Princeton Election Consortium

Innovations in democracy since 2004

Outcome: Biden 306 EV (D+1.2% from toss-up), Senate 50 D (D+1.0%)
Nov 3 polls: Biden 342 EV (D+5.3%), Senate 50-55 D (D+3.9%), House control D+4.6%
Moneyball states: President AZ NE-2 NV, Senate MT ME AK, Legislatures KS TX NC


October 23rd, 2012, 3:30pm by Sam Wang

For new readers (of which there are many!), I make a brief introduction to our highest-value product.

The Princeton Election Consortium’s main contribution is to give you a high-quality “thermometer” of exactly where the Presidential race is, based on state polls – and therefore the Electoral College, which is how the Presidency is determined. We do not use econometric indicators or national polls because these are only indirect measures (“What political science models really tell us,” August 28).

The track record of our state-polls-only approach in past years is quite good. In 2004, on Election Eve, the median was Bush 286 EV, Kerry 252 EV – the exact outcome. In 2008, we predicted Obama 364 EV, McCain 174 EV – just 1 EV off from the final outcome of Obama 365 EV. (For true nerds: in the latter case, the error was effectively 0.1 sigma, which suggests that we might be overestimating the 95% confidence interval.)

In addition to this, our time resolution is excellent. Other aggregators (Pollster, FiveThirtyEight) use methods that typically must be integrated over at least a week to reveal conditions. In contrast, we give a highly precise statistical snapshot of the race based on polls only.

This means we can test the recent assertion that the Romney campaign has the momentum. In 2004, Joe Lieberman’s Presidential campaign claimed to have “Joementum” to indicate they were going places. Where they went was a “three-way tie for third place.”

So…does the Meta-Analysis show real momentum? Or “Ro-mentum”? (update: bonus InTrade takedown at the end!)

History of electoral votes for Obama
Here’s a history of the Electoral Vote estimator. It achieves high resolution because we use poll margins in all states to calculate a win probability – and then we evaluate all 2.3 quadrillion possible outcomes, using a math trick to do it quickly. All the swings of the race are made visible. This year, the largest effects came after the Ryan VP nomination, the Democratic convention, and the first debate.

What is apparent is that the large plunge after Debate #1 came to a stop last week, right around the time of the VP debate. After that and Debate #2, Obama made some recovery. Now we are at a plateau, in which Obama is slightly – but decisively – ahead.

How far ahead? That can be answered using the Popular Vote Meta-Margin:
History of Popular Meta-Margin for Obama
The Meta-Margin shows how much the race would have to swing to create an electoral near-tie. It is precise to within 0.5% or less.

Today, the race is quite close. However, note this. In terms of the Electoral College, President Obama has been ahead on every single day of the campaign, without exception.

I would then give the following verdict: Indeed the race is close, but it seems stable. For the last week, there is no evidence that conditions have been moving toward Romney. There is always the chance that I may have to eat my words – but that will require movement that is not yet apparent in polls.

The popular vote is a different story. I estimate an approximately 25% chance that the popular vote and the electoral vote will go in opposite directions – a “Bush v. Gore scenario”. I regard this as a serious risk, since it would engender prolonged bitterness.

In summary: Ro-mentum!

Update: Via Marginal Revolution: for about $1,250 it is evidently possible to manipulate InTrade. This morning’s swing toward Romney was caused by one trader’s manipulation. Ro-mentum!!


To learn more details, read our FAQ, or for true geeks, the statistical methods.

Tags: 2012 Election · President

139 Comments so far ↓

  • madguy

    Is “Ro-Mentum” being used sarcastically here? Otherwise, I don’t get it!

    • Mr. Patient

      Yes, Sam is comparing it to Joementum, which was a rather sad attempt at spinning a primary loss into a win.

    • Bobo

      It’s the Mittmentum I worry about. Cheating. And cheating fits Mitt’s MO to a tee.

      We saw what happened last time we were this close. The guy who lost Romneyed his way into the WH. And boy did the USA pay for that. 911, a war based on lies and a cratered economy.

      Who will monitor the automated vote-count in Ohio? Is it a legitimate concern, or have I gone McCain?

    • SkepEng

      It’s satirical. Have a look at

      “Since then, the term has become popular among some bloggers as a satirical reference to both Lieberman himself and his election efforts, usually referring to the perceived lack of potential for success of a campaign or endeavor”

    • Ian

      No . This is not sarcasm.
      Prof. Wang is being redicurus

    • Ms. Jay Sheckley

      Bobo- I’m worried, and though it’s outside the purview of poll aggregation I’d really like Sam to cobble up some odds on a national case of Romtheivia

    • voter suppression

      I can’t speak for Prof. Wang, but it might be of some comfort to know that Nate Silver’s (relatively pessimistic) forecast already factors me in.

  • Jay

    What seems a possibility is that Obama will have the advantage on sites like this until a few days out from the election, and then suddenly the EV divide will break for Romney, and once again, an “accurate prediction” can be garnered with a Romney victory.
    I just hope this isn’t the case.

    • Ohio Voter

      That’s a really, really big EV divide to overcome. You’re talking 2% in a matter of days. As Dr. Wang says, it’s all in the oven, I can’t imagine a situation after the first debate that would prompt anyone to join the Romney wagon if they already hadn’t.

      And let’s not forget. There are people voting right now.

    • Jay

      Ohio Voter: Wasn’t Kerry winning (barely) in Ohio for many weeks leading into the election? — and then suddenly a few days out, he was losing in the polls.
      Ohio can change at any time. I’m sure you can attest to the fact that Ohio is somewhat like all of America packaged in one state. You have the industrial union-friendly Cleveland. You have the hi-tech and large university city of Columbus. You have the pseudo-Bible-belt dynamic of greater Cincinnati. You have farming communities. You have struggling towns like Youngstown/Toledo/Dayton. Anything can swing in that state. It just matters who actually votes.

    • Ohio Voter

      Kerry pulled into striking distance (and was leading in a few polls) in early/mid October before Bush regained his footing heading right into election day.

      So yes, it does look a little Obama is Bush.

    • Ohio Voter

      For some reason, it cut off my last comment. But yes, it does look at little like 2004, except that Romney is Kerry and Obama is Bush.

