Lies, Damned Lies, and Politico

Politico is generous enough to admit that Empty Chair’s lead has slipped to one point, nationally:

A new POLITICO/George Washington University Battleground Tracking Poll of likely voters shows Obama ahead of Mitt Romney 49 percent to 48 percent nationally, a statistical tie and a percentage point closer than a week ago.

But wait, there’s less! Get to the internals if you want to get to the truth. First, let’s look at voter enthusiasm:

Only 73 percent who support Obama say they are “extremely likely” to vote, compared to 86 percent who back Romney. Likewise, 84 percent of Republicans say they are extremely likely to vote, compared to 76 percent of Democrats.

Ten points? That’s newsworthy, yes? And let’s get to this year’s Broken Glass voters:

Among those extremely likely to vote, Romney actually leads Obama 52 percent to 46 percent. That’s up from a 2-point lead last week. Obama led 50 percent to 47 percent among this group three weeks ago.

Now, Politico is going to make you do a lot of scrolling to get to this next bit, which is why you can leave it to highly-trained scrolling professionals like myself to get the facts out to you:

Romney now leads among independents by 16 points, 51 percent to 35 percent. This is up from 4 points last week. But he still trails in the overall head-to-head numbers because of near monolithic support for Obama among minority Democrats.

But how likely is that monolithic block to vote? Not nearly as likely, in a paragraph Politico very strangely left separated from the last one I quoted:

The percentages among key Democratic constituencies who say they are extremely likely to vote should cause concern in Chicago: While 82 percent of whites (who break for Romney by a 15-point margin) say they’re “extremely likely” to vote, only 71 percent of African-Americans and 70 percent of Latinos do. And just 68 percent of 18-to-29-year-olds, another key Obama constituency, put themselves in the “extremely likely” to vote category.

You’d think that would be better reflected in Politico’s Likely Voter model, because there’s just no way to square Romney’s impressive lead with Indies and in enthusiasm, with Obama’s 1-point “lead.” Unless, of course, you’re oversampling Democrats significantly. Not that we’ve seen anything like that this year.

Now for the kicker:

This poll was in the field last Monday through Thursday, but about 85 percent of the calls were made before the debate on Wednesday night. The final night of tracking was good for Romney, but it’s not a big enough sample to report. So this does not reflect any momentum Romney might get from his performance in Denver. [Emphasis ever-so-sweetly added]

Why did they even bother publishing a poll that was outdated five nights ago? I think you’ll find the answer in this next line:

Regardless of who they’re supporting, 61 percent of those who replied to the poll said they think Obama will eventually prevail compared to 31 percent who said the same for Romney.

Huh. I wonder how people might ever have gotten that impression.