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

Post-debate reaction

October 3rd, 2012, 10:33pm by Sam Wang

Seems like Romney did well. Landed some good punches, didn’t look like he was circling the drain. Obama was more natural and got off a few…zingers. To tell the truth, I started zoning out after about 30 minutes.

It’s not enough to change the basic dynamics. Stalemate. I suspect the race will stay where it is now, maybe narrow by a point. Two points if we believe the pundit reactions. This is not enough for Romney. The Meta-Margin is currently at Obama+6.2%.

I do wonder if it will help the GOP downticket. Despite your obsession with the Presidential race, readers, the downticket is far more interesting in the coming weeks. Those races are at center stage (ActBlue / CrossroadsGPS).

Update 1: Commenter JohnJacobs asks about undecided voters. They are now 5.0 +/- 1.0% (n=13 polls, Read my previous take on undecideds from a week ago – about halfway down the essay. Also see my 2008 essay on how undecideds break. On average, the net benefit to Romney from undecideds will be 0.3 +/-1.1%. If he matched the largest recorded break (3:1 for the challenger), he would gain 2.5% of margin.

Update 2: On this comment thread anything goes, including opinions, glee, panic, whatever. Better to channel it, though.

Update 3: The pesky issue of falsehoods arises. See Krugman and Digby. The “Romney-lied” backlash to “Romney-kicked-a**” begins…

Tags: 2012 Election · President

124 Comments so far ↓

  • BillSct

    Perhaps last night was “Rope-A-Dope” in round one. Certainly there is plenty of fodder for the Ad men. Stewart and Colbert are a must watch tonight to see what they do with the new Mitt versus the old Mitt. I had to take a phone call about twenty minutes in so I missed most of it. My wife, who is definitely not a political junkie, but who always votes, got bored and changed channels. Anybody know what the ratings look like? Maybe other people got bored too.

  • KEL

    What about fundraising. How do the debates effect donations? How do political spending, fundraising, advertising effect the polls. What about the super PACs?

  • Brad Davis

    I didn’t feel that Obama did as badly, certainly not as badly as the media would have you believe. To invoke a football analogy, don’t play like you’re losing when you’re winning. Taking your chances on dramatic ‘hail mary’ plays may excite the crowds, but they only invite disaster in the form of unexpected turn overs. Instead, keep putting the football in the hands of your running back and run the clock down. It seems to me that is exactly the strategy that Obama and the DNC took. If Obama had been too combative and aggressive he would’ve risked looking like an ‘angry black man’.

    Romney, on the other hand, is fighting from behind. His only option for winning is to continue throwing ‘hail mary’ passes. As it stands, Romney has already lost the election, things are not going to get a lot worse for him. He has a lot to gain and very little to lose if he’s a little fast and loose with the facts. And just like in football, it can be a bit frustrating when you see the opposing team moving down the field, putting points and slowly narrowing the lead- but you’ve got to play smart and take time off the clock. Let the other team gain meaningless yards and focus on winning the game.

    • Richard Vance

      Excellent analogy. Don’t make mistakes and run the clock out. However I think this ignores the down ticket races. The President needs a Democratic congress to avoid stalemate so I would like him to talk over Romney’s head towards the Republican party planks, point them out and make Romney either accept it or deny he’s a loyal Republican. Whatever Romney chooses he loses and the downticket campaigns can use that in their ads.

    • Brad Davis

      @Richard, I agree- the approach (and the analogy) totally ignores the down ticket races, and the down ticket benefits are most likely to be felt on domestic issues rather than foreign policy (the upcoming debate). I note that the the Meta Margin has dropped by a huge amount this morning.

      Honestly, I don’t know why Obama isn’t trumpeting more of his accomplishments. The Cash for Clunker’s program reduced America’s dependency on foreign oil substantially- it reduced oil consumption levels down to 1997 levels, which is huge! It means fewer dollars are being sent over seas to the oil producing nations! I think a similar argument can (and SHOULD) be made about ‘green’ solar and wind energy. Yes, that energy costs more, but the US economy is capturing a much larger fraction of those dollars than energy from oil imports. That means more good paying jobs for Americans.

