Princeton Election Consortium

A first draft of electoral history. Since 2004

Swing state advantage

October 2nd, 2012, 5:44pm by Sam Wang


Via Andrew Sullivan, several claims that the Presidential race may be tightening. Maybe at a national level, but not where it counts:

History of Popular Meta-Margin for Obama

In national polls, the median for polls from 9/26 to 10/1 is Obama +4.0+/-0.6% (n=11). Yet the Meta-Margin, which measures what it would take to flip the swing states, is at Obama +6.0%, including a substantial uptick today.

Although the Meta-Margin, which is based on state polls, can sometimes take a little while to catch up with national polls, it is extremely precise – within a few tenths of a percentage point. Its sharp rise indicates that something is afoot. Obama appears to be significantly overperforming in critical battlegrounds compared with the rest of the country. Is this because of the much-vaunted microtargeting by the Obama campaign? Whatever the cause, this hill is a very tall one for Romney to climb in October.

Tags: 2012 Election · President

16 Comments so far ↓

  • Peter D

    I was thinking about this today. I think that strong voter polarization, in addition to giving us stable preferences, can cause strange equilibria where the national popular vote is much closer than any particular state. This will give you a very stable electoral college outcome and reduce Meta Margin volatility.

    Groupthink caused by perceived homogeneity and the belief that one is making a moral/ethical choice would be the primary cause.

    Just a theory.

  • Michael Worley

    Interesting that this is exactly the same as Gallup is right now–Obama +6.

  • 538 Refugee

    While I’m happy Obama may have nothing to worry about, there is the issue of coattails. I hope he and his campaign have an eye on the makeup of the congress he will be working with too. This isn’t only about winning the presidency but also about being able to do something with the office once you have it. The opposition party said they “got it” when they got swept out but we all saw the sandbagging and the large step to the right they have taken since then.

  • Steve16748

    In fact, Ohio has Mt. Bankrupt Detroit.
    Florida has Mt. Medicare.
    Many more states have rocky peaks with grades of over 47% if you want to climb to the top.

  • Joel

    I wonder if this is the vaunted ground game in effect? Just looking at the OFA website, it looks like Obama has 34+ field offices in Ohio alone. I know that Romney is nowhere close to that number.

  • Tapen Sinha

    Indeed the hill is tall for Romney (or should I say Mr. Romney).

    (1) Obama campaign has spent a good deal of money on targeted ads in the (ever expanding) battleground states.

    (2) They also have a tremendous advantage at the ground game.
    http://1.bp.blogspot.com/-LVZ3lnwz2DY/UEZyrT5_eCI/AAAAAAAAE0A/LJIQfXhtdtE/s1600/field+offices+2012.png

    Tapen

    • Anbruch

      The ground game is one reason it might not make sense to move all contributions away from the President and to the House. The ground game will have effects all along the ticket, the money to the House candidate will only affect that particular race. I would be curious if one could devise an optimal contribution strategy from public information.

    • Olav Grinde

      Interesting point, Anbruch.
      I would be astonished (and disappointed) if the Obama campaign did not make a significant effort to maximize the coattail effects. Especially in states where Obama is more popular then than Democratic Senate/Congressional representative, I would expect that he would use that for what it’s worth. After all, to implement his policies, President Obama is dependent on approval in both houses of Congress — or at least decreased obstruction!

  • Some Body

    Here’s an educated guess: As Nate Silver noticed not long ago (couldn’t locate the link on the fly), there was a visible gap in state polls between robopolls and standard methodology polls recently. And there seems to be a similar amount of polls of both types released.

    When there is such a gap in the middle, a median should “jump” halfway across the gap every time a new poll makes one of the two heaps exactly equal to, or larger by exactly one, than the other. In other words, the median margin in some of the swing states may be “overreacting” to the fact that a few more standard methodology state polls than usual were released in the last couple of days.

    That’s one of the situations in which an average may be more well-behaved than a median.

  • thingsbreak

    Off topic: What happened over the past week with the RAND American Life Panel?

  • Matt McIrvin

    The RAND poll had what looked like real movement toward Romney around the time of the conventions; it looks to me like an actual, very short-lived Republican convention bounce, since it was too late to be just the Ryan selection. It may be that RAND was just the only survey with sufficient sensitivity to day-to-day changes to detect it.

    Anyway, RAND currently shows the race tightening up just a tiny bit, with a slight movement back to Romney. The lack of motion in the respondents’ own predictions of who will win suggests that the poll-truther line isn’t getting any increased traction (still, they’ve got about 39% of their sample asserting consistently that Romney is going to win).

    • Olav Grinde

      I suspect some of that movement is hopeful thinking.

      The GOP campaign has greatly increased the public’s expectation of how Mitt Romney will perform in the first debate. If he delivers the “knock-out punch” some conservative pundits are hoping, then he’ll probably surge in the polls. Well, actually, a bounce significantly less than the Meta-Margin is more likely the best possible scenario…

      If his performance is anything less than stellar, then Mr Romney will have failed to fulfill those inflated expectations — and he’ll sink further.

      We’ll soon see what happens.

Leave a Comment