They adjust their numbers based on demographics. They have to, because they would otherwise undersample the young and minorities and oversample the elderly.
In other words, they’re saying “Pollsters expect x% of blacks to vote and y% of Hispanics, but really it will be x-3 and y-7.”
Do pollsters really weight polls this way?
That’s right. Many do weight polls that way, but not all.
The thing is, this kind of weighing works. In aggregate, they get it right, even when the election is exceptional. If anything was going to test that, it would be 2008. In theory, having the first black man on the presidential ballot should have made the demographic margins really hard to predict. But, as it turned out, the aggregate did just fine after all.
They do. They don’t just call 3000 people and whoever answers that’s who they go with. They weight the polls in various ways to make the sample representative of America(or a particular state). that involves making some assumptions. On one hand, it makes the polling more scientific, but on the other it can mean the pollsters can get it wrong.
No one’s made money betting against the polls, but that does not mean they are always right. Most of them are working on the assumption that 2008 will see growth in minority turnout comparable to the last few Presidential elections. Problem is, 2008 saw a big jump in minority turnout and a big fall in white turnout, so it’s more likely that the electorate in 2012 will look more like the 2004 electorate than the 2008 electorate. That’s the assumption Gallup is working with, which is why they’ve been showing better results for Romney than even Rasmussen most of the time.
What I want to know is how Nate Silver readjusts his system if a minority of pollsters got it right while the majority got it wrong. Is there even a way to account for that? He works from an assumption that an average of polls will probably get it right, but if the majority of pollsters are doing it wrong, then that doesn’t work.
This is inaccurate. The jump from '04-'08 is exactly the same as the jump from '00-'04.
ETA: To be more precise, the increase in the actual absolute turnout level (% of the electorate comprised by each demographic) was perfectly steady. The turnout rates (% of eligible voters who vote) accelerated slightly, but this did not alter the overall increase in proportions.
According to Pew’s numbers, I’m right:
The white share of the vote fell 1.5% from 2000 to 2004, but a whopping 2.9% from 2004 to 2008. A lot of that was whites staying home in 2008. If the assumption is that whites will turn out at the same rate as they did in 2008, then that assumption could be wrong.
The exit polls had it as a 3.3 drop election-over-election for both cycles. Not sure how Pew is measuring it if not by exit polls.
Reading the full report, it looks like it’s based on their own polling data. Might be more reliable than exit polls, I don’t know.
Speaking as an American, I can think of no reason why Canada should want to be part of the USA. They are pretty much a European style country, while American is sliding rapidly into Third World-hood.
Define ‘right’ and ‘wrong.’
It seems like a lot of people are saying calling the winner correctly is the measure - it’s not. Getting it “right,” for a pollster, is the vote percentage being within the margin of error of your poll. If a pollster says Romney is ahead by 1% with a 3% margin of error, but Obama wins by 1%, the pollster’s poll was correct.
If Nate Silver thinks a pollster is poor, he drives them to obscurity or out of business (Research 2000, Strategic Vision, Zogby).
Individual polls are “right” if they get within the margin of error, but Nate Silver assumes that an average of polls weighted by reliability gives you a good idea of who is going to win. If that average is wrong, then Silver’s system fails. He has to readjust for 2016.
Nate’s model is still good if the state predictions are close to actuality. If he predicts a state will be close then it’s not a bad model if it goes the opposite direction of his guess.
It would be a bad model if he predicted Virginia or Colorado were going to Romney with 95% certainty and they went to Obama. It would be a bad model if he predicted Illinois would be close and it went 63/35.
For this discussion to make sense, I need to know what’s ‘right’ or ‘wrong’ for the various pollsters that you talk about - a minority get it right, and a majority get it wrong.
He does list his margin of error for state polls, and I assume that’s based on taking all those polls in aggregate. For example, his projection for Ohio is 51.1-47.9 Obama, with a 2.9 percentage point margin of error.
If the majority of pollsters are doing it wrong, then Nate will analyze exactly what was wrong with the methodology of those polls, and account for the possibility of that type of error in subsequent predictions. He has said that much of the current percentage he is allowing for the results to be different from those predicted by the polls comes from the possibility of polling errors. If results from this election show that polls are less accurate than his current model assumes, he will adjust the model for that. That is why he is not making yes/no predictions, but instead providing percentages.
That’s the 16% chance Romney has of winning. He said a couple days ago that that number represents the chance the polls are statistically biased.
Newt Gingrich says Romney has lost.
Errant Email to Newt Gingrich Supporters Says ‘Obama Is Going to Win’
Today’s story of Romney’s internal polls showing him leading all over the swing states made me do a little nostalgia diving:
Time’s “Swampland” blog, Nov 2, 2008:
Letter by McCain pollster Bill McInturff on National Review Online, Oct 28, 2008:
Stop me if any of this sounds familiar…
You’re burying the lead. Apparently Obama will steal an unconstitutional third term and stay in office until 2020. Why he’s going to stop then is unclear.
Because FEMA only has eight years of body bags.
One of the campaigns is still polling PA, I can report tonight. I got a poll call from a pollster using a phone number that also did polling of CO on July 29, and the only questions were about the presidential race. No public poll had calls in the field on that date, so I guess it’s a campaign.
Meanwhile, Obama is in Wisconsin and Bill Clinton is in Pennsylvania; states Obama won comfortably in 2008. Last year at this time guess where Obama was? Indiana. Makes one wonder, doesn’t it?