Impact of the sampling process on the measurement uncertainty, a case study: physicochemical parameters in diesel

Abstract

The result of a chemical measurement is consequence of a set of steps that begins with sample selection. The aim of this study was to evaluate the relative importance of the sampling in overall measurement uncertainty, taking as study case physicochemical measurements of diesel. The empirical method used was by duplicates, and the treatment of data sets was performed by four different statistical techniques: classic analysis of variance, robust analysis of variance and two different Range Statistics models. The choice must be backed by previous evaluation of data set in terms of normality, independence, presence of outliers and homogeneity of variance. It was found that Range Statistics produced most reliable results under different premises and for the variety of physicochemical parameters evaluated. Robust analysis of variance failed indicates proper influence of sampling even it is a test designed to work under the lack of some of the proper premises.