PSLE Science Reality Lab Vol No.079 | “Four Samples Were Mixed Before Testing” — Does One Pooled Result Describe Every Location?

PSLE-SCI-REALITY-0079

Wait, What? One laboratory result can come from four places at once—and then stop telling you what happened at any one place.

A school garden report says, “Soil moisture across the four plots was 24%.” The number looks wonderfully precise. There is only one problem: the team did not test four soil samples separately. They collected equal spoonfuls from Plots A, B, C and D, mixed them thoroughly in one container, and tested the mixture once.

The reported result may be useful. It may even be exactly the result the investigation wanted. But it belongs to a pooled or composite sample. It does not automatically mean every original plot had 24% moisture.

Reality Lab Vol No.079 teaches one real-world evidence-transfer job: when several original samples are mixed before measurement, keep the later result attached to the mixture and do not silently send it backward to each original source.

Quick Answer

Trace the sample history. Were the original samples measured separately or physically combined first?
Name the measurement unit correctly. After mixing, the measured object is the pooled sample.
Ask what pooling preserves. With equal, suitable portions and an appropriate measurement, a composite can help estimate an average condition.
Ask what pooling hides. It can hide which original location was high, low or unusual.
Do not assign the pooled value to every source. “The mixture measured 24%” is not “every plot measured 24%.”
Match the design to the claim. If the claim is about each location, measure each location—or preserve a way to identify them.

Reality Lab habit: A result belongs first to the thing that was actually measured.

The Owned Learner Job — and the Boundary

This page does not re-own sample provenance, combining samples or generalisation. Those underlying reasoning skills already have canonical owners. Reality Lab Vol No.079 applies them to a distinct communication object: a report, infographic or claim that presents one pooled laboratory result as though it described every original place or item separately.

Vol No.023 asks whether many tested specimens represent a wider population. Vol No.079 asks something different: what information is lost when several original specimens become one physical mixture before measurement?

Original Reality Lab Case: The Four Garden Plots

This is an original composite teaching case. No real environmental report is being reproduced.

A science club wants a quick picture of soil condition across four equal garden plots. From each plot, students collect 50 g of soil from a defined depth. Instead of sending four containers to the laboratory, they combine the four 50 g portions, mix them, and send one 200 g composite sample.

Original location	Amount placed into pool	Individual value known?
Plot A	50 g	No
Plot B	50 g	No
Plot C	50 g	No
Plot D	50 g	No
Mixed pool	200 g	Measured result: 24 units on the chosen test

The 24-unit result is evidence about the mixed material under that test. Depending on the property, equal portions and method, it may provide an estimate related to the average condition across the sampled material. But the mixture has erased the identity of the four contributions. You cannot reconstruct four individual values from one pooled result.

Observed, Claimed and Inferred

Layer	What can be said
Observed	One mixed sample made from equal portions of four plots produced a result of 24.
Claimed	All four plots had a value of 24.
Inferred	The pooled result is being assigned back to each original plot separately.

The observation can be completely accurate while the backward assignment is invalid.

Build the Sample Family Tree

A useful way to reason is to draw the evidence path:

Plot A sample + Plot B sample + Plot C sample + Plot D sample → mixed composite → laboratory measurement → reported result.

Once the branches merge, later evidence belongs to the merged object unless the method preserves separate identifiers or uses additional tests to recover them.

This is a provenance problem. Ask, “Which physical material actually reached the measuring instrument?”

Why Scientists Pool Samples

Pooling is not automatically bad science. It can be a deliberate design choice. Official U.S. Environmental Protection Agency sampling guidance describes composite sampling as physically combining several samples before analysis. One major advantage is that researchers can cover more locations while performing fewer laboratory analyses. When the study goal is an average concentration or overall condition, a composite design can be useful.

The trade-off is information. You save analyses, but you may lose detail about the individual sources.

What Pooling Can Answer

What was measured in the combined material?
What is the overall condition represented by the pool under the sampling design?
Is a substance detectable somewhere in a pooled screening sample, for methods where pooling is suitable?
What approximate mean condition might the sampled material represent when equal contributions and the measurement support that interpretation?

