Youthful beauty queens do not prevent murder by scalding

2008 was a turbulent year in many ways. Our banking system almost crashed, and had to be rescued at taxpayers’ expense. It was also the year which marked the end of the previously impressive correlation between the age of Miss America and the rate of murder in the USA by steam and hot objects. Up until then, for every year with a youthful Miss America, the murder rate dropped, while when a slightly older woman was victorious, the murder rate increased. The correlation is quite impressive. I have seen a lot worse in plots in scientific papers.

But of course, correlation does not imply causation. There seems no obvious mechanism whereby the election of a youthful Miss America can discourage people from killing with steam, and it also seems unlikely that people would react to a Miss America who as old as 24 years by offing someone with a hot poker.

The plot above is by Tyler Viglen, who has basically downloaded a heap of American data and then plotted the correlations. The basic idea is the same as that used to apparently (not really) show that a dead salmon can recognise emotion. This is that with a big enough pile of data it is inevitable that some things will be correlated with random other things, for just the same reason that in a group of 50 friends it is likely that at least one or two will have a birthday in common.

His website has many examples, some of which are just great. It is worth a look, both because it is fun, and because the simplest way to get that correlation and causation are not the same is by examples where that is pretty obvious.