What is a fair test in PSLE Science, how do pupils identify changed and unchanged variables, and what should they do when a graph looks unfamiliar? The useful answer is to start with the question the investigation is supposed to answer. Change the factor being tested, keep other relevant factors comparable, measure a suitable response, and decide whether the evidence supports the conclusion. This MOE/SEAB Science Syllabus Guide turns that logic into practical strategies, tables, original worked questions and safe home observations for Primary 3–6 children preparing for PSLE 2027.
The MOE Primary Science syllabus sets a clear progression: by the end of P4, pupils should recognise a fair test through changed and unchanged variables; by the end of P6, they should be able to design fair tests. The same official progression includes recording observations, using tables, bar and line graphs, analysing evidence and communicating reasoning. SEAB’s Science specification from the 2026 examination assesses predictions, interpretation, evaluation and explanation. These are not obscure enrichment extras; they are part of the reasoning expected when syllabus knowledge is applied to a new situation.
- Identify the changed condition, what was measured and what remained comparable.
- Read axis labels, units and starting values before interpreting data.
- Separate fair testing from repeating trials.
- Explain why a conclusion does or does not follow from evidence.
- Use safe original experiments and paper-based scenarios.
Start here if a pupil can recite Science definitions but keeps losing marks on experimental questions. Learn to separate what was deliberately changed, what was measured and what should have stayed comparable; practise reading graph axes and units; evaluate unusual results without secretly deleting them; and use a clear explanation linked to the investigation. Every practice scenario below is original, not an official SEAB item or a claim about an unreleased 2027 paper. The 2027 format must be checked with SEAB when published.
A quick navigation guide
- Official MOE science-inquiry progression
- Fair tests, tables, graph scales and causal conclusions
- Thirty-six original examination-style scenarios
- Ten-step learning recovery plan, FAQs and official links
An experiment begins with a question
Good investigation design starts with a precise testable question. ‘Does air movement affect evaporation rate?’ is clearer than ‘What happens to water?’ because it identifies the relationship to investigate. It suggests a deliberate change in air movement and a way to observe or measure loss of water over time. The pupil can then think about temperature, container surface, starting volume and measurement duration as conditions that should be comparable. Without a clear question, children are easily impressed by apparatus instead of evaluating whether the evidence answers anything.
Changed, unchanged and measured: the school-friendly language
MOE’s progress table explicitly describes fair tests in terms of changed and unchanged variables. Some teachers use additional labels such as independent, dependent and controlled variables; they can help organise thought, but a child should not need specialist vocabulary to identify what changes and what gets measured. In a light-and-shadow test, distance between source and object may be changed, shadow size measured, and the object shape and screen arrangement kept comparable. Always interpret the diagram: a factor assumed unchanged is not automatically controlled in an incomplete description.
Why changing two things at once is a serious problem
Two plant pots differ in water supplied and amount of light. If one grows more, which factor caused it? The comparison does not isolate either. Many pupils answer with a genuine fact, such as ‘plants need light’, yet the claim about this experiment remains unsupported. To improve it, keep plant variety, age, starting size, soil, water and observation period sufficiently comparable, then deliberately vary only light availability. A scientific statement can be true generally and still be an invalid conclusion from a particular experiment. That distinction is central to evaluating investigations.
Fairness is not the same as repeating trials
Repeat measurements may show whether results are consistent. They cannot fix a basic design in which two major factors change at once. Imagine comparing a broad shallow water tray on a hot balcony with a deep cup in a cool room. Repeating both arrangements ten times does not prove that temperature alone causes any rate difference. First isolate the intended condition, then repeat or refine measurements to assess reliability. Pupils who know this difference can answer ‘suggest an improvement’ questions more precisely than those who always say ‘repeat three times’.
