question: Does exposed surface area affect the rate of evaporation? The changed variable is exposed surface area. The measured outcome might be loss of water mass over a fixed time. Conditions such as starting water amount, container material, temperature, air movement and time interval may need control.
Now change the scientific question: Does air movement affect the rate of evaporation? The apparatus might look similar, but the variable roles change. Air movement is now the changed variable, while exposed surface area becomes a condition to control. This is why memorising “the variable in this experiment” from one diagram is fragile. Variables belong to the question, not permanently to the object.
The Four-Part Investigation Frame
| Investigation job | Question to ask | Example |
|---|---|---|
| Scientific question | What relationship is being tested? | Does number of leaves affect water loss from a shoot? |
| Changed variable | What is deliberately different? | Number or total area of leaves |
| Measured outcome | What observation or quantity answers the question? | Decrease in water mass |
| Controlled conditions | What else must remain sufficiently similar? | Plant type, time, light, temperature, starting water volume |
Worked Investigation 1: Evaporation From Two Dishes
Dish A and Dish B contain equal masses of water. Dish A is wide and shallow; Dish B is narrow and deep. Both are placed side by side for one hour. Dish A loses more mass.
Scientific question: How does exposed surface area affect evaporation over the same time?
Changed variable: exposed surface area of the water.
Measured outcome: decrease in water mass after one hour.
Important controls: initial water mass, water type, location, surrounding temperature, air movement and time.
Conclusion: Under these test conditions, the larger exposed surface area was associated with a faster loss of water by evaporation.
Why “Same Container” Can Be Wrong
Students are often told to keep the container the same, but that instruction only makes sense if container shape is not the variable being tested. If the investigation is about surface area, different container shapes may be necessary to create different exposed areas. A controlled variable is not “something we always keep the same”. It is a condition that must stay sufficiently similar because changing it could interfere with the relationship being tested.
Fair Does Not Mean Perfectly Identical
Living systems create an important complication. Two leaves, flowers or human beings cannot be perfectly identical. Scientific control therefore means making relevant conditions similar enough for a useful comparison. Researchers may use plants of the same species and similar size, select flowers at similar developmental stages, or compare a group rather than one specimen.
This matters in Primary 5 topics such as reproduction and transport. Biological variation is normal. The correct response is not to pretend it does not exist, but to design the investigation so one unusual specimen is less able to distort the conclusion.
Worked Investigation 2: Does Leaf Area Affect Water Loss?
Two similar leafy shoots are placed in identical containers with equal water volumes. A thin oil layer covers the water surface in both containers. Shoot A retains many leaves; most leaves are removed from Shoot B. After two hours, the container with Shoot A loses more water.
Changed variable: leaf area.
Measured outcome: decrease in water amount or mass.
Why use oil? The oil reduces direct evaporation from the container surface. That removes a competing pathway for water loss, making the measured decrease a better indicator of water moving through and leaving the shoot.
Validity lesson: A control is useful only when it blocks a plausible alternative explanation.
Control Set-Up Versus Controlled Variable
These terms are easy to confuse. A controlled variable is a condition kept sufficiently similar across compared setups. A control set-up is a comparison condition designed to show what happens when the tested factor is absent or when the normal condition is maintained.
For example, when testing whether a material conducts electricity, first closing the test gap with a known conductor confirms that the rest of the circuit works. That is a useful control check. It is not the same thing as saying “the wire is a controlled variable”.
Worked Investigation 3: Pollinator Access and Fruit Formation
A group of flower buds is covered with mesh before opening. A second similar group is left uncovered. The mesh blocks large insect visitors but allows light and air through. The number of flowers that later develop into fruits is recorded.
Question: Does access by large insect visitors affect fruit formation?
Changed variable: access by large insects.
Measured outcome: proportion or number of flowers that later form fruits.
Possible limitation: The mesh might also alter airflow, humidity or physical contact around the flower. A careful conclusion should therefore say the result supports a role for insect access rather than claiming absolute proof from one design.
