Stable ID: PSLE-SCI-REALITY-0539
PSLE Science is not only about knowing a fact. A strong Primary 5 or Primary 6 learner also has to judge whether evidence is usable, what a scientific result actually says, and where a conclusion must stop. This Reality Lab looks at a real-world laboratory communication problem that can make an apparently simple table surprisingly difficult: a result is present in the report, but beside it sits a quality code saying that the result was rejected or unusable.
That situation is a powerful PSLE Science evidence-and-inquiry exercise because the number, blank cell or status code can tempt us into a false shortcut. A learner may think, “If the laboratory rejected the result, perhaps the substance was not there.” Another may make the opposite mistake: “The laboratory must have found the substance but simply did not like the number.” Neither conclusion follows automatically. The correct scientific habit is to separate the sample, the measurement attempt, the quality evidence and the claim we want to make.
This matters directly to the current PSLE Science emphasis on applying scientific knowledge and inquiry: interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning. It also fits the 2023 Primary Science syllabus emphasis on healthy scepticism—questioning observations, methods, processes and data rather than accepting a label or number at face value. The aim here is not to turn a Primary 6 learner into a laboratory analyst. It is to practise one durable evidence-transfer job: when scientific data are rejected, do not convert “we cannot rely on this result” into “the amount was zero” or “the substance was absent.”
Wait, what? A result can exist and still be unusable?
Imagine a school environmental club sends four water samples to a fictional laboratory. The report arrives with this compact table:
| Sample | Reported result | Status |
|---|---|---|
| A | 2.1 units | Accepted |
| B | < reporting limit | Valid non-detect |
| C | 3.4 units | Estimated |
| D | — | Rejected: quality-control failure |
A learner is asked: “Which sample proves that none of the target substance was present?” The most tempting choice is Sample D because there is no usable number. But that is exactly the trap. Sample D does not tell us “none was present.” It tells us that the laboratory could not use that measurement attempt to make the required decision reliably. The evidence failed before the claim could be made.
Quick Answer
Rejected data are not negative evidence by default. A rejected or unusable result means the measurement did not satisfy the relevant quality conditions strongly enough to carry the intended conclusion. Depending on the system, a serious quality-control problem can mean that the presence, absence or amount of the target cannot be determined from that result. The right response is usually to find out why it was rejected and what new evidence—such as reanalysis, a new sample or another valid measurement—would be needed.
Keep three statements separate:
- Measured zero: the method produced a valid result of zero, if zero is a meaningful possible result for that measurement.
- Valid non-detect: the target was not detected under a method with stated detection capability. This still is not automatically the same as “absolutely none exists.”
- Rejected/unusable: the measurement evidence is not fit for the intended decision. The result cannot simply be treated as zero or absence.
Your learner job—and what this guide does not own
Your job in this Reality Lab is narrow and practical: read a laboratory result together with its quality status and decide whether the result is allowed to support the claim being made. You are not learning a universal system of laboratory qualifier letters. Different organisations can use different codes. You are not memorising an examiner keyword. You are not learning analytical chemistry procedures. You are not being asked to decide whether water, food or any product is safe for personal use.
Those boundaries matter. The United States Environmental Protection Agency, for example, explains that data qualifiers can differ across datasets and that users should consult the metadata or documentation that defines the codes. Its current guidance includes examples in which rejected data should not be used and where the evidence cannot determine presence or absence. The scientific habit is therefore not “R always means exactly one thing everywhere.” The habit is: read the key, read the reason, and match the conclusion to the evidence status.
Case File 1: The school pond report
Let us rebuild the problem as an original composite case. A class is investigating whether a dissolved substance is detectable in four pond-water samples. The laboratory normally runs the samples together with quality-control checks. In one batch, the control check falls outside the laboratory’s stated acceptance condition. The report marks Sample D “rejected—do not use.”
Now sort the information into three boxes.
| Observed or documented | Claim someone wants to make | Inference that still needs support |
|---|---|---|
| The sample was analysed in a batch. A quality-control check failed. The reported result was rejected. | “The substance was absent from Sample D.” | Whether the substance was absent, present below a limit, present above a limit, or present at some amount. |
Notice the gap. The quality failure is evidence about the reliability of the measurement process. It is not automatically evidence about the substance’s true amount in the pond. One piece of information has been aimed at the wrong question.
A four-gate way to read any rejected result
Instead of memorising codes, use four gates. The gates are questions, not a magic answer template.
