Princeton Election Consortium

A first draft of electoral history. Since 2004

For the junkies

October 18th, 2012, 6:00pm by Sam Wang


I smell change in the air. Also, polls are much more frequent. For example, the Meta-Margin has been jumping around today, with a bit of a trend toward Obama. So…Andrew F. has added two new daily updates to the Meta-Analysis. The schedule is now: 8AM, 10AM, noon, 3PM, 5PM, 8PM.

Neuroscience note: Rewards are most effective at shaping behavior if they are given intermittently (which maintains their unexpected nature). For example, a rat will learn to press a lever repeatedly if something tasty comes out sometimes rather than always.

You may now reload your browser.

Tags: 2012 Election · President

109 Comments so far ↓

  • Ms. Jay Sheckley

    BF Skinner- Sorry, right, _reinforcers_: Like Facebook friends who agree with with one’s most arcane suspicions, not to mention the cure for cancer is being kept secret. My friends all agree! Damn you, scheming scientific journals! Some day we will take over Congress… Wait.

    • BFSkinner

      Skinner’s radical behaviorism ain’t Watson’s methodological behaviorism… Http://www.ejoba.org/PDF/2001_2/Moore_2001.pdf.

      But yes, “truth by agreement” is pretty silly.

  • BFSkinner

    Reinforcers, not rewards!

  • Dinsdale Piranha

    Hey Sam, love the site, and especially love the more frequent updates on the meta-margin. But I’d be surprised to hear that I was the only one wondering what prompted each fluctuation. Do you think you or Andrew might be able to give a quick blurb of justification, or at least give us a dump of all the date that *might* have caused each change?

    • Sam Wang

      If you mean which individual polls…gosh, no. That’s opposite to why I set up the calculation! However, we might set up a data feed so you can see recent polls.

      Generally, any uptick today is driven largely by post-debate#2 data.

    • Dinsdale Piranha

      Sam – I hear you, and I also hear you about ignoring the disparity and noise in all the polls out there, and that’s why I and so many others have become so dependent on the meta-margin. It’s just that the dependence and need for frequent updates comes from a need for understanding and sense, so anything you can do to help with that would be icing on the cake.

    • Ms. Jay Sheckley

      Sam- Post debate 2 data coming today? I though you said we were finally seeing ApolJo bounce

      Cant wait to hear you on NPR!

  • Ralph Reinhold

    I think this volatility in the polls is related to a rule of thumb I was told many years ago (1972). I was given a reference at the time, but I don’t remember what it was. The rule is that when the respondents to your poll fall below 10%, your numbers are no longer reliable.

    Wikipedia cites Bradburn and Sudman (Asking Questions) as saying that the non-response error does not affect the margin of error, but the citation does not provide an original source for this conjecture.

    I tried making the assumption that the ‘real’ set of respondents was related to being plus or minus the square root of the total sample set, but this grew large enough to be declared a problem long before the sample set was below 10%.

    Pew recognizes that non-response is a problem but does not provide quantification of the error. http://www.people-press.org/2012/05/15/assessing-the-representativeness-of-public-opinion-surveys/

    • Olav Grinde

      How would you go about quantifying that error? Even theoretically… I am curious.

  • NC Obama Guy

    Have you all seen RAND today? It just made a sharp move towards Romney. Can anyone explain that???

  • pechmerle

    Olav, a Republican House can decline to fund some of Obamacare. They can do that right now. Neither Obama nor Romney could force the House to appropriate funds the House refused to appropriate.

    But, critically, there are large elements of Obamacare that don’t depend on federal budgeting. For example, the new rule that a young person can stay on their parent’s health coverage until the age of 26 isn’t a budgeting matter. So, no House funding action would not affect that. Similarly, the requirement that health insurers cannot turn anyone away because of preexisting health conditions isn’t a budgeting matter. Again, House funding action would not affect that.

    So funding for Obamacare is important. But neither Romney nor a Republican House can simply undo it via use of the budget. This requires a Democratic Senate, of course. If the Republicans control both houses of Congress and the Presidency (and leaving the filibuster rule aside for the moment), then of course they could undo all of Obamacare.

