28 Days: Romney definitely gains in polls, with *lots* of caveats

I’ll do my best to give you as much polling information in as tight a package as possible. In service of that, here’s the executive summary: Mitt Romney gained on Barack Obama in tracking polls once they had 2-3 days worth of data, and a few swing states have tightened. It looks to be +2-3 points nationally for Romney and around that in the few states. But it wasn’t a big enough bounce for Romney to take a lead nationally other than with Rasmussen. Ipsos/Reuters, RAND, and Gallup showed gains, but already had Obama with a lead large enough to absorb a 2-3 point tightening.

There are two takeaways here. First, I’m personally wondering if this change in polling is more the good debate performance for Romney, or the final deflation of the convention bounce for Obama with until the debate, had not subsided at all. The reason for that is the second takeaway: Romney isn’t back where he started with this bounce, he’s actually worse off than before the conventions.

Here are the most recent tracking polls I have, as of 1am EST, October 8th. These polls vary in length from three days to seven (rolling averages), but all of them include Saturday the 6th. This story will be published much later in the day, possibly after more polls are available:

Rasmussen: Romney +2
Ipsos/Reuters: Obama +2
Gallup: Obama +3
RAND: Obama +3.8

Rasmussen has shown leads for Romney before while no other pollster has. Because of that, that result from them isn’t that big of a deal. Going back before the debates, Gallup had shown Obama with leads as big as 6 points, as did RAND, and Ipsos/Reuters was hanging around 5 points, while Rasmussen was down around +2 for Obama. The drops are relative and consistent; Rasmussen is simply leaning pretty far to the right these days and with the other pollsters showing results that close, this is what you can expect because this is what Rasmussen has been doing all year.

Here are the last ten polls before the Republican National Convention, tracking and non-tracking:

Rasmussen: Obama +2
ABC News/Washington Post: Romney +1
CNN/Opinion Research: Obama +2
FOX News: Romney +1
Resurgent Republic (R): Obama +1
Associated Press: Obama +1
NBC News/Wall Street Journal: Obama +4
Monmouth/SurveyUSA/Braun: Obama +1
LA Times/USC: Obama +2
Rasmussen: Romney +2

Obama’s average lead without Rasmussen before the conventions was 1.85 points, before the debate it was 4.1 points, and now after the first debate, it’s 2.93 points.

I’m not excluding Rasmussen to artificially hold Romney’s favorable results down, but because it’s difficult to take them seriously when they are often the only pollster in America showing the kinds of results they are, and it makes it very difficult to compare Obama’s lead during all three time periods. Amongst trackers, three show a lead for Obama and one for Romney. At many times, it’s been 15-to-1 amongst all pollsters. And it’s not just that other pollsters show a different leader. With RAND, Gallup, and Ipsos/Reuters, they agree with each other within the margin of error of each poll, which makes them far more reliable together than when considered alone. The confidence of three polls that agree to within such a small margin makes Rasmussen’s results even more of an outlier.

That said, it’s clear that Romney gained ground. There are two debatable questions from that: why, and does it matter?

You can argue about why. What matters is does it matter. While Romney gained ground in a couple of battleground states, he didn’t gain enough to turn a deficit into a lead in any of them with certainty. While confidence of calling tossups should be lower, which ought to increase Romney’s chances of winning (probably still well below 30%), it shouldn’t change any electoral college projections.

There are three ways of looking at state polling. Most recent results, trends, and averages.

Recent Results

For example, the last five polls for Colorado showed Obama leading by 6 and 3 points, Romney leading by 4 and 3 points, and now the University of Denver is showing Obama leading again by 4 (October 4-5). That last poll encompasses two days of post-debate sentiments. There are a few issues with those polls, though. One by McLaughlin (a conservative pollster, but that doesn’t mean biased) only sampled 300 people which is way too few. And they paradoxically showed a 4 point lead for Romney *before* the debate. The University of Denver only sampled 604 people, which isn’t invalid, but is about 300 fewer than you’d see from the New York Times, and is less than half of what you’d see from Quinnipiac.


