|
This was a "Change" election, they say. We'll see how "change" affects U.S. policy. But one big change is the relationship of statistics to elections.
Nate Silver runs the blog FiveThirtyEight.com. He is a math/statistics whiz. He developed his own statistical analysis for baseball to run a fantasy baseball team at the age of 13. Now he applies baseball statistical analysis to politics. And predicted virtually every race accurately. Voter turnout, election results for president, which senators would win. I don't know if anything went wrong for him. The chaos of election campaigns was transformed into the orderly elegance of sophisticated statistical techniques, made available to the world for free.
The media quotes poll results all the time. Fox quotes Republican leaning polls. MSNBC quotes Democratic leaning polls. Many polls are just statistical garbage. But Silver punched through the delusions and biases of media commentators and virtually anybody who followed the campaigns.
I quoted Rasmussen polling firms a lot during the summer, because they had a good record and are well regarded. Rasmussen also tended to lean Republican as compared to other polls in the past, but the Republican lean was generally accurate in 2004.
Pollster.com and Real Clear Politics have composites of polls. They combine all poll results and post them as a composite, which is much more accurate than any other poll. This is a major advancement from looking at one poll, or a series of tracking polls. It is Polling 2.0.
Nate Silver developed Polling 3.0. He digs deeply for statistical trends based upon massive data, including polls, but also past election results, demographics, cell phone use, etc. He weighted good polls heavier and garbage polls lighter. And it worked very well. On November 3, 2008, he had given Obama a 98.1% chance of winning based upon his machine running 10,000 regression analyses each day. He is also an unapologetic Obama supporter. But that should not reduce the significance of his work any more than Rasmussen's Republican lean did.
Basically the rub of it is... you cannot maintain your delusions or misjudgments in the face of Nate Silver's numbers. Who would have thought the emotionalism of political campaigning could succomb?
Now FiveThirtyEight.com will apply its statistical techniques to predict whether certain bills will pass. I have no idea how Silver will do that. But I am interested to see how math will bring order to that particular sausage grinding machine.
Trackback(0)
 |
TrackBack URI for this entry
Subscribe to this comment's feed