A semi-quantitative assessment process for improved use of the expected value of information measure in safety management

Abstract

For safety-related decision problems, investment in reliability information could lead to more informed decision-making. The investment in such information could add value by allowing for reduction and prevention of failures and errors leading to adverse events, depending on whether the additional information is worth the acquisition cost. A common measure used to assess this is the expected value of information (EVOI). However, the traditional EVOI approach has limitations in the way it reflects risks and uncertainties. In particular, the traditional approach of expressing the uncertainties exclusively with probabilities could fail to reflect important aspects of risk related to the strength of the background knowledge on which the probabilities are conditioned (e.g. assumptions and models). An extended, semi-quantitative, approach is suggested, which allows for improved decision-making in the sense that aspects of knowledge strength are included in the decision basis. The article illustrates how the extended approach can improve EVOI assessment in safety management by using an example where there exist an option of acquiring human reliability information prior to a decision between alternative safety educational programs.