We've already gone through two previous game-changing moments in the Kansas Senate race this month; first, Democratic candidate Chad Taylor dropped out, clearing the way for the better-financed independent Greg Orman to have a clear shot at beating Republican incumbent Pat Roberts, so we had to switch over to only the handful of polls that had looked at a head-to-head Orman vs. Roberts race. Then Republican Secretary of State Kris Kobach ruled that Taylor didn't follow proper dropout procedures and would stay on the ballot, at which point we had to switch back to the few polls that had Roberts, Orman, and a disclaimer that Taylor wasn't running but still allowing respondents to pick him.
Last Thursday, though, the Kansas Supreme Court ruled that Taylor's name should be removed from the ballot. Republican Secretary of State Kris Kobach is insisting that the Democrats name someone else to replace Taylor on the ballot, but he's also out of time; federal law requires him to send out ballots very soon to reach overseas voters, so at least the first round of ballots will be printed without any Democrat on the ballot.
The Supreme Court didn't even reach the question of whether any Democrat should be on the ballot, only ruling that Taylor himself shouldn't be. A random Democrat (whose son works for Sam Brownback) is suing to get the Democrats to name a replacement candidate, so it's possible we'll hear from the Kansas Supreme Court again ... meaning it's entirely possible that we may be writing another post in a few more weeks about how the model is changing to accommodate an entirely new Democratic candidate!
Considering that the ballots are about to ship, though, we've had to switch back to using the polls of only an Orman vs. Roberts race. Three more polls of that matchup came out last week: a PPP poll showing Orman up 46-36, a Fox News poll showing Orman up 48-42, and a Rasmussen Reports poll showing Orman up 45-40. That combination of polls gives Orman a very strong chance of winning: 94 percent. Unfortunately, though, the odds of a "Democratic" victory in Kansas haven't improved at all, going from 42 percent last Thursday to 40 percent now.
That's because Orman has stated he'll caucus with whichever party already has the majority in the Senate. (He hasn't stated what he'll do if it's a 49 D/50 R split and he gets to decide control of the Senate by himself, so we assign him 50-50 odds of going each way.) So even though, at this point, Orman is very likely to win, the odds of the Democrats winning at least 49 other seats in the Senate have declined noticeably. There are two conditions necessary for Kansas to be a "Democratic" victory (Orman has to win, and the Democrats have to get at least 49 seats other than Orman), and while the odds of the first condition are as strong as ever, the odds for the second condition are lower.
We'll discuss why that's the case over the fold, along with the gubernatorial races:
When the Democratic odds in the Senate were at their highest, a mere week ago, it wasn't because the Democratic candidates were doing better, across the boards, in every race. If you
compare the chart then with now, you'll notice that only three races have changed significantly: Colorado (where odds of a Democratic victory were at 78 percent), Iowa (60 percent), and Louisiana (and 49 percent).
Today, odds in Colorado are down to 37 percent, odds in Iowa are 39 percent, and in Louisiana it's 16 percent. In each of those cases, though, that drop was caused by polls that don't really meet the smell test. We discussed last week that the Iowa drop was caused by a Quinnipiac poll with a small Joni Ernst lead, while the Louisiana drop was caused by a Fox News poll with a huge Bill Cassidy lead.
And now, we've added that Quinnipiac poll in Colorado that gave Republican Cory Gardner an improbable 48-40 lead. (Two partisan polls with smaller sample sizes have come out since that one: a poll from Gravis, a Republican pollster, with Gardner ahead, and a Democratic internal poll from Keating Research. Those two essentially cancel each other out, though, leaving Quinnipiac to have an outsized impact on the Colorado trendlines.)
We suspect that once we get some fresh polls in Colorado and Iowa, those will move back toward the norm. In the meantime, though, the downturn in those two races make an outcome of 48 or 49 Democratic seats, rather than 50 or 51, likelier. As you can see if you click through to the Election Outlook permanent page, 48 is both the median number of Democratic-held seats, and if you look at the histogram, the modal number as well. (Overall odds on the Democrats retaining control are currently 41 percent.)
There's still a pretty robust path to 49, though, which means that the odds of Orman being the decisive vote in the Senate are larger than ever. In 15 percent of all simulations, the result is 49 D, 50 R, and then an Orman victory. That's a very unusual circumstance, where we're going into an election without a clear picture of which party is likely to control the Senate, but also without a clear picture of what party a candidate who's in potentially the decisive race will caucus with.
Finally, let's turn to the gubernatorial side. It's been an unusually quiet weekend in terms of gubernatorial polls, so the numbers have hardly changed in the last half a week; the median and modal number of Democratic-controlled state