Primary 5 Science Learning Guide | Experimental Design & Evaluation Application Lab
A fair test is not “keep everything the same”. It is a designed comparison in which the intended cause changes, the relevant outcome is measured, and competing explanations are controlled well enough for the result to mean something.
Wait, What? Most Investigation Questions Are Diagnosis Questions
Students are often taught to memorise one improvement—“repeat the experiment”. But an investigation can fail for many different reasons: two variables changed at once, the wrong outcome was measured, the instrument was too coarse, timing was inconsistent, the sample was too small or the control did not actually remove an alternative explanation.
This lab trains targeted repair. First identify the exact weakness. Then choose the improvement that fixes that weakness.
The Experimental-Design Routine
- Write the question as a relationship.
- Identify the changed variable.
- Identify the measured outcome.
- List relevant controlled conditions.
- Choose a suitable instrument and range.
- Plan timing and recording.
- Decide whether repeats or more specimens are needed.
- Evaluate validity before writing the conclusion.
Lab 1: Surface Area and Evaporation
Question: How does exposed water surface area affect evaporation rate?
- Changed variable: exposed water surface area.
- Measured outcome: water mass lost in a fixed time.
- Controls: starting water mass, location, airflow, starting temperature and time interval.
- Validity check: do not change airflow at the same time as area.
Lab 2: Invalid Evaporation Test
A wide tray is placed beside a fan. A narrow cup is placed in still air.
The test changes both surface area and airflow. Even if the result repeats perfectly, it cannot isolate surface area. Reliability does not repair invalid design.
Lab 3: Wrong Outcome
A student tests whether more batteries make a bulb brighter, but records only the number of batteries.
The changed variable was recorded instead of the outcome. Brightness must be observed or measured because that is what the question asks.
Lab 4: Instrument Too Coarse
Expected water loss is 2–4 g. The balance reads in 10 g steps.
Repeating the same measurement cannot recover detail the instrument cannot detect. Use a finer-resolution balance.
Lab 5: Poor Timing
Setup A is measured after exactly 30 minutes. Setup B is measured “about half an hour later”.
Timing inconsistency weakens the comparison. Use the same defined interval and a consistent start/stop method.
Lab 6: One Plant Per Group
One large-leaf plant and one small-leaf plant are compared.
Individual plant health could affect the result. Use several similar specimens in each group where practical to reduce the influence of biological variation.
Lab 7: Repeating One Person
A student measures one person’s pulse after exercise ten times and concludes that all children recover in the same way.
Repeated measurement can improve confidence for that person, but it does not make one participant representative of all children. More participants are needed for broader conclusions.
Lab 8: Conductor-Test Control
An unknown material fails to light a bulb.
Before calling it an insulator, replace it with a known conductor. If the bulb lights, the rest of the circuit is working. This control removes several alternative explanations.
Lab 9: Plant Water-Loss Control
A leafy shoot stands in an open beaker of water. The water level falls.
The result combines direct evaporation from the beaker with water movement through the plant. Covering the water surface reduces the direct-evaporation alternative.
Lab 10: Pollination Covering Problem
Covered flowers form fewer fruits than open flowers.
The covering may reduce insect access, but it may also alter airflow, humidity, temperature or light. A stronger design reduces unintended changes while controlling pollinator access.
Lab 11: Reliability
| Trial | Water loss |
|---|---|
| 1 | 8 g |
| 2 | 9 g |
| 3 | 8 g |
The repeated results are reasonably consistent. Reliability asks whether repeated observations produce a stable pattern.
Lab 12: Reliability Problem
| Trial | Water loss |
|---|---|
| 1 | 8 g |
| 2 | 2 g |
| 3 | 17 g |
The spread is large. Inspect the method, environment and measurements before trusting the average.
Lab 13: Accuracy and Zeroing
A balance reads 5 g when empty.
Every reading may be shifted. Zero or calibrate the instrument according to its correct use before collecting data.