The trend in Colorado is recently unstable. Of the last five polls, Obama has lead 3 and Romney 2. But before that, Obama had lead 11 of the previous 12 polls. The trend favored Romney, if not for the UoD result. If it had been a 1-2 point lead for Obama or tied, it wouldn’t have mattered. But to show a 4 point lead right after two good polls for Romney makes it too short to say anything for sure.

If I’m Romney, I don’t like that UoD poll one bit.


Pollster averages polls and the average still favors Obama, 47.6 – 46.5. It wouldn’t take much to tip it either way.

Obama’s leads in Colorado ranged from 1 to 6 points going back to April. It was probably a safe state before, less so now, but still not quite a tossup. The size of the leads are tough to resolve. +1 Obama here, +2 Romney there, that makes some sense. Noise and errors can account for that, as can the quality of a pollster and their legitimately different methodology. But I don’t think that any of those things alone can explain Obama +3, Romney +4, Romney +3, and Obama +4 over the course of ten days. Half those polls should be close to their margin of error and that kind of disagreement is manic.

My first reaction is to tell you to ignore everything between the 25th of September and 5-6th of October, until more data comes in. Especially with low-credibility pollsters like Gravis Marketing and McLaughlin accounting for some of the paradox data.

Push comes to shove, I’d put Colorado in the barely “Obama leans” column because the established history is Obama winning this state by about 3 points, and we don’t know how long Romney’s bounce will last, even though we now know how big it is. Romney’s biggest fear should be that his bounce was from his debate performance, and not Obama’s convention debate subsiding. Because if that’s the case, then when Romney’s debate bounce subsides, Obama’s convention bounce will return and may end up being permanent.

All of that said, here are some swing states that have recent data:


Rasmussen and We Ask America polled Virginia on Thursday and found a 3 and 1 point lead for Romney respectively. Before that, Obama had lead 14 straight polls between 2-8 points. PPP has already broken the new trend, and shown a 3 point lead for Obama. Just like we saw in Colorado with the University of Denver.


A similar story. Rasmussen and We Ask America polled Florida on the 4th, and found a 3 and 2 point lead for Romney. Before that, Obama had lead in 9 straight polls.


We Ask America polled on the 4th and found Romney up by 1 point, which is crazy when you look at more recent pre-debate results:

NBC/WSJ/Marist: Obama +8
PPP: Obama +4
Columbus Dispatch: Obama +9
Quinnipiac/NYT/CBS: Obama +10
Washington Post: Obama +8

Then the debates gave us:

Rasmussen: Obama +1
We Ask America: Romney +1

It’s unlikely that Romney’s debate performance didn’t just wipe out an 8-10 point lead, but reversed it. Even a 1 point Obama lead is stretching things. Before this, Obama had led 16 straight polls in Ohio. Very questionable results here.


Neither We Ask America nor Rasmussen polled Wisconsin after the debate, but PPP did, and found Obama leading by 2 points, down from 10, 11, 12, and 7 points in the four previous polls. WAA was the one that gave Obama the 12 point lead, September 20-23. PPP’s results are genuinely surprising and it makes you wonder why Rasmussen didn’t get in there first.


Gravis Marketing hit Nevada the day of the debate and found Obama leading by 1 point, but that almost certainly does *not* include debate sentiments. That’s another very strange result since We Ask America found Obama up by 11 about a week before that. Obama has lead or tied in 21 straight polls (two ties) there. The 11 point Obama lead from WAA was the highest all year.

Neither Rasmussen nor We Ask America polled Iowa where Obama has been pulling away since early August.

So what does all that bullshit mean?

Two things: yes, swing states are tightening in favor of Mitt Romney. And what nobody is saying: nearly *all* of that is the result of two polling firms, We Ask America which I’d not heard of until this year, and Rasmussen, which has documented bias and inaccuracy issues. Only PPP has confirmed tightening, and only in a single state with a single poll.

I do think some swing states are genuinely tightening, as are national polls. But here’s the bottom line. Without Rasmussen, Romney is further behind after his boost from the debate than he was before the conventions, and Rasmussen is responsible for like half or more of the tightening polls we’re seeing in swing states. What we may be seeing is a biased exaggeration of a genuine tightening.

