Things Not in the Model-Black Swans-Fat Tails Election Weblogging
Rick Perlstein: Romneyland

Today's War on Nate SIlver: Quiet Flows the Don Edition

In the Serafimovich bridgehead on the right bank of the Don River, the Sam Wang Front--which has been hobbled by logistical difficulties--begins operations against the underequipped and underdressed forces of Karl Rove:

Ro-mentum watch: Karl Rove: Karl Rove surprises basically nobody by predicting a Romney win. His reason? He cites a Romney lead in some national polls….

If state polls are accurate on the whole, then Obama will win. However… national polls since October 14th give a tied median…. [T]he discrepancy… has been over 2.0% all season.

What is going on?… Do differences in national and state poll methods account for the discrepancy?… Answer: no. Are state polls slow to catch up?… Answer: no….

Answer: national polls do about 2.5x worse at predicting the popular vote outcome than expected if the wisdom of crowds of pollsters were perfect…. But if state polls use the same methods, why would they do better than national polls?… State polls target more homogenous populations, which poses fewer technical problems to the pollster…. [S]wing states… are done more frequently. This focuses the data where information is most needed. As for why the weighted sum of state polls gives a result that differs from national polls, the only reason I can think of is… state polls might be technically easier to conduct and weight. Still mulling that one.

BOTTOM LINE…. Because state poll aggregation is so powerful, the result based on state polls is likely to be more accurate. That is what I would call The Math.

Seth Mandel--well, it is not clear what Seth Mandel is trying to do, other than warn Nate Silver that it is to his long-term career advantage to trim back his Obama probability estimates somewhat:

Latest Defense of Nate Silver: Even When He’s Wrong, He’s Right: At this point in the election, both the national tracking polls and, generally speaking, the battleground state polls show an extremely close race. This has inspired much criticism of stat-man Nate Silver. Liberals, as Jonathan wrote recently, may be looking for a scapegoat if Barack Obama loses, and will wonder why Silver insisted that statistically Obama was an overwhelming favorite to win re-election right up to the end. Conservatives say that statistics don’t factor in momentum, that any predictive model that ignores trends should be taken with a grain of salt, and that current trends don’t back up Silver’s predictions of Obama as the heavy favorite…. [Ezra Klein] basically boils down to: Don’t blame Silver if he’s wrong, because he’s relying on other people’s work, and no matter what happens, Silver wasn’t wrong, because he said it could happen, and it did.

Mark Coddrington's horde of cossacks crosses the Volga to conduct a "deep battle" penetration raid on the epistemological foundations of "savvy" journamalism:

Why political journalists can’t stand Nate Silver: The limits of journalistic knowledge: Journalists get access to privileged information from official sources, then evaluate, filter, and order it through… “news judgment,” “news sense,” or “savvy.” This… is how political journalists say… “This is why you can trust what we say we know — because we found it out through this process.”…

Silver’s process… is almost exactly the opposite… public… systematic… scientifically based…. we can trace how he applied that judgment to reach his conclusions….

David Brooks… There’s no place in Brooks’ journalistic way of knowing for quantifying… so he can’t see how anyone else can know…. Joe Scarborough gets us even closer to the clash….

Nate Silver says this is a 73.6 percent chance that the president is going to win? Nobody in that campaign thinks they have a 73 percent chance — they think they have a 50.1 percent chance of winning. And you talk to the Romney people, it’s the same thing.

How does Scarborough know that Silver’s estimate is incorrect? He talked to sources in both campaigns. In Scarborough’s journalistic epistemology, this is the trump card…. These are the savviest, highest inside sources…. When journalistic objectivity is confronted with scientific objectivity, its circuits are fried.

Perhaps the most interesting thing is that we all know that their "savvy" sources are lying to the "savvy" Brooks and Scarborough. everybody on both campaigns agrees with Nate Silver: they all think that Obama does not have a lock on next Tuesday's election, but that he does have the better position. No political professional in either camp right now would bet on Romney at 1-1 odds, or would bet on Obama if they had to give 10-1 odds. Every Republican hopes that Gallup's likely voter screen is accurate, but thinks it a bad bet. Every Democrat fears that Gallup's likely voter screen is accurate, but thinks it not terribly likely

So why does Scarborough think that it is 1-1 when none of the professionals he talks to does? Why does Brooks think it impossible to estimate probabilities when everybody he talks to thinks Romney's odds are worse than 1-1 and better than 10-1?

It is a great mystery…

Sean Davis assaults the tractor factory with no mental weapons whatsoever, showing for an MBA an extraordinary lack of knowledge as to how and where black swans roost:

Is Nate Silver's value at risk?: [H]umans are desperate to view the world as far more rational and predictable than it actually is…. Silver stormed onto the scene in 2008…. I spent a few hours re-building Nate Silver’s basic Monte Carlo poll simulation model… [T]he winning probabilities [my model] produced for Obama and Romney were nearly identical to those reported by FiveThirtyEight…. So what gives? If it’s possible to recreate Silver’s model using just Microsoft Excel, a cheap Monte Carlo plug-in, and poll results that are widely available, then what real predictive value does Silver’s model have?…

[T]he FiveThirtyEight model is a complete slave to state polls. When state polls are accurate, FiveThirtyEight looks amazing. But when state polls are incorrect, FiveThirtyEight does quite poorly…. [O]f the five major state races in which polls were wrong over the last four years, Silver only got one right… batting .200 when it counts won’t get you into the big leagues….

