Small Group Tutorials

Here to help students catch up, keep up, and move ahead. Book a consultation here.

How to Answer State, Describe, Explain, Compare and Predict Questions in PSLE Science

Wait, What? Two Questions Can Be About the Same Science and Still Need Different Answers

A learner can know the concept and still answer the wrong question.

Imagine a diagram showing a plant under two different conditions. One question asks what happened. Another asks why it happened. Another asks what would happen if a condition changed. The scientific topic may be identical, but the reasoning job is not.

The first skill is not writing more. It is recognising what kind of scientific response the question is asking you to produce.

Quick Answer

For PSLE Science, treat command words as signals about the job your answer must do. State usually asks for the required fact, result or conclusion. Describe asks you to report a pattern, change or feature from the evidence. Explain asks you to connect a cause or scientific mechanism to the outcome. Compare asks you to place two cases against the same basis. Predict asks you to use the evidence, pattern or concept to state a likely outcome under the stated condition.

This is a reading-and-reasoning guide, not a universal marking-code glossary. SEAB’s official 2026 Science syllabus states the broader assessment objectives—application, prediction/hypothesis, interpretation/analysis, evaluation, and communicating explanations and reasoning—but it does not mean every school must use one rigid sentence template for every command word.

The Official PSLE Science Frame

For examination from 2026, the PSLE Science Paper assesses attainment in the 2023 Primary Science syllabus. The official assessment objectives include knowledge with understanding and application of knowledge and scientific inquiry. Scientific inquiry includes making predictions and formulating hypotheses, interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning.

That official frame matters because a strong answer is not merely a collection of scientific words. It must do the reasoning job requested by the question.

Owned PSLE Science Learning Job

  • Recognise the response job before answering.
  • Separate observation from explanation.
  • Match the amount of reasoning to the question.
  • Use evidence when the question gives evidence.
  • Compare using the same scientific basis.
  • Predict from a stated condition rather than from memory alone.
  • Avoid over-answering a simple question and under-answering an explanation question.
  • Check that the final sentence actually answers the command.

The PSLE Science Response Chain

READ THE COMMAND → IDENTIFY THE OBJECT OR RELATIONSHIP → READ THE GIVEN EVIDENCE → SELECT THE RELEVANT CONCEPT → PRODUCE THE REQUIRED RESPONSE TYPE → CHECK THAT THE ANSWER MATCHES THE COMMAND.

The command word does not replace Science. It tells you what to do with the Science.

1. State — Give the Required Scientific Result Clearly

A “state” question usually asks for a direct response: a value, observation, relationship, name, result or conclusion that the evidence supports.

Suppose a table shows that Plant A grew 8 cm while Plant B grew 3 cm over the same period. If asked to state which plant grew more, the direct answer is Plant A. You do not need to invent a long explanation unless the question asks why.

Common State Error: Explaining Instead of Answering

Learners sometimes write several sentences and bury the required result. This increases the chance of adding an unsupported statement.

Repair: write the direct scientific result first. Add explanation only if another part asks for it.

2. Describe — Say What the Evidence Shows

“Describe” usually requires you to report an observable pattern, difference, trend, sequence, shape or relationship. The safest source is the diagram, table, graph or observation given.

Example: a graph shows temperature rising from 25°C to 60°C over five minutes. A description could state that the temperature increased over the five-minute period. If the graph shows different rates, plateaus or reversals, include those features if they matter.

Description answers “What does the evidence show?” Explanation answers “Why did that happen?”

Common Describe Error: Sneaking In an Unproven Cause

If the graph shows a decrease, you may describe the decrease without knowing why it occurred. A cause becomes part of the answer only when the question or evidence supports it.

3. Explain — Build the Causal Bridge

An explanation normally needs more than the final fact. It connects a scientific mechanism or relationship to the condition and outcome in the question.

A useful Primary-level chain is:

GIVEN CONDITION → RELEVANT SCIENTIFIC PROCESS OR PROPERTY → CHANGE → OUTCOME.

Suppose identical wet cloths are placed in different conditions and one dries faster because it is spread out more. A complete explanation should connect the larger exposed surface area to faster evaporation and then to less liquid water remaining after the same time.

Common Explain Error: Concept Name Without Mechanism

“Because of evaporation” may be relevant but incomplete if the question asks why one condition changes the rate. The answer needs the relationship between the condition and the process.

4. Compare — Use the Same Basis for Both Cases

A comparison becomes scientifically useful only when both cases are compared using the same variable, property, feature or outcome.

Weak comparison: “Plant A is in sunlight. Plant B grew less.” This changes the basis midway.

Stronger comparison: “Plant A increased in height more than Plant B over the same period.” If the question then asks for a reason, add the scientific explanation separately.

