On second thought . . .

The web site Real Clear Politics  lists a bunch of state poll results, mostly for 23 battleground states.  If you don’t weight by population, the average swing toward Romney (compared to 2008) was 6.8%, not enough to overcome the fact that McCain lost by 7.3%.  But here’s what really struck me:

1.  The Electoral College had a strong Democratic tilt in 2008, as best we can tell (it’s harder to estimate in lop-sided elections.)  If you adjust each state evenly, the popular vote would have had to move 9.6% in McCain’s direction for him to eke out a EC victory.  That would have given him a 2.4% popular vote margin, which is quite unusual.  So that explains most of the mystery in the previous post.

There were 4 states with unusually large Romney swings from the McCain election to the current RCP state poll consensus:

Missouri (+11.1%), Wisconsin (+11.6%), Michigan (+12.5%), and Indiana (+13.5%).  That makes sense, as the Midwest is less polarized then other parts of the US, with more open-minded voters, or more voters who are open-minded, or both.  I’d guess that in Mississippi you could guess how 90% of people will vote by just looking at them.  So this should be great news for Romney—he’s picking up most strongly in the parts of the country where he most needs to do better than McCain, the Midwest.  Except for one state.  If you had to name one state that was most similar in terms of industrial mix to the 4 states mentioned above, it would be Ohio.  And in Ohio Obama only won by 4.6% in 2008.  Easy pickings for Romney.  But for some strange reason Romney is only doing 2.3% better than in the McCain election.  Given me any explanation you want, and I can blow it out of the water by pointing to Romney’s huge gains Indiana and Michigan.

The election really does come down to one state.  And because of Ohio the Democratic EC tilt this year may actually be larger than the already huge 2.3% of 2008.  Romney should consider running commercials saying he’ll move the national capitol to Dayton, or Toledo, so it’ll be more centrally located.

Oh wait, he still doesn’t have Virginia wrapped up . . .

PS.  I still don’t think the South can explain the big EC imbalance.  Consider the 12 Midwestern states.  As things stand now Romney will win 6 and Obama will win 6.  But all of Romney’s wins will be double-digit, and only one of Obama’s wins will be double digits.  It’s bad luck for Romney that in the Midwest all 5 of the single-digit states seem likely to break for Obama.  It’s odd to win 6 of 12 states in a region, and not win a single one by a margin of less than 10%.  Even in the South Romney’s likely to win lots of vote-rich states by narrow margins (Florida, Virginia, North Carolina.)



36 Responses to “On second thought . . .”

  1. Gravatar of adam adam
    27. October 2012 at 16:01

    Explanation: Obama’s targeted ground/ad game is just that good. He’s targeting only the states he needs, and nothing more. There is a lot of evidence to indicate that his campaign is light-years more sophisticated and targeted than Romney’s is.

    I don’t know if this is the explanation, but it is AN explanation.

  2. Gravatar of Greg Ransom Greg Ransom
    27. October 2012 at 16:39

    Wake up and smell the preference cascade … Onama paper after Obama paper is breaking Romney and endorsing the challenger.

    And the there is the continuing drip, drip, drip of Benghazi …..

    A well under 50% incombent this late in the game isn’t winning anything.

    Ohio is explained by months of anti-Romney hit jobs via TV advertising by Obama.

  3. Gravatar of Brett Brett
    27. October 2012 at 16:47

    DeLong looks at the RAND continuous polling data (which tracks a defined set of respondents over time) and suggests that “A resurveyed electorate becomes an informed electorate, and a high-information electorate likes Obama.”

    If you are looking for a the big difference between Ohio and Indiana, that difference is that anyone who lives in Ohio has more or less been forced to pay attention.

    Is this the explanation? I don’t know. But the data seem to point in this direction.

  4. Gravatar of Yellow Dog Yellow Dog
    27. October 2012 at 16:50

    New polls out showing Obama surging in Virginia. Up now between 4 and 5 points in the latest polls. I think that Mourdock rape comment is hurting Romney among women more than he is letting on.

    Another thing is that almost all of the polls showing Romney with a lead are expecting a huge “enthusiasm gap” in favor of Romney whereas the latest WaPo Virginia poll shows Obama voters more enthusiastic.

    I think this ends in an Obama win with 300+ electoral votes.

  5. Gravatar of ssumner ssumner
    27. October 2012 at 16:59

    I see lots of wishful thinking on both sides in the comments.

    For the record, I’ve predicted Obama to win from the beginning.

    Greg, Do you seriously expect Benghazi to sway one vote?

  6. Gravatar of Rajat Rajat
    27. October 2012 at 17:21

    Yes Scott, maybe I missed some explanation in between but what made you write-off Romney in mid Sept post QE3?

