How to Tell a Replication From an Extension in a PSLE Science Follow-Up Investigation

Wait, what? “Do the investigation again” and “test it under a new condition” are not the same scientific follow-up.

Both can be useful. But they answer different questions. One asks, “Does this result appear again when the same scientific question is tested comparably?” The other asks, “Does the relationship still hold when we deliberately broaden or change what is tested?” If you mix those jobs, you may collect more data but learn the wrong thing from them.

Quick Answer

In this guide, a replication-style follow-up means repeating essentially the same scientific question under suitably comparable conditions to check whether the result or pattern is consistent. An extension-style follow-up deliberately adds a new range, condition, case, specimen type or context to test whether the earlier conclusion travels further. Replication can strengthen confidence in consistency; extension can broaden the evidence. Neither automatically proves the other.

The words “replication” and “extension” are useful learning labels here. This guide does not claim they are compulsory PSLE command words or official SEAB answer phrases.

Owned PSLE Science Learning Job

This guide owns one learner job: how a Primary 5/6 learner distinguishes a follow-up investigation whose main purpose is to check consistency from one whose main purpose is to broaden the tested scope, then interprets what each kind of evidence can and cannot support.

It does not own generic repeatability, fair tests, experimental design, or generalisation by themselves. Those already have separate owners in the eduKate Science estate. This page connects them through one PSLE Science follow-up decision.

Why This Matters in the Current PSLE Science Frame

The 2026 PSLE Science examination assesses the 2023 Primary Science syllabus. The official assessment objectives include prediction and hypothesis, interpretation and analysis of information, evaluation of observations, information and methods, and communicating explanations and reasoning. Scientific inquiry is therefore not only about carrying out a first test. Learners also need to understand what additional evidence would strengthen, challenge or extend a conclusion.

The Core Mechanism: Follow-Up Design Depends on the Uncertainty Left Behind

Before designing a follow-up, ask what the first investigation still leaves uncertain.

Uncertainty after the first investigation	Most useful follow-up job
“Was this result a one-off, recording problem or unstable outcome?”	Replication-style check under comparable conditions
“Would this relationship still appear at another tested level?”	Extension to a broader range
“Would it also apply to another similar case?”	Extension to a new case or specimen group
“Two explanations still fit the original results.”	A discriminating follow-up that changes the condition needed to separate them
“The method itself was flawed.”	Repair or redesign before treating more repetition as useful evidence

The follow-up is therefore chosen by the remaining scientific uncertainty, not by a rule that “more trials are always better” or “more conditions are always better”.

Replication-Style Follow-Up: What Stays Comparable?

A replication-style follow-up asks substantially the same scientific question again. The variables keep the same roles. The important comparison is preserved. The measurement method remains suitably comparable. The tested range usually remains the same or close enough that the follow-up still addresses the same relationship.

This does not mean every physical detail must be identical. A different day, learner or set of similar specimens may be deliberately used, depending on the question. What matters is that the follow-up is still testing the same scientific claim rather than silently changing it into a broader one.

Extension-Style Follow-Up: What Is Deliberately Broadened?

An extension deliberately moves beyond the original evidence. It might test an additional level of the changed condition, a longer or shorter duration, a new similar specimen group, another environment, or a different but related case.

The purpose is not merely to see the same result again. The purpose is to ask whether the original relationship remains useful when the scope changes. That means the conclusion after an extension must be read carefully. A relationship that survives one new case has broader support than before, but it still does not automatically become universal.

Replication and Extension Strengthen Different Claims

Feature	Replication-style follow-up	Extension-style follow-up
Main question	Does the result appear consistently again?	Does the relationship extend to a broader condition or case?
Scientific question	Essentially the same	Related but deliberately broadened
Tested range	Usually kept comparable	May be deliberately expanded
Evidence strengthened	Consistency/repeatability of the finding	Scope/generalisation of the finding
Major trap	Thinking repeats repair an unfair comparison	Thinking one extra case proves a universal rule

The PSLE Science Reasoning Chain

READ THE ORIGINAL QUESTION AND RESULTS → IDENTIFY WHAT THE EVIDENCE ALREADY SUPPORTS → NAME THE UNCERTAINTY THAT REMAINS → CHOOSE A FOLLOW-UP JOB → PRESERVE OR DELIBERATELY CHANGE THE RELEVANT CONDITION → MEASURE COMPARABLY → INTERPRET THE NEW EVIDENCE → UPDATE THE CONCLUSION WITHOUT OVERCLAIMING.

Worked Example 1: Repeat the Same Conditions or Add a New Level?

An original practice investigation tests three levels of condition X—low, medium and high—and measures outcome Y. The results suggest that Y increases across those three tested levels.

