On the eve of the US presidential election, experts at Princeton university decided that Donald Trump had a 1 per cent chance of being elected. Before the last general election, Populus, the opinion poll firm, gave David Cameron a 0.5 per cent chance of winning a majority. Much is made of the need to look at ‘the data’ when considering political arguments, but so often it is a wildly inaccurate guess with a decimal point at the end to give an aura of scientific specificity.