A result can be reliable but answer the wrong question
Suppose a pupil repeatedly observes that a metal spoon gets warm in soup. Those measurements may be consistent, but they cannot by themselves compare heat conduction with a plastic spoon unless a proper comparison is included. Similarly, a graph can be drawn accurately while measuring a response unrelated to the investigation’s question. Scientific validity asks whether the design and measurement meaningfully address the claim. This does not require advanced statistical terminology. Ask, ‘If we get the same reading again, will it actually tell us what we wanted to find out?’
What to read on a graph before interpreting it
Read the graph title, horizontal and vertical axes, units, scale, plotted points and any legend. The horizontal axis often shows time or a condition, but not always. The vertical axis often shows the measured response; confirm rather than assume. Check whether the scale starts at zero, whether intervals are equal and whether two lines refer to different trials. Only then describe a rise, fall or unchanged value. A memorised sentence about ‘the steeper line’ is meaningless unless the quantities and units have been correctly identified.
Bar graph or line graph: avoid automatic rules
MOE expects pupils to interpret information from both bar and line graphs, charts, tables and diagrams. Bar charts are useful for comparing categories, while line graphs can show a measured response across ordered or continuous values such as time. In a test question, the representation is already given: the task is to read it faithfully. Don’t force every set of observations into a slope explanation. A bar with a greater height may represent more water remaining, not more water lost; the direction of the scientific conclusion depends on the label.
Change from starting to ending readings
Many tasks measure change, not final size. A seedling that goes from 9 cm to 15 cm grows 6 cm; one that goes from 11 cm to 16 cm grows 5 cm. Although the second ends taller, the first grows more during the interval. For a water-loss experiment, subtract final volume from initial volume; for a cooling experiment, compare temperatures in the stated time. Check units and time before averaging anything. A pupil can possess the correct plant or heat concept and still lose the argument because the arithmetic compares the wrong quantity.
A rate requires attention to time
If two similar amounts of water evaporate, the one lost over a shorter interval indicates a higher average rate during that interval. A 10 mL loss over one hour and a 20 mL loss over four hours do not imply the second was faster; the averages are 10 mL per hour and 5 mL per hour. Use simple numerical reasoning when appropriate, but avoid inventing an official formula not required by a given task. The important idea is that change and time both matter when comparing rates.
Unusual readings and errors
An unexpectedly different result is a reason to investigate, not permission to erase inconvenient data. Check the instrument, unit, reading, recording, setup, and any condition that may have changed. A repeat measurement can help assess whether the reading was anomalous, but the original should not be silently rewritten. An unusually high thermometer reading may be caused by a malfunction or a genuine temporary change; the observation alone cannot distinguish the two. Pupils should give a plausible check directly connected to the equipment and method described.
Prediction versus conclusion
A prediction is made using existing knowledge before or without seeing the result. A conclusion interprets observed data after an investigation. If air movement is increased while other conditions are comparable, a pupil may predict faster evaporation. If measured loss is greater in the moving-air trial, the recorded result supports that expectation under the design. But do not convert a single trial into a law about every situation, or claim the observed difference proves an unmeasured internal process. Conclusions should stay within the evidence.
Correlation is not proof of one cause
If taller plants also receive more sunlight, it is tempting to say sunlight alone made them tall. But soil, water, plant age and other conditions may differ. Primary learners can understand the issue without the word ‘correlation’: two changes happening together do not establish which one caused the other. Ask them to name another factor that could explain the pattern and design a fairer comparison. This is helpful beyond exams, including when children encounter online claims about food, exercise, consumer products or the environment.
Controlled comparisons in light and shadows
The P4 syllabus includes investigations into variables affecting shadows, including object shape and size, position, and distances between light source, object and screen. If a pupil wants to test how one distance affects shadow size, the other distances and object should be kept comparable. Trace light traveling in straight lines rather than relying only on memory of ‘nearer equals larger’. When two distances change simultaneously, the direction of shadow-size change may not be uniquely attributable to one. Draw a clear source-object-screen setup with labels before writing.