Gate 1: What was actually measured?
Identify the sample, target quantity, units, method and time. A status word without an object is incomplete. “Rejected” might refer to one analyte, one sample, one run, one calculation or an entire batch. Never widen the rejection beyond the report’s scope.
Gate 2: Why was it rejected?
Look for the reason or the report key. Was there a failed blank, failed calibration check, unacceptable control recovery, sample identity problem, holding-time problem, instrument malfunction or another documented issue? The exact cause matters because it tells us what part of the evidence chain is weak.
Gate 3: What decision can the result no longer support?
This is the most important gate. Sometimes the quality problem prevents reliable quantification. Sometimes it prevents a presence/absence decision. Sometimes a result may still be usable only with a qualification. Do not decide this from the word “rejected” alone; use the report’s documented meaning.
Gate 4: What evidence would repair the decision?
The scientific response to weak evidence is not panic and not pretending. It is a repair plan: reanalyse if a valid retained sample and method allow it, collect a new sample if the original cannot support a valid result, check an independent method where appropriate, or state that the claim remains unresolved. A Primary learner does not need to choose an actual laboratory procedure; the reasoning job is to recognise that a new valid line of evidence is needed.
Rejected is not the same as non-detect
This distinction deserves its own section because the two labels can look similar in a table. Both may lack an ordinary numerical concentration, yet their meanings are different.
| Status | What the evidence says | What you must not say automatically |
|---|---|---|
| Valid non-detect | The method did not detect the target under the stated conditions and detection capability. | “The true amount is exactly zero everywhere.” |
| Estimated result | A result exists, but some limitation means the number carries additional qualification. | “It is identical in evidential strength to an ordinary accepted result.” |
| Rejected/unusable | The result fails the quality conditions needed for the intended use. | “The target was absent,” “the target was present,” or “the amount was zero” unless separate valid evidence supports it. |
For a deeper application of non-detects and reporting capability, route to PSLE Science Reality Lab Vol No.111. For the different job of reading an explicitly estimated number, route to PSLE Science Reality Lab Vol No.112. Those pages own those specific evidence patterns; this article owns only the transfer problem created by rejected or unusable laboratory evidence.
Why “no usable result” is not “zero”
Suppose your ruler breaks before you measure the length of a pencil. You would not record 0 cm. The failed measurement does not make the pencil length zero. It leaves the length unresolved by that measurement attempt. The same logic travels to more complex laboratory evidence.
The key word is epistemic: a rejected result is telling us something about what we are justified in knowing from the evidence. It is not necessarily telling us that the physical quantity itself is absent. Primary pupils do not need to memorise the word epistemic, but the distinction is worth understanding: a limit on evidence is not automatically a limit on reality.
Case File 2: The control sample failed
A fictional laboratory measures a target substance in fruit-juice samples. To check the procedure, it also analyses a control material whose expected performance range is known. In one run, the control result falls outside the stated acceptable range. The laboratory rejects the associated sample result.
A student says, “The control was wrong, so the sample must contain no target.” That conclusion fails because the control is checking the measurement process. A control failure can weaken trust in the sample result, but it does not set the sample’s true value to zero.
Another student says, “The control was wrong, so the sample definitely contains a lot of target.” That also fails. A failed quality check is not evidence for the opposite claim. Good scientific reasoning does not turn uncertainty into whichever answer feels more interesting.
The best statement is narrower: “Because the relevant quality-control check failed and the laboratory marked the result rejected, this result should not be used to decide the target’s presence, absence or concentration. We need valid replacement evidence.”
Case File 3: A number is still printed beside the rejection
Sometimes a data system preserves a numerical value even when a qualifier says the result is unusable for the intended analysis. This creates an especially strong visual temptation. Imagine the table prints “7.8” in one column and “REJECT” in another.
Which field should control your conclusion? Neither field should be read alone. The number tells you what the system recorded; the qualifier tells you how that number may be used. If the documentation says rejected results must not be used because required quality conditions failed, copying 7.8 into an average simply because it looks precise is not scientific care. Precision in appearance does not rescue invalid evidence.
This is a powerful lesson for PSLE Science tables too. A value can look exact while the evidence needed to interpret it is weak. Digits are not authority. A neat decimal is still only as useful as the measurement chain that produced and qualified it.
