TPM just published the full text of Research 2000 president Del Ali's response to Kos' lawsuit. In it he said:
Yes we weight heavily and I will, using the margin of error adjust the top line and when adjusted under my discretion as both a pollster and social scientist, therefore all sub groups must be adjusted as well.
I think he just admitted that he has manipulated data in the polls he did for Dailykos.
This sentence isn't so hard to parse. If Ali really means what he said here, that he applies weighting heavily and at his discretion, it implies that he can change the weighting algorithm by the seat of his pants. In Kos' words, that would be manipulating data "beyond recognition."
According to his description, he applies the weighting algorithm in the following steps:
- Using his intuition as both a pollster and social scientist, he will change the top line number up to the margin of error. So suppose the actual result is 45% and the margin of error is 4%, but he feels the correct number should be 51%, he will adjust the number to 49%.
- After the top line is adjusted, he then has to adjust the subgroup numbers so that they add up correctly to the total percentage.
This may explain the even-odd strangeness. One can imagine that he tried to keep the men-women difference in the favorables the same while adjusting the top line. So he may have simply took the top line number, and added and subtracted the half of the men-women difference to get the subgroup numbers for men and women. With rounding to integers that would guarantee the same evenness of men-women numbers.
Let's take the above hypothetical case as an example again. Suppose the actual subgroup favorable numbers were men 43% and women 46%, assuming same number of men and women, which should average to 44.5% but was rounded to 45%. If the top line was adjusted to 49% you have to also adjust men and women numbers to make them consistent. You take the difference between the men and women, which was 3%, divide by two, which gives you 1.5%, round it to an integer percent (2%), and add it or subtract it from 49% to get men 47% and women 51% favorables. Now the pair are both odd numbers.
His full statement:
Let's state for the record that as of this email, my legal counsel has not been sent this suit via email or anything! That should speak volumes about Kos and there counsel, it just adds further evidence in our pending suit.
I am floored that the any media outlet would print anything Kos releases publicly because just looking at the suit there are so many lies and fabrications that will expose them in litigation, not in the media. Additionally, this Kinkos story and this alleged two weeks to get raw data to them is the best illustrations not only in our defense but in our suit against them. The fact that Kos would publicly release cherry picked emails and not disclose others and to not think aggressive legal action would be taken is just mind boggling. We are not going to litigate or release any key evidence we have collected on Kos and others over the past two weeks. I think you would agree that it could compromise our case as well as give Kos and supporters in the media witch hunt for them time to spin their way out of it. Every charge against my company and myself are pure lies, plain and simple and the motives as to why Kos is doing it will be revealed in the legal process and not before that. I will share one little minor reason that Kos is doing this and it pertains to the fact they owe us a significant sum of monies that is in the six figure category and payment was on June 15, 2010. There are several more sinister ones that will come out.
Let me make an important point to you and NCOP. The raw data, the report used by the Kos bloggers posing as statisticians in which we have found that several of our numbers (the too clean ones) they used in this smear are FALSE too slander me, the use of Gallup's numbers (not attacking Gallup at all, I am defending them, just making a point that any vindictive phoney posing as a statistician could do the same to them) in the report to make a case, yet one university professor (will not go on record now because she fears a slander job by Kos and public attacks by supporters) with a PHD in Research Methods pointed out that Obama's fav rating dropped 26 points over 15 months in our Kos polling. The professor pointed out that Gallup's own tracking has shown what some could say are odd anomalies in identical results for several weeks: "Gallup's tracking from June 1-8, Obama's approval 46% on the June 1, 46% on June 8, June 22, 46% and June 28 46%". To further add on this point, if Obama or for that fact any other poll question measured only goes up one point or down one point in a week, why should any one expect there to be major shifts in any of the demographic groups? If we were calling this a true random sample un weighted, then I perhaps I would have had concerns as well.
I also have several issues with NCOP that I will address after this is settled. I will address their concerns at that time as well and not before. If Kos wants to make their case by blogging slander and lies even though they were served via email "Cease and Desist Demand" (we can not find a physical address for him and he will not give it to our counsel - also speaks volumes) that is great and so helpful to us.
Finally, it has been brought to my attention that NCOP as well as others who have asked publicly, why not just release the data and explain your methodology as well as this this garbage statement by some that our data is "too clean". This is clear dog whistle support for Kos and a disregard for the legal process and our case. I say this because for discussion purposes, let us assume that we gave the Kos what they wanted, Kos himself stated on his Tuesday blog that it would NOT vindicate us. We already knew that and as I stated the reasons will come out during litigation. As I said, his account of the data and the Kinkos story is slander and will be proven. It is also important to know I was never warned about a report, what Kos said in an email to me is that the last thing he needs is for someone showing a report that says his data is flawed the past two years. It is also important to know, that Kos himself stated in his slander on us that he was going to begin slandering us on Friday, but he waited until the next week to get better coverage out of it. We did not receive this report until 2 AM in the morning this past Monday! On the data is too clean crap, let me say this and I challenge anyone to then look at comparable data from other firms, not one or two but many others. As I stated, using Gallup one could question the frequency of 46% on Obama's approval. Regardless though. to you so-called polling experts, each sub grouping, gender, race, party ID, etc must equal the top line number or come pretty darn close. Yes we weight heavily and I will, using te margin of error adjust the top line and when adjusted under my discretion as both a pollster and social scientist, therefore all sub groups must be adjusted as well. I would have gladly gone over with Kos before his accusation in a vile email on June 9. However, it is clear that no matter what, Kos was going to go the route they have not just to get out of paying their bill but as stated for several other sinister reasons that have come to light.
Let's also be honest here and state that many in this witch hunt are in the tank for the Kos. If they were truly fair and objective, they would allow this matter to go through the legal process. Most importantly, what would they all report if Kos did receive what they wanted? I can answer that, they would two things, first they would find a biased statistician that would find a subjective flaw in the data and two, even if the data met the satisfaction of all, our vindication would be on page 7 in a small blurb in the Metro section and Kos would not even receive a slap on the wrist from any of them. Most importantly, our reputation will never be repaired even after we are vindicated, so in that sense Kos has already been successful
This is why this will be my only response on this matter until after we are vindicated. No more inquiries or questions will be answered until after this is settled.