    • Olav Grinde

      Pechmerle, thanks for your clarification!
      I would be really interested in seeing a more detailed breakdown if one exists, so as to have a more objective understanding of the impact of de-funding.

  • bsk

    Firstly, can anyone explain to me why the meta-margin isn’t at exactly Obama +2% right now? As far as I can tell, he’s up at least 2% in states worth over 270 EV, according to the “power of your vote” bar.

    Second, how did the EV estimator decline from its position at 290 yesterday? No states seem to have flipped since then. I can understand a change in the confidence bands, but not the point estimate.

    Third, why was Ohio still at 2% yesterday evening (18/10/12)? Taking all polls on Pollster with sample end dates from the 11th onwards gives Obama +3.5%. Does Sam use the date of publication rather than the sample dates?

    As someone who follows this site regularly but is getting a little frustrated at these apparent anomalies (which are probably just me failing to understand the methodology), any answers would be much appreciated.

    Thanks,
    bsk

    • Richard Vance

      Don’t quote me but I think Sam uses a moving median type of thing so that it takes a bit of heavy movement to gel into the median EV.

    • badni

      Regarding your second question: a state doesnt need to flip to have this effect. If a state drops from a 80% likelihood of an Obama win down to a 60 percent likelihood then when you calculate the chances of every possible electoral vote outcome, Obama will come out with a lower average ( and, more importantly, median) number of electoral votes.

    • bsk

      badni: So it’s the median outcome of the simulation, not the sum of the individual state medians – that makes sense.

      Any ideas about the others? Anyone? Should I re-ask this question in the most recent thread?

  • Eric Walker

    Let us not forget that either party could do away completely with the filibuster if that party has a simple majority of votes (and can command them all) and acts within the first two weeks of the new Senate session.

    If the Democrats have 50 votes solid, with or without the “independents”, they can change the rules and eliminate filibusters. (Ditto for the Republicans, but despite their historical objections, it is doubtful they would do so now.)

  • IBP

    Uconn national poll, obama 48-45, 55-38 among women. Gender gap live and well for O. Pre debate 2, and if anyone lost women votes after that debate it surely would be Romney.

    http://www.courant.com/news/politics/hc-president-poll-uconn-courant-20121018,0,6416855.story

    • Olav Grinde

      That’s very interesting, because elsewhere I seem to recall claims that Romney was closing the gender gap.

      I’m looking forward to more state-wide polls. And I would not be surprised if new national polls show that much-quoted recent Gallup poll to be quite an outlier!

  • Tony

    Romney cannot win if Obama holds on on Ohio, Wisconsin, and Iowa.

    As of today, Romney is losing to the president to all three States.

    The election is up to voter turn ups!

    If Obama gets his base in voter, he will definitely win..

    If that awful Romney gets the vote out, he could win….

    I am not freaking out as a democrat. Bit the the so call “Biased medis
    is not making the point that Obama is true to polling….

    My opinion on this?

    It is all about money….even pro Obama polling is going to put the vote to Romney…..

    And that is lies.

    So… demo, vote early!

  • Tony

    Senate filibuster can kill Romney overturning obamacare because is a supreme court decision.

    And I believe they can’t take away medicare or SS or mortgage interest deduction.

    A win on the top of the ticket and both houses may be able to destroy the student loan program.

    Without the student loan program, our children cannot pay for the ridiculous price of a college tuition…especially the Romney position is to cut education, thus increasing the university prices significantly higher.

    • Olav Grinde

      @Tony, I have a question: Is there a possibility that Romney can simply de-fund the Affordable Health Care Act?

      If so, this admittedly imperfect legislation would be a pyrrhic victory; the public might not ever see more of the positive effects of so-called Obamacare.

    • Richard Vance

      Obamacare was passed as a budget item did not need 60 votes. Thus it can be overturned the same way.

    • Olav Grinde

      Richard, are you saying that a simple Republican majority in the Senate (and the House) would do the trick?

