Princeton Election Consortium

A first draft of electoral history

← Stat-urday: Lombardo’s Quadratic Molasses Michelle Obama, the Great Persuader →

The GOP convention negative bounce: a final look

September 7th, 2012, 6:23pm by Sam Wang 54 Comments -->


Tweet
(Greetings Political Wire, Andrew Sullivan, Reddit, Fark, & others. For what our low-noise approach recommends for your Dem/GOP donation, click here.)

Soon, post-DNC state polls will begin to arrive. So here is the best glimpse we are going to get of the negative post-GOP-convention bounce. Basically, their convention appears to have helped…Obama. It’s the part in red at the end.

spacer

Some notes and caveats…

As always, this EV history is viewed through state polls only, using Electoral College mechanisms. Therefore it is most sensitive to swing-state changes. In practice, changes in this index resemble changes in national polls. Since the state polls sample more voters, the Meta-analysis has considerably less noise.

The horizontal axis reports the date that polls were released, and therefore when they entered the Meta-analysis. This accounts for the 2-3 day delay for a news event to exert its effect. I have color-coded the events for clarity.

The following events appear to be notable: the Bain Capital/tax return attacks, the Ryan VP pick, the Todd Akin “legitimate-rape” comment, and the last day of the Republican National Convention.

The negative GOP bounce. As I stated before, the GOP convention was of no help to them in the Electoral College. Indeed, it appears that the race shifted towards President Obama by 6-15 EV, or about 1.0% of Popular Vote Meta-Margin. From an analytical perspective, a negative bounce is quite remarkable because all the talk in recent weeks has been of bounces being smaller or zero, but always in the hosting party’s favor. It is all the more remarkable because of the relatively small number of state polls over the last week, so that the Meta-analysis’s inputs have not fully turned over (for discussion see comments). So the negative bounce may be larger than what is shown in the graph. Such an event would have been missed in past years (and even this year) because national polls don’t have the best resolution.

The natural question arises: why would the Republicans be hurt by their own convention? Two answers come to mind.

(1) The Ryan-VP bounce effectively used up whatever room there was for a bounce. This year, opinion seems to be fluctuating in a very narrow range: Obama up by 1.0-5.0%. Maybe there was no room for improvement.
(2) The GOP convention was not particularly inspiring. Indeed, the most notable event was Clint Eastwood’s empty-chair routine, which overshadowed Romney’s acceptance speech.

Bain Capital, tax returns, and “you didn’t build that.” As I noted at the time, the Bain/taxes attack appeared to be very effective. I note that July 13th, which is nearby, was when President Obama made the “you didn’t build that” remark. So whatever we think of that remark, we can safely discard the hypothesis that it was of immediate help to Romney.

The Ryan VP pick and Todd Akin. The pick of Paul Ryan on the GOP ticket led to major media saturation, and positive attention for Romney/Ryan. This drowned out other stories. As I pointed out at the time, the political media seems to be unable to cover more than one story at a time. This could account for the large, sudden shift. The Ryan bounce ended at the same time as Todd Akin’s remark that “legitimate” rape victims don’t get pregnant, which either hurt Romney directly, or just reset the media’s mindless attention.

And now we wait for the post-DNC bounce. I wrote the other day that given the narrow band of opinion movement, I will be surprised to see the bounce to go above 5.0% in Meta-margin or 330 EV. On the other hand, it was a very strong convention. Also, the negative RNC bounce is not fully apparent in the Meta-analysis. So perhaps there will be a big shift next week after all.

Tags: 2012 Election · President

54 Comments so far ↓

  • spacer Xtalographer // Sep 7, 2012 at 7:38 pm

    Prof. Wang-

    I wish your blog were required reading for media types who cover polls or write about them.

    Anyway, we know that there are far fewer polls this election cycle than last time around. How much of an effect is this having on the sensitivity and confidence of the meta analysis?

    Reply
    • spacer Sam Wang // Sep 7, 2012 at 8:29 pm

      Xtalographer – Thank you. It’s a good question about the low density of state polls.

