92
7 Validation of the Measurement Procedure
algorithms. This applies, for example, to using the various tests for rejecting outliers,
various classifying tests, whether the results belong to the same population, different algorithms depending on the determination of the calibration, as well as various
methods of estimating uncertainty. The primary criterion is that for the given type
of measurements, the same algorithms should always be used. Moreover, the analyst
must prove that the algorithm is correct for a particular measurement procedure.
This requirement stems from the fact that, depending on the intended purpose of the
research, there are different methods of calibration, a different number of repetitions,
and differences depending on the detector response to the actual sample composition.
Thus, the calculations involved must be selected individually, and the legitimacy of
the choice must be confirmed in the validation process.
In practice, the term ‘software validation,’ is used, which refers, for example, to
spreadsheets. In this case, it should be considered as a mental shortcut. The user of the
spreadsheet does not need to validate the software as such, provided that the software
is delivered by a renowned manufacturer or a source for which the manufacturer has
supplied a suitable report confirming that the program meets certain requirements
(verification).
The above-mentioned term applies in practice to the validation of algorithms
included in a spreadsheet and used for the conversion of the inserted raw data. In
commercially available spreadsheets, there are algorithms for rounding numbers, and
for executing multiple operations that may affect the final result of the calculation.
In practice, so-called ‘software validation’ is carried out by performing a model
calculation (for the prepared set of data), both using the software program as well as
another program or a calculator. The comparability of both results within the range
of accepted uncertainty is considered to be confirmation that the requirements for
the intended use have been met.
7.6 Standardized Versus Laboratory-Developed
Procedures
It is clear that the analyst should use analytical methods and measurement procedures
appropriate for the expected application. According to the requirements of ISO/IEC
17025:2005, methods published in the appropriate international, regional or national
standards or by reputable technical organizations are preferred. This is mainly due to
the fact that standardized methods have undergone the laborious process of validation
by expert laboratories and are considered optimal for a particular purpose. Moreover,
the use of the same analytical procedures in laboratories conducting the same kind
of testing (e.g., the monitoring of surface water pollution, food inspection, clinical
trials), is the right basis for the metrological comparison of results, especially when
they are dependent on the applied measurement procedure. However, it is important
to note that the use of a standardized method does automatically ensure the correct
result; the laboratory should confirm (verify) that their conditions are appropriate for
7 Validation of the Measurement Procedure
algorithms. This applies, for example, to using the various tests for rejecting outliers,
various classifying tests, whether the results belong to the same population, different algorithms depending on the determination of the calibration, as well as various
methods of estimating uncertainty. The primary criterion is that for the given type
of measurements, the same algorithms should always be used. Moreover, the analyst
must prove that the algorithm is correct for a particular measurement procedure.
This requirement stems from the fact that, depending on the intended purpose of the
research, there are different methods of calibration, a different number of repetitions,
and differences depending on the detector response to the actual sample composition.
Thus, the calculations involved must be selected individually, and the legitimacy of
the choice must be confirmed in the validation process.
In practice, the term ‘software validation,’ is used, which refers, for example, to
spreadsheets. In this case, it should be considered as a mental shortcut. The user of the
spreadsheet does not need to validate the software as such, provided that the software
is delivered by a renowned manufacturer or a source for which the manufacturer has
supplied a suitable report confirming that the program meets certain requirements
(verification).
The above-mentioned term applies in practice to the validation of algorithms
included in a spreadsheet and used for the conversion of the inserted raw data. In
commercially available spreadsheets, there are algorithms for rounding numbers, and
for executing multiple operations that may affect the final result of the calculation.
In practice, so-called ‘software validation’ is carried out by performing a model
calculation (for the prepared set of data), both using the software program as well as
another program or a calculator. The comparability of both results within the range
of accepted uncertainty is considered to be confirmation that the requirements for
the intended use have been met.
7.6 Standardized Versus Laboratory-Developed
Procedures
It is clear that the analyst should use analytical methods and measurement procedures
appropriate for the expected application. According to the requirements of ISO/IEC
17025:2005, methods published in the appropriate international, regional or national
standards or by reputable technical organizations are preferred. This is mainly due to
the fact that standardized methods have undergone the laborious process of validation
by expert laboratories and are considered optimal for a particular purpose. Moreover,
the use of the same analytical procedures in laboratories conducting the same kind
of testing (e.g., the monitoring of surface water pollution, food inspection, clinical
trials), is the right basis for the metrological comparison of results, especially when
they are dependent on the applied measurement procedure. However, it is important
to note that the use of a standardized method does automatically ensure the correct
result; the laboratory should confirm (verify) that their conditions are appropriate for