Princeton Election Consortium

A first draft of electoral history. Since 2004

Bush v. Kerry revisited?

October 12th, 2012, 2:15pm by Sam Wang


In national surveys, Romney’s up by a median of +1.0+/-1.0% (n=10). Yet the Popular Vote Meta-Margin, which is nearly all caught up, is at Obama +0.84%. What accounts for the difference?

Normally I don’t engage in this kind of detailed combination-dissecting, since there are so many possibilities. The whole point of the Meta-Analysis is to avoid such activity. But it does illustrate the difficulties faced by a candidate, even when a race is very close. Also, the risk of nonmatched popular vote and electoral vote is a recurring theme in recent Presidential politics (2000 and 2004).

This is Romney’s best scenario with current polls:

The Electoral College contains a structural advantage that favors Obama at the moment. Click the map once to get current probabilities, twice to get a version you can experiment with yourself. Interactive versions are also available in the right sidebar.

The most plausible paths for Romney at the moment are OH or hitting the double: a NV-WI combination. All three of those states have totally post-debate polls, and they are all blue. It would take a pretty hard push to drive the Meta-Margin to the red side. Possible, but something big (in addition to current trends) would have to happen. Watch the major candidates’ travel schedules.

Tags: 2012 Election · President

87 Comments so far ↓

  • Insidious Pall

    Seems like the other ‘easy’ path for Romney is to win IA and NV or WI. Either of those scenarios could negate the Ohio firewall.

    • Obama 2012

      Problem for Romney is Obama has consistently lead polls in all three of those states. And there is a very good chance that Obama will outperform Nevada polling (he did so by 6 points in ’08… and Reid greatly outperformed polling in ’10 … Nevada polling is undersampling Democratic strengths for some reason or another – this was even true going back to ’04)

    • Sam Wang

      Polls also wrongly estimated the 2010 Senate race, showing Angle ahead of Reid. Nevada has an unusually mobile population – lots of people coming and going. These people are hard to reach by any survey means, whether landline or mobile phone.

    • Stuart Mills

      Nate just posted an interesting discussion about voters that only speak spanish – they are not represented in current polling.

      See http://fivethirtyeight.blogs.nytimes.com/2012/10/13/oct-13-arizona-and-the-spanish-speaking-vote/#more-36004

      Might help explain Nevada a bit

  • NY Romneyite

    Starting with your map, paths that work for Romney:

    The Ohio Method
    1) OH

    The Wisconsin Method
    2) WI + NV
    3) WI + IA
    4) WI + NH

    The One No One is Talking About:
    5) NV + IA

    • Insidious Pall

      NY -

      Your #5 scenario seems the most likely. IA is a mystery but there’s bleeding in Nevada and with their economic foibles, it may turn into a hemorrhage.

    • Froggy

      I’d say that Ohio flipping for Romney is still far and away the most likely of the five scenarios. Romney has yet to lead in any NV poll this year, and led in WI only in two August polls taken in the immediate aftermath of the Ryan announcement, and then by only by 1%.

  • Froggy

    NH and WI is another possibility to overcome OH. The two fully post-debate polls in NH have it tied (Ras, 10/9) and Romney +4 (ARG, 10/9-11). (This means that the next update switches NH from O+6 to tied.)

  • theDAWG

    Is Romney taking VA so clear cut with Virgil Goode on the ballot?

    Does Goode get a larger response when pollsters mention him by name (like it will appear on the ballot)?

    • theDAWG

      Put another way, does “Other” get a larger share when pollsters ask:
      Romney/Obama/Stein/Goode

      vs. when they ask:
      Romney/Obama

      and what is the effect on who gets the plurality? The HuffPost aggregator isn’t clear on whether pollsters give prompts for the 3rd party candidates.

    • Froggy

      PPP polled Virginia recently both ways (http://www.publicpolicypolling.com/pdf/2011/PPP_Release_VA_1007.pdf ). When asked Obama vs. Romney, the result was Obama ahead 50-47, with 3% undecided. When Johnson, Goode, and Stein were included as options, Obama’s lead ticked up to 48-44, with Johnson getting 4%, Goode 1%, and Stein 0%, and with 2% undecided. It doesn’t look like Goode will have any significant effect on the race.

