Check it out here: http://www.freerepublic.com/focus/f-news/2123557/posts
Good Hunting... from Varmint Al
- M.E.
POLLS!...SMOLLS!
VOTE NO TO SOCIALISM TOMORROW!
So the polls you disagree with are nonsense and the polls you agree with are not? Does that pretty much sum it up?
Welcome to FR.
I hate to say this but for just about every national poll to finish on election eve with movement towards Obama is not good news. EVERYONE here said they would tighten up in the final days of this campaign and they have not. We have seen movement in the state polls towards McCain but I just cannot imagine that all of these national polls with Obama +5 and up are going to hang their reputations on the line just to TRY to help him get elected.
Get out and vote, drag everyone you know to the polls to vote for McCain/Palin because I think it’s gonna take a miracle now. We are in for 4 or more years of complete and utter poltical devastation the likes of which none of us have seen.
Welcome to FR.
polling is a science. It’s only as reliable as the person applying the science.
The underlying assumptions on most of the polls are what skews it to Obama in numbers that are too large IMO.
When McCain pulls this out, a lot of pollsters will need to really evaluate how they weigh certain assumptions after all is said and done because they would have done a great disservice to the country because there will be riots, and people will die.
I worked with pollsters and most of them are egghead types, people that get too buried in numbers to actually see whats going on. There is also the agenda factor that’s in play too and there is no equation that can brush away personal bias.
Check it out here: http://www.freerepublic.com/focus/f-news/2123557/posts
Good Hunting... from Varmint Al
You have understated the effect of oversampling.
As explained in the post, How to Adjust for Oversampling, a one percent oversampling error results in a reduction of about 0.7 percent from one side (not every person counted will vote against the oversampled party) AND an addition of about 0.7 percent to the other side.
Therefore, in your example, if the ABC Poll has indeed oversampled the Democratic side by nine points, each side will change by about 6.3% -- for a total swing of 12.6%. In other words, Barry drops from an 8 point lead to a 4.6 point deficit.
Similarly, a ten point Gallup oversampling error results in a 14 point swing -- and Barry's eleven point lead drops to a 3 point deficit. Well withing the MOE of both polls.
You might find it interesting to adjust the polls for their oversampling bias and note they are actually pretty close when they don't try to cook the books, and they almost all favor McCain.
excellent post! ty!
Welcome to FR!
It’s broken glass time! Everybody turn out! Turnout is what will turn a loss into a win or a squeaker into a blowout! Let’s cream Obama!
It's an interesting way to forecast tomorrow, to see how well pollsters did in predicting the Hillary vs. Obambi races. I wonder which pollster did the best.
It seems to me that if there is still indeed a Bradley factor, it would have been most evident in those contests.
Not to put too fine a point on it, the polls are not nonsense. If you take the data the pollsters collect and weight it using their formulas on voter turnout, the polls are accurate reflections of the data. The validity of the weighting factors creates the polling variation on the Dan Rather TANG document: polling data can be accurate but incorrectly skewed. 8-)
What the headlines are not telling you is that the pollsters' underlying assumptions on voter turnout are sheer nonsense. There is no scientific means of determining in advance the percentage of Demonrat voters vs. Pubbie voters in the general election. By ignoring historic voter turnout norms from 2000 and 2004, this year's polls are skewed and likely inaccurate.
There is no way to determine how the voters will vote for those who hangup when the pollsters call. An assumption is made that they will reflect the general population's leanings. Believe this one at your own risk.
Bottom line: the polls have entertainment value and this year (as in most years) give great comfort to the left. But the polls have NO predictive value whatsoever unless their turnout models happen to fall in line with the actual voting population.
WELCOME TO FR, FRiend!!
Big Bump!!
UNION VOTERS
PUMAS