  7. Gravatar of Greg Ransom Greg Ransom
    27. October 2012 at 18:45

    Your ads totally suck for anyone using an iPad — really obnoxious.

    Do you know how hostile your aggressive & user unfriendly ads are — please stop abusing your readers with these pop-ups.

  8. Gravatar of Greg Ransom Greg Ransom
    27. October 2012 at 18:51

    7 points in 3 days.

    And the only thing that is playing out right now are the facts about Benghazi.

    The low information voters are not turning out as they did in 2008.

    And the highly emotional facts about Benghazi are pouring out.

    Just wondering if you are one of the low information voters on this …

    My guess is you are on this one.

  9. Gravatar of Greg Ransom Greg Ransom
    27. October 2012 at 19:26

    “in the Rasmussen survey, only 47% approve of Obama’s job performance, while 52% disapprove. There are two plausible explanations for Obama’s current slide: voters may be disgusted by the mean, petty campaigning he is running, or they may disapprove of his handling of the Benghazi crisis. Or both.”

    A 47% approval incombent is not an incombent who will have. 2nd term in a 2 man race.

  10. Gravatar of David R. Henderson David R. Henderson
    27. October 2012 at 19:28

    If you give me 70/30 odds, I’ll bet you $30 that Romney wins at least 280 electoral votes.

  11. Gravatar of wufwugy wufwugy
    27. October 2012 at 19:33

    When you scour the internals of each poll, you find that the ones favoring Romney have worse methodology. One example (but there are many): CNN recently gave Romney +1 among LV in FL, but they did so by cutting out a ridiculously huge amount of RV respondents (approximately 1/3rd) which had Obama +7 in RV.

    So the truth of the CNN poll was that properly weighted LV model would have had Obama around +3, CNN merely lied about it just to drum up the Ro-mentum so people would watch their crappy “journalism.”

    The list of polls that have done shady things like this is almost endless, and most of them boost Romney’s numbers.

  12. Gravatar of Yellow Dog Yellow Dog
    27. October 2012 at 19:34

    Rasmussen is a hack pollster.

    Check this out from 2008 and you’ll know all you need to know about Rasmussen:


  13. Gravatar of Greg Ransom Greg Ransom
    27. October 2012 at 19:47

    Romney needs Iowa & Wsconsin to take the EC without Ohio — the Des Moines Register just endorsed Rmney, and the state is in the margin of error with Romney coming on and his base energized, while most 2008 Obama voters are not.

    Wsconson has gone Blue repeatedly the last 3 years ….

  14. Gravatar of Yellow Dog Yellow Dog
    27. October 2012 at 19:51

    I don’t get this meme that Obama voters are not energized. Here is an example from the most recent poll of Virginia which Romney has to have:


    If anything, Romney is a little more vulnerable to losing last-minute fence-sitters.

    Obama supporters: 94% will definitely vote for Obama, 1% good chance of changing mind, 4% unlikely to change mind, 1% no opinion

    Romney supporters: 92% will definitely vote for Romney, 3% good chance of changing mind, 5% unlikely to change mind

    Obama supporters: 70% very enthusiastic, 25% fairly enthusiastic, 3% not too enthusiastic, 1% not enthusiastic at all, 1% no opinion

    Romney supporters: 56% very enthusiastic, 37% fairly enthusiastic, 4% not too enthusiastic, 2% not enthusiastic at all

  15. Gravatar of Andy Harless Andy Harless
    27. October 2012 at 20:28

    12 Midwestern states. As things stand now Romney will win 6 and Obama will win 6. But all of Romney’s wins will be double-digit, and only one of Obama’s wins will be double digits.

    I don’t think this is the right way to look at it. Romney will win 4 small, consistently Republican Midwestern states plus Missouri and Indiana. Obama will win 5 large Midwestern states plus Iowa. If we look only at the large states, Romney has a large margin in two, but Obama has a truly huge margin in the largest of them all.

    Based on size and past Presidential voting patterns, I think we should see Kansas, Nebraska, and the Dakotas as a Republican block that is thrown in with (and dominated by) the rest of the Midwest but doesn’t really belong there. They’re small in population, and they vote like Texans, not like the vast majority of Midwesterners.

  16. Gravatar of Greg Ransom Greg Ransom
    27. October 2012 at 22:34

    The big question — are the polls dramatically over-sampling and over-weighting Democrat voters?

    Lots of analysts conclude that they are.

  17. Gravatar of Greg Ransom Greg Ransom
    27. October 2012 at 22:38

    Romney might win the Presidency simply from the wave of vote switchers on all of the newspaper editorial boards flipping from Obama to Romney this year ….

