Nate Silver ‘A Bit Terrified’ by Current Polling Landscape


NYT election forecaster Nate Silver, who has got Obama currently at a 63.3% chance of winning, looks at the bog of state and national polls coming in for the last week and says it's going to take more data to decode:

Anyone in my business who is not a bit terrified by this set of facts is either lying to himself — or he doesn’t know what he’s doing.

There are three ways out of the stalemate. First, the state polls could move toward Mr. Romney. Second, the national polls could move toward Mr. Obama. Or third, we could receive more emphatic evidence that the difference between state polls and national polls in fact reflects a potential difference between the popular vote and the Electoral College. (This latter case, importantly, would require evidence that Mr. Romney was running well in noncompetitive states along with evidence that Mr. Obama was performing well in swing states.)

He says an ABC News poll today could be of some help with the forecast:

The forecast does not yet account, however, for a national poll conducted for The Washington Post and ABC News, which was released early on Monday morning. That poll gave Mr. Obama a lead of 3 percentage points in the national race among likely voters, and a 7-point lead among registered voters, both figures tying for his largest of the year in that survey.

We will see what the rest of Monday’s polling data brings, but The Washington Post and ABC News poll has the potential to be influential on the forecast, in the same way that a Pew Research poll showing a sharp break to Mr. Romney was last week.


  1. MarkUs says

    A “poll taker” or “forecaster” who describes himself as “terrified” has outed himself as what most knew from the beginning: he’s a partisan hack.

  2. Charlie says

    The full paragraph said
    “There have been plenty of elections before when the outcome was highly uncertain down the stretch run or on Election Day itself. But I am not sure that there has been one where different types of polls pointed in opposite directions. Anyone in my business who is not a bit terrified by this set of facts is either lying to himself — or he doesn’t know what he’s doing.”

    I think what he is concerned about is that state polls and national polls show different results.

  3. Butch says

    The composite over at Daily Kos is clearly trending Obama; I don’t get where Silver obtains his numbers.

  4. Sanderson says

    Butch/Markus: anyone who follows Nate Silver knows where he gets his numbers. If you don’t know, go look. I’ve yet to find any pollster who takes the holistic and broad approach to making election predictions as Nate Silver.

    In any case, a 3% lead or more in national polling is agreed by most to mean that there is no issue with the Electoral College. No one has ever won by more than 2% and not won the Electoral College. So a 3% lead in polls for Obama should be sufficient to win the electoral college.

    Right now, Ohio is slipping further and further from Romney’s grasps. If he loses Ohio to Obama, he would have to flip Wisconsin (which is polling even better for Obama than Ohio), New Hampshire, Virginia, and Colorado from Obama. No one who has yet to hit 50% every in a single national poll has ever managed to do that. If it happens, then none of these pollsters mattered much because this election was clearly nothing compared to any that has come before it.

    With Ohio, Obama essentially only needs to win Wisconsin, Colorado, Iowa, New Hampshire OR Virginia and he’s at 270. I think it is good to have people worried so they stay involved, continue to contribute time/money, but until Ohio polling shows Romney making some serious ground up there, I’m not yet concerned.

  5. Butch says

    Are you finished with your lecture? Silver is only one of a lot of statisticians, and he’s not infallible. Rand American Life Panel composite is showing Obama up by nearly 5.

  6. Caliban says

    Markus, you tool, Silver isn’t “a bit terrified” by polls showing that Obama’s reelection could be threatened, as a statistician he’s “terrified” by polls that are all over the map and show contradictory results.

  7. MarkUs says

    Andrea Mitchell and Chuck Todd both said this morning Colorado, Florida, and Iowa have all “slipped out of reach”. It’s funny all-important Florida with nearly DOUBLE the electoral votes of Ohio doesn’t matter anymore.

  8. BRAINS says

    The polls will continue to swing, as they always do!

    However, as voters become repulsed by the flip-flopping liar Romney is they will flip towards President Obama, permanently!

  9. jamal49 says

    Recognizing that Silver is a partisan hack and the NYT is a hack newspaper, I’m not quite sure of what Mr. Silver is “terrified”. Beware wussy pollsters who use the word “terrified” when discussing polling results. It means his latte got cold before he could finish it.

  10. RyanInWyo (formerly RyanInSacto) says

    JAMAL49: What Caliban said above to Markus, applies to you too.

    Take a reading comprehension course or something.

  11. mike8787 says

    @Sanderson I’m as excited by that news as you are, but to be fair, the cohort that makes up early voters is much more likely to be composed of Obama supporters.

  12. BRIGGS says

    i listened to the morning discussion on msnbc with chuck todd of the current state of the race–taking into account the new washington post poll–which made it all sound very discouraging for obama at this poiint”

    imagine my surprise, when i actually looked at the poll later on the internet and discovered that the poll actually gave obama a point higher lead over rommney then in late september–when obama was supposed to be surging– and that his lead in the swing states stands at 6 percent, while the view of the economy and the direction the country is going in has improved measurably.

    what kind of story is todd and msnbc trying to run with, given the distortion of the poll findings i observed this morning.

