In Nate I trust.
This last week before a close presidential election is always a maelstrom.
We are bombarded with polls and with sound clips from feverish candidates shouting hoarsely at huge crowds.
If you care about the election result in any way, it’s a lot to digest.
Which is why I trust in Nate.
Nate Silver, that is, the nerdy, brilliant proprietor of the FiveThirtyEight blog on the New York Times web site.
Nate Silver started analyzing election polls as a solo act in 2007. In 2008, he accurately predicted the presidential result in 49 of 50 states, and called all that year’s Senate races correctly. He was way more accurate than any name pollster.
This is the kind of performance that gets you your own blog on the New York Times.
But Nate had been a minor god to me long before. He is revered among baseball stat-geeks for what he did in his first career, which was to create PECOTA, the most accurate statistic ever for predicting future performance of major leaguers.
What Silver does with elections is this: He does a poll of polls, weighting each poll based on its past records for accuracy and tilt to either party.
Here’s the thing about polls, which we all know, but sometimes forget in the heat of an election: They are all guesses. And those guesses have become more error-prone recently.
One reason is this: about a third of Americans now have only cell phones, not landlines. Polls tend to miss those people. And cell phone owners, since likely to be younger, are more likely to be Obama voters. Silver estimates that produces survey results that understate the Obama vote by 2 percent; his analysis factors in how different polls correct for the cell phone factor.
Similarly, different pollsters adjust differently for the black turnout mystery. Will Obama benefit the same historic turnout of black voters he did in 2008, or will the turnout be closer to the norm? How you answer that makes for big variations in poll results.
This is the thing about Nate Silver: He cares way more about the integrity of the analysis than he does who wins. So much election talk you now hear is tinged by the emotions of the talker. Silver is loyal to the facts, because they sing to him.
So how does this prodigy see the race? Well, at this moment, FiveThirtyEight (and, in case you didn’t figure out, 538 is how many electoral votes there are) forecasts a 74.6 percent chance of an Obama win, with a likeliest electoral college total of 296 votes. Romney’s numbers looked quite a bit brighter early last week.
Whether you’re an anxious Obama backer or a hopeful Romney one, Nate Silver’s blog is an antidote to the nonstop, fact-free palaver of the cable talkers.
UPDATE: On the day I posted this commentary, 538.com came out with a really interesting analysis of Pennsylvania’s history as a swing state. Not by Nate Silver, but one of his contributors.