On October 31 2012 03:03 Praetorial wrote: [quote]
Er. I'm not even conservative, but a 6% for Romney lead is kinda ominous for Obama.
A 6 percent lead for Romney gleaned from a non-representative sample of 15% of early voters who voluntarily declared their vote does not indicate anything.
You may want to reread what Gallup is saying. They asked the people in the sample whether they had voted early, whether they were going to vote early, and whether they intended to vote on election day. Romney leads by 7 in the first category, is tied with Obama in the second category, and leads Obama by 6 in the last category (which is the largest category). That points to Romney winning the popular vote by 5+ points.
Also, the Republican party released data today showing an 18+ point lead for Romney in Pennsylvania in absentee ballots (these are votes that have already been counted).
The pain train is coming for Obama.
Wasn't Romney supposed to win Penn anyways?
No. Pennsylvania is a blue state that should have been an easy Obama win. If Obama loses there, he's done.
Huh? Penn is worth like 20, Obama projected to get 290, need 270 to win. Doesn't seem to me like he needs it at all.
In the scenario that PA flips Romney, a load of other states (Ohio, Wisconsin, Minnesota) would flip Romney too.
(Out of the last 40 polls in PA Romney has been up once, so that's not gonna happen, but thats what xdaunt means)
Why? The states' electoral votes aren't tied to each other at all.
No, but PA is considered more democratic than most swing states. The situation where Obama loses PA but still wins enough other states to win the presidency is extremely unlikely. The logic is essentially is that OH votes more republican than PA, thus if Obama loses PA, he is extremely likely to lose a large portion of the other swing states that lean democratic.
He still won't lose it tho ;p. The absentee ballots are largely meaningless because unless every absentee ballot up to this point was counted, and not just some non-random part of it, the result will skew tremendously.
Is exactly what I'm thinking is happening.
Explains the 1) Extremely low number of ballots counted thus far 2) the massive skew
Edit: To me this just sounds like more Republican boasting about how they're so comfortably in the lead and going to win. It's just a campaign strategy... not something that holds any relevance.
I've ran an AB op before. It's important and telling. A strong AB showing means a strong GOTV showing. These two go hand in hand because the contact databases are linked. GOTV is what this election is going to hinge on. If someone is up by that much, it tells you a LOT about voter contact levels and election excitement. If Romney is leading by that much, there is a good chance his GOTV is going to be much stronger than Obama's in that state.
On October 31 2012 03:06 farvacola wrote: [quote] A 6 percent lead for Romney gleaned from a non-representative sample of 15% of early voters who voluntarily declared their vote does not indicate anything.
You may want to reread what Gallup is saying. They asked the people in the sample whether they had voted early, whether they were going to vote early, and whether they intended to vote on election day. Romney leads by 7 in the first category, is tied with Obama in the second category, and leads Obama by 6 in the last category (which is the largest category). That points to Romney winning the popular vote by 5+ points.
Also, the Republican party released data today showing an 18+ point lead for Romney in Pennsylvania in absentee ballots (these are votes that have already been counted).
The pain train is coming for Obama.
Wasn't Romney supposed to win Penn anyways?
No. Pennsylvania is a blue state that should have been an easy Obama win. If Obama loses there, he's done.
Huh? Penn is worth like 20, Obama projected to get 290, need 270 to win. Doesn't seem to me like he needs it at all.
In the scenario that PA flips Romney, a load of other states (Ohio, Wisconsin, Minnesota) would flip Romney too.
(Out of the last 40 polls in PA Romney has been up once, so that's not gonna happen, but thats what xdaunt means)
Why? The states' electoral votes aren't tied to each other at all.
No, but PA is considered more democratic than most swing states. The situation where Obama loses PA but still wins enough other states to win the presidency is extremely unlikely. The logic is essentially is that OH votes more republican than PA, thus if Obama loses PA, he is extremely likely to lose a large portion of the other swing states that lean democratic.
