nate33 wrote:...in a recent Public Policy Polling poll (which had Clinton up +5), their sampling incorporated 52% people who voted Obama, 41% McCain, and 7% "don't remember". Why on Earth would they use that as a sample when in the actual election, we know it was 51-47 Obama over McCain? Basically, all of those points favoring Hillary were due to a sampling that was 7% heavier weighted in Democrats than Republicans than in the last election. (It's actually more likely that Democrats will be slightly less enthusiastic in turnout because there is no Obama on the ticket.)
I love the "7% don't remember!" Funny....
I don't think the 2d bolded point can be concluded as you do. To my knowledge, people run polls because they need real information -- the best they can get. Hence this was certainly a randomized poll, not a weighted sample chosen by the pollster. I presume that, in addition to getting an answer on the respondant's preference in the coming election, the poll also asked whether the person voted, and who for, in '08.
People's answers to a question like this are not always reliable -- not just the 7%, some of whom may very well remember who they voted for but don't want to say for one reason or another. Or didn't vote at all. And some of those who say they voted for Obama may actually have voted for McCain. Or vice versa. Or stayed home for that matter.
It's my understanding that, given a poll has been randomized in a professional manner, the rest of the factors (including that randomization itself surely can't be 100%) go into making up the margin of error. Which is considerable.
As it happens, I know a techie high up in a company that does big data analysis for organizations along one of the two big vectors (progressive - conservative). He says that everyone in the industry, whoever they do this kind of work for, is kind of freaked out about polls these days. Why? Because no one answers their landlines any more! I've had two calls I didn't answer while writing this post!

IOW, who the f#@k knows what's going on?!?! Weird times, that's for sure.