Abstract
In decision and risk analysis together with operational research methods, probabilistic modelling of uncertainties provides essential information for decision-makers. As uncertainties are typically not isolated and simplifying assumptions (such as independence) are often not justifiable, methods that model their dependence are being developed. A common challenge is that relevant historical data for specifying and quantifying a model are lacking. In this case, the dependence information should be elicited from experts. Guidance for eliciting dependence is sparse whereas particularly little research addresses the structuring of experts’ knowledge about dependence relationships prior to a quantitative elicitation. However, such preparation is crucial for developing confidence in the resulting judgements, mitigating biases and ensuring transparency, especially when assessing tail dependence. Therefore, we introduce a qualitative risk analysis method based on our definition of conditional scenarios that structures experts’ knowledge about (tail) dependence prior to its assessment. In an illustrative example, we show how to elicit conditional scenarios that support the assessment of a quantitative model for the complex risks of the UK higher education sector.
Disclosure statement
No potential conflict of interest was reported by the authors.
Notes
1 In the case-study of this paper we elicit conditional exceedance probabilities. Briefly, a conditional exceedance probability is the probability of a variable exceeding (or equalling) a certain percentile given that the variable we condition on also exceeds a certain percentile. The definition is the same for both variables being below a certain percentile threshold.
2 The experts were given as well the corresponding monetary values for the specific percentiles by the facilitator.