Case File 4: One rejected result inside a larger dataset
Now imagine ten measurements from a monitoring programme. Nine are accepted and one is rejected. A learner has to evaluate a headline saying, “All ten samples met the target.”
Can we count the rejected sample as a pass? No. If its status means the result cannot support the required decision, then we do not have ten valid passing results. But we should also resist the opposite exaggeration: one rejected sample does not by itself prove that the sample failed the target. The correct report might be, “Nine valid samples met the target; one sample was unresolved because its laboratory result was rejected.”
That sentence is less dramatic, but it is more scientific. It preserves the strength of the nine accepted results and the uncertainty of the tenth. Scientific communication often improves when we stop forcing every observation into only “pass” or “fail.”
Representation check: where did the qualifier go?
A result can be handled correctly in a laboratory report and then miscommunicated later. Picture this chain:
- The laboratory table marks one sample rejected.
- A spreadsheet imports only the numeric column and drops the status column.
- A graph is made from the spreadsheet.
- An infographic shows the graph without the qualifier key.
- A headline summarises the infographic.
The scientific problem can therefore be a representation problem, not just a laboratory problem. Ask whether the chart, dashboard or article preserved the quality information that controlled how the result should be used. A number without its qualifier can become a different—and misleading—communication object.
For the broader job of asking what a data quality flag says about a displayed scientific value, use PSLE Science Reality Lab Vol No.040. This page is narrower: it asks what to do when the quality system has gone so far as to reject the evidence for the intended decision.
Comparison and baseline check
Rejected results can also break comparisons. Suppose Site East has five accepted measurements and Site West has four accepted measurements plus one rejected result. Someone compares the two five-row tables and says, “Both sites were measured five times, so the evidence is equally complete.” That is not necessarily true. Five attempted measurements are not the same as five usable measurements.
Before comparing groups, ask:
- How many measurements were attempted?
- How many met the required quality conditions?
- Were exclusions handled by the same documented rule?
- Could missing or rejected results occur more often under one condition?
- Does the conclusion depend strongly on the unresolved cases?
This does not mean every dataset with a rejected value is useless. It means the denominator and evidence status matter. A transparent comparison tells the reader what was attempted, what was usable and what remains unresolved.
Method and variable check
When a result is rejected, do not invent a cause. A class may know that temperature, timing, contamination, calibration, sample identity and instrument performance can affect measurements, but the report must tell us which issue actually triggered the rejection. Listing every possible variable is not the same as identifying the documented problem.
A disciplined answer can use conditional language: “If the rejection was caused by a failed calibration check, measurements from the affected period may not support the intended quantitative claim. If it was caused by uncertain sample identity, the problem is instead whether the result can be linked to the correct sample.” Different failures attack different links in the evidence chain.
Alternative explanations: two different questions
Primary Science encourages learners to consider more than one plausible explanation when evidence does not uniquely decide a question. A rejected result is a perfect place to practise that habit. If Sample D cannot be interpreted, at least four broad possibilities remain open:
- The target may truly be absent or below the method’s useful capability.
- The target may be present at a measurable amount, but the failed run could not establish it reliably.
- The sample itself may have changed, been compromised or been misidentified before measurement.
- The instrument or analysis process may have failed in a way that prevents trustworthy interpretation.
These are possibilities, not conclusions. The point is to resist pretending that a failed measurement chooses one explanation for us.
What evidence would strengthen the claim?
Suppose the claim is “The target was not detected in Sample D.” Evidence becomes stronger when the new measurement is produced under documented acceptable quality conditions, the sample identity is secure, the method has appropriate capability for the claim, relevant controls perform acceptably, and the result is reported with its detection or reporting information. An independent repeat or a new sample may add confidence when appropriate.
Suppose the claim is instead “The target was present at 5.0 units.” Then a valid quantified result is needed, not merely a statement that a previous run was rejected. The exact repair depends on the claim. Scientific evidence is fit-for-purpose: an observation that answers one question may not answer another.
What evidence would weaken the claim?
The claim weakens if the supporting result is rejected, if the qualifier key is missing, if the sample identity is uncertain, if a relevant control fails, if reanalysis produces a conflicting valid result, or if the claimed conclusion demands more sensitivity than the method can provide. The important move is to state which link weakens and why.
A vague sentence such as “The experiment is unreliable” throws away too much information. A better sentence is: “The rejected result cannot support the stated presence/absence conclusion because the relevant quality-control evidence did not meet the acceptance criteria; the other accepted samples are not automatically invalidated.” Scope is part of scientific accuracy.
