The evaluation of uncertainties in the analysis of calibration data

The National Physical Laboratory is involved, both internally and in collaboration with other National Measurement Institutes, in activities to promote sound methods of uncertainty evaluation. Some of the results of this work are described in this paper. A classification of measurement system models is presented that is more general than that considered in the ISO GUM. For each model type it is indicated how the uncertainty in the measurement result can be evaluated. The model types are illustrated using examples of real measurement systems. Sampling techniques are discussed and it is indicated how they can be used to estimate the statistical distributions of the model outputs, i.e., the results of measurement, without making some of the assumptions required by the mainstream GUM approach. The confidence intervals obtained by application of the GUM are compared with those obtained from sampling techniques for some of these examples.