How to Keep a PSLE Science Investigation Question, Prediction and Conclusion in Different Jobs

Wait, What? A Good Prediction Is Allowed to Be Wrong

A learner predicts that Set-up P will produce the greatest measured change. The results later show Set-up Q produced the greatest change. The learner quietly rewrites the original prediction so that it now says Q.

The page looks cleaner, but the scientific reasoning has been damaged. A prediction belongs before the result. A conclusion belongs after the result. If the prediction is changed after the data are known, the investigation loses the useful comparison between what was expected and what was actually observed.

The investigation question has a different job again. It tells you what relationship is being investigated. Strong PSLE Science inquiry keeps these three roles connected but separate.

Quick Answer

The investigation question asks what relationship will be tested. The prediction or hypothesis states what you expect before seeing the results, with a scientific reason when required. The conclusion states what the actual evidence supports after the investigation. All three should refer to the same scientific variables and conditions, but they are not interchangeable.

The reasoning chain is: QUESTION → IDENTIFY VARIABLES AND CONDITIONS → PREDICT BEFORE RESULTS → RUN/READ THE INVESTIGATION → INTERPRET THE DATA → COMPARE RESULT WITH PREDICTION → CONCLUDE ONLY WHAT THE EVIDENCE SUPPORTS → EXPLAIN THE MECHANISM IF REQUIRED.

Owned PSLE Science Learning Job

This guide owns one exact job: keeping the scientific roles of investigation question, prediction/hypothesis and conclusion distinct while aligning them to the same tested relationship.

It does not replace the existing guides on planning an investigation, answering prediction and hypothesis questions, comparing a prediction with the actual result, or writing an evidence-bounded conclusion. Those owners remain canonical. This page teaches the handoff between the three inquiry roles.

Why This Matters in the Current PSLE Science Frame

The 2026 PSLE Science examination assesses attainment in the 2023 Primary Science syllabus. SEAB includes scientific inquiry such as making predictions or hypotheses, interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning.

These are not one undifferentiated “experiment skill”. Scientific inquiry has stages. The meaning of a statement depends partly on when it is made and what evidence is available at that point.

The Three Jobs at a Glance

Scientific role	When?	Main question	Evidence available?
Investigation question	Before testing	What relationship are we trying to find out?	Background information may be available, but results are not yet known
Prediction / hypothesis	Before results	What do I expect under these conditions, and why?	Prior knowledge, patterns or supplied information
Conclusion	After results	What do the actual observations or measurements support?	The investigation results

The same variables can appear in all three. What changes is the epistemic job: question, expectation, then evidence-based judgement.

Worked Example 1: One Relationship, Three Roles

An original investigation varies the number of layers of an insulating material around identical cups and measures the decrease in water temperature after the same time.

Investigation question: How does the number of insulating layers affect the decrease in water temperature over the stated time?

Prediction: The cup with more insulating layers will show a smaller temperature decrease because the insulation reduces heat transfer to the surroundings.

Conclusion after results: Across the tested numbers of layers, cups with more layers showed smaller temperature decreases, if that is what the measured data actually show.

Notice how the prediction may contain a mechanism before the results, while the conclusion must be checked against the data. If the data show the opposite pattern, the conclusion must follow the data rather than loyalty to the prediction.

Worked Example 2: The Prediction Is Wrong but the Investigation Still Teaches You Something

A learner predicts that Material A will allow the greatest measured outcome because of a remembered property. The actual results show Material C has the greatest outcome.

The correct response is not to erase the prediction. Record the mismatch:

Prediction: A would be greatest.
Actual result: C was greatest among the tested materials.
Conclusion: C produced the greatest measured outcome under the tested conditions.
Next reasoning job: check whether the original scientific reason was wrong, whether another condition mattered, or whether the evidence needs further investigation.

Science improves when expectations are allowed to meet evidence honestly.

A Prediction Is Not a Promise

A prediction says what should happen if your current understanding and the stated conditions are adequate. It is not a commitment that the result must agree.

When learners treat a prediction as something that must be “correct”, they may ignore awkward data or reinterpret results to protect the answer. A scientifically useful prediction creates a testable expectation, not a mark of personal success.

The Investigation Question Must Not Contain the Answer

Compare these:

Question: How does the number of insulating layers affect temperature decrease?
Loaded version: Why do more insulating layers always reduce the temperature decrease?

The second version already assumes the direction and universality of the result. A scientific question should leave room for the investigation to establish the relationship.

Some school questions may supply an expected relationship for students to test, but the learner should still distinguish the claim being tested from the result observed.