    • Obama 2012

      There’s definitely no clear cut Romney win in Virginia. That’s crazy talk. Obama has seen some very strong post-debate polling in VA (NYT/Q = Obama +5). And yes Goode could take a 1% or so from Romney which may be enough to flip it if it gets extremely close (see Nader / Florida / 2000)

  • Steven J, Wangsness

    Frankly, this is getting scary. I keep waiting for Romney to revert to his mean. (Not his mean self. You know what I mean.) I’m hoping that he peaked yesterday just shy of 48%. What’s really scaring me is Obama’s drop. It should have ticked up by now if all we were looking at was a Romney bounce, shouldn’t it?

  • Obama 2012

    Fear not! Yes the election is looking to be much closer than we (Obama supporters) were hoping it would be… but Obama is still in the better position with Romney’s path to 270 still being quite precarious.

    Plus Biden kicked serious butt last night. I really think it’ll stop any downward trend for Obama/Biden.

    • Sam Wang

      Proving my point about the post hoc, propter hoc fallacy.

      Come to think of it, it’s actually pre ante hoc, propter hoc.

    • Arbitol

      I fear it is starting to look like this may be the only point of yours proven this cycle …

    • Pat

      Yes. I am starting to think that the overconfident tone on this site, telling us for a long time that Obama is pretty much a lock and it’s no use giving money for that campaign, and being pretty dismissive of more cautious voices, and to “snap out of it” when more and more bad polls were coming… all of this was maybe not too appropriate.

    • Sam Wang

      I am glad you are not questioning the numbers. I agree the situation is very close. However, Democrats are currently still better off with their Presidential numbers than Republicans.

      When things get tight, fight harder. Thus the Moonstruck reference.

    • Renee

      True, it seems that when things get tough for Dems, they start wringing their hands and fretting and pointing fingers and the worse…THROWING IN THE TOWEL…while still in the thick of the game (for example, if the Dems were the Cardinals, the Cards would have given up and stopped swinging their bats in the 4th inning or the 9th, not go on to win in the last inning… sorry first analogy that came to mind since I stayed up late to watch the game)! That is a very weak quality, turns people off, and is demotivating. This is just based on appearance but it seems like a lot of Obama’s supporters are of that stripe.

    • wheelers cat

      As predicted here at PEC, Debate1 bounce appears to be over in RAND.
      https://mmicdata.rand.org/alp/index.php?page=election#election-forecast
      The nat’l polls are sampling from a closed subset of responders.
      This is a bigger problem every year.
      Just because someone is unlikely to take a poll doesnt necessarily mean they are unlikely to vote.
      In the past “enthusiasm” was the best correlate with liklihood of voting. That may no longer be accurate because of cell-onlies(~32%) and smartphone users(~47%) who never get asked if they are likely to vote because they dont accept polls.

    • P.B.

      I don’t question the numbers per se, and I’ve taken great interest in and comfort from the discussions on this terrific site. I do, however, question the certainty ascribed to what amounts to a model of expected voter behavior, at best proximate to actual patterns. I guess I’m saying the following: When your model was predicting a 90-97% chance of an Obama victory, those odds were mitigated (in my mind) by the substanial possibility that your model contained flaws — perhaps 2012 would behave rather differently from 2000-2008, for instance, providing revelatory new data. In short, I’ve always understood your impeccable logic, but could never buy its hand-to-glove fit on reality, worrying as I do that there are more things in heaven and earth, Sam Wang…

      Am I being innumerate here? Please let me know.

    • Sam Wang

      I recognize your concern. You can’t possibly think I am immune to what is unfolding.

      The facts at present, to a colder calculation, are:

      1) Obama’s been ahead all season. He’s currently in a place I said he’d be about 14% of the time.

      2) Something big happened last week. We don’t know which way things will move.

      3) In past races, where the race is now is not very predictive about where in the range it will be, 3 weeks out or longer. Closer than 3 weeks, the current state of the race gets slowly more predictive.

      4) The people surveyed in June-September are the same ones surveyed postdebate. To me this suggests movement back toward the mean, Obama +3.0%.

      Yes, there is a chance it will not happen and Romney will win. Storms can turn. Currently I think his probability is about 1 in 6.

