EDUKATE SINGAPORE · GOLD STANDARD PRACTICAL CASE STUDIES · CASE 128
A laboratory result marked ‘not detected’ does not prove that a chemical is absent.
A fictional water-monitoring team receives eight nondetect reports, but the laboratory’s reporting limit is four times higher than the programme’s intended decision threshold. This Gold Standard case explains uncertainty, method sensitivity and accurate public claims.
All locations, chemical thresholds, sample results and institutions are fictional. This is scientific evidence-management education, not drinking-water guidance, a health-risk finding or an interpretation of any actual water source.
Chapter index
- 1. The word ‘not detected’ becomes a promise of clean water
- 2. A detection result is not a zero concentration measurement
- 3. The worked numbers reveal why the green result is inconclusive
- 4. What EPA says about detection and quantitation
- 5. The actual sample history still matters
- 6. An inconclusive result requires a different status from a pass
- 7. Choose measurement capability before ordering the next test
- 8. Keep detection, decision and safety claims in separate columns
- 9. Test the report using an intentionally insufficient limit
- 10. Practice: nondetect above the decision threshold
1. The word ‘not detected’ becomes a promise of clean water
A fictional community science partnership collects water samples from an education centre’s demonstration reservoir. Its preliminary laboratory contract asks whether a particular fictional chemical can be detected under a standard screening method. The laboratory sends back a report marking the substance ‘not detected’ for eight tested samples. The project dashboard records eight green results.
The centre’s publicity team prepares a leaflet stating that the samples prove the water contains none of the chemical and is safe for every intended use. A scientist then asks about the laboratory’s reporting limit. The particular analytical method could not report the chemical below 0.020 milligrams per litre under the relevant sample conditions. The project had also been using a separate hypothetical decision threshold of 0.005 milligrams per litre in its educational planning.
The laboratory may have performed its agreed screen correctly, yet the screen is not sensitive enough to answer the more demanding question the publicity team is asking. The Gold Standard issue is the unsupported step from ‘the method did not detect it’ to ‘the substance is absent or below a lower decision threshold’.
2. A detection result is not a zero concentration measurement
The report contains sample identifier W1, collection and handling record W2, analytical method W3, sample-specific reporting limit W4 and result W5. W5 is reported as not detected under the chosen method. That language describes a property of the measurement, not a guarantee that the true concentration is exactly zero.
Under a simplified teaching interpretation, a report shown as ‘less than 0.020 mg/L’ does not determine whether the true concentration is 0, 0.003, 0.012 or another value below the reporting boundary. The detailed analytical meaning may depend on the method, detection and quantitation definitions, matrix effects and laboratory reporting practice. A competent scientist must interpret the actual report.
GS-R08 of eduKate’s Gold Standard framework requires the public claim to remain within the evidence’s demonstrated boundary. GS-R02 requires the original question to be specified before picking the method. If the project wants evidence about 0.005 mg/L, it cannot automatically rely on a method whose relevant reporting limit is four times higher.
An honest nondetect is useful information. It becomes misleading when it is treated as a precise zero or as proof that a different, more demanding criterion was met.
3. The worked numbers reveal why the green result is inconclusive
The invented project decision threshold is 0.005 mg/L, while the laboratory’s sample reporting limit is 0.020 mg/L. In this example, 0.020 divided by 0.005 is 4. The method’s reporting boundary is therefore four times the level about which the project wanted to decide.
Consider a possible true concentration of 0.012 mg/L. That value lies below 0.020 but above 0.005. A ‘less than 0.020’ report therefore cannot exclude values that exceed the fictional decision threshold. The example is a logical illustration, not a statement that 0.012 was present in any sample.
The reverse also matters. A nondetect at a higher limit does not establish that the chemical was present above the criterion. Both a value beneath the criterion and a value above it could remain consistent with the simplified unresolved range. The outcome under the intended threshold is unknown on the current evidence.
The correct calculation concerns the adequacy of the method for the decision, not an arbitrary conversion of the nondetect into a half-limit value. Statistical methods for censored observations require an appropriate design and expertise.
4. What EPA says about detection and quantitation
The US Environmental Protection Agency’s Method Detection Limit frequently asked questions describe the method detection limit as a statistical analytical characteristic that can distinguish sample measurements from method-blank results with a specified level of confidence.
EPA’s Detection and Quantitation guidance distinguishes different detection and quantitation concepts used across methods and programmes. Some SW-846 contexts emphasise a lower limit of quantitation, while other regulatory methods use method-detection-limit concepts. The specific definition in the laboratory’s actual report therefore matters.
EPA’s guidance on chemical concentrations near detection limits explicitly warns against equating nondetects automatically with zero in risk assessment. These are authoritative US technical references, not standards assigning the fictional reservoir any legal chemical threshold.
The Gold Standard draws the management conclusion without inventing a health decision: the measurement’s sensitivity, reporting definition and sample conditions must be suitable for the question the organisation wants to answer.
5. The actual sample history still matters
Even if the laboratory used a suitably sensitive method, the project would need a credible sample chain. Collection location, timing, sample preservation, laboratory receipt, blank controls and other relevant procedures may affect what the measurement can establish. A beautiful analytical result from the wrong sample would answer the wrong environmental question.
In this fictional case, the eight labelled bottles were collected under a planned demonstration. Their identities and locations are documented. The immediate problem is the analytical reporting limit, not an alleged sample switch or a known contamination incident. The review should avoid inventing a broader failure unsupported by the record.
