How to Decide When PSLE Science Repeated Results Should Be Averaged — and When They Should Stay Separate

Wait, What? An average can make evidence clearer—or erase the scientific difference you were supposed to notice.

Students often learn that repeated results can be averaged. That is useful, but incomplete. Before you calculate any average, you have to ask whether the values belong together scientifically. Results from repeated trials under the same condition may sometimes be summarised. Results from different conditions, different times, different quantities or different scientific jobs usually should not be mixed into one number just because they appear in the same table.

Quick Answer

Average only after you have established that the results are comparable repeats of the same measured quantity under the same intended condition and role. Keep values separate when they represent different test conditions, different variables, different time points, different specimens that are not meant to be pooled, or a known invalid run. Always keep the raw results visible enough to check what the average hides.

The PSLE Science Learning Job This Guide Owns

This guide owns one learner job: deciding whether repeated numerical evidence may legitimately be combined into an average or must remain separate. It does not replace the existing guide on how to interpret an average that is already given, nor the guide on how to read repeated results that vary. Its job comes earlier: deciding whether the values belong in the same summary at all.

For the 2026 PSLE, Standard Science assesses the 2023 Primary Science syllabus. The official assessment objectives include interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning. That means the meaning and origin of data matter before any arithmetic is performed.

First Ask: What Does Each Number Represent?

Before averaging, label each value with five pieces of information:

Object or specimen: what was measured?
Quantity: what scientific variable was measured?
Unit: in what unit was it recorded?
Condition: under what test condition was it produced?
Time or trial role: is it a repeated trial at the same point, or a different time point, stage or condition?

If those labels do not match in the way the investigation intends, do not average merely because the numbers are nearby.

When Averaging Can Be Useful

Suppose the same measurement is repeated three times under Condition A using the same method, same unit and same intended comparison. The values are 18, 20 and 19 units. A summary average of 19 units may be useful because the values are repeated observations of the same scientific quantity under the same condition.

The average does not become a new observation. It is a calculated summary of the three observations. The individual values still matter because they show the variation that the average hides.

When Results Must Stay Separate: Different Test Conditions

Imagine a plant model is tested under three light conditions and produces values of 4, 8 and 12. These are not three repeats of one condition. They are evidence about three different conditions. Averaging them into 8 would destroy the relationship the investigation is trying to study.

The question is not “What is the average of all numbers in the table?” The scientific question is “How does the measured outcome differ across the tested conditions?” Keep the condition–outcome pairs together.

When Results Must Stay Separate: Different Time Points

A temperature measured at minute 0, minute 5 and minute 10 is a time series. Those values describe change through time. Averaging them into one temperature can erase the very pattern the question is asking you to interpret.

Repeated measurements over time are not automatically repeated trials. The time point is part of the condition. Keep the sequence unless the question specifically asks for a summary that makes scientific sense.

When Results Must Stay Separate: Different Quantities

A table may show temperature, distance and time for the same set-up. These numbers cannot be averaged together. They describe different quantities and may use different units. The fact that all three are numerical does not make them one data set.

When a Known Invalid Run Should Not Be Quietly Pooled

If one trial has a documented procedural failure—for example, the timer was started late or the wrong quantity was recorded—do not simply mix it into the average and hope the summary “smooths it out”. First diagnose the trial. If there is a justified reason that it did not follow the intended method, repair or repeat that trial according to the investigation design. If no such reason exists and the result is merely unusual, preserve it as evidence rather than deleting it for convenience.

Averaging Does Not Make Different Things Comparable

Suppose Set-up A was tested for five minutes and Set-up B for ten minutes. Each has three repeated measurements. Averaging the three values within each set-up may be reasonable if those are valid repeats. But comparing the two averages may still be unfair if elapsed duration matters and was not matched.

Aggregation comes after scientific comparability, not before it.

The Pool-or-Separate Decision Table

Check	If it matches	If it differs
Same measured quantity?	Continue checking.	Keep separate.
Same unit and measurement meaning?	Continue checking.	Convert only when scientifically appropriate; otherwise keep separate.
Same intended test condition?	May be repeated evidence.	Keep condition groups separate.
Same time point or trial role?	May be repeated evidence.	Keep the time structure visible.
Valid runs of the intended method?	Summary may be useful.	Diagnose the invalid run first.
Does an average preserve the question’s relationship?	Calculate if useful.	Do not average away the relationship.

Worked Example 1: Three Repeats Under One Condition

A learner measures the same outcome three times under Condition P: 24, 25 and 23 units. The trials use the same method and condition. Averaging can be a useful summary, but the learner should still notice that the raw results vary from 23 to 25.

