This idea that existing data on nuclear disasters amounts to a sufficiently large and representative sample to extrapolate a hazard trend from - shouldn't reall

This idea that existing data on nuclear disasters amounts to a sufficiently large and representative sample to extrapolate a hazard trend from - shouldn't really float in statistically literate conversation. Estimation of the likelihood and error bars on different classes of potential nuclear disasters, involves confounding complexity and argument. The worst conceivable nuclear disaster types have thankfully not been realized to date. The problem with the "death rate" measure is it simply assumes they never will be realized. A most flattering assumption that is widely taken as granted along side genuine hope and enthusiasm for a nuclear solution.