SARAH PALIN DRAWING LARGE CROWDS ALL ACROSS AMERICA
http://www.jeffhead.com/palin/palin-hiding.htm
One look at those crowds and people will know the real truth about what is happening on the ground in this campaign right now, and puts the lie to the MSM push polls.
Ping to self for later read.
The polls -- commissioned, after all, by media outlets -- are designed to extract the most advertising dollars from both campaigns. Presidential elections are the political Super Bowl for the MSM, and political advertising differs from the run-of-the-mill since it has to be paid for in advance. It's cold, hard cash for a media sorely in need of it.
So you bet they'll make it look close until the bitter end, on occasion showing one candidate or the other building a lead so the other candidate increases his advertising buys.
These polls are all about money for the Drive-Bys.
I'd add that as the election grows closer, and the MSM and their hired pollsters decide they've wrung every last dollar out of this thing, the polls will begin to more clearly reflect reality, in order to maintain the polling firms' reputations.
Poller: are you planning on voting?
H-mama: You betcha!
Poller: Will you be voting for McCain or BO?
H-mama: My father and I, and at least five of my “D” neighbors will be voting for McCain.....Guess that means, according to Murtha, we're living in a “racist neighborhood”!
Poller: sputtered.
I’m telling you, there is a quiet, and big, movement to McCain since last week, even before the debate. Dayton Daily News came out with a poll Sunday showing McCain winning Obama by 2. I have info that he may be doing even better than that. If he’s doing that well here, he has to be doing pretty well in PA.
The claim seems to be that because there are more of one group polled than another, the reported poll results will necessarily be skewed by the same amount.
In reality, any pollster would adjust the numbers to reflect the actual ratios of R, D, and "Other" voters, based on voter registration statistics, for example. A simple-minded approach would be to use registration stats to scale both samples to a theoretical 1000-person sample.
There are less tractable issues, such as (perhaps) differences in how easy it is to contact an R vs. a D -- are the former more likely to have unlisted numbers, for example? Are Republicans more or less likely to respond to a poll?
There are plenty of reasons to be skeptical of the poll results, but I seriously doubt that a failure to account for sampling differences is one of them. There is too much at stake, business-wise, for a pollster to make such a basic error.
Of course they over sampled Democrats. We have a huge advantage in registrations resulting from the need to rid Washington of the incompetent Bush administration that has been in charge of the stock market crash and jobs depression of the past 8 years.
And don’t forget the Bradley Effect.
Sample 2000 people. Take the 1000 that give the desired results.
How many do this?
Why do they over-sample? Is it real vote vs fake dem vote? Good Grief.
bttt