Measurements are often an objective support for decision-making. If a measurement result can affect the decision, the measurement uncertainty associated with the result can be useful to manage the risk of a wrong decision. This is applicable to measurements in all fields, scientific, industrial or human-related. In the last case particular care has to be taken when a wrong decision can affect health, life or environment. Now, measurement-based decisions may be, with fair generality, reduced to a conformity assessment problem. Typically the state of the process under investigation is monitored by a set of measured parameters and their belonging to a given “safe” subset of the parameter space has to be checked. Decisions are thus strongly affected by measurement uncertainty. Common decision rules are based on expressing measurement uncertainty as intervals of values (expanded uncertainty), which results in an on-off criterion, with an uncertainty region. Due to the importance of the problem, a more sophisticated approach, based on probability, merits investigation, aiming at improving the quantification of the risk associated with each decision. In the article, such an approach is addressed in general terms. The role of uncertainty evaluation is then discussed and an algorithm for evaluating uncertainty in terms of probability density is presented and the advantages of this approach are discussed.
Keywords:
measurement vs. decision support, uncertainty of measurement, uncertainty evaluation by probability densities, risk management