      My frank opinion is that aggregators flowered under boom conditions in 2004-2010 — so many polls. As a result, many of them did not develop skills to extract maximum information from what was available.

      The Meta-analysis is still remarkably low-noise. However, it doesn’t always reflect the full change. For example, right now it probably reflects only about half of the true post-RNC change. Per state it uses the last 3 polls or 1 week’s polls, whichever is greater. In key swing states, RCP shows polls ending after 8/30 in the following quantity: 1 in CO, 2 in FL, 1 in NC, 1 in OH, and 0 in VA.

      Response time right now is up to two weeks to get a full effect, less time to see the direction of change. It showed the Ryan bounce with the same speed as Palin 2008. So it’s still fairly good.

      Where the low poll density really hurts is in Senate and House races. That has been a bit frustrating.

  • spacer Steve16748 // Sep 7, 2012 at 8:55 pm

    I very much enjoy finding this site. So I supposes, the best I can do is get my personal business done by Oct. 15, (including a corp. tax return), fly into Las Vegas, file for an absentee ballot and vote for Obama. (Actually, I was in Las Vegas for a trade show in August, and except for the time I was out at the beautiful Valley of Fire State Park reading petraglyphs, I was bombarded by ultra conservative messages on the video, (including the Ceasars Corp. internal feed at the greatly dimminished Rio. I remember how Harry Reid beat the poll odds in 2010. Thanks to God red blooded Americans are robust in Vegas and blow away the turnout models.

    Reply
  • spacer wheelers cat // Sep 8, 2012 at 10:24 am

    well here comes some new polling.
    It will be interesting to see if this weeks battleground ad-blitz can change the seeming immutable spread. You can bet there will be many polls commissioned to test the permeability of the electorate to this big-money message.
    andrewsullivan.thedailybeast.com/2012/09/ad-war-update-7.html
    generic ad formula–>
    “Here in [state], we’re not better off under President Obama. [list of problems Obama is implied to be responsible for].
    Romney’s plan? [list of generic GOP solutions], create [number] new jobs for [state].”
    This is a test of two Team Romney hypoths, really.
    H1: Can post citizens united spending make a difference?
    H2: There are enough undecided voters left in the battleground states to change the EV maths.

    cant wait for fresh state polling, for battle ground states at least.
    this is a Koch-funded attempt to step on Obamas convention bounce.
    Interesting times.

    Reply
  • Romney Got a Negative Bounce | Hotspyer – Breaking News from around the web // Sep 8, 2012 at 1:01 pm

    [...] Princeton Election Consortium finds Mitt Romney actually got a negative bounce coming out of his [...]

    Reply
  • spacer Tapen Sinha // Sep 8, 2012 at 1:08 pm

    https://mmicdata.rand.org/alp/?page=election#shifts-between-candidates
    Look at the scale of vote switching. Basically one percent. And this is VERY
    different from other polls as they track the SAME people over a long time.
    Bottom line: People have made up their mind and the model of Sam Wang
    rules.

    Reply
  • spacer Robert // Sep 8, 2012 at 1:33 pm

    Here is another proof that Obama has got the real bounce after convention. Look at this chart and latest polls: www.pollheadlines.com/obama-vs-romney-polls.html

    Reply
  • Romney Got a Negative Bounce // Sep 8, 2012 at 1:45 pm

    [...] of America Stadium » Romney Got a Negative Bounce Originally Posted September 8, 2012 The Princeton Election Consortium finds Mitt Romney actually got a negative bounce coming out of his [...]

    Reply
  • spacer D Yancey // Sep 8, 2012 at 2:31 pm

    Excellent work! Makes an ancient econometrickster proud…

    A dumb question:

    Why is Ohio included in your “as of today (sep 8th)” EV projection (of 309) but shown as a white toss-up in the state-by-state map?

    Reply
    • spacer Sam Wang // Sep 8, 2012 at 2:38 pm

      Indeed why! Read FAQ please. Probabilistic model, very precise EV estimate without specific state assignment.