He still won't lose it tho ;p. The absentee ballots are largely meaningless because unless every absentee ballot up to this point was counted, and not just some non-random part of it, the result will skew tremendously.
Is exactly what I'm thinking is happening.
Explains the 1) Extremely low number of ballots counted thus far 2) the massive skew
Edit: To me this just sounds like more Republican boasting about how they're so comfortably in the lead and going to win. It's just a campaign strategy... not something that holds any relevance.
I've ran an AB op before. It's important and telling. A strong AB showing means a strong GOTV showing. These two go hand in hand because the contact databases are linked. GOTV is what this election is going to hinge on. If someone is up by that much, it tells you a LOT about voter contact levels and election excitement. If Romney is leading by that much, there is a good chance his GOTV is going to be much stronger than Obama's in that state.
You ignore that this is a potentially non-random sample.
Edit: Doesn't really matter, though. People are going to vote, we're going to see who wins.
Romney wins, I get more wealthy, social issues go to shit.
Obama wins, I get to feel good about myself for helping out gays and whatever.
It's almost a win-win for me. Only downside is that I'm potentially voting for the loser, which no one likes.
On October 31 2012 03:13 xDaunt wrote: Also, the Republican party released data today showing an 18+ point lead for Romney in Pennsylvania in absentee ballots (these are votes that have already been counted).
Counted? Do you have a source for that?
Here in Ohio, early and absentee votes are not counted until election day.
"I have no idea nor am I in the least bit interested."
Smacking down any attempt to tie the Romney campaign in to positively dealing with the disaster. Not even a "I'm sure if Romney were President he would be just as attentive to the disaster but obviously right now he's irrelevant to the issue of getting emergency relief" but instead just dismissing him entirely. Obviously Romney is irrelevant to the Presidential response to the disaster but still, after the Fox host tried to tie him in Christie could have gone with it.
I like Christie on this, and I generally don't like the man. Romney has an interesting opportunity to dodge the privatization of relief soundbite, while Obama is otherwise distracted.
It's not only Romney and Obama that have their jobs on the line here. One of these two will fall after Nov. 6:
1.) Gallup/Rasmussen 2.) Nate Silver/PPP
Each has gone so far out on a limb for their respective side that it seems the loser will take a crippling credibility hit.
Gallup has been around for 50 years now. If one were to place trust somewhere, it might be there. But you never know. Nate Silver is barely older than the average TL poster. PPP and Rasmussen have only been around for a decade. PPP and Silver were successful in 2008. Dems fared well and their Dem-leaning results matched that. In the present election, I think Repubs are slightly more energized than the Dems. Dem early voting is falling short of where it was in 08. This may ultimately explain the gap between the actual 2012 results and Silver/PPP's projection models that are based on a 2008 election with strong Dem enthusiasm.
Votes come from three groups: 1) your base 2) crossover 3) independents. Romney seems to have a huge independent lead (double digits in some cases). I believe he has more of an energized base since Repubs are angry and Dems are complacent. It's hard to see enough Repubs crossing over to Obama at this point. I will also add that I think early voting does not work in favor of the incumbent. Voting on election day is a process that appeals to people who want the status quo. People who vote early are angry or enthused and can't wait.
On October 31 2012 03:48 jdsowa wrote: It's not only Romney and Obama that have their jobs on the line here. One of these two will fall after Nov. 6:
1.) Gallup/Rasmussen 2.) Nate Silver/PPP
Each has gone so far out on a limb for their respective side that it seems the loser will take a crippling credibility hit.
Gallup has been around for 50 years now. If one were to place trust somewhere, it might be there. But you never know. Nate Silver is barely older than the average TL poster. PPP and Rasmussen have only been around for a decade. PPP and Silver were successful in 2008. Dems fared well and their Dem-leaning results matched that. In the present election, I think Repubs are slightly more energized than the Dems. Dem early voting is falling short of where it was in 08. This may ultimately explain the gap between the actual 2012 results and Silver/PPP's projection models that are based on a 2008 election with strong Dem enthusiasm.
