PSLE Science Reality Lab Vol No.107 | “QA/QC Passed” — Does That Mean Every Result Is Correct?

PSLE-SCI-REALITY-0107

Wait, What? A Green “QA/QC Passed” Box Can Be True Without Meaning “Nothing Could Be Wrong”

A scientific dashboard displays twenty measurements. At the top is a reassuring green label:

QA/QC PASSED

A reader sees the label and says, “Good. That proves every number is correct.”

That is too strong. Quality assurance and quality control are ways to design, monitor and check a measurement process. They can provide important evidence that the process behaved acceptably for particular purposes. But a quality-control check does not turn science into certainty. A blank can check one contamination pathway. A replicate can reveal one kind of variability. A spiked sample can test how well a known addition is recovered. A calibration check can test instrument performance. None of these, alone or together, proves that every possible error has been eliminated.

The Reality Lab habit is simple: when you see “QA/QC passed”, ask what was actually checked, what the pass criterion meant, and whether those checks are strong enough for the claim being made.

Quick Answer

Find out which quality checks were actually used.
Match each check to the problem it can reveal.
Ask what result counted as a pass and whether that criterion suits the intended scientific use.
Keep other failure paths alive: sampling bias, wrong method, sample change, instrument limits, transcription mistakes and inappropriate conclusions can survive some QC checks.
Treat “passed” as evidence that specific controls met specified criteria—not as a universal certificate that every number and conclusion is flawless.

The Exact Learner Job This Page Owns

This page owns one real-world communication problem: a laboratory report, monitoring dashboard, infographic or product comparison uses “QA/QC passed” as though the phrase itself proves complete correctness.

Reality Lab does not take over the canonical PSLE Science jobs of controls, repeated trials, fair testing, accuracy, precision, calibration or method limitations. It applies those ideas to a real quality-status label that learners may encounter outside a worksheet.

Original Reality Lab Case: The River-Monitoring Dashboard

This fictional teaching case uses original data and does not reproduce a real laboratory report.

A community science project measures Indicator K in four river samples. Its public dashboard displays the following quality-control panel:

Quality check	Result	Status
Field blank	Low background	Pass
Duplicate pair	Close agreement	Pass
Reference check	Within project criterion	Pass

A headline above the chart says: “QA/QC passed, so all four sites have been measured perfectly.”

The quality checks are useful. The headline is not. The blank gives evidence about certain contamination pathways. The duplicate pair gives evidence about agreement under the tested duplicate conditions. The reference check gives evidence that the measurement system handled a known reference acceptably. But none of these proves that four sampling locations fully represent the entire river, that no sample changed during transport, that the chosen indicator answers every environmental question, or that every transcription and interpretation step is error-free.

Observed, Claimed and Inferred

Layer	Statement
Observed	The listed QC checks met the project’s stated criteria.
Supported interpretation	Those checks did not reveal the particular problems they were designed to detect beyond the chosen limits.
Stronger claim	Every measurement is correct.
Even stronger claim	The dataset proves the condition of the whole river.
Problem	The stronger claims require evidence about error paths and scope that the QC box does not automatically provide.

QA and QC Are About a Measurement System, Not a Magic Stamp

Quality assurance is the larger planning and documentation system used to make data suitable for an intended purpose. Quality control includes specific checks that help reveal whether parts of the process are behaving as expected. The exact terminology and procedures differ across scientific fields, but the reasoning pattern is durable.

The U.S. EPA’s quality-assurance guidance, for example, describes QC sample types such as blanks, split samples, co-located samples, replicates and spiked samples. Each one has a different diagnostic job. The important lesson is not the vocabulary. It is that a good check is designed around a particular possible failure.

One Check, One Main Question

Example check	Main question it can help answer	What it does not automatically prove
Blank	Did contamination or background appear along a specified pathway?	That every sample result is accurate.
Replicate or duplicate	How much agreement or variability appears when part of the process is repeated?	That the average is representative of the whole system.
Spiked sample	Can the method recover a known added amount under the tested conditions?	That the original sample contained that amount, or that a treatment removed it.
Reference material/check standard	Does the system measure a known reference acceptably?	That every unknown sample has no matrix or handling problems.
Calibration check	Is instrument response behaving acceptably relative to references?	That the chosen method is suitable for every scientific question.

A quality programme becomes powerful because these checks complement one another. But “many checks” still does not mean “omniscience”. Science remains a process of bounding uncertainty.

