Princeton Election Consortium

A first draft of electoral history. Since 2004

Anatomy of a bounce

October 21st, 2012, 8:00am by Sam Wang


In the wake of his improved debate performance, President Obama’s recovery is now apparent. It is most clear when viewed in terms of the Meta-Analysis of state polls. Over the last four days, the Popular Vote Meta-Margin – the amount of swing it would take to create an electoral near-tie – has moved by over 1.0%. Today, the President’s effective lead, using Electoral College mechanisms, is Obama +1.8%. A rapid move like this can continue for a few days as polls catch up with the nation’s mental state.

But why do national polls continue to look so close, and in about half of cases good for Mitt Romney? To answer that, let’s take a look back at the effects of all three debates so far.

This is a graph showing the average national margin during a 4-week period when 50 national polls were taken. (If you wonder why poll aggregation is needed, that fact alone answers it.)

The graph is easier to interpret than most aggregates (compare with RCP and Pollster). To calculate it, I spread out each poll’s data across days when the pollster made contact. For example, the ARG Oct. 11-14 poll is split across four days. To capture the Debate #1 change, I left out a few polls where contacts were equally spread before and after October 4. The gray zone is the 1-sigma confidence interval.

The post-debate-1 correction was a nearly five-point swing. It was complete within one day. This means that the post-debate-day media meltdown could not have caused the swing – though it certainly helped cement perceptions.

What caused this crash? Considering the polarization of voters this year (only a small fraction are persuadable), it seems likely to be caused by a change in morale on both sides: hope among Romney’s supporters and despair among Obama supporters, and a consequent change in whether they meet the criteria for being a “likely voter.” Imagine that an Obama supporter was 60% likely to vote before Debate #1, and then 57% afterward – and vice versa for a Romney supporter. That could fully account for the change.

There could also be some fraction of voters whose minds were changed by suddenly-moderate-Mitt and stumbling-Barack.

Since that time, there’s been a reversal in the direction of change. On October 5, Romney had a narrow lead, about 1.0%. Today, President Obama is back in a razor-thin lead at the national level – about 0.5-1.0%.

Because the race is so close, individual polls will inevitably be all over the place. And news organizations love the outlier data points, like the Gallup poll showing Romney +6%. I find this to be a particularly unattractive trait in media coverage. It was what led me to start the poll meta-analysis in 2004.

But here is something interesting. National polls do not match the state polls – and it is state races that determine the outcome, via the Electoral College. In the Meta-Analysis that Andrew Ferguson and I report on this website, Obama has been ahead all along.

You can see this especially clearly in the Popular Vote Meta-Margin, which I have defined as the amount of voter swing needed to create a tossup as defined by the Electoral College. It is an extremely sensitive measure, precise to within 0.2-0.5%.

Here is what it looks like. I shifted it one day to the left to match the national-poll data. The general pattern is clear: viewed through polls that focus directly on electoral mechanisms, Obama performs 1-2 points better than in national polls.

The likeliest cause of this discrepancy is that in states where it matters most such as Ohio, Nevada, and Colorado, the two candidates are genuinely overperforming/underperforming. Also, as I noted in 2008, winners in non-swing states often outperform polls. So there is some question about the accuracy of likely-voter screens when a local result is very unequal. Which leads us back to state polls as a better measure of the race.

In this graph the Meta-margin appears to be a lagging indicator. However, it’s not an apples-to-apples comparison, since I am showing the calculation as it unfolded, day by day. It can move quickly if more state polls are reported – as has occurred since Debate #2.

I will be interested to see if on Election Day, the national vote and the electoral vote count still show this discrepancy. If both are accurate, President Obama’s re-elect probability is about 90% – but his probability of winning the popular vote is lower, about 70%.

Update: for hobbyists, the MATLAB source code is here.

Tags: 2012 Election · President

231 Comments so far ↓

  • Lisa

    After visiting this website several times, I can clearly see differences between Repubs & Democratic voters. Generally speaking, Dems are quick to succumb to any type of negativity. Any change in the polls sends them off the deep end of the ocean. Repubs on the other hand will fight until the end. They don’t care if they appear to be losing, they hav, e a lot of fight in them. As an Independent who is going to vote for PO, I have complete confidence that he will win if we quit the mass hysteria over the polls and simply VOTE, encourage others to vote. Take people to the polls. Do something encouraging to ensure victory. Temporarily act like a Repub and fight to end and of course vote Democratic. Change your mindset and fight. Yes, the first debate did not go well. But does anybody really think that the people who appeared to have shifted were really solid PO voters? Of course not. The thing to do is too appreciate Nate Silver & Dr. Wang for wait they do but hey are not magicians, they can’t assure the future. But we can, if we let go of the hysteria and do something about it. I don’t know about you all, the polls mean nothing if we don’t vote. And I intend to fight until the last vote is counted.

