Dielectric measurement: error analysis and assessment of uncertainty

The advantages and limitations of using partial differential analysis to assess the methodological uncertainty associated with the measurement of the dielectric properties of a material are discussed and an alternative pragmatic approach is proposed. It relies on repeat measurements of the dielectric properties of reference liquids and an analysis to estimate random and systematic uncertainties. Examples of measurement uncertainty are provided for well-defined monomolecular materials and for less homogeneous materials at microwave frequencies. All examples relate to measurement with an open-ended coaxial probe but the methodology is not specific to this technique. Examination of the components of uncertainty in the dielectric properties of biological tissue shows that, where the system is free of methodological bias, random fluctuations originating from sampling and natural inhomogeneity dominate the uncertainty budget. In such cases, the mean value of the measured parameter and the standard error of the mean can be taken as a good measure of the true value and its associated uncertainty.