The Fit-for-Purpose Check: Good Enough for What?

A measurement can be carefully traceable and still be unsuitable for a particular decision if its uncertainty is too large or its method does not address the question being asked. NIST makes this point explicitly in its guidance on metrological traceability: traceability alone does not automatically guarantee that a result is fit for a given purpose or free of mistakes.

For a Primary learner, translate that into an everyday scientific question: what decision is this number supposed to support?

If two products differ by a large amount, a moderately precise method might be enough to support the comparison. If they differ by a tiny amount, the same measurement uncertainty could make the ranking uncertain. “Passed QC” has meaning only alongside the measurement need.

The Representation Check: The Green Tick Can Hide the Scope

A dashboard can compress a dozen QC results into one green tick. That may be useful for navigation, but the icon hides detail. A careful reader asks:

Which checks are included in the green status?
Which checks were not performed?
What limits counted as passing?
Was the status assigned to every sample, one batch, one instrument run or the entire project?
Does the public claim go beyond what those checks evaluate?

A status indicator is a summary of evidence. It is not the evidence itself.

The Sampling Check: Perfect Measurement of a Poor Sample Can Still Answer the Wrong Question

Suppose every laboratory QC check passes beautifully, but all water samples came from one calm corner of a large pond while the headline claims to describe the whole pond. The laboratory may have measured those bottles extremely well. The bigger conclusion can still fail because the sampling design does not support it.

This is why Reality Lab keeps ownership boundaries clear. QA/QC of measurement does not replace the canonical question of representativeness.

The Method Check: Quality Control Cannot Rescue the Wrong Measurement

If the scientific question asks about one property but the method measures a different property, excellent repeatability is not enough. A sensor can repeatedly measure the wrong proxy. A scale can be precise but unsuitable for a tiny mass. A beautifully controlled demonstration can still fail to test the explanation claimed.

Quality is therefore layered: the method must fit the question, the sampling must fit the system, the instrument must fit the quantity, and the controls must fit the likely failure modes.

Alternative Explanations That Can Survive a “Pass”

The sample was collected from an unrepresentative location.
The target changed during storage before analysis.
The method measures a proxy rather than the phenomenon claimed.
An interference affects samples but not the clean reference material.
A transcription or unit error occurs after the laboratory checks.
The numerical result is sound but the headline generalises too far.
The pass criterion is suitable for one purpose but too loose for another.

The existence of alternatives does not mean the data are bad. It tells you what additional evidence would make the claim stronger.

What Evidence Would Strengthen “QA/QC Passed”?

The report names the checks rather than giving only a badge.
Acceptance criteria are stated or linked to a suitable method or data-quality plan.
QC samples are distributed through the relevant parts of the measurement process.
Failures and corrective actions are reported rather than silently discarded.
Measurement uncertainty, range and limitations match the intended comparison.
The sampling plan and sample handling are documented separately.
Independent review or inter-laboratory comparison is used when the scientific job requires it.

What Would Weaken the Claim?

“QA/QC passed” appears with no description of what was checked.
Only one easy control is shown while other obvious failure paths are ignored.
QC results are averaged away even when they show disagreement.
The report uses laboratory quality as proof of sampling representativeness.
The method’s uncertainty is too large for the tiny difference being advertised.
A failed check is excluded from the public report without explanation.

Worked Case 1: The Clean Blank

A field blank produces almost no signal, and the dashboard says QC passed. This strengthens confidence that the blank’s contamination pathway was controlled. It does not prove that the real samples were collected from representative locations, that their target amount remained stable in storage, or that another substance could not interfere with the measurement.

Worked Case 2: The Reference Standard

A known reference should read 50 units. The instrument reports a value inside the laboratory’s accepted range. That is useful evidence that the measurement system is behaving appropriately on the reference. It still does not prove that every unknown sample behaves identically, especially if the unknown samples have different physical or chemical properties that affect the method.

Worked Case 3: The Perfectly Repeated Wrong Question

A sensor gives nearly identical readings every time. QC for repeatability looks excellent. But the public claim says the sensor directly measures “plant health” when it actually measures one optical signal affected by several plant and environmental properties. Repetition makes the signal more consistent; it does not magically convert a proxy into the full biological idea.