    • Strabo

      “Ohio Voter: Wasn’t Kerry winning (barely) in Ohio for many weeks leading into the election? — and then suddenly a few days out, he was losing in the polls.”

      Not really. Taking the polls from 13th of October to 22nd of October 2004 as example, Bush led by 0.2 % points. The race fluctuated and Kerry was a bit ahead the next week, but it was less than 0.5 %.

      Compare this with Obama’s 2-3 % lead in Ohio (1.9 % on RCP, higher with other aggregate sites).

    • MarkS

      Ohio Voter wrote “I can’t imagine a situation after the first debate that would prompt anyone to join the Romney wagon if they already hadn’t.”

      Then you didn’t watch MSNBC’s focus group of 8 “undecided voters” after the 3rd debate. 7 thought Obama won, but 2 of these were now leaning towards Romney.

      No, I don’t understand it either.

    • P G Vaidya


      Let us, like, Lojn Cleese, look at the “bright side of things”.
      Just look at the Gallup numbers today.
      Obama likability has jumped from 49 to 51 and “dislikability” has gone down 1. These are averaged over a 3 day cycle. The dreaded likely poll has Obama numbers improving today by 1.

      There are three narratives from a signal processing point of view to look at moving averages.

      1: Averages are a form of an integration and thus reduce noise.
      2: These averages can be seen as difference. Thus in the Rand Poll the seven day average for Oct 15 to Oct 21 was 48.06. The same thing for Oct 16 to 22 is 48.26. All this says is that the sample on the 22nd was 0.2 time 7 = 1.4 greater than the one on October 15. This process is similar to differentiation and thus increases noise and therefore is less accurate. Now, does it mean, for Gallup, that Obama’s approval numbers jumped by 6 and in the likely voter case, jumped by 7 over the corresponding days which were dropped? Not quite, because there is an even greater round off error because of the less precise nature of Gallup data which is in an integer format.

      3 We can take a longer Gallup data set and use the information they have provided for the sequence of 7 day averaging, plus the tidbits they have been giving about at least four sets of three day sub-samples of the data. Set this all as a straightforward under-determined problem and solve using a Moore-Penrose generalized inverse, and we can get the least norm solution.

      In any case, this is a cause for optimism.

      There is another reason to cheer today. Rassmusen numbers have gotten more favorable for Romney. Victory must be near!

  • Ohio Voter

    I know many folks around here think that Romney’s “momentum” is simply a result of MSM wanting to see a horse race.

    While that’s certainly possible, I think it’s mostly the MSM’s (and bloggers, actually) desire to only focus on national polls, and more over, daily tracking polls.

    Though I’m pretty confident if the President wins the Electoral College and loses the Popular Vote, the likely Republican House will be introducing the “Electoral College Abolishment Bill” on Day 1 (pay no attention that the EC gave us Bush in 2000)

    • Matt McIrvin

      There’s already an ingenious National Popular Vote movement to effectively abolish the EC at the state level.

      The idea is that various states pass a law stating that once the whole collection of states that have passed the law are capable of swinging the presidential election, they’ll all simultaneously start assigning 100% of their electoral votes to the national popular-vote winner.

      (Until that point, the law has no effect, which solves the collective-action problem that a piecemeal approach would reduce any given state’s influence.)

      There’s some controversy over whether the whole thing is constitutional, but several states have passed it. I suspect that an Obama win with a PV minority would lead to a groundswell of support for it in the red states, and we’d get to see whether it actually works.

    • InmanRoshi

      The GOP Congress placating a cry from the far right to abolish the Electoral College would be the best possible scenerio for Democrats.

      Imagine the advantage Democrats would have in future popular vote based elections if voters in New York or California felt their votes actually mattered in Presidential Races, and if the Democrats were running a candidate who’s name/skin color alone wasn’t a complete dis qualifier through huge swaths of the country.

    • Matt McIrvin

      Congress couldn’t actually do it with a federal law; it’d take a constitutional amendment. (The NPV movement gets around this by exploiting the way the Constitution grants states the ability to choose electors however they like.)

    • Matt

      Matt’s right, and not just because he has a right on name.

      The movement gained traction after 2000 in a number of blue states. What could happen is some red states join in and the movement gets to 270 EV and becomes the law of the land.

    • JamesInCA

      @Matt McIrvin – Why do you say “exploiting”?

      The Constitution doesn’t merely give state legislatures the “ability” to determine how a state’s electors are appointed; it requires them to do so, as it provides no other means by which those electors might be appointed.

      So why is exercising that Constitutional authority an “exploit” any more than, say, awarding all the state’s electors to the winner of the popular vote? Or awarding an elector to the popular vote winner in each congressional district?

    • Matt McIrvin

      I didn’t mean “exploit” negatively, more as appreciation of a clever move. I quite like the whole business.

      There’s a potential problem in that a really close popular-vote count could prove far harder to verify than a really close electoral-vote count, but making elections work better is a problem that we’re long overdue fixing anyway.

      Another potential problem that occurred to me is that nonparticipant states might try to sabotage the whole thing by keeping their vote counts secret, but that would have disincentives of its own.

    • Some Body

      James – a thought experiment, if you will. Imagine that following a Tea Party resurgence in Ohio in the low-turnout 2014 election, Ohio’s legislature passes a law that grants the state’s electors to whoever is the Republican party’s nominee for president (i.e., regardless of how Ohioans vote, the Republican always wins the state).

      Now, Dunce SCOTUS will have to hear the case on whether this new rule for determining the winner is constitutional, and, of course, four of the five Conservative Justices will say it is (states’ rights and all), but one of the five is nevertheless likely to join the other four in striking this provision down.

      But why? Well, because the new rule essentially disenfranchises the entire population of Ohio. The state’s vote in the EC no longer depends on how its residents actually voted.

      But then, you should also look at the analogy with the EC abolishment bill. That bill also disenfranchises the residents of the states that accept it, in that it ties the state’s electors not to the results of the election in the state, but to the *national* popular vote. You can of course argue that the national popular vote also includes the votes cast in the state. But hey, the results of the Republican primaries also reflect the will of Ohioans, *in*directly.