  • Michael S

    Ipsos claims that Obama is holding steady, a four-point swing of undecideds going to Romney with a one point swing of undecided to Obama

    The net gain is a two point swing to Romney.

  • pechmerle

    I take Sam’s point that we are just entertaining ourselves with our conflicting impressions of who “won” the debate. And I have previously taken his advice to “channel it” by making a few contributions to knife-edge down ticket races.

    Still, I call your attention to this:

    “Lingistics professor George Lakoff, well known for his work on how politicians use language effectively or not in making arguments, has analyzed President Barrack Obama’s performance in last night’s debate and he’s not impressed. Here’s what he published on today:

    You don’t win a presidential debate by being a policy wonk. Obama violated all the basics of presidential debating.”

    As Xian, and I, commented a couple of days ago, facts have little to do with campaigning. Instead, you have to concentrate on setting the “frame” of the discourse favorably. Obama’s team has actually been doing this with some success since Romney became the nominee (defining Romney negatively on Bain, taxes, an out of touch fat cat). But then Obama the candidate did almost none of that last night.

    It’s frustrating for two reasons, neither of which is he will lose the election: (1) It hurts in the down ticket critical races. (2) It is the mistake he has been making in governing in most of the past three years.

    You want to talk opportunity cost: If Obama’s personally weak performance in these six weeks contributes to not having a Speaker Pelosi after Nov.7, a huge opportunity will have been lost.

    • wheelers cat

      There is going to be pushback against Romney’s epic dishonesty on stage.
      And I’m unimpressed with Lakoff. Demagoguery has been around since Kylon and Pythagoras.
      Half the population is below the mean of the bell curve of IQ.
      Obama is a gamer.
      He is very risk-adverse, and at the debate he was playing min-max. The left thinks he should have been playing tit-for-tat.

    • kel

      In 1994 Ann Richards, then governor of TX, debated challenger G W Bush. A great governor and charismatic speaker capable of dishing out her fair share of zingers, she roundly thrashed Bush on policy issues. Despite this Bush succeeded in framing him self as the champion of conservative values.

      If conservative policy is a form of addictive behavior, then Texas was the gateway drug. I believe the likelihood of GW winning the presidency without having first been governor would have been low.

      Is there some way of using the meta-margin to test Lakoff’s framing hypothesis?

  • Ms. Jay Sheckley

    The Xbox Debate Poll. Sorry I’m late, I just heard about this. One article I saw said Romney won– Some don’t mentin winning. Maybe gamers don’t see that as the big issue in “THE MOST TWEETED POLITICAL EVENT EVER”. [More than Iran? ]
    All figures and statements below from link at bottom:
    Xbox Live polls show 59 percent of viewers would vote for Obama if the election where that day, as compared to 27 percent for Romney.
    Spot polls consistently showed that more than half of those watching felt Obama’s positions on issues debated were more in line with their thinking.
    A suggestion that Romney be pushed for details regarding what programs or services he would eliminate to ”pay for tax cuts for millionaires” was re-tweeted more than a thousand times.
    Those making comments appeared to seize on points or counterpoints supporting their sides, Romney seeming to ”bully” the moderator, Obama appearing weary.
    ”This Romney dude is just rude,” one tweeter chirped .
    Google xbox debate or use this link:

  • wheelers cat

    Poor Nate. Obama’s win prob has improved to 87.1% percent in the post debate forecast, the high of the season so far.
    Hilarity ensues.
    He is getting a beat down on twitter because conservatives believe Obama should be plummeting . Never mind that he doesn’t use flash polls or Luntz focus groups in his model.
    More empirical proof that conservatives don’t understand either statistical forecasting or polling.

  • Matt McIrvin

    We’ve got Thursday polls from WeAskAmerica showing Romney ahead in Ohio, Virginia and Florida (and Rasmussen polls showing Obama barely ahead in Ohio and Romney ahead in Virginia).

