Gravis Working Group versus the New York Times, letter to NYTimes Ombudsman
<Because the work product to get here was shared I give credit to the unnamed assistants of the GWG but
in this case the arguments are mine, so any errors are mine alone. It is doubtful that a single letter to the
NYT is going to get far. If you are interested in sharing your opinion on the subject Margaret Sullivan is the
NYT Ombudsman and her email address is public@nytimes.com.
Also I don't have time to crosspost this elsewhere. I waive all copyright and thank you in advance for crossposting elsewhere, please let us know if it gets any traction >
Dear Margaret,
I am writing to raise serious questions about the NYTimes use of polls from the time of August through today.
I have discovered and exposed one of the most frequently used and heavily weighted pollsters used by the NYT (and more specifically Nate Silver) as someone who is a prolific liar with no post high school education (although multiple CVs claim 'attending' various diploma mills like DeVry) and no previous professional association with a pollster. With a group of citizen investigators we have pieced together the 'pollsters' last 10 years of professional work and know that he is a frequent target of federal and state investigations because he violates federal laws with junk faxes and robo calling. In fact he is a robo callier, not a pollster, confirmed by the source himself. I have detailed these charges in a 10 article expose at Democratic Underground and you can find an index to the articles here:
http://www.democraticunderground.com/10021568200
If you are short on time but have some interest start with article;
X. Gravis Working Group: Douglas Kaplan confirms our charges he is not a pollster.
http://www.democraticunderground.com/10021709599
After our constant bombardment Douglas Kaplan admits the following
1) He is not a professional pollster
2) He has no academic training in polling
3) His earlier polls were substandard
4) He has 'had to hire' a statistician and a 'political adviser' because these are areas outside of his expertise.
To the other charges, contained in a hundred pages of published material, he doesn't comment on, but no one has refuted a single one of our charges or facts.
This is not a trivial issue about a habitual liar (in an interview with the Voice of Russia Kaplan boasts that he has run both Presidential and State campaigns) who punked Nate Silver.
In the new world of post Citizen United we are faced with vast war chests of hundreds of millions of dollars entering the field from specific moneyed interests like Koch (heavy industrial) or Adelson (gambling). Does the NYT think that this wash of money is only going to be used to buy ad time?
It is not.
The Romney campaign has shown a propensity of buying supporting campaign infrastructure. Need a hundred thousand twitter followers? Buy them. Can't find editorial cartoons that make pro Romney cartoons? Set up a faux editorial cartoon site, hire a cartoonist that works for the Mormon Church and pretend that it has no connection to the Romney campaign.
http://mittfitts.com/test-2-2/
We have seen with the hateful antics of James Keefe III that there are resources to fund the type of activities that the NYTimes used to win Pulitzer Prizes for exposing ala Liddy and the Plumbers that were part of the Nixon White House. Now they appear to be part of the acceptable landscape and after having taken out an important advocate the poor, ACORN we are now seeing AA standing in 5 hour lines to vote just 12 years after we vowed that it would never happen again.
So it must be asked if the New York Times is missing the big picture, is there a campaign to use vast amounts of money to shape the geography of the campaign for the special interests, beyond simply buying ad.
Now we come to the more specific question of the polls that are used by Nate Silver. While not a statistician I understand his model to be based on the baseball idea model of a virtual conveyor belt of unending statistics. That model worked well in 2008 where there was a vast and unending number of polls due to two contested primaries and the historically unique campaign of electing our first African American President. This year not so much.
In August Gravis Marketing started publishing polls. A lot of them. In certain swing states they were the most frequent. In some cases they had the widest margin for Romney. Silver gave them the highest weight, higher than well established, but Republican leaning Rasmussen. Not only that there were a large number of other Right Wing polls (ARG, Rasmussen, Purple Strategies, et al, ad naseum). In 2008 you could argue that the polls balanced each other out. In 2012 that claim can no longer be made. Zogby is gone.
Silver does weight the polls. Curiously Rasmussen, who is a real pollster and gets paid for it, does not receive full credit because he leans consistently Republican. Rasmussen is honest about his bias and while not academically trained in statistics he has a known background and did in fact attend college.
Mr. Silver gives a heavier weighting to Mr. Douglas Kaplan. Kaplan is a proven liar who never achieved any academic distinction, and admits it. Mr. Kaplan's experience is that of junk faxing and robo calling primarily around marketing dubious travel packages on unsuspecting dupes. He has been the target of federal and state complaints as well as lawsuits by Disney.
Mr. Silver may not be aware of his past but Mickey Mouse is.
As I said I am not a statistician. I am however (or technically was some years ago) a certified inspector in the quality control system known as ISO 9001. In the terms of ISO 9001 the 538 pollster column is in 'significant non compliance' with any known quality system. Mr. Silver's product is based on the raw material he receives. If he doesn't control the BS he gets in he cannot control, no matter how much weighting he attempts, the quality of the product. The first step that Mr. Silver must undertake is to be sure that the pollsters he is using are in fact pollsters.