    • Laura

      Good comment, Lisa, and I agree with you, for what that’s worth. We should all make some calls, donate a little money, and make sure everyone we know is voting.

    • A New Jersey Farmer

      Thank you Lisa. My sentiments exactly. Or you could blog and send your thoughts out into the world.

    • mnpundit

      As I’ve said elsewhere. Republican governance damages the country so badly and causes so much suffering it makes a lot of sense for Dems to fear it.

  • Lisa

    Sorry for the typos. I was typing fast. Also, I believe some of the pundits on MSNBC did more to hurt POTUS by two week rant on their perception of how bad he did in the first debate. Their constant rants instead of support of the POTUS, caused hysteria and premature visions of Nov 06, 2012. If you r a solid Dem or Independent who is sold out for our Prez, you knew something was off in that first debate and he would come out fighting for his base in the second debate. And he did. This last and final debate means nothing to me because I have seen enough of mittens to know he is unstable and not fit to lead our country. Look at the pictures of the long lines today, who do you see? The mass majority are people with Obama buttons, etc. The enthusiasm is STILL there! You can’t believe all of the hype and mind games put out by the Repubs, just fight and show up to vote. When it’s all said and done, I really think the Prez is going to win more than we think. But doesn’t really matter, he will win if we show up. The Repubs are definitely going to show up! JUST VOTE & FIGHT! CHANGE THESE NEGATIVE WORRISOME ATTITUDES….IT SPREADS LIKE A DISEASE . And we can’t afford a disease like mittens right now, we don’t need to help him by our incorrect thinking and quick to worry attitudes.

  • Lisa

    @ A New Jersey Farmer…more than happy to put out a blog…What site? I am serious soldier for this election. Too much is at stake! I have been calling and encouraging everybody I see to vote and to check online to make sure they haven’t been dropped due to the true to vote nuts! I am fighting my—-off. My MIND is made up. I will not stand by and not give my all.

  • Reason

    MM down .8 in one day? K. Worried some.

  • Pankaj Sharma

    Today was a slow news day as far as polls are considered (exception NBC/WSJ poll). Why was there decrease in MM today. Nate Silver also dropped 0.3% in his forecast. One other site I follow http://www.cabpolitical.com/electoral-forecast/ also mentioned a small decrease in Obama’s national lead. When a slow day like today makes changes, it is very unnerving

    • Ms. Jay Sheckley

      New polls are loaded in as frequently as they arrive. The numbers go up and down, but in the main seem to trend upward. Do not be unnerved; step back a see: The same candidate has been ahead the entire election.
      I recommend you stick to this poll aggregate, going deep here rather than spread yourself across several sites. Sam does the best job of looking at everything so you don’t have to. not only will yu be better informed, you’ll be less anxious.

    • Matt McIrvin

      Also, as a further cautionary note: Sam’s only just started trying to make a Nov. 6 projection this year. It’s an experiment. I’d trust the day-to-day Meta-Analysis more than his projections, because the Meta-Analysis is what has a track record of great success. Of course, on election eve the two should converge.

    • wheelers cat

      correct me if Im wrong, but wasn’t PEC accurate to a single EV last cycle, while Nate whiffed by a whole state?

    • Matt McIrvin

      Yes, though I think he had to do a little bit of adjustment from the raw Meta-Analysis results to make that guess. But he didn’t make an Election Day prediction last cycle until the very end of the race, just “if the election were held today”.

    • MAT

      Well, in fairness to Nate, he did make 53 individual predictions (50 states, DC, Nebraska-Omaha district and national overall) while Dr. Wang made 1. And Nate’s miss was Indiana, which was very close. So while I prefer the PEC methodology, I wouldn’t beat up on Nate too much here for his 2008 results ;-)

    • badni

      Apples to oranges.

      Nate calls individual states as well as doing a probablistic EV count and Sam only does the probablistic EV count.