      So the constitutionality of the EC abolishment law is a bit dubious. If the US wants to scrap the EC (as it very well should), this has to go all the way through to a constitutional amendment.

    • Terry Doyle

      Well, if it’d make you feel any better we would never have had Bush II and maybe not 9/11 and certainly no Iraq war.

    • JamesInCA

      @Some Body – Thank you, that’s an interesting thought experiment, and a good plausible case against the national popular vote strategy. I do, however, take some issue with a couple of points.

      Firstly, I’m not sure the putative disenfranchisement is necessarily any worse than that which already occurs. In Nebraska and Maine, they award one elector to the winner of the vote in each congressional district, and then the two at-large electors to the statewide winner. You’ll recall that one of Obama’s electoral votes came from NE in 2008 through this mechanism.

      Now, I live in a conservative congressional district in CA. In comparison with Nebraska’s system, our winner-take-all mechanism will effectively “disenfranchise” my neighbors, who reliably vote Republican, by awarding all the electors, including “theirs,” to Obama. It’s not clear to me that our state legislature is constitutionally empowered to disenfranchise some voters this way, but not empowered to disenfranchise them by awarding electors based on the national vote.

      Secondly, I’m not sure the thought experiment you propose is actually unconstitutional. Egregiously wrong, perhaps, but not necessarily unconstitutional. As the Supreme Court reminded us in 2000, “The individual citizen has no federal constitutional right to vote for electors for the President of the United States unless and until the state legislature chooses a statewide election as the means to implement its power to appoint members of the Electoral College.” Egregiously wrong, yes, but that’s my reading of the Constitution as well.

    • Some Body

      James – I’d say that deciding the winner irrespective of the vote raises issues that have to do with the core of how democracies are supposed to work. In the end, the Supreme Court is supposed to interpret the constitution under the constraints posed by these core principles. Exactly how they might come about doing that, assuming they would, is a different question; perhaps Congress can help them out by extending the Voting Rights Act to all US states, and then they won’t need to invoke the constitution directly. But in the end, if a state can get away with a rule such as “the Republican gets the electors” then something is very deeply and essentially wrong with the US as (no longer) a democracy.

      Where the letter of the constitution might matter more is with comparing the EC Abolishment Bill to the individual district level situation you described. Here the point is that the US constitution leaves the choice of electors with the states, not the congressional districts, nor with the people as a whole. So, again, if you really want to get rid of the EC (and, again, I think it’s indeed better to get rid of it), there’s no way around changing the constitution to say that the people, not the states, elects the President.

    • Arbitol

      Some Body,

      Oddly enough, you do not have a constitutional right to vote for president. Article II, Section 1 states in relevant part that “Each State shall appoint, in such Manner as the Legislature thereof may direct, a Number of Electors, equal to the whole Number of Senators and Representatives to which the State may be entitled in the Congress.”

    • Beaucon

      I am struggling to understand how article 4 section 4 of the constitutional applies here. Doesn’t the national vote movement negate the guarantee of a republican form of government?

    • Some Body

      Arbitol – I don’t have such a right anyway (not a US citizen), but you (assuming that you are a citizen) have that right, whether or not it is explicitly stated in the constitution. It’s a basic feature of any democratic system of government that the executive branch is chosen by all citizens in an election, directly or indirectly, but you can’t just turn the general election into a beauty contest.

      But because the US constitution moves the responsibility for choosing electors to the states, the same argument has to apply at the state level too, which is, again, where the bill in question gets stuck.

      I guess we’re going a bit overboard with a purely theoretical argument here, though :-)

    • Arbitol


      The Supreme Court has decided (without much good reason) that the republican form of government clause was not intended to be enforced by courts.

    • Beaucon


      Wow, that’s very interesting. I don’t suppose this is not the place for a more in depth discussion on this topic, but I am very interested to learn more about that. Recently I heard an apparently well informed person discussing the historical connection between this National Vote movement the movement to reverse reconstruction after the Civil War. It leads me to believe that we should approach this topic with a sufficient level of caution.

    • Beaucon

      Opps, I will have to be sure to proof read better. It seems there is no edit function once the post is submitted. Should be “I don’t suppose this is the place…”

    • JamesInCA

      @Some Body – it’s a bit of beating an off-topic dead horse at this point, but note again my quote from Bush v. Gore, twelve short years ago: “The individual citizen has no federal constitutional right to vote for electors for the President of the United States….” We don’t need to theorize about whether the right exists, or how the Court might theoretically rule. The Constitution’s text is clear, and the Court has answered the question quite recently.

      As to the Supreme Court interpreting the Constitution in light of these diffuse core democratic principles, sure – but only after a plain reading of the text. If the text is clear, as it is in the case of how electors are chosen (specifically, as to who decides how electors are chosen), then we can’t reach for “core principles” to contradict the plain meaning of the text.

      This is why, for example, nobody has ever succeeded in nullifying the Second Amendment with the argument that “things are different now.” Sure, they’re different, but the words say what they say, until amended. Likewise, we didn’t get rid of slavery by ignoring the relevant constitutional provisions because they were wrong, even though that was obvious to many at the time. We excised it by amending the text. We instituted popular election of Senators by amending the text, not reaching for “core principles” in contradiction of the text. This rule of the law and the supremacy of what the text says is what makes our democracy safe and stable. Otherwise we could find principles to negate any part of it we found inconvenient at any time.

      Lastly, it isn’t undemocratic or un-republican (small-“r”) to leave the decision of how electors are appointed to the state legislatures. It is, in fact, republican by definition. And it is democratic in that, if some state legislature were to pass that law, they’d be summarily thrown out by the voters.

    • Karen

      “Though I’m pretty confident if the President wins the Electoral College and loses the Popular Vote, the likely Republican House will be introducing the “Electoral College Abolishment Bill” on Day 1 (pay no attention that the EC gave us Bush in 2000)”

      You are so right, Ohio voter!

      Repugs have amnesia about the Bushie years so it would stand to reason they’ll forget how he got in in the first place.

      And of course if that does happen they’ll also so Obama has no mandate ignoring that he did when he smoked McCain but they still obstructed from Day 1.

