Why it matters
ISO/IEC 17025 clause 7.6 requires laboratories to identify the contributions to measurement uncertainty and evaluate it. Beyond compliance, uncertainty tells you how confident you can be when a result is compared with a legal limit.
- No uncertainty estimate, or one fixed value applied to every parameter
- Significant sources — sample preparation, recovery, matrix effects — left out
- No documented rationale or data source behind the numbers
- No decision rule for results close to a limit
Approaches we use
GUM (modelling) approach
Based on JCGM 100:2008. A measurement model is built and each input's uncertainty is combined using sensitivity coefficients. Common in calibration and physical measurements.
Nordtest (data-based) approach
Following Nordtest TR 537: within-laboratory reproducibility combined with bias from reference materials, recovery or proficiency testing data. Practical and defensible for chemical testing.
Eurachem/CITAC guidance
Guidance on quantifying uncertainty in analytical measurement, supporting the use of validation data.
Sampling uncertainty
Cases in environmental work where uncertainty arising from sampling needs to be assessed separately.
How to calculate measurement uncertainty
- Specify the measurand and how the result is calculated.
- Identify sources with a cause-and-effect (fishbone) diagram.
- Quantify each source from validation, calibration certificates, QC and PT data.
- Convert and combine into a combined standard uncertainty.
- Expand, typically with k=2 (approximately 95% coverage).
- Report and define the decision rule for statements of conformity (clause 7.8.6).
The strongest input to an uncertainty budget is a well-designed method validation; proficiency testing results can supply the bias component.
Deliverables
- Uncertainty budgets per method with a rationale document
- Protected calculation templates your staff can reuse
- Measurement uncertainty procedure and decision rule
- Hands-on calculation training for your team