Controlled comparisons in water evaporation
The P5 Standard syllabus specifically includes investigating wind, temperature and exposed surface area as factors in evaporation rate. Each can be treated as a changed condition in a suitably controlled design. If a broad tray sits next to a fan and a narrow cup is protected from the fan, the observed rate cannot isolate surface area. A good answer can identify both changes and suggest making container dimensions comparable if air movement is the intended variable. Explain what is measured—such as liquid water volume lost over a fixed duration—rather than making the unspecific claim that one ‘has more evaporation’.
Controlled comparisons in plant photosynthesis
The P6 syllabus includes investigating water, light energy and carbon dioxide requirements for photosynthesis, with the course-specific outcomes determining scope. A test of light should not also vary plant species and water supply. Evidence may be recorded using a school-approved method, but the pupil must know what that method actually shows. Leaf colour, overall height and starch-test results are not interchangeable measures. Do not assume a coloured leaf is guaranteed to have a particular photosynthesis rate; read the specific outcome measured and its limitations.
Controlled comparisons in electricity
P5 circuit investigations vary the number of suitable batteries arranged in series or the number of bulbs in specified arrangements. A fair comparison keeps other circuit components and conditions suitably matched. If one drawing has two cells and a bright bulb while another has one cell and an entirely different bulb type, brightness cannot be attributed only to cell count. Pupils should trace complete paths and identify whether any connection is open before discussing brightness. Any hands-on work must be teacher-approved, low-voltage and safe; never short batteries or test household mains wiring.
Controlled comparisons in forces
P6 force situations can involve how surfaces affect a sliding object’s motion. If the child compares a heavy object pushed hard on rough carpet with a light object nudged on a smooth desk, the trial cannot isolate friction due to surface type. A better paper design keeps the object, method of starting motion and measurement distance consistent while changing the surface. Forces do not become measurable from a photograph simply because two objects stop at different positions. The strongest explanation identifies the observed motion difference and the limited inference that controlled conditions support.
Design for animal and plant welfare
Environmental investigations must protect organisms. Do not ask children to deprive animals of food, trap wildlife, damage ecosystems or cut plant roots unnecessarily to demonstrate a theory. Observation, photographs, model habitats and published classroom data often provide sufficient evidence for inquiry practice. The MOE framework includes care for living things and responsibility towards the environment, alongside scientific objectivity. A well-designed investigation does not need an unsafe or harmful activity to be ‘authentic’. At home, paper scenarios and benign weather observations can be equally effective.
Why a good Science table is transparent
Every table should make it clear what was changed, what was measured, when, and in what units. A caption such as ‘Results of Test A’ is less informative than ‘Volume of water remaining after 60 minutes at different air-flow conditions’. Use row and column headings consistently. If the starting volumes differ, include them rather than giving only final values. If a table includes trial repetitions, show how the summary is calculated. Pupils should be able to reconstruct the comparison without needing to guess what an unlabeled number represents.
How to write an evaluation answer precisely
A useful response has three components: identify the problem, explain the consequence for interpretation, and propose a specific repair. For example: ‘The two containers have different exposed surface areas, so the result cannot isolate the effect of wind. Use identical containers and keep the starting water and temperature comparable while changing wind speed.’ This is better than ‘The experiment is not fair’ because it identifies the flaw and fixes it. The purpose is to make the reasoning testable, not to fill lines with generic advice.
How to recognise when there is insufficient evidence
Some tasks tempt pupils to invent facts the graph does not report. If the graph contains only plant height, it cannot by itself measure root absorption rate. If the table reports ‘bright’ and ‘dim’, it does not give an exact current reading. If a food web omits alternative food sources, the child cannot assume they exist, but also should not overstate certainty about what happens over time. Good scientific judgement distinguishes what the data establish, what is a reasonable prediction and what additional observation would be needed.