If the learner worries that the results may be unstable because each level was tested only once, repeating the same three conditions with comparable methods is a replication-style follow-up. It asks whether the observed pattern appears again.

If the learner instead wants to know what happens at an even higher level of X, adding a fourth level is an extension. It asks whether the relationship continues beyond the original tested range.

These are not interchangeable. Repeating low, medium and high cannot directly tell you what happens at the new higher level. Testing only the new higher level cannot tell you whether the original three results were consistently repeatable.

Worked Example 2: Same Material, New Material

A student compares two thicknesses of the same insulating material and observes that the thicker sample produces a smaller temperature change over a fixed period.

A replication-style follow-up might repeat those two thickness conditions with another set of comparable samples to check whether the difference appears again.

An extension might test a third thickness, or test whether a similar relationship appears in another material. The second extension broadens the case more strongly because it changes not only thickness but also material identity. The conclusion must therefore be phrased at the level the new evidence supports.

Worked Example 3: More Repeats Cannot Repair the Wrong Comparison

Suppose Set-up A differs from Set-up B in two scientifically relevant conditions at the same time. The learner repeats both set-ups ten times and gets very consistent results.

The repetition may show that the same difference appears consistently. It does not isolate which of the two changed conditions produced that difference. A replication-style follow-up has strengthened consistency but has not repaired the causal comparison.

The useful next investigation may require redesigning the comparison so only the relevant condition differs while other relevant conditions are controlled. This is why repeatability and fairness are different scientific jobs.

Worked Example 4: Extending to Another Similar Specimen Group

An original investigation uses one group of similar specimens and finds a relationship between a tested condition and an outcome. A follow-up uses a second group of similar specimens under the same set of tested conditions.

This follow-up can serve more than one purpose depending on how it is framed. If the main purpose is to see whether the same result can be obtained with another comparable group, it functions like a replication. If the second group deliberately represents a new population or category and the question asks whether the relationship extends to that group, it also carries an extension purpose.

Real scientific designs do not always fit one label perfectly. The learner’s job is to identify the dominant evidence question, not to force every scenario into a vocabulary box.

Worked Example 5: Two Explanations Still Fit

Imagine that the original results are consistent with Explanation A and Explanation B. Simply repeating the same set-up may produce the same ambiguous result again. The follow-up needs a condition under which A and B make different predictions.

That follow-up is not “more data” in the generic sense. It is a discriminating extension or redesign chosen to make the competing explanations separate. This shows why a good follow-up begins from the uncertainty that remains.

Observable Failure Signatures

The learner says “repeat the experiment” no matter what the weakness is.
The learner changes the tested condition and still calls the new run an exact repeat.
The learner adds a new condition but compares it with results measured by a different method.
The learner thinks more repeats make an unfair test fair.
The learner thinks a consistent repeated result proves the causal explanation.
The learner thinks one new tested case proves a universal rule.
The learner cannot state what uncertainty the follow-up is supposed to reduce.
The learner proposes extra measurements that do not distinguish among the remaining possibilities.
The learner broadens several conditions at once and then cannot tell what the extension actually tested.

Earliest Weak-Link Diagnosis

Failure	Earliest weak link	Repair
Cannot decide what follow-up is needed	Uncertainty not identified	State what the first evidence still cannot decide.
Repeats a flawed comparison	Method evaluation failure	Repair the design before repeating it.
Adds new conditions without purpose	Question-design failure	Write the follow-up question first.
Cannot compare old and new results	Measurement alignment failure	Keep quantity, unit, timing and method comparable where required.
Overgeneralises after one extension	Evidence-scope failure	State exactly what new case or range was tested.
Confuses consistency with causation	Claim-type failure	Separate “appears again” from “this condition caused it”.

Misconception Repair

“More repeats make every conclusion stronger.” Repeats help most when the original comparison is meaningful. Repeating a confounded or badly measured design can reproduce the same weakness.

“More test conditions are always better.” Extra conditions are useful when they answer a real extension question. Otherwise they may add complexity without reducing uncertainty.

“A result that repeats must have the proposed cause.” Consistency does not by itself isolate causation.

“If the extension gives the same pattern, it is now true everywhere.” The evidence has travelled one step further, not infinitely far.

A Follow-Up Investigation Protocol

Write the scientific question the first investigation answered.
State what the first evidence supports.
Name what remains uncertain.
Ask whether you need consistency, broader scope, or discrimination between explanations.
If checking consistency, preserve the scientific question and relevant comparison.
If extending scope, state exactly what new range, case or condition will be added.
Keep the outcome measurement comparable unless changing it is part of the new question.
Predict what different follow-up results would mean.
Collect or interpret the new evidence.
Update the conclusion only as far as the combined evidence supports.

