Research Library · Literacy
How to evaluate an analytical claim
Most claims can be sorted in under two minutes, without any chemistry, by asking what would have to exist for the claim to be checkable.
Three verdicts, not two
Claims about testing are usually sorted into “trustworthy” and “dubious”, which is a poor scheme because it forces a judgement about the material from evidence that is only about the documentation. A better scheme has three outcomes:
- Verifiable. Every element needed to check the claim is present and could be followed up independently.
- Partially verifiable. Some elements are present; you can state exactly which are missing.
- Unsupported. Nothing in the claim can be checked by anyone outside the party making it.
“Unsupported” is not an accusation. It is a statement about what has been shown, and keeping it that way is what makes the assessment defensible.
The five-element test
For any testing claim, look for five elements. Each is a yes or no.
| Element | Present when… |
|---|---|
| 1. A subject | A named material with a batch code, not a generic “our peptides” |
| 2. A method | The technique and its key parameters are stated |
| 3. A result | A measured value appears, distinct from a specification or target |
| 4. An author | A named laboratory, ideally with accreditation details covering the test |
| 5. A date | Analysis and report dates that sit plausibly against the batch date |
Five out of five is verifiable. One or two missing is partially verifiable, and the missing ones are worth naming out loud — “a result with no method” is a much more useful description than “looks vague”. Missing a subject or an author is close to fatal: without them there is nothing to check and no one to ask.
Phrases that carry less than they appear to
| Phrase | What it actually asserts |
|---|---|
| “Lab tested” | A test happened somewhere, on something, at some point |
| “Third-party tested” | An unnamed independent party tested an unspecified sample |
| “99%+ purity” | A specification or aspiration, unless a measured value is given |
| “COA available on request” | A document exists; its contents remain unknown |
| “Pharmaceutical grade” | No defined meaning outside a pharmacopoeial monograph and a compliance claim against it |
| “Batch verified” | Someone holds a record; independence is unstated |
| “Research grade” | A use restriction, not a quality level |
None of these phrases is inherently dishonest. Each simply asserts less than a reader typically takes from it, and the gap is where most misplaced confidence lives.
Two failure modes to avoid in yourself
Over-reading. Treating a document as an approval. Certificates report measurements; they do not establish safety, suitability or legality for any purpose, and no analytical result can.
Over-dismissing. Treating an incomplete document as proof of bad faith. Most research certificates are incomplete because full characterisation is expensive and was never commissioned. The correct conclusion is “these questions are unanswered”, which is precise, rather than “this is fake”, which usually is not supported either.
Writing the conclusion down
A useful habit is to state the finding in one sentence that separates document from material. For example: “The report gives a purity result with a method and a named laboratory, but the sample identity does not include a batch code, so the result cannot be attributed to this container.”
That sentence is checkable, contains no accusation, and tells you exactly what would resolve it. It is also, in the end, all that analytical literacy is for. The COA guide covers the field-by-field mechanics, and batch traceability covers the attribution step.
References
- Magnusson B, Örnemark U (eds). Eurachem Guide: The Fitness for Purpose of Analytical Methods, 2nd edition. Eurachem, 2014
- International Organization for Standardization. ISO/IEC 17025:2017 — reporting of results and statements of conformity. iso.org
- United Kingdom Accreditation Service. Checking an accredited scope of testing. ukas.com