The role of predictive modelling: social and scientific problems of radiation risk assessment

This chapter starts from the premise that current predictive modelling of radiation hazards fails to reflect accurately the status of the science of radiation toxicology. It surveys scientific problems facing the construction of robust models, arguing that uncertainties are such as to confine 'prediction' to order-of-magnitude analysis which can provide little more than a range of plausible risk estimates. It examines the origins of controversy over these issues, briefly analysing the former in terms of their cognitive and institutional dimensions. These problems are discussed in relation to recent concern over raised incidence of child leukaemia around Sellafield nuclear fuel reprocessing plant. If predictive modelling is to be constructive in debates over nuclear technology, the chapter argues decision-makers and risk analysts should have greater regard for the extent of social and scientific dissensus surrounding the issue of radiation hazards. This would best be achieved within the context of a less technocratic approach to risk assessment, and the establishment of formal mechanisms for public and worker involvement in nuclear power regulation. (author)