This takes us back to Nassim Taleb’s key insight: despite our best efforts, we humans are just not that good at predicting the future. The main assumption underlying Nate Silver’s Obama bet this year is that the state polls will be correct. Maybe they will be, even though three states were wrong in 2010, two states were wrong in 2008, one state was wrong in 2004 (WI), and a very important state in 2000 was incorrectly called…. But if they’re wrong, it’ll be Nate Silver whose value is at risk. If that happens, I have a great title for his next book: “The Snake and the Oil.”

And Matt Lewis of the "Daily Caller" reveals himself as another unarmed man entering the battle of wits: he hates Nate Silver, but cannot spell "prophecy":

On hating Nate Silver: when the public sees that a prominent New York Times writer gives Barack Obama a 70 percent chance of winning, that can become a sort of self-fulfilling prophesy. It has consequences. It drives media coverage. It dries up donations. Whether Silver likes it, or not, people do interpret his numbers as a “prediction.” They see this as election forecasting….

Conservatives are, by nature, skeptical of utopian schemes. The “best and brightest” always believe they can invent a perfect analytical model that will solve all the worlds problems. This, of course, is a fatal conceit. Models break down. This model might be breaking, too…. Gallup wrote, “Romney currently leads Obama 52% to 46% among voters who say they have already cast their ballots”… the energy will be on Romney’s side… polls using a 2008 turnout model are, in fact, skewed, and that Romney’s turnout will be higher than anticipated.

If polls are skewed, then (as Klein states), Silver’s model is wrong. Garbage in, garbage out.

There is another reason that conservatives have grown to disdain Silver, and that has to do with the condescending attitude put forth by media elites. What is more, conservatives sense his friends in the media will protect him, no matter what…

Approaching the front, Russ Roberts diverts himself onto a dead-end spur and begins denying that you can calculate the probability of a unique zero-one event before asserting via the Principle of Insufficient Reason that the real odds are 50-50:

Nate Silver and objectivity: Nate Silver is generating a lot of excitement on the internet. I’ve been a fan of his for a long time. He was a superb writer on baseball and he’s taken his knowledge of statistics and applied it to politics. The last time I looked, he was estimating a 75% of Obama being re-elected. That seems a little high but it could be right, whatever “right” means where Obama either wins or loses. The state polls suggest Obama is going to win. The national polls suggest Romney is going to win. Of course it’s possible that Romney wins the popular vote and loses the electoral college but that’s unlikely given the size of Romney’s lead in the national polls. We’ll know more on Tuesday.

The fact that Romney does not have a current lead in the national polls is, to Roberts, apparently neither here nor there--Wednesday's national likely-voter polls gave Obama a median national lead of 1%.

Even though the Politico Brigade has been rated "combat ineffective" by Army Group B Staff, General Seydlitz orders them to return to the attack, and they comply with another effort to prove via the principle of insufficient reason that the presidential race is a 50-50 tossup:

Dylan Byers and Mackenzie Weinger: Media stumped by 2012 outcome: Gov. Mitt Romney’s campaign says it still has momentum. President Barack Obama’s campaign says that’s all spin. Meanwhile, there isn’t a single well-informed pundit between them who can tell you who’s right.

“The problem is: there are so many variables. And now, with the storm, turnout may become an issue in closer Obama-leaning states like Pennsylvania,” Time Magazine’s Joe Klein told POLITICO. “Polling is inexact, especially with the cell phone factor—not enough data over time for pollsters to be absolutely sure they’re getting it right.” In a column headlined “I don’t know,” Klein wrote earlier this week, “Anyone who claims to know who is going to win is blowing smoke.”

Let me note that when Joe Klein is your Chief Statistical Analysis Officer, you are in huge trouble indeed.

Indeed, from Fox News to MSNBC to the Wall Street Journal to the New York Times, all but those aligned or associated with the campaigns — and even some who are — admit to being deeply uncertain about Tuesday’s outcome. “I think more than any other race I’ve covered this is one where both sides genuinely seem to believe they’re going to win. That’s different,” ABC News correspondent Jon Karl said. “Given that, it’s hard for somebody covering the race to make a call. I’m completely confused. I have no idea who’s going to win. And I usually have a sense of who’s going to win.”… “It really is too close to call this time, as it was in 00 and (a bit less so) in 04,” Bill Kristol, the founding editor of the Weekly Standard told POLITICO. “People are saying it’s too close to call because it is.” “The key is that no one on anyone’s own side is calling it,” said John Podhoretz, the New York Post columnist. “2008 featured a lot of Republicans talking about ‘headwinds’ and McCain saying it was a 6 point race. That was the tip-off. Now, nobody knows anything.”

