So who's going to win the election? (POLL)

[Lathum]

It is amazing to me anyone thinks Romney has a shot in hell of winning.

[JPhillips]

Originally Posted by Leroy Veritas

Most polling is terribly inaccurate as it oversamples democrats. I think Obama wins pretty handily

Huh?

Outside of the Rasmussen poll, and this Silver guy, there's not much in the polling realm that is worthwhile.

HUH?

[stevew]

Originally Posted by molson

But doesn't the margin of error cover those few points and those several hundred thousand votes? Sure, if we knew for a fact that if the election was held today, one guy would win by a few points and a few hundred thousand votes, things ain't gonna change much by Tuesday, but we don't know that, all we have is polling.

I kind of think that for there even to be a chance for a dude to win, he should probably lead a poll at some point, or at least have it show a dead heat. Not just be within a margin of error.

[molson]

Originally Posted by larrymcg421

Margin of error covers that in one poll, but is less likely a problem when you have several polls. For example, Nate Silver points out that the Ohio polls in the last week combined make for 1.5 pt. margin of error.

If the poll has you up by 2 points 4 or 5 days before the election and there's a 1.5 margin of error I don't think it's impossible the other guy wins, even if there's hundreds of thousands of people in those points. They're polls, not real time voter numbers, you don't necessarily need hundreds of thousands of people to change their mind or not show up, that's all I'm saying. Unless Nate Silver refutes that. But I have to believe even he was off by 2 points in at least one state race in 2008.

[Lathum]

I'm giving serious thought to betting everything I have in my sportsbook acount on Obama at -300

[Apathetic Lurker]

Originally Posted by bronconick

They want Obama gone so badly etc...

It's amazing how close this year feels like 2004 with everything flipped.

I think this race feels closer to 1860 than anything......

[Raiders Army]

Originally Posted by Apathetic Lurker

I think this race feels closer to 1860 than anything......

Wha

[larrymcg421]

Originally Posted by molson

If the poll has you up by 2 points 4 or 5 days before the election and there's a 1.5 margin of error I don't think it's impossible the other guy wins, even if there's hundreds of thousands of people in those points. They're polls, not real time voter numbers, you don't necessarily need hundreds of thousands of people to change their mind or not show up, that's all I'm saying. Unless Nate Silver refutes that. But I have to believe even he was off by 2 points in at least one state race in 2008.

Well it's important to note that the MOE doesn't cover 100% of the situations. It's based on a 95% confidence interval. So when you're up outside the margin of error, there are two ways you could still lose: 1) One of the 5% situations happened or 2) The polls could be statistically biased due to bad samples, bad methodology, bad weighting, etc.

For instance, Silver has Obama up outside of the MOE in Ohio. However, he has Romney with 15% instead of 5%. That's because, despite all the criticism for him being liberally biased or a numbers nerd who lives in his mom's basement, he is taking into account the chance that most of the state polls could be statistically biased against Romney.

[JediKooter]

I have a feeling that Obama and the democrats will totally Norv Turner the election. I see too many similarities in how their campaign has gone and how Norv Turner prepares the Chargers and the resulting product on the field ends up with too many games that should have never have been lost.

[Big Fo]

Originally Posted by Lathum

It is amazing to me anyone thinks Romney has a shot in hell of winning.

That doesn't surprise me, he is a longshot but it's not impossible. But over a dozen people on here thinking Romney has a even or better chance is kinda surprising.

[sterlingice]

Even at a 1 in 5 shot, that's not that long of odds. Yeah, I'd rather be on the 4 in 5 shot, but it's not something even remotely outside of the realm of possibility.

SI

[Passacaglia]

538 has Romney with a 13.7% chance this morning, so it's looking more like a 1 in 7 shot. Incidentally, this is about what it was right before the first debate.

[Crapshoot]

I'm kinda shocked by how big the gap is here. Guys I respect a lot on the right like Dan McLaughlin (you guys know him as baseball crank, like Nate Siliver a baseball guy first) see this as breaking pretty clearly to Romney, as do the likes of Michael Barone (300 EV + results). Guys I respect on the left like TNR et all see this ending up pretty clear on the Obama side. Someone is meaningfully wrong, and I'm fascinating (more from a statistical and strategic perspective) to see what it is; in theory, the core goal should be go to get it right.

[lighthousekeeper]

same here. especially professional polling groups. i would think some professional polling group must lose credibility post-election.

[lighthousekeeper]

for some strange reason i like to rely on drudgereport as a news source. i know it's always been heavily right leaning, but lately it's gotten especially ridiculous regarding the skewing of the truth. only the polls that favor romney the most are cited - all others are ignored. when bush was president, especially during his second term, reporting was a bit more even-handed (though of course still right leaning).

[stevew]

I used to go to Drudgereport a lot as well, but it's become pretty garbage lately.

