I have been a leading proponent in the Bayesian statistics community of the use of elicitation to formulate expert knowledge about uncertain quantities.
I can offer various options for training in elicitation. I have established training courses in the theory and practice of elicitation, and can also arrange mock elicitation sessions to further develop facilitation skills. Following training given to key staff, the SHELF protocol is routinely used in several companies to quantify uncertainty about critical parameters, including Novartis and GlaxoSmithKline. I also developed an online e-learning course for people who will participate as experts in an elicitation exercise, teaching them how to make probabilistic judgements.
I have conducted expert elicitations for Her Majesty's Inspectorate of Pollution, Risk Management Solutions Ltd, AstraZeneca, Pfizer, the European Food Safety Authority, GlaxoSmithKline and the Centre for Workforce Intelligence. Methods that I developed have been coded into software which has been widely used in the water and rail industries, and I have advised a number of users of those methods.
My one-day general course on elicitation has been given to Evira (Finland), Pfizer (UK), Procter and Gamble (USA), the University of Texas at Austin (USA), the University of St Andrews (UK), the Netherlands Forensic Institute (Netherlands), Humboldt University (Germany), Chalmers University (Sweden), GrĂ¼nenthal (Germany), L'Oreal (France) and also as an open-entry course at Warwick University.
My three-day intensive course has been given several times as an open-entry course to a total of more than thirty individuals, and also as an in-house course to GlaxoSmithKline (UK and USA), the University of St Andrews (UK), Chalmers University (Sweden), Novartis (Switzerland and USA).
Risk assessment and management is an important field that has special challenges. Experts are often called upon to assess risks smaller than can be informed by real data. Risk typically depends on the occurrence of extreme events, and particularly on the concurrence of several extreme values. From the perspective of elicitation, this poses particular difficulties in assessing the tails of probability distributions and correlations between variables. I can offer some novel methods to address these challenges.
I am happy to consult on these or other problems in elicitation.
![]() |
![]() |
![]() |
![]() |
![]() |
![]() |