A laboratory may have followed the original purchase order, which merely requested ‘standard screen for substance X’. The science adviser needed the result to support a lower-threshold decision, but this criterion was not passed to procurement. Each team may have executed its local instructions while the final answer remained insufficient.
GS-R07 protects the transfer of decision-relevant criteria from the person asking the scientific question to the analyst performing the work.
6. An inconclusive result requires a different status from a pass
The dashboard initially shows eight green ‘clean’ results. The project lead corrects its status to explain that eight samples were nondetect under the reported method, while the current analysis does not determine whether the fictional lower planning threshold was met. The original laboratory values and timestamps are preserved unchanged.
Management should not recolour every sample red as though the chemical were confirmed present. That would be an equally unsupported inference. It also should not invent a health or consumption advisory from a teaching example. Decisions concerning actual water quality require the appropriate public-health, environmental and laboratory authorities and the specific governing standards.
The centre can describe the limitation publicly in plain language: ‘Our screening did not detect the substance at the laboratory’s stated reporting limit, but that limit was above the level relevant to our question. We therefore cannot infer the lower-level result from this test.’
GS-R09 requires transparent management of uncertainty. A useful ‘inconclusive for this decision’ status can be more honest and operationally valuable than a binary pass-or-fail icon.
7. Choose measurement capability before ordering the next test
The project revises its laboratory request process. The responsible expert defines the analyte, appropriate sample matrix, decision question, required sensitivity and reporting expectations. A competent laboratory then determines whether a suitable analytical method is available and what limitations may apply under the actual sample conditions.
The manager must not specify a fanciful precision beyond what a defensible method can provide. Nor should the centre select the cheapest test automatically when the reported results cannot inform the intended decision. Method selection is a scientific and procurement question connecting resources with purpose.
If the current samples are still usable under applicable analytical and preservation conditions, an authorised professional can consider whether further analysis is possible. If they are not, the project may need a properly designed new sampling programme. This article does not prescribe which action is valid for real materials.
GS-R05 asks whether resources and methods are ready for the claimed task. GS-R11 asks whether repeated monitoring has sustainable capability rather than relying on one impressive but insufficient certificate.
8. Keep detection, decision and safety claims in separate columns
The revised management report includes sample identity, analyte, method, detection or reporting boundary, observed result, specific decision criterion where legitimately applicable, and a qualified interpretation. It shows whether the method can actually discriminate around the question of interest.
A result can be technically valid but unsuitable for a particular regulatory or educational inference. A method might report a compound reliably at relatively high concentrations and still be valuable for screening, while not providing sufficient information for a lower threshold. That is not evidence of laboratory dishonesty.
The centre should also distinguish a one-time sample from a statement about every day or location. A result at one point cannot automatically establish the condition of an entire water system through time. Evidence must remain bounded by where, when and how it was collected.
GS-R08 is the central safeguard: a measurement’s limitations belong beside the number every time the result is used to justify an action or public claim.
9. Test the report using an intentionally insufficient limit
The quality reviewer receives synthetic laboratory reports: one nondetect with a reporting limit appropriately below a hypothetical decision level, another nondetect with a reporting limit above it, and a third with a positive measurement and associated uncertainty. The reviewer should distinguish what each case supports without deciding actual health safety.
A second exercise supplies a sample-specific dilution that changes the applicable reporting limit. The report must use the true limit for that sample rather than a generic method figure copied from an old brochure. The authorised scientist should be able to explain why a test is or is not suitable for the required decision.
The test also includes a publicity sentence asserting ‘zero contamination’ from a nondetect. Staff must qualify the claim and preserve the laboratory’s original language rather than silently recast it as a measured zero.
A successful exercise supports the organisation’s ability to recognise sensitivity limits; it does not certify an actual water source, analytical laboratory or future monitoring programme.
10. Practice: nondetect above the decision threshold
Exercise. Eight samples are reported as not detected with a relevant laboratory reporting limit of 0.020 mg/L. The project’s fictional decision question concerns 0.005 mg/L. Write the supported finding, one calculation, immediate public correction and retest. Do not claim that any particular value is present in the water.
A reasoned answer is: ‘The laboratory reported nondetection under the stated method. Its reporting boundary is four times the hypothetical decision level, so the results do not establish whether the samples meet that lower criterion. The centre should remove the unqualified zero-contamination claim and seek competent review of a method appropriate to the actual question.’
The Gold Standard lesson is that an accurate nondetect can still be an incomplete answer. Measurement excellence comes from selecting evidence fit for the decision, not from making a laboratory’s green result sound stronger than it is.
Analytical method references and related cases
- US EPA · Method Detection Limit FAQs
- US EPA · Detection and Quantitation
- US EPA · Handling Concentration Data Near Detection Limits
- Gold Standard · Measurement, Thresholds and Performance Review
- Gold Standard · Evidence, Claims and Verification
- Gold Standard · Information and Record Integrity
- Case 51 · River Survey Turned Units into a False Trend
- Case 118 · Microplastic Samples and Missing Blank Controls
- Gold Standard Reference Casebook
- Gold Standard Master Hub
EPA resources provide authoritative US analytical-method context, not a legal or health conclusion about the invented samples. GS-R01–GS-R12 is eduKate’s educational management reference.