New Robust Method for the Treatment of Analytical Data

摘要

A new robust method is proposed for the calculation of the median and the relative standard deviation for a set of data containing unsymmetrically placed outliers. Examples of the advantages of the new method are given, the findings being confirmed by Monte Carlo tests. At Mintek mass-spectrometric data are often unsymmetrical, and all outliers are often at one of the extremities, either all high or all low. In some instances the outliers can be seen visually, and would therefore be rejected on sight. However, when large sets of results are processed, such as those from mass-spectrometric, X-ray fluorescence, and neutron-activation analyses, or in fields such as the assignment of values to reference materials, the sets of results are often too large for visual inspection and are processed within the computer. (Atomindex citation 14:773041)