    • wheelers cat

      @PB
      What if the model of expected voter behavior is corrupted by Bayes and Gauss?
      What if the evolution of technology and the evolution of the electorate has changed what we can assume about sampling and forecastng?
      Its not like Dr. Wang’s turning storm– there are interactions and dependencies all round us.
      In my Proving Progams Correct class it is said you can prove there are no detectable errors in your code– you cannot prove there are no undetectable ones.

    • wheelers cat

      @Pat
      as predicted on this site, RAND this am.
      https://mmicdata.rand.org/alp/index.php?page=election#election-forecast

  • hummbumm

    looking at palin/ryan bounce max bounce was 11 days, so hopefully for me as an obama fan, romney bounce will peak tomorrow or sunday and then hopefully the reversion to the mean anchor point of +3 will begin.
    Looking at Ryan bounce meta margin got down to +1 so this has been more trying and obviously closer to election, but if pick up begins on sunday, then things will be acting true to form.

  • Olav Grinde

    Dr Wang: Do you have any numbers on what percentage of people in key swing states voted before the Presidential Debate and thus before Obama’s precipitous drop?

    If a significant number of voters did cast their vote before this, how might that influence your predictions?

  • chris brandow

    I don’t trust my perceptions at all under circumstances of close elections, but it has seemed as though a larger than typical percentage of the polling results have come from less-known pollsters or historically R-leaning pollsters, which would make this bounce seem a little larger than perhaps warranted. Do you have any measures on this?

    • Stuart Mills

      It does seem like we’ve only gotten cr*p polling at the state level so far.

      PPP in ohio today looks good though.

  • pechmerle

    Ohio early voting has begun, and should be a significant pro-Obama effect. I think about 18% of the votes are already in in Ohio. Among those polled who said they had already voted, I think the figure I saw was 63% voted Obama.

    (Somebody straighten me out if I’ve hallucinated any of this.)

  • 538 Refugee

    Is there any evidence yet that there a polls put out just to try and influence the poll aggregate sites? This could possibly justify at least some of Nate Silver’s tinkering model.

  • Richard Vance

    Nobody is talking about the Colorado flip to red. That went fast and far!! I was looking at CO as a firewall as I saw that when Obama gets NH and CO and NV then he can lose OH and FL and VA and still win. Its shaping up as the small EV states needing attention. If I were he (Obama) I would be in NH, NV, and Colorado…. spend tons of $$ and time to bring CO back blue then you don’t care about VA, OH, FL..

    • Sam Wang

      Perhaps it is time for me to remind everyone of jerseyvotes.

    • Richard Vance

      Jersey votes don’t measure the effect of a $1 only the effect of a human vote. Correct? So, where should I invest my $$ in last weeks of campaign.. I need a dollar value meter..

    • Olav Grinde

      @Richard Vance: Your dollar influences that one decisive vote. Hence the Jersey Vote is, essentially, the same measure.

  • wheelers cat

    ummm….just because someone refuses to take a poll doesnt necessarily mean they arent going to vote.
    sure, enthusiasm was the best correlate for liklihood of voting in the past…but who can say that is true now?
    That is why I think RAND is a truer measure of the underlying structure of the data.

  • NickT

    Sam Wang, shame on you for muddling your Latin prepositions and prefixes.

    Ante hoc, please.

  • Howie Weiner

    My interest has always been the emotional or “irrational” as it applies to both social situations and scientific analysis. I’ve always felt that sometimes a deep political or emotional impulse can oftentimes supercede rational thought. The debate may have been one of those events. Keep waiting for the Obama rebound but it’s hard to shake that final image of him at the debate almost pleading us to give him another chance. There has been a turn towards Romney and I think it has come from some deep impulse that released many voters from a certain sense of obligation towards him as a person and a leader. The Republican ticket expresses the most radical agenda since….I can’t think of anything but pre-Civil War. Obama can recoup some of his leadership in the 2nd debate but my guess is it will be at best a draw.
    The election was always going to be really, really close and the break on the last week before Nov 6th will be decisive. Who wants it more? Who will put it all on the line, that is what confronts all of us.

    • Olav Grinde

      Perhaps, in order to swing back, the voters who are adrift or have swung toward Governor Romney truly need to see a President Obama who is “all fired up and ready to go”.