    WeAskAmerica seems to have an R house effect, but a lot smaller than Rasmussen or Gravis. Other polls aren’t post-Wednesday yet.

    So I guess we have our delta function effect: 2 points to Romney in the Meta-Margin, Virginia going to Romney and Florida softening.

    I’m not convinced this is the end of it, though; since this is just the effect of a few polls so far, it seems much worse than Prof. Wang is predicting. If other polls follow WeAskAmerica and we don’t have a rapid reaction over Big Bird or what have you, we could see Romney ahead in the EC within the week.

    Post-debate memes are already established. Anyone care to speculate about further strengthening, decay lifetime, etc.?

    • Paul

      I would caution against reading too much into the We Ask America numbers.

      For example, in June WAA had Romney +5 in Virginia, which is tied with a Gravis poll for the highest Romney margin all cycle, among over 50 polls in the HuffPo database. The 3 previous and 3 following polls by date in the HuffPo database all had Obama ahead, by an average margin of 4 points.

    • wheelers cat

      The jobs report is going to step on Romney’s debate bounce. I think Obamas win prob is going to break 90% in the 538 forecast, because while Luntz focus groups and flash polls are not part of his model, job reports are.
      RAND is showing the opposite of what you are predicting.

    • Froggy

      I would also caution not to read too much into this. The Meta-Margin was poised for a big drop after its long run-up, as I noted in my comment yesterday in this thread. Even if the only poll this morning had been an Obama +4 in Ohio, I think we would still have seen a 2% drop in the Meta-Margin. This looks to have been a very odd situation where a single poll could cause a big shift in the Meta-Margin.

      In addition, today’s releases have been a virtual who’s who of Republican-leaning pollsters (Ras, We Ask, and Gravis). Together with the unusual situation of the Meta-Margin tottering at the edge of a cliff, it’s been kind of a perfect storm for movement of the numbers.

      PPP is indicating that the initial results in their ongoing new surveys show a 1-pt shift for Romney in Va, and a 2-pt shift for Romney in Wisc. If that keeps up, then that would be Obama +5 in both states. Let’s wait and see.

    • Joel

      McLoughlin and Associates dumped several more polls (using 300 voter samples (!!)) to thumb the scale, too.

    • Matt McIrvin

      True, the Rand poll shows almost nothing happening as a result of the debate. There was some insignificant net switching toward Romney right before, and the situation after is basically the same. Not sure why the election prediction bounced upward.

  • Some Body

    The meta-margin has very dramatically dropped from more than 6 to 3.72 in one update. Is this the effect of the debate? My guess is no. It’s the effect of Rasmussen and We Ask America (both robopols with a clear R lean) releasing swing state polls before the rest got to it.

    I wrote in another comment here about the poll results in swing states having a “hole” in the middle, which makes the median “jump” a bit too wildly (while an average would move more smoothly). This looks to me like another consequence of the same phenomenon.

    PS: I saw that over at 538, wheelerscat wrote a comment, saying that with the jobs data, Nate’s forecast is going to move in the direction of the now-cast and the PEC. Well, the movement appears mutual ;-)

  • Olav Grinde

    Froggy: “In addition, today’s releases have been a virtual who’s who of Republican-leaning pollsters (Ras, We Ask, and Gravis).”

    That’s quite an interesting coincidence…

    I’m keeping my head cool until we see a significant number of post-debate and post-latest-jobd-report polls from a broad range of swing states.

  • Steve in Colorado

    Interesting that the RAND survey went back up for Obama on the 4th. I was thinking about the analysis I have been seeing recently that Americans are more optomistic about the economy lately, which has been laid at the feet of feeling better because they support Obama. Well, if we had an extra 400,000 jobs added over the last year that went unreported, was it possible that the Americans felt this and therefore that their feelings about the economy were actually more accurate than the data we were relying on?

    • Matt McIrvin

      Also, remember RAND only surveys one-seventh of their population on any given day, so any effect is going to be smeared out over a week.

    • Sam Wang

      Yes, but the onset of the effect should be sharp and easily visible.

