Statistically Speaking

Nate Silver, a brilliant statistician whose must-read FiveThirtyEight blog appears in the New York Times, says President Obama is an 80% favorite to win the electoral college on Tuesday. He based it on a proprietary formula which combines state polls, national polls, and other data to winnow it down to a single prediction.  Silver's numbers have pointed towards Obama's re-election throughout the year.

Naturally, right-wingers (who hate math almost as much as science) dismiss Silver's stats, despite the fact that, he correctly called 49 out of 50 states in the 2008 presidential race, as well as the winners of all 35 Senate races. Those attacking Silver act as if he's guaranteeing an Obama win, which he isn't.

In a profile on "Sunday Morning" today, after Silver repeated his 80% figure, CBS correspondent Martha Teichner asked him, "Are you absolutely convinced that Obama will win?"  Silver replied, "Oh, no. I'm convinced that if offered even money I would be happy to bet on Obama. I would need a pretty good price to bet on Romney, two-to-one wouldn't do for it for me. Three-to-one might." Teichner then went on to compare Silver's number with those of a 7-11 coffee cup poll and which candidate's masks sold more for Halloween, anecdotal nonsense which is neither scientific nor relevant.

What is relevant is understanding what Silver's number means.  As a poker player (like Silver), I know that an 80% chance of victory is about where you are when you get all the money in pre-flop in hold'em with a pair of aces versus a pair of kings.  And I've seen the opponent spike a king often enough to know you shouldn't stack your winnings until they're pushed towards you, but that shouldn't stop you from being all-in and letting the cards & chips fall where they may.

Labels: