By using this site, you agree to our Privacy Policy and our Terms of Use. Close

Forums - Politics Discussion - Biased polls? Yes...

Wow, Gallup might have been one of the worst losers on election night. And Google one of the biggest winners.



Around the Network
theprof00 said:
GameOver22 said:
Taking polls from the last 21 days is a bit strange. I would have used them from the last week, where a lot of polls were picking up a late Obama surge.


>arguing with Nate Silver about polling data.

Actually he's argueing with him about methodology... not the data.

Nate Silver basically says he used 21 days because some people think pollsters cook the numbers at the end of an election.

While gameover is suggesting that even if such a thing is true, to use 21 days would be a flawed method because there had been actual changes that would outlook.  Like the hurricane.

Gameover actually has a valid point, which is why if you actually read the Nate Silver piece you'll note he doesn't actually argue for that point in anyway.  Just saying it's a common measure... and agrees that Gameover has a point.

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure."

Put it at 1.5... and the numbers change quite a bit.

So i mean... if your going to argue... argue with some substance.   Rather then relying on appeal to authority of something you haven't read.

It's actually an interesting article, so it's worth a read rather then just assuming it's right without knowing the content.  It talks about methodology problems involving things like cellphones effecting certain kinds of polling.



theprof00 said:
GameOver22 said:
Taking polls from the last 21 days is a bit strange. I would have used them from the last week, where a lot of polls were picking up a late Obama surge.


>arguing with Nate Silver about polling data.

Um....yes because the polls over the final week show Obama picking up momentum and winning. It would just be more accurate to use polls over the final week...not the final three weeks. A lot of stuff happens over the last week, like the undecided voters making up their minds, and this was showing movement in Obama's favor. Granted, there does seem to be a Republican bias to the polls, but its nowhere near as drastic as the article makes it seem. Just for instance, he has Gallup at +7.2 and Rasmussen at +3.7 for Romney when both of their final polls show Romney winning by +1.

RCP-You can see the trends pretty clearly.

Edit: Kaz beat me to it. : )



Kasz216 said:
theprof00 said:
GameOver22 said:
Taking polls from the last 21 days is a bit strange. I would have used them from the last week, where a lot of polls were picking up a late Obama surge.


>arguing with Nate Silver about polling data.

Actually he's argueing with him about methodology... not the data.

Nate Silver basically says he used 21 days because some people think pollsters cook the numbers at the end of an election.

While gameover is suggesting that even if such a thing is true, to use 21 days would be a flawed method because there had been actual changes that would outlook.  Like the hurricane.

Gameover actually has a valid point, which is why if you actually read the Nate Silver piece you'll note he doesn't actually argue for that point in anyway.  Just saying it's a common measure... and agrees that Gameover has a point.

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure."

Put it at 1.5... and the numbers change quite a bit.

So i mean... if your going to argue... argue with some substance.   Rather then relying on appeal to authority of something you haven't read.

It's actually an interesting article, so it's worth a read rather then just assuming it's right without knowing the content.  It talks about methodology problems involving things like cellphones effecting certain kinds of polling.

Except gameover said to take the poll data from the last few days and Nate says polls didn't keep up with the last few days.



theprof00 said:
Kasz216 said:
theprof00 said:
GameOver22 said:
Taking polls from the last 21 days is a bit strange. I would have used them from the last week, where a lot of polls were picking up a late Obama surge.


>arguing with Nate Silver about polling data.

Actually he's argueing with him about methodology... not the data.

Nate Silver basically says he used 21 days because some people think pollsters cook the numbers at the end of an election.

While gameover is suggesting that even if such a thing is true, to use 21 days would be a flawed method because there had been actual changes that would outlook.  Like the hurricane.

Gameover actually has a valid point, which is why if you actually read the Nate Silver piece you'll note he doesn't actually argue for that point in anyway.  Just saying it's a common measure... and agrees that Gameover has a point.

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure."

Put it at 1.5... and the numbers change quite a bit.

So i mean... if your going to argue... argue with some substance.   Rather then relying on appeal to authority of something you haven't read.

It's actually an interesting article, so it's worth a read rather then just assuming it's right without knowing the content.  It talks about methodology problems involving things like cellphones effecting certain kinds of polling.

Except gameover said to take the poll data from the last few days and Nate says polls didn't keep up with the last few days.

I said the last week, and there most definitely is data from the last week.



Around the Network

Doesn't really matter. Fox told all it's viewers not to believe the polls anyway.



GameOver22 said:
theprof00 said:
Kasz216 said:
theprof00 said:
GameOver22 said:
Taking polls from the last 21 days is a bit strange. I would have used them from the last week, where a lot of polls were picking up a late Obama surge.


