In an interesting and enlightening new book on the interaction between statistics, predictions, data and human behaviour, Nate Silver (of the New York Times) asks the question “Why is baseball easier to predict than presidential elections?”. It’s just one case study among many that he looks at to try and understand whether humans have got better at predicting outcomes as data and stats have got better, and whether human behaviour trumps all at the end of the day or if the biggest skewing factor in predictions made by humans is humans themselves.
Meanwhile, some ‘events’ are easier to predict, or at least work on statistically, by virtue of the type of data they throw up. For instance, Silver’s comparison between baseball games and Presidential elections. While US Presidential Elections come around only every four years, there are around 162 baseball games in just one year alone. So the frequency of data to work with is much better, and finding patterns may be easier.
Try get a copy of this statistician and New York Times contributor’s book ‘The Signal and the Noise: Why So Many Predictions Fail – But Some Don’t’. For a quick intro, watch this – http://www.bbc.co.uk/news/magazine-19982180
(featured image courtesy The Guardian UK Data Blog)
Here’s an interesting piece by David Brooks of NYT on the addiction to Polling data:
Poll Addict Confesses – http://www.nytimes.com/2012/10/23/opinion/books-poll-addict-confesses.html?smid=tw-share