Experimental designs for evaluation of uncertainty — Use of factorial designs for determining uncertainty functions

This document specifies experimental procedures and statistical analysis for the determination of measurement uncertainty in situations where the following conditions are fulfilled:

Condition 1: The level of the measurand is non-negative, e.g. concentration level of a contaminant in a sample.

Condition 2: Measurement error consists of two independent components: for one of these components the relative standard deviation is constant (that is, the absolute deviation is proportional to the level of the measurand), whereas for the other component the absolute standard deviation is constant (that is, independent of the level of the measurand).

Condition 3: Samples for different levels of the measurand can be made available; if the level of the measurand is the concentration of a chemical substance, samples could be obtained e.g. by fortifying (spiking) blank samples.

Conditions 1 and 2 are met for most applications of instrumental chemical analyses. Condition 3 can be met for chemical analyses if blank samples are available.

This document can also be used to determine precision data for a particular laboratory for different technicians, different environmental conditions, the same