Honestly it took me only 30 minutes to figure out that Douglas Kaplan was a fraud. I explain it in detail in the above link but I after I clicked on Mr. Kaplan's media link and listened to an interview he did on Voice of Russia where in answer to the question (which was the announced topic of the show) "What is the importance of the South Carolina Primary?", Mr. Kaplan completely disassembles and proffers "South Carolina was the first state to secede from the Union" I knew that Kaplan was not only a fake but not a very intelligent one at that.
More investigation and more questionable details about Kaplan flowed out without much effort. Under the steady drumbeat of facts Kaplan acknowledges that he is in fact, not a pollster and that his early polls 'were not very good'. And yet those polls were fine for Mr. Silver.
On Mr. Silver's behalf I know that his model requires a vast number of polls to improve the prediction quality. But, quite frankly, Douglas Kaplan has successfully punked Nate Silver.
It is obvious that Silver's justification is that 'if it looks like a good poll, numbers match like a good poll, and it walks like a good poll' it is a good poll.
This is ridiculous. Clearly we have people doing polling not to reflect opinion but to move perception, or make opinion. Mr. Silver acknowledges this fact every time he gives Rasmussen polls a lower weight.
If you were interested in using a fake pollster, oh say like you are using a fake editorial cartoonist, the way that you would buy your way in is to offer up a number of polls that matched the expected average, then when needed you can start publishing numbers that pull the averaged poll numbers your way.
Mr. Kaplan started publishing polls only 77 days ago (this was 6 months after he had gone on Voice of Russia and talked about his extensive experience as a Presidential campaign manager and glibly lying about dozens of polls that he had done, including one that was coming out just as soon as he got off the air with the English speaking Russophiles. Lie after lie after lie, including detailed discussions about his work in certain counties in Iowa. All now acknowledged by Mr. Kaplan as not being true. He admits he is just a robo caller who wanted to get a bigger slice of the pie.
Again all of this was found out by spending 20 minutes listening to the tape that Kaplan had linked to his own website.
Which brings us to Mr. Silver's probability score on North Carolina.
North Carolina is actually more important than Ohio. Mr. Romney needs to flip a number of states from Mr. Obama's 2008 performance in order to have any level of credibility in the Electoral College race. He has Indiana. After that he has North Carolina. If he can't flip North Carolina (that was decided by only 25,000 votes) then the likelihood that he can flip Virginia, Ohio, Iowa, Nevada, or Wisconsin are non existent.
North Carolina is the key to maintaining a facade of Electoral College viability. Enter Gravis Marketing and a whole bunch of unknown right wing pollsters. I looked at Gravis because a) most frequent b) farthest out. At the same time that PPP is showing NC a tie. Gravis continues to show it a Romney runaway up by +9 +8 +4. Just as the NYTimes and other outlets are reporting that Romney is having problems raising money Gravis comes in with magic numbers in North Carolina.
Talking about money we should also note that we have also documented that Mr. Kaplan has hundreds of thousands of dollars of IRS liens, multiple foreclosures and has said that he 'isn't making a dime out of the polls'. The kind hearted Kaplan is favoring us with independent polls, paid out of his own pocket, even hiring statisticians and political advisers that coincidentally serve to maintain Romney's Electoral College viability and help keep his fund raising ship afloat.
Hey, even Mickey Mouse wasn't fooled by this guy.
Nate Silver maintains that North Carolina has a 80% probability of Romney winning (ok today he lowered it to 77%). Look at the early voting results. North Carolina is a 50/50 tossup, but given the number of new voters that the Obama campaign has brought in so far, you would have to give a slight edge to the President taking North Carolina.
Why does it matter? Its just one state.
We don't need Nate Silver to be right about Hawaii or Utah. His value, and his only value is in the close states. And in all of the close states, North Carolina is the most important for reasons noted above.
For the record we obtained emails of various editors including political desks and inside emails to Nate Silver offering all of the information about Doug Kaplan and Gravis Marketing before we blogged the results and never got an answer. Further these issues have been discussed in various parts of the blogosphere, including discussion threads at Intrade and we have to assume that Silver and the political desk did know that highly dubious facts about Gravis were available. There are more details to come as well.
At this point none of the charges or facts that we have documented have been refuted and in his last interview Kaplan confirms that he is not trained (either in an academic or professional setting) in polling that he learned it 'from scratch' and that the quality of some of his polls were not good.
Silver might have had a question when one of Gravis' polls showed a majority of African Americans supporting Mitt Romney, but even that obvious brick didn't change the fact that for most of September Gravis was the most prolific and therefore the most used pollster in polling averages and had become, for a few weeks at least "America's most powerful Pollster".
There is more to this story but I will leave you with just one question. At the same time that Kaplan is launching his faux polling company he also launches this Political Action Committee, The Protect Candidate Speech PAC
http://www.electionfund.org/Committee/Protect-Candidate-Speech-Pac
Nate, New York Times, why would a legitimate independent pollster start a PAC? Has this ever happened before? How would that be consistent with someone that was doing legitimate polling? Is there someway that a dishonest person could profit from such an arrangement? All questions for the future.
Nate Silver and the New York Times didn't just drop the ball, you dropped it in the ninth inning of the seventh game of the World Series.
Sincerely
grantcart
= new reply since forum marked as read