Lab 14: Validity
A method is valid when it actually tests the intended relationship. A perfectly repeatable method can still be invalid if it measures the wrong outcome or changes multiple causes.
Lab 15: Better Improvement Language
Weak: “Repeat to make it more accurate.”
Better: “Repeat the trial several times to check whether the result is consistent and reduce the influence of one unusual trial.”
Different weakness: “Use a finer-resolution balance so small mass changes can be detected.”
Lab 16: Plan the Table Before the Test
A good results table contains the changed condition, measured outcome, units and trial structure before data collection begins. Planning the table forces the learner to decide exactly what evidence is needed.
Lab 17: Range of the Changed Variable
Testing only “low” and “high” airflow can show a difference but not the shape of the relationship. Using several safe levels can reveal whether the effect increases steadily, plateaus or behaves differently.
Lab 18: Negative Result
If changing leaf area produces no clear difference across repeated trials, do not invent a positive trend. Report the result and consider whether the effect is small, the method insensitive or another variable dominant.
Lab 19: Anomaly Handling
An anomalous trial should be recorded, investigated and repeated where appropriate. Deleting it because it “spoils the pattern” weakens scientific integrity.
Lab 20: Scope of Conclusion
A test of three materials cannot justify “all materials behave this way”. A study of one student cannot justify “all humans recover in six minutes”. The conclusion must fit the sample and conditions.
Lab 21: Safety as a Design Constraint
A scientifically interesting electrical test is not acceptable if it uses household mains electricity. A valid school investigation also has to be safe. Use low-voltage teacher-approved equipment only.
Lab 22: Ethical and Practical Limits
Some scientific questions cannot be investigated directly in a classroom. Human reproduction is taught through established scientific knowledge and models, not through direct reproduction experiments. Scientific method includes choosing methods appropriate to the subject.
Evaluation Checklist
- Does the changed variable match the question?
- Does the measured outcome answer the question?
- Are competing variables controlled?
- Is the instrument suitable?
- Is timing consistent?
- Are repeats or more specimens needed?
- Is the test safe?
- Does the conclusion match the actual evidence?
Misconception Repair Set
- Fair test means literally everything must be the same.
- Repeating fixes every weakness.
- More data automatically means better data.
- A finer instrument fixes an invalid comparison.
- A valid method must give identical results every time.
- One organism can represent the whole species.
- An anomaly should be deleted.
- A negative result means failure.
- Safety is separate from experimental design.
Exam Answer Control
- Name the exact weakness.
- State how it can affect the result.
- Give one targeted improvement.
- Explain why the improvement helps.
- Do not use generic “more accurate” wording unless accuracy is actually the issue.
Model Limit
Primary Science uses simplified experimental-design language. Real research uses formal statistics, calibration, randomisation, replication and uncertainty analysis. The Primary goal is the same in spirit: build a comparison whose evidence can answer the question honestly.
Delayed Return Challenge
One week later, diagnose five flawed investigations without notes. For each, identify changed variable, measured outcome, one control problem, one measurement issue, one targeted improvement and the maximum conclusion the method could justify.
Experimental-Design Receipt
- I identify changed, measured and controlled variables from the question.
- I distinguish reliability, accuracy and validity.
- I match improvements to specific weaknesses.
- I choose suitable instruments and time intervals.
- I account for biological variation.
- I use controls to reduce alternative explanations.
- I preserve anomalies and negative results.
- I keep safety and scope inside the method.
Official Reference Routes
- Singapore Ministry of Education — Primary Science Teaching & Learning Syllabus 2023
- Singapore Examinations and Assessment Board — 2026 PSLE Science Syllabus
Continue the Batch 11 Laboratories
- Primary 5 Science Learning Hub
- Data, Graphs & Evidence Application Lab
- Open-Ended Answer Construction Application Lab
- Mixed-Concept Transfer Challenge Lab
The Quiet Return
Good experimental design is controlled scepticism. Ask what else could explain the result, then build the method so that the intended relationship has a fair chance to speak clearly through the evidence.