Votes come from three groups: 1) your base 2) crossover 3) independents. Romney seems to have a huge independent lead (double digits in some cases). I believe he has more of an energized base since Repubs are angry and Dems are complacent. It's hard to see enough Repubs crossing over to Obama at this point. I will also add that I think early voting does not work in favor of the incumbent. Voting on election day is a process that appeals to people who want the status quo. People who vote early are angry or enthused and can't wait.
This is exactly why I don't trust the polling out there and I don't really care what Silver has to say. I just don't believe that Party ID doesn't matter as Silver and others would have us believe. Gallup and Rasmussen are both predicting a far better than republican showing this year than in 2008, which is to be expected.
On October 31 2012 03:48 jdsowa wrote: It's not only Romney and Obama that have their jobs on the line here. One of these two will fall after Nov. 6:
1.) Gallup/Rasmussen 2.) Nate Silver/PPP
Each has gone so far out on a limb for their respective side that it seems the loser will take a crippling credibility hit.
Gallup has been around for 50 years now. If one were to place trust somewhere, it might be there. But you never know. Nate Silver is barely older than the average TL poster. PPP and Rasmussen have only been around for a decade. PPP and Silver were successful in 2008. Dems fared well and their Dem-leaning results matched that. In the present election, I think Repubs are slightly more energized than the Dems. Dem early voting is falling short of where it was in 08. This may ultimately explain the gap between the actual 2012 results and Silver/PPP's projection models that are based on a 2008 election with strong Dem enthusiasm.
Votes come from three groups: 1) your base 2) crossover 3) independents. Romney seems to have a huge independent lead (double digits in some cases). I believe he has more of an energized base since Repubs are angry and Dems are complacent. It's hard to see enough Repubs crossing over to Obama at this point. I will also add that I think early voting does not work in favor of the incumbent. Voting on election day is a process that appeals to people who want the status quo. People who vote early are angry or enthused and can't wait.
A previous post has shown how off Gallup has been in recent times. Nate Silver, on the other hand, was pretty darn accurate for the 2008 election.
Nate Silver has no side. He's not a partisan hack. He's a pure statistician. Enough of this liberal bias crap.
I can't see Gallup taking too big of a hit even if their polls are off by a lot. They have decades' worth of strong polling behind them. If their polls are off this time, they'll retool them and the public will give them another chance.
Regarding Silver and other aggregate modelers, the correct way to judge them isn't who wins or loses the presidential. For example right now, 538 says Romney has a 27% chance of winning. While that makes him an underdog, that means it is something we should expect to happen 1 out of 4 times. That is like the probability of flipping 2 coins and both landing on heads. You shouldn't be SHOCKED if that happens.
A better way to judge them is to look at every state. Was he right 3 out of 5 times for states with around 60% confidence? 3 out of 4 times for states with around 75% confidence? etc etc. Do that for the electoral college and for the senate races. (alternatively, add all of the confidence estimates together, and that should equal the total number of races he got right if the confidence was correct)
However I do think that people will look at the model and say Silver either had a crystal ball or is a complete fraud, since most people have little understanding of what probability means, so his fate as an influential political writer may come down to being right or not.
On October 31 2012 03:56 Signet wrote: I can't see Gallup taking too big of a hit even if their polls are off by a lot. They have decades' worth of strong polling behind them. If their polls are off this time, they'll retool them and the public will give them another chance.
Regarding Silver and other aggregate modelers, the correct way to judge them isn't who wins or loses the presidential. For example right now, 538 says Romney has a 27% chance of winning. While that makes him an underdog, that means it is something we should expect to happen 1 out of 4 times. That is like the probability of flipping 2 coins and both landing on heads. You shouldn't be SHOCKED if that happens.