      They most definitely did not see that in the first debate.

      However, a re-energised President Obama may not be enough. Romney/Ryan and GOP superPACs are poised to carpet-bomb the airwaves with their campaign commercials in the next three weeks.

      Unlike many people, I do not have confidence in the integrity of the voting and vote-counting process. Much has been invested in undermining it. I believe that is why, in order to win, Obama needs to win by a very significant margin.

      Time will tell whether we will see the pendulum swing back.

  • BillSct

    I’ve been looking at the Pollster graphs for all the battleground states and every single one, with the exception of Iowa, shows an Obama bounce beginning in late August/early September and hitting a maximum between 21 and 24 September then subsiding over about a week to more typical levels. In all cases, the end of the decline appears to coincide with Debate Number 1 and things stabilize after that. From this it looks like perhaps as much as half of the precipitous drop in the meta-margin was the end of the No Republican Convention Bounce/Big Democratic Convention Bounce and the rest is due to momentum effects from Debate Number 1, the polls dumped right after Debate Number 1, and a weeks worth of reporters fanning the flames that spontaneously erupted in the hair of scared Democrats and other left of center folks. What is interesting is that from Debate Number 1 on all of the graphs from the Battle Ground States are on track with their past results. Iowa, which doesn’t show any bounce, shows both Obama and Romney trending up on about the same slope but with Obama holding a steady 4% lead since early November.

    It would be useful to see all of the battleground graphs next to each other (i.e., what E.R. Tufte refers to as small multiples) to better take in the comparison.

  • BillSct

    I meant “with Obama holding a steady 4% lead since early September.”

  • Mike Davis

    All right, for the first time in freaking forever, this update didn’t go down any further from the last one. Let’s let this be the bottom, OK? I will even be happy if they don’t change at all.

    • Anbruch

      Actually, there have been several reversals/plateaus update-update, but I don’t think we’ve made it through a day since the debate where the end of the day hasn’t seen significant erosion from the beginning of the day.

    • Sam Wang

      Yes, but look at the dates on the polls. I agree that the MM is moving toward Romney, but only after the fact will we be able to say when (and if) it hits a minimum and reverses. By then the story line will be written, whether true or not.

    • MikeDavis

      I’m ready for it to plateau! No more bad news is totally fine with me. (I’m less commenting and more praying to the gods of politics, I guess)

  • Reason

    Not sure what you meant by the pre hoc, propter hoc.

  • AlpsStranger

    Here’s AlpsStranger.Common.Sense.Wake.Up’s prediction:

    Romney will win. Republicans will take the senate and house, and Todd “Achin’ Toad” Akin will win handily.

    Take that to the bank, friends :(

    • wheelers cat

      no thnx.
      I for one do not priviledge “common sense” over IQ, experience, and education.
      That is a trait associated with organic conservative tendency.

  • Opinion

    The way I see it, many people were looking for a reason to not vote for Obama. Instead he convinced them that Romney was not qualified and was dangerous through ads and Romney’s 47% statement.

    Then the debate happened and people realized they were actually looking for a reason to not vote for Obama, and he fell fast and hard.

    In 90 minutes, he reminded them it was him they were actually looking to vote against, plus, he gave them reasons to actually vote against him.

    I still think Obama wins (and I am voting for Obama), but I think he needs to remember that people are always looking for a fresh start in times of crises and are actively looking for a legitimate reason to vote against him. He shouldn’t give them one by being sheepish, incoherent and absent-minded.

    • John Jacobs

      Completely agree with this analysis.

      I think its more than the crisis though. The Republicans have managed to build a compelling narrative in the American psyche – they’ve been trashing Obama for years as an incompetent who was elected based on star power and racial guilt.

      More people remember that he was a community organizer (2 yr post college job) than his years as a lawyer/ professor.

      Awful move: walking right into the narrative by seeming less competent than a rich white former financier (Americans instinctively think rich = smart/competent, and Romney certainly has the smarts and business record to back the money up)

    • hk

      I think you are spot on in terms of debate impact. Some of the president’s edge prior to the debate was due to voter’s whose support for him was soft. They are the one’s who jumped ship post-debate. The soft Romney supporters on the other hand became more entrenched and those on the hard right became more enthustiastic.

