Nov 04 2012

Print this Post

A Last Look at the State Polls

Share to Google Plus

Two Can Argue But Only One Can Be Right

It is the most over-used cliché in elections, but it does indeed come down to turn out.

A majority of the state polls that are commonly sighted to track the state of the presidential race in the battlegrounds include a modeling sample that mirrors or exceeds Democratic turnout from 2008.

Most of the critics of these polls- myself included – have looked at likely voter models presented by Gallup, Pew and Rasmussen (which generally agree that party ID for participating voters this year will be roughly even) and are thus skeptical of these polls’ results.

It is a stark divide that will only be resolved on the morning of November 7th. Someone is going to have serious egg on their face.

That said, here are the latest state polls, pointing out the modeling distortion(s), for your edification.


Columbus Dispatch poll: President Obama is leading Mitt Romney 50-48 percent. The Margin of Error is 2.2 percent, so the race is a statistical dead heat.

Now look at the modeling sample.

The make-up of the survey sample is D+4, 39 percent Democrat/35 percent Republican. In 2008, President Obama won Ohio by 4 points with a D+5 Party ID advantage. So the Dispatch model is looking at near identical turnout compared to 2008.

But there are tangible mitigating factors.

As Adrian Gray has pointed out, the GOP has increased their early turnout in Ohio by more than 100,000 from 2008. Democratic turnout, in contrast, is down by 150,000. That is 250,000 net gain to Romney, virtually wiping out Obama’s 260,000 vote majority in ’08. In 2008, McCain lost the early vote to Obama, but won among voters who showed up on Election Day. Is there any reason to believe that won’t be the case on Tuesday?

And there are other notable anomalies.

The Dispatch has Indies going for Obama by +10. Almost all recent polls show Indies trending with Romney.

And then there is the religious vote.

McCain won Ohio Protestants 52-48 percent in 2008. He won Ohio Catholics 52-47 percent. Together, these two groups represented 81 percent of the Ohio electorate.

In the Dispatch poll, Romney wins even bigger among Protestants (60-39 percent) and Catholics (55-44 percent) than McCain did in ’08. But in the Dispatch poll, Protestants and Catholics make up only 68 percent of voters in the religious category. Indeed, those who identified with “other” or “none” went from 17 percent of the electorate in ’08 to 29 percent in the Dispatch poll.

That big a jump in only four years is simply unlikely.


I am of the opinion that MN is a stretch for the GOP, but the latest PPP poll bearings further scrutiny.

The PPP poll has Obama ahead of Romney by +8, 53-45 percent. The sample for the poll is D+8. In the wave election of 2008, the Democratic Party turnout advantage was D+4.

So PPP is estimating that double the number of Democrats will turn out this year compared to ’08?

Good luck with that.

If there wasn’t a tangible advantage, why would Team Romney send VP nominee Paul Ryan to Minneapolis on the Sunday before the election? It could be a two-fer, since Minneapolis serve part of the Wisconsin media market, but it remains a telling use of the highest prized resources.

The GOP may not win MN, but the final result won’t look like this PPP poll.


The Des Moines Register poll has President Obama leading Mitt Romney by five points, 47-42 percent. The poll sample favors Democrats, D+5. In 2008, Democrats outnumbered Republicans in turnout by D+1. In the midterms of 2010 (a less reliable indicator) the turnout was R+4.

So, the Register expects 4x the number of Democrats to turnout against Republicans compared to the watershed year of 2008.


Back to PPP here. In 2008, President Obama beat McCain by a daunting 10.5 points in PA. Now, PPP has Obama beating Romney in PA by eight points, 53-45 percent.

In 2008, the Democrats had a strong turnout advantage – D+7 over the GOP.

However, to get Obama to an eight point lead in its survey, the PPP model has Democratic turnout at D+10; an expected 30 percent greater turnout advantage for Democrats on Tuesday than in the wave election of ’08.

Again, does anyone see and expanded Democratic turnout advantage this year?

And while you are chewing on that, consider that in the midterms of 2010, the Democratic turnout edge dropped to was D+2. That was enough for PA to elect a Republican governor.


PPP has President Obama leading closely here, 51-48 percent.

In 2008, POTUS won by a staggering 14 points in WI. The Democratic Party turnout advantage was D+6. PPP seems to be more cautious with its current sample for WI, predicting a D+3 advantage for Obama. But even that more conservative model is problematic.

In the PPP poll Obama’s approval is at a net negative – by 49-46 Wisconsin voters disapprove of his job performance. Dig a bit deeper and you see that 12 percent of Democrats disapprove of Obama’s job performance. Yet, PPP has independents going to the President by 17 points. That is inconsistent with all other polling done over the last month, and is a particularly striking gap when the Democratic base is unsettled.

But perhaps the most important inflection point for WI, that which sets it apart from other swing state analysis, is that WI held its most recent statewide election only four months ago with the recall election against Governor Scott Walker; an election that Walker won decisively. It makes this turnout model more predictive than the midterms or ’08 that we have been using in other states.

And in the June recall election, the turnout advantage was R+1, a four point swing away from PPP’s current sample. In the recall, Walker won 53-46 percent. In fact, running against the same opponent from 2010, Walker won by a bigger margin in the recall. He won eight percent of the Democratic vote and he won Independents by nine points.

This is the coalition that is available to Mitt Romney, if his GOTV efforts are competent and effectively deployed.

In sum, we are looking at two different views of reality. Only one can be right.

If these polls are right, then there is a highly motivated and deeply organized cohort of Americans, heavy on single women and minorities, which will turn out en mass on Tuesday and push the President to victory.

Nixon once talked of the “silent majority” of Americans who deeply disagreed with war protesters and social liberalism, but otherwise remained quiet and indoors. For these polls to be right, there would have to be a an equally massive and mostly under-the-radar progressive cohort, that has shunned early voting and has intentionally sat out Obama rallies and other public metrics of enthusiasm, events that ironically are designed to attract and energize just these kind of voters.

They were everywhere in 2008. They are mostly nowhere to be found today.

If they exist, and the polls are right, Romney can’t win.

But tangible evidence beyond the polls does not support these kind of numbers for the President.

We’ll know who’s right on Wednesday morning.



Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>