The Two-Sided Test

  • Did I mention both objects or conditions?
  • Did I compare the same measured or observable feature?
  • Did I state the direction clearly: greater, less, faster, slower, higher, lower, more, fewer?
  • Did I accidentally explain when only comparison was required?

5. Predict — Carry the Science Into a New Condition

A prediction is not a guess. It is a proposed outcome supported by a pattern, concept, relationship or prior evidence.

Suppose an experiment shows that increasing the number of identical batteries in a particular allowed arrangement increases the brightness of a bulb within the conditions shown. If a new but similar condition is introduced, the learner may predict the likely outcome—but only if the scientific relationship and circuit arrangement remain relevant.

A safe prediction structure is:

IF THIS CONDITION CHANGES → THE RELEVANT SCIENTIFIC RELATIONSHIP SUGGESTS → THIS OUTCOME IS LIKELY.

Common Predict Error: Continuing a Pattern Beyond the Evidence

Not every trend continues forever. A graph that rises over a small range does not prove indefinite increase. Prediction must stay inside what the concept and evidence can support.

Worked Example — One Setup, Five Different Question Jobs

Two identical containers hold equal amounts of water. Container A has a larger exposed surface area than Container B. After the same time under the same surrounding conditions, Container A contains less water.

  • State: Which container has less water remaining? → Container A.
  • Describe: How did the amount of water compare after the same time? → Less water remained in A than in B.
  • Explain: Why did A have less water? → Its larger exposed surface area allowed faster evaporation under the stated conditions, so more liquid water changed into water vapour during the same time.
  • Compare: Compare the amount of water remaining. → A had less water remaining than B.
  • Predict: If A’s exposed surface area were made even smaller than B’s while other relevant conditions stayed the same, predict the likely result. → A would be expected to lose water more slowly by evaporation than B.

The Science did not change. The response job changed.

Command Words Are Not Magic Passwords

Do not memorise a fixed number of sentences for each command word. A one-mark direct response and a multi-step structured question can use the same command word but require different depth because the evidence and task are different.

The stronger rule is: identify the scientific job, then give enough reasoning to complete that job and no unsupported extra claim.

Observation, Inference and Explanation Inside Command Words

A question may ask you to describe an observation, infer an unseen process, or explain why the process produces the result. These are not interchangeable.

  • Observation: what was seen, measured or directly reported.
  • Inference: what the evidence suggests about something not directly observed.
  • Explanation: the scientific relationship that connects cause, process and outcome.

If you write an inference as though it were directly observed, your reasoning becomes less precise.

How to Read Multi-Part Questions

Multi-part questions often change the job from part to part. Part (a) may ask you to state. Part (b) may ask you to explain. Part (c) may ask you to predict under a changed condition.

Reset before every part:

  • What is the command now?
  • What new evidence is introduced?
  • What earlier information still applies?
  • What exact output is required?

Failure Signatures and Earliest Repairs

“I wrote a correct fact but it was not enough.” The weak link may be response type. Ask whether the question wanted an explanation rather than a statement.

“I explain everything and run out of time.” The weak link may be over-answering. State direct results directly.

“My comparison is unclear.” The weak link may be inconsistent basis. Compare the same variable for both cases.

“My prediction sounds reasonable but is not supported.” The weak link may be evidence boundary. State the scientific relationship that supports the prediction.

A Six-Question Checking Receipt

  • What command word is used?
  • What scientific object or relationship is being asked about?
  • What evidence is given?
  • What concept is relevant?
  • Did I produce the correct response type?
  • Does my last sentence answer the exact question?

Retrieval Practice

Take one unfamiliar diagram and write five different questions about it: state, describe, explain, compare and predict. Then answer each one. This teaches you that the same evidence can support different scientific jobs.

Unfamiliar Transfer Test

Use a topic you have not revised recently. If you can identify what a command asks you to do before recalling the topic details, the reading skill is becoming independent of familiar chapter wording.

Delayed Independent Return Test

Two days later, take five mixed questions without notes. Before solving, write a tiny label beside each: S / D / E / C / P. After answering, check whether your response actually performed that job.

Parent and Tutor Teaching Guide

When a child gives a weak answer, do not immediately provide a model sentence. Ask first: “What is this question asking you to do—state, describe, explain, compare or predict?” Then ask: “What evidence belongs in that type of answer?”

This diagnosis separates a Science-knowledge failure from a response-control failure. A learner may know the concept but answer the wrong job.

Authoritative References

The Quiet Ending

Strong PSLE Science answers begin one step before writing. First decide what kind of thinking the question asks for. Once that is clear, the Science has somewhere precise to go.