2012 Polling Accuracy

[RabidMongoose]

Originally Posted by buckshot24

I was hoping that this post didn't really exist.

It does. But you as an apparent conservative may have a different reality than the rest of the world. Minorities exist. You will need to interact with them. 90% of Romney's constituency was white non-hispanic. Apparently even their pollsters reflected that same constituency.

[jackstar7]

What motivates someone to defend a pollster (any of them)?

[Athena]

Originally Posted by buckshot24

The only poll I was basing anything on was their party affiliation poll which was pretty accurate going back 10 years in predicting voter turnout.

As I understand it, Rassmussen has been off in the past three elections.

I really think that anyone who is interested in this stuff should go back and read Nate Silver's columns -- at least going back to the spring. Unlike the talking heads on TV, he did a thorough job of explaining his conclusions.

Mathematics is not a "belief system". One can understand mathematics and still believe in God -- anyone here ever read anything by Bertrand Russell?

[randomrogue]

It's tough to scientifically identify why the Republicans are "in their own bubble world" but that's precisely the reason that their polls were wrong. Rather than use science they used "feelings" to identify voter turnout and groups. There was never any basis in reality for what they thought was going to happen. They should have stuck to reality and not made up numbers. Even after they lost they still struggle with the real numbers.

[shira]

Originally Posted by randomrogue

It's tough to scientifically identify why the Republicans are "in their own bubble world" but that's precisely the reason that their polls were wrong. Rather than use science they used "feelings" to identify voter turnout and groups. There was never any basis in reality for what they thought was going to happen. They should have stuck to reality and not made up numbers. Even after they lost they still struggle with the real numbers.

The right uses the same approach to climate change. When confronted by an overwhelming consensus of scientific data that demonstrates that greenhouse gases released as a consequence of human activity are a significant factor in the observed increase in global temperatures, the right cherry-picks the conclusions of the tiny minority of climatologists that disagree.

Just as they ignored the consensus polling data showing that Obama would be an easy winner, the right "know" their right-wing positions are correct because they can always find "fringe data" to reinforce their beliefs.

[techs]

Is it that people don't understand that pollsters ask questions and study previous patterns to determine turnout? They don't only rely on the the last election. The question of how many and which people are actually going to vote is nearly as important to pollsters as the polling on who people say they are supporting.

The fact that its being treated as simply as, "they relied on the turnout" from 2008 is absurd."
Not only do they apply statistical models and questions from 2008 but they have a whole new set of polling data from 2012 that they use to compare and determine how well their predictions worked in 2008.

What I saw was that once the polls continued to show a persistent Democratic advantage the Republicans pushed for some of the polls to use different data. And like a self fulfilling prophecy the Republicans took their biased polls and convinced themselves they were the correct polls.

[Jhhnn]

Originally Posted by randomrogue

It's tough to scientifically identify why the Republicans are "in their own bubble world" but that's precisely the reason that their polls were wrong. Rather than use science they used "feelings" to identify voter turnout and groups. There was never any basis in reality for what they thought was going to happen. They should have stuck to reality and not made up numbers. Even after they lost they still struggle with the real numbers.

It's not tough to identify why. Sam Wang references a known phenomenon, "motivated reasoning"-

http://election.princeton.edu/2012/1...bug/#more-8830

http://en.wikipedia.org/wiki/Motivated_reasoning

I just call it living in denial, believing what you want to believe. Our own resident righties exemplify that on a daily basis.

[piasabird]

Maybe a lot of voting machines were rigged?

[Charles Kozierok]

Originally Posted by randomrogue

It's tough to scientifically identify why the Republicans are "in their own bubble world" but that's precisely the reason that their polls were wrong.

Epistemic closure.

[buckshot24]

Originally Posted by RabidMongoose

It does. But you as an apparent conservative may have a different reality than the rest of the world. Minorities exist. You will need to interact with them. 90% of Romney's constituency was white non-hispanic. Apparently even their pollsters reflected that same constituency.

They used 2008 as a model of minority turnout which ended up being wrong. Definitely isn't any kind of conspiracy involved.

I'm perfectly fine with minorities, I love minorities, my daughter is a minority!

[buckshot24]

Originally Posted by Athena

As I understand it, Rassmussen has been off in the past three elections.

I really think that anyone who is interested in this stuff should go back and read Nate Silver's columns -- at least going back to the spring. Unlike the talking heads on TV, he did a thorough job of explaining his conclusions.

Mathematics is not a "belief system". One can understand mathematics and still believe in God -- anyone here ever read anything by Bertrand Russell?

I know all about Silver's treatment on Rasmussen and the prior election (2010 basically).

The poll I have been talking about is the party affiliation poll they run. Up until last tuesday it had pretty good accuracy in predicting turnout. It was horribly off this year.

Oh and I'm not defending Rasmussen. He was off this year on his national poll and his partisan affiliation poll. All I was saying is that they underestimated turnout among young voters and minorities by basing it on percentages of turnout in 2008.

[lothar]

This is the only poll that mattered.
All others are irrelevant.

