The West Midlands group held a meeting on 9 May 2013 hosted by the University of Warwick. The speaker was Simon French, director of the Risk Initiative and Statistical Consultancy Unit (RISCU) at the Department of Statistics, University of Warwick.
Simon first focused on his experience in the former USSR with nuclear emergency management after the Chernobyl accident. Through an historical overview of the Russian society at that time, he motivated the forces that were driving decision making during and after the accident. He described the key issues with the methodology used, the attributes that were considered relevant, the alternative strategies the decision making body had available and the analyses that were performed during decision conferences in the former USSR.
Simon then focused on the uncertainties associated with the decision analysis developed after the Chernobyl accident. As a key example, he presented the Linear Hypothesis, which assumes a linear relationship between health risks and radiation dose. It is not commonly understood that for very low levels of radiation this is truly a hypothesis, and that its use to predict long term health impacts consequently introduces considerable uncertainty. He further presented other examples of uncertainties in a nuclear emergency domain, as the ones that derive from forecasts, data collection and disagreement between different models.
The last part of the talk described the uncertainties associated with decision analyses in any domain. Simon argued that the uncertainties can be categorised into two broad classes: the ones associated with values, which can be modelled using value-focused thinking, and the ones related to science, that can be addressed through probability. For the latter case, the uncertainties may however be too deep, as for example when experts disagree, and no probabilistic technique can be used. In such situations, Simon suggested the use of Scenario Planning, a qualitative methodology that can then be mixed with a quantitative analysis.
Simon concluded his talk by stating that uncertainty permeates emergency management and that in these contexts decision analysis practitioners need to properly address it.
Report by Manuele Leonelli, University of Warwick