| Popis: |
The search for sources from which small changes have an influence on the response of an analytical system, the ruggedness test, has become an essential part of a validation study. However, the `definition' of ruggedness and how to perform a ruggedness test are still ambiguous. In this paper a strategy is proposed to perform a ruggedness test for chromatographic methods. Because the strategy is based on saturated designs, ANOVA cannot be applied to estimate the error variance. Therefore a graphical technique, the half-normal plot, is used to detect influential effects. In order to identify influential effects, the algorithms of Lenth and Dong are applied to four different examples. If a main effect is found to be significant, the confounded two-factor interactions are resolved by a follow-up design. The strategy of this sequential design is implemented in a SAS® software application called TRIC, testing ruggedness in chromatography. |
Full Text from ScienceDirect