  18. Gravatar of Mike Sax Mike Sax
    27. October 2012 at 23:21

    The idea that Benghazi-of all things-will win it for Romney is wishful thinking. If it’s such a smoking gun why did Romney fail to engage it on Monday?

    There’s nothing big coming out about it. The truth is that things like this are fluid situations where more information comes out and more is learnt.

    For instance the GOP was belaboring Susan Rice’s comments as either ignorant or deliberate lies. Yet it turns out that what she said on that day in September is largely in line with what the CIA was writing in it’s “talking points” memo on the very same day.

    If Obama has Ohio-and he does-it’s over.

  19. Gravatar of Mike Sax Mike Sax
    27. October 2012 at 23:24

    And as far as editorial endorsements are concerned the most embarassing one for Romney-and the most telling for his ill-fated chances in places like Ohio, Wisconsin, and Michigan-is the one the Detroit Press gave him.

    It’s a conservative paper and you know it will endorse him. Yet even in giving it, they still said that Romney was flat wrong about the auto bailout and the President was exactly right.

  20. Gravatar of Max Max
    27. October 2012 at 23:27

    If the EC result diverges from the popular vote for the second time in 12 years, maybe we’ll see a constitutional amendment to change the system. Of course this requires that the consequences of the change by unpredictable, otherwise the party expected to be harmed would block it.

  21. Gravatar of Mike Sax Mike Sax
    27. October 2012 at 23:34

    Yeah Max, I don’t think the Democrats will be in such a hurry. The GOP had no problem with it in 2000.

    As a Democrat I probalby would have liked your idea in 2000 but now I wonder why we only need such a thing when it might benefit the Democrats

    Stil we don’t know that Romney will win the popular vote. Cumuatlively he has no more than a 1 point advantage-and Nate actually has Obama up by a very small margin-something like 0.4.

  22. Gravatar of Kevin Donoghue Kevin Donoghue
    28. October 2012 at 00:58

    WaPo: Romney’s leads in Virginia by -4. #TheOldJokesAreThe Best

    Really, all this popular vote discussion is pointless. Politicians work with the system which is in place. If the popular vote determined the result, Obama would be dividing his time between New York and California. Republican candidate Rick Perry would be working the South. Romney would be installing a new jacuzzi for Rafalca.

  23. Gravatar of Simon Simon
    28. October 2012 at 01:24

    Like I said in my comment to your previous post, OhioOhioOhio…

  24. Gravatar of Saturos Saturos
    28. October 2012 at 02:00

    Everyone, Marcus Nunes’ latest is brilliant: http://thefaintofheart.wordpress.com/2012/10/27/the-housing-boom-financial-crisis-the-great-recession-3/

  25. Gravatar of Saturos Saturos
    28. October 2012 at 04:18

    Scott, I’m afraid you are going to have to respond to this: http://marginalrevolution.com/marginalrevolution/2012/10/would-abolishing-cash-help-cure-ad-problems.html

    I tried to help you out a bit in the comments.

  26. Gravatar of Saturos Saturos
    28. October 2012 at 04:23

    I also find Ryan Avent’s attitude in this piece strange: http://www.economist.com/blogs/freeexchange/2012/04/monetary-policy-0

    He talks as if what a central bank needs to do is “create inflation”, in order to lower real interest rates, in order to get people spending, in order to boost growth. Talk about a Rube Goldberg contraption… Also he is dumbing down the issues re the ZLB, which he shouldn’t need to do as he’s been blogging on this topic for years now.

  27. Gravatar of ssumner ssumner
    28. October 2012 at 06:06

    Mike Sax, I’m not surprised, I always suspected you had no principles.

    Andy, I disagree. I think and EC tilt is 100% due to the favored party winning lots of states by narrow margins and the other party winning slightly fewer by big margins. That’s all it is. If both guys win identical votes in the Midwest (which possible, but unlikely) Obama will have far more electoral votes.

    Saturos, I already have a post up on Tyler’s post. Ryan was presumably referring to expected inflation. But expected NGDP would obviously be preferred.

    David, I prefer to bet with anonymous people. I have my Obama bet at the IEM. Have you considered the same?

    But if you just want bragging rights, I’ll do a gentlemen’s bet where I lose 70 reputational points if wrong, and you lose 30 reputational points.

  28. Gravatar of Patrick R. Sullivan Patrick R. Sullivan
    28. October 2012 at 07:29

    ‘If Obama has Ohio-and he does-it’s over.’

    Not true. Romney can win without Ohio.