What Pooling Cannot Answer by Itself

Which original location had the highest value?
Which location had the lowest value?
Whether every location met a threshold.
Whether one extreme result was diluted by several ordinary ones.
Whether all original sources were similar.
The exact individual value of any source that was never tested separately.

The Hidden-Extreme Problem

Suppose, for a simple imaginary additive property, four equal portions would have produced values of 10, 10, 10 and 50 if measured separately. Their equal-weight average is 20. A properly mixed composite might therefore produce a result around 20 for a measurement that behaves that way.

A report that says “the composite was 20” may be correct. But “each location was 20” would be badly wrong. Three were lower; one was much higher.

This is why pooling is naturally suited to some average questions and poorly suited to questions about local extremes.

Representation Check: One Number Can Hide a Map

Imagine four sampling points on a map, all pointing toward one laboratory bottle. If the final infographic prints only “24” in large type, the reader may forget that four spatially different sources were physically merged.

A better representation shows the sampling path: four collection points → one composite. This does not change the data. It restores the evidence history.

Comparison and Baseline Check

If two pooled results are compared, ask whether the pools were built in comparable ways. Did each contain the same number of original samples? Were equal masses or volumes contributed? Were the same kinds of locations included? Were the samples collected at comparable depths, times or conditions?

“Pool X = 24 and Pool Y = 18” is only a fair comparison when the sampling design makes those pools comparable for the intended question.

Method and Variable Check

Pooling adds a method layer before measurement. That layer can affect interpretation.

Were equal amounts combined?
Was the mixture homogenised enough for the laboratory subsample to represent the pool?
Could the measured property change during mixing or storage?
Were unlike materials combined even though the test assumes comparable material?
Was the property additive or interpretable as an average under the chosen method?
Was the purpose screening, estimating an average, or classifying each original unit?

The word “pooled” tells you how evidence was combined. It does not automatically tell you that the design was suitable for every scientific question.

Source and Provenance Check

When a real report uses a composite sample, look for the sampling method. How many increments or locations contributed? How much material came from each? Were locations selected systematically or randomly? Was one laboratory test performed per pool? Without that provenance, one neat number can appear more specific than the evidence really is.

Worked Case 1: The Playground Soil Pool

Five equal soil portions from five points are mixed. The composite result is 12 units. A poster says, “Every point measured 12.”

Evaluation: unsupported. Only the composite was measured. The correct bounded statement is that the mixed sample produced 12 under the stated method. The individual points remain unknown unless separately analysed.

Worked Case 2: A Good Use of Pooling

A gardener wants an overall nutrient estimate for one small, fairly uniform bed and collects many equal subsamples according to a planned pattern, mixes them, and sends the composite for a suitable test. The question is about the bed’s overall condition, not the exact value at each centimetre.

Pooling can be sensible because the design and the claim align.

Worked Case 3: A Bad Use for a Threshold Claim

Four product batches contribute equal material to one pooled sample. The mixture is below a stated limit. The label then says, “Every batch passed.”

The pooled result cannot by itself prove that every original batch was below the limit. One higher batch might have been diluted by lower ones. A claim about each batch requires evidence that preserves batch identity.

Worked Case 4: Positive Pool, Unknown Source

A screening method combines small portions from several units. The pool gives a positive result for a target feature. What can you conclude?

You have evidence that the pooled material contains the target under the method. You do not yet know which original unit contributed it. Follow-up individual testing may be needed if the learner job is to identify the source.

Alternative Explanations and Hidden Structure

One location may be much higher than the others.
One location may be much lower.
Several different combinations of individual values can produce the same pool result.
Unequal contribution amounts can make some sources influence the pool more than others.
Poor mixing can make the laboratory subsample unrepresentative of the composite.
The measured property may not combine as a simple arithmetic average.

The important point is not to invent which explanation happened. It is to recognise that the pooled result does not distinguish among them.

What Evidence Would Strengthen “The Whole Area Is About 24”?

A sampling design that covers the target area appropriately.
Equal or otherwise justified contributions to the composite.
A measurement for which compositing is scientifically suitable.
Good mixing and documented laboratory subsampling.
Replicate composites or some separate samples to check variation.
A conclusion framed as an overall or average condition rather than an individual-location guarantee.

What Would Strengthen “Every Location Meets the Limit”?