Worked Case 4: The Tiny Product Difference

Product A scores 50.2 units and Product B scores 50.5 units. Both were measured in a QA/QC-controlled process. Can we advertise B as scientifically superior? Only if the measurement uncertainty and method performance are small enough to support that 0.3-unit difference and the comparison conditions are otherwise fair.

Tempting Reasoning That Fails

“QA/QC passed, therefore the number is exact.” Quality controls bound some errors; they do not remove all uncertainty.
“The lab followed a procedure, therefore the scientific conclusion must be correct.” A procedure can be followed while the claim still overreaches the data.
“A clean blank proves no contamination occurred anywhere.” It supports the pathway represented by that blank, not every imaginable contamination path.
“Duplicates agree, therefore the whole environment is uniform.” Duplicate agreement addresses repeatability or local variability under specific conditions, not broad representativeness.
“One QC failure means all data are useless.” A failure is evidence to investigate. Its impact depends on what failed and which results it affects.

Model and Measurement Limits

No quality system can test every possible mistake on every sample. Quality programmes therefore choose checks based on the measurement process, risks and intended use. A stronger claim may require stricter limits, more controls or an independent method.

This is not a weakness of science. It is one of science’s strengths: a good system states what has been checked and leaves room for what has not yet been ruled out.

How Far Can the Conclusion Travel?

If the QC plan is appropriate, the checks pass and the method fits the question, we may have strong evidence that the measurements are suitable for their intended analysis. Moving from that to a broader claim—every item is safe, every location is the same, the product is universally better, the mechanism is proven—requires additional evidence for those specific conclusions.

PSLE-Style Transfer Case

A class measures temperature at four points in a model system. A reference thermometer check passes, and two repeated readings at one point agree closely. The class claims, “QA/QC passed, so every point in the model must have the same temperature.”

Question: Why does the QC evidence not support that conclusion?

Reasoned answer: The reference check and repeat readings give evidence about measurement performance, but they do not show that different locations in the system have equal temperature. Spatial variation is a separate scientific question that requires measurements at the relevant locations.

Explained Practice

Practice A: A report says “QC passed” and shows a clean blank. What can you say? The blank did not reveal concerning contamination through the pathway it represents. You cannot conclude that every other possible error was absent.

Practice B: Two duplicate measurements agree closely. What does that strengthen? Confidence in repeatability or local agreement under those duplicate conditions. It does not by itself prove accuracy against the true value.

Practice C: A result is traceable to a reference but has uncertainty larger than the difference between two products. Is the ranking secure? Not necessarily. Traceability does not make uncertainty disappear.

Delayed Independent Return: The Q-U-A-L-I-T-Y Check

Q — Question: What scientific claim is the data supposed to support?
U — Use: What level of quality is needed for that purpose?
A — Actual checks: Which QA/QC checks were performed?
L — Limits: Which errors can those checks detect, and which can remain?
I — Interpretation: What does “pass” actually mean?
T — Transfer: Is the claim travelling beyond the samples, method or conditions checked?
Y — Yet unknown: What important alternative has not been ruled out?

Parent and Tutor Teaching Guide

Give a learner four cards labelled BLANK, DUPLICATE, REFERENCE and SPIKE. Give a second set of cards labelled CONTAMINATION, REPEATABILITY, KNOWN VALUE and RECOVERY OF A KNOWN ADDITION. Ask the learner to match the main jobs. Then deliberately offer a ridiculous extra card: “PROVES EVERYTHING”. The learner should be able to explain why no single QC card matches it.

A second activity is to show a fictional report with a green QA/QC badge but a weak sampling design. Ask, “Can the laboratory be careful and the broad conclusion still be too strong?” That question teaches a mature scientific habit: different parts of an evidence chain can have different strengths.

Authoritative Sources

The official Singapore Science frame is a good fit for this problem. The 2026 PSLE Science assessment objectives include evaluating observations, information and methods and communicating reasoning. The 2023 Primary Science syllabus asks learners to exercise healthy scepticism, consider assumptions and uncertainty, and understand how Science is presented in different forms and media. A QA/QC badge is exactly the kind of scientific communication object that should be interpreted rather than merely trusted or distrusted.

The Quiet Return

Quality control is not a promise that science has no uncertainty.

It is evidence that scientists looked for particular failures instead of pretending failures cannot happen.

When a report says “QA/QC passed”, ask the scientific question hiding behind the green tick: passed which checks, for which purpose, and what remains possible?

Small Group Tutorials