  • Dave Kliman


  • Ken

    “Prolonged bitterness” is a certainty no matter who wins or if the electoral college and popular vote go in opposite directions.

    • wheelers cat

      Prolonged bitterness is the outcome no matter what what.
      I’ve said this so many times I feel like a talking parrot.
      This election is really about demographics.
      2012 is the last presidential election that the GOP has a chance to win with an all white base.
      The increase of majority-minority markets, the continuing decline of non-hispanic caucasians as a percentage of the electorate, the increased power of social media and internet access all contribute to demographic evolution and a speedup in memetic evolution and transmission.

      Neither side will say this for different reasons, but presidential elections from now on are entirely dependent on liberal voter turnout.
      …unless the GOP can integrate by 2016.

    • skmind

      @wheelers cat

      “Prolonged bitterness is the outcome no matter what what.”

      I agree, with a small note. If Romney wins, the bitterness would be masked by the brazen hostility of the Republicans, and the caving in of the Democrats, with bitterness, on basically every issue. You can expect Reid and co. to agree to tax cuts for the wealthy, a war in Iran, and more digging for oil in Alaska.

      If Obama wins, then you’ll see the House impeach him no later than 2014. No, there won’t be any real charges, just the bitterness of failing to make Obama a one-termer. Also, no co-operation at all.

      Unless the Democrats, for once, do the right thing and change the Senate rules on day 1, to reform this one-man filibuster provision that hides the real extent of the gridlock.

      At least make Jim DeMint read the phone book.

    • wheelers cat

      since my background is game theory, i think what we are seeing is an iterated game set up by the founders. the tension between the red and the blue is actually a fitness enhancer, just like it was in the EEA (environment of evolutionary adaptation).
      Both genotypes were successful, and the tension between them made homosapiens sapiens successful too.
      its all about the diversity.

    • MarkS

      It’s a bit of a mystery to me why national elections are so close. Why should the nation be so evenly divided?? I wonder if there’s some sort of sociopolitical “fixed point” at equal division …

    • orchidmantis

      I agree that prolonged bitterness and muttering about how the result is invalid is inevitable no matter who wins how.

      In the Obama EV win, PV loss scenario: Bush governed like he had a mandate when it happened to him. Obama came in with a mandate but neither he nor the Republicans in Congress acted like it. So I think acting like one has a mandate, regardless of whatever actually happened at the polls, is the critical thing here.

    • JamesInCA

      @MarkS – It would be much more peculiar if national elections were routinely lopsided.

      If they were typically lopsided in favor of one side, you’d expect smart people on the other to find ways to encroach on the majority’s vote share. There would be no reason for a party to voluntarily adopt permanent rump-minority status. (Unless that party were the California state GOP, but I digress.)

      If they were randomly lopsided, that would suggest an electorate wildly changing its mind from one election to the next, which isn’t really consistent with observed behavior.

    • skmind

      @wheelers cat

      Interesting take.

      I am afraid though that the liberal gene pool is going extinct, what with the Democrats freely breeding with the RightWingNuts.

      Think about it this way. Six months before the DNC, 52% of the country was in favor of allowing gays to marry.

      The Democrats were apparently almost ready to make it part of their platform. 45% of them were in favor.

      Think about that, the general population favors a socially liberal position more than the Congressional Democrats, who have to “think about it”. Even Obama’s evolved position is that he is now willing to leave it to the states, but is personally “for it.” WTF?

      This was not that isolated. Remember the DADT repeal? 80% of the nation was in favor of it, and guess how long and how many Democrats it took to finally do something *that* popular.

      These Democrats are right of Reagan conservatives. Which is why the country is in a gridlock. They are too chicken to stand up. They do not have to stand tall, just stand up.

      Reform the filibuster such that you at least require more than the one objection it takes today to stall a bill.

    • wheelers cat

      MarkS, i have an idea.
      At the beginning of the election cycle a lot of us wondered why Obama wasnt farther ahead in national polling. I mean, +95% of blacks, 70% of hispanics, 55% of women….
      So here is my hypoth.
      The nat’l polls are only polling enthusiasm. They make a guesstimate about the non-responders. And I suspect enthusiasm is asymmetrical.

    • NC Dem

      orchidmantis: The idea that Obama and the Congress didn’t act as if they had a mandate seems to me ludicrous. One of the causes of right wing extreme bitterness is that Obama and Congress “forced” Obamacare and other legislation through using their majority, and circumventing the 60-vote nonsense with parliamentary maneuvering. They acted exactly like they had a mandate: “I won,” Obama said in explaining why he would do things his way.

    • wheelers cat

      no, assortative mating will prevent that.
      congressmen are motivated by survival. they want to retain their seats.
      They are not a true barometer of public sentiment, just of what they think they need to survive.
      Why do you think “conservatives” are always trying to “recapture” culture?
      Culture isnt a steering wheel even though republicans think it is. That is why they are always bitching about liberal media bias, and bitching that academe is painted blue and hollywood too.
      Culture is both a semi-permeable reflective surface, and the leading edge of the waveform.
      Society isnt shaped as much by culture as society shapes culture according to its needs.
      That is why I’m not worried about poll shapers refining their techniques down the road.
      Because we will have better sampling methods the next time, based on the outcome of this election.

    • orchidmantis

      As Dan Savage has pointed out, the 52% who favor same sex marriage are not evenly distributed across the country. They’re concentrated in blue states. Making Dems up and down the ticket defend same sex marriage in <50% purple states and districts is a counterproductive grand gesture when your goal should be to elect people solid on gay rights (or better than the alternative on gay rights and other issues) and let those steep trend lines on same sex marriage have a few more years to inch all the states further over.

    • Some Body

      Cat – I think the exactly split electorate favours a game-theory explanation (political parties, in a two-party system, over time, tend to split the votes equally, even as the population’s views change, because they shift the line of demarcation between the two parties accordingly).

