I'd decided this story would be about Mitt Romney's clear surge to the lead (albeit a small lead inside the margin of error and nothing like the 4-5 point lead that Barack Obama enjoyed throughout September) in national polls, and the tightening of several swing states. Because when I made that decision on Saturday, that was reality.

Today it's not. Mitt Romney's gain from the first presidential appears to be fading in some polls and holding in others. Not the kind of trend Romney supporters would like to see.

RAND's five-day tracker of likely voters saw Obama fall from a pre-debate high of +6.3 points on September 30th to a low of +1.8 points on the 10th and 11th of October. Obama has made steady gains every single day in that poll ever since: +2 on the 12th, +3.1 on the 13th, +3.7 on the 14th, and +4.6 on the 15th. That's the best result Barack Obama has had from RAND since October 5th.

There hasn't been any movement from Gallup in either their likely or registered voter models in the past 4-5 days, but that doesn't mean Obama isn't gaining. Gallup uses a seven-day average and takes the longest of all the trackers to show big changes. Ipsos/Reuters maddeningly doesn't report results on the weekend, but they showed a strong lead for Obama before the debates, were slower to show Romney's gains, but did eventually show a +3 Romney lead on the 11th. That lead dropped to 1 point on the 12th, the last day they've reported data (we should get more data today.)

Which leaves Rasmussen.

We know that Rasmussen had a pro-Republican bias (4-5 points) in 2010 and missed some races by ridiculous margins (40, for one race in Hawaii) making them one of the worst pollsters of the cycle. But I think it's safe to assume that their bias is the result of a pro-conservative model and methodology, not data tampering, which means their poll result bias should at the very least be stable, and therefore meaningful from a relative perspective. So what Rasmussen, on any given day, it'd mean less than they showed candidate X leading by Y points, and it'd mean more than a comparison of the most recent two polls showed candidate X gaining or losing on their lead.

Rasmussen had Obama up by 2 points (and all other trackers had him up by 4-5) on October 4th, then had Romney leading by 2 points on the 5th and 6th, a tie on the 7th and 8th, leading by 1 point on the 9th and 10th, Obama leading by a point on the 11th, and Romney leading by 1 point on the 12th, 13th, and 15th (I forgot to write down the results on Sunday. If anyone knows what it was, please email me.)

The takeaways from the trackers: RAND's six-day tracker is showing strong gains for Obama over the past five days and a robust lead. Ipsos/Reuters will probably show either a tie, or a small Obama lead today if that trend is real. Rasmussen should have shown it before anyone, with their three-day survey, but again, Rasmussen has accuracy problems. Gallup will take several more days to show anything meaningful, if there is anything to show.

For what it's worth, the Investor's Business Daily tracker, which I have zero faith in, is also showing Obama gaining and Romney falling. Obama leads by 1 point, 10/8 - 10/13, after this poll showed a 5 point lead for Romney back on the 10th. That's a big swing. Real big, and consistent with what RAND is showing.

That leaves us with swing state polls, and non-tracker national polls.

A Politico/GWU/Battleground poll showed Obama leading by 1 point, 10/7 - 10/11.

An ABC News/Washington Post poll showed Obama leading by 3 points, 10/10 - 10/13.

This is after FOX News and Pew had Romney leading by 1 point (10/7 - 10/9) and 5 points (10/4 - 10/7), and the Washington Times showed a tie, 10/5 - 10/7.

What I find interesting and important from the RCP average is that in the last few days, the change from Romney leads to ties and Obama leads have come almost entirely from Obama gaining directly on Romney. The percentage of poll respondents saying they'll vote for Romney hasn't changed, but the percentage saying they'll vote for Obama has increased. Whereas Obama may have been behind 46 to 47.3 (avg) on the 13th, it's now a tie at 47.3 each.

Are Democrats becoming more energized, or are undecideds suddenly breaking for Obama? And why?

It could have been the VP debate. CNN did a snap poll of 381 registered voters after the debate, and found that Paul Ryan "won" 48-44. But there were problems. Remember when Republicans started denying polls back in September (funny how they are quiet now) because they thought pollsters were over-sampling Democrats? Well, it appears that CNN actually did over-sample Republicans. A CNN spokesman said the 33% GOP sample "...indicates that the sample of debate watchers is more Republican than an average of recent CNN polls of all Americans." CNN then stuck this disclaimer on their poll: "...the sample of debate watchers is about eight points more Republican than an average CNN poll of all Americans, so the respondents were more Republican than the general public."

CNN refused to release the internals of that poll, which casts even more doubt on it. Because a coverup is confidence inspiring, right?

Make of that what you will, my only beef with that poll is that it's meaningless because it was of registered voters, most of whom had already made up their mind. CBS sampled more people (500) and surveyed only undecided voters, 58% of the independents, which is where the gold is. And they found that Biden "won" the debate 50-31.

Make of that what you will as well.

Reuters/Ipsos did a special poll for the VP debate of 629 registered voters (meh) and found that Biden "won" 42-35.

The CNN and Ipsos polls clash while using roughly the same voter model, where party ID was 49% Dem, 39% GOP, 9% Indy, and 3% None Of That Bullshit, Son. If you take away leaners, it's 41% Dem to 30% GOP, about the same.

So 2/3 polls show that Biden "won", but not by the Romney landslide. Trackers show movement towards Obama, as do national polls.