    • wheelers cat

      @hk
      How do we know ANY voters have jumped ship?
      The polls may just polling a subset of people who will take polls.
      The subset of people who will accept a polling call is likely correlated with enthusiasm as well as landline bias.
      That is why I think RAND is more accurate if you are looking a real shift from one candidate to another, because they are sampling a fixed population that has nearly 98% response. The population of RAND panelists was set up randomly, but RAND pays $2 for each returned survey. Pretty much deals with the non-responder problem.
      That and the biological basis of behavior.

  • Patrick

    Your site usually keeps me calm, but the falling MM is really working me up. Do you still think that we’ll be seeing a reversal, and that Romney’s debate high has already peaked?

  • Jun Talabucon

    Quite surprised why Obama’s EVs was down to 277 here. It turned out, it’s due to ARG’s NH poll.

    Prof. Wang is too generous giving this poll credit, and so, their other polls as well, considering the item below:

    http://ajacksonian.blogspot.com/2007/12/problem-with-american-research-group.html

    Or he’s just playing coy by patronizing the Romneyites at this point of the cycle…

    He’s even acommodating enough to accept:

    “Ok, it’s down to OHIO now…”

  • minimus

    I gave all my political contributions to Congressional Democrats in tight races, since this site was assuring me that Obama was a lock for relection and he didn’t need my money or help. Much better to funnel efforts down ballot. So, all is not lost. Obama may lose, but hey, all is not be lost… Heidi Heitcamp may be next junior senator from North Dakota. Yippee!

    • Sam Wang

      Still a better use of your money, unless you were going to give the President some Diovan or debate coaching. I appreciate it very much.

  • Bill N

    Sam,

    When are you going to update your election prediction? It has been about 10 days since the first debate. My guess is that this race is now what you would call “knife edge.” I cannot see it being any other way at this point. Your last forecast was, I think, about a 90% chance of re-election. Nate Silver’s forecast is now approaching only 60%.

    I have to say I have been shocked at the rapid decline in the meta-margin the last ten days and the re-shaping of the EV out look. I did not think it could change with such velocity. Colleagues of mine, as well as myself, have believed it nigh on to impossible for a candidate to win a presidential election who flip-flops on issues, tells demonstrable lies, tells folks he will let you know what he will do on any number of issues only after he is elected, and in general behaves as Romney has done. Yet, my gut is telling me he will be the next president of the U.S. If this gut hunch is correct, at this point the only thing that could put the brakes on a dramatic change sweeping this country, and saving the social safety net we have had for so many years, is a Democratic Senate. This soon may be a country which is not a good place to grow old in.

    • Sam Wang

      It is updated four times a day. Click on the time series graph for the 68% strike zone. Eyeballing the graph it looks like 84% at the moment.

      My gut does not say what your gut says.

  • Bill N

    Thanks for your quick reply. You seem to be an early bird like me.

    My gut will bet your gut a beer on this one.

  • Bill N

    Agreed. Donate and volunteer in any way possible.

    What brand of beer?

  • Ray

    Hi Sam! I really love this website. Can you comment on the similarities of this with the Bush/Kerry graph in 2004. They look eerily similar right now.

    • Sam Wang

      I suspect most reelection races look similar. There’s a great little book by Erikson and Wlezien on the subject.

      Compared with Bush v. Kerry, this year’s race is shifted about 2-3 points toward the Democratic candidate. That race was a near tie and the lead switched several times.

      This year is considerably less variable – on average it has moved over half the range. In 2004, a 50-EV shift was not unusual.

      Some shift points are similar: conventions and debates. That year, SBVT appeared to be critical. Bain was this year’s SBVT, maybe 47% as well if Team Obama keeps its head.

  • Leo McNellis

    I don’t doubt that Sam’s posted numbers have ever been lacking in integrity or reliability at any point in time for the two months I have been daily following this website. What his — or any other person’s model — do not and probably can not accurately take into consideration, however, is the propensity of candidates’ uncharacteristic behavior and uncontrollable “acts of god.” Nobody can predict with any strong degree of accuracy the next San Francisco earthquake, the next suicide bombing in Damascus, or the next time a sitting president thinks it good practice to ingest four ounces of cannabis before an important game changing debate.