      A genetic explanation, though, would have a problem here. Because with all the differences that there are between Liberals and Conservatives (as they’re known in the US; elsewhere in the world “Liberals” are the right-wingers, and “Socialists” are on the left, and don’t feel that label is a term of abuse), also in terms of how their brains work, the right/left divide passes in very different places in different countries and cultures, and periods too. What’s considered far-right in Norway is well left of centre in Israel, but both countries are relatively evenly split (well, maybe not Israel, but in Israel you have an overblown right wing, not left wing, especially among ethnic Jews, those same ethnic Jews who are overwhelmingly on the left practically everywhere else in the world…)

      Also, a genetic explanation has a tough nut to crack when it comes to the gender gap in the US. The difference in political preferences between women and men is too small to be explainable by the Y chromosome (otherwise you’d see practically all men voting for one party and practically all women voting for the other), but then it is also too large to fit with an hypothesis of genetically-determined political preferences.

      Sophistry aside (we still know little enough about genes and how they work for any statement of the form “P is determined by people’s genes” to be impossible to either conclusively confirm or conclusively refute by empirical evidence), your different claims in this very comment section to seem to be in some logical tension.

    • Some Body

      Quick self correction: of course, not *any* statement of the form “N is determined by the genes” is inconclusive. One’s eye colour is determined by the genes, and not much to doubt there; one’s native language is not (sorry, Noam Chomsky), and that’s very clear too. But when we come to Ns that involve complex behaviours, and socially, ethically, or politically determined definitions – then I’d still stick to what I said before.

    • Some Body

      No, Cat, he doesn’t even try to argue that people’s native tongue is determined by their genes. That’s an absurd claim.

      That there is some genetic basis to the human ability to use language is self-evident. That genetic factors affect specific trends in the development of languages is debatable. That genetic factors affect features highlighted in linguistic analysis, such as syntax, used to be the orthodoxy in some circles, but is highly dubious. That genetic factors (rather than environmental ones) actually decide which of the world’s languages will be the first one you master is evidently false.

  • Mike Chapman

    Great coverage of the polling wars. I appreciate the sanity you present here.

  • ChrisD


  • Slightly Skeptical

    Sam, I was wondering if there already is or maybe you could add a way to see the winning probabilities if the election were today. In other words, the numbers corresponding to the chart marked “All possible outcomes” which is visible on the right, summing up the percentages for a Romney or an Obama win.

    I’m sure It would be generally appreciated by your readership. I would appreciate it even more, because I’m skeptical about the way your prior is based on the state of the race in the summer and early fall (I don’t think O+3% is a reasonable baseline at this point, though maybe my thinking on this is flawed). So such numbers would be more significant for me than the predictions you have for election day.


    • MarkS

      ???? Sam uses a flat prior, using only polls within the past week (more or less). Perhaps you’re thinking of Drew Linzer at

      You can make a rough estimate of today’s win probability by eyeballing the 95% confidence interval (gray region).

    • john b

      or you could read the FAQ

    • Slightly Skeptical

      MarkS: I don’t think that’s right (regarding the prediction for election day), at least based on my understanding of this explanation . As for the gray zone, thanks, though I was aware of that already.

      John: Thanks. I guess Sam has already decided against it, so we can use the histogram data. FYI, I’ve read the FAQ several times before, but I don’t have perfect recall. Thanks, anyway.

  • Steve in Colorado

    There was an interesting piece on Huffpost saying something similar, that Romney campaign is puffing itself up to get media play, but it’s just a bluff. But if it works, and they win on Nov 6, it will all have been validated…

    • orchidmantis

      Chait is saying the same thing at NYMag. Citing Rove’s idea that late deciders break for the person they think is winning. (The huge fanfare around moving one entire staff person out of NC is a strong indication that feeding the media Mittmentum stories is an active tactic.)

  • Olav Grinde

    Let’s say, just for the sake of argument, that Governor Mitt Romney should win Ohio (getting a return on Tagg Romney’s investment in voting machines, help from the Republican secretary-of-state, or receiving divine intervention).
    Say he also takes Virginia and holds every “red” state, including Florida…

    But that President Obama wins New Hampshire…

    The result?

    Obama 272
    Romney 268

    • Ohio Voter

      I imagine Ohio, New Hampshire, Iowa, Nevada, and Colorado will be seeing a LOT of the President over the next two weeks.

    • Olav Grinde

      I would really really like to see reliable numbers on voter roll purges for Colorado, Nevada, Ohio and Florida. As well as “accidents” such as giving Latinos the wrong voting date.

      Don’t be surprised if there is planted evidence of Democratic fraud.

      And don’t be shocked if this election ends up being decided by the US Supreme Court, or as a cumulative effect of lower-court decisions.

    • Slightly Skeptical

      Olav: Don’t assume Obama wins IA and CO. Especially if you’re doing a hypothetical about Romney winning OH. It’s hard to imagine him winning OH and not winning at least one out of IA, CO, NH. Given the polling in the last month or so, it’s far more likely the other way around: he might well win one or two of them but lose Ohio.

    • ChrisDC

      Romney has never led in Ohio (in poll averages, not individual, one-off polls). He has, at most, pulled within a statistical tie at his highest point, now receding back to a 1-2 point deficit. Ohio’s “natural state” in this election seems to be a narrow but decisive Obama win, and there’s nothing short of a bizarre and unexpected shock that could change that dynamic now, with 14 days to go.

    • JamesInCA

      @Slightly Skeptical – Agreed. I have the same objection to scenarios that start along these lines: “Suppose Romney wins Florida, Virginia, and Wisconsin….”

      If Romney wins Wisconsin, most of the rest of the scenario is moot, because he probably will have picked up other middling and most right-of-center states.

    • Olav Grinde

      Slightly Skeptical, please note the possible reasons I mentioned for Romney winning Ohio – none of them due to a majority of the votes actually cast.

      Such intervention in Ohio — by Karl Rove & Co, or Tagg Romney’s helpers — might not be applied in equal measure in the other states you mentioned.

    • Slightly Skeptical

      Sorry, I thought you were at least half-kidding about those. If divine intervention gives Ohio to Romney, what reason is there to think it’d stop at that?

    • Olav Grinde

      Basically I am optimistic that the voting and counting process mostly has integrity. By that I mean the cheating is limited to producing small swings. This factor is more likely to be decisive in closely fought battles.