  • Matt

    I’m confused about these numbers. How do you get 0.5% Obama in CO:

    10/16 – 10/18 Obama +3
    10/15 – 10/16 Obama +3
    10/15 – 10/15 Romney +1
    10/5 – 10/11 Obama +2
    10/4 – 10/11 Obama +3
    10/9 – 10/10 Romney +1

    Everything else is clearly more than a week old

    Other states like WI, OH and IA don’t make much sense to me either. What am I missing?

    • ChrisD

      Just speculating here. The latest poll in the list (PPP) is labeled NEW by Pollster. Perhaps the prior six polls make up the group that was averaged. That includes another R+1 (Qinnipiac). Throwing out the two O+3′s and the two R+1′s leaves an O+2 and an R+1, whose average is O+0.5.

    • Matt

      I think you are correct that it is using older numbers to retain a 0.5% in CO. I guess I’m not understanding when and how the margins are updated. Is the EV and Meta Margin using the latest numbers, while the power of your vote and state maps lag?

  • Reason

    Except the slide ended and the latest rash of polls show a slight uptick to O. As for him losing college women, I have no idea where you are getting that from.

  • Spiny Norman

    Ram, almost none of what you say here makes sense. And the phrase “very profound” is “very stupid.”

  • Ram

    The polls don’t smell good for O. This was a very stable race till first debate with MM of +3 for Obama. The loss of MM from +6% to +1 to 1.5 % is very profound. The shift seems to be because of independents and college educated women.

    If the MM doesn’t revert to say around +3% by this weekend, the shift is probably permanent. There is a very strong possibility R can prevail with up to 295 EVs if the loss college educated women from Obama’s coalition is deeper than what it is now. The polling numbers from rush belt indicate structural shift in voting coalitions not because of local factors (Wisconsin-Ryan, auto bailout-MI/OH etc). The campaign schedules indicate current battlegrounds are confined to just NV, CO, OH, IA and NH.

    I also don’t buy the theory of swing states and national polling need not converge. Obama’s slide has been national and will eventually reach battleground states.

    • Ms. Jay Sheckley

      Ram, I’m here because there’s a measurable likelihood that Sam is onto something. See the 2008 site in the upper left margin. From what I’m learning here, a Romney win is a definite possibilty, but except by you, I havent seen it called a “strong possibility”. In the same way, many of your statement have an unusual twist. I dont know where you get your figures from, concerning what this weekend’s MM means. If the figures are off the top of your head, it would make more sense to ask questions than make declarations.

    • Olav Grinde

      Ram: “I also don’t buy the theory of swing states and national polling need not converge. Obama’s slide has been national and will eventually reach battleground states.”

      I think Ms Jay Sheckley (and Sam) are onto something.

      But you have a point. In fact, in the next week-and-a-half, I think we will see the national and state-level polls converge — and to Obama’s advantage.

      Yes, the first debate was catastrophic for him.

      However, almost just as many people watched the second debate, where Obama clearly bettered Romney. More importantly, Barack Obama appeared presidential and totally in command.

      And regarding women — Romney blew it. He told an extremely awkward story about “binders of women” that has become a national meme. (Check out http://bindersfullofwomen.com/) Worse: Romney’s story was proven a lie; this did not happen at his initiative.

      Romney’s other debate answers on women’s issues and family issues are likely to neutralize his recent gains in this demographic. Heck, the former Governor of Massachusetts can’t even say he supports the Lilly Ledbetter Fair Pay Act — and his aides keep contradicting each other!

      Ram, let’s agree on one thing: let us let the polls speak for themselves. Their story will soon become more clear.

    • Richard Vance

      I agree. In the first debate Romney was near death going in and Obama resuscitated Romney giving him a Presidential look, RR got their foot into the door that should have been slammed shut.

      Now as a good door to door Mormon Romney used that to get into the kitchen to talk to the Missus.. It was a serious and permanent damage that was done.