READ GIVEN INFORMATION → IDENTIFY THE QUANTITY → CHECK THE CONDITION → CHECK THE TRIAL ROLE → CALCULATE A SUMMARY → RETURN TO THE RAW EVIDENCE.

Worked Example 2: Three Different Conditions

Condition P gives 10 units, Q gives 14 and R gives 18. These are not repeats. Their differences are the evidence. Averaging all three to 14 would hide the relationship across conditions.

Worked Example 3: Repeats at Several Time Points

At minute 0, a learner has three repeated readings. At minute 5, there are another three. It may be useful to calculate a separate summary for the three comparable repeats at minute 0 and another for minute 5. It would usually be misleading to average all six together if the scientific job is to study change over time.

This shows that “average or not” is not a property of the numbers alone. It depends on how the evidence is organised.

Worked Example 4: An Unusual Result

Four repeats give 12, 13, 12 and 20. Do not automatically delete 20 and average the remaining three. Check the record, apparatus, procedure and conditions. If no justified error is found, the unusual value remains part of the evidence and should affect how confidently you summarise the result.

Failure Signatures

Averaging every number in a row without asking what the row represents.
Averaging values from different test conditions and erasing the trend.
Averaging different time points and erasing change over time.
Mixing values with different units or different measured quantities.
Dropping an anomalous result simply to make the average look neat.
Treating an average as another independent measurement.
Comparing two averages from set-ups whose conditions were not scientifically matched.

Earliest Weak-Link Diagnosis

What does one row or one group of values represent?
Which values are true repeats of the same condition?
Which values represent different conditions, times or quantities?
Are any runs known to be methodologically compromised?
What relationship is the question asking you to preserve?
Would one average clarify that relationship or hide it?

Misconception Repair

“Repeated numbers should always be averaged.” No. First decide whether they are repeated evidence of the same scientific quantity under the same intended condition.

“The average is more real than the raw results.” No. It is a calculated summary. The raw results show the evidence from which it was made.

“Averaging removes errors.” It can reduce the influence of ordinary variation in some situations, but it does not repair unfair comparisons, shared measurement bias or invalid methods.

“If two averages are close, the set-ups are scientifically the same.” Not necessarily. You must still inspect variation, conditions, units and what the method can support.

The PSLE Science Averaging Protocol

IDENTIFY EACH VALUE → GROUP ONLY COMPARABLE REPEATS → PRESERVE CONDITION AND TIME → CHECK FOR KNOWN INVALID RUNS → CALCULATE A SUMMARY ONLY IF IT SERVES THE QUESTION → RETURN TO RAW RESULTS → INTERPRET THE SUMMARY WITH ITS LIMITS.

Practice Sequence

Round 1: sort values into “same repeat group” or “must stay separate”.
Round 2: explain why the grouping is scientific, not merely numerical.
Round 3: calculate an average for one valid repeat group, then state what the raw spread still tells you.
Round 4: transfer to a table with several conditions and time points.
Delayed return: decide the grouping from a fresh table without prompts.

Unfamiliar Transfer Challenge

A results table contains three readings for Set-up A at minute 5, three for Set-up A at minute 10, and three for Set-up B at minute 5. Which values may be summarised together? The answer depends on the scientific question, but one safe starting rule is that minute-5 repeats for A form a different evidence group from minute-10 repeats for A and minute-5 repeats for B. Preserve those labels before any calculation.

Answer-Checking Receipt

I know what each value represents.
I know which values are genuine repeats of the same scientific condition.
I have not averaged away a condition or time difference the question needs.
I have not mixed different quantities or units carelessly.
I have preserved unusual results unless there is a justified reason to treat a run as invalid.
I remember that the average is calculated from evidence; it is not new independent evidence.
I can still inspect the raw results after calculating the summary.

Parent and Tutor Teaching Guide

Before asking a child to calculate, cover the numbers and ask what each row and column represents. Then reveal the values. This prevents arithmetic from taking control before the scientific structure is understood.

Use two contrasting tables: one with repeated trials under the same condition and one with different test conditions. Ask, “Which numbers belong in the same scientific group?” Only after the learner can justify the grouping should averaging enter the task.

Avoid teaching “always average repeats” as a universal rule. Teach the more durable question: what is being repeated, and would this summary preserve the relationship we need to see?

Useful Internal Routes

Authoritative References

Quiet Return

Averaging is not a command that arrives whenever several numbers appear. It is a scientific decision about which observations belong together. Preserve the object, quantity, condition and time first. Then summarise only when the summary keeps the question’s meaning intact. Good data handling begins before the arithmetic.

Small Group Tutorials