In addition to tight poll numbers, pundits know that the trajectory of the campaign can turn on a dime. A change in a candidate’s message, new revelations unearthed by the media, an unfortunate gaffe — all of these things could change voter sentiment in Ohio, Florida, Virginia, or any of the other battleground states and tip the scales in the last days of the election…. The Obama camp has been forced to compete against pro-Romney ads that have recently gone up in Pennsylvania, Minnesota, and Michigan…. Jennifer Rubin, the Washington Post’s pro-Romney blogger, wrote on Oct. 16, that if her candidate had “a solid final debate,” worked “on turning out the vote” and stayed positive, “in all likelihood he will win the presidency.” One Red State blogger went further: “Romney will win,” the blogger argued. Touré, the liberal MSNBC co-host, struck an equally confident tone for President Obama in a recent editorial for Time Magazine. “Fortunately there are many factors that allow me to sleep at night, because they lead me to think Obama will win on November 6,” he wrote on Oct. 26….

MSNBC’s Joe Scarborough said earlier this week that “anybody that thinks that this race is anything but a toss-up right now is such an ideologue, they should be kept away from typewriters, computers, laptops and microphones for the next 10 days, because they’re jokes.”… “I stopped trying to predict these things many cycles ago,” Klein said. “I stopped making predictions for anything—who’ll win Iowa, who’ll win the nomination, who’ll with the presidency—a long time ago. We are excellent at reporting the past, not bad reporting the present. But we absolutely suck when it comes to the future.”

Alexandra Petri at the Washington Post claims:

Nate Silver’s weatherman problem - ComPost: The mere fact of rolling a three tells you precious little about how many threes were on the die. At least one. But beyond that, it’s difficult to say.

Really the only way for the public to start to tell if Silver’s model worked or not would be to have this election a hundred or so times. Even then you might not know.

And that’s not a fate I'd wish on anyone.

The fact that Nate Silver's poll-aggregation and demography approach has been validated in a host of previous elections before this one is, apparently, not something that she knows, or bothered to learn. Is it an accident that her weblog is called "Compost"?

Robert Schlesinger on why the right-wing has chosen to fight its war on arithmetic at Silvergrad:

Mitt Romney’s Electoral Problem and the War on Nate Silver: Conservatives, especially ones who believe in the myth of Mitt-mentum, are not big fans of Silver. So he's come under increasing fire from the right as merely a partisan hack. The most astounding and bizarre such attack came from one Dean Chambers, who criticizes Silver for assigning different weights to different polls, an astounding critique given that Chambers started a whole Web site devoted rewriting poll results to fit his partisan proclivities. (His attack was bizarre because he seemed to suggest that Silver's analytical skill is marred by what Chambers sees as his effeminate qualities.)…

Conservatives might dislike and disagree with the numbers Silver is pushing, he is not alone in pushing them. There are in fact several Web sites and/or scholars who push statistical models aimed at making similar estimates about who will be the next president, and they all give an edge to President Obama. The Princeton Election Consortium… Votamatic… Real Clear Politics…. And the major online betting markets all give Obama pretty good odds of re-election (Intrade puts it at 63.3 percent chance, and Betfair says 68 percent).

Rather than the figure conservatives portray—a lonely voice inveighing against the oncoming Romney juggernaut, giving liberals a last beacon of hope before the inevitable change—Silver and his percentages are comfortably in line with a host of other prognosticators…. So why is Silver getting the attention from conservatives and others? That's simple: He writes for the Times…. If the newspaper had hired Wang two years ago we'd likely be reading "Sam Wang: One-term celebrity?" in Politico instead of the recent Silver-focused piece.

And the partisan troops No More Mister Nice Blog begins sniping at the rear-area Romanian Halperin Division:

No More Mister Nice Blog: On Morning Joe today, Mark Halperin explained what he says is Beltway conventional wisdom: Romney may lose because Romney is perceived as losing:

MARK HALPERIN: I'm going to take a risk here and peel back the curtain to tell you what insiders are thinking…. [I]nsiders… look at the last few days of swing state polls and say, 'The President may have this'…. [E]lites are starting to think… that these leads are, as the Obama campaign has said… small but persistent and consistent…. Romney needs to go into the weekend, for his own sake, with the race tied in the conventional wisdom….

I'm not aware of any moment when Obama or a surrogate told a crowd at a rally, "Vote for us -- we're winning." But Romney has said that a lot. And Team Romney has worked very hard to persuade the press that he's winning. Which means that if Obama wins, the right will insist that stories saying Romney was losing caused Romney to lose. There'll be right-wing books with titles such as In the Tank: How the Mainstream Media Stole the 2012 Election for Barack Obama. Large portions of these books will be devoted to the notion that Nate Silver, above all others, ruined Romney's chances with his evil math…

Comments