  • A New Jersey Farmer

    Nobody said anything about a lock. The point is that given all of the variables and the polls, the likeliest outcome is an Obama victory. This was always going to be a close race. If the mm is +.25 on November 6, most of us would be very happy.

  • Lee Hostettler

    Thanks you Sam Wang, although I am no longer on the political high I was on two weeks ago, I feel it may be best…I started to be a little over confident in Obama re -election…It is time to fight on and to realize we do not have this is the bag…Ohio will remain our fire wall..Florida and Virgina are still clearly within our reach, but it is going to take work.It is time to roll up our sleeves and get busy

  • wheelers cat

    Again, if there are flaws in the 21st century polling methodology, how do we know? The CLT and the Law of Large Numbers are wholly based on the idea that the underlying structure of reality conforms to Gauss.
    All polling is based on that.

  • NC Obama Guy

    Could anyone please comment on the RAND survey? It seems to have ticked back up yesterday. I suspect that is because the surveys from the day after the first debate (the strongest for Romney) fell off yesterday. It seems interesting that that survey has not shown the same changes that we see in all the other polls. Can anyone conjecture as to why?

    • RDT

      My understanding is that it may be less subject to swings in voter enthusiasm since it polls the same people over and over.

    • Matt McIrvin

      I’ve been commenting on this periodically. I personally think the RAND sample may differ from the general population by virtue of being more politically plugged-in, simply because they are in the RAND sample and are being asked about politics constantly.

      This may make them less likely to have gotten their first good look at the candidates at the debate. Also, what RDT said: swings in enthusiasm will change what they say about likelihood of voting, but not whether they respond to the poll itself.

  • Jen

    Sam, thank you so much for your thoughtful analysis this election season. It is rare to find and so much appreciated! I am bummed about recent events….very bummed, actually. But I trust your judgment and instincts as much as ever.

  • Obama 2012

    I know you don’t talk about individual polls – but that Obama +2% in Arizona (all post-debate sample) is pretty awesome. I really think folks are underestimating how much the GOP is hurting themselves in heavy Latino states.

    I see Nevada & Colorado as states where Obama is almost certainly going to out-perform the polls (look at 2008 & 2010 for evidence of this – Dems greatly outperformed the polls in White House/Senate races) … and with a very strong Hispanic Senate candidate running in Arizona (and no homegrown Senator on the ticket) … Arizona could be the Indiana of 2012. The surprise no one sees coming.

  • Ms. Jay Sheckley

    To be clear: today Sam Wang wrote: “… To me this suggests movement back toward the mean, Obama +3.0%. ¶ Yes, there is a chance it will not happen and Romney will win. Storms can turn. Currently I think his probability is about 1 in 6.”

  • Ms. Jay Sheckley

    So you’re saying Obama 5 to 1… as of now, so it seems.

  • Olav Grinde

    I do notice that the Rand Survey’s current prediction has Obama at +3%.

    Which is the mean that Dr Wang suggests the margin will swing back toward.

  • NC Obama Guy

    Sam – I have a question about the RAND poll conflicting with most of the other data coming from other traditional polls at this time. Is it possible that a huge shift in voter enthusiasm, ( both increase from Republican side and decrease from the Democratic side) after the first debate could have resulted in a huge swing in willingness to participate in polling? To me this could explain the changes also seen in “party identification” (increases in Republican ID) which are present in a lot of the traditional polling we have seen over the last week. The RAND poll would not be affected by this “enthusiasm for taking a poll” shift because the same individuals are always included in the poll. If this makes any sense to anyone I would love to hear what you all think.

  • betsy teutsch

    Great site! I loved looking at great numbers for Obama . The last 10 days has certainly been a shocker. The conventional wisdom was that debates have no impact. My theory? That this became the close race which was always predicted, and Republicans were longing to not be embarrassed by their nominee. Romney fired them up at the debate by looking presidential and feisty, and they got their groove back. Doesn’t mean that Romney suddenly turned into a great candidate, just a more plausible one. My gut tells me it will fade. We still haven’t heard a tax plan, nor have we seen tax returns. Fingers crossed, and just volunteered for more GOTV shifts here in Pennsylvania! Complacency might have been more dangerous for the Obama team than a bad debate. Maybe it was a good thing.

Leave a Comment