Cross-posted from
MN Campaign Report
Every time a new poll comes out, the big-race campaigns spin it to the best of their ability - "this poll shows a rapidly tightening race," and so on. This spin, however, is only that - spin. Each pollster uses a different methodology - interactive online, automated phone polling, phone interview, random digit-dialing, etc - and thus each pollster is going to come up with slightly different results.
If you read MNCR on a regular basis, you've seen the poll graphs. Rasmussen and Zogby have polled the Senate and Gubernatorial races several times each now. What if we remove all other polls and pollsters and focus on a single pollster/methodology dyad? Interesting results.
In an effort to keep down the file size on this image, the actual numbers aren't here - you can refer to my most recent Poll Update post for those. (Sorry Kossacks - I really should get a photobucket account. Check the image though, it's telling). But there are some interesting trends here. Zogby's Senate numbers haven't moved a whole lot - the Klobuchar campaign's repeated pronouncements of an eight- to nine-point margin appear to be dead on with Zogby's results. Both Klobuchar's and Kennedy's numbers have risen and fallen, but Zogby's margin has stayed remarkably consistent.
On the gubernatorial side, while Zogby shows Pawlenty dropping from and rising to a deadlock over time, the split has never been outside the margin of error. These results fit nicely with conventional wisdom on these races.
Compare these smooth lines to Rasmussen's results. Although the general results are the same (Klobuchar strong, Kennedy weak, Hatch and Pawlenty locked in a cage match), their numbers have really been all over the place. It's difficult to believe that Rasmussen's methodology isn't introducing some kind of error, since in the absence of some world-rocking event, big swings like these don't usually happen. Perhaps Zogby really is on to something with their new-fangled interactive online polling. Or perhaps Rasmussen simply jumps around within that margin of error more. We'll see in twenty-two days which has a stronger correlation to the final results.
As any political professional or junkie will tell you, a poll is merely a snapshot in time. Poll lines like those you see from time to time on MNCR average all pollsters and methodologies together, and give you a good general idea of where the race is. But don't believe any campaign that tells you that one poll represents a "rapidly tightening race." It really is just spin.