Obama’s campaign seems pretty confident that they’re ahead. And both campaigns seem to be strongly contesting states that Romney’s leading in, like FL and NC, while ignoring states like MI and PA.
It’s not bad strategy. If Obama loses either MI or PA it probably doesn’t matter what he does win. Romney making those states close is just a moral victory.
Right now, Romney is poised to fall just short all over the midwest. Minnesota he’s only three behind according to a Star Tribune poll.
Of course, I’d love it if he actually won those states. While losing Ohio, just to make Axelrod look bad.
Of course you’d love it- just like I’d love it if Obama wins NC, FL, and a surprise state like AZ. But it’s not likely to happen.
The likelihood of a Romney popular vote win and Obama taking the presidency on the strength of the EC is getting more likely all the time. 
Hey, I think I was in that poll!
Not sure how serious you are Jackmannii, but I would take that poll in isolation no more as the best approximation of the truth than I would the CNN poll that puts Obama up by 4. They have overlapping error bars (3.1 to 3.5%). Just those two together average at the same roughly 2.4% Obama lead that 538 comes up with using a much larger universe of polls. And today’s 538 has an interesting take on that sort of lead at this point -
Also a bit of some interest in the CNN polling data.
30% of votes cast in Ohio were cast early in 2008 and Reuters quotes experts expecting that number to be 35 to 40% this time nationally (and if anything likely higher in Ohio). If those latter numbers and the polling data they report are anything close to accurate the math then results in a solid Obama win at the end of the day.
According to the CNN poll, 40% of likely voters had already voted. But the county election offices have that number at between 15% and 20%, give or take. Even assuming CNN’s numbers are correct with regards to who early voters have voted for and who those who plan on voting on election day plan to vote for, and accepting that 20% of likely voters have already cast their ballots, then Romney would lead 48.4% to 44.6%. Which is why you shouldn’t put much stock into polls asking respondents if they’ve already voted early, as often times they are incorrect.
Can you point me to where it says that? I don’t see it. Not in the article or in the crosstabs.
I do see the 59 to 38% Obama lead among those who have or plan to vote early.
A poll by several major Ohio newspapers, including the Cleveland Plain Dealer, has Obama and Romney tied in the state at 49% each: Ohio Presidential Poll: Presidential race in battleground Ohio is dead even - cleveland.com
Nate Silver says Ohio is no longer a tossup. If true, 2nd term is a sure bet.
Yeah, that’s the poll Jackmannii linked to.
Where are you getting this?
Well, I sure hope so. I went to help phonebank at the local Obama campaign office on Sat. night and found, to my surprise, that they weren’t phonebanking that night. Looked like they were closing up for the night at just a little after 7pm. Hmm. Incompetence, overconfidence or something else…?
Nate’s current take on Ohio including that most recent tied poll:
Is 73% a tossup?
Yeah, Silver also said in his latest update that “it is normal for some polls to show him tied or trailing by a point or so instead in contrast to others that might put him four or five points up.”
Edit: Beat me to it.
We’ve kind of moved past **adaher’**s insistence on “The national polls say Romney’s ahead, so Nate Silver is wrong.” Methinks, though, there’s a serious misunderstanding in what margin of error means.
I just looked at the top five national polls listed on the RCP website. Four of those polls show Romney with a lead. All but one of those, though, had the lead within their margin of error. Lemme 'splain … that means those polls showing Romney ahead admit that their numbers might be off by enough that Obama is actually leading. That’s what margin of error means.
For example, Rasmussen shows Romney up by 3, 50-47. The margin of error is 3. Romney might have anywhere from 53% to 47%, while Obama could be anywhere from 50% to 44%. Those overlap … hardly a convincing Romney lead. Likewise with ABC/WaPo, showing Romney up by 1, with a 3.5 margin of error; AP/GfK showing Romney up by 2, with a 4.2 margin of error; and IDB/TIPP, showing Obama ahead by 1 with a 3.5 margin of error. Only Gallup has Romney’s lead more than the margin of error.
So just pointing at those polls saying, “Those numbers show Romney is going to win the popular vote!” is wrong and misguided. The pollsters themselves admit, by publishing a margin of error, that voters might go either way. Nobody is actually predicting what’s going to happen (well, except for unskewed polls, but that guy’s nuts). These are all educated guesses at best, and they can and do turn out wrong sometimes.
I guess I should point out that the RAND poll has Obama nearly up 7. RCP doesn’t include that poll in its national aggregate, though, so the race is ostensibly much closer than you’d think just going off of RCP.
Okay. Doing some math those results of those who “have voted or will vote absentee or early” (italics mine) come up with that 40% who have or who plan on voting absentee or early. You do understand why the votes that have already been received being less than the number of those who have voted early and those who intend on voting early is what would be expected, yes?
That’s not how margin of error works. The margin of error reflects the likelihood that random variation alone (you got unlucky and called too many Republicans, say) could explain the result. The 95% confidence interval means that 95 out of 100 random samples will be in that interval.
But it does not mean that all the results in that interval are equally likely. To the contrary, the result you got is the most likely reflect actual reality. That’s because the odds of randomly calling too many Republicans is low in a decent sample size, just like the odds of flipping a fair coin heads 5 times in a row. Moreover, when you aggregate polls, you effectively reduce the margin of error for the aggregate, similar to the way you would do so by doubling your sample.
You cannot therefore reason that a robust average of polls showing Romney +1 means he might just as well be -1. That’s not how it works.