  29. Gravatar of David R. Henderson David R. Henderson
    28. October 2012 at 10:52

    I looked quickly at IEM site and couldn’t find that bet. Notice that I said 280 or more electoral votes, not 270 or more.
    I prefer to bet with non-anonymous people because I bet mainly for fun and to test myself publicly.
    So that does lead to reputational points. So OK. We’re on. On this condition: both winner and loser agree to do a blog post telling the outcome relative to their bet. Otherwise reputation won’t be much affected. Deal?

  30. Gravatar of ssumner ssumner
    28. October 2012 at 11:00

    David, Yes, that’s a deal, as long as you promise to send me a reminder email when I lose. I don’t want to welch out on a deal through memory-loss.

    Apologies to my readers from Wales.

  31. Gravatar of David R. Henderson David R. Henderson
    28. October 2012 at 13:11

    Good, Scott, will do. I’ll also send a reminder “when” I lose.

  32. Gravatar of mpowell mpowell
    29. October 2012 at 07:59

    Use a real model like Nate Silver’s not RCP. Romney is winning this election because of his polling improvement in Michigan and Indiana compared to Obama in 2008? Really? Do some more work besides just glancing at unweighted state poll averages from one other election or don’t bother putting up a post.

  33. Gravatar of mpowell mpowell
    29. October 2012 at 08:06

    Also, if you want to know why Silver puts more weight on state based polls as opposed to national tracking polls, read his site. The short story is this: state polls are more reliable because they usually represent more and better polling. So yeah, he uses state polling to construct a true national average which is 1-2% away from a simple RCP average. You might not like his model, and it may be wrong about some things, but there are actual good reasons for the stuff he has put in there which anybody who actually does high level statistical modelling should be able to appreciate. You can’t just reject the model because 30 minutes of your investigation of unweighted polling aggregates suggests a different conclusion.

    Finally his model only has Ohio swinging the election 50% of the time, which is huge for a single state, but nowhere near being the only ‘plausible scenario’ for victory for either candidate.

  34. Gravatar of Saturos Saturos
    29. October 2012 at 23:44

    Scott, do you like this argument?

    Also: http://econlog.econlib.org/archives/2012/10/why_we_wont_eli.html

  35. Gravatar of ssumner ssumner
    30. October 2012 at 05:49

    mpowell, I’m arguing Romney is losing the election, not winning it. Indeed I bet money on Obama. Don’t you read my posts?

    The question of whether Ohio “swings the election” is different from whether someone can win without Ohio. You seem awfully cocky for someone who doesn’t know much about electoral math. It’s very possible that Obama will with without “needing” Ohio, but getting it anyway. My claim is that it was very unlikely that the winner will not win Ohio. That’s different from the odds of it being the swing state in the sense of providing the 270th EC vote.

    Saturos, I think he’s 100% wrong. If making special interest promises could move even 2% of the vote, (much less 20%), Romney would be making all sorts of desperate promises to Ohio, and the other states he needs, as would Obama. So his argument is a much stronger critique of the EC than it is a critique of a popular vote system.

  36. Gravatar of mpowell mpowell
    30. October 2012 at 08:56

    Perhaps I’m misunderstanding some of the claims that you are making. I interpreted your comment about Romney’s poor polling performance in Ohio as suggesting that the polling was off. Clearly that was not the case.

    But your main point is that you are expecting a huge electoral college advantage for the Democrats this year. And that is based on an RCP national polling edge for Romney and an Ohio state polling edge for Obama. And my argument is that this is a poor basis for that expectation. A better election predictor, like Silver’s model, constructs a new expected national average based partly on state polling data. For example, his model gives Obama a 1.4% edge today nationally and a 2.1% edge in Ohio, which if true, suggests a much smaller EC advantage than you assume. This election is unlikely to be a big blowout so we should actually get a pretty good idea of the true Dem EC advantage in this one afterwards. I expect Silver’s approach will fair much better than yours.

    And regarding how crucial Ohio is, of course in most of the scenarios, the winner carries Ohio. But this is similarly true of a number of other swing states. You cannot point out that Ohio goes with the winner a great portion of the time and then use that observation to claim that the election comes down to Ohio when a similar claim can be made about other states as well. The point remains that according to Silver’s model, in cases where a single state determines the election, that state is Ohio only half the time. This is huge for a single state, but is still not definitive. And thinking about campagin investments, if Romney had dedicated huge amounts of resources in moving Ohio, say 5%, but this had cost him 0.5% in the national polling average, it almost certainly would not have been worth it. That’s because if you move Ohio just a tiny bit away from where it is with respect to the national average, it suddenly becomes much less important.

Leave a Reply