2.8 Other Analytical Properties
85
Fig. 2.20. Relationships between robustness and other features of analytical methods
Figure 2.20 relates robustness to other analytical features. The most immediate relation is to precision (and, indirectly, to accuracy). In fact, robustness
is measured by the difference produced by random changes in the experimental
factors that introduce variability in the results; consequently, robustness is also
related to uncertainty.
The reliability of a method is defined as its ability to preserve its accuracy and
precision over time. This feature, which is directly related to robustness, was
formerly referred to as "safety". At present, this word is used in the context of
accessory analytical properties to describe the effects of a method on human
health or the environment.
The transferability of a method is a practical property that represents its
capacity to provide consistent results when used by different laboratories. It
relies heavily on robustness and on a detailed description of the method (the
"procedure" included in the hierarchy of Fig. 1.16). Transferability ensures
comparability among laboratories, an ancillary property that leads to mutual
acknowledgement of their results.
Box 2.17
A study of robustness should begin with a careful identification of those experimental (physical, physico- chemical and chemical) factors that are crucial to the analytical
method.
A univariate study, where each individual factor is altered while all others are kept
constant, is labour-intensive, time-consuming and thus inadvisable. For example, with n = 7
factors, an overall i = 128 experiments are required.
It is preferable to conduct a multivariate study by using an experimental design approach
or even an expert system.
The YOUNDEN-STEINER procedure is widely used in this context on account of its simplicity.1 t involves the followi ng steps:
1. The variables or factors to be processed (n = 7) are strategically selected.
2. A high value (A - G) and a low value (a - g) for each factor are chosen.
85
Fig. 2.20. Relationships between robustness and other features of analytical methods
Figure 2.20 relates robustness to other analytical features. The most immediate relation is to precision (and, indirectly, to accuracy). In fact, robustness
is measured by the difference produced by random changes in the experimental
factors that introduce variability in the results; consequently, robustness is also
related to uncertainty.
The reliability of a method is defined as its ability to preserve its accuracy and
precision over time. This feature, which is directly related to robustness, was
formerly referred to as "safety". At present, this word is used in the context of
accessory analytical properties to describe the effects of a method on human
health or the environment.
The transferability of a method is a practical property that represents its
capacity to provide consistent results when used by different laboratories. It
relies heavily on robustness and on a detailed description of the method (the
"procedure" included in the hierarchy of Fig. 1.16). Transferability ensures
comparability among laboratories, an ancillary property that leads to mutual
acknowledgement of their results.
Box 2.17
A study of robustness should begin with a careful identification of those experimental (physical, physico- chemical and chemical) factors that are crucial to the analytical
method.
A univariate study, where each individual factor is altered while all others are kept
constant, is labour-intensive, time-consuming and thus inadvisable. For example, with n = 7
factors, an overall i = 128 experiments are required.
It is preferable to conduct a multivariate study by using an experimental design approach
or even an expert system.
The YOUNDEN-STEINER procedure is widely used in this context on account of its simplicity.1 t involves the followi ng steps:
1. The variables or factors to be processed (n = 7) are strategically selected.
2. A high value (A - G) and a low value (a - g) for each factor are chosen.