Research and analysis

The probabilistic elicitation of subjective data

Probabilistic elicitation provides a means to elicit the uncertainty surrounding subject matter experts’ best guesses for numerical values. This pioneering research is highly relevant to government work.



This paper feeds directly into the government’s Analytical Quality Assurance book.

Much of Dstl’s advice is driven by research that utilises subjective quantities, as determined by experts. At present, these subjective quantities are often reported as single-point figures; it is not known how confident the experts are in their estimates. Consequently, important decisions might be based on estimates for which the experts are unsure.

This report presents an established methodology and highlights best practice for representing the uncertainty surrounding experts’ subjective judgements. It gives decision-makers a means of assessing the levels of risk that they are taking by using expert judgement to inform the decision-making process.