A better way to judge them is to look at every state. Was he right 3 out of 5 times for states with around 60% confidence? 3 out of 4 times for states with around 75% confidence? etc etc. Do that for the electoral college and for the senate races. (alternatively, add all of the confidence estimates together, and that should equal the total number of races he got right if the confidence was correct)
However I do think that people will look at the model and say Silver either had a crystal ball or is a complete fraud, since most people have little understanding of what probability means, so his fate as an influential political writer may come down to being right or not.
If it turns out that he's wrong about the Party ID thing, I'll have no problem putting him in the fraud category.
On October 31 2012 03:48 jdsowa wrote: It's not only Romney and Obama that have their jobs on the line here. One of these two will fall after Nov. 6:
1.) Gallup/Rasmussen 2.) Nate Silver/PPP
Each has gone so far out on a limb for their respective side that it seems the loser will take a crippling credibility hit.
Gallup has been around for 50 years now. If one were to place trust somewhere, it might be there. But you never know. Nate Silver is barely older than the average TL poster. PPP and Rasmussen have only been around for a decade. PPP and Silver were successful in 2008. Dems fared well and their Dem-leaning results matched that. In the present election, I think Repubs are slightly more energized than the Dems. Dem early voting is falling short of where it was in 08. This may ultimately explain the gap between the actual 2012 results and Silver/PPP's projection models that are based on a 2008 election with strong Dem enthusiasm.
Votes come from three groups: 1) your base 2) crossover 3) independents. Romney seems to have a huge independent lead (double digits in some cases). I believe he has more of an energized base since Repubs are angry and Dems are complacent. It's hard to see enough Repubs crossing over to Obama at this point. I will also add that I think early voting does not work in favor of the incumbent. Voting on election day is a process that appeals to people who want the status quo. People who vote early are angry or enthused and can't wait.
A previous post has shown how off Gallup has been in recent times. Nate Silver, on the other hand, was pretty darn accurate for the 2008 election.
Nate Silver has no side. He's not a partisan hack. He's a pure statistician. Enough of this liberal bias crap.
Of course he has a side. Nobody is a pure anything unless they're completely impartial about the results of the US election. Which I highly doubt that he is.
All I mean to say is that if Obama loses this election in spite of Silver's 75% probability, it will be because his statistical models were based on a 2008 election that had an energized Dem base (which is the opposite of the present situation). People are quick to dismiss Rasmussen and Gallup for certain past failures, but forget that they place their faith in someone whose only success is to have called one election pretty close when he was in his 20s.
On October 31 2012 03:48 jdsowa wrote: It's not only Romney and Obama that have their jobs on the line here. One of these two will fall after Nov. 6:
1.) Gallup/Rasmussen 2.) Nate Silver/PPP
Each has gone so far out on a limb for their respective side that it seems the loser will take a crippling credibility hit.
Gallup has been around for 50 years now. If one were to place trust somewhere, it might be there. But you never know. Nate Silver is barely older than the average TL poster. PPP and Rasmussen have only been around for a decade. PPP and Silver were successful in 2008. Dems fared well and their Dem-leaning results matched that. In the present election, I think Repubs are slightly more energized than the Dems. Dem early voting is falling short of where it was in 08. This may ultimately explain the gap between the actual 2012 results and Silver/PPP's projection models that are based on a 2008 election with strong Dem enthusiasm.
Votes come from three groups: 1) your base 2) crossover 3) independents. Romney seems to have a huge independent lead (double digits in some cases). I believe he has more of an energized base since Repubs are angry and Dems are complacent. It's hard to see enough Repubs crossing over to Obama at this point. I will also add that I think early voting does not work in favor of the incumbent. Voting on election day is a process that appeals to people who want the status quo. People who vote early are angry or enthused and can't wait.
A previous post has shown how off Gallup has been in recent times. Nate Silver, on the other hand, was pretty darn accurate for the 2008 election.
