« BACK  |  PRINT

RS

MEMBER DIARY

WPA’s Election Update: 2 Weeks Out

Sometimes even when you lose, you win.  After a debate that has been described by most in the media as a “narrow loss,” Mitt Romney gained ground in all of the major national polls.

But, some of the state-by-state polls suggest that Romney still trails in the Electoral College.

So what’s really going on?

With just two weeks to go until the election, we look at the Presidential trend, the Electoral College, and dig into the data from one key state.

The Presidential Race

National Trends: Romney on the rise.

Putting aside the question of who “won” last week’s debate—and we made the case that Romney won on the issues that matter last week in our blog—Mitt Romney clearly won the week.

Romney now leads by a fraction of a point in the realclearpolitics.com polling average.  But if you exclude the Investors Business Daily poll that is an outlier every year, Romney leads by an average of nearly three points across all of the national polls taken after the debate.

And, no matter which model you believe is right, the trends are good for Romney:

  • Romney gained three points in Gallup’s tracking, moving from a four point lead among likely voters before the debate to a seven point lead after.
  • The NBC/Wall Street Journal poll shows a dead heat compared to a three point Obama lead at the end of September.
  • Romney held steady with a two point lead in Rasmussen’s tracking.

The Electoral College: Obama with an advantage?

We know the popular vote doesn’t pick the President, so let’s look at the Electoral College:

  • North Carolina is all but certain to go for Romney, with recent polls of all partisan types showing him with leads ranging from two to nine points (with nine being the most recent).
  • He now leads Obama by two points in Florida, based on an average of recent polls, with a range of a five point lead to a single point deficit.
  • Romney leads in Virginia in the last three polls conducted by a range of one to three points.

If the election were held today and we assigned each state based on the average of post-debate polls, here’s how things would look:

Romney

Obama

Probable Wins

206 EVs

(AL, AK, AR, AZ, GA, ID, IN, KS, KY, LA, MO, MS, MT, ND, NE, NC, SC, SD, TN, TX, OK, UT, WV, WY)

237 EVs

(CA, CT, DE, DC, HI, IL, MA, MD, ME, MI, MN, NY, NJ, NM, OR, PA, RI, VT, WA)

Narrow Leads

46 EVs

FL, NH, VA

49 EVs

CO, IA, NV, OH, WI

Total

252

286

These numbers are a lot less positive than the national ballot trends.  But there are a few caveats we should include:

  • We assigned every state to a candidate, even when the polling margins are very small.  Someone has to win and this is the “best guess” of the moment.  But anything in the “narrow lead” category is definitely still up for grabs.
  • The state-by-state polling is questionable, to say the least.  In our first update we discussed the issues with the partisan make-up of the major university/news polls in many of these states.  Those issues haven’t changed.  The alternative to those polls is IVR polls which still show very high variability.

A Detailed Look: Ohio

Looking at the electoral math above, it’s clear that Ohio’s 18 Electoral votes could be pivotal.  Those votes alone changing columns would give Mitt Romney the 270 he needs to be elected President.

So let’s look more at Ohio.

First, here are the recent polls:

Pollster

Dates

Romney

Obama

PPP

10/18-20

48%

49%

CBS News/Quinnipiac

10/17-20

45%

50%

Gravis Marketing

10/18-19

47%

47%

Fox News

10/17-18

43%

46%

Rasmussen Reports

10/17

48%

49%

The range of these polls is pretty broad, with a tie outcome in one and a five point lead for Obama in another.

But, as we’ve pointed out previously, there’s something odd going on with the models that are being used in Ohio.  Here are the partisan mixes:

Pollster

Dates

Republicans

Democrats

PPP

10/18-20

34%

42%

CBS News/Quinnipiac

10/17-20

26%

35%

Gravis Marketing

10/18-19

32%

41%

Fox News

10/17-18

34%

42%

Ignoring Rasmussen, because he won’t release his demographics except behind his pay-wall, we see a spread of either Democrats +8 or Democrats +9.

The trouble is that this mirrors the 2008 exit polls (31% Republican/39% Democrat) but is very different from 2004 (40% Republican/36% Democrat) or less relevant but still interesting distributions in 2006 (37% Republican/40% Democrat) and 2010 (36% Republican/36% Democrat).

So what happens if the electorate is different than 2008?

Here’s a table that explores some scenarios.:

2008 Electorate (Fox News Poll)

2004 Electorate

Mid-Point 2004/2008

Romney

43%

48%

46%

Obama

46%

42%

44%

If we assume Democrats have an eight point advantage, as these pollsters did, Obama leads.  All other things being equal, if we assume Republicans with a four point advantage as was the case n 2004, Romney leads and if we split the difference and assign Democrats a two point edge, Romney leads by a narrow margin.

In addition to the polls, we also have early voting data to examine.

This may give us some insight into what the actual electorate will look like.

We examined the data provided in the amazing United State Elections Project data set maintained by Michael McDonald at George Mason University.

Below are the nine counties that are making up substantially less of the early vote so far this year then they did in 2008.  In other words, these are counties that so far are seeing much lower turnout relative to the rest of the state.

Counties

% of Early Vote County Contributed in 2008

% of Early Vote County has Contributed in 2012

Difference

2008 County Vote Margin

Lake Cnty

2.87%

0.42%

-2.45%

Obama +1

Licking Cnty

1.98%

0.82%

-1.16%

McCain +16

Portage Cnty

1.33%

0.61%

-0.71%

Obama +9

Cuyahoga Cnty

18.66%

18.04%

-0.62%

Obama +39

Lucas Cnty

4.44%

3.91%

-0.53%

Obama +31

Summit Cnty

6.15%

5.63%

-0.52%

Obama +19

Stark Cnty

3.00%

2.53%

-0.46%

Obama +6

Greene Cnty

1.67%

1.22%

-0.45%

McCain +19

Lorain Cnty

3.25%

2.81%

-0.43%

Obama +18

Seven of the nine counties that are significantly lower in turnout (so far) compared to four years ago are Obama counties, including two that went for Obama by more than 30 points.

This is only one small piece of evidence, but it is an additional suggestion that the 2008 electoral model we’re seeing in the polls is not realistic.

If trends continue, this election may very well come down to Ohio.  And based on this analysis, we think that Ohio may already be much better for Romney than the polls would suggest.

Follow Ryan on Twitter

Get Alerts