Use graphs to check your own story
After explaining a result, ask if its direction agrees with the plotted pattern. If a line shows temperature falling, an answer claiming heat flowed into the object from cooler surroundings may contradict the case. If a bar shows a larger final water volume, don’t immediately claim faster evaporation. If measured shadow size increases as an object moves, verify that the diagram’s actual movement matches the proposed mechanism. A final glance at axes and units can catch an entire category of reasoning errors before submission.
How parents can test understanding in five minutes
Give the learner a tiny invented table with two starting and ending values. Ask what was measured, what deliberately changed, which factors should be constant, and what conclusion is justified. Then deliberately make the experiment unfair by changing a second condition and ask whether the conclusion survives. Encourage the pupil to say ‘not enough evidence’ when appropriate and justify why. This builds a habit that will transfer beyond one curriculum chapter to future science, everyday health claims and information encountered online.
When a child needs targeted help
If a pupil can explain evaporation but cannot read a volume table, assigning more evaporation facts is unlikely to repair the graph skill. If the data are read correctly but the student cannot identify a control variable, teach fair-test logic using a simpler context. If the design is correct but answers are vague, practise the exact link from evidence to mechanism. A good tutorial records which step fails, demonstrates a repair and checks a new unprompted example later. Progress means the student can reason with different objects and data independently.
Original fair-testing and data-analysis question clinic
The questions in this section are new examples created to teach inquiry skills. They are not actual SEAB questions. Each one can be answered without unsafe equipment; a teacher’s diagrams or a parent-drawn data table are enough. The ‘changed-context’ prompt is essential because a child might otherwise memorise the sample result rather than understand the method.
1. Wind and evaporation
Situation: Two equal water dishes are observed for 40 minutes, one with moving air and one with still air. Question: Identify the changed condition and a suitable measured result.
Model answer: Air movement is intentionally changed; water volume lost over the same period is a suitable response to measure.
Typical mistake: The mistake is naming water amount as the changed variable even though the starting amounts are equal.
Independent extension: Keep temperature and exposed surface area comparable, then test a new air speed. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
2. Changing two conditions
Situation: Dish A is wider and warmer than Dish B, and it loses more water in one hour. Question: Can temperature alone be credited?
Model answer: No. Exposed surface area also differs, so the test does not isolate temperature’s effect.
Typical mistake: A generally correct rule about warmth is not evidence that warmth was the only cause here.
Independent extension: Use equally wide dishes while deliberately varying one temperature condition. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
3. Replicating an unfair test
Situation: A pupil repeats the same two-factor comparison six times. Question: Did it become a fair test?
Model answer: No. Repetition may help assess consistency, but the confounding difference remains.
Typical mistake: Confusing reliability with controlling variables prevents a valid causal claim.
Independent extension: Modify the setup before discussing repeat measurements. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
4. Growth increase
Situation: Plant A grows from 12 cm to 17 cm; Plant B grows from 9 cm to 15 cm. Question: Which increased more?
Model answer: A grew 5 cm and B grew 6 cm. B increased by 1 cm more.
Typical mistake: Comparing final heights would incorrectly favour A.
Independent extension: Create a third plant and rank growth changes independently of final size. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
5. Unequal observation time
Situation: A towel loses 8 g of water in one hour while another loses 15 g over three hours. Question: Which has greater average loss per hour?
Model answer: The first: 8 g/h versus the second’s 5 g/h.
Typical mistake: The larger total loss need not mean a higher rate.
Independent extension: Equalise observation time or calculate both rates before concluding. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
6. Shadow experiment
Situation: A torch, cup and screen are used. The cup moves closer to the torch and the screen stays fixed. Question: Which factor was deliberately changed?
Model answer: The distance between the cup and torch changed; shadow size can be measured.
Typical mistake: The pupils may name screen colour even though it never changes.