      Ohio’s secretary-of-state has already indicated that the integrity of the voting process is not his highest priority. Tagg Romney’s involvement on the ownership side of voting machines is well documented, not conspiracy myth. Hence I think Ohio is a top priority.

      But, yes, I don’t think it stops there. Some powerful people have invested a lot of money in the outcome of this election, they want a return on their investment.

      Just like they got a return on their investment when injecting money in state-level and House races in 2010. Do you think it’s coincidence that swing states were so high on their priority list? I certainly do not.

      It would, for instance be very interesting to see reliable tallies of voter roll purges for swing states. I have searched, but cannot find them.

      Perhaps someone here can help with those numbers?

  • dwgsp

    While one option would be to abolish the electoral college, another possible outcome would be more movement at the state level to abolish the winner take all method of allocating electoral votes.

    • Olav Grinde

      Southern states first, would be safest.

    • 538 Refugee

      I don’t think the political parties really want to do away with the electoral college. Absolutely no interest in it after Bush/Gore. The parties don’t want to have to run national campaigns.

    • Matt McIrvin

      Adopted everywhere, that would have the net effect of a huge bias toward Republicans, since it would enfranchise rural R-voting areas in blue and purple states without removing the overrepresentation of red states in the EC. As it is, things are pretty much a wash.

      But it’d be unlikely to get much traction at the state level, because in any individual red state, it’d have the effect of an advantage for Democrats. Another collective-action problem.

  • skmind

    “After that and Debate #2, Obama made some recovery. Now we are at a plateau, in which Obama is slightly – but decisively – ahead.”

    This is the only part I do not fully understand. Obama did make some recovery, and from either chart it appears that following that recovery, he lost half of what he recovered.

    The plateau, whatever it is has been, has been for a day. Two tops.

    He may be decisively ahead, but I do not see that plateau being large enough. Certainly not as large as the drop following his recovery

    • orchidmantis

      I’m not sure what the point of worrying about the size of the plateau is, rather than focusing on the fact that it leveled out? Being 6 or 10 pts ahead is nicer, of course, but not likely 2 weeks before an election in a down economy. Being consistently ahead of an opponent is good. Being consistently ahead of an opponent who is not rising in the polls is even better. Relying on Obama not to do anything first-debatesque for two weeks is not a huge bar.

    • Tractarian

      @skmind STATISTICAL NOISE.

  • ChrisDC

    Doubts about the Romney campaign’s loud bluffing about “momentum” seem to be gaining a bit of salience in the media and among election prognosticators. It’s something denizens of this site and 538 have known for some time, but now journalists are beginning to pick up on it.

    Per usual, the insular Beltway press corps. – cocktail circuit chatterers like Mark Halperin and the Politico boys – will be the last to realize this, but I think it will gain mainstream credence over the next few days.

    How pundits on TV can continue to assert as fact that Romney has “momentum” (whatever the hell that means) is beyond me. Not beyond me is why. So I suppose that answers the question “how” – where there’s a will (and a lack of critical faculties), there’s a way.

  • don in fl

    how about obama wins wi. va. thats 260 then work from there.cant believe obama loses va.fl. maybe but not va.

  • NC Obama Guy

    I just read an interesting article on Huff Po for anyone who would like to read. It kind of says what I have thought and mentioned a few times as have others on this site in the last few days.

    • skmind

      Man, that’s a long (but interesting) read.

      This is a well concealed nugget from the piece:

      “Well-informed poll-watchers in the US, such as Nate Silver (New York Times) and Mark Blumenthal (Huffington Post) say response rates to US telephone polls are now below 10%. So pollsters fail to reach more than nine out of ten Americans they try to contact. It takes only a tiny change in relative response rates to have a large impact on the results.”

      10% is scary low. For pollsters. Throws a lot more uncertainty in their numbers, and if they are wrong, and separated from the pack, well… they can always hook up with FOX “News” and join Rasmussen and Luntz, I guess.

    • Matt McIrvin

      His defense of party weighting is interesting, and definitely not a majority opinion. Part of the problem with it is that you can get huge differences in absolute results depending on what you assume the proper party weighting is, which is basically where Rasmussen’s famous Republican house effect comes from.

      It also tends to damp out fluctuations. I guess Keller’s thesis is that those fluctuations are illegitimate and come from variations in response enthusiasm (if those correspond to variations in voting enthusiasm, though, maybe they’re not a bad thing to let affect a poll).

      The usual claim is that R vs. D fluctuates a lot with voting preferences, so if you weight for it you’re just artificially damping out real changes.

    • Some Body

      @Matt. I noted this too, but I also noted the crucial difference between how they are applying party ID weighting in the UK and how Rasmussen and others do it in the US. In the UK they don’t ask people how they identify themselves right now. Rather, they ask people how they have voted in a given recent election in the past. Now, *that* is something on which there is solid demographic data, and which does not change (assuming people don’t lie and don’t make things up too much) with the news.

      Their US methodology is also different (and close to the RAND one): they are recontacting the same people and doing a longitudinal study, in which case party identification can actually be compared with how people identified last time around. This again bypasses the problem of assigning party ID by fiat to be what you think/hope it would be (which is what Rasmussen does and the poll-unskewers demand even more of).

  • mnpundit

    Plateau? The graph itself looks more like Obama lost about half the ground he’d regained after the nadir with the margin trending downward once again. You can’t really call it momentum for Romney since he’s still behind however.

  • Jeff in CA

    Given that MSM can only have one narrative going at a time, I imagine it’ll be a week of Romney momentum, then a week of Obama momentum, and at the last second, Romney wins.

    Or not. :}

  • E L

    If James Follow’s Iron Law of the Media: “The story must change” is valid, then the RoMo story should show up in about three days and the piling on begin about a week before the election. Then… all the WaPo and Politico types can preen their predictive plumage while cooing about being geniuses when the election results come in.

  • Les Honig

    I went to bed here on the West Coast with the consensus saying Obama had won the debate, with Sam’s ratings holding pretty firm and with Nate Silver bumping up his O prediction to 70%. In Trade had raised his prob of winning up to 61%. I checked today’s polls just a while ago and they seemed to produce roughly the same mixed results; yet I checked In Trade and the odds are suddenly down to 57%. Anything happen poll wise or with post debate sentiment that can account for this? I am confused.

