I have never believed Gallup polls. They were near the bottom in accuracy in 2008.

Why would I believe any article written by one of the Gallup spinmeisters to justify their inaccurate polling?

I wouldn't.

Anyone who reads this and can think for themselves will realize that what this guy is saying is not only true but also makes all poles irrelevant. If party affiliation varies so much, it is useless and void to use 2008 party identification to measure ones support in 2012.

''Figures never lie, but liars often figure''

Poles irrelevant? Then why are the candidates in Cleveland? *rimshot*

This article is Gallup defending Gallup ie protecting their revenue stream. They've already shown themselves to be corrupt and cowardly. Why believe them now?

#4 - I thought the same thing, but wanted to see what others thought too.

Sorry, but the spin made me dizzy. I think this guy is falsely representing what people have been saying about the polls in the first place. He is using totally fales premises. Then he states some obvious facts about polls and questions that COULD be asked (not that were) to qualify party affiliation. Then he finally concludes that the polls are something that they really are not based on the scenario he spun.

Sorry, but as far as I am concerned, all polls are a joke and when someone puts this much into "explaining" his polls, I know that he is doing even more misleading than he claims others did. Someone needs to tel this guy that nobody is claiming to ask people their party affiliation in 2008. That wasn't what the concern was in the first place. The question is how many democrats overall were polled vs. Republicans and Independents. And if you are using skewed numbers like D+7 or more every time, then something isn't right.

Oops, I meant "he is using a totally FALSE premise" -- not fales

The only time I was called in a presidential political poll, I was asked if I had time to answer a few questions. I said "yes". The first question was who would I vote for for president. When I answered, "George Bush", I was told they had enough of that choice, and they wouldn't need to take up any more of my time.

So much for polls -- I don't trust any of them!

Some polls ask ''who did you vote for in 2008?"

That is often interesting. How many folks have changed sides?

Regardless, only after say the second debate will public opinion finally start to focus on this election.

Sounds like CYA to me...

OP, agreed, this seems more self-serving than a McDonald's [repulsive, usually] soda bar.

I kept asking myself, "Then why do you ask? Why?" And at the end, we get to FTA: "I don’t personally put a great deal of stock in survey-to-survey variations in party identification. All of our weighting focus is on the effort to bring more solid demographic variables into alignment with census figures -- including in recent years cell phone and landline phone use. We don't find that party identification is stable enough to be of much use when it comes to comparing sample-to-sample variations, or sample to exit poll Newport admits it's not much use.

And these guys are touted as the Gold Standard? This is "scientific"?

Waiter? Please get me a break, an extra-large one, if you would. Thanks.

In polling, a sample is a sample. What would skew the sample would be any unaccounted for variables in access or response--i.e., which people answer the phone, which people take the time to respond to questions.

Is it possible that more self-labeled Democrats make up a larger portion of Ohio voters today? Is it possible (or just self-servingly snotty) to wonder if more of the un- & under-employed who are willing to find work have left Ohio for the opportunities "red" states offer? Did they thus leave an increased percentage of those who are attached to the federal or union teat?

Or, is it possible that more conservatives are working several part-time jobs and therefore less willing to give time to the pollsters?

IMHO, Newport is making a sound (if limited) point about polling methodology. There are lots of other ways to end up with a skewed poll.

If it were just one poll complaining about the sample might make sense. But it is virtually every poll. Even if you look at the Rasmussen state by state Obama is far ahead.

#13, Obama and Romney are tied overall in Rasmussen, and if you look at Rasmussen's state by state polls, there are far too many even-up battleground states to declare anyone the winner at this point.

Frank Newport does not explain why Obama is tied with Romney in Rasmussen (the most accurate poll over the past several national elections), but has a big lead in all the polls conducted by left-leaning news organizations and universities. And yet the explanation is obvious -- those organizations are deliberately oversampling Democrats in order to publish news stories that Obama is winning.

D+8? D+11? Surely you jest.

I really doubt that party identification is as fluid as he claims.

What is fluid is turnout. What are the chances that turnout for Obama will be as high as it was in 2008? If the answer is "not likely" then why are polls assuming turnout for Obama even higher than 2008?

If party identification is just a will-of-the-wist, here today gone tomorrow, ephemeral sort of thing, why do you guys weight your poll results with such a dicey statistic as party identification four years ago? Isn't that likely to yield dicey results?