    • Ms. Jay Sheckley

      Before discussing what Richard said, there’s an old Buddhist tale I like. A farmer is upset because his only son has broken his leg. “I’ll never get in the harvest” he tells an old monk, “This is terrible!” The monk says, “Maybe.”
      Then suddenly the nation is at war. “This is terrible!” the farmer tells the monk. The monk nds and says “Maybe.” Then all the other men’s sons are conscripted into the army, but not his because his leg is broken. The farmer’s son on horseback directs women and children to bring in the finest harvest ever, and his yield and services are in high demand in a region struck by war, where others’ sons never came home. “This is wonderful!” says the farmer. Quoth the monk, Maybe.

      Richard Vance’s own clever Saleman’s Tale beginning: “I agree.” ie alng with his Ram his take is also that essentially that things are terrible.

      I empathize, my gut is much like yours. But in the Samland Funhouse, we peer into a strange and strangely calming world with _less_ distortion than our own.

      How is it done? Slowly. Sam’s Looking Glass is carefully made so own narratives don’t come in with us. Here we see only what we can be most sure we know.

      So we notice new true things.

      Richard writes: “In the first debate[…Obama resuscitated Romney… RR got their foot into the door that should have been slammed shut.”

      _Should_ is beyond the purview of fact and likelihood. Beforehand, we weren’t looking at Obama’s usual numbers, but at a number inflated by certain events and which were likely to return to a lower figure. More importantly, they’d already begun this downtick _before_ the first debate.

      Richard writes: “Obama resuscitated Romney giving him a Presidential look.”

      It’s well known that ALL debates give advantage to the challenger, as the levelish playing field makes him appear presidential or nearly so. The effect is heightened at the first debate, which tends to poll strongest.

      It is _possible_ (I won’t call it true) that Obama had further to fall for several reasons:
      - Romney had, unusually, no post-convention bounce, which inflated at least the appearance of the post DNC bounce,
      -With the Akin incident so soon on the heels of his bill coauthor Ryan’s apearance and various other incidents, we did not see these numbers settle, but that is their tendency, or so I believe I’m learning.
      - Obama _may_ have looked more presidential at his convention. Accepting the nomination, he didnt need to explain anything. Clinton and others did the explaining far less tediously than Obama is usually required to do. Who knows? There were so many factors! We should ask Sam

      So the door was never slammed shut, whether or not it should have been. But it wasn’t Romney well-shined shoe that got in the door. The post DNC bounce down tick had ALREADY BEGUN before the first debate.

      That’s right, Obama’s numbers were down BEFORE Mitt’s surprising debate debut.

      Richard concludes there was “…Serious and permanent damage… done.”

      Well…. The narrative itself can contribute toward the narrative you posit. But the special thing here in Samland is we know that we don’t know yet. We wait and see.

      We say, “Maybe.”

      But Sam is likely better than maybe. There are statistical probabilities of certain outcomes based on the numbers we see today.

      Sam points out the the simple fact that Obama has always been ahead. All races go up and down. Is it realistic to select the moment of greatest comfort as being a norm, yet simulaneously a state of grace from which we have fallen?

      Nothing will unbreak the farmer’s son’s leg.
      On the other hand, I too am concerned about the “coattail” races of the House and Senate. Some however, are doing nicely on their own! Poor Akin!

      Tune in today at 2:15 Eastern time to hear Sam and another poll consolidator discussing their methods, algorithm and goals on NPR.

      I will. Probably!

  • Tony

    Don’t forget the Senate needs a 60-40 majority to be filibuster proof.

    Even if the unlikely scenario that the Senate goes republican and Romney wins happens, there’s no way the Senate will be filibuster proof.

    You can be sure that the Democrats in the Senate will use that filibuster power whenever the far right tries to do anything extreme…

    • Matt McIrvin

      The budget reconciliation process is not subject to filibuster, and most of the Ryan Plan is sufficiently budgetary that it could be rammed through that way.

      I’d guess that with a Romney presidency, a Republican House and a slim Republican majority in the Senate, we’d see much of Congress’s contentious legislative activity phrased somehow as budget bills so that it could go through reconciliation.

