Response to comment on ‘Empirical versus modelling approaches to the estimation of measurement uncertainty caused by primary sampling’

Abstract

A binomial model is used in the analysis of our original data by Geelhoed (Analyst, 2009, DOI: 10.1039/b812422a). However, in assessing the standard uncertainty of sampling pistachio nuts for aflatoxins, the real sampling target will be much more realistically described by a continuous distribution of analyte concentration amongst the batch of nuts rather than being made up of just ‘contaminated’ and ‘uncontaminated’ nuts. The binomial assumption is therefore unlikely to be realistic enough to describe the analyte concentration distribution within a real sample population. Doing so can lead to a substantial overestimate of both the uncertainty and its confidence interval and is therefore ineffective for explaining the discrepancy between the empirical and modelling estimates in our case.