Comparison of different approaches to the statistical evaluation of proficiency tests

The objective of this study was to compare different z-score calculation for the results from proficiency testing programmes. The comparison was carried out on the data from proficiency tests for four different matrices and for all analytes included in the proficiency testing programmes organised by UKZUZ institute from the year 2005 to 2007. It represented 99 samples (three periods per year and eleven samples distributed in each period); the number of determined analytes was 108 in each period. The total number of investigated data sets was 953. Three different approaches to z-score calculation were compared. The results gave reasonably comparable evaluation in most cases but sometimes UKZUZ method proved to be rather strict. The procedure using the robust average and robust standard deviation calculated according to Algorithm A described in ISO 13528 was found to be the most suitable for matrices and level of measurands in UKZUZ proficiency tests.