Exploratory study on median-based robust regression methods for linear calibration in atomic absorption spectrometric analysis

Abstract

A suitable has been developed either to avoid the influence of outliers or to reject them in the case of homoscedasticity in routine analysis. Several robust methods based on the median, namely the single median, the repeated median and the least median of squares, are introduced and evaluated. These methods were used for the analysis of real data obtained from atomic absorption spectrometric analysis. Their features and limitations in simple regression are discussed.