Statistics Meeting Book (May 15, 2019)

Metamers • ‘C’ metamer: Traditional AOAC approach where data in measured concentrations ‘C’ are statistically analyzed. • ‘pC’ metamer: Data is first transformed to pC = - log 10 (C) before statistic analysis. • pC metamer is preferred, particularly here where Matrix and Concentration are confounded. pC statistics may be compared and composited across concentrations. copyright 2018 by Least Cost Formulations, Ltd., all rights reserved. 5 General assumptions (both metamers) • ‘Independence’: Sources of variation do not interact. The underlying model is additive and main effects only. • ‘Homogeneity’: Sources of variation are the same for each Institution, characterized by a common variance. • ‘Normality’: Sources of variation are governed by a univariate normal distribution, so only means and variances matter.

6

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