You would have loss that bet This is from a NBC /WSJ poll from 11-2,had Obama + 8,had more dems then repubs in the sample.Obama won by +7 Dem/ lean dem -45 Repub/lean repub -35
I was typing quickly. So you did a good job interpreting what I was saying. Well it does look to be like a big dem sample, but my understanding is that the pools are sample a blend of dems, republicans and independents. So your poll might have accidently turned out correct because independents turned out for Obama. Something we all know they are not saying they will do in the polls right now. I have trouble believing you are arguing this point. Are you saying that a poll which over samples dems by 7 to 11 points... is the most accurate way to predict the presidential race?
Not the results, the sample. If the sample is drawn randomly from the population in question, and is of large enough size, its composition can be relied on to a known degree of uncertainty to be representative of the population from which it's drawn.
The public does not show up to vote in proportion to the the population. (there are a lot of dems who don't vote and dont exist) Additionally, that strong democrat sample (over 50% of the dems sampled) is not representative of what is going on in the electorate right now. Obama is losing counties to inmates and "other". The only strong democrats are the crazy leftist spokesman we see on et and in the media. Most democrats are embarrassed democrats but forced to vote for the govt handouts.
Of course they do, mathematically they must. Using example only numbers, 30% of the population will vote, 50% could have voted, and the rest can't vote at all. If you took a large, random sample from the census, and eliminated all those in the latter category, you're left with a sample of "can vote". Randomly sample that and ask each of those drawn, "are you likely to vote?" Eliminate those who say, "no". The remainder, if you started with a large enough initial sample and have ended up with a statistically reliable (the validity is determined by the randomness) sample size, you also have a representative sample, whose members will be (this can be mathematically demonstrated) a proportion of the general public, and a proportion of the general public who can vote, and a proportion of the general who can vote and who will likely vote. Reexamining the results, after the event occurs, you should see, with a variance of "plus or minus three percentage points", that your sample voted just like the general population (who can vote and did). This is elementary statistics. Now, for those in the sample you mention, of course not all dems will vote, but neither will all reps vote. As for nonexistent members of the sample, maybe, maybe not, prove it. From what I can see, the various polling agencies are bouncing around a mean and I'm not seeing a consistent bias that would indicate anything besides random sampling. Admittedly, I'm only glancing at the results at this point, it's early.
I should have been more specific. Dems do not show up in the same proportion. We speculate it is because dems are famous for using groups like acorn to register voters... registered voters who are sometimes dead and sometimes frequently mickey mouse. However, the ratios could be off for other reasons. I do not "know" the reasons dems dont show, we just know they don't show up as much. pollsters are able to provide useful polls by doing polls which mirror previous participation rates.
http://www.bloomberg.com/news/2012-...s-romney-s-bain-record-over-job-creation.html Attack ads/Bain Capital under Romney
Thanks to rampant democrat voter fraud, an undeniable truth. Hard to vote from prison, cartoon land or the cemetery.