Seminar Series 2010
Expert Opinion Elicitation: Art or Science?
Thursday, 25 November 2010
Conference Room, Wagga Wagga Agricultural Institute
Elicitation of formal expert opinion, through expert judgement exercises, has been used to populate simulation, mathematical and conceptual models in animal and human health fields. Numerous techniques have been developed to enable openness, transparency and documentation within expert opinion, and to address and document biases as they arise.
The aim of this study was to obtain data through implementation of an expert opinion questionnaire, enabling the initial, and fully updateable, parameterisation of a quantitative risk assessment model in the area of acquisition of Methicillin-resistant Staphylococcus aureus (MRSA) in pet dogs.
An emailed questionnaire was developed and administered to 34 targeted multidisciplinary experts. The outcome of the study was a set of composite expert estimates of prevalence, environmental contamination and transmission variables.
Calibration variables, expert weightings and other recognised techniques were utilised to increase the accuracy and reduce the bias associated with the responses. However, while many questions returned narrow and informative combined distributions, markedly divergent opinions were obtained for some estimates, resulting in wide and minimally-informative combined expert distributions. In particular, the responses for questions estimating environmental contamination and the routes and probabilities of transmission of MRSA between humans, the environment and dogs, obtained responses with marked variation in confidence, uncertainty and absolute values. These findings were also reflected in the difficulty reported by the respondents in completing the questionnaire with respect to estimation of transmission variables.
The results of this exercise highlight the benefits and pitfalls of expert opinion elicitation, along with the difficulty in defining probabilities of transmission of pathogens between humans and animals, which are unlikely to be resolved through either observational or experimental studies.
For further information about this seminar please contact:
A/V ref: MG-C