Independent extension: Change object-screen distance in another comparison and redraw the rays. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
7. Shadow with two movements
Situation: The torch moves nearer while the screen also moves farther away. Question: Can one movement’s effect be isolated?
Model answer: Not without separating the changes, because both may affect shadow geometry.
Typical mistake: Knowing light travels in straight lines is not enough to validate an uncontrolled test.
Independent extension: Hold the screen fixed while changing only one distance. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
8. Light availability in plants
Situation: Two similar seedlings receive different light levels but comparable water and soil. Question: What is the intended changed factor?
Model answer: Available light is deliberately varied; a relevant measured outcome must be specified.
Typical mistake: Using ‘plant health’ without a measurable definition makes interpretation weak.
Independent extension: Choose an appropriate teacher-approved indicator and state what it actually measures. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
9. Water availability and photosynthesis
Situation: Plant species and light are comparable but watering amounts differ. Question: Which factor is being investigated?
Model answer: Water availability; other relevant conditions should be held comparable.
Typical mistake: The pupil may mistakenly list water as an unchanged variable.
Independent extension: Explain how a prediction would follow if water becomes insufficient. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
10. Starch test claims
Situation: A school-provided leaf diagram shows a positive starch-test result. Question: Can the pupil calculate photosynthesis rate from the colour alone?
Model answer: No. The test gives evidence of starch under its conditions, not an exact rate without further measurements.
Typical mistake: A positive outcome should not be treated as a numerical measurement it does not provide.
Independent extension: Ask what extra timing and quantitative data would be necessary for a rate claim. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
11. Conductor experiment
Situation: A gap in a simple circuit is filled with metal and the bulb lights. Question: What does the observation support?
Model answer: Under a working comparable setup, the metal sample can conduct electricity and complete the path.
Typical mistake: The circuit must be checked: a faulty contact may imitate insulation.
Independent extension: Use a known conductor as a control and change only the sample. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
12. Faulty bulb control
Situation: Every tested object seems non-conducting because the bulb is damaged. Question: What should be checked first?
Model answer: Verify the circuit works using a known conductor before judging unknown materials.
Typical mistake: The mistaken conclusion attributes a system failure to all samples.
Independent extension: Repair the bulb, establish a baseline, then retest fairly. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
13. Different bulb types
Situation: A circuit comparison changes cell number and bulb model together. Question: Is the effect of battery count isolated?
Model answer: No. Bulb characteristics also differ and may affect brightness.
Typical mistake: A bright appearance alone does not show which changed factor caused it.
Independent extension: Keep bulb type and other circuit conditions comparable. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
14. Friction on surfaces
Situation: A block is launched from identical starting conditions onto carpet and tile. Question: What is changed and measured?
Model answer: Surface type changes; distance travelled before stopping is one possible response measure.
Typical mistake: Different pushes would confound the effect of surface type.
Independent extension: Explain why launch method and object must remain comparable. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
15. Different masses
Situation: One heavy box on carpet moves less than a lighter box on tile. Question: Is tile automatically the sole reason?
Model answer: No. Object mass, surface and perhaps applied force differ.
Typical mistake: A single observation does not isolate friction from other differences.
Independent extension: Use the same object and a consistent starting method. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
16. Temperature units
Situation: A cup cools from 58°C to 44°C over ten minutes. Question: What change can be directly reported?
Model answer: Its temperature decreased by 14°C over ten minutes.
Typical mistake: Degrees Celsius here are not a unit for quantity of heat transferred.
Independent extension: Explain the direction of heat transfer without inventing an energy value. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
17. Water-loss subtraction
Situation: Container A holds 60 mL initially and 37 mL later. Question: How much water is no longer in the container?
Model answer: 23 mL, calculated from 60 minus 37.
Typical mistake: A careless student may report 37 mL as the loss.
Independent extension: Use a second container with a different starting amount to test understanding. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
18. Different scale axes
Situation: One graph uses 0–50 units and another 0–500 units with identical page height. Question: Can the visual steepness be compared directly without reading scales?
