CHEMOMETRIC PROTOCOL TO VALIDATE AN ANALYTICAL METHOD IN THE PRESENCEOF CORRIGIBLE CONSTANT AND PROPORTIONAL SYSTEMATIC-ERRORS

A statistical methodology to verify the trueness of an analytical meth od in the presence of corrigible systematic errors is presented. This protocol enables detection of constant and proportional components of error. By using the data set obtained in the Youden calibration with d ifferent sample test portions, the constant component of the error (Yo uden blank) can be determined. An analysis of covariance was applied t o 3 calibration curves established with standard solutions and with st andard additions to 2 different sample test portions. The slopes were compared, and the presence of any matrix-analyte interaction was detec ted. A method for removing the numerical components of systematic erro rs is proposed: a calculation procedure to obtain a correct analytical result and a statistical test to verify the correctness of analyte co ntents obtained from different calibrations. For demonstration purpose s, the protocol was applied to spectrofluorometric determination of ox alates in spinach leaves.