General Discussion
Related: Editorials & Other Articles, Issue Forums, Alliance Forums, Region ForumsScarborough endorses PPP polls.
Joe Scarborough ?@JoeNBC
A series of PPP polls out last night also have the president winning in almost all swing states. It's a D poll but was accurate in 2010.
Retweeted 97 times
https://twitter.com/search?q=ppp%20poll&src=typd
powergirl
(2,393 posts)and tweeted. LOL
malaise
(269,054 posts)Trust me
DarthDem
(5,255 posts)His mood tomorrow (I assume he's on) will be a leading indicator.
Honeycombe8
(37,648 posts)and others have said, that have built statistics based on the polls. The stats/probabilities will be wrong if the polls are wrong. Not likely, but possible, esp in a close race.
DarthDem
(5,255 posts)What's the point in being pessimistic now, at this stage? Especially when all rational evidence indicates that the president is going to win.
Honeycombe8
(37,648 posts)All rational evidence does indicate Obama will win. But crap happens. Who could have predicted Sandy?
It ain't over 'til it's over. I'm not going to tempt fate.
MissMarple
(9,656 posts)CTyankee
(63,912 posts)It doesn't seem possible what with the science behind modern day polling. Will some "universal" glitch in what is standard polling practice just suddenly pop out with all the of polls just going haywire at the same time?
MissMarple
(9,656 posts)That's what his inside polling guy is telling the campaign. His strategists aren't buying it, neither is Nate Silver. A better underpolling senario is with cell phone users with no land lines, they are underpolling them. And they tend to be younger and less affluent. I want to see how Georgia goes. Now that could be interesting.
Here's a link to the romney meltdown on polling.
http://www.democraticunderground.com/10021719750
Honeycombe8
(37,648 posts)It's only the battleground states that are in question, and only some of those.
Nine counties have been involved in fraud by Republicans, and things are looking fishy in Ohio.
Obama will still probably win. But it's not nice to tempt fate.
CTyankee
(63,912 posts)Mutiny In Heaven
(550 posts)but in his heart, it was SO real.
aaaaaa5a
(4,667 posts)instead of the bogus poll out of PA, which has been proven to have a terrible track record.
Hmmmmmm.....
Texas Lawyer
(350 posts)consistently favored Romney.
I think the outdated methodology probably balances out the D-lean.
grantcart
(53,061 posts)They get the D leaning image because there are so many RW polls that when they publish real numbers they are always 2-3 points to the left, but that is usually where the polls end up before the RW polls do their last minute pre poll shift to get their numbers in line with what turns out to be the correct result.
PPP has NC as a tie. The RW polls show it a blow out for Romney. We will see who is right.
Texas Lawyer
(350 posts)are pretty consistently accurate.
Here is Nate Silver's discussion of why he assigns PPP a pro-D in-house effect: http://fivethirtyeight.blogs.nytimes.com/2012/06/22/calculating-house-effects-of-polling-firms/
grantcart
(53,061 posts)He is using a consensus of surveys. What if one side throws millions into surveys not to read opinion but to help shape it. That would pull his consensus that way.
Well if you are adding Gravis, and others like it then it will pull it way to the left. I think that this is much more damning than anything I have read about Nate. He isn't basing it on historical performance but on the average of all of the surveys now being issued. What if the left stopped with guys like Zogby and the right doubled up.
It will be interesting to see. I am guessing that PPP is much closer to the middle than Rasmussen is.
As you can see, there is a fairly wide spread in the polls this year. For instance, the firm Public Policy Polling shows results that are about three percentage points more favorable to Mr. Obama than the consensus of surveys.
Texas Lawyer
(350 posts)susceptible to manipulation by a scam artist fraudster/pollster like Gravis who pitches in bogus numbers to move the needle).
Nate Silver weighs the polls (partly based on methodology and margin of error and a pollster's history of accuracy), and then he accounts for a pollster's in-house effect (so he accounts for the fact that Rasmussen is habitually skews a few points below the consensus for the President and Pew skews a few points above the consensus for the President).
I agree that PPP will be closer to the middle than Rasmussen because both mainly rely on land-line-only robocalls. I believe this somewhat dated methodology skews naturally toward Romney because we kick ass among the younger demographic that uses cell phones as their only phone service (and is under-counted by this method). The pro-Romney land-line-only methodology balances out any pro-Democrat in-house effect at PPP so I think PPP comes close to hitting the target (as it has done historically). On the other hand, the pro-Romney land-line-only methodology exacerbates any pro-Repub in-house effect at Rasmussen so it will miss the mark (just as Gallop will miss the mark).
