Saturday, July 3, 2010

Was Markos Made the Fool?

It's hard to say what happened with Markos and the Research 2000 polls he commissioned for over 18 months.

He's filed suit and I understand the pollster is counter-suing. Some statistics geeks have had at the R2K formulae and declared them suspect, but anyone who was watching the results being headlined daily at Daily KOS had to be thinking at the very least that Markos was getting what he paid for, poll results that pleased his fancy.

I go back a long way with dKos. My user ID is 501; it would be lower but for the fact that my computer at the time was acting up and it wouldn't let me sign up to the Scoop site when the changeover happened, and I had to wait till the next day -- as I recall. It was a long time ago. At any rate, I was using a different screen name at the time, one that I still maintain for certain very limited purposes. I was active at dKos almost from the beginning of its emergence, made what I thought of as a number of online friends there, but most of them, like me, have since departed the place. In my case, I cut back commenting to nearly nothing after Markos pretty much turned the site over to the notoriously abusive Armando Llorens (aka "Big Tent Democrat") who was simply allowed to run wild for a couple of years -- until I guess he had to go into rehab. He eventually managed to get his ass banned by Markos, but only after an enormous number of complaints about his atrocious behavior had been sent to Markos. So far as anybody knew, he ignored them all. Armando was his man. And nobody could say anything against him.

Very early on, Markos declared himself a "poll junkie," and he showed an almost child-like belief in the "magic" of polls. I was puzzled. Markos clearly didn't know anything about polls or polling, and he would often misuse the results of this or that poll to reinforce whatever political belief or candidate he was promoting at the moment, even if the polls actually showed something quite different than what he believed. He would routinely tout polls that reinforced his belief, denounce polls that showed something else, and he seemed to be completely ignorant of how polling was done, how the raw data was massaged or manipulated, let alone how questions could control outcome, and how really slimy some pollsters were.

It was all "magic" to him. And he seemed to love the notion that polls could accurately predict the future. Of course, they often didn't, but that was a mere detail.

When he hooked up with Research 2000, I thought maybe he had learned something what with all his heady talk of "internals" and "crosstabs" and such. But then, the results he touted from R2K seemed to be intended to promote Obama and Democrats when there was no other indication that the public was quite so enamored.

He was getting what he was paying for, much like the clients of other polling outfits got what they were paying for. A tool with which to market political product.

But that's not to say the R2K polls were "fraudulent" as Markos is now claiming they were. My suspicion was they that they were designed to produce the results they did, much like other polls are, and that they were valid enough within their strict boundaries.

When Markos dropped R2K, the rationale was that the polling wasn't sufficiently close to voting results, particularly in the Halter/Lincoln Primary Thing in Arkansas. R2K's polling was off, but that's not unusual in the polling business. What seemed to make Markos wake up to the idea that something was out of whack with R2K was the fact that Nate Silver declared R2K's results to be essentially an outlier, not just on the Halter Thing but through and through, and thus unreliable.

The bias and unreliability of polls in general has long been a feature of the trade, however, not a bug. What was so special about R2K's bias and failings? That is still not clear, although statistics geeks have come up with plenty of charts and graphs that suggest that R2K may not have been doing any actual polling at all but might have been simply pulling numbers out of the air.

If that is so, wouldn't it be amazing? Actually, no. Not necessarily. Since they were consistently producing results Markos wanted to see -- much as other polling outfits do for their clients -- how they got those results really doesn't matter.

Polling in general really isn't reputable any more, if it ever was. "There are lies, damned lies, and statistics," isn't just a catch-phrase. Markos' child-like faith in polls and polling, in contrast to his general distrust and skepticism of nearly everything else, seemed odd years ago, and this episode has a distinctly odd aroma to it.

There are apparently those who are developing an elaborate conspiracy theory about Markos and R2K. That it was all a scam from the outset, and Markos was in on it, and he would still be using R2K to scam the public if it weren't for the "statistical anomalies" that were being pointed out far and wide. Under the circumstances, he has to make a big show of denouncing R2K and suing their ass; otherwise Markos's reputation is in the toilet, too.

Well, I think that goes too far. Way too far. Manipulated polls are a fact of life, have been for a very long time. So has using manipulated polls to market political products. Anyone who puts their faith in polls and polling because it supports their political ideology is, in my opinion, a fool. Rather than scamming, it is more likely that Markos was merely fooled by polls that showed what he wanted to see.

On the other hand, R2K may be a complete fraud.

No comments:

Post a Comment