It is more useful to constrain uncertainty analysis to the same standards of rigor to which deterministic analysis in science and engineering is held, namely that analyses be phrased only in terms of falsifiable claims. The design engineer is unlikely to know these experimenters well enough to interpret their subjective uncertainty assessments correctly. Moreover, to phrase these uncertainties in subjective terms, such as the personal betting preferences of the contributing experimenters, is an untenable strategy.
What he or she needs is a distribution of rational belief over the continuum of possible lift coefficient values. This engineer is not interested in determining which of two hypothetical values of the lift coefficient is better supported by the wind tunnel data. For example, the aerospace design engineer will need to propagate uncertainty about the aerodynamics of an aircraft together with uncertainties about the structure and control systems to determine how much belief is justified in the proposition that the proposed design will function as required. Two basic requirements pervade uncertainty quantification problems in engineering: first, that uncertainty be represented in a comprehensive fashion second, that uncertainty be expressed in meaningful (i.e. The question addressed in this paper is how that inferential uncertainty can be usefully quantified.
These random errors will cause inferential uncertainty in the experimentally based estimates of aerodynamic parameters such as the lift-curve slope or the stall angle of attack. To assess the aerodynamics of the airframe, this engineer will likely use wind tunnel data, which will be subject to random errors. For example, consider an engineer analyzing a proposed aircraft design. The purpose of the theory of confidence structures is to provide adequate mathematical support for engineers and applied scientists who must quantify and propagate inferential uncertainty.