Repeated Trials: What They Actually Fix
Repeating an investigation can reveal whether a result is consistent. If one trial gives a very different reading from the others, the learner can investigate possible causes instead of treating the single value as certain. Repeats are especially useful when measurements naturally vary.
But repeated trials do not automatically make a badly designed experiment valid. If every trial changes two variables at once, repeating the same unfair comparison simply repeats the same design problem.
More Specimens Versus More Measurements
These are different strategies. If the concern is natural variation between living specimens, using more similar specimens can help. If the concern is random measurement variation, repeating measurements on the same setup may help. A good Primary 5 answer identifies the source of uncertainty before suggesting “repeat the experiment”.
Accuracy, Reliability and Validity
| Idea | Main question | Typical improvement |
|---|---|---|
| Accuracy | Is the measurement close enough to the true value for the purpose? | Use a suitable instrument and correct reading method |
| Reliability | Would the result be reasonably consistent if repeated? | Repeat trials or use more specimens where appropriate |
| Validity | Does the investigation actually test the intended relationship? | Control competing variables and measure the right outcome |
Do not collapse these into one word: “fair”. A result can be reliable but invalid. A measuring instrument can be precise while the comparison still tests the wrong thing.
Choosing What to Measure
The measured outcome must answer the scientific question. If testing evaporation rate, final temperature may be interesting but does not directly answer how much water was lost. If testing whether more cells make a bulb brighter, counting the number of cells simply restates the changed variable. A suitable light measurement is closer to the required outcome.
This sounds obvious when stated directly, yet it is a common source of examination error. Students can become so focused on the apparatus that they forget what evidence the question actually needs.
Worked Investigation 4: Number of Cells and Bulb Brightness
A student wants to test whether the number of identical cells affects the brightness of one bulb in a fixed circuit arrangement.
- Changed variable: number of identical cells.
- Measured outcome: brightness measured with a light sensor at a fixed distance.
- Controlled conditions: same bulb, same cell type, same circuit arrangement, same sensor position, same room lighting.
- Reliability improvement: repeat the reading at each cell number.
- Safety boundary: use only low-voltage classroom components within teacher and equipment limits.
Procedure Steps Are Not Results
“Measure the temperature every minute” is a procedure. “The temperature increased from 24°C to 38°C” is a result. “The liquid was heated more quickly in Set-up A” is an interpretation. Keeping these jobs separate makes investigations much easier to analyse.
Prediction, Result and Conclusion
A prediction is made before the result is known. A result records what was observed or measured. A conclusion interprets the result in relation to the scientific question. If the result disagrees with the prediction, do not rewrite the prediction afterward. Compare them honestly and consider whether the model, method or one trial needs further investigation.
Unexpected Results Are Information
Suppose three evaporation trials show expected water losses of 8 g, 9 g and 2 g. The 2 g result should not be deleted merely because it is inconvenient. Check for an explanation: was the container partly covered, was the balance read incorrectly, was the setup moved, or was the recording wrong? If no cause is found, report the unusual result and repeat the test.
Scientific honesty matters. Data are not decorations added after the conclusion. The conclusion must return to the evidence.
How to Evaluate a Method
- Restate the scientific question.
- Identify what was changed and what was measured.
- Check whether the measured outcome actually answers the question.
- Look for other conditions that also changed.
- Check whether the observation period is appropriate.
- Check whether the instrument has suitable range and resolution.
- Consider natural variation if living specimens are used.
- Ask whether repeats or more specimens are needed.
- State one specific improvement and the exact problem it fixes.
Specific Improvements Beat Generic Improvements
Weak: “Repeat the experiment to make it more accurate.”
Better: “Repeat the measurement several times and calculate a representative value so one unusual reading has less influence on the conclusion.”
Weak: “Use better equipment.”
Better: “Measure water loss with a digital balance instead of judging the water level by eye so smaller changes can be measured more consistently.”
Question-Language Control
- Identify a variable: name the condition and its role.
