This is a neat article about how to use (and not abuse) statistics that happens to overlap with most of my beliefs. Of course, this is more than just about statistics: Socrates, paraphrased, said that to understand anything is to know how much you don’t know, or, in other words, proper use of statistics is at the foundation of all understanding–as opposed to mere “knowledge.”
How much traction this set of beliefs has with real life is anybody’s guess, including even FiverThirtyEight, though. In the end, we want to sell people how much we know, for selling how much we “understand” is far more complex process. So we show that we are smart not by understanding how weird things (like Sanders and Trump) can crop up in elections, but by cutting through all the political trash talk and summarizing numbers to predict election results. But, for what it is worth, last two elections that made FiveThirtyEight’s reputation were also among the most predictable.2016 should be a demonstration of how important understanding the inherent instability in politics is, how we cannot know, regardless of how fancy our algorithms are and how much data we have–if the cat in the box is alive or dead. Instead, of course, people will talk about how even stats gurus failed. Of course, I’d talked about in another post, how use of massive data without an understanding is potential recipe for willful and knowing ignorance–if you make following the “price fluctuations” so cheap and easy, without understanding the underlying causes, it destroys the incentive for learning about the underlying causes and, in the long run, destroys the usefulness of the price system itself–even if it will be a better source of information, relatively speaking, than anything else still. This is something to ponder about.