[Blackadar]

Originally Posted by Crapshoot

I'm kinda shocked by how big the gap is here. Guys I respect a lot on the right like Dan McLaughlin (you guys know him as baseball crank, like Nate Siliver a baseball guy first) see this as breaking pretty clearly to Romney, as do the likes of Michael Barone (300 EV + results). Guys I respect on the left like TNR et all see this ending up pretty clear on the Obama side. Someone is meaningfully wrong, and I'm fascinating (more from a statistical and strategic perspective) to see what it is; in theory, the core goal should be go to get it right.

Someone is wrong, but you're looking at apples to oranges. One side is predicting from the gut. The other is using statistical data.

The only way this breaks for Romney is if the polls are fundamentally flawed. It's one thing to say that a poll is within the margin of error, so it's a proverbial dead heat. But when virtually every close poll is now leaning Obama, that margin of error really doesn't apply anymore.

Take Colorado polls over the last week or so. Ipsos, which had been +Romney, is now dead even. YouGov, PPP, CNN, CallFire, SurveyUSA, WeAskAmerica, Grove and Purple Strategies are all +Obama. Only Rasmussen (which has a known +Republican polling methodology) and American Research Group (the oldest poll, taken on 10/28) are +Romney. That's a very clear trend.

Ohio is even worse for Romney. Wenzel shows +Romney. Rasmussen shows it as a tie. Every other poll taken in the last week - 12 of them - shows +Obama.

Virginia is solid +Obama for the last week in every single poll. Yes, the numbers are slight (+1 to +2%), but every poll says the same thing.

When you start looking at polls from an aggregate standpoint, there really isn't much margin of error anymore. So there's no evidence to say that the polls support Romney. So while right-wing pundits that you respect may be saying "Romney", there's no evidence to support it. The only way they can do that is to discount (or cherry pick) polling data. So as I said before, looking at the data from a logical, aggregate standpoint, the only way Obama loses is if the polling data is fundamentally flawed.

So what's more likely? Someone's gut feeling? Or sound statistical methodology?

[mckerney]

Originally Posted by Blackadar

Someone is wrong, but you're looking at apples to oranges. One side is predicting from the gut. The other is using statistical data.

The only way this breaks for Romney is if the polls are fundamentally flawed. It's one thing to say that a poll is within the margin of error, so it's a proverbial dead heat. But when virtually every close poll is now leaning Obama, that margin of error really doesn't apply anymore.

Take Colorado polls over the last week or so. Ipsos, which had been +Romney, is now dead even. YouGov, PPP, CNN, CallFire, SurveyUSA, WeAskAmerica, Grove and Purple Strategies are all +Obama. Only Rasmussen (which has a known +Republican polling methodology) and American Research Group (the oldest poll, taken on 10/28) are +Romney. That's a very clear trend.

Ohio is even worse for Romney. Wenzel shows +Romney. Rasmussen shows it as a tie. Every other poll taken in the last week - 12 of them - shows +Obama.

Virginia is solid +Obama for the last week in every single poll. Yes, the numbers are slight (+1 to +2%), but every poll says the same thing.

When you start looking at polls from an aggregate standpoint, there really isn't much margin of error anymore. So there's no evidence to say that the polls support Romney. So while right-wing pundits that you respect may be saying "Romney", there's no evidence to support it. The only way they can do that is to discount (or cherry pick) polling data. So as I said before, looking at the data from a logical, aggregate standpoint, the only way Obama loses is if the polling data is fundamentally flawed.

So what's more likely? Someone's gut feeling? Or sound statistical methodology?

So you're saying when George Will is predicting Minnesota to go for Romney he doesn't have any data to back that prediction up?

[Jas_lov]

George will is a hack. He had Romney winning by over 100 electoral votes. It's the same debate as baseball. Stats vs gut, grit and heart. These are at best coinflips and Romney needs to win too many of them. Obama isn't going to lose every state he has a slight lead in. Nevada seems to be in his column now, possibly Iowa as well. It might turn out that he doesn't even need Ohio.

[RendeR]

Originally Posted by mckerney

So you're saying when George Will is predicting Minnesota to go for Romney he doesn't have any data to back that prediction up?

pretty much, yes. Minnesota is about as likely to go Romney as I am to run naked through my neighborhood declaring "I'm a right wing nutjob and I LOVE IT!"

[JonInMiddleGA]

Obama wins, Romney manages to pick up 1-2 states that McCain couldn't.

Obama raw vote count noticeably lower than four years ago, Romney virtually identical to McCain. Overall turnout down.

[Philliesfan980]

Originally Posted by JonInMiddleGA

Obama wins, Romney manages to pick up 1-2 states that McCain couldn't.

Obama raw vote count noticeably lower than four years ago, Romney virtually identical to McCain. Overall turnout down.

Sounds about right - well said.

[PilotMan]

Originally Posted by RendeR

pretty much, yes. Minnesota is about as likely to go Romney as I am to run naked through my neighborhood declaring "I'm a right wing nutjob and I LOVE IT!"