Model answer: No. Axis scales and units must be considered before quantitative comparison.
Typical mistake: The larger-looking line may be an artefact of graph presentation.
Independent extension: Read numerical values from both axes and compare actual changes. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
19. Bar graph labels
Situation: A bar labelled water remaining is tallest for Container C. Question: Which container has the most remaining water?
Model answer: C, if the chart shows comparable units and time. This is not necessarily the greatest water loss.
Typical mistake: The question asks about what is left, not what evaporated.
Independent extension: Add starting amounts and calculate actual water lost. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
20. Unusual thermometer reading
Situation: Repeated room readings are 23°C, 23°C, 45°C, 24°C. Question: What action is scientifically responsible?
Model answer: Check the instrument, conditions and recording of the 45°C reading; repeat if appropriate and disclose the original.
Typical mistake: Discarding a result because it disagrees is not objective evaluation.
Independent extension: Ask how to document an error correction transparently. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
21. Average of measurements
Situation: A simple measurement is repeated with readings 7 cm, 8 cm and 9 cm. Question: Find the arithmetic mean.
Model answer: The average is 8 cm, assuming the readings measure the same quantity under comparable conditions.
Typical mistake: An average of incomparable measurements is not a meaningful summary.
Independent extension: Ask what happens when one measurement was taken using different units. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
22. Mass units mixed
Situation: One trial reports 250 g, another 0.3 kg. Question: Can the numbers be compared as written?
Model answer: Convert units first: 0.3 kg is 300 g, which exceeds 250 g by 50 g.
Typical mistake: Treating 0.3 as less than 250 without units reverses the comparison.
Independent extension: Rephrase both values in kilograms and check the same result. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
23. Observation or inference
Situation: Droplets form outside a chilled bottle. Question: Identify the observation and possible inference.
Model answer: Droplets on the outer surface are observed; cooling of air’s water vapour and condensation explain them.
Typical mistake: The cause is not itself seen by merely looking at droplets.
Independent extension: Give a misted mirror and ask for the observation before its explanation. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
24. Prediction before test
Situation: A pupil expects a wider water surface to lose more water over the same time. Question: What type of statement is this?
Model answer: A prediction based on the exposed-surface-area relationship under comparable conditions.
Typical mistake: It is not an observation until a suitable measurement is made.
Independent extension: Ask for a controlled comparison that could test the prediction. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
25. Unsupported absolute conclusion
Situation: One plant does poorly in a shaded corner. Question: Can this prove all plants die without direct sunlight?
Model answer: No. Other factors and the distinction between light availability and direct sunlight matter.
Typical mistake: An absolute statement exceeds a single observation.
Independent extension: Suggest a careful comparison of suitable light conditions and plant types. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
26. Reading a line plateau
Situation: A line graph shows a water level constant across one measured interval. Question: What is directly observed?
Model answer: The recorded level did not change over that interval at the graph’s resolution.
Typical mistake: Do not assert that no microscopic activity occurred or that every later interval must be unchanged.
Independent extension: Ask for one limitation of the measurement precision. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
27. Time intervals unequal
Situation: The graph records results at 0, 1, 3 and 6 hours. Question: How should a pupil compare rates?
Model answer: Account for unequal time intervals instead of treating consecutive plotted points as equally spaced in time.
Typical mistake: A steep-looking segment without scale consideration can mislead.
Independent extension: Calculate average change per hour across two different intervals. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
28. Evidence of leaks
Situation: A container loses water but a crack is visible. Question: Can all loss be called evaporation?
Model answer: No. Leakage is another plausible path, so evaporation is not isolated.
Typical mistake: The child may trust a result simply because it has a measured number.
Independent extension: Replace the cracked container before comparing evaporation conditions. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
29. Two outcome measures
Situation: A seedling investigation records height and leaf colour. Question: Should both be treated as identical outcomes?