One note: in the past, I have seen where Rasmussen seems to be way off the mark throughout the election cycle until the last poll, when their polls suddenly veer much closer to the mark. That makes me wonder if Rasmussen has his thumb on the scale.
grantcart
(53,061 posts)Look at NC
http://fivethirtyeight.blogs.nytimes.com/author/nate-silver/
Gravis has NO HISTORICAL RECORD.
Their first poll was on August 20th.
Two months later Gravis on 10/24 that has Romney up by 8.
There is no historical basis for Gravis and Gravis is way out of sinc with other polls
Now look at the weighting. While other polls diminish in weight Gravis maintains a higher weight.
Gravis +8 on 10/24 has maximum weight.
PPP on 10/28 has half of the value.
Silver is making judgements on what a pollster is doing by what they say they are doing, and not by any real evidence of what they are doing.
Obviously I have made the argument that Silver should have done a little checking on who the principles actually are.
But even if he hadn't done that he should have questions raised by Gravis polls in NC and in CO.
Here is one example
On October 3/4 Gravis publishes a poll from CO with this results
Black 40.34 Obama 2.12 Other 57.54 Romney 100.00
http://www.realclearpolitics.com/docs/2012/Gravis_CO_1006.pdf
Despite the fact that Gravis produces the only poll in history that shows Romney out polling AA he still gets 4 big bars on that and every Gravis poll.
In reality there is very little difference between RCP and Nate Silver.
RCP has Romney at +3, Silver has had him +3 but today went to +2.
In reality it is a distinction without a real difference and there isn't any evidence that Silver is using historical performance.
Fordham rates Pew as the most accurate in 2008 National Polls and Nate has him as the most biased
http://www.fordham.edu/images/academics/graduate_schools/gsas/elections_and_campaign_/poll%20accuracy%20in%20the%202008%20presidential%20election.pdf
When it comes to state races and state polls, I don't know of anybody that outperforms PPP, if there is such a liberal bias then why were they by themselves predicting a Brown victory in MA?
PPP first entered prominence through its performance in the 2008 Democratic primaries between Barack Obama and Hillary Clinton. The company performed very well, producing extremely accurate predictions in many states ranging from South Carolina to Wisconsin, many of which featured inaccurate results by other pollsters.[7][8][9]
In 2010, PPP was the first pollster to find Scott Brown with a lead over Martha Coakley in the Massachusetts Senate special election; Brown ultimately won in what was considered an upset.[10]
In 2011, PPP was praised for its accuracy in polling primaries and special elections, which are notoriously hard to predict. The contests they accurately predicted include the West Virginia gubernatorial primaries, special elections in New York and California,[11][12] as well as all eight Wisconsin recall elections.
Texas Lawyer
(350 posts)what he does and Votamatic does and what RealCearPoltics does and what Pollster does, etc.
http://election.princeton.edu/2012/11/04/comparisons-among-aggregators-and-modelers/
I agree that Gravis is a fraudster, but I think I read where Silver tags him with a big Pro-R in-house effect.
grantcart
(53,061 posts)Wang has one map that has NC/FL even (which is fair) and another that they are slight pink
When you hit on Wang's NC link it brings you to this
http://elections.huffingtonpost.com/pollster/2012-north-carolina-president-romney-vs-obama
Gravis showing Romney +4 (likely repositioning before the election after consistently having it +8/9
PPP has it even, which they have had it even for a long time.
One of the two is going to be more accurate.
As for the big Pro-R in-house effect I have also heard that, and that may be in the national polls but in the state polls, according to the bars that are attached, he gives Gravis a greater weighting.
If you are correct, and it is entirely possible, then at best Silver's presentation misrepresents his actual calculation.
Texas Lawyer
(350 posts)time so it may look like he is assigning more weight to a poll as compared with another poll, but you may be comparing a "brand new" Rasmussen poll against a week old Pew poll. It is not a negative judgment on the Pew methodology that Rasmussen has more weight -- it would be a function of which poll is more recent.
Also, the weighting for in-house effect is separate from the weight bars.