I smell a challenge!!!!

[Crapshoot]

Originally Posted by Blackadar

Someone is wrong, but you're looking at apples to oranges. One side is predicting from the gut. The other is using statistical data.

The only way this breaks for Romney is if the polls are fundamentally flawed. It's one thing to say that a poll is within the margin of error, so it's a proverbial dead heat. But when virtually every close poll is now leaning Obama, that margin of error really doesn't apply anymore.

Take Colorado polls over the last week or so. Ipsos, which had been +Romney, is now dead even. YouGov, PPP, CNN, CallFire, SurveyUSA, WeAskAmerica, Grove and Purple Strategies are all +Obama. Only Rasmussen (which has a known +Republican polling methodology) and American Research Group (the oldest poll, taken on 10/28) are +Romney. That's a very clear trend.

Ohio is even worse for Romney. Wenzel shows +Romney. Rasmussen shows it as a tie. Every other poll taken in the last week - 12 of them - shows +Obama.

Virginia is solid +Obama for the last week in every single poll. Yes, the numbers are slight (+1 to +2%), but every poll says the same thing.

When you start looking at polls from an aggregate standpoint, there really isn't much margin of error anymore. So there's no evidence to say that the polls support Romney. So while right-wing pundits that you respect may be saying "Romney", there's no evidence to support it. The only way they can do that is to discount (or cherry pick) polling data. So as I said before, looking at the data from a logical, aggregate standpoint, the only way Obama loses is if the polling data is fundamentally flawed.

So what's more likely? Someone's gut feeling? Or sound statistical methodology?

Hey, my bias is strongly, strongly towards the stats side (as anyone who has read anything I've posted here on baseball can confirm ). But I think Dan McLaughlin made some reasonable, statistical points about the value of the data we're measuring and whether we can be really sure if we're capturing the core components; ie, is it possible that the strong decline in poll response rate (below 10%)? Dan''s a smart guy who understands math, not the unskewed idiot or Sean Hannity or Dick Morris. I'm definitely not saying I agree, but I'm intrigued. My money, as it were, is on Obama 2008, losing in NC / IN / VA /FL, but winning 290 to 248. My map is here: 2012 Presidential Election Interactive Map and History of the Electoral College

On Polling Models, Skewed & Unskewed | RedState


Again, I'm fundamentally a numerically oriented data junky - but I think an important is learning what we're not capturing accurately in our models, to improve the proverbial R-squared. I

[sabotai]

Originally Posted by mckerney

So you're saying when George Will is predicting Minnesota to go for Romney he doesn't have any data to back that prediction up?

Inconceivable!

[Passacaglia]

538 has Obama with a 92% chance to win. So now we're talking 1 in 12 shot.

[Crapshoot]

The guys at RedState + Karl Rove have it for Romney, albeit at less than 300 Ev. Michael Barone has Romney with 320+ EV.

[mckerney]

RedState has a Romney win hinging on winning Wisconsin, seems like it might be more wishful thinking than anything else.

[Izulde]

Originally Posted by mckerney

RedState has a Romney win hinging on winning Wisconsin, seems like it might be more wishful thinking than anything else.

Indeed. People are still pissed at Scott Walker for what he did to the unions and though he survived a recall election, that was less about support for Walker and more about folks hating to have a recall election.

[Blackadar]

Originally Posted by Crapshoot

The guys at RedState + Karl Rove have it for Romney, albeit at less than 300 Ev. Michael Barone has Romney with 320+ EV.

Yes, but that's like saying the people at the DNC have it for Obama. The great thing about 538 is that there's no bias - it's an objective look at the data and not just another talking head spouting an uninformed/biased opinion.

[stevew]

I have to work all night. What site would be good to follow for reasonably unbiased election results?

[Crapshoot]

CNN and the NYT or WSJ. Good news orgs. Fox will be in palpitations or in celebratory throngs.

[Scoobz0202]

Originally Posted by stevew

I have to work all night. What site would be good to follow for reasonably unbiased election results?

I plan on using the link I just posted a couple posts back. If it updates regularly I should be good to go. No punditry, just numbers.

e: Er... it's in one of these threads. Too many right now. The NYT map.

[mckerney]

Originally Posted by stevew

I have to work all night. What site would be good to follow for reasonably unbiased election results?

NPR or BBC

[DanGarion]

Originally Posted by mckerney

NPR or BBC

Come on everyone knows that NPR is a left wing jizzfest.

(listens to NPR and isn't a left winger)

[JonInMiddleGA]

Originally Posted by stevew

I have to work all night. What site would be good to follow for reasonably unbiased election results?

If you only want top line numbers, maybe try BBC - Homepage

edit to add: They also usually do some analysis & fairly straight news writethrus as well, but that might be a way to avoid some of the standard U.S. media bias.

[sterlingice]

The Brits probably just sit at home, sipping tea, and laughing at the Florida-style cluster eff from the colonies.

At least that's how it happens in my mind.

SI