- State a control: name a relevant condition that must remain similar.
- Explain why it is controlled: show how changing it could affect the measured outcome.
- Suggest an improvement: fix a specific weakness in accuracy, reliability or validity.
- Evaluate: judge what the method allows the learner to conclude and what remains uncertain.
Worked Answer Surgery
Question: Why should the two water samples start at the same temperature when comparing container surface area?
Weak: “To make it fair.”
Better: “Temperature can affect the rate of evaporation. Keeping the starting temperature the same helps ensure that differences in water loss are more likely to be due to exposed surface area rather than temperature.”
Common Misconceptions and Repairs
- Every condition must be identical. Repair: only relevant competing conditions need control; the changed variable must differ.
- A variable always has the same role. Repair: its role depends on the scientific question.
- Repeating makes an unfair test fair. Repair: repeats improve reliability, not a flawed comparison.
- One living specimen is enough. Repair: biological variation may require multiple similar specimens.
- A control set-up is the same as a controlled variable. Repair: one is a comparison setup; the other is a condition kept similar.
- Any measurement is useful. Repair: the measured outcome must answer the question.
- An unexpected result should be removed. Repair: investigate it, report it and repeat when appropriate.
Model Limit: School Fair Tests Simplify Real Science
Real scientific investigations can involve many interacting variables, statistical uncertainty, instrument calibration and complex controls. Primary 5 fair-test models simplify this so students can learn the logic of isolating a relationship. Use the simplified framework carefully without assuming all real science changes exactly one variable at a time.
Transfer Across Primary 5 Topics
| Topic | Possible changed variable | Possible measured outcome |
|---|---|---|
| Water | Exposed surface area | Mass of water lost in fixed time |
| Reproduction | Pollinator access | Number of fruits formed |
| Plant transport | Leaf area | Water loss / uptake over time |
| Human systems | Activity level | Breathing or pulse rate |
| Electrical systems | Number of cells | Bulb brightness |
Delayed Return Test
Three to five days later, take one unfamiliar investigation and do not answer the science question immediately. First write only four lines: scientific question, changed variable, measured outcome, controlled conditions. Then identify one reliability issue and one validity issue. If you can do this without topic-specific memorisation, the investigation framework is becoming transferable.
Primary 5 Investigation Receipt
- I identify variables from the scientific question, not from memorised apparatus.
- I can explain why a controlled condition matters.
- I know the difference between a control set-up and a controlled variable.
- I choose an outcome that actually answers the investigation question.
- I distinguish reliability, accuracy and validity.
- I know when more specimens are more useful than repeated measurements.
- I can keep prediction, result and conclusion in different jobs.
- I can respond to unexpected data without deleting it automatically.
- I can suggest a specific improvement and explain what problem it fixes.
Parent and Tutor Teaching Guide
Before discussing the answer, cover the labels in an investigation diagram and ask the child to reconstruct the scientific question. Then ask what must be deliberately different and what evidence would answer that question. This prevents the common habit of identifying variables by position alone.
When a child writes “keep everything the same”, ask which conditions are scientifically relevant and why. When they write “repeat for accuracy”, ask what kind of weakness the repetition actually addresses. Precision grows when every improvement has a reason.
Official Reference Route
Singapore Ministry of Education — Primary Science Teaching & Learning Syllabus 2023
This is an independent eduKate Sengkang learning guide. School practical work must follow teacher instructions and appropriate safety procedures.
Continue the Primary 5 Science System
- Primary 5 Science Learning Hub
- Diagrams, Tables, Graphs & Evidence
- Structure, Function, Systems & Cause-and-Effect
- Explain, Compare, Predict & Evaluate Questions
The Quiet Return
A good investigation makes reasoning visible. Begin with the question. Assign each variable a job. Choose evidence that actually answers the question. Control plausible competing causes. Repeat only when repetition addresses a real source of uncertainty. Then return the conclusion to the data. That is the beginning of scientific judgement.