Nate Silver has no side. He's not a partisan hack. He's a pure statistician. Enough of this liberal bias crap.
Of course he has a side. Nobody is a pure anything unless they're completely impartial about the results of the US election. Which I highly doubt that he is.
All I mean to say is that if Obama loses this election in spite of Silver's 75% probability, it will be because his statistical models were based on a 2008 election that had an energized Dem base (which is the opposite of the present situation). People are quick to dismiss Rasmussen and Gallup for certain past failures, but forget that they place their faith in someone whose only success is to have called one election pretty close when he was in his 20s.
He has a side but that does not mean it's portrayed in his statistics. There are people out there who have integrity, y'know.
And really, what does age have anything to do with it? First the racism and now the ageism. oboy. Nate Silver wasn't just some random 20-something-year-old who said, "I think the Democrats will win!" He constructed a statistical model that was able to predict the Electoral College within five votes (and he understated it!).
Anyway, yes, we will see how much credibility Nate Silver has after this election. Just remember, as Signet said, he never guessed that Obama has a 100% chance of winning. The biggest factor we should review after the election is his methodology and see how accurate the predictions it produced were across the board.
On October 31 2012 03:56 Signet wrote: I can't see Gallup taking too big of a hit even if their polls are off by a lot. They have decades' worth of strong polling behind them. If their polls are off this time, they'll retool them and the public will give them another chance.
Regarding Silver and other aggregate modelers, the correct way to judge them isn't who wins or loses the presidential. For example right now, 538 says Romney has a 27% chance of winning. While that makes him an underdog, that means it is something we should expect to happen 1 out of 4 times. That is like the probability of flipping 2 coins and both landing on heads. You shouldn't be SHOCKED if that happens.
A better way to judge them is to look at every state. Was he right 3 out of 5 times for states with around 60% confidence? 3 out of 4 times for states with around 75% confidence? etc etc. Do that for the electoral college and for the senate races. (alternatively, add all of the confidence estimates together, and that should equal the total number of races he got right if the confidence was correct)
However I do think that people will look at the model and say Silver either had a crystal ball or is a complete fraud, since most people have little understanding of what probability means, so his fate as an influential political writer may come down to being right or not.
If it turns out that he's wrong about the Party ID thing, I'll have no problem putting him in the fraud category.
You can do whatever you want, but even right-leaning statisticians have described why Party ID isn't a reliable variable.
On October 31 2012 03:56 Signet wrote: I can't see Gallup taking too big of a hit even if their polls are off by a lot. They have decades' worth of strong polling behind them. If their polls are off this time, they'll retool them and the public will give them another chance.
Regarding Silver and other aggregate modelers, the correct way to judge them isn't who wins or loses the presidential. For example right now, 538 says Romney has a 27% chance of winning. While that makes him an underdog, that means it is something we should expect to happen 1 out of 4 times. That is like the probability of flipping 2 coins and both landing on heads. You shouldn't be SHOCKED if that happens.
A better way to judge them is to look at every state. Was he right 3 out of 5 times for states with around 60% confidence? 3 out of 4 times for states with around 75% confidence? etc etc. Do that for the electoral college and for the senate races. (alternatively, add all of the confidence estimates together, and that should equal the total number of races he got right if the confidence was correct)
However I do think that people will look at the model and say Silver either had a crystal ball or is a complete fraud, since most people have little understanding of what probability means, so his fate as an influential political writer may come down to being right or not.
Well, it's true that his model does not claim to actually predict, but merely states "the probability of something happening based on my model". The problem is--what if that model is worthless because it was constructed on the basis of one or two recent elections? What is it ignoring? What is it giving too much credence to?
On October 31 2012 03:56 Signet wrote: I can't see Gallup taking too big of a hit even if their polls are off by a lot. They have decades' worth of strong polling behind them. If their polls are off this time, they'll retool them and the public will give them another chance.
