Some polls now have Romney ahead.

[Double Trouble]

Intrade is moving towards Romney.

54.7 - 45.5

There's been a major shift on intrade over the past few weeks, a move that puzzles me because I don't see it justified/supported by the data on RCP and 538. Unless there's a major shift in Ohio, Wisconsin, Pennsylvania and VA, I don't see how the math adds up to Romney winning anything. Since I haven't seen any indications of such a shift occurring, I"m puzzled by the rapid intrade shift.

 

[buckshot24]

There's been a major shift on intrade over the past few weeks, a move that puzzles me because I don't see it justified/supported by the data on RCP and 538. Unless there's a major shift in Ohio, Wisconsin, Pennsylvania and VA, I don't see how the math adds up to Romney winning anything. Since I haven't seen any indications of such a shift occurring, I"m puzzled by the rapid intrade shift.

I think people may have been reading this thread.

 

[randomrogue]

2008 Rasmussen wasn't heavy Republican.

According to PPP Rasmussen turned out some of the worst results in the last election. Check out their home page.

 

[buckshot24]

According to PPP Rasmussen turned out some of the worst results in the last election. Check out their home page.

According to this page Rasmussen was off 5.8% in 2010 with a bias towards Republicans 3.9% which means they had a bias towards Democrats by 1.9%. So on average they were 2% more favorable to Republicans.

The best firm in 2008 was Quinnipiac and they had an error rate of 3.3%

However if you go to

http://www.fivethirtyeight.com/2010/06/pollster-ratings-v40-methodology.html

You see that the average error of all polls for non presidential races is pretty high.

House 6.0%
US senate 5.1%

*National polls for president have an average error rate of 2.8% while state polls have a 3.8% error.

Dismissing Rasmussen out of hand for producing error rates that are average in one non presidential election is probably not warranted. In the last presidential race they were pretty spot on nationally.

 

[xBiffx]

According to PPP Rasmussen turned out some of the worst results in the last election. Check out their home page.

One pollster declaring another pollster's results are crap. News at 11.

 

[mshan]

FWIW, I've now seen two different declarations that the race is over (Obama has won), I think because there are too few truly undecided voters remaining (http://video.msnbc.msn.com/the-daily-rundown/49086377#49086377) too little time left for Romney to change the terrain before Nov. 6:

- http://www.predictwise.com/node/1003

- http://election.princeton.edu/2012/10/23/debate-3-obama-wins/

Seems like what remains at state is control of the House, which may be coarsely correlated with Obama's national polls margin (http://election.princeton.edu/2012/10/06/predictions-october-6th-house/)


The (Possible) Coming Obama Landslide?:

"So if Pennsylvania is off the boards, let’s look around.

Imagine it’s election night, say 10:45 east coast time. Four eastern states haven’t been called yet: Ohio (18), Virginia (13), North Carolina (15), and Florida (29). Also, in some Western states, the polls haven’t closed, or the races are too tight to project just yet—Colorado and Nevada, say. Arizona has just been called for Romney. At this point, Romney actually leads, 188 to 182. In this scenario I’m assuming Obama has won Iowa (6), which is admittedly close but where his lead has been stable at three or four points, and New Hampshire (4), where Obama has a similar fairly small but stable lead, and Michigan (16), where the gap appears to be opening up a little.

So it’s a six-vote Romney edge. They’re feeling great up in Boston. Especially with the big Eastern four still up in the air. Right?

Not really. Let’s look at these West Coast states. Even though they’re still voting in California, obviously Obama is going to win it (55). And equally obviously, he’s going to win Washington (12) and Oregon (7), where neither side even bothered to spend a dime. Throw in Hawaii (4). Those 78 votes haul Obama up to 260. That’s something to keep in mind for election night: Whatever Obama’s number is at 10 pm Eastern, add those 78 EV’s—they’re a mortal lock, and a hefty insurance policy. If he wins Nevada (6) and Colorado (9), it’s over.

In other words, Obama can lose the big Eastern four—Ohio, Virginia, North Carolina, and Florida: all of ’em!—and still be reelected."

http://www.thedailybeast.com/articl...y-on-the-possible-coming-obama-landslide.html

Citi: http://www.zerohedge.com/news/2012-...nore-pundits-partisans-huffpo-data-says-obama

 

[xBiffx]

http://www.colorado.edu/news/releas...ting-model-still-points-romney-win-university

Never been wrong. Granted this is about popular vote so it doesn't necessarily mean Romney wins EC. They do extrapolate an EC figure, however.

 

[mshan]

That model has never predicted anything, in advance... (i. e. IIRC, it didn't exist before last election cycle)

First revision used economic data from around May / June, when there was some suggestion economy might be teetering back into recession, and second revision I think used September data, which probably reflected transient weakness in August.

And again, they retrofit their model, after the fact, to the data, not the other way around.

 

[xBiffx]

That model has never predicted anything, in advance... (i. e. IIRC, it didn't exist before last election cycle)

First revision used economic data from around May / June, when there was some suggestion economy might be teetering back into recession, and second revision I think used September data, which probably reflected transient weakness in August.

