Two reasons a lot of people are doing that -
Then you get 2000 Florida situation, where those Nader voters clearly meant to vote for Gore.
Over here we have cardboard boxes... and not even the ones with thick walls.
While the second one is an obvious concern, there really isn't evidence of actual oversampling, at least not for recent presidential elections, and definitely not consistently over history.
CNN just released a poll yesterday that showed obama and romney tied if you oversample democrats by 11 points. Eleven. You can't tell me there's not a problem with party ID oversampling. Furthermore, the same reports are showing "independents" - which we're told are the key to every election, after all - are going toward romney by 22.
The proof is in the pudding, as it were. The nice thing about the statistical models is that we can find out how accurate they are when the votes are all tallied. Believe it or not, many pollsters are genuinely trying for accuracy. As for Nate Silver, he has been repeatedly saying that for Romney to win, polls have to be statistical biased towards Obama. He also points out that historically that degree of bias has not occurred AND he accounts for specific polling firms' bias in his own models.
You are aware how statistics work, right? Even a 90% predicted victory from 538 doesn't mean that a landslide is being predicted.
Whoa, whoa, whoa. Gas once read a thing on the internet, man. You can't go discounting that!
That looks more like a case of bad touch screen recognition on likely outdated tech than a vote being changed.
Posted in another thread, but it's really not a problem, since it shows the check. The voter can go to a precinct official and have the machine fixed or taken out of service.
I identify independent, which to me mostly means I'll never punch a partisan ticket.
I have a feeling we're going to see a lot more of that video, along with a lot of people late to the party and with their fingers in their ears.
The single most definite indicator of something's veracity is that CDS thinks it's wrong.
I'm sorry I insulted your man-crush, MD.
I didn't say it's going to be a landslide either way. I said the data we have is unreliable and some people go out of their way to make it moreso. Polls aren't about showing what people think anymore (if indeed they ever were), they're about trying to make people think they way the pollster wants them to be thinking.
Probably, but only if it ends up in a swing state, and that state goes for Romney by a very small margin.
Fortunately (?), at this point, the only state projected to be close enough for faulty voting and recounts to matter would be Florida.
Nope. It's gonna happen no matter who wins and it's going to make us look like idiots. The American system of voting is a joke. It's partisan, outdated, and rife with stupidity.
I actually know nothing about Silver at all. But he does have a clear grasp on statistics and statistical theory (I teach it myself) and he explains his methodology, both qualities that don't exist in some predictions people are making.
Whether there is oversampling or not, that democrats were represented at D+11 over Republicans in a poll is not proof of it; it's merely proof that there were more Democrats in the poll.
From what I've read, those new "I" affiliations sound more like disaffected "D" types rather than crafty "R" types.
C'mon guy, this is GB, he knows that very well... never stopped him before.
I really thought he would be way off on his state projection, but wow was I wrong.
