What this run supports, and what it does not

The ten-replicate guided example supports two narrow, defensible conclusions: this run was precise (%CV 1.57%) and showed a small, non-actionable bias (plus 0.5%) against the assigned target. The four-data-set exercise supports a classification of each data set by precision and bias and a defensible next investigation for each. Neither exercise establishes the method's within-laboratory precision, its reproducibility, or a formal measurement uncertainty estimate, because those require a structured multi-day, multi-level design, not a single run or a single day's comparison.

Measurement uncertainty is a quantitative expression of the dispersion reasonably attributable to a reported result. It is typically built from a combined estimate of precision and bias information and reported as an expanded interval using a stated coverage factor. It is method-specific, concentration-specific, and often laboratory-specific; it is not one fixed number that applies to every analyte or every laboratory, and it is not the same model as a total allowable error calculation. The relationship between measurement uncertainty and total error remains an active, unresolved discussion in the peer-reviewed literature; no single framework has replaced the other.

Passing a proficiency testing criterion does not prove a method has no measurement uncertainty, and a proficiency testing failure does not by itself say whether the cause was imprecision, bias, or both; that determination needs a precision and bias breakdown, not the pass or fail score alone.

Use the laboratory's own QC rules, analyte-specific allowable-error limits, and measurement-uncertainty procedure. The glucose criterion shown here does not transfer to other analytes.

State what a data set supports and stop there; do not let a single run's precision or bias stand in for a formal precision study, a formal bias study, or a measurement uncertainty estimate.