One of the more interesting side-stories to yesterday's U.S. presidential election (congratulations to Barack Obama, by the way) was watching Nate Silver's projections narrow in and ultimately predict perfectly the eventual outcome. Silver does not do any polling himself, but he analyzes the polls, corrects for built-in bias, and uses the data to create an overall projection. Though some wags are saying (with good reason) that "the numbers have a liberal bias" what we actually have here is a triumph of the meta-analysis over mere numbers - it's the relation and interplay between the different polls that creates the projection, not simply the raw data produced by counting people. And indeed, it shows how a naive reliance on the numbers produces a misleading reasult (such as the numerous pundits saying the race was "too close to call" while all along Silver predicted a fairly comfortable Electoral College victory for Obama).
Today: 4 Total: 105 [Direct link] [Share]