    • Olav Grinde

      Matt, in the scenario you describe, could the Republicans also use the reconciliation process to de-fund the Affordable Health Care Act?

    • Matt McIrvin

      According to Jonathan Chait, they can not only defund the ACA that way, they can voucherize Medicare too:

      http://nymag.com/news/politics/elections-2012/obama-romney-economic-plans-2012-10/

    • Matt McIrvin

      …note, there’s even precedent for this. Ever wonder why, when you lose your job, you buy health insurance from a venomous snake? COBRA was the name of a 1985 budget reconciliation law.

  • kinglerch

    Can someone explain to me how a poll of 500ish people in a national poll is representative? Let’s say there are 50 states :P and each made up 10 votes out of the 500…isn’t that subject to where those 10 landed? Were all 10 in NE Ohio or South Ohio? And what happens if 20 end up in California?

    500 participants in a coutrywide landscape like we have, seems to have a very large margin or error, or am I missing something?

    • Olav Grinde

      This is one of many reasons why Dr Wang’s prognosis relies on state polls only. National polls are not taken into account.

    • Jim in CA

      500 is a relatively small sample, but once you get up to 1,000 (or better yet 3,000) people, the results become relatively precise.

      You can get a sense of the effect of sample size by plugging in different numbers here:

      http://americanresearchgroup.com/moe.html

      The population size in this case is something like 100,000,000. Once you get above a million or so, it matters very little how large you make that.

  • A New Jersey Farmer

    Don’t make any life decisions based on the polls until Monday. That’s when most pollsters will have done surveys after the debate.

    • Ms. Jay Sheckley

      Sam, anyone, do you agree with what Jersey Farmer says? I really would like info on how fast/slow info arrives here.
      As for life decisions, I’m just trying to help until there’s no use and I’m off the hook.
      My dad who is 85 takes a sardonic view of my efforts. And when one considers the billions being spent this week alone, versus my nillions, ok he’s likely right. But the impossible is really the most fun: if you fail, who can blame you? If the result suggests you may have succeeded, OMG! Like art (or the bad deal of gambling) Sam is so right about the endorphin pumping hook of intermittent rewards. My birthday is November 6th, and I want to share the joy of renewal with everyone.

  • Ms. Jay Sheckley

    Proof I’m a PEC junkie: I noticed right off which times were new. Speaking of salivating and swing states, Ohio has gone crazy. Paul Ryan did a scrub & dash photo op at the Youngstown St Vincent DePaul soup kitchen, and the surprising pushback now angry Republicans are people defunding thekitchen which feeds 100,000 meals a years. Maybe this election has driven me insane, but I melt down so easy now that I phoned Brian Antal who runs it and got him to agree to attach his email to paypal if a buck arrives at bantal@svdpusa.org . FWIW, you heard it here first. And don’t forget what started all this: Fund the party of your choice right here using Act Blue or the link below it, on the left margin of this PEC page .
    Lastly, but most importantly, don’t fail to hear Sam Wang interviewed on NPR tomorrow- I think at 11:15 am West Coast Time, 2:15 Jerseyvote time.

    • Terry

      Jay,

      Just sent a donation to St. Vincent de Paul Society Mahoning County.

      It made me feel good and my mother would be proud.

    • badni

      Ms Jay:
      Could you explain what you mean by “and the surprising pushback now angry Republicans are people defunding the kitchen”?

  • Peter Principle

    Squeak! Squeak, squeak, squeak! Scurry.

  • Mike

    Why did the meta-margin go down after two great polls for us from Maris???

    • ChrisD

      Romney’s +4 in PA by Susquehanna (for the Republican State Committee)?

    • Mike

      That’s an internal poll… it shouldn’t even count!

    • ChrisD

      True, but Pollster is showing it. Does Dr. Wang use Pollster’s polls or RCP’s?

    • Mike

      Pollster is showing it? What BS. RCP is not. If other internal polls aren’t included, then Dr. Wang needs to remove this one as well. It was not a poll for public release.

    • Matt

      I don’t understand it either. The numbers don’t make sense for me in IA, WI nor CO for that matter. See my comment below about CO numbers.

