A practical explanation of how to determine whether two laboratory results are genuinely comparable before interpreting their difference.
Establish whether the results are comparable
When two laboratories report different results, the first question is not which laboratory is wrong. First determine whether both results describe the same sample population, measured under sufficiently similar conditions. Compare product identity, lot or batch, sampling location, sampling date, sample size, and the way specimens were selected. Results from different batches, or from non-equivalent samples within one batch, may answer different questions.
Next compare the reported measurand: the exact property being measured and how it is defined. Similar-looking test names can conceal differences in units, endpoints, detection limits, calculation rules, or pass-fail criteria. A numerical comparison is meaningful only after these definitions align.
Compare methods and preparation
Review the method version, equipment configuration, environmental conditions, conditioning time, specimen dimensions, preparation steps, and any departures from the stated procedure. Drying, cleaning, cutting, storage, or temperature differences can change what reaches the instrument. Also check calibration status, controls, blanks, repetitions, data exclusions, and rounding practices where the reports disclose them.
Interpret uncertainty before judging the difference
A reported value is not perfectly exact. The NIST Technical Note 1297 guidance explains the general role of measurement uncertainty in expressing the dispersion reasonably associated with a result. Compare each laboratory’s stated uncertainty, coverage basis, repeatability information, and reporting resolution. Two central values can differ while their uncertainty ranges still indicate no clearly resolved conflict.
If comparability remains uncertain, describe the discrepancy without declaring laboratory error or product failure. A defensible next step is a documented review of sample traceability and methods, followed where necessary by testing retained or newly selected specimens under an agreed protocol.