Kansas and incumbency advantages in our model

I said last time that we have no clue what's going on in Kansas, and it just got harder to tell, since yesterday the Kansas Supreme Court ruled that Democrat Chad Taylor could have his name removed from the ballot. With Taylor presumably out of the picture (to be replaced by "random Democrat", if Secretary of State Kris Kobach's next order stands), the only polls that make sense now are the ones that ask about a matchup between independent Greg Orman and Republican Sen. Pat Roberts.

 

Unfortunately, that doesn't make matters any simpler. To begin with, very few polls asked about what would happen if it were just Orman vs. Roberts; in fact, the YouGov / New York Times / CBS poll didn't even include Orman in their question, as it was conducted when Taylor was still in the race. 

Then there's the question of Libertarian candidate Randall Batson. No poll has asked about an Orman-Roberts-Batson matchup only; whenever Batson was included, the pollsters also asked about Taylor (because why would you include the Libertarian if you weren't going to include the Democrat, who would presumably garner more of the vote?). And usually I wouldn't bat an eye at that. But in our model, which uses an estimate of what percentage of "undecided" or "other" voters in a poll will actually end up voting for a third-party candidate, Batson's absence is important because there's now only one major-party candidate on the ballot, and Orman, despite running for the Democratic nomination in 2008, isn't interchangeable with a Democrat. As a hypothetical Kansas Libertarian-leaning Democrat, I might think the spoiler effect would be greatly attenuated if there isn't a Democrat whose votes are being spoiled. Batson's vote share could greatly increase in the month and a half before Election Day. 

 

But for now, all we've got are Orman-Roberts matchups. Two of them. In combination with the fact that we have no polls with just an Orman-Roberts-Batson race, there's a lot of uncertainty--not just in whether Orman's 7-point lead in our average is accurate, but also in the number of Batson voters who actually will end up voting for Batson. Do note, as well, that our model still lends quite a bit of power to political experience, especially multiple-term incumbency. Thus, when Taylor's name had to remain on the ballot and the polling was close, we still gave a substantial edge to Sen. Roberts despite Orman remaining narrowly ahead.

 

Not anymore. While we expect the majority of undecided voters to break for Sen. Roberts, Orman's lead in the polls can't be ignored as it stands. It's a crazy race, but one thing we're fairly sure of is that Pat Roberts is no longer leading. This race we're now labeling a TOSS-UP.