148
8 Measurement Uncertainty
– From the assessment of the results of inter-laboratory comparison (with the
use of CRM).
The uncertainty that estimates the influence of all random effects (u rand ) can
be delivered from the data:
– Of test samples (stable over the time of collecting data);
– Of a blank sample referring to the entire analytical procedure;
– Of a reference material with a certified content of the analyte.
The procedure used for the evaluation of uncertainty from validation data:
1. Defining the measurand (analyte, sample type);
2. Calculating recovery value R and the assigned uncertainty;
3. Calculating laboratory reproducibility;
4. Evaluating the standard uncertainties of the systematic and random effects;
5. Calculating the combined standard uncertainty: using the law of propagation;
6. Calculating the expanded uncertainty with the selected coverage factor k.
Data from the control charts
When the evaluation of uncertainty is performed within validations, useful data are
collected over the relatively short period of time in which the validation was conducted. In laboratory practice, validation experiments are often conducted by one
employee, with the use of one set of glassware, reagents and instruments. This means
that the dispersion of data reflects the uncertainty related to the random effects occuring over the validation time.
In the long-term, a larger dispersity of results can be expected, due to the higher
possible variability of the measurement conditions. With this in mind, whenever
possible, it is recommended that data (e.g., standard deviation) from control charts
is used. When the control charts were completed over a long time or a large enough
amount of data was collected, it can be assumed that standard deviation reflects the
intra-laboratory reproducibility, encompassing the changeability of the following
parameters:
– Personnel conducting the test (when more than one person is authorized to conduct
the test);
– Environmental conditions (temperature, humidity);
– Equipment (stability, repeatability);
– Reagents (stability, purity);
– Standards (uncertainty of certified value, matrix).
8 Measurement Uncertainty
– From the assessment of the results of inter-laboratory comparison (with the
use of CRM).
The uncertainty that estimates the influence of all random effects (u rand ) can
be delivered from the data:
– Of test samples (stable over the time of collecting data);
– Of a blank sample referring to the entire analytical procedure;
– Of a reference material with a certified content of the analyte.
The procedure used for the evaluation of uncertainty from validation data:
1. Defining the measurand (analyte, sample type);
2. Calculating recovery value R and the assigned uncertainty;
3. Calculating laboratory reproducibility;
4. Evaluating the standard uncertainties of the systematic and random effects;
5. Calculating the combined standard uncertainty: using the law of propagation;
6. Calculating the expanded uncertainty with the selected coverage factor k.
Data from the control charts
When the evaluation of uncertainty is performed within validations, useful data are
collected over the relatively short period of time in which the validation was conducted. In laboratory practice, validation experiments are often conducted by one
employee, with the use of one set of glassware, reagents and instruments. This means
that the dispersion of data reflects the uncertainty related to the random effects occuring over the validation time.
In the long-term, a larger dispersity of results can be expected, due to the higher
possible variability of the measurement conditions. With this in mind, whenever
possible, it is recommended that data (e.g., standard deviation) from control charts
is used. When the control charts were completed over a long time or a large enough
amount of data was collected, it can be assumed that standard deviation reflects the
intra-laboratory reproducibility, encompassing the changeability of the following
parameters:
– Personnel conducting the test (when more than one person is authorized to conduct
the test);
– Environmental conditions (temperature, humidity);
– Equipment (stability, repeatability);
– Reagents (stability, purity);
– Standards (uncertainty of certified value, matrix).