But none of that actually matters. The states decide the election, so what are the state polls showing?

Here's what Colorado looked like before the Presidential debate: Obama +3, +4, +3, +5, +1, +2, +5, Romney +1, Obama +5, +3, +4, +6, +3, Romney +4, +3.

I hate to start sounding like conservative poll deniers, but it's worth noting that all the Romney leads during that span came from pollsters with a known Republican bias, like Rasmussen (+1), McLaughlin (on behalf of a Republican lobbying firm, +4), and Gravis Marketing (+3), a shady group that doesn't release full results or its methodology. McLaughlin is particularly questionable because it only surveyed *300* people.

Most of my election stories get more eyeballs than that, for comparison.

Here's Colorado after the debate: Obama +5, Obama +1, Romney +4, Romney +1, Obama +1, Obama +2.

Things appear to be swinging back towards Obama in Colorado.

Here are the most recent ten polls for all the states that CNN has considered a tossup since the summer:

Nevada: Obama +9, Tie, Obama +7, +2, +11, +1, +1, Tie, Obama +2, +4
Ten poll average: Obama +4
Five poll average: Obama +2

Colorado: Obama +6, +3, Romney +4, +3, Obama +4, +1, Romney +4, +1, Obama +1, +2
Ten poll average: Obama +1
Five poll average: Tie

Iowa: Obama +5, +8, Romney +3, Obama +7, Romney +1, Obama +4, +7, +4, +2, Tie
Ten poll average: Obama +3
Five poll average: Obama +3

Wisconsin: Obama +5, +7, +12, +11, +10, +7, +2, +4, +3, +2
Ten poll average: Obama +6
Five poll average: Obama +4

Ohio: Obama +1, Romney +1, Obama +1, +4, Romney +1, Obama +1, +6, Romney +1, Obama +1, +5
Ten poll average: Obama +2
Five poll average: Obama +2

Virginia is still being dominated by pro-Republican pollsters, with the last three coming from George Allen's Senate campaign, Rasmussen, and ARG. I don't doubt that Mitt Romney may be leading in Virginia, I just doubt that data coming from these three pollsters. Decide for yourself.

Virginia: Obama +1, Romney +3, +1, Obama +3, Tie, Obama +5, Romney +1, +7, +2, +1
Ten poll average: Romney +1
Five poll average: Romney +1

North Carolina, on the other hand, I have no problem believing. Obama only won it by 14,000 votes in 2008, but it is getting more blue by the day as NC moves away from tobacco and manufacturing to high tech, medical, and finance jobs in Research Triangle Park. But it's not there quite yet. What you might take away from this is any Obama lead should worry Mitt Romney, because he needs to win states like North Carolina to hit 270 in the electoral college. Simply having a tight race in this state means he's further behind than he's comfortable being, because it's that much more distance to make up and money spent that shouldn't have to be spent in a state that's traditionally as red as this. The district I live in makes some parts of Texas look liberal by comparison.

North Carolina: Obama +4, Obama +2, Tie, Romney +4, Obama +2, Romney +4, +9, +3, Obama +1, Romney +2
Ten poll average: Romney +1
Five poll average: Romney +3

Obama can't like some of this, but the trend favors him as Romney's lead has been shrinking.

Florida: Obama +1, Romney +3, +2, Obama +4, +1, Romney +7, +3, +4, +1, +1
Ten poll average: Romney +2
Five poll average: Romney +3

New Hampshire: Obama +1, Romney +3, Obama +7, +7, +7, +5, +15, +6, Tie, Romney +4
Ten poll average: Obama +4
Five poll average: Obama +4

Bottom line: there are 96 electoral votes in play in these states and Obama leads in 5 of them, taking away 44 of those votes. Romney leads in 3 states taking away 57 votes (29 alone from Florida which is trending back towards Obama.

Colorado is split.

Based on CNN's map, that's 237+44 (281) for Obama, and 191+57 (248) for Romney. Even if you give Colorado to Romney, he still can't win. It's tight, but not really close, if you will.

Put another way, there are 96 votes in play in those states and Mitt Romney needs 79 (82.2%) of them to win. With Florida trending back towards Obama, if he gets a lead there, those numbers change to 237+73 (310) for Obama, and 191+28 (219) for Romney. So you can see how important Florida is. If Obama wins Florida, that's the election. It doesn't much matter to Obama what some of these other swing states do, and he can afford to lose quite a few of them. Romney still can't win without Florida unless he takes a lead in almost all of these tossup states.

So this is worth saying over and over again: If Romney wins Florida, he has a real shot at winning the presidency, but he's still behind. If Obama wins Florida, it's all over by a wide margin.

Here's how things look right now, if tossups are pushed to the current poll leader:

evmap 2012-10-15

Update: As promised, here's the new tracker poll chart with two recent days of data from Ipsos/Reuters that I just discovered:

Ipsos shows a 2 point lead for Mitt Romney transforming into a 2 point lead for Barack Obama, confirming the movement seen in the RAND tracker and the new national poll by ABC News/Washington Post showing a 3 point lead for Obama nationally. I think it's very safe to say now that either Romney's debate bounce is fading, on its way to being gone, or that Biden gave his ticket a debate bounce of his own. Either way, Obama is back on the rise as of Monday, and back in the lead.



Recent Entries

Close