150
8 Measurement Uncertainty
Proficiency testing
The main objective of the proficiency testing (PT), executed via ILC, is to evaluate
the competences of a laboratory; thus the results of PT tests can also be a good
source of data for the evaluation of uncertainty of measurements conducted in a given
laboratory. The use of the results of the PT is subject to the same conditions as the use
of any other results of ILC. The only limitation, described in various documents (e.g.,
EA-4/16 ‘EA Guidelines on the Expression of Uncertainty in Quantitative Testing’)
is that in some cases the proficiency testing is not conducted regularly enough so that
it could be possible to collect a representative set of data.
Another limitation with the use of the data from PT for evaluation of the uncertainty is that the tested object does not sufficiently reflect the properties of samples
that are routinely tested in a given laboratory. In that case, the uncertainty budget
should include the predicted differences in the behavior of the object tested in the
laboratory with regard to the routine samples. This should be also considered when
the uncertainty varies over the concentration range for which the procedure is applied
in the given laboratory.
8.22 Conclusions
In practice, four approaches for evaluation of uncertainty could be applied. The
particular approach can be selected, depending mainly on the purpose of conducting
measurements and depending on the availability of source data. For this reason, the
use of a combined approach is considered to be the most effective.
It should be highlighted, that apart the different sources of uncertainty components, the general process in always the same for all four approaches. The process
always starts with the specification of measurand, which means there is a necessity
to clarify and define the quantity intended to be measured.
Measurand: the quantity intended to be measured.
Clause 2.3; ISO/IEC Guide 99
Whenever the measurand was defined, the measurement procedure should be
selected and described in the form of the mathematical function (model equation).
The measurement procedure should be selected considering the measurand of interest as well as target uncertainty and available resources. The measurement function should first of all ensure the proper calculation of the final results and should
be updated when all possible sources of uncertainty are identified. In this respect,
understanding the entire analytical procedure and especially all possible effects that
could affect measurements is a crucial part of the evaluation of uncertainty.
The next step is related to the quantification of the uncertainty components and
is different for all possible approaches, as described above.
8 Measurement Uncertainty
Proficiency testing
The main objective of the proficiency testing (PT), executed via ILC, is to evaluate
the competences of a laboratory; thus the results of PT tests can also be a good
source of data for the evaluation of uncertainty of measurements conducted in a given
laboratory. The use of the results of the PT is subject to the same conditions as the use
of any other results of ILC. The only limitation, described in various documents (e.g.,
EA-4/16 ‘EA Guidelines on the Expression of Uncertainty in Quantitative Testing’)
is that in some cases the proficiency testing is not conducted regularly enough so that
it could be possible to collect a representative set of data.
Another limitation with the use of the data from PT for evaluation of the uncertainty is that the tested object does not sufficiently reflect the properties of samples
that are routinely tested in a given laboratory. In that case, the uncertainty budget
should include the predicted differences in the behavior of the object tested in the
laboratory with regard to the routine samples. This should be also considered when
the uncertainty varies over the concentration range for which the procedure is applied
in the given laboratory.
8.22 Conclusions
In practice, four approaches for evaluation of uncertainty could be applied. The
particular approach can be selected, depending mainly on the purpose of conducting
measurements and depending on the availability of source data. For this reason, the
use of a combined approach is considered to be the most effective.
It should be highlighted, that apart the different sources of uncertainty components, the general process in always the same for all four approaches. The process
always starts with the specification of measurand, which means there is a necessity
to clarify and define the quantity intended to be measured.
Measurand: the quantity intended to be measured.
Clause 2.3; ISO/IEC Guide 99
Whenever the measurand was defined, the measurement procedure should be
selected and described in the form of the mathematical function (model equation).
The measurement procedure should be selected considering the measurand of interest as well as target uncertainty and available resources. The measurement function should first of all ensure the proper calculation of the final results and should
be updated when all possible sources of uncertainty are identified. In this respect,
understanding the entire analytical procedure and especially all possible effects that
could affect measurements is a crucial part of the evaluation of uncertainty.
The next step is related to the quantification of the uncertainty components and
is different for all possible approaches, as described above.