    • Matt McIrvin

      It’s also not clear to me from a first read of their description if the Oct. 4th point actually comes from post-debate responses, or if we’ll have to wait for tomorrow for that.

  • MAT

    Ras sample sizes were 500. I would wait a few days and a few more polls before drawing any conclusions.

    • wheelers cat

      Remember the conventions?
      First Ras put a post saying conventions dont matter, then pumped out a bunch of polls showing a modest bounce for Romney.
      Last week, he put out a post saying the debates dont matter, then hey presto! a sharp bump for Romney.
      He doesnt have to put a thumb on the scale, his crappy LV model underweights cell phone demographies. That is why he whiffed in 2010 in Nevada and Colorado.

  • ed groome

    The Meta-margin has been posted as nearly half of yesterday’s. Any particular explanation? Typo?

    • Sam Wang

      Automatic. Also, the calculation is quite resistant to individual errors. However, see other comments re R-leaning pollsters – three of them dropped data at once today. Clever of them.

    • Olav Grinde

      Indeed. There is of course the remote possibility that those R-leaning polls become self-fulfilling, by driving the public discourse.

    • wheelers cat

      Interestingly 538’s forecast and nowcast are pretty much unchanged. I dont think Nate incorporates flash polls or Luntz focus group data into his model.
      What does factor into 538’s model is the decline or increase in jobs.
      I bet 538s Obama win probs go up tomorrow on the jobs report.

  • Olav Grinde

    NBC has a very interesting analysis of 1.3 million social media comments. It’s quite interesting. As fact-checking began to play a role, the perception of “who actually won the debate” changed significantly.

    The article’s charts give a striking indication how perceptions changed as the candidates’ statements were weighed against the facts.

    Social media analysis: Who really won the debate?

    • Sam Wang

      Romney did what he had to do. His only chance was to appear as if he had won. This involved saying whatever it would take to be seen as the winner. And he did well. As you point out, time will tell.

  • Ed Groome

    The most recent posted item is not accepting comments at present, so I’ll put one here. I’m new, so is there a place on the site where I can educate myself as to which are the R-leaning and which are the L-leaning polls?

  • Ed Groome

    Thanks, and that’s a helpful summary. I’m a bit confused by your no-truck comment, though, given that you raised the subject. In any event, it doesn’t surprise me that three R-leaning polls would emerge simultaneously if by their lights they considered the debate would be Romney’s last throw of the dice and were thereby prepared to release quickly. If Obama ends up needing to do more than run out the clock, it will be interesting to see if Pew, PPP et al are similarly aggressive.

    • wheelers cat

      You are assuming a blue bias for Pew and PPP.
      I doubt that is true.
      OTOH it is empirically evident that Rasmussen has a red bias.
      Otherwise he would have changed his LV model after whiffing on CO and NEV in 2010.

    • Sam Wang

      Probably not PPP at least. See Jackman.

  • Ed Groome

    wh. cat, I am not assuming anything, simply referring to Dr. Wang’s reference to the “house effects” over at nytimes:


    I would hope that there are D-leaning polls in the PEC metrics to balance out the Rasmussens, otherwise there would be a rightward bias to the Meta-margin, surely? Otherwise stated, we could simply ignore the effects of the three new polls if there were no compensatory data. I imagine I’m understanding the thing imperfectly, but it seems an interesting question to me.

    • Sam Wang

      I take ’em all. Give it a few days. I think the PPP drop is tomorrow.

    • wheelers cat

      It doesnt work like that, Ed.
      All information is valuable. And if the sample size is large enough there isn’t even need to use nonparametrics. But unlike weather forecasting, which has null effect on the weather, political forecasting does affect politics. Enthusiasm, fundage, etc.
      Nate believes like you do, that red/blue house effects cancel each other out. He uses parametric statistics. He believes in symmetry (/sneer).
      I think the way Dr. Wang does it is better….because he isn’t forcing symmetry. How do you set the midpoint of no house effect?
      Its arbitrary, and its arbitrary to assume symmetry.

Leave a Comment