And again, they retrofit their model, after the fact, to the data, not the other way around.

http://www.denverpost.com/recommended/ci_21373080

The professors who created the model, Ken Bickers from CU-Boulder and Michael Berry from CU-Denver, say it correctly forecast every winner of the electoral since 1980.

Unless the story is incorrectly using the word "forecast" I believe you are incorrect.

 

[mshan]

IIRC, it is after the fact, in that model was not up and running before last cycle.

You could probably argue it has been retrofit to select economic variables to be consistent with previous cycles and to get result they want...

 

[Charles Kozierok]

Biff, your own link says right at the top that the model was just announced in August of this year. They went back and applied it to the preceding eight elections.

mshan is correct.

 

[Farang]

That Colorado 'prediction' model is the dumbest and worst reported on thing in all of polling this cycle.

It literally means as much as this: http://news.yahoo.com/your-nfl-team-can-predict-the-election.html

But journalists bungled the reporting and confused a bunch of people.

 

[xBiffx]

Biff, your own link says right at the top that the model was just announced in August of this year. They went back and applied it to the preceding eight elections.

mshan is correct.

Check again. The August reference refers to when they released/updated their prediction based on the model for this election cycle. Not to when the model was created.

 

[randomrogue]

According to this page Rasmussen was off 5.8% in 2010 with a bias towards Republicans 3.9% which means they had a bias towards Democrats by 1.9%. So on average they were 2% more favorable to Republicans.

The best firm in 2008 was Quinnipiac and they had an error rate of 3.3%

However if you go to

http://www.fivethirtyeight.com/2010/06/pollster-ratings-v40-methodology.html

You see that the average error of all polls for non presidential races is pretty high.

House 6.0%
US senate 5.1%

*National polls for president have an average error rate of 2.8% while state polls have a 3.8% error.

Dismissing Rasmussen out of hand for producing error rates that are average in one non presidential election is probably not warranted. In the last presidential race they were pretty spot on nationally.

Are you comparing popular vote to EC again?

 

[mshan]

"47% is probably the most famous number to come out of the 2012 presidential campaign. That’s the percentage of Americans Mitt Romney suggested he was writing off when captured on that famous secretly recorded video during a Florida fundraiser.

But 47% also represents the critical dividing line between how the Romney and Obama campaigns gauge the President’s chances of re-election, viewed through the prisms of their closely guarded internal polling data. In general, the Republicans see a ceiling of around 47% for Obama’s share of the vote in most of the nine battleground states that will decide this contest. The Obama team says it consistently measures their man above that magic line, at around 48-50%.
Both campaigns employ top-flight pollsters who are paid to scientifically take the pulse of the electorate in the battleground states and deliver to the candidate and his strategists an unvarnished view of the state of the race at any given moment. It doesn’t well serve either the incumbent or the challenger to sugarcoat the situation by producing poll results that overstate support (as some pollsters are famous for doing). The key differential between the Chicago and Boston methodologies appears to be assumptions about the final makeup of the electorate in the nine battleground contests.

If the President’s internal data is correct, he is indeed likely to win, pushing off a floor above 47%; using his superior get-out-the-vote operation to dominate early and absentee voting and hold his own on Election Day; and taking a significant share of the small remaining undecided vote. These numbers apply to Chicago’s data from the battleground states, where the President’s support in his campaign’s internal research has consistently outperformed his national poll standing."

"As one Obaman told me: “First of all, there aren’t that many undecideds. And let me say one things about the undecideds …. The undecideds— the structure of these undecideds are such that they very much reflect the electorate as a whole. They’re not like a divergent. And, while our standing isn’t as good with these undecideds as they are with people who are voting for us, our standing with these voters is at least as good or better than Romney’s.”

http://thepage.time.com/2012/10/24/...n-is-still-so-confident-about-beating-romney/

 

[lotus503]

I think people may have been reading this thread.

I think people know math and odds and with a huge swing in Obamas favor a romney bet is more attractive.

at some of the earlier intrade odds, I had considered owning me some Romney

 

[Charles Kozierok]

Check again. The August reference refers to when they released/updated their prediction based on the model for this election cycle. Not to when the model was created.

August is the first time they told anyone about this alleged perfect model of theirs. If it was so accurate on the previous eight elections, why did they keep it to themselves until now?

Seems pretty strange to me.

 

[Charles Kozierok]

If the other polls stabilize at Romney being in the +5 or +6 range over the next few days, then I'll be happy to admit I was wrong and reassess. But if that happens without significant movement in the state-level polls, it will just leave the situation muddled.

Gallup is down to +3 Romney today.

And has Obama's approval up to a whopping +11.

 

[Farang]

http://thepage.time.com/2012/10/24/...n-is-still-so-confident-about-beating-romney/

More important than what you quoted, I think, is this:

Said one senior official: “But the most important thing about early vote is one thing and one thing only: are you getting your sporadic voters to vote? Because if it’s just chasing people who are going to vote anyway than it’s just…a zero sum game. But all the data I see says we are getting our sporadics to vote at a higher rate than they are, which, especially for any Democratic candidate, is a bigger challenge because we have lower propensity voters. That’s exactly what we are doing and we feel great about that.”