  • Felix

    time to refocus on the Senate and the House, as Sam said many, many weeks before!

    • Matt McIrvin

      Speaking as a Democrat, I’d want to see it a lot higher before I’m convinced Obama is out of the woods. Romney’s ramping up to maximum ad blitz right now.

      But don’t ignore the Congressional situation, certainly. If Obama loses, a Senate simple majority is the difference between passage and blockage of all the Ryan budget stuff, including repeal of Obamacare, voucherization of Medicare, etc.; it’s apparently constructed to go through the budget reconciliation process to be filibuster-proof.

  • Tony

    The new Marist poll is well within the moe…

  • pechmerle

    Olav, Unfaithful electors who are pledged/state-law mandated for Obama and unfaithfully vote for Romney: Yes, hypothetically possible, but.not.going.to.happen.

    • Matt McIrvin

      I could see a crazy situation where the vote is on a knife edge and one of the blue-state electors decides to go for Jill Stein or something.

    • Olav Grinde

      Look, billionaires are dropping 10,000,000 and 100,000,000 checks. To them this election and Karl Rove’s part of it is “an investment” — and they want a return. If the EV margin is a handful votes or less, then I wouldn’t be willing to bet my life savings on it not happening.

      And that’s just one way to steal an election.

      Obama needs a very clear margin.

    • Olav Grinde

      $

  • don in fl

    new to site and actually illiterate to what the site does.is the+1.46% meta-margin good or bad for obama?

    • RDT

      It would be better if it were higher… As I understand it, it is the swing in opinion needed to change the winner.

      I’m trying to understand the changes in the last update — predicted EV went down, but meta-margin went up. Does this mean a small state just moved significantly in Romney’s direction, but a larger state just moved more toward Obama?

    • Craigo

      1.46% is the swing towards Romney needed to produce an electoral college tie.

      Polling results actually describe a range of results, with differing probabilities. For example, if a poll shows 50-47 for X-Y, with a margin of error of 3, then there is a 26.6% chance that the true value for X is 50%, and 40% that the true value for Y is 26.6%.

      Same does this calculation for every state, using the recent polling data. He then uses that information calculate the probability that a candidate will win a particular state. He then uses *that* information to calculate the probability that a candidate will win N number of electoral votes.

      The median result – combination of electoral votes – that cuts the probability distribution exactly in half – is the one displayed at the top of this site. If the election were held today, that’s the single most likely result.

      So to produce a tie – a scenario where the median is 269 electoral votes for each candidate – would require a national shift of 1.46% in the popular vote.

      Obviously, higher is better for Obama. 1.46% isn’t bad, but he peaked at about four times that number. The average has been about twice what it is now, so Sam expects him to continue to rebound somewhat.

    • Jack Rems

      Craigo- Are there typos here?
      “Polling results actually describe a range of results, with differing probabilities. For example, if a poll shows 50-47 for X-Y, with a margin of error of 3, then there is a 26.6% chance that the true value for X is 50%, and 40% that the true value for Y is 26.6%. “

    • Craigo

      Yes, there are! If the margin of error is 3 pts at 95% confidence, then there is a 26.6% chance that the mean is the true value. There, never trying to type a comment on my phone again.

  • Olav Grinde

    Tony: “I believe with Ohio, Wisconsin and Iowa in Obama’s column, we don’t need the rest of the swing States to get to 270 electoral votes.”

    You’re right! Even if Romney takes Colorado, Nevada and New Hampshire, in addition to Florida, Virginia and North Carolina — President Obama still captures 271 Electoral Votes.

    However, unless something drastic happens, the map won’t be that red. Moreover, I think it’s imperative that Obama win both the popular and electoral vote by a clear margin.

    First of all, he President Obama needs the mandate. And second, the coat tails are important — i.e. decisive control of the Senate, and flipping (or almost flipping) the House of Representatives.

    • Olav Grinde

      And thirdly, with a narrow EV victory, I would still be worried about unfaithful voters in the Electoral College. Although that has yet to flip a Presidential Election, it can happen.