  • Judson

    What’s the probability that no state is more than two state borders away from RED or BLUE?
    (ignore AK, HI)

    Or how about the probability that every state is bordered by at least one RED and BLUE state? (oops this would be 0% if you count Maine..) but you get the idea..

    • JamesInCA

      Curious about the reasoning behind the question. You could calculate that figure by adding bordering-state information to the database and counting up the results, but I’m not sure what you’d know afterwards.

      If you’re wondering whether blue and red are clumping geographically, the map will answer it visually.

  • Reason

    So is it sarcasm or not? Some of you here seem to know Dr. Wang’s sense of humor more than a noob like myself.

  • John

    How do we –or do we?–account for the modest, but potentially significant given the tightness of some state races, impact of third-party candidates? They’re not included in most state polls, so understandably not included in Dr. Wang’s analysis. What I can’t get my head around here, though, is that it’s actual votes at play, rather than, say, vague econometric measures that may in fact be trailing indicators, leading to a sort of double-counting in other models. I’m coming at this as a finance guy, so I’m doubly sensitive to the leading-vs-trailing indicator concept.

    Gary Johnson is really the only one that could possibly matter here, and even if he only gets 1-2% of the vote (though it wouldn’t surprise me if he got a shade more than that), in some states that may be enough to tilt the margin given that most of his support will come out of Romney’s potential base, which may go some distance towards explaining why in several states, such as Ohio, the GOP has tried to get him off the ballot.

    Any thoughts? Anyone? Anyone?

    • Olav Grinde

      Virgil Goode may matter in Virginia, if the outcome is really close.

    • Reason

      I believe Goode and Gary Johnson is on the ballot here in Va too.

    • JamesInCA

      You’d have to go state-by-state to see where a third-party candidate could garner more votes than the difference between the major candidates. And then you’d have to be pretty sure you knew who that third-party candidate was drawing more votes from. To my mind, that’s not at all obvious with Gary Johnson, a Republican (when elected) governor of a blue state who is a champion of marijuana legalization.

      As Olav suggests, that calculus is easier with Goode, but the question of whether he draws enough votes to affect the outcome is unclear.

  • tzx4

    ” I regard this as a serious risk, since it would engender prolonged bitterness.”
    Yeah right?
    We have already experienced nearly four years of prolonged bitterness.

  • Alan Cobo-Lewis

    Hey, who ya callin a “true nerd”?

    I resemble that remark.

  • A New Jersey Farmer

    Hello all. Nice to see that the conspiracies are not just for the right. Some thoughts:

    Kerry was ahead slightly going into the last week, but remember that bin Laden released a tape on the Sunday before the election and both campaigns noted that the numbers moved sharply back towards Bush. That was enough for him to win Ohio.

    Another way to look at the state of the race is that Romney has never rebounded from the ad blitz in the summer that defined him as unacceptable, the negative convention bounce and the 47% comment. He’s come close, but after the month that Obama’s had starting with the debate, he should be ahead. Why isn’t he? Because the economy has improved just enough and Romney is still seen by many independents as a rich white guy who talks good but will take your money and call it even.

    We need to keep our eyes on the fundamentals. Romney has never led in Ohio and he won’t because the Obama ground game has more resources there. They knew that Mitt was holding out money for the last two weeks, so they moved heaven and earth to get their people to the polls before today. That plus GM will win Obama the state.

    Nevada will stay with Obama because of Reid and right now Obama leads more there than Mitt does in FL. Iowa will stay with Obama because it’s been trending D for the past couple of cycles and again, Obama leads by more there than Mitt does in CO.

    That’s 271. Everything else is gravy.

    You may feel free to make life decisions now.

    • ChrisDC


    • Amitabh Lath

      Hello NJ Farmer, yes I remember the Bin Laden tape, timed perfectly to keep in office the best recruitment tool he ever had. Somehow I doubt there will be a repeat this year.

      Even without the tape, I don’t think there was any hope. Kerry never ever felt plausible. According to Sam’s calculation he was behind since late August.

      Where is your farm? Would love to bring the kids.

    • Mark F.

      I anticipate Obama will have a good ground game in Ohio. However, GOP turnout in Ohio was miserable in 2008 (way below the levels of 2004) , and will likely be much better next month as every Republican I know is foaming at the mouth to vote against Obama. Obama can’t control good GOP turnout and that might be enough to tip the state towards Gov. Romney.

    • A New Jersey Farmer

      The Farmer thing is just a moniker. I have some tomato plants in the summer, though.

  • Steven J. Wangsness

    Does anyone know how RCP calculates its “averages”? Tske New Hampshire, for exaample. They have a +1.4 Obama average lead, even though the list of polls is mostly a sea of blue Obama leads ranging up to +15 and only a couple of +R entries, the highest of which is +4. I don’t get it.

  • Scott Mulder

    There is a serious black swan in this statistical equation: the passing of severe voter control laws in several swing-states, and emerging accusations of manipulation and fraud, particularly in the states of Arizona and Virginia. These factors cannot be accounted for in a statistical program, and will almost certainly weigh in on election night.

    • Michael

      Hasn’t Arizona been polling solidly R throughout this campaign? Why should anyone be overly worried about voter suppression there? Virginia has a new voter ID law, but it isn’t particularly onerous — they mail a voter card to every registered voter, and that card is sufficient to vote. Many other forms of ID are also acceptable — doesn’t have to be government issued. And finally it’s a fairly minor change — Virginia voters have been asked to show ID for many years, but in the past you could vote without one if you signed a sworn statement that you were who you claimed to be. And you can still vote that way, but it will be a provisional ballot and you’ll have to provide the ID later to have it counted.

    • ChrisD

      AZ has been polled just twice this month, the last showing O+2:

      VA’s voter ID was reviewed by DOJ and approved. It’s actually workable and fair. Too bad FL, OH, and PA didn’t come up with something similar.

    • Matt McIrvin

      There have been several cases of outright fraud in Virginia, though, things like people holding registration campaigns and then throwing the registrations in the trash. Their horror of a state AG refuses to even pursue any claims of party involvement, the argument being that since people don’t register with parties in Virginia they can’t possibly have partisan motivation.