  13. Marc C says

    @CALIBAN: You have to understand MARKUS isn’t one of our more critical thinkers here at Towleroad.

    …or in the United States for that matter.

  14. shle896shle says

    Conducting a poll every time one of the candidates has a bowel movement tells us NOTHING. STFU already Nate! Obama and McCain were neck and neck at this point in 2008 and President Obama won handily. All this hand-wringing is ridiculous.

  15. Bob R says

    If you notice the graphics and have been following the polls, you’ll note that Obama was trending up, while Romney was trending down. Then came the “debate”. Obama’s dismal performance created a seismic shift in the trend lines. Many voters were really “re-exposed” to the candidates personally. So much for those that insist debates don’t matter.

    Now note the downward trend for Obama and upward trend for Romney seem to have reversed or gained momentary equilibrium. This change is due to the VP Debates that Biden won.

    Trends from this Obama/Romney first debate and forward will be more telling since the majority of voters who aren’t “political junkies” or partisan loyalists really don’t give Presidential politics any attention until after Labor Day. Thus, Obama’s loss at the first debate had a major impact.

    I think you’ll find that whoever is ahead in the polls on October 23-25th (barring some major catastrophic event) will go on to win the election. Normally, very little changes in the two weeks prior to election day.

  16. jb says

    If you actually read Silver’s blogpost (a tall order for most of you, I know) instead of just reading the word “terrified” out of context, it’s pretty clear what he’s terrified of. He is terrified that he won’t be able to predict the outcome of the election correctly because the data are behaving so weirdly in his mind. His professional reputation depends on his model getting the outcome right, and as of now it’s looking like it won’t be able to do that. If the data weren’t so mixed, he would not be “terrified”–even if they clearly pointed toward a high likelihood of a Romney win.

  17. Samuel says

    No matter Romney surge lately, Obama has a better chance of winning this election. To me, he has to be a bit aggressive but not disrespectful and have his facts right.

  18. jamal49 says

    RYANINWYO I read it all: the article here, the article at the NYT website, all the comments, including what Caliban said to Markus.

    I am also a graduate of the Copacabana School of Fine Arts, majoring in How to Read and Write Really Good. So there. I got it.

    What I don’t get is why Mr. Silver is “terrified”. That’s a bit much to say, isn’t it?

    What’s he terrified of? The polls? The difference between state and national polls? What? That this may effect an outcome where the election is thrown to the Electoral College?

    Those are my questions. They ain’t been answered. And, I’ve just reread everything again.

    So, darling, perhaps you could enlighten me, since obviously I ain’t as brilliant as thineself.

    Ever so. Thank you.

  19. Mike says

    I love when people with no grasp of political science or polling try to talk about it like they know what they’re saying. Pollsters do this for a living, you’re sitting behind a computer typing away like a snide little prick.

  20. Caliban says

    OK, so “terrified” may be somewhat of an exaggeration, a perhaps too dramatic word to use. But analyzing statistics is what Nate Silver does- he has been lauded as the most accurate predictor of the 2008 election. He first gained fame as a stunningly accurate BASEBALL analyst then moved over to political analysis and his ability to incorporate so many factors into his statistical models was seen as revolutionary.

    So polls with contradictory results, whether national or state, are making it difficult for him to accurately forecast the results. So he’s “worried” about the accuracy of his statistical models and therefore his reputation. Don’t get too hung up on the word “terrified.”

    An interesting question would be WHY the polls are so erratic and contradictory. Are some of the questions or the results deliberately skewed to present a certain picture and that’s what’s throwing off Silver and other statisticians’ ability to forecast?

  21. emmy says

    Silver doesn’t need to be “outed” with respect to his politics. He’s open about it, and he believes that analysts *should* be open about their biases rather than pretending that they don’t have any. That said, he’s no “partisan hack”. He’s an independent who skews “Libertarian” on social issues and “centrist” on economics. And the only reason that people pay such careful attention to his predictions is that he has an excellent track record.

  22. Jay says

    He is far from a statistician, anyone who claims he is, doesn’t understand statistics. And is swallowing his one sided journalism hook line and sinker. HE NEVER DOES WRITE UPS AS A PROPER STATISTICIAN SHOULD.

    Look up his academic history, BA Economics, I’m not sure he even took a calculus class. He certainly wasn’t required too. He’s just riding the money ball wave.

    He, I’m guessing, used a linear regression to find statistically significant indicators for winning games, that recruitment methods, at that time, weren’t exploiting. Mainly for a fantasy baseball addiction, found a coach desperate enough, it was successful, the million dollar franchises got real mathematicians and he got incredibly overrated publicity as a “mathematician.”

    Anyone ever wonder why he’s now a political BS analyst, and not the MLBs foremost expert on sabermetrics? :)

    Yeah he’s awesome alright…