[RabidMongoose]

Originally Posted by buckshot24

They used 2008 as a model of minority turnout which ended up being wrong. Definitely isn't any kind of conspiracy involved.

I'm perfectly fine with minorities, I love minorities, my daughter is a minority!

They used racist models.

[sactoking]

Not presidential, but we had a Congressional race here that pitted out outgoing State Senate Majority Leader against a Republican candidate from a well-known family. The district was gerrymandered to be pretty heavily D, but the D candidate was somewhat disliked as he is just a jerk, so when the polls had the R up a few points it was only a mild surprise. As the election season went on the R candidate increased his lead in the polls 'til pretty much every poll had him up by 6-8%.

On Tuesday the D candidate won by 8%; that was a 14-16% swing from every poll.

[Athena]

Originally Posted by buckshot24

All I was saying is that they underestimated turnout among young voters and minorities by basing it on percentages of turnout in 2008.

And many would say that the decision to base the entire campaign on an expectation of lower than 2008 minority turnout, was in fact based on racist assumptions. Those assumptions should have at least been cross-checked with 2010 census data. They didn't do that because... well, they didn't think the minority vote mattered enough to warrant a closer look.

Then again, perhaps looking at the data was what mobilized all the voter suppression efforts in the so-called swing states.

On average, Romney campaign staffers were paid twice what Obama's were and I think we can all agree that Obama had demonstrably better talent. The people who developed the differing turnout models all went to the same schools and had access to the very same data. For some reason, the Romney folks just didn't see what the Obama team did.

Obviously, it's not all about racism, some of it is just hubris. Some of us suspect though, that the Republicans' two-decade long policy of thumbing their noses at black voters had something to do with that blind spot.

[Charles Kozierok]

Originally Posted by buckshot24

They used 2008 as a model of minority turnout which ended up being wrong. Definitely isn't any kind of conspiracy involved.

Remember when you tried to mock me by saying I thought I was smarter than Gallup and Rasmussen?

[Fern]

Seems premature to come to any firm conclusion about the polls' accuracy. They haven't even counted all the votes yet. And I'm guessing we don't have firm data on turnout yet either (D's versus R's).

http://nbcpolitics.nbcnews.com/_news...n-to-tell?lite

Fern

[RabidMongoose]

Originally Posted by Fern

Seems premature to come to any firm conclusion about the polls' accuracy. They haven't even counted all the votes yet. And I'm guessing we don't have firm data on turnout yet either (D's versus R's).

http://nbcpolitics.nbcnews.com/_news...n-to-tell?lite

Fern

Yeah, it's definitely too early. Maybe Romney did get more votes.

[zinfamous]

Originally Posted by buckshot24

I was reading Rasmussen's take on why he was off and it was turnout of the young and minority voters. He was off by a few percentage points on both which underestimated Democratic turnout. As long as you get your demographics right it shouldn't matter how you contact them.

can't you only make such estimates by contacting these people? that is the exact reason for broadening the method of contact, which, again, seems to be why such polling has trended towards inaccuracy.

[buckshot24]

Originally Posted by RabidMongoose

They used racist models.

Quit throwing around "racist" for every fucking thing.

[buckshot24]

Originally Posted by Athena

And many would say that the decision to base the entire campaign on an expectation of lower than 2008 minority turnout, was in fact based on racist assumptions. Those assumptions should have at least been cross-checked with 2010 census data. They didn't do that because... well, they didn't think the minority vote mattered enough to warrant a closer look.

Then again, perhaps looking at the data was what mobilized all the voter suppression efforts in the so-called swing states.

On average, Romney campaign staffers were paid twice what Obama's were and I think we can all agree that Obama had demonstrably better talent. The people who developed the differing turnout models all went to the same schools and had access to the very same data. For some reason, the Romney folks just didn't see what the Obama team did.

Obviously, it's not all about racism, some of it is just hubris. Some of us suspect though, that the Republicans' two-decade long policy of thumbing their noses at black voters had something to do with that blind spot.

We're talking about the polling outfits, not the campaigns.

[RabidMongoose]

Originally Posted by buckshot24

Quit throwing around "racist" for every fucking thing.

The models were racist because they took GOP ideals of the non-existence of minorities and actually applied them to polling. It's so racist.

The GOP's electorate is 90% white non-hispanic. Their models reflect this, too.

[buckshot24]

Originally Posted by RabidMongoose

The models were racist because they took GOP ideals of the non-existence of minorities and actually applied them to polling. It's so racist.

Now I know you're trolling. Ok, thanks for making it clear.

[Athena]

Originally Posted by buckshot24

We're talking about the polling outfits, not the campaigns.

Well in that case, it turns out that all the polls overstated support for Romney (I'm sure that we'll read all sorts of explanations for that in the months to come) but in aggregate they still predicted an Obama win. That's the point that most people here are talking about...why/how the campaign's model came up with different forecasts than the statisticians using the very same data.

The Romney campaign took data that was already skewed in his favor and adjusted it because, in their minds, it wasn't favorable enough.