Our statistical concept with all calculation formulas and the basics for the laboratory evaluation is described in detail in our statistical protocol.
The statistical protocol in revision level 30 is available for download (listed on the left). The latest edition can only be provided to registered users in our online system ODIN.
If you have basic questions about statistics or laboratory evaluation, please feel free to contact us.
Below you will find some general information about our statistical concept:
In principle, we evaluate proficiency tests simultaneously with different statistical models. This is based on the fact that the ideal statistic does not exist, just as the ideal data set for an evaluation does not exist.
In the ideal model it is assumed that for measured quantities such as “concentration” or “weight” the dispersion of the data in a data set is normally distributed. Nevertheless, in real data sets outliers or effects may exist which influence data systematically or unsystematically. The aim of the various statistical procedures is to eliminate the outliers or to detect and exclude the errors of a data set in order to determine the assigned value.
The following approaches are available, all statistical methods have their advantages and disadvantages:
- Sensitive statistics (all values)
- Sensitive statistics (without outliers)
- Robust statistics (Hampel estimator, Q method)
- Robust statistics (median, MAD/nIQR)
- Expert laboratory (expert determination)
We calculate the laboratory performance with the help of the so-called z’score, which you will also find on the respective certificates.
If permissible you will receive an additional calculation of the z-score and the CRD value, which provides useful information for the laboratory routine.