    • Michael

      There have been several cases of outright fraud in Virginia, though, things like people holding registration campaigns and then throwing the registrations in the trash. Their horror of a state AG refuses to even pursue any claims of party involvement…”

      The horror is investigating now:–election.html

  • Obama 2012

    I would love to buy Obama shares at $5.70 on InTrade but it’s a pain to do so… that’s a good thing really, I do think it should be illegal as I agree with Mr. Wang said regarding betting on politics – but man… it’s tempting to make money off of right wing hubris.

  • Elizabeth Duvert

    Anyone concerned about Romney’s family connection to the company that manufactures the voting machines used in parts of Ohio and contractors who count votes?

    Any role here for OSCE?

  • Amitabh Lath

    Several posts ago I mentioned that R-moneybags could easily manipulate Intrade. The Iowa Electronic Markets are less manipulated and currently they show 62 Obama while Intrade has 57.

    If the plutocrats manage to push it down to 55 then I would be sorely tempted to go in and take their money.

  • Analytical

    I am begnning to think the polls are all over place. There are a couple of nice polls (YouGov and RAND) that seem to provide more realistic picture of the race. Please RAND poll data below
    Date, Obama, Romney, Difference
    3-Oct, 49.17, 44.83, 4.34

    11-Oct, 48.17, 46.15, 2.02

    16-Oct, 49.07, 45.16, 3.91

    22-Oct, 48.16, 46.36, 1.8

    The first debate probably cost Obama about 1% while Romney gained 1.53%. Since October 10, there are only minor fluctuations more because of minor shifts between candidates than anything else. Please see the analysis by You Gov President Peter Kellner.

  • David

    An additional potential indicator that there is “Ro-mentum” – note that the trendline reversed the day before the debate. This is a large group that Rand has followed day by day

  • don in fl.

    be good to see MM rise to 2or better.make a few of us less nervous i suspect.and stay at 2 or better.

  • L. Murray

    Thank you to Dr. Wang and all of the math and political nerds on this site who have helped to keep my blood pressure and consumption of Valium at manageable levels.

  • Terry Doyle

    So we have a 25% chance of increased whining for the next four years? I can’t imagine the cacophony of Republican sniveling to come. What would be interesting to know is whether this trend toward chance of whining has increased in the last couple weeks or not.

  • Ron Gee

    Now we have the added uncertainty of unprecedented magnitude of voter suppression and fraud perpetrated by the GOP (!). Various individuals have been caught but GOP-run states refuse to prosecute these crimes.
    Wonder what that does to the manipulated vote counts in the end.

  • Bill

    In 2004 based on your analysis, I was convinced Kerry would win until about 8:30 pm CST. There may have been a last tilt towards Bush, but it was unexpected. Will the same thing happen again?
    PS: I supported Kerry because of disgust with Bush, but that is not my natural place. Maybe this effect will shift other prediction models?
    Navy (bayonet and horse) vet

    • Sam Wang

      Bill – I feel your pain, but please, please read the documentation. It is so tiring to be asked this again and again.

      The answer is no. That year, I added an additional assumption about turnout/undecideds that was wrong. The current approach is polls-only. Thus it shall ever be at PEC.

    • Matt McIrvin

      Looking at the trends this year, it occurs to me that the conventional wisdom about undecideds breaking for the challenger is probably actually true–except that in a modern presidential election it happens well before election eve, and you can see it happening in real time. It’s not something that suddenly happens when people walk into the voting booth.

      In previous cycles this may have been missed just because there wasn’t enough polling going on. 2004 was really the first year that we had something like the wealth of information we have today, probably as a response to the debacle of 2000.

  • JohnJacobs

    Here’s your chance Amitabh. What’s going on Intrade currently is either blatant manipulation or Donald Trump’s attempt at blatant manipulation

    • wheelers cat

      nah. Intrade is just more sensitive to asymmetrical enthusiasm.
      Look what happened with Roberts and the ACA.

  • Intradeisweird

    Intrade has the republican candidate winning Ohio at 56.9%….a ton of movement today. Am I missing something or did something big happen in Ohio to affect enthusiasm

  • MAT


    Wow, that was an interesting article, which has challenged me to think thru some of my prior assumptions on weighting. Highly recommended.

  • Andrew

    Ro-mentum, baby! Yeah!

  • gina ligouri

    Why has no one in the news media, TV or newspapers, picked up on the fact that Romeny’s son now owns the company making the machines in the Ohio election? This is Diebold and BUSH again. Why isn’t this being reported except on facebook and other Internet sites? Thanks….please find a way to show this. Thanks.

    • Elizabeth Duvert

      I brought this up earlier on this thread, but no one has commented. OK. Does that mean it’s not important?

    • BruceMcF

      One point is that the “ownership” is not precisely a direct controlling interest, but the bigger point is that its just two counties. Swinging a 2% to 5% Obama win in Ohio to a Romney win, based just on Cincinnati and a rural county, without triggering a recount which would reveal differences between machine tallies counted and the results of the printed paper tape, would be quite difficult.

      And winning Ohio doesn’t land Romney the White House, it just brings him into position to win the White House. The more states he would try to steal, the greater the chance of the attempt falling apart.

      That’s why they are trying to steal the race the old fashioned way, with voter suppression.

  • Ms. Jay Sheckley

    TRUMP’S BORING ANNOUNCEMENT The supposed game changer is Donald Trump will give $5 million to Obama’s charity of choice if Obama releases his college and passport applications and records. ::Yawn:: Meh. So what. Funny-looking idiot.

  • Amin

    OH is now O +2.1, from the current O+1. This should mean we will see an uptick in the MM once it is updated at 5PM!

  • Steve Schlichtenmyer

    This is only slightly off topic and speaks to the issue of Romney’s “momentum.” Regarding the Median Electoral vote estimator, there is clearly a downturn for Obama after the first debate, yet in the graph of the aggregated swing state polls on (linked to this site) this drop appears to begin to occur during the prior week. Am I missing something? Does anyone else see the same thing? There must be a simple explanation.

Leave a Comment