The present paper deals with three important topics: calibration data, recovery assays, and comparison of analytical methods. Linear regression is applied in all the cases in order to check if the analytical model applied is correct, and the parameters obtained are subject to statistical tests. Residual analysis is also applied in order to check that the proposed model is correct. However, statistical tests on the slope and intercept of a straight line obtained by linear regression are not entirely reliable when the strong correlation existing between the slope and intercept is not taken into account. That is to say, the confidence rectangle should be substituted by a confidence ellipse, which describes the behaviour of the system in a more rigorous way. The methodology followed could be applied to several fields of analytical importance: determination of statin drugs by liquid chromatography coupled with triple quadrupole tandem mass spectrometry, and copper analysis by atomic absorption spectroscopy in the field of calibration; determination of perfluorooctanoic acid in biological samples and trigonelline in coffee for the study of recovery assays, as well as determination of cadmium in urine in the field of comparison of methods, among others.
Buy this Article