      It is hypothetically possible.

  • Steven J, Wangsness

    Great. Now I gotta get up at 5 am (PDT) for my first fix — and nothing from late afternoon till the following day.

  • Florida Swing Voter

    http://i.imgur.com/bzTU8.png

    Sam and others, how do you feel about people worrying/celebrating over Gallup’s recent polling results while ignoring that we still have an electoral college? Ohio, Nevada, Colorado, Iowa and Wisconsin ≠ the South.

  • Jim in CA

    Hey! It’s 8:02 and no update. What gives?

  • Dave K.

    I qualify as a victim of this operant conditioning. Not just for numbers but exciting blog posts. Rss just isn’t SATISFYING enough. I’ve had to add all the other sites to my quadrennial addiction to make sure and soak up every little detail.

  • Paul

    I know Sam hates looking at individual polls, but I have a question about the Marist polls, specifically the new one from Iowa and the previous one from Ohio.

    They say that in Iowa 34% of likely voters in their sample have already voted and another 11% plan to vote early. Since 36% of Iowans voted early in 2008, a total of 45% early voting this year is not implausible. However, according to Michael McDonald’s early vote tracking, the current early vote in Iowa as of today only equals about 18% of the 2008 total vote, which should presumably be roughly similar to the 2012 total vote. Moreover, the Marist poll was conducted Mon-Wed when the total was a little lower (though maybe some ballots were in the mail which would cancel out that effect somewhat.)

    A similar question can be asked about the recent Marist Ohio poll, which said that around 20% of likely voters had already voted. At that time McDonald only showed about 5% the 2008 total vote in Ohio early voting . McDonald notes that his totals for Ohio are incomplete as some mainly smaller counties don’t update, but even so a factor of 4 is inexplicable.

    So while as an Obama voter I hope the Marist polls are correct, I am curious as to how Marist would explain why their samples appear to contain more early voters than their really are.

    • NC Obama Guy

      I was wondering the same thing. Sam can you help us out on this? If more early voters are Obama supporters doesn’t that mean that these polls (please forgive me) are “skewed”? OMG I can’t believe I just said that…

    • Craigo

      Hi Paul:

      When asked, there will always be more people who claim to have voted than actually did so. There are some areas where respondents will often lie to pollsters – this being one of them. Here’s a recent paper on the phenomenon:

      http://www.eitanhersh.com/uploads/7/9/7/5/7975685/misreporting08_v2.2.pdf

      All the literature that I’ve seen has focused on all voters, but I don’t think it’s a stretch that this will apply to early-voting as well.

      Combined with the random error of a small subsample, and I think this is easily explained.

    • Sam Wang

      A difference of that kind could be explained if it takes time for votes to be counted. If the two numbers never converge, then maybe there’s a problem.

    • Tony

      Didn’t someone said that it takes days for a casted vote to count?

      Marist is conducted Oct 15-17. That’s probably more recent that a district actually counting an already cast vote.

      Correct me if I’m wrong.

    • Matt McIrvin

      Besides the mechanisms other commenters have given, it’s possible that their LV screen is actually undercounting likely voters who haven’t voted, which would make the people who already voted a larger fraction of the total.

  • Neil

    Every update needs a neuroscience note.

  • Olav Grinde

    Tony, those are good numbers.

    I’m hoping to see more non-Rasmussen polls for other swing states tomorrow — Florida, New Hampshire, Virginia and North Carolina. An Arizona poll from someone other than Behavior Research Center would be nice too. I am also curious about Missouri.

    And there are certain Senate races we haven’t seen in a long time!

    • Tony

      I believe with Ohio, Wisconsin and Iowa in Obama’s column, we don’t need the rest of the swing States to get to 270 electoral votes.

      With the New NBC/Wall STREET Journey/Marist POLL, Oct. 15-17…

      Iowa: Obama-51, Romney-43

      Wisconsin: Obama-51, Romney-45

      These new poll numbers are about Obama’s best numbers in those swing States…

    • bks

      The difference between the EV call at electoral-vote.com with and without Rasmussen data is 6 which move from Obama to toss-up. I am going to give up ragging on Rasmussen until I see a more convincing critique.

