It’s worth approaching these results with a note of caution. Recent polls actually show a remarkably consistent level of support for Obama (InsiderAdvantage: 42%; PPP: 43%; SV: 42%; Rasmussen: 43%; Quinnipiac: 44%). Even the outlier – SurveyUSA – puts Obama at 38%.

On the other hand, when it comes to Hillary, we see more variation: Insider Advantage: 45%; PPP: 46%; SV: 47%; Rasmussen: 48%; Quinnipiac: 50%; SurveyUSA: 56%. Why the spread? Well, there’s a fairly strong inverse correlation with the number of voters selecting neither candidate and Hillary’s level of support. IA: 13%; PPP: 11%; SV: 11%; Rasmussen: 9%; Quinnipiac: 6%; SurveyUSA: 6%.

I’d suggest that what we’re seeing here has been a constant feature of rust-belt polling this cycle. Hillary has an early lead, and Obama’s levels of support are low. Obama then builds his level of support – in this case, it appears he’s in the low 40s. That leads to headlines trumpeting a “tightening race.” The problem is that undecided voters this cycle have broken heavily for Hillary. She seems to be the default option – if Obama can’t win over a voter by the time we get within a day or two of the election, they’re likely to vote for Hillary. So Hillary stages a “comeback.”

Of course, the underlying reality is different – in Ohio, for example, an election eve Zogby poll found a tie race, with each candidate garnering 44% of the vote. That was roughly the average level of support for Obama found in polls taken in the final week, and in fact, exactly what he received on election day. The problem was the other 12% of voters, who broke overwhelmingly for Hillary. It’s one reason why SurveyUSA has been so accurate this cycle – they don’t offer “undecided” as an option in most of their polling, and they force voters to commit to one candidate or the other. I’m not endorsing their latest poll – it certainly looks like they’re at least slightly underestimating Obama here. But I suspect that if the election were held tomorrow, Obama would get about 44% - and Clinton the rest.

I am willing to believe that undecided voters are more apt to put that X next to Clinton's name, for a bunch of different reasons. Name ID is one major reason for honest to god undecideds. She is a known quantity, more or less, for someone who can't make up their mind.

I don't know that there is any way on earth to really come to some agreement over how many voters that represents. At this stage, I would imagine more voters are decided than not.

When Hillary was up 20+ points, we didn't believe the polls.

When Obama was up by a point, we didn't believe the polls.

The only poll that matters is on 4/22.

Chuck Todd says campaign internals say the race is within single-digits and closing. That's good enough for me.

I'm as sick as anyone else of the people constantly shrieking that TPM is in the tank for Hillary, but that said, the analysis in the last paragraph really is twisted into pretzels to obtain a favorable result for her.

A more realistic analysis to me is that Obama's support among people who want the contest to be over is just barely lower than among people who want it to continue. Considering that if Obama wins Pennsylvania, the contest is over, he should be able to pick up considerably more support from the people who think it should be, and I imagine that as more people make the argument that Hillary needs a win in Pennsylvania to survive, and that there's no realistic scenario in which she can claim a win before early-to-mid-June, the more these people will move to Obama.

As an Obama supporter I was very disturbed by yesterday's SUSA poll which showed Obama trailing Clinton by 18 points -- actually regressing 6 points compared with SUSA's tracking poll of the previous week. This was in shattering contrast to yesterday's encouraging Quinnipiac poll which showed Obama only 6 points behind and apparently closing. What a difference! And who am I supposed to believe? SUSA has a very good reputation, and my instinct was to credit its results over Quinnipiac's pretty much without question. However, I just read an interesting take my Mark Blumenthal over at Pollster.com. He argues that that the Q poll might be more reliable than SUSA in this case. Here is his rationale:

{F}irst, the Quinnipiac University poll has been conducting very large samples of likely Democratic primary voters lately (ranging from 1,304 to 1,549 interviews on the last three surveys) that reduces the sampling error within smaller subgroups. Second, Quinnipiac has long and extensive experience in Pennsylvania. The pollsters I know who have worked in the state hold the poll in high regard.

Third, perhaps indicative of the first two factors, the current Quinnipiac results -- showing Hillary Clinton leading Barack Obama by six points (50% to 44%) -- track fairly closely with our overall trend estimate based on all polls (51% to 43%). Finally, Quinnipiac is one of the only polls that regularly tabulates its results by education and Doug Schwartz, the director of the poll, has been kind enough to share additional tabulations that do not appear in their public release.

Informed comments, anyone?

Here's the link: http://www.pollster.com/blogs/update_pennsylvania_results_by.php

Let's all agree that a political campaign is a complex human event, involving tons of objective and subjective variables (demographics, prejudices, momentum and the like). No two races are alike, even though we all try compare the last states (Ohio and Texas) to this one.

Having said that, allow me to share these observations. Two weeks out from the Ohio primary, Obama was 14 points out in the average of polls on the RCP site. Now he's 7 points out in the same average in Pennsylvania, which is about where he was in Texas at the same point. Obama eventually surged ahead in Texas, and got closer in Ohio, and then fell back in both states amid Clinton's ad and the NAFTA-Canada thing, and other stuff. But I think the real reason for Obama's late drop in TX and OH was the whole "default" argument: people just don't know enough about the guy, and aren't willing to sign up for the wedding yet. So they go with the proven girl: Hillary.

Logic and human nature would likely have it that this reluctance to embrace the newcomer-frontrunner will be less of a factor as time wears on. It seems likely to me that Obama will continue to pull a bit closer to Hillary and that the final deciders will break for Hillary, but perhaps not as dramatically as in New Hampshire, Ohio and Texas. If all other things remain constant, I'd expect Hillary to win by just two or three points: in the end, Hillary's demographic advantage will only slightly outweigh Obama's money/advertising advantage. The other factors that may shift the race one way or another are 1) a new scandal/gaffe; and 2) voter turnout, and the youth vote.

I guess we'll see in 13 days.

Off topic but Politico has a good article about Hillary's managerial incompetence:

"Clinton leadership a study in missteps"
http://www.politico.com/news/stories/0408/9478.html

Very good point.

All the demographics in Pennsylvania heavily favor Senator Clinton, and all the big political operations are working to turn out the vote for her.

Hillary should win at least 60% of the vote in Pennsylvania.

Anything less than that will be a disaster for her, given all the advantages that she has in that state, and she has the two term President campaigning heavily there.

Anything less than a 20% margin of Victory for Hillary in a state that is completely set up for her, will expose her for always going wobbly coming down the stretch. That would make her a terrible candidate to bet on for the big race in November.

"Give me more white people! We need more white people!" This is being reported in a Pennsylvania paper as what was being shouted by an event coordinator at one of Michelle Obama's political events in Pittsburgh, Pa. They wanted more white people to be seen in the crowd behind Mrs. Obama.

http://dyn.politico.com/popvideo/kotecki.cfm