With the country focused on Hurricane Sandy, the election has hit the backburner a bit today. We do have a few new state polls though, and also finally have two new non-tracker national polls. The numbers are below, followed by my analysis.
* FLORIDA: Romney 50%, Obama 49% -- No change since 10 days ago (CNN)
* OHIO: Romney 50%, Obama 48% -- Romney +2% since 1 week ago (Rasmussen)
* COLORADO: Romney 48%, Obama 47% -- Obama +3% since 3 weeks ago (ARG)
* NORTH CAROLINA: Obama 45%, Romney 45% -- Obama +4% since 2 months ago (Elon)
* NEVADA: Obama 50%, Romney 46% -- No previous poll (CallFire)
* NEW HAMPSHIRE: Obama 47%, Romney 42% -- (Lake Research)
* NATIONAL: Obama 47%, Romney 47% -- Obama +4% since 3 weeks ago (Pew Research)
* NATIONAL: Obama 49%, Romney 48% -- Obama +3% since 2 weeks ago (Politico)
The two new national polls, both of which show shifts to Obama, are the first new national numbers in awhile that aren't from the daily tracking polls. Coupled with the state polls that we've been seeing over the past few days, it's pretty clear that Obama has gained some since the final debate last week. Romney got his first good poll number in awhile with Rasmussen's Ohio poll which shows him gaining 2% to take a 2% lead. Rasmussen has had about a 3% Republican lean this cycle, but more important than the margin is that the poll shows Romney gaining 2% compared to last week. Of course, out of the last 18 Ohio polls, Rasmussen is the only one to show a Romney lead, and most other pollsters have been showing Obama increasing his lead in the state, not decreasing. Point being, we'll need to see some other data support what Rasmussen's poll finds before we treat it as anything but a usual Rasmussen outlier. The number you should follow is the polling averages in each swing state. Currently they are:
* OHIO: Obama +1.4%
* FLORIDA: Romney +0.3%
* VIRGINIA: Obama +2.0%
* NEVADA: Obama +3.2%
* COLORADO: Obama +1.8%
* IOWA: Obama +2.3%
* WISCONSIN: Obama +2.2%
* NEW HAMPSHIRE: Obama +3.0%
* NORTH CAROLINA: Romney +2.5%
This is why the Elon poll showing North Carolina tied, while good for Obama in the same way that the Rasmussen Ohio poll is good for Romney, should be taken with a grain of salt because Romney remains ahead in the North Carolina polling average by more than 2%.
It's important to also acknowledge that Rasmussen has a very bad track record in Presidential election polling, especially when they are an outlier compared to the consensus (as most polls are when they are outliers vs. the consensus). They missed badly in pretty much every swing state in 2008. Here are some examples.
* OHIO - Rasmussen's final 2008 poll: TIE; Actual 2008 result: Obama +4.6%
* FLORIDA - Rasmussen's final 2008 poll: McCain +1%; Actual 2008 result: Obama +2.8%
* COLORADO - Rasmussen's final 2008 poll: Obama +4%; Actual 2008 result: Obama +9.0%
* NEVADA - Rasmussen's final 2008 poll: Obama +4%; Actual 2008 result: Obama +12.5%
My point here is not to make Rasmussen look bad, but to illustrate that it's the polling averages that matter a lot more than single poll numbers. If you have 20 polls showing one result in Ohio, and one showing another, or 10 polls showing one result in North Carolina, and two showing another, the final result is likely to be closer to the majority polls, not the minority. It's the same reason why the RAND tracking poll, which has Obama up 6%, should be treated like an outlier, just as the Gallup LV poll, which shows Romney up by 5%, also should be treated as an outlier. The national polling average shows the race basically tied, and that's probably closer to the real picture.
I also should note that polling may get very shaky in the coming days due to the hurricane. The states impacted by the hurricane, Virginia in particular, are going to become very difficult to poll accurately in the coming days, as will national polling due to the large number of people expected to lose power. It's very possible that we have seen our last truly accurate polling for the next several days.
Be sure to follow me on Twitter for more of my election analysis.