in Uncategorized

The State of the Race: Debunking Nate Silver

Look, it’s not complicated. America’s had a lousy four years, but they still weren’t willing to abandon Obama until they were comfortable with the other guy. The debate made them realize Romney’s a more-than-acceptable alternative, so now the president finds himself swimming upstream. Simple as that.

Romney’s new poll numbers aren’t a bounce; they’re a seismic shift.

In record numbers, voters discovered for themselves that the real MItt bears no resemblance to the Monopoly Man caricature that the Obama-ites have spent hundreds of millions of wasted dollars creating. So regardless of what happens in the debates going forward, or how hard Obama tries to get you to believe him instead of your lying eyes, there’s no going back to Grey Poupon Romney. Although it’s almost comical to watch Obama try. Does anyone believe, for instance, that in an era of massive economic uncertainty, unemployment, and unrest abroad, that undecided voters are going to cast their votes based on the fate of Big Bird?

The job numbers were supposed to turn it all around for Obama, and, indeed, the media has been doing yeoman’s work to try and make that happen. But economic data doesn’t move the electorate as much as it impresses the talking heads. I learned that on the Friday after the Democratic Convention, when the catastrophic September jobs report came on the heels of Obama’s acceptance speech.  I was sure it would squelch any Obama bounce. It didn’t. Neither, despite much wishful thinking on the part of MSNBC et al, has this statistically anomalous 7.8% unemployment figure done anything to replace the fallen bottom of the Obama campaign boat.

7.8% eliminates the “40 months above 8%” talking point, but it’s still a terribly depressing figure, and everything else in that report was awful, too. Job creation has slowed for three months straight as we continue to march like lemmings off of January’s fiscal cliff. People aren’t impressed with the figures when they’re living with the economic realities, which remain unyieldingly lousy.

All this has stopped the gloating among most of my lefty friends, but some stubbornly cling to wunderkind Nate Silver as proof that Obama is going to pull this out. Nate Silver, curator of the New York Times’ much vaunted 538 election forecast blog, made quite a name for himself with his analysis of baseball stats, and he has applied a very sophisticated model to predict the outcome of the 2012 election. To bolster claims of Silver’s infallibility, 538 acolytes point to his impeccable track record of predicting the outcomes in 49 of 50 states in 2008 accurately, along with picking the winners of all 35 senate races in the same year. Currently, he insists that Romney’s “bounce” has only marginally shifted the race, and that Obama is still a prohibitive 3-1 favorite to win reelection. So every time I think Romney’s doing well, I’m told to pay attention to Nate Silver, because the race is really over, and if I were as smart as Nate Silver, or at least not so silly as to ignore Silver’s sagacity, I’d know that and would therefore stop making such a slobbering fool of myself.

Let’s take a closer look, shall we?

Suppose I were to accurately predict the weather for tomorrow. How impressed would you be? I’m thinking not very. Our temporal proximity to tomorrow gives us enough information that most of us, even without meteorological training or satellite data, could safely predict that a 90 degree day won’t be followed by a blizzard within 24 hours. I could tell you that tomorrow would be pretty much like today, and, 9 times out of 10, I’d probably be right. However, if I could tell you, with precision, what the temperature will be a month from now, I’m betting you’d be tremendously impressed. It takes far more skill – and requires far more risk – to make long-term forecasts.

Nate Silver isn’t capable of making long-term forecasts.

That doesn’t stop him from creating the illusion that that’s what he’s doing, though. After all, if the election were held today, according to Silver, Obama would win handily. But there’s no way to prove that hypothesis, because the election isn’t being held today. And Silver, like any good statistician, alters his numbers when he gets new data, so he doesn’t have to stick by his predictions, even for 24 hours. By the time Romney’s imminent victory is as close to us in time as tomorrow’s weather is, no doubt Silver’s impressive model will have caught up with reality, and he’ll get kudos for calling the race accurately, just as he did last time. Except is there anyone in America who wouldn’t have been able to call the race for Obama over McCain  at any time in November 2008? Silver is getting credit for long-term foreknowledge on the basis of his unimpressive and easily imitated short-term predictions.

