Statistics Meeting Book (May 15, 2019)

pC metamer • Useful for micro to trace analyte (C < 0.10 g/g) across a wide range of concentrations. • Results for different concentrations generally comparable. • Data may be subject to platykurtosis (failure of normal distribution assumption). Skewness in C metamer (a more serious problem) will be corrected by pC transform. • Example: pH transform of hydrogen ion concentration.

7

copyright 2018 by Least Cost Formulations, Ltd., all rights reserved.

Reported statistics for pC metamer

• pAve: estimate of mean of pC data • pSD(r): estimate of pC data repeatability standard deviation • pSD(L): estimate of pC data among-laboratory standard deviation • pSD 5 HVWLPDWH RI S& GDWD UHSURGXFLELOLW\ ¥> pSD(L) 2 + pSD(r) 2 ] • pBias = pAve – log 10 (TrueValue)

8

copyright 2018 by Least Cost Formulations, Ltd., all rights reserved.