The Prediction Must Answer the Question You Actually Asked

If the question asks how light level affects a measured growth outcome, a prediction about water amount answers a different question. Even scientifically true information becomes irrelevant when it does not match the tested relationship.

Before writing the prediction, name:

the deliberately changed condition;
the measured outcome;
the expected direction or result;
the scientific reason, if required.

The Conclusion Must Answer the Same Question Again

A conclusion is not a summary of everything that happened. It should return to the relationship the investigation was designed to test.

If the question asks how surface area affects a measured outcome, a conclusion about colour change is not sufficient unless colour change was the stated measured evidence for that outcome. Keep the variables aligned from question to conclusion.

Worked Example 3: Prediction About the Wrong Outcome

Question: How does the length of time under Condition X affect the mass of Object P?

Weak prediction: “Object P will become drier.”

Why weak? The investigation measures mass. “Drier” may be a possible interpretation, but it is not yet tied to the measured quantity.

Stronger prediction: “As the time under Condition X increases, the measured mass of P will decrease,” followed by a scientifically appropriate reason if the question requires one.

The prediction is now testable with the stated measurement.

Worked Example 4: Conclusion That Overclaims Beyond the Tested Range

An investigation tests 1, 2, 3 and 4 layers. The results show smaller temperature decreases with more layers across those values.

Weak conclusion: “The more layers there are, the water will always lose less heat.”

Stronger conclusion: “Across the tested values from 1 to 4 layers, increasing the number of layers was associated with a smaller temperature decrease over the stated time.”

The latter conclusion respects the range actually tested and does not convert one school investigation into a universal physical law.

Question, Prediction and Conclusion Can Use Different Tenses Without Changing the Science

Question: How does X affect Y?

Prediction: If X increases, I expect Y to decrease because…

Conclusion: As X increased across the tested values, Y decreased.

The wording changes because the evidence state changes. The underlying variables remain aligned.

Prediction and Hypothesis: Use the Official Frame Carefully

SEAB’s inquiry frame includes predictions and hypotheses. In Primary Science teaching, schools may use these terms with different levels of formality. Do not invent one universal sentence pattern.

The important scientific discipline is that the expectation is stated before the outcome is known and is connected to a testable relationship or condition. A hypothesis may be framed as a testable relationship or explanation; a prediction states an expected outcome under specified conditions.

When the exact school task uses one term, answer that task rather than arguing about labels.

Worked Example 5: Same Variables, Different Evidence Role

Stage	X	Y	What is known?
Question	Changed condition	Measured outcome	Relationship is to be investigated
Prediction	Expected change in X	Expected response in Y	Expectation based on current knowledge
Conclusion	Actual tested values of X	Actual measured Y	Relationship supported by results

This table shows why the variables should stay recognisable from beginning to end even though the evidence role changes.

Do Not Rewrite a Prediction After the Result

If a prediction was poorly worded, you can annotate it during correction: “My original prediction did not specify the measured outcome.” But preserve what you actually predicted before seeing the result.

Then write a corrected future prediction for a new question. That keeps learning honest while still improving the skill.

Do Not Force the Conclusion to Agree With the Prediction

The conclusion belongs to the evidence. If the data contradict the prediction, say so. Then investigate why.

Possible reasons include:

the scientific idea in the prediction was wrong or incomplete;
a relevant condition was overlooked;
the measurement method had limitations;
the result included variation or an unusual observation;
the prediction applied outside the conditions actually tested.

Do not assume a mismatch proves the experiment was “bad”. The mismatch itself is information to analyse.

Worked Example 6: Unexpected Result

Question: How does Condition X affect the measured time?

Prediction: Greater X will produce a shorter time because of the learner’s current scientific model.

Result: Greater X produces a longer time across the tested values.

Conclusion: The measured time increased as X increased across the tested range.

Next step: revisit the mechanism or method. The conclusion must not be rewritten to protect the prediction.

Result, Conclusion and Explanation Are Also Different

The raw result may be a set of numbers. The conclusion compresses those results into an evidence-bounded relationship. The explanation connects that relationship to scientific concepts and mechanisms.

This gives a fuller inquiry sequence:

QUESTION → PREDICTION → METHOD → OBSERVATIONS/RESULTS → CONCLUSION → EXPLANATION → EVALUATION.

Not every school task asks students to write every stage. But when stages appear, keep their jobs distinct.