      Nate called Indiana for Mccain and it barely tipped Obama. Sam made no comparable prediction so how can you say Nate did worse on that? I suppose you could go and look at Sam’s final map and see which states were leaning which way, but that would not be fair to Sam since that was not the point of his method.

      Sam and Nate are two honest and smart statistical analysts. I am sure that if they were competing they would want honest and smart comparison metrics (which would have to be established in advance) rather than cherry-picked, asymmetrically-tested “gotchas”.

    • wheelers cat

      not apples and oranges.
      the Nowcast is EV votes.

      c’mon, admit Nate has been infected with market forces this cycle. hes pandering half the half the time so the grey lady can harvest pageclicks from the FOXnews fanboiz.

    • badni

      You weren’t referring to the nowcast. You said he whiffed on a state.

      I understand Nate has fallen out of favor with you, but that’s no reason to start playing a shell game with your comments about him. Given the subject matter and gist of your criticism, your slippery answer is especially ironic.

    • wheelers cat

      Its my remembrance Nate called Indiana wrong.
      The nowcast is state based polling isnt it?

    • wheelers cat

      and this ironing is delicious.
      ;)

  • Darren

    I find your work extremely interesting Dr. Wang and I hope your current odds of Obama being re-elected of 90% turn out to be correct.

    I am therefore assuming that you are going to bet a small fortune somewhere on Obama winning since he is currently just under a 2/1 favorite to win?

    • Ms. Jay Sheckley

      Dr Wang says gambling is a vice. The odds on In-trade favored him better months ago, but instead Sam’s interest lies in science. He is a professor of neurology at Princeton, and willing to share what he knows about numbers and probability. This is our gain and no one’s loss.

  • Jun Talabucon

    At this point…,

    There seems to be a consensus that the most competitive swing states are nine(9), namely: WI, NV, IO, OH, NH, CO, VA, FL & NC.

    And if most forecasting models aggregate polls, why not aggregate different forecasting models as well?

    Here’s my take, let’s aggregate (average) Pres. Obama’s chances in the following forecasting models, specifically, for these nine (9) states:

    http://fivethirtyeight.blogs.nytimes.com/2012/10/20/oct-20-calm-day-in-forecast-but-volatility-ahead/

    http://www.predictwise.com/politics/2012presidentstates

    http://research.uvu.edu/DeSart/forecasting/october/1021.html

    By tracking these models separately on a daily basis, the (big) similarities and (little) differences become more apparent… while the daily average paints a better picture.

    Here’s what I see, with due respect to Prof. Wang’s quadrillion more possibilities:

    WI, NV and IA are about 3:1 odds (>75%-ave.) for Pres. Obama; translated to EVs = 259 votes.

    Now, combined these with either/or:

    1.) OH = 18 EVs
    2) NH (4) + CO (9) = 13 EVs
    3.) VA = 13 EVs
    4.) FL = 29 EVs
    5.) NC = 15 EVs

    Five (5) different/short paths to more than 270 EVs and Pres. Obama being re-elected as President of the USA…

  • Debbie

    Excellent site Dr Wang. Anxiously watching this entire election unfold from New Zealand.

    We might be unable to vote, but that does not dampen our enthusiasm or support for your current President. Mr Romney was never considered a serious contender, still isn’t, and I can assure you there are some frayed nerves around the place that our worst nightmare may unfold (to say we are unimpressed with Mr Romney aligning himself with the likes of John Bolton is an understatement).

    Following this site and reading the comments is a great tonic. We are all crossing our collective fingers that President Obama can get this thing done. I was getting slightly manic, constantly refreshing pages whilst following all the polls. Thank you for the sanity and perspective.

  • wheelers cat

    I still dont understand the irrational exuberance of the conservatives. Why are they giddy?
    Is it magical thinking or are they simply incapable of evaluating logic and facts that are contra their belief/desire edifice?

    • Olav Grinde

      I would think much depends on your news source. Three guesses as to the spin if your perception of reality is filtered through Fox News…

    • LiberalActuary

      Using the Socratic method of learning? :-)

    • LiberalActuary

      That question was for wheelers cat, btw.