How far can the conclusion travel?
Even after a successful replacement test, keep the conclusion inside its evidence boundary. One valid sample from one location and time does not prove the same condition everywhere. A valid laboratory measurement does not automatically prove the cause of the result. A non-detect does not automatically prove zero. A rejected result does not automatically prove absence. These boundaries are different, but they share one habit: do not make evidence travel farther than its design allows.
Tempting reasoning that sounds sensible—but is not
| Tempting statement | Why it fails | Better move |
|---|---|---|
| “Rejected means zero.” | It confuses unusable evidence with a physical measurement. | Read the rejection reason and preserve the result as unresolved for the affected decision. |
| “Rejected means detected.” | Quality failure does not automatically support the opposite claim. | Separate evidence quality from target presence. |
| “There is a number, so we can use it.” | A displayed number may carry a qualifier that limits or forbids its use. | Read the number and status together. |
| “One rejected result makes the whole study false.” | The rejection may be local to one sample, analyte, batch or decision. | Identify exact scope before widening the criticism. |
| “The laboratory is accredited, so no result can be rejected.” | Quality systems exist partly to detect and manage problems. | Treat transparent rejection as part of evidence control, then evaluate the specific result. |
A PSLE-style transfer case
This is an original practice case, not a past examination question.
A student tests water collected from three containers for substance Z. The laboratory report states:
| Container | Result for Z | Report note |
|---|---|---|
| P | 4.2 mg/L | Accepted |
| Q | Not detected | Quality checks acceptable |
| R | — | Rejected because the batch control failed |
Question: A classmate says, “Container R definitely had no substance Z because the laboratory did not report a concentration.” Explain why this conclusion is not supported.
Explained answer: The result for R was rejected because the batch control failed, so the measurement does not provide valid evidence for deciding the presence, absence or concentration of Z. A missing usable concentration is therefore not the same as a valid result of zero or a valid non-detect. New valid evidence would be needed.
Notice what the answer does not do. It does not claim that R contained Z. It does not claim that R was clean. It does not invent a laboratory rule. It explains the evidence boundary caused by the failed quality check.
Transfer Case 2: The bar chart with a hidden rejected sample
A class receives a bar chart showing the mean result from five samples at Site A and five at Site B. Later they discover that one Site B result had been rejected but was entered as zero before the mean was calculated.
What is the scientific problem? Entering the rejected result as zero creates numerical evidence that was never measured. The mean may be pulled downward, and the chart hides the quality problem. A better approach is to follow the documented data-use rule, show the number of valid results and preserve the rejected case as rejected rather than silently manufacturing a zero.
This also connects to the general Reality Lab warning that a graph can hide important provenance. For a different but related non-detect substitution problem, see PSLE Science Reality Lab Vol No.130.
Practice laboratory: classify the evidence status
For each original mini-case, decide whether the evidence supports the claim, weakens it, or leaves it unresolved.
Practice 1
A report says “Target not detected; all required quality checks acceptable.” A poster says, “The test did not detect the target under the stated method conditions.”
Answer: Supported, provided the poster preserves the method and detection boundary. It still should not strengthen the claim into “the true amount was exactly zero everywhere.”
Practice 2
A report says “Result rejected; sample identification uncertain.” A student says, “The value proves the sample from Location A was high.”
Answer: Not supported. If sample identity is uncertain, the result cannot safely be attached to Location A.
Practice 3
Eight results are accepted. One result is rejected. A headline says, “All nine measurements prove the same pattern.”
Answer: The word “all” is too strong. The eight accepted results may support a pattern, but the rejected result cannot be counted as another valid confirming measurement.
Practice 4
A rejected result is later repeated with a new properly identified sample. The new result passes all relevant quality checks and differs from the old printed number.
Answer: The new valid evidence should carry the decision. The old rejected number does not gain authority merely because it appeared first.
Practice 5
A dashboard removes all rejected records without showing how many were removed. The remaining graph looks smooth.
Answer: The graph may still contain useful accepted data, but the communication is incomplete. The reader needs enough provenance to know the exclusion rule and whether missing/rejected observations could change the interpretation.
Delayed independent return
Come back to this section later without looking at the earlier explanation.
A fictional soil report shows three fields: Sample S1 = 12 units, accepted; Sample S2 = 10 units, estimated; Sample S3 = no usable value, rejected because required quality-control evidence failed. A social post states: “S3 was the cleanest because its value was zero.”