This is what makes Ohio unwinnable for Romney, in my view, unless he can pull into a clear lead immediately. Every day that doesn't happen, Obama remains the favorite. Meanwhile Romney is still on defense in FL, VA, and CO, states Obama doesn't necessarily need

 

[lotus503]

I think this year Ohio decides it and 538 has Obama at %70 likley win in Ohio.

2 months ago I was pretty sure Romney would win, based on Money.

 

[mshan]

"Chicago is keying off of a daisy-chain of educated assumptions and analysis of existing data to inform their view of the race: the demographic groups that disproportionately back the President will make up a sufficient contribution to the total vote to provide the margin of victory; both new registrants and the early votes now banked are coming disproportionately from those same groups, many of whom are low propensity voters who might not otherwise cast ballots in traditional patterns; the make up of the small remaining undecided bloc is not starkly adverse to the incumbent; and five of the nine battleground states are near-locks for the President, enough to make it impossible for Romney to reach 270 electoral votes."

Intrade:

On Monday night, after the debate, Barack Obama was leading Romney on Intrade by around 60 percent to 40 percent. But at around 10:00 a.m. on Tuesday morning, Romney surged to 48 percent. Was this evidence that the conventional wisdom was wrong? Had Romney actually won the debate handily? Or, alternatively, was the nosedive in the stock markets putting a dent in Obama’s re-election chances?

Neither. As economist Justin Wolfers pointed out on Twitter, the huge swing toward Romney appears to have been driven by a single trader who spent about $17,800 buying up Romney shares and pushing the Republican candidate’s chances on Intrade up to 48 percent. But the surge only lasted a few minutes before other traders whittled the price back down to what they saw as a more accurate valuation. Romney’s odds of winning are currently back at around 41 percent.

…As Wolfers pointed out, this mysterious trader ended up overpaying by about $1,250 for shares that quickly collapsed in value. Was this just someone who made a bad trade? Or was somebody trying to influence Intrade odds in order to sway perceptions of the race? And if so, was it worth $1,250 to jolt the markets for less than 10 minutes?"

http://marginalrevolution.com/marginalrevolution/2012/10/intrade-manipulation-fail.html

London Betting: http://news.yahoo.com/britains-bookmakers-see-obama-returning-white-house-092228046--finance.html

 

[Todd33]

According to this page Rasmussen was off 5.8% in 2010 with a bias towards Republicans 3.9% which means they had a bias towards Democrats by 1.9%. So on average they were 2% more favorable to Republicans.

The best firm in 2008 was Quinnipiac and they had an error rate of 3.3%

However if you go to

http://www.fivethirtyeight.com/2010/06/pollster-ratings-v40-methodology.html

You see that the average error of all polls for non presidential races is pretty high.

House 6.0%
US senate 5.1%

*National polls for president have an average error rate of 2.8% while state polls have a 3.8% error.

Dismissing Rasmussen out of hand for producing error rates that are average in one non presidential election is probably not warranted. In the last presidential race they were pretty spot on nationally.

I think you are confused and just throwing numbers around. Can you elaborate on the bold part? They were wrong by 5.8%, with a bias towards the Republican. This has to do with their models.With the link you provided (did you read the text?), they tend to assume there are more Republican than really are and their polling methods also favor Republicans (ie. old white people sitting by phones). You then go on to compare the actual error rate (ie. how far they were from the real results) with the margin of error to individual polls, which is a function solely of how many participants are used (margin goes down with more people).

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

I would say you are reading the data wrong and seriously don't understand it.

 

[xBiffx]

August is the first time they told anyone about this alleged perfect model of theirs. If it was so accurate on the previous eight elections, why did they keep it to themselves until now?

Seems pretty strange to me.

Time to stop spewing the bullshit. The model has been around before August and they haven't kept anything to themselves.

http://spot.colorado.edu/~mcnownr/working_papers/presidential_voting_mscpt.pdf

 

[mshan]

Are you sure that the 2008 paper was a model, or perhaps just previous research / foundation upon which this supposed model is now based (e. g. intro to article says it went back to 1972, yet their current study only claims predictive power since 1980).

Plus from their updated press release:

The authors also provided caveats. Their model had an average error rate of five states and 28 Electoral College votes. Factors they said may affect their prediction include the timeframe of the economic data used in the study and that states very close to a 50-50 split may fall in an unexpected direction due to factors not included in the model.

“As scholars and pundits well know, each election has unique elements that could lead one or more states to behave in ways in a particular election that the model is unable to correctly predict,” they wrote.

All 13 election models can be viewed on the PS: Political Science & Politics website at http://journals.cambridge.org/actio...nts-romney-win-university-colorado-study-says

The fact that they are stilling putting Pennsylvania in Romney's column should tell you something...

 

[xBiffx]

Are you sure that the 2008 paper was a model, or perhaps foundation upon which this supposed model is now based (e. g. intro to article says it went back to 1972, yet their current study only claims predictive power since 1980).

The paper describes the model. They only claimed its predictive power back to 1980. Perhaps because it was wrong before then.

The fact that they are stilling putting Pennsylvania in Romney's column should tell you something...

Nothing is certain at this time?