TABLE OF CONTENTS The repeatability of EMC testing depends on many factors that affect the measurement result. Some factors are not well understood or not documented in terms of the nature and magnitude of their contributions to the total uncertainty. The treatment of measurement uncertainty involves the use of statistics to estimate the probable uncertainty and associated confidence level with regard to a particular measurement or set of measurements. Statistics and probability theory is a considerable subject in its own right and far too large to deal with adequately in a book such as this, which is primarily concerned with the varied aspects of EMC testing. Useful summaries of the main theories and analytical techniques are given in a number of texts [1 5] and are recommended to those particularly interested in this aspect of mathematics, or those who are determined to achieve a sound grasp of the basics before considering how to apply them to the understanding of EMC measurement uncertainty.
The application of statistics to experimental measurements has been dealt with specifically by Box et al. [6] and Topping [7], and a good understanding of the subject can be achieved via these texts. Early work carried out in the British Calibration Service by Dietrich [8] resulted in the development of a statistical approach to combining measurement uncertainties based on the assumption of probability distributions for all the contributing components. In 1977 a useful code of practice was produced by Harris and Hinton [9] for the treatment of uncertainty in...
