The En value can be different because the values used by you are not the values used by NAPT. Numbers found on the preliminary or final report are not the exact values NAPT uses. These numbers are rounded. The only way that you are going to guarantee that your En calculations matches NAPT calculations is to use the exact same numbers. Feel free to contact NAPT for the exact numbers used in any calculation.
The reference value is a value against which your results are compared. In any proficiency test, the results are only as good as the established values assigned.
NAPT's process for establishing the reference value was designed with the help of lead statisticians in the metrology community plus technical oversight by statisticians at NIST. Using standard technical practices, an analysis is performed and an appropriate reference value is assigned to the data set. We ensure that reference values and their associated uncertainties assigned are stable and worthy of comparison.
Pivot and reference labs are accredited, helping to ensure credibility of reference values. To further analyze the results we also perform the following statistical reviews: Two Sigma, Three Sigma, Chauvnet Criterion, Sample Median, Trimmed Mean, Interquartile, Q-Test, and Thompson Technique. This is done in all cases to assure sound and meaningful results are published. During the course of each proficiency test, comprehensive technical reviews are conducted before, during, and after a kit is put into distribution. This is done to assure test integrity, establish and/or validate reference data and check for trends and/or anomalies in the data. Our technical advisors believe our technical review is unmatched by any ILC Provider.
These requirements are described in NAPT Quality Procedure 407-1 Data and Statistical Analysis Procedure. Only after a careful review of the data does NAPT assign an established reference value. To prematurely assign a reference value could be inaccurate and may result in a value that would not pass a robust analysis. Doing so would not ensure confidence in the reference value assigned. Making an assumption that a single measurement is the correct measurement, is not a technically sound process for ensuring the validity of the data. (No single laboratory is infallible.) That is why only after a thorough review of all data, should a reference value be assigned.
Copyright 2020 National Association for Proficiency Testing (NAPT)
Website developed by 10 Pound Gorilla