Aggregating expert judgement

Abstract

In a paper written some 25 years ago, I distinguished three contexts in which one might wish to combine expert judgements of uncertainty: the expert problem, the group decision problem and the textbook problem. Over the intervening years much has been written on the first two, which have the focus of a single decision context, but little on the third, though the closely related field of meta-analysis has developed considerably. With many developments in internet technology, particularly in relation to interactivity and communication, the textbook problem is gaining in importance since data and expert judgements can be made available over the web to be used by many different individuals to shape their own beliefs in many different contexts. Moreover, applications such as web-based decision support, e-participation and e-democracy are making algorithmic 'solutions' to the group decision problem attractive, despite many results showing we know that such solutions are, at best, rare and, at worst, illusory. In this paper I survey developments since my earlier paper and note some unresolved issues. Then I turn to how expert judgement might be used within web-based group decision support, as well as in e-participation and e-democracy contexts. The latter points to a growing importance of the textbook problem and suggests that Cooke's principles for scientific reporting of expert judgement studies may need enhancing for such studies to be used by a wider audience.

A practical guide to structured expert elicitation using the IDEA protocol

“…As with other structured protocols (Cooke, ; Keeney & von Winterfeldt, ; Mellers et al., ; O'Hagan et al., ), IDEA promotes the use of multiple experts based on empirical evidence that indicates that while criteria such as age, experience, publications, memberships and peer‐recommendation can be useful for sourcing potential experts, they are actually very poor guides to determining a priori someone's ability to provide good judgements in elicitation settings (Burgman et al., ; Shanteau et al., ; Tetlock & Gardner, ) and may result in the unnecessary exclusion of knowledgeable individuals (French, ; Shanteau et al., ). The best guide to expert performance is a person's previous performance on closely related tasks, which is rarely available a priori.…”

“…Rather, it should meet the requirements of more rigorous definitions of structured elicitation protocols. That is, it should treat the elicitation of expert judgements in the same regard as empirical data, by using repeatable, transparent methods and addressing scientific questions (not value judgements) in the form of probabilities and quantities (Aspinall, ; Aspinall & Cooke, ; French, ; Morgan, ). Importantly, it should account for each step of the elicitation including the recruitment of experts, the framing of questions, the elicitation and aggregation of their judgements, using procedures that have been tested and clearly demonstrated to improve judgements (e.g.…”

The Development of Progress Plans Using a Performance‐Based Expert Judgment Model to Assess Technical Performance and Risk

“…The result is a structured and theoretically justified subjective approach to performance‐based weighting expert judgment that can be used to predict a future quantitative parameter. Considered relatively easy to understand and implement, Cooke's classical model for aggregating expert judgment has been widely used in the field of risk analysis for over 20 years [Clemen, ; Cooke, ; Cooke and Goossens, ; French, ]. Unfortunately, because of the unique characteristics of developing technologies in acquisition programs, the creation of relevant seed variables for expert judgment weighting is difficult to achieve.…”

A Method for Reducing Noise and Complexity in Yield Analysis for Manufacturing Process Workflows

“…In such a case we aggregate the latent sources using an opinion pool. Several pooling methods are presented in [52], but extending our IID assumption, our aggregate probability is a linear sum of all incident latent sources. These assumptions allow us to remove source vertices and outgoing edges corresponding to latent effects from G 3 but incorporate their influence on their direct successors using auxiliary variables p e i (k) for all vertices e i that correspond to latent effects.…”