>arguing with Nate Silver about polling data.

Actually he's argueing with him about methodology... not the data.

Nate Silver basically says he used 21 days because some people think pollsters cook the numbers at the end of an election.

While gameover is suggesting that even if such a thing is true, to use 21 days would be a flawed method because there had been actual changes that would outlook.  Like the hurricane.

Gameover actually has a valid point, which is why if you actually read the Nate Silver piece you'll note he doesn't actually argue for that point in anyway.  Just saying it's a common measure... and agrees that Gameover has a point.

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure."

Put it at 1.5... and the numbers change quite a bit.

So i mean... if your going to argue... argue with some substance.   Rather then relying on appeal to authority of something you haven't read.

It's actually an interesting article, so it's worth a read rather then just assuming it's right without knowing the content.  It talks about methodology problems involving things like cellphones effecting certain kinds of polling.

Except gameover said to take the poll data from the last few days and Nate says polls didn't keep up with the last few days.

I said the last week, and there most definitely is data from the last week.

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure.

Nonetheless, polls conducted over the final three weeks of the campaign had a two-point Republican bias overall, probably more than can be explained by the late shift alone. In addition, likely voter polls were slightly more Republican-leaning than the actual results in many races in 2010."





theprof00 said:

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure.

Nonetheless, polls conducted over the final three weeks of the campaign had a two-point Republican bias overall, probably more than can be explained by the late shift alone. In addition, likely voter polls were slightly more Republican-leaning than the actual results in many races in 2010."



Key word....some. There are plenty of polls conducted over the final week and the final days. Just look at the data. RealClearPolitics

Edit: To get an even better idea of all the polls conducted over the final days Polltracker



GameOver22 said:
theprof00 said:

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure.

Nonetheless, polls conducted over the final three weeks of the campaign had a two-point Republican bias overall, probably more than can be explained by the late shift alone. In addition, likely voter polls were slightly more Republican-leaning than the actual results in many races in 2010."



Key word....some. There are plenty of polls conducted over the final week and the final days. Just look at the data. RealClearPolitics

Edit: To get an even better idea of all the polls conducted over the final days Polltracker


Yep... it's funny.  People like to repeat that saying their are three kinds of lies, lies, damned lies and statistics...

the truth is, statistics don't like.  It's just most people don't understand the statistics and don't really understand what's being said.

 

It would be really worth it to replace a lot of Alegebra and Geometry in jr highschools and highschool with math logic and statistics classes.  Much more applicable day to day.



Kasz216 said:
GameOver22 said:
theprof00 said:

"Some of the overall Republican bias in the polls this year may reflect the fact that Mr. Obama made gains in the closing days of the campaign, for reasons such as Hurricane Sandy, and that this occurred too late to be captured by some polls. In the FiveThirtyEight “now-cast,” Mr. Obama went from being 1.5 percentage points ahead in the popular vote on Oct. 25 to 2.5 percentage points ahead by Election Day itself, close to his actual figure.

Nonetheless, polls conducted over the final three weeks of the campaign had a two-point Republican bias overall, probably more than can be explained by the late shift alone. In addition, likely voter polls were slightly more Republican-leaning than the actual results in many races in 2010."



Key word....some. There are plenty of polls conducted over the final week and the final days. Just look at the data. RealClearPolitics

Edit: To get an even better idea of all the polls conducted over the final days Polltracker


Yep... it's funny.  People like to repeat that saying their are three kinds of lies, lies, damned lies and statistics...

the truth is, statistics don't like.  It's just most people don't understand the statistics and don't really understand what's being said.

 

It would be really worth it to replace a lot of Alegebra and Geometry in jr highschools and highschool with math logic and statistics classes.  Much more applicable day to day.

Couldn't agree more. As long as you know what the stats represent, the stats can't lie. The problem is, the meaning can sometimes be misleading and counterintuitive.

One example that I've been reading up on lately and that's relevant to politics is what people mean when they identify as conservative or liberal. After the election, I remember Ari Fleischer talking about how the conservative ideology dominates liberal ideology, so the country is still a center-right country. The problem is, it seems people don't have politics in mind when they identify with conservatism. Conservatism actually loads more onto a religious/social/familial dimension while liberalism is associated moreso with the counterculture of the 1970s, race riots, environmentalism, welfare exploitation, etc. This is the reason you tend to see Democrats with a party ID advantage even though liberalism gets trounced 2:1 by conservatism. Long story short, there's a big difference between someone identifying as a republican and a conservative or vice versa.