      –bks

    • Matt McIrvin

      I’ve never been convinced by the claims that Rasmussen’s numbers are completely fraudulent, or that they skew them way Republican until late in the cycle and then tweak them to be accurate at the last minute to save their credibility. To me, it appears that they just use a somewhat-worse-than-usual weighting procedure that gives them a Republican house effect. But it hasn’t been huge this cycle.

      I think a valid critique is that their practice of re-weighting the data to a model of the party breakdown of the electorate is not doing anything good. In practice, what it does is to just artificially damp out fluctuations, until such time as they update the party model.

  • 538 Refugee

    Doesn’t the schedule negate the intermittent/sometimes part of the equation?

    Wait just a cotton picking minute. I think we’ve all been had. We are all part of some experiment Sam is running……

  • Brian

    For any of you running Google Chrome — you can install my Princeton Election extension, which puts a little icon in your browser that, when you click it, gives you the up-to-date Meta-Margin, Obama EV, Romney EV, the median-EV history graph and a link to Sam’s latest blog post. If you are the rat, this can be your lever.

    https://chrome.google.com/webstore/detail/princeton-election-consor/mopbhdfigdcejolaebakppgnffbhlamd

  • Tony

    Good news!
    New NBC/Wall STREET Journey/Marist POLL, Oct. 15-17…

    Iowa: Obama-51, Romney-43

    Wisconsin: Obama-51, Romney-45

    These new poll numbers are a lot more important than the Gallup Poll imo.

    If Obama wins Ohio, Wisconsin AND Iowa, he wins the election.

    • James Moore

      You have been reading the posts where everyone on the planet agrees that individual poll results are about as interesting as, hmm, something that’s completely uninteresting, right?

    • Matt McIrvin

      Yeah, I’d wait for more data before busting out the champagne. Still, they’re not bad. Especially given the scary Marquette poll in WI.

    • Matt McIrvin

      Ah hell, Republicans tout *their* individual data points to the heavens, even if they’re obvious outliers. So THESE ARE THE MOST AUTHORITATIVE NUMBERS EVER!!!

    • Jen

      I don’t agree with James. These numbers are VERY interesting to me! And I completely agree about their significance. Florida, North Carolina, etc stopped mattering a while ago….

      Would be great to see some polling out of Nevada and Colorado.

    • Craigo

      Well, let’s talk about that. Yes, you should not take any one poll as gospel, for any number of reasons. Random error, house effect, and the probability that you will, subconsciously or not, begin to discount polls unfavorable polls and trumpet favorable results.

      This especially goes for national polls, which have a higher volume than any single state, and thus providing a reasonably good sample size of results.

      But there have been only six polls conducted in Iowa during the month of October, and only eight in Wisconsin. (According to the Pollster database.) We’re basically starved for post-debate data at the state level, compared to the dozens of national surveys in the same timeframe.

      Every little piece helps – as long as you’re looking at the picture itself.

  • piktor

    More, more! Faster, faster! Cain’t get enough of the jumping Estimator, the Meta-Margin, the works!!

  • Joel

    I feel my dopamine level spiking!

  • A New Jersey Farmer

    I was wondering why I’ve been salivating more lately. Looking forward to the updates.

  • Stuart Mills

    I need more! Every 15 minutes should be OK.

    Is there anyway to provide page with a list of all the polls that went into the updates so we can get a feel of the marginal contributions?

    Thanks much (even if the answer is no)

    • MAT

      The polls are loaded from Pollster. They have a list of new polls in the right column of the main page.

  • Independent Left

    what’s a browser? i get all my news via RSS…haha…push announcement of new content rather than hopeful refreshing…

    that said, i have some websites (this is one of them) dialed into a combo of muscle memory and google chrome auto-complete. my favorite is when i type the address for the page i’m already on.

    eagerly awaiting…

  • Anbruch

    So what you are saying is that if you want to spike the page loads, you should offer updates at random times…