Browse by author
Lookup NU author(s): Professor Kevin Wilson, Dr Malcolm Farrow
This work is licensed under a Creative Commons Attribution 4.0 International License (CC BY 4.0).
We consider the problem of aggregating the judgements of a groupof experts to form a single prior distribution representing the judgements of thegroup. We develop a Bayesian hierarchical model to reconcile the judgementsof the group of experts based on elicited quantiles for continuous quantities andprobabilities for one-off events. Previous Bayesian reconciliation methods have notbeen used widely, if at all, in contrast to pooling methods and consensus-basedapproaches. To address this we embed Bayesian reconciliation within the probabilisticDelphi method. The result is to furnish the outcome of the probabilisticDelphi method with a direct probabilistic interpretation, with the resulting priorrepresenting the judgements of the decision maker. We can use the rationales fromthe Delphi process to group the experts for the hierarchical modelling. We illustratethe approach with applications to studies evaluating erosion in embankmentdams and pump failures in a water pumping station, and assess the properties ofthe approach using the TU Delft database of expert judgement studies. We seethat, even using an off-the-shelf implementation of the approach, it out-performsindividual experts, equal weighting of experts and the classical method based onthe log score.
Author(s): Wilson KJ, Farrow M, French S, Hartley D
Publication type: Article
Publication status: Published
Journal: Bayesian Analysis
Year: 2024
Pages: Epub ahead of print
Online publication date: 05/12/2024
Acceptance date: 14/11/2024
Date deposited: 15/11/2024
ISSN (print): 1931-6690
ISSN (electronic): 1936-0975
URL: https://doi.org/10.1214/24-BA1497
DOI: 10.1214/24-BA1497
ePrints DOI: 10.57711/3ry7-7662
Altmetrics provided by Altmetric