  • spacer Scott // Sep 8, 2012 at 2:38 pm

    As you alluded to, TV journalism is so DUMB now. Makes anyone who bathes in it, with any channels or programs, relatively less well informed and less intelligent. Blogs and articles like this, and your competitors ;) make people smarter and more well informed.

    Reply
  • spacer Tapen Sinha // Sep 8, 2012 at 4:05 pm

    Sully sees the light. He should have paid more attention to Sam’s website!
    andrewsullivan.thedailybeast.com/2012/09/have-the-conventions-changed-the-race.html

    Tapen

    Reply
  • Have The Conventions Changed The Race? | Mental Health BlogMental Health Blog // Sep 8, 2012 at 4:05 pm

    [...] I.e. GOP-FAIL. Sam Wang who analyzes state polling and thereby the electoral college vote has a fascinating graph of the shape of the race for the past few weeks. Here's how his Electoral College measurement [...]

    Reply
  • spacer Jack Moss // Sep 8, 2012 at 4:28 pm

    Of course for Romney there is history. macsmind.com/wordpress/2012/09/according-to-gallup-romney-should-win-the-election/

    Reply
    • spacer Sam Wang // Sep 9, 2012 at 12:18 am

      Jack Moss – the statistic you cite in favor of your conclusion is a single measurement, 47-46 Romney-Obama. Statistically, there is an approximately 40% chance that it sampled from a population that in fact preferred Obama. So I would not get too excited.

      The Obama win probability of 88% on my site is based on a similar argument that you propose, but on a far larger quantity of data. This will be a close election, but it is unlikely to go in the direction you suggest.

  • spacer D Yancey // Sep 8, 2012 at 5:16 pm

    (Head bowed in FAQ’d up shame):

    Ah, probability, not the mundane bean-counted reality that your 60% or better blues add up to 291, and 309 if Ohio’s 18 are added.

    Clearly, a (non-Sandra) fluke.

    Thanks!

    Reply
  • spacer badni // Sep 8, 2012 at 6:02 pm

    Jack Moss- putting aside everything else that’s wrong about that article (which is nearly everything), where does that 95% come from? Is yhe author asserting that 12/15=.95? And if the history is 12/15 how are there only two that do not follow the pattern? That article is so weirdly wrong I’m wondering if it’s a spoof

    Reply
  • Princeton Election Consortium finds Mitt Romney actually got a negative bounce coming out of his convention. « Egyptday1 // Sep 8, 2012 at 6:41 pm

    [...] The Princeton Election Consortium finds Mitt Romney actually got a negative bounce coming out of his convention. “Indeed, it appears that the race shifted towards President Obama by 6-15 EV, or about 1.0% of Popular Vote Meta-Margin. From an analytical perspective, a negative bounce is quite remarkable because all the talk in recent weeks has been of bounces being smaller or zero, but always in the hosting party’s favor. It is all the more remarkable because of the relatively small number of state polls over the last week, so that the Meta-analysis’s inputs have not fully turned over… So the negative bounce may be larger than what is shown in the graph. Such an event would have been missed in past years (and even this year) because national polls don’t have the best resolution.” [...]

    Reply
  • Daily 2012 Polling for September 8, 2012…Obama ‘bounce’ arrives across the board….Romney ‘negative bounce’? | Politicaldog101.Com // Sep 8, 2012 at 9:37 pm

    [...] The Princeton Election Consortium finds Mitt Romney actually got a negative bounce coming out of his convention. [...]

    Reply
  • spacer Tim // Sep 8, 2012 at 10:53 pm

    Sully’s giving you some serious exposure, right up there with Nate Silver. I’ve been sending him links to your site for weeks now.

    Congrats. You deserve it.

    Our country desperately needs this kind of artful empiricism.

    Reply
  • spacer

gipoco.com is neither affiliated with the authors of this page nor responsible for its contents. This is a safe-cache copy of the original web site.