If you doubt that, consider Silver’s long-term foreknowledge in 2009, when he was describing the Republican “death spiral” that would surely spell GOP defeat in the next year’s mid-term elections. Why were Republicans doomed? Because they had voted en masse against a bill subsidizing the transition from analog to digital television. Ummm… what? Does anyone remember that? They surely didn’t in the actual 2010 elections, when Republicans retook the House and made significant gains in the Senate. Of course, Silver’s wacky prediction was forgotten and replaced with his more reliable statements made closer to the actual election.

InTrade does the same thing, by the way. InTrade is startlingly accurate on the day before election day, Prior to that, it varies wildly, and, even on election day, it can shift from 0 to 100 almost instantly. The day before McCain selected Sarah Palin as his running mate, InTrade said Romney was 90% likely to be the 2008 veep. Palin was at 0%. Both Silver and InTrade, then, are lagging indicators, not predictors of future events. And, again, there’s no way to verify their accuracy in the weeks and months leading up to an election.

Another interesting tidbit about Silver. In 2008, he was employed by the partisan hacks at the Daily Kos prior to the New York Times. It was there that he signed a confidentiality agreement with the Obama campaign in order to gain access to their extensive internal polling. Campaigns guard their internal poll numbers with their lives, and they’re far more reliable than what’s peddled to the public by media outlets. How accurate would your predictions be if you could look at Obama or Romney’s internal polling? Silver, incidentally, has no such access this time around.

This is not to accuse Silver of corruption or bias or anything else nefarious. He’s certainly very bright, and I’ve no doubt that he’ll get it right in the end. But it’s foolhardy to assign any weight to his entirely unprovable assertions a month out from the election as if it can compensate for the reality of the total implosion of Obama’s campaign.

That said, watch for Joe Biden to do surprisingly well on Thursday, which will provide another thin reed for Silver and Obama supporters to hold on to as the president continues his inexorable sink into the mire.

Other Campaign News
Remembering The Message

Leave a Reply

  1. Silver is a liberal shill.

    The “49 out of 50 states” and “all 35 senate races” line is completely overblown. Right now, I could correctly predict about 30 or 35 states in the 2016 presidential election … and I don’t even know who either candidate will be! That’s because so many states are either reliably blue or reliably red. So “49 out of 50” isn’t nearly as impressive as it sounds, even if the predictions were made months in advance (which Silver’s were not).

    As for the 35 senate races Silver correctly predicted in 2008, in only 4 of those 35 races did the victor win by less than 5% (see link below). That is, in 31 of the 35 contests, the race wasn’t particularly close. So I’m again unsure why I should worship at the Nate Silver altar when anyone would have made almost the same predictions he made. I don’t need a fancy model to predict the obvious.

    So with Silver, he’s not just getting credit for making late predictions with no meaningful horizon, he’s getting credit for correctly predicting outcomes that aren’t even in doubt!

    http://en.wikipedia.org/wiki/United_States_Senate_elections,_2008

    • If during the 2008 presidential election my “model” were to predict that the winner of each state would be candidate from the party that won that state in 2004, I would have (by my count) correctly predicted 40 out of 50 states. That is, even in a year when (a) both presidential candidates were nominated for the very first time (such that they were not relevant to the 2004 results), and (b) the outcome of the race was vastly different than it had been 4 years previous (e.g., the Republican candidate won in 2004, while the Democratic candidate won in a landslide in 2008) … even the fact that there were fresh faces on both tickets and there was a complete reversal of fortune for the two parties … my “model” would have been correct on 40 out of 50 predictions.

      A middle-school education and 10 minutes of looking at the most basic polling data would have likely allowed me to correctly predict at least 45 out of 50 states.

      Forgive me for not waiting with baited breath for every Nate Silver pronouncement.