Reading Technical Evidence

Write the decision as a question. Are you selecting a test method, accepting a delivery, reviewing a process change or explaining a performance claim? Each can require different evidence.

Identify the requirement and the consequence of an incorrect conclusion. A report cannot answer a question that falls outside its subject, method or sample.

Record the exact product, configuration, material, process or population concerned. Then identify the object described by the evidence. Check model references, sample descriptions and any differences that could affect relevance.

A report on a component is not automatically evidence about the finished assembly. A measurement on one sample does not by itself establish the distribution of a characteristic across production.

Keep the method designation, edition, conditions and reporting basis together. Two values may use the same unit while describing different conditions or quantities.

Check whether the result is a measured value, a limit, a range, a declaration or a calculated value. A result reported as “less than” a limit is not an exact value at that limit.

Measurement uncertainty is part of the information needed to interpret a measurement. NIST's guidance addresses the evaluation and reporting of uncertainty; its application must remain within the measurement and decision context. NIST Technical Note 1297.

Identify who produced the information and what role that organization performed. A laboratory, report issuer, publisher and later commentator can be different parties.

Use the original document where permitted. If you have an excerpt, identify the missing sections and whether they could change the conclusion. A document number helps identify a report; it does not prove every statement copied from it is accurate.

Compare the evidence with the requirement and intended use. Ask:

• Does the sample represent the object of the decision?
• Is the method appropriate to the characteristic?
• Are the relevant conditions compatible?
• Is the document complete enough for this purpose?
• Have later changes affected its relevance?

Where the answer is unknown, record the gap and its effect. Do not resolve an unknown by assuming the most favorable case.

A useful conclusion states what the evidence supports, the conditions attached and what remains unresolved. For example, an internal review might conclude that a reported result is relevant to one configuration, while leaving the effect of a different material unresolved. That is more informative than an unrestricted statement that a product is “verified.”

Keep the source edition and reasoning with the decision. Revisit the conclusion when the object, requirement or evidence changes.

Use the Evidence Review Worksheet. For a related distinction, read Standards, Certification and Membership.

This guide is a professional reading aid. It does not determine compliance, authenticate a report or approve a product.