Is it all about the ratings? Stay with CNN to find out.

CNN's Campbell Brown was on The Daily Show the other night promoting her new show (No Bias, No Bull), and during the interview described CNN as the new bastion of the middle ground in politics. FOX on the right, MSNBC on the left, and the neutral CNN right smack in between. If they don't owe allegience to the right or left, though, is it possible that they may owe it to the ratings? While most national polls still show an Obama lead from anywhere between 3 points (FOX) to 15 (Pew), CNN's "Poll of Polls" is showing what they are calling buyer's remorse for Obama. They now have him winning by only 5 points nationally, down from 8 just last week.

I commented (which, of course, did not make the moderator's cut since what I said wasn't incendiary or vacuous) that the buyer's remorse stipulation was used by Howard Wolfson after Super Tuesday, as well as by CNN during the closing primaries that Hillary Clinton won. In other words, CNN is recycling an old Clinton talking point that they already had lifted for the sake of fabricating a story of polling tension. It goes farther than this, of course, because that Poll of Polls they do (notoriously inaccurate, by the way) had to come from somewhere. Let's take a look, shall we? Here's their chart.

Polls used are: FOX, ARG, ABC/WaPo, Zogby, Gallup, Hotline, and IBD.

The first problem here, is that they don't specify which Gallup poll they are using. The "traditional" model for likely voter estimates will strike out any respondent who did not vote in 2004, which seems absolutely crazy to me considering the way in which Obama has gotten this far seems to have relied heavily on what CNN calls the "League of First Time Voters", yet this model still has Obama +5 today. However, if we look at the expanded model, which includes people who didn't vote in '04, but say they really swear they will this time, it becomes Obama +7.

Next problem, IBD/TIPP. They've been in a bit of hot water as of late (that's two links there) for having atrociously atrocious atrocities in their voter samples. Not to mention that they've had a relatively significant house-effect towards McCain this election cycle. In other words, I just wouldn't include them in my top 5 pollsters if I were, say, a political editor at a respectable news outlet like CNN. I might go with the pollster they decided to reject instead, like - oh... I dunno - Rassmussen. In fairness, they did snub Battleground as well, but that is deserved just like a snub of IBD would have been. In fairness to that fairness, they also snubbed CBS/NYTimes, whcih is actually even more deserved.

Let's take a peek at that FOX/Opinion Dynamics poll now. You may have read in previous postings how much respect I have for this pair, but allow me to give one more example of just why I feel that way. You'll need acrobat reader to view it (it's a .pdf), but here is the detailed poll they released. Notice anything a little strange? Apparently more Republicans are "extremely interested" in this election than Democrats, by eight freakin' points! How can that be, considering the widely reported-on enthusiasm gap in favor of the Dems this time around? The answer is actually pretty simple, if you scroll down to their sample. 41% Democrat, 39% Republican, 16% Independent. Problem with that is that the country looks more like 40/30 for Dems/Repubs this time around. In other words, FOX/OD fudged their numbers to make the race seem tighter.

That leaves us with ARG, Zogby, Hotline, ABC. Let's check out what Nate Silver thinks of these guys. Out of 32 ranked pollsters on fivethirtyeight.com, these 4 make up #16 (ABC/WaPo), #17 (Zogby), #23 (ARG), and Hotline doesn't even make it. CNN's own polling comes in at #21. The snubbed Rasmussen comes in at #3. Way to go on that one, CNN.

So CNN has the race to O+5, which is 3 points lower than they had it a week ago. Assuming CNN used the "traditional" Gallup poll, and excluded Rasmussen, Battleground, and CBS/NYT, we wind up at an average O+5.4. Using the "expanded" Gallup it comes to O+5.7. My guess is they chose to round down the 5.4 rather than the 5.7, which means they picked the traditional model despite it's seeming ridiculousness. To pre-empt anyone who might say "what if they used both?", that averages to O+5.6.

This means that they took a field of generally bad pollsters, and made them worse, then published the poll under a reused headline to try and show the race as tightening dramatically. Just what, exactly, was that O+8 just a week ago, then? As it turns out, there were some different pollsters in the field that cycle. Pew (O+15), Newsweek (O+12), ABC/Washington Post (O+7), Zogby (O+4.7), Gallup (O+7), Diageo/Hotline (O+7), and IBD/TIPP (O+3). Those Pew/Newsweek numbers are enough to get this out to an average O+7.9. In other words, CNN's Poll of Polls has been telling us more about the house-effect or slant that the pollsters in the field that day have more than they do about who is actually winning.

That being said, when even the right-leaning pollsters have Obama at +5, things look pretty grim for the Republicans. Never fear, though, folks like CNN will be insisting that there's a close race all the way to November 5th. In the mean time, all they ask is that you don't touch that dial.