Survey USA's PA poll released yesterday had Hill up by 12 over Obama.

www.surveyusa.com/.../PollReport.aspx

Survey USA's been on right on the money this election season.  PPP, on the other hand, has had an 8 pt. margin of error on GOOD days.

Looper is right- as little as I pay attention to the polls, I know PPP is consistently off.

But I think the broader point is correct- PA has been trending toward Obama, which generally happens the more people get to know him.  I doubt it will be enough to win, though.

No, he's not gonna win here. If he does, its just a massive repudiation of Clinton. Probably just voters feeling that the race is really over - not so much that they prefer him. But I don't see it or feel it. I think she wins and probably by close to 10 points. If he keeps it under 10, good for him. If she gets above, say, 12 or 15, very good for her. This is based on nothing more scientific than my gut and knowing the similarities between here and Ohio - and she has better infrastructure here than Ohio and no independents mucking things up. If he's closer here than he was in Ohio, that's VERY good for him. Ohio ended up at what, 10 points?

I don't think Hillary is going to take the state by double digits.  Obama has plenty of time to make his presence felt in the state, and plenty of money to spend.  My gut tells me he'll narrow her lead considerably.

Also: "wolls around"?  Are you Barbara Walters or Lili Von Schtupp now?

I think ramboorider's right. I don't know if he's in Eastern Pennsylvania, but I'm in Western PA near the Ohio border. Things appear to be trending toward Obama, which is remarkable in the aftermath of the Wright controversy and given the fact that this a blue collar area that has not historically been progressive in matters involving race (I understate my case). Still I don't see Obama carrying Western Pennsylvania, which is not yet fertile soil for a new generation of Democratic Party politics. This is still very much old school Hillary country.

scott, and I am not supposed to be surprised that you are a Hillary supporter. Amazing the assertions people make. This rock solid poll of yours is based on 588 Likely Voters. 588, not 588 thousand, but just 588 people, out of a state with what 10 million people. If they were right in the past, they were lucky because 588 people are evidently nowhere enough people to get an accurate impression.

regardless, of course even if Hillary wins Pa. by a few points, it is meaningless. She needs to win by at least 20 points, preferably 30 points, to make a true difference and then she needs to eke out a win in North Carolina. Until then, Hillary doing a happy harpy dance after Pa. due to a 8 point win is meaningless, like a football player doing a dance in the end zone even though his team is down by 30 with 5 minutes left.

Actually, PPP has been consistantly on the money this election. In Texas, PPP's final poll had Clinton winning with +6%. In Ohio PPP's final poll had Clinton up +9.

Honestly, I had never heard of PPP until this year, but they have had a great track record so far.

"Amazing the assertions people make. This rock solid poll of yours is based on 588 Likely Voters. 588, not 588 thousand, but just 588 people, out of a state with what 10 million people. If they were right in the past, they were lucky because 588 people are evidently nowhere enough people to get an accurate impression."

Blackton,

Actually its not very amazing if you took to the time to understand the statistics.  The margin of error for even such a small poll is still pretty small (4.1%).  Plus as the polls get larger there are diminishing returns with respect to increasing accuracy (A poll of 1549 people like the last Quinnipiac one gets you down to about 2.5%).

The only amazing assertion here is to compare the sample size of 588 with the size of the state.  The margin of error of a poll has nothing to do with the size of the population.  The SurveryUSA poll would still have a 4.1% margin of error whether the population of PA was 1,000,000, 10,000,000, or 100,000,000.

Maybe some remedial statistics classes are in order?

right now, I would say that if HC only wins by 5 points, that is a huge Obama talking point. Remember, polls in CA and OH had Obama really close or even ahead and on primary day, HC lambasted Our Chosen One....

I will "exhilerate per The Grudge Holder" only win I see the actual vote count numbers...

I should add one caveat to my comment above: population size doesn't matter so long as it is large relative to the sample size.

Intuitively, it seems that a larger sample size in relation to the population will produce a more accurate result.  But when you're talking small sample in what essentially becomes an infinite population (588 versus 10 million) it doesn't really come into play.  So amarkle is correct, but this really only applies to the probability that the sample population will be weighted towards any one outcome.  Realistically there are too many factors that cannot be adequately accounted for, such as the phrasing of the questions, geographical weighting and so on.  Those are all arbitrary and prone to human error, and the true margin of error can never be calculated.

amrkle, hey I never claimed to be a statistician, I teach language, but I still am not going to place to much faith in a poll with such a small sampling size, especially in a state as large and diverse as Pa. Just who are these 588 people? the breakdown was only by age and sex, but not by ethnic background or economic status, also at what time was the poll taken? As I said, I don't do statistics, but you know the old line, there are three kinds of lies: white lies, damn lies, and statistics.

I think Clintons overblown daring in Bosnia has and will continue to have an effect. Any, and I mean any, other politician would have been drummed out of contention for such a 'misspeaking' . I just don't think the whole story telling sits well with people. Big talking bullshitters are just that and little more. They are easy to discount and deservedly so.

During my time in the Navy our ship had occasion to hit General Quarters with full battle regalia and train our resources upon Soviet spy planes that were engaging in a pissing match common enough for the era, circa Vietnam. I was charged to employ some lethal gear for their attention and consideration as a due course in flipping them the collective bird. Nothing more and nothing less. Now I suppose that I  could, through various descriptive means, employ through the suggestive drama of a bona fide battle station action to lead you to believe high drama and imminent danger. That would make me a bullshitter. To suggest that hot fire conditions existed would make me a liar.

oh, and amrkle: Maybe some remedial statistics classes are in order? Hey, I haven't even gotten up to remedial yet. Math has always been my decided weak suit. It is funny, I had to edit a paper for another University professor who specializes in statistics, didn't understand a bit of it, but I was able to tell if he had an incorrect noun clause marker, but as bcbaird has come to my rescue somewhat, I will leave the math to others. But what I don't understand is, if statistics are supposed to be so precise, why the hell have so many polls been all over the place?

bcbaird is completely correct.

I guess the larger point I wanted to make is that people frequently harp on the small sample size of polls when really even (seemingly) small samples can be virtually as good as larger samples.  The larger sources of variance between polls (as bcbaird notes) are the phrasing of questions, robo or in person, how they define likely voter, etc.

How they define likely voter is the major weakness in polls that look at likely voters rather than registered voters.  Risky business at this point.

Roi - they state "likely primary voters" which I hope would be comprised solely of registered Democrats, seeing as the PA primary is closed.  I didn't see any elaboration on the term, so I can't be entirely sure they made the distinction.

Different polling firms identify likely voters in different ways.  Some have formulas they use to predict who will actually vote based on projected turnout and other factors, while others take the simpler approach of simply asking people if they intend to vote in the primary.  I'm not sure how PPP does it.

Winning would be nice, but he's just there to scrape away at the lead, and take as many delegates as he can. He works hard even in states he's not going to win. That's part of why he's winning in delegates.

Those interested in becoming more poll-literate can find a wealth of information here:

www.pollster.com/.../FAQ.php

"By the time the primary wolls around, expectations could change to the point where it would be a huge Clinton win."

Huh? Since Clinton is significantly behind in delegates and popular vote, she needs to make major inroads in those areas. In any expectations scenario, and eight point win doesn't do much for her. In fact it helps Obama to run out the clock on her.