I am currently preparing a presentation about the value of more complex (specically: non-Gaussian) statistical inference. I thought it might be interesting to start the presentation with a small real-world example about disasters or mistakes caused by over-ambitious simplification.
Unfortunately, it quickly turned out that there do not seem to be that many stories about statistics disasters, which came as quite a suprise to me. One story I found relates the 2008 financial crash to the oversight of fat-tailed interactions caused by using Gaussian copula models. That's already not bad, but there must be better anecdotes. Do you know of any other entertaining/interesting stories in which statistical simplifications or errors led to unexpected results?