Different evidence. You would need location-level measurements, a validated sampling rule that supports that classification, or another method designed to establish the condition of each original unit. A single pooled mean-like result is usually not enough to make an every-location statement.

What Would Weaken the Claim?

The report hides that samples were mixed.
Contribution amounts differ without explanation.
Sampling locations were chosen only from convenient areas.
The pool is used to make a claim about maxima, minima or every individual source.
The property changes during storage or mixing.
One composite is treated as proof of uniformity.

How Far Can the Conclusion Travel?

A composite made from four sampled points may tell you something about that composite and, with a defensible design, something about the overall sampled area. It does not automatically support claims about unsampled places, future conditions, every original point or a different sampling depth.

The conclusion can travel only as far as the sampling plan and measurement allow.

Model and Measurement Limits

Our simple examples often treat an equal composite result as though it were an ordinary average. Real measurements can be more complicated. Some properties do not combine linearly. Chemical reactions, particle size, moisture loss, uneven mixing and detection limits can affect the result.

Primary learners do not need specialist sampling statistics. They do need one durable rule: do not reverse a many-to-one measurement into several invented one-to-one measurements.

PSLE-Style Transfer Case

Four equal water samples, W, X, Y and Z, are collected from different tanks. Equal volumes are mixed. The mixture gives a colour intensity reading of 30 arbitrary units. A student concludes, “Each tank has a colour intensity of 30 units.”

Evaluation: The conclusion is not supported. The measured object was the combined sample, not each tank sample separately. Different individual readings could produce a mixture with the same overall reading. To know each tank’s value, the samples should be measured separately using a suitable method.

Tempting Reasoning That Fails

“One result represents four samples, so all four equal the result.” Representation is not identity.
“Mixing gives an average, therefore every part is near the average.” An average can hide wide variation.
“Pooling is less scientific because it loses detail.” It can be an efficient, valid design when the question asks about an average or overall condition.
“A negative pooled result proves every source is negative.” That depends on the method, detection limits, dilution and intended classification rule.
“A positive pool tells us which source was positive.” Source identity was lost unless a follow-up design recovers it.

Explained Practice

Practice A: Three equal leaf extracts are mixed and the composite contains 18 units of a measured substance. Can you state that every leaf extract contained 18 units? No.

Practice B: Ten subsamples are deliberately pooled to estimate the overall condition of one uniform material pile. Is pooling automatically a flaw? No. The design may fit the overall-mean question.

Practice C: A pooled result passes a maximum limit. Can you conclude no individual source exceeded the limit? Not without evidence that the pooling and decision method justify that individual-level claim.

Delayed Independent Return: The P-O-O-L Check

P — Provenance: Where did the original portions come from?
O — Object measured: Was each sample tested, or only the combined pool?
O — Outcome needed: Is the claim about an overall condition or each individual source?
L — Lost detail: What information disappeared when the samples were mixed?

Return to a new case later. If you can keep the pooled result attached to the pooled object without inventing individual values, the learner job is secure.

Parent and Tutor Teaching Guide

You can model this safely with four cups of coloured water. Use different intensities of food colouring, take equal spoonfuls from each, and combine them in a fifth cup. Ask the learner whether the colour of the fifth cup tells them the exact colour intensity of each original cup.

Then reverse the task. Show only the mixed cup and ask the learner to invent two different sets of original cups that could produce a similar-looking mixture. The point is not to calculate a perfect colour average. The point is to experience information loss: many different source patterns can collapse into one combined observation.

Finally, ask when pooling would actually be useful. A strong learner should say something like, “When I want an overall result and do not need to know every source separately.”

Authoritative Sources

EPA guidance explains that compositing physically combines samples and can reduce analytical cost while estimating an overall mean-like condition, but it also notes that information is lost. SEAB’s current PSLE Science assessment objectives ask learners to interpret and analyse information and evaluate observations, information and methods. The Reality Lab transfer is therefore simple: always ask what object the measurement actually belongs to.

The Quiet Return

Science often combines information to see a larger pattern. That can be efficient and powerful.

But combining also has a price. Once four samples become one bottle, one result cannot magically remember every individual value that was never measured.

Follow the sample. Keep the result attached to the thing the instrument actually saw.

Small Group Tutorials