The Horserace

So everyone's sick of polls by now. Slate has basically given up, after Florida shifted from Bush to Kerry back to Bush in a matter of hours, and Ohio and Wisconsin basically did the same. They correctly advised, "if you don't like the results, wait an hour."

But as this site is intended to be more than a news filter, I want to point out that, depending on the results tuesday night, we may see a major transformation of the way polling is done and reported based on Sam Wang's work at Princeton on allocating undecided voters. Follow the link for indepth information, but the gist is this: decided voter polls, which is what every media organization currently uses, and which obviously produces a media-feedback effect, has the campaign going strongly for Bush. Here are some numbers:

Median outcome, decided voters: Kerry 252 EV, Bush 286 EV

Popular Meta-Margin among decided voters: Bush leads by 0.9%

Predicted median with undecideds: Kerry 280 EV, Bush 258 EV

Electoral prediction with undecideds and turnout: Kerry 323 EV, Bush 215 EV

Popular vote prediction with undecideds and turnout: Kerry 50%, Bush 48%

So as you can see, and Wang has basically clung to this all along, that because of a statistical observation termed "Cook's Law," the so-called "undecided voters" are not statistically split down the middle, but historically favor the challenger by almost a 3:1 margin. In an election this close, such a seemingly insignificant detail changes the entire landscape. With decided voters only, Bush always wins in a landslide. With undecideds and turnout predictions included, Kerry wins both the popular vote, and in something of an electoral college landslide. Wang personally puts Kerry's odds of winning at 6:1 in favor, and is putting his own money on it.

Because perhaps the most important factor in contemporary politics is the phenomenon of "momentum," it stands to reason that the race would be very, very different if news organizations across the world - with the exception of Fox, of course - were reporting a Kerry landslide on the basis of these numbers. How or whether it would change actual voting depends on your understanding of mass psychology. If voters, on the whole, are more drawn to the fight because it's so close, or because they believe Kerry is losing, then the change might work against Kerry. If, however, you believe as I do, that we have a lot waffling voters who don't like Bush, but are afraid to vote for Kerry - and who are getting psychological reinforcement for this belief on the basis of media predictions that everyone else feels the same way - then bringing in the undecided calculation could work massively in Kerry's favor. Media commentators would all begin to chime in that "the country is moving in a particular direction" - in this case, towards the idea not simply that Bush is bad, which it has already decided, but that Kerry is an acceptable replacement, which it has not.

So after all the results come in on Tuesday night, whether you're elated or angry as hell, or - most likely - that you're gripped to the news listening to all the insanity of multiple law suits and countersuits, check back here and see whether Sam Wang was right or wrong, because it could lead to significant changes in 2008 and beyond. -andrew