I read with interest this morning about the kerfuffle New York Times blogger Nate Silver is causing. A Financial Times article (subscription required) explained that Silver has been using statistical analysis to consistently predict an Obama win in the coming US elections. His FiveThirtyEight blog is now being roundly criticised by the right-wing - and the criticism is getting pretty personal.
The unfortunate reality is that this "effeminate man with a soft-sounding voice” (Dean Chambers) has a pretty amazing track record of success. He started plying his trade forecasting the performance of Major League Baseball players and since 2008 has turned his attention to politics. In 2008 he correctly predicted the presidential winner of 49 out of 50 states. As a side note, the same blog (but different blogger) also correctly predicted the narrow win for the ruling Australian Labor party in 2010.
As far as I can see, Silver is only guilty of using open data to make systematic and scientifically based predictions using probability theory. You can check-out his methods for yourself.
I recommend his blog to anyone interested in stats and polling. TL;DR? If that's too old-school for you gen-Y twits (?), then he tweets as well.
His book 'The Signal and the Noise: Why So Many Predictions Fail-but Some Don't' has just been published and I'm ordering it today.
It's good to see that mathematically-based poll analysis continues to be politicised in the States. Way to go America - who needs facts?