Regarding Silver and other aggregate modelers, the correct way to judge them isn't who wins or loses the presidential. For example right now, 538 says Romney has a 27% chance of winning. While that makes him an underdog, that means it is something we should expect to happen 1 out of 4 times. That is like the probability of flipping 2 coins and both landing on heads. You shouldn't be SHOCKED if that happens.
A better way to judge them is to look at every state. Was he right 3 out of 5 times for states with around 60% confidence? 3 out of 4 times for states with around 75% confidence? etc etc. Do that for the electoral college and for the senate races. (alternatively, add all of the confidence estimates together, and that should equal the total number of races he got right if the confidence was correct)
However I do think that people will look at the model and say Silver either had a crystal ball or is a complete fraud, since most people have little understanding of what probability means, so his fate as an influential political writer may come down to being right or not.
Well, it's true that his model does not claim to actually predict, but merely states "the probability of something happening based on my model". The problem is--what if that model is worthless because it was constructed on the basis of one or two recent elections? What is it ignoring? What is it giving too much credence to?
His model isn't based on one or two recent elections, though...
On October 31 2012 03:56 Signet wrote: I can't see Gallup taking too big of a hit even if their polls are off by a lot. They have decades' worth of strong polling behind them. If their polls are off this time, they'll retool them and the public will give them another chance.
Regarding Silver and other aggregate modelers, the correct way to judge them isn't who wins or loses the presidential. For example right now, 538 says Romney has a 27% chance of winning. While that makes him an underdog, that means it is something we should expect to happen 1 out of 4 times. That is like the probability of flipping 2 coins and both landing on heads. You shouldn't be SHOCKED if that happens.
A better way to judge them is to look at every state. Was he right 3 out of 5 times for states with around 60% confidence? 3 out of 4 times for states with around 75% confidence? etc etc. Do that for the electoral college and for the senate races. (alternatively, add all of the confidence estimates together, and that should equal the total number of races he got right if the confidence was correct)
However I do think that people will look at the model and say Silver either had a crystal ball or is a complete fraud, since most people have little understanding of what probability means, so his fate as an influential political writer may come down to being right or not.
Well, it's true that his model does not claim to actually predict, but merely states "the probability of something happening based on my model". The problem is--what if that model is worthless because it was constructed on the basis of one or two recent elections? What is it ignoring? What is it giving too much credence to?
What would Nate Silver's model being constructed "on the basis of one or two recent elections?" look like? You keep saying this in vaguely hypothetical terms; point at some aspect of Silver's method (it is rather transparent) and qualify what you are saying.
On October 31 2012 03:56 Signet wrote: I can't see Gallup taking too big of a hit even if their polls are off by a lot. They have decades' worth of strong polling behind them. If their polls are off this time, they'll retool them and the public will give them another chance.
Regarding Silver and other aggregate modelers, the correct way to judge them isn't who wins or loses the presidential. For example right now, 538 says Romney has a 27% chance of winning. While that makes him an underdog, that means it is something we should expect to happen 1 out of 4 times. That is like the probability of flipping 2 coins and both landing on heads. You shouldn't be SHOCKED if that happens.
A better way to judge them is to look at every state. Was he right 3 out of 5 times for states with around 60% confidence? 3 out of 4 times for states with around 75% confidence? etc etc. Do that for the electoral college and for the senate races. (alternatively, add all of the confidence estimates together, and that should equal the total number of races he got right if the confidence was correct)
However I do think that people will look at the model and say Silver either had a crystal ball or is a complete fraud, since most people have little understanding of what probability means, so his fate as an influential political writer may come down to being right or not.
Well, it's true that his model does not claim to actually predict, but merely states "the probability of something happening based on my model". The problem is--what if that model is worthless because it was constructed on the basis of one or two recent elections? What is it ignoring? What is it giving too much credence to?
And what if he just has a much better understanding of statistics, probability and elections than you do?
EPT
