From Measurement to Decision with the Analytic Hierarchy Process: Propagation of Uncertainty to Decision Outcome

Measurement is customarily performed to acquire information useful to support decisions. More and more frequently, the complexity of analyzed systems and problems requires to measure various heterogeneous properties and to process the returned results in order to obtain information directly exploitable to support decision-making activities. However, despite a high relevance for many engineering applications, methods for rigorously processing data about heterogeneous empirical properties have not been properly analyzed yet. As a result, decision makers attain their conclusions by subjectively fusing together the available measurement results without following a formal procedure, which has a negative impact on the verifiability and the reliability of the whole evaluation process. This paper proposes a simple and effective multicriteria decision-making method that can be applied whenever measurement results are available for the considered decision criteria. Moreover, unlike most of the scientific literature on decision making, the proposed method provides not only the most desirable decision outcome, but it also returns information on the outcome reliability. This is achieved by leveraging on measurement fundamentals. In particular, by expressing measurement uncertainty using interval values, the related interval priorities of the considered decision alternatives can be determined. The proposed methodology is finally applied to a dependability-benchmarking case study.