Most of my recent stories have been focused on polling, but the only meaningful polls for the next week or so are the snap polls that were done to gauge voter response to the second Presidential debate. Those will only take a minute to report, and then for the next week you can pretty much forget about polling. It'll be Saturday at the earliest that any tracker will have a full sample of post-debate data (Rasmussen). Ipsos/Reuters won't have a full sample to report until Monday, although they did a post-debate snap that I'll report below. Gallup won't have a full sample until Wednesday.

I put so little faith in the Investors Business Daily tracker that I don't even know what their sample length is, and I don't really care.

As far as the trackers go, Ipsos has shown a crystal clear swing away of likely voters away from Mitt Romney, +3 on 10/11, towards Barack Obama, +3 on 10/16. Rasmussen bounced between +1 and +2 for Romney since the 12th. RAND had shown a steady build for Obama, from +1.8 on the 10th, to +5.1 yesterday, but that fell a bit today (from polling yesterday) to +3.91.

Gallup has me worried. Their likely voter model from 2010 supposedly missed the mark by something like 9 points to the GOP side. Their results reported today (10/10 - 10/16) are very quite strong for Mitt Romney, 51-45 in the likely voter model, and now +2 in the registered voter model. I'm worried because that's a pretty big movement that has no explanation, and contradicts most other polling. Ipsos and RAND show Obama gains, Rasmussen has been flat for about 10 straight days, some national and state polls show a small but steady rebound for Obama.

How is it only Gallup sees such a big movement towards Mitt Romney, and why? Even with Romney's first debate bounce, which every pollster recorded, Gallup never showed Romney with a lead in its registered voter model. But now, with one day of a seven day sample dropping off, and one day being added, Romney gained +4 in RV, and +4 in LV? When basically nothing happened? When all other pollsters show either no movement at all, or movement towards Obama?

Gallup has Romney up by 6 while Rasmussen only has him up by 1?

That, folks, is the definition of an outlier. And it makes me question whether or not I should even be reporting Gallup anymore. If Gallup and Rasmussen agreed within a reasonable margin, I wouldn't have a problem here. But when Gallup is showing a five point skew beyond a pollster that was missing Senate races by *40* points in 2010, you've got to seriously consider dumping them as a junk shop.

You really do.

Here's all the snap polls I could find. These are "who won" or "who did better", not "who are you voting for":

CBS: Obama 37%, Tied 33%, Romney 30%.

CNN: Obama 46%, Romney 39%.

Sample: 33% Democrat, 33% Republican.

PPP in Colorado: Obama 48%, Romney 44%.

Tabs: Obama 58%, Romney 36% among Independents; Obama 63%, Romney 27% among moderates; Obama 49%, Romney 46% among women; Obama 48%, Romney 43% among men.

Sample: 33% Democrat, 36% Republican, 31% Independent.

CBS5/SurveyUSA in California: Obama 56%, Romney 32%, Tie 12%.

Tabs: Romney 44%, Obama 40% among Independents.

Google Survey: Obama 48.5%, Romney 31%, Tie 20.5%.

Ipsos/Reuters: Obama 48%, Romney 33%.

No snap poll showed Romney winning the debate, but none showed a blowout like they did last time around. That's not news Romney wants to hear, because he was losing pretty soundly in the electoral college and by a stable out-of-MoE in national polls before the first debate, and was still losing in the EC even after his debate win. With some signs of polls already moving back towards Obama as that debate bounce faded, Romney needed another outright win, and didn't get one.

That's part of the incumbents advantage, and Obama's advantage specifically. He was already winning before the first debate and therefore could absorb some damage without falling over. Now he's recovering from that at the same time he just "won" a marginal victory in debate #2.

All things considered, I agree with Nate Silver that we'll probably land on a consensus of +2 Obama from here to election day. But that doesn't matter as much as state polls. Dedicated polls of Colorado, Ohio, Florida, and Virginia will be key over the next seven to ten days. Watch for those and you'll know what's going on.

* * *

While writing this and looking for any snap polls I may have missed, I came across this explanation of that odd Gallup poll showing +6 for Romney amongst likely voters. Here is the poll by region:

East: Obama +4

Midwest: Obama +4

South: Romney +22

West: Obama +6

Make of that what you will, but Romney needs to see gains in Colorado, Nevada, Iowa, and Ohio if he wants to win this election, along with holding Florida, and only Florida is in the South. Here are the last four polls of Florida, none newer than 10/14:

ARG: Romney +3

Rasmussen: Romney +4

PPP: Romney +1

Gravis Marketing: Romney +1

I've said it before and I'll say it again: Mitt Romney can't win the election without carrying Florida. If he loses Florida, he's done. But he can lose even if he wins Florida. That's the peril of being behind. That's why he really needs Florida, and then Ohio, Iowa, and either Colorado or Nevada. Chances are he'd only get one of them if the election were held today, and maybe none if Obama gets even a 2 point bounce from last night.

I'll do graphs tomorrow or later on. Other things to do today and as I said, most polls other than the snaps that we've already had won't mean much for a while anyway.

* * *

On a side note, anyone want to help me with a polling project I might play with? No skills required, just a little bit of your time. Let me know.



Recent Entries

Close