Model answer: No. Growth in height and leaf colour reflect different observed properties and may have different causes.
Typical mistake: The word ‘health’ does not make two measurements interchangeable.
Independent extension: Specify which measurement tests the proposed hypothesis. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
30. Choosing a measuring tool
Situation: A pupil wants to compare the volume of water in two cups. Question: Which type of tool is appropriate?
Model answer: A suitable graduated measuring container can measure liquid volume with relevant units.
Typical mistake: A ruler alone measures length, not liquid volume without an additional justified method.
Independent extension: Check scale markings and read at eye level in a safe classroom context. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
31. Apparatus limitation
Situation: A measuring cylinder has markings only at 10 mL intervals. Question: Can a pupil claim 0.01 mL precision?
Model answer: No. The instrument cannot justify an unrealistically precise reading.
Typical mistake: Writing more decimal places does not improve measurement accuracy.
Independent extension: Select a tool suitable for the required precision and describe its limitations. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
32. Graphs and missing labels
Situation: A plotted curve shows values but no unit or y-axis label. Question: What prevents full interpretation?
Model answer: The measured quantity and unit are unknown, so a valid numerical conclusion is limited.
Typical mistake: A good-looking curve is not a substitute for a labelled measurement.
Independent extension: Write the missing headings needed to make the graph interpretable. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
33. Food web numerical counts
Situation: A class records 15, 20 and 18 insects in three small sample areas. Question: Does this establish every insect population in the park?
Model answer: No. Those counts describe only the observed sample areas under the method used.
Typical mistake: The temptation is to extend a local sample into a universal count.
Independent extension: State what sampling information would be needed for a broader estimate. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
34. Fairness when observing animals
Situation: Two bird feeding sites are compared at different times of day with different food types. Question: Can time alone explain count differences?
Model answer: No. Time and food type differ, and bird availability may also vary.
Typical mistake: A field observation with multiple changes cannot isolate one factor.
Independent extension: Use non-invasive observation and hold planned factors comparable where possible. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
35. Everyday air quality claim
Situation: An advertisement says one room plant removes all indoor air pollutants based on one observation. Question: Is the evidence enough?
Model answer: No. A broad claim requires relevant pollutant measurements, controls and repeated evidence across conditions.
Typical mistake: A real biological process does not guarantee unlimited real-world effectiveness.
Independent extension: Ask the child to name which outcome should be measured before making the claim. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
36. Explain improvement exactly
Situation: A student is asked to improve a test with different container shapes and wind levels. Question: What is a specific method improvement?
Model answer: Use containers with equal exposed surface areas and comparable starting volumes and temperatures while changing only wind.
Typical mistake: ‘Do the test properly’ is not actionable scientific design.
Independent extension: Ask which measurement should be taken at the same time for both trials. Ask which information from the new setup changes the response and which scientific claim remains supported.
Evidence check: make the learner name the deliberate change, measured outcome and a suitable comparison condition where relevant. Then ask whether the response is observation, interpretation, prediction or evaluation. If the data are insufficient, a reasoned ‘not enough evidence’ is more accurate than inventing a cause.
Ten-step improvement plan across Primary Science topics
Step 1: map one actual error
Take a returned school paper and sort the issue into wrong variable, misunderstood instrument, graph scale, calculation, conclusion or explanation. Diagnose the first wrong step before adding more homework.
Step 2: teach the three roles plainly
Ask what is deliberately changed, what is measured and what stays comparable. Use cards or a one-row table rather than introducing an entire vocabulary lecture about independent and dependent variables.
Step 3: test the graph-reading routine
Give a small labelled line graph. Ask for the axis names, units, exact two readings, and direction of the change before discussing any science process.
Step 4: practise a fair and an unfair design
Show two near-identical experiments, one where only a single factor changes and one where two factors change. Ask why the conclusions differ in strength.