Earliest Weak-Link Diagnosis

Failure signature	Earliest weak link	Repair
Your prediction answers a different variable than the investigation question.	Question alignment	Rewrite the relationship as changed condition → measured outcome.
You change the prediction after seeing the results.	Evidence timing	Preserve the original; compare it with the result.
Your conclusion repeats the prediction even when data disagree.	Evidence ownership	Build the conclusion only from actual results.
Your conclusion includes an untested universal claim.	Scope	Bind it to the tested objects and range.
Your prediction has a mechanism but no expected measured outcome.	Testability	State what the measurement should show.
You write an explanation where a conclusion is required.	Response role	State the data-supported relationship first.

Misconception Repair: “Correct Prediction” Is Not the Goal of Science

In classroom practice, students can become afraid of making a prediction that turns out to be wrong. That encourages safe guessing or hindsight rewriting.

A useful prediction should be scientifically reasoned and testable. Its value comes partly from what happens when reality answers back. If evidence disagrees, the learner has something to investigate and repair.

A Three-Column Inquiry Check

During practice, use three columns:

Question	Prediction before results	Conclusion after results
What relationship is tested?	What do I expect and why?	What did the data actually support?

Then draw lines under the changed condition and measured outcome. If those variables change identity between columns, the reasoning has drifted.

Question-Reading Protocol

What exact relationship is being investigated?
Which condition is deliberately changed or compared?
Which outcome is measured?
Is the current task asking for a prediction, a result, a conclusion or an explanation?
At this point in the investigation, what evidence is actually available?
Does my statement refer to the same variables as the investigation question?
Am I preserving the tested range and conditions?

Practice Sequence

Round 1: Match prepared questions, predictions and conclusions to one another.
Round 2: Find examples where the prediction answers the wrong measured outcome.
Round 3: Compare predictions with actual results that agree and disagree.
Round 4: Repair overstrong conclusions without changing the results.
Round 5: Build all three roles for a new unfamiliar investigation, then keep the prediction hidden until after the data are interpreted.

Unfamiliar Transfer Challenge

A fictional material has a property called Z. An investigation tests how condition X affects measured Z. You are told a new rule suggesting that greater X should reduce Z.

Write:

the investigation question;
a prediction based on the supplied rule;
then read fictional results showing Z actually increases;
write the conclusion without rewriting the prediction.

You do not need familiar chapter knowledge to perform the inquiry roles correctly. This tests whether the structure transfers.

Delayed Independent Return Test

Several days later, provide a new investigation set-up and ask the learner to write the question and prediction before revealing the results. After the results are shown, the learner writes a conclusion.

Check whether the same variables remain aligned across all three statements and whether the conclusion follows the evidence even when the prediction is contradicted.

Answer-Checking Receipt

What role am I writing: question, prediction or conclusion?
Am I using the same changed condition and measured outcome throughout?
Was the prediction made before the results?
Does the prediction state an expected measurable outcome?
Does the conclusion come from the actual data rather than the prediction?
Is the conclusion limited to the tested evidence?
If the prediction was wrong, did I preserve the mismatch and learn from it?
If an explanation is required, did I add the mechanism separately?

Common Traps

Putting the expected answer inside the investigation question.
Predicting a different outcome from the one actually measured.
Editing the prediction after seeing the result.
Forcing the conclusion to agree with the prediction.
Writing a mechanism instead of a conclusion.
Making a universal conclusion from a narrow tested range.
Treating an incorrect prediction as proof that the whole investigation failed.

Parent and Tutor Teaching Guide

During home or tuition practice, physically cover the results before asking for a prediction. This protects the timing of the reasoning and prevents hindsight.

After the results are revealed, do not ask first, “Was your prediction correct?” Ask, “What do the results show?” Then compare the conclusion with the earlier prediction.

Praise accurate revision of understanding, not merely correct anticipation. A learner who says, “My prediction was wrong because I overlooked this condition” is doing more scientific thinking than a learner who guessed the right direction without a mechanism.

Keep the language flexible. Do not force one school-specific template unless the child’s teacher has explicitly required a local format. The scientific roles matter more than memorising one sentence pattern.

Useful Internal Routes

Authoritative References and Evidence Boundary

The role-separation framework in this article is an educational scaffold, not an official national-examination response template. Prediction and hypothesis terminology can be taught at different resolutions in different classrooms; learners should follow the wording and evidence demands of the actual task while preserving the underlying scientific roles.

The Quiet Return

A scientific investigation is a conversation between a question, an expectation and the world.

Ask the question clearly. Make the prediction before you know the answer. Then let the evidence write the conclusion. When those three jobs stay distinct, being wrong becomes useful because reality is allowed to correct the model.

Continue learning: Return to PSLE Science: investigations, variables and fair tests, or choose English, Mathematics and Science.

Small Group Tutorials