The influence of different evaluation techniques on the results of interlaboratory comparisons

Abstract

 The influence of different evaluation techniques on the results of an interlaboratory comparison for the determination of nutrients in ground- and surface water was investigated. The outlier-test procedure was found to influence the interlaboratory standard deviations (SDs), but not the averages. It was shown that even small differences in the numbers of outliers detected can change the SD severely. Comparing the outlier-test procedures of Hampel, Grubbs and Graf-Henning, it was found that Hampel's test detected the most outliers, thus generally resulting in smaller SDs between interlaboratory comparisons. The Graf-Henning test detected the fewest outliers and its application resulted in the highest SDs of the three test procedures investigated. The comparison of different summarising indices, namely the rescaled sum of z-scores, average of absolute z-scores and average deviation showed no comparability. Possibilities to improve the comparability of interlaboratory comparisons and to minimise misunderstandings are suggested.