Step 5: practise a measured difference
Use simple starting and ending figures. Make the pupil calculate growth, loss or change correctly, then change the initial values so visual shortcuts no longer work.
Step 6: separate data from explanations
Ask the pupil to write ‘what the results show’ and ‘why they may show it’ in two sentences. Check that the second depends on the first and fits the actual controls.
Step 7: identify one limitation
Introduce an unusual reading or a faulty instrument. Ask what should be checked and why deleting the reading without explanation would be unreliable.
Step 8: change the chapter
Repeat the same fair-test reasoning using circuits, shadows or plants instead of evaporation. Improvement is real when the skill transfers beyond the first context.
Step 9: check the learner independently
Wait until another day, provide a novel mixed table and ask the child to identify variables, perform one calculation and evaluate the conclusion without prompts.
Step 10: use the result for the next week
If the child calculates correctly but still assumes causation from any pattern, revise controls rather than arithmetic. If axis reading fails, practise scales rather than memorising more chapter definitions.
Ready-to-copy investigation planning table
| Investigation question | Changed condition | Measured result | Keep comparable |
|---|---|---|---|
| Does wind affect evaporation? | Air movement | Water lost in a fixed time | Container area, starting volume, temperature |
| Does distance affect shadow size? | One specified distance | Shadow size on screen | Object, light source, other distances |
| Does number of cells affect bulb behaviour? | Number of suitable series cells | Brightness or approved measurement | Bulb type, wires, connections |
| Does surface type affect sliding? | Surface material | Distance or motion outcome | Object and starting method |
Frequently asked questions
What is a fair test in Primary Science?
An investigation in which the intended factor is deliberately changed while other relevant conditions are kept comparable, so the measured result can be interpreted more confidently.
Does MOE require the words independent and dependent variables?
The 2023 Primary Science progression states ‘changed/unchanged variables’. Technical labels may be useful teaching aids, but the central syllabus skill is recognising and designing fair tests.
What should P4 pupils already know?
MOE expects pupils by the end of P4 to recognise a fair test and conduct guided investigations, including simple analysis of data and information.
What changes by P6?
By the end of P6 the progression includes designing fair tests, using and selecting apparatus, analysing data across tables, bar and line graphs, charts and diagrams, and defending ideas with evidence.
Is repeating a test the same as making it fair?
No. Repetition can show consistency but cannot remove a second uncontrolled changed factor.
How do I read a graph correctly?
Inspect title, axis names, units, scales, legend, and actual data values before describing patterns or inferring causes.
Should every unexpected reading be deleted?
No. Check the instrument and conditions; record the anomaly and justify whether a repeat measurement or correction is needed.
Can a Foundation Science pupil use this guide?
Yes, for the inquiry skills within their syllabus, but use the course-specific MOE topic outcomes and SEAB Foundation 0039 assessment format.
Are these genuine PSLE questions?
No. All exercises are original teaching scenarios, not copyrighted national-paper reproductions or forecasts of the unreleased paper.
What is the quickest way to improve?
Correct the first reasoning error in one familiar situation, then test the same skill in a changed context a day later. Quantity of worksheets is not a substitute for independent evidence.
Sources and next eduKate reading
- MOE 2023 Primary Science Syllabus: Ways of Thinking and Doing, Table 1, pages 16–19
- SEAB official PSLE Science and Foundation Science 2026 exam specifications
- Open-ended answering: explanations, command words and common mistakes
- Heat and water cycle: 29 original worked data and reasoning questions
- Photosynthesis and plant systems: evidence-based investigations
- eduKate reference: diagnosis, sequenced practice and tutorials
Verified 12 November 2026. Official learning outcomes come from the 2023 MOE Primary Science syllabus, and examination-format statements refer to the 2026 SEAB documents applying from 2026. This independent guide does not assert that any example is an actual or guaranteed future PSLE question; check the specific 2027 examination documents when available.