Can you identify the exact reasoning error in one sentence?
Return answer: The post has converted an unusable measurement into a measured zero; rejection means the result cannot support the intended conclusion, not that the physical amount was zero.
Model and measurement limits
No short classroom model can represent every laboratory quality system. Codes differ. Acceptance rules differ. Some data can be used with qualifications; other data must be excluded for a particular analysis. A result may be rejected for one use but preserved for audit or investigation. Therefore this guide deliberately avoids teaching a universal letter-code dictionary.
The transferable scientific reasoning is more stable than any one code: identify what was measured, recover the qualifier definition from the relevant source, understand what quality condition failed, decide which conclusion that failure prevents, and seek appropriate valid evidence rather than inventing zero, absence or certainty.
How this connects to PSLE Science without becoming an exam trick
The current 2026 PSLE Science assessment framework states that candidates apply scientific inquiry by making predictions and hypotheses, interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning. None of that requires a pupil to memorise laboratory jargon. The transfer comes from the structure of the reasoning.
- Interpreting: read the result and qualifier together.
- Analysing: distinguish accepted, estimated, non-detect and rejected evidence.
- Evaluating: ask whether the method evidence is strong enough for the stated conclusion.
- Communicating: say “unresolved by this rejected result” when that is the honest conclusion.
There is no universal examiner phrase such as “rejected means cannot conclude” that should be inserted blindly. The explanation must match the evidence supplied in the question or source.
Parent and tutor teaching guide
A useful teaching sequence takes about fifteen minutes and does not require specialist laboratory equipment.
- Step 1: Show four cards: Accepted, Non-detect, Estimated, Rejected. Ask the learner what each status allows them to say.
- Step 2: Give one number beside each card. Watch whether the learner overvalues the printed number and ignores the status.
- Step 3: Remove the qualifier column and ask how the conclusion changes. The correct response is that important evidence has been lost.
- Step 4: Replace “Rejected” with a specific reason such as “sample identity uncertain.” Ask which link in the evidence chain is now broken.
- Step 5: Change the claim. First ask about presence/absence; then ask about exact concentration. Learners should notice that different claims require different evidence.
- Step 6: End with a new context—soil, air, food, a material test or a sensor—so the habit transfers beyond the original story.
Do not reward the child merely for saying “cannot conclude.” Ask, “Cannot conclude what, from which evidence, because which quality condition failed?” That pushes the learner from slogan to scientific reasoning.
A compact decision card
When you see a rejected scientific result, run this quiet sequence:
- Read the object: What sample and quantity are we talking about?
- Read the status key: What does rejected mean in this dataset?
- Locate the failed link: Identity, method, control, calibration, handling or another documented condition?
- Bound the claim: What can no longer be decided from this result?
- Do not invent a value: Rejected is not zero.
- Ask for repair evidence: What valid observation would settle the claim?
Authoritative source trail
The official 2026 PSLE Science syllabus from the Singapore Examinations and Assessment Board states that the paper assesses attainment in the 2023 Primary Science syllabus and includes interpreting and analysing information, evaluating observations, information and methods, and communicating explanations and reasoning.
The Ministry of Education’s 2023 Primary Science Teaching and Learning Syllabus explicitly includes healthy scepticism: questioning observations, methods, processes and data, as well as reviewing one’s own ideas.
The United States Environmental Protection Agency’s current Data Considerations guidance explains the purpose of data qualifiers, warns that different datasets can use different qualifier systems, and gives examples where rejected data should not be used because quality-control criteria were not met and presence or absence cannot be determined from the data.
These sources support the evidence habit in this guide. They do not turn this page into an official examination answer key, and this page does not reproduce proprietary test questions or laboratory procedures.
Quiet return: do not turn an evidence failure into a physical fact
A rejected result can feel like an empty space, and empty spaces invite the mind to fill them. That is exactly when scientific discipline matters most. If the evidence cannot support presence, absence or amount, keep the answer open. Do not manufacture zero. Do not manufacture detection. Do not punish the whole dataset without checking scope. Name what failed, preserve what is still known, and ask what valid evidence would settle the question.
That habit is small enough for a Primary 5 learner to practise and strong enough to carry far beyond PSLE Science: when the measurement is rejected, the honest result is not “nothing.” The honest result is “this evidence cannot answer that claim yet.”