How to Read the Follow-Up Result

Follow-up result	Careful interpretation
Comparable repeat gives similar pattern	The pattern appears more consistent under those conditions.
Comparable repeat gives different pattern	Investigate variation, method, conditions and evidence quality before choosing a cause.
Extension gives same pattern in new range	The relationship now has evidence over a broader tested range.
Extension gives different pattern	The earlier relationship may have a boundary, changed condition or different mechanism that needs explanation.
Redesigned discriminating test supports A over B	The new evidence helps distinguish the explanations, within the design’s limits.

Original Practice Sequence

Use short scenarios rather than memorising the two labels.

Round 1: Read an original investigation and state the remaining uncertainty.
Round 2: Choose between repeating the same conditions, adding a new condition or redesigning the comparison.
Round 3: Explain what evidence the chosen follow-up would add.
Round 4: Predict what a confirming and a conflicting result would mean.
Round 5: Switch to a different Science theme while keeping the follow-up job the same.
Round 6: Remove the labels “replication” and “extension” and ask the learner to design the follow-up from the uncertainty alone.

Unfamiliar Transfer Challenge

An unfamiliar system is tested at three levels of condition P. A clear pattern is observed, but each condition was tested only once. The learner is offered three possible next investigations:

repeat the same three levels using comparable measurements;
test a fourth, higher level of P;
repeat the original three levels but change the measurement method at the same time.

The first mainly checks consistency. The second mainly extends the tested range. The third changes two things in the evidence system: it may answer a useful question, but direct comparison with the original results becomes harder unless the methods are calibrated or otherwise connected. The learner should choose based on the uncertainty the question asks them to reduce.

Delayed Independent Return Test

After a meaningful gap, give four follow-up proposals without using the words replication or extension. Ask the learner to write, for each proposal: What uncertainty would this reduce? What stays comparable? What is deliberately new? What stronger claim, if any, could the result support?

Success means the learner can identify the evidence job from the design itself rather than from vocabulary cues.

Answer-Checking Receipt

I can state what the first investigation already showed.
I can state what remains uncertain.
I know whether the follow-up is mainly checking consistency or broadening scope.
I kept the same scientific question when consistency was the goal.
I can name exactly what new range, case or condition an extension adds.
I kept measurements comparable where the comparison requires it.
I did not assume repeated results repair a flawed design.
I did not turn one extension into a universal law.
I can explain what different follow-up outcomes would mean.
I can update the conclusion without exceeding the evidence.

Common Traps

Repeating one trial when the whole comparison is flawed.
Changing several new conditions at once in an extension.
Changing the outcome measure and treating old and new numbers as directly comparable.
Calling every repeat a fair-test improvement.
Calling every new condition a “better experiment”.
Using the same conclusion after the tested scope has changed.
Ignoring an unexpected follow-up because it conflicts with the first result.
Choosing a follow-up because it is easy to carry out rather than because it reduces the relevant uncertainty.

Parent and Tutor Teaching Guide

When a learner proposes “repeat the experiment”, ask one question: “What would repeating it help us find out?” If the learner cannot answer, the follow-up has not yet been designed scientifically.

Then ask, “If we wanted to know whether the relationship also works under a new condition, what would we change deliberately?” This separates a consistency check from an extension without forcing formal terminology.

If the learner proposes more repeats to repair a comparison in which two relevant conditions changed at once, return to fair-test logic. If the learner proposes a new condition when the first result itself may be unstable, return to repeat evidence. Match the teaching prompt to the uncertainty.

Evidence Limits

Repeated agreement increases evidence of consistency, but it does not automatically prove accuracy, fairness or causation. Extending a tested range gives evidence about the added range, but conclusions still depend on the objects, conditions, measurements and sampling used. A conflicting follow-up is not automatically “wrong”; it may reveal variation, a boundary condition, a method problem or an incomplete explanation.

Useful Internal Routes

Authoritative and Research References

The official Singapore sources define the current curriculum and assessment frame. The replication/extension labels and follow-up protocol are teaching tools used here to make evidence purposes visible; they are not compulsory examination terminology.

Quiet Return

A follow-up investigation should not merely be “more experiment”. It should answer the question the first evidence could not.

If you need consistency, repeat the scientific question carefully. If you need broader scope, extend it deliberately. If the design itself is weak, repair it first. Then let the new evidence change only the claim it is actually capable of changing.

Series route: Previous: How to Diagnose a PSLE Science Failure After a Delay · Return to the PSLE Science Learning Guide

Small Group Tutorials