I want to admit to my addiction. I have become a rabid fan of Nate Silver's FiveThirtyEight (http://http://fivethirtyeight.blogs.nytimes.com/) blog on the presidential and senatorial polls that culminate in the November 6 election. I want to go on record as endorsing Silver's model before we know whether it will actually work as well this time as it did in 2008, because I don't want to appear that I went on the record only a posteriori. Silver's approach to the statistical problem of extracting a small signal (if there is any signal at all) from all the noise that approximately six billion dollars invested in obfuscation, character assassination, and distortion can buy. I think that, if polls can tell us how elections will go, Silver's quite scientific approach will do as well as can be done. in "The Signal and the Noise", Silver describes how statistical methods can be applied to the prediction of economic collapses, earthquakes, weather, climate change, terrorism, poker, and sports, among others. Silver has personal experience with several of these. Besides his political polling blog, he has earned a living in poker and developed and sold a system for the evaluation of baseball players (it is called PECOTA -Player Empirical Comparison and Optimization Test Algorithm - and is still for sale, although no longer owned by Silver). The best parts of the book are those about polling, poker, and baseball, but I enjoyed all of it because he applies a consistent scientific approach to all of them. This book is very relevant to the teaching of science - more so than most books on the teaching of science. I think that, if you are reading this, you will enjoy "The Signal and The Noise".
The Signal and the Noise: Why So Many Predictions Fail - But Some Don't