    • wheelers cat

      Olav, it is logical that poll aggregates would give a more accurate picture than singleton which could be outliers.
      Yet, like Dr. Wang pointed out, it is the conservatives that are giddy and optimistic.
      Mathematically and statistically Obama’s odds are far better than Romney’s….6:1 from Dr. Wangs calculations.
      So LiberalActuary… from a game theoretic perspective…does this mean that conservatives are bad at games? Because bidding theory should mandate reduced optimism on those odds…. or does it mean that they are good at evolutionary games, in that privileging emotion over reason is a fitness advantage for red phenotypes.

    • THE

      Problem is, you keep thinking in terms of two populations WC, instead of a distribution, with lots of people in the middle who swing one way then the other.

      Romney won a lot of people over, in that first debate, who previously thought he was hopeless and out-of-touch. They have been taking him more seriously ever since.

    • wheelers cat

      Disagree THE.
      you are the one thinking the underlying structure of carbon-based reality is gaussian, with red and blue phenotypes neatly conforming to a symmetrical normal distribution.
      try Benoit Mandelbrot instead.

    • THE

      @Wheelers
      I just think there’s a lot of people in the middle who don’t have any fixed political commitment at all. To me, your thinking sounds not so much fractal as binary.

      Some voters have special interests that tweak their decision. I am very uncommitted politically myself. I have switched vote preference in the past on single (but critical to me) issues. Often I don’t know until quite late in a campaign, which party I will vote for.

    • wheelers cat

      @THE
      ahh, but they do have a fixed GENETIC position.
      The biological basis of behavior.

    • THE

      It’s not as simple as that. It’s not really genetic determinism. For most social factors, genetics explains only a part of it.

    • Reason

      NJF, not sure what this poll shows. Is it shifts from R to O and vice versa? If so, is it good or bad for O or R?

    • Ohio Voter

      This is showing that more Romney voters are going to Obama and fewer Obama voters are going to Romney, bucking a trend from the first debate where more Obama voters were going to Romney than vice versa.

      If we’re looking at this as “good vs. bad” for Obama, it would have to be classified as “good”, but more so because the trend is swinging his way again.

    • Reason

      OV. I figured as much. Too early. There is too much blood in my caffeine system. Many thanks.

    • Olav Grinde

      Yes, the Rand swing to Obama is good. First time in three weeks that he’s had a net gain. The widening gap (now 5%) in likelihood to vote is not so good.

      I think we’ll be seeing some interesting movement in the Rand poll in the days ahead.

    • Matt McIrvin

      It could well be that voters newly swinging to a candidate have a lower likelihood of voting, since they’re closer to being undecided/disenchanted.

  • Reason

    I just read Nate Cohn’s article about Obama’s apparent struggles with Likely Voters. Yet he leads in Registered Voters in most polls. My question to Dr. Wang and anyone else who follows this site, is does the MM take into account this scenario regarding LVs? Thanks.

    • Matt McIrvin

      The meta-analysis uses LV polls whenever it can, I believe. That means that LV screens will affect it. However, he doesn’t use the Gallup national tracker, which has the most dramatic LV screen effect.

    • Olav Grinde

      Dr Wang doesn’t use any national trackers.

  • Rick

    MM gonna tick up big time. New CBS/Quinnipiac poll of Ohio!

    • bsk

      Indeed, there should be an absolutely massive, positively gargantuan increase, the likes of which Ohio has never seen! From 1% to… 2%.

      :)

    • bsk

      And yes, the MM should rise above 2% again :)

    • Reason

      I just went back to the history of the MM in regards to the polling. Anyone else notice that R seems to poll a little better when it is a Saturday/Sunday poll? Is it because they catch more R’s at home with a land line? Dr. Wang’s probabilities has me intrigued, like a Black Jack player trying to figure the odds.

    • Reason

      I suppose this is good and bad. The last poll they did had O up 10 as opposed to 5.

    • Matt McIrvin

      I think at this point we’re getting so emotionally invested in individual polls that we’re seeing ghostly patterns emerge out of random noise.

    • Slightly Skeptical

      Two new polls have Ohio tied: one by Suffolk and one by Angus Reid. Median should be back to O+1.

  • Olav Grinde

    Excellent Ohio poll from CBS/Quinnipiac today, with Obama +5%. Glad to see more non-Rasmussen polls from that key swing state.

  • Olav Grinde

    Anyone know whether CBS/Quinnipiac will be releasing polls from other swing states?

    Would love to see their figures from Florida, New Hampshire, Colorado, Nevada, Wisconsin, Virginia and North Carolina, Iowa and Pennsylvania!