Rasmussen was accused of flooding the zone in 2010, meaning they were putting sometimes as many as ten times the polls into the field as anyone else did, in order to game the media narrative. If you had five recent Rasmussen polls showing GOP candidate X with a lead, and one poll showing something else, it’s hard for the ADD media to get the right story. Rasmussen got a lot right in 2010 in the horse race elections, but missed the final result by large margins. In other words, they called the correct winner with decent interval, but missed the actual result way too often.

This may be a smaller version of that where you pretty much know that you’re going to get a result favorable to Mitt Romney by default, and Rasmussen intentionally set out to do that in swing states to generate a narrative that’s friendly to the Romney campaign.

Or they could be right, and literally every other pollster in the country is wrong.

We’ll know this week as other pollsters start releasing their results.

Update: Rasmussen has gone from a two point Romney lead on Sunday to a tie today; Obama has gained two points in the Gallup tracker from yesterday’s average, and Obama is down 0.1 points in RAND’s tracker. The consensus gain was 2-3 points from Romney, closer to 3 than 2. But Romney has already lost 2 points of that *today*. Was his bounce that fragile? Did the big drop in unemployment nuke the bounce? Rasmussen (take with a grain of salt) reports that consumer confidence jumped 10 points after the Friday job report. This whole thing may have all been for naught, we’re close (1-2 points) to being right back where we were a week ago.

* * *

And the news…

– The Times took a hard look at Mitt Romney’s claim to having a strong bi-partisan relationship with Massachusetts Democrats while he was Governor. What they found was far from pretty. Some (unnamed) sources say Romney’s relationship with state Democrats back then was about the same as Obama’s relationship with House Republicans today.

– Two-in-three Republicans in Minnesota believe the Bureau of Labor Statistics manipulated the unemployment rate to help Obama. The AP mocked this crowd in the vein of Bigfoot believers and people who deny that NASA put a human being on the moon.

That number is not trivial. 66% of Republicans in any state is just insane. Republicans in very large numbers now deny: evolution, climate change, Obama’s birth place, Obama’s religion, polls, and now job reports from non-partisan career civil servants, many of whom served under President Bush and would serve under President Romney if he wins.

Although poll denial has already began to subside now that Romney is doing better in them since the debate.

– NBC is telling the Obama campaign to stop using a 5-8 second clip of one of its cable shows in its campaign ad, where Romney is attacked for his statements about his tax plan during the debate. (Apropos of nothing, I’ve seen this ad non-fucking-stop all weekend. It’s the kind of hard hit that Obama wouldn’t throw during the debate itself.) They’ve done the same thing with the Romney campaign, showing in my personal opinion how extremely rich and powerful corporations think they own literally everything, and how NBC in particular has no reasonable understanding of either the first amendment or Fair Use.

– The money war just changed in an epic way. Obama+ have reported raising $181 million in September, after raising “just” $114 million in August, and losing the race for several previous months. The Romney campaign hasn’t released their numbers yet. It makes you wonder if the “October surprise” we’ve been hearing about from conservatives may be the Romney campaign getting crushed to shit in campaign fund-raising and spending. Perhaps not what they were hoping for.

– This is from Dkos, for what that’s worth: The Senate race is very fluid. Democrats could still lose it in a bad election, but could actually gain ground in a good one.

– (No link): Some things I’ve seen on Twitter raise an interesting point. It’s a right-wing meme now that the unemployment rate has been faked to help Obama, even though that’s retarded. But it wasn’t even 48 hours earlier that Mitt Romney was using the BLS unemployment rate to bash Obama over the head during the first debate, nor has it been uncommon for the Romney campaign and like every Republican in the country to bring up that rate every day since Obama was elected.

It’s just like the “all polls are skewed” meme that is now quickly dying because the polls are changing after Romney’s good debate performance. It appears that after expressing a great deal of concern over Romney’s electability early in the primaries, the conservative base, pundit class, and professional political class have completely and seamlessly embraced and adopted Romney’s flip-flopping.

More later.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.