The liberal Daily Kos website got defrauded by the pollster they’ve used for quite a long time. Many people, especially on the Right, have been skeptical over their consistently liberal-favoring numbers, but I could not have imagined the scope of the fraud. From my somewhat mathematical, if not statistical, background it looks like the numbers have been either manually adjusted or completely made up without regard for good “fake” statistically possible results.
To Kos’s credit he lays the case out openly on the front page of his site – Research 2000: Problems in plain sight – where he explains the statistical incongruities, and More on Research 2000 with more information.
It is an interesting read. He will be suing the firm. Should he have looked at the results earlier based on their liberal bias? I am not sure, as any poll can be made to skew to a particular side depending on methodology. That does not make a poll invalid, just biased.
Considering the facts, I think he handled it well, and though I despise the liberal ideology and most of the stuff published on that site, I think the way he did this should be recognized.