    • Tom Gavin

      Muhlenberg +5 Obama. Considered a good local poll.

    • Olav Grinde

      Pennsylvania? Was this poll conducted prior to the second Presidential Debate?

    • ChrisD

      The Muhlenburg’s poll dates are 10/17-21. The five polls that figure into Dr. Wang’s current PA average are O+5, O+7, O+4, O=4, and R+4. That last poll (10/11-13) is the only PA poll running back through July that has Romney in the lead. It’s by Susquehanna for the Republican State Committee.

    • Tom Gavin

      Just out

    • Tom Gavin

      Not a bad result for Obama considering some local talk that it”s been tightening.

  • Paul

    I have a question about Ohio versus Colorado.

    Conventional wisdom is that the median electoral vote, and therefore the Presidents clearest path to re-election lies in Ohio. Indeed, if you give Obama the uncontested blue states plus Wisconsin and Nevada, where he seems favored, that would leave him with 253 electoral votes. The 18 from Ohio would do the trick. But another path would be the 17 from Iowa+NH+Colorado.

    Is Ohio really an easier lift than Iowa+NH+Colorado? My understanding is that the Obama campaign is fairly confident in Iowa+NH, though some polls have shown them close. Which leaves Colorado, where the RCP poll average is Romney +0.2. However the Colorado polls underestimated Obama by 3.5 in 08 and again Bennet by 3.9 in 2010 (when the RCP average was Buck by 3).

    It is well known that Nevada polls underestimated both Obama in 08 and Reid in 10 by 8-10 points each time. Is there a smaller scale version of the same phenomenon in the somewhat similar Colorado?

    Another reason to think Iowa-NH-Colorado might be an easier lift is that these are all states Obama won by 9+ points last time, whereas he only won Ohio by 4.8.

    Indeed, the path I describe (uncontested states plus Winsconsin, Nevada, Iowa, NH, Colorado) consists of exactly those states Obama won by 9+ last time, and is gof 272 electoral votes.

    What do you guys think, Ohio, or Iowa-NH-Colorado?

    • Matt McIrvin

      My gut feeling is that NH really is tied up. Maybe it’s just because the part of NH closest to me is a forest of Romney yard signs.

      …Which raises a point. Everything I’ve read about political campaigns suggests that yard signs are basically useless in a presidential race, since the main utility of the things is to get name recognition for obscure local candidates.

      But it seems to me that the Romney campaign (and very much *not* the Obama campaign) has gone all-out blasting yard signs all over the country in the past couple of weeks. They’ve been going around whole neighborhoods unsolicited, asking people if they can put these signs on their yards.

      Meanwhile, Obama’s campaign is spending the money on actual GOTV instead, which what experience says is more effective.

      But I think the Romney campaign is trying to create the appearance of an enormous Romney/Ryan groundswell, and I think it’s working: every Democrat who feels depressed about the presidential race cites these huge swarms of yard signs that they see everywhere. I wonder if a yard-sign campaign on this level might actually be effective as psyops aimed at depressing opposition enthusiasm? I suppose it is, but there’s the question of whether it’s the best operation for the dollar.

  • Beaucon

    Wheeler Cat,:

    Your use of the term phenotype and the formula that follows it leads me to understand that you are referring to some well established research that, as a novice, I am not aware of. Do you have a link that could get me up to speed.

  • Rick in Miami

    I like looking at both sites. To my point of view, the strength of Sam’s approach is that it makes as few assumptions as possible. When I wonder about house effects, the timeliness of state polls, etc., I look to see differences in the 538 forecast since Nate Silver makes lots of adjustments and corrections to the data – examples include synthesizing the national and state polls, extrapolating an election day forecast also using economic data, adjusting to remove the average convention bounce, etc.

    As a result of all those assumptions at 538, the error bars are enormous – not a lot of confidence in that forecast there! Alternatively, a reader here can fret about the timeliness of state polls and the “lost” information buried in national polls, but there isn’t a black box churning out the forecast. When both sites are showing a similar multi-day trend, I have confidence in that trend’s reality.

  • RDT

    Something that worries me: I went back and looked at the 2010 results and saw that the individual district polls painted a more favorable picture for House races than the overall national polls, and more favorable for Dems than the final outcome.

    Could the same thing be happening in state vs. national polls now?

  • Michael Worley

    MM will be below 1 at 1 barring last minute OH poll showing O+2 or more

Leave a Comment