The Illinois Senate race keeps going back and forth. Republican Mark Kirk led a while, then Democrat Alexi Giannoulias took it back, but now having won three consecutive polls including PPP's latest, it seems that Kirk is definitely on top again.
It's so close though that the polling of third party candidates is a serious issue. It may not matter in the end, though.
So yes, technically we do have a four way top line here: Kirk (R) 40, Giannoulias (D) 36, LeAlan Jones (G) 8, and Mike Labno (L) 3 (MoE 4.5). That sends Kirk all the way up to a 67% lead probability in my estimate, per this poll alone.
But wait: Do we really think Jones and Labno are capable of going that high? According to Wikipedia (yes, call me lazy, but I wanted the numbers quickly) in 2004 against Barack Obama and Alan Keyes there was no Green but the Libertarian got 1.3 and an Independent got 1.6. In 2008 against Dick Durbin and Steve Sauerberg the Green got 2.2, the Libertarian got 0.9, and the Constitution party candidate got 0.5.
8 and 3 would seem to be far too high for the Green and Libertarian this year, but the plain fact is that Kirk and Giannoulias are not popular candidates. I can see both third party candidates overperforming the trend. Just not this high. 3-4 seems right to me for Jones, and 1-1.5 would be more like it for Lanbo. If we then crudely re-assign those points from the Green to Giannoulias, and from the Libertarian to Kirk, we still show Kirk on top, albeit with a smaller lead.
So is it fair to say that Kirk actually leads? I think so, because it's the very best case for Giannoulias to get the entire share of the Green vote that changes at the last minute. A fair number of those people might stay home if they feel depressed enough about the blowout Governor's race, and what they perceive to be two dissatisfying choices for Senate.
So for now, Kirk seems to be back on top, though far from a range that would seem to avoid recounts or accusations of fraud. I'm not looking forward to lawsuits, but we just might see them here and in Nevada, the two states I'm predicting to be closest in the final count.