Decision rule pursuant to DIN EN ISO/IEC 17025:2018
Decision rule not taking account of the measurement uncertainty
In case a directive or specification says the measurement uncertainty does not need to be considered or has been taken account of when calculating the requirement (limit value) – e.g. in a pharmacopeial monograph or a standard – the decision rule will be applied without considering the measurement uncertainty. The only action will be to compare the limit values with the analysis result and/or the average of all analysis results.
Unless you wish to use another decision rule, we will continue to apply the above rule to all quantitative analyses.
Decision rule taking account of the measurement uncertainty
Another option is to assume the extended measurement uncertainty as k=2 and the confidence level as approximately 95%. Comparing the specifications will then yield one of four results:
1) Both the measured value and the measurement uncertainty are out of the target range (result is “reliably out of range”):
Non-conforming sample. The risk of a wrong evaluation is very low.
2) Measured value is out of the target range, measurement uncertainty is partially within range (result is “not reliably out of range”):
Non-conforming sample, since the measured value exceeds the target value. Considering the extended measurement uncertainty (k=2), there is still a chance of the value meeting the requirements, though.
3) Measured value is within the target range, measurement uncertainty is partially outside range (result is “not reliably within range”):
Conforming sample, since the measured value is within the target range. Considering the extended measurement uncertainty (k=2), there is still the risk of the value failing to meet the requirements, though.
4) Both the measured value and the measurement uncertainty are within the target range (result is “reliably within range”):
Conforming sample. The risk of a wrong evaluation is very low.
Feel free to contact us if you have any questions.
Your Analytics Team