The Mark Kirk Senate campaign, hiding behind the NRSC, sent this out:
*Executive Summary**:* The ultra left-wing Web site DailyKos commissioned a poll by Research 2000, which was conducted in Illinois Feb. 22-24, 2010. Just as it did before the 2008 Congressional election, DailyKos commissioned a skewed survey to create a false impression of the race for Illinois Senate. First, the survey over-sampled voters in Central Illinois and the City of Chicago, and under-sampled voters in the Collar Counties surrounding Chicago. Second, the survey over-sampled Democrats and under-sampled Republicans.
Skewed Regional Sample*: ** *According to the database of actual registered voters, 23% of voters live in the Collar Counties while 14% of voters live in Central Illinois. The DailyKos/Research 2000 poll filled sample quotas differently, showing 19% of those surveyed living in the Collar Counties and 23% living in Central Illinois. The Collar Counties remain Kirk’s greatest regional strength while he remains less well-known in Central Illinois. In addition, the survey makes no distinction in its Cook County sample between suburban Cook County (22% of total votes and stronger Kirk) and the City of Chicago (16% of total voters and stronger Giannoulias) – thus raising questions of over-sampling Democrats in the city. Skewing the sample away from Kirk’s stronger regions will obviously skew the poll results toward Giannoulias.
* *
Skewed Party Sample*: * Throughout previous surveys in Illinois, we found on average that the state breaks down as follows: 36% Democrat, 30% Republican and the rest Independent. The DailyKos/Research 2000 poll filled sample quotas differently, showing 44% Democrats, 27% Independent and only 29% Republicans. Surveying more Democrats and less Republicans will obviously create an artificially different ballot test.
Oooh, "ultra left wing"!
Such silliness. Only an idiot would equate "registered voters" with "actual voters". Otherwise, you could all the election today based on voter registration numbers. And yes, Democrats would clean house, even in the South. Unfortunately for our team, that's not how it works.
Here, I'll show you how this works in the real world, looking at the exit polls for the last two elections:
In other words, actual vote results the last two election cycles have zero similarity to the numbers that the Kirk campaign want people to use. In fact, they are conveniently demanding a full 10 points less Democrats than we've seen in two straight elections. That would certainly skew things in their direction.
Now polling is an educated guess, since no one really knows who will turn out in November. But R2K is certainly basing its numbers on historical trends and not on irrelevant voter registration statistics. (And as a reminder, R2K, an independent pollster used by traditional media outlets, makes all decisions about sample compositions. We have no say in the matter.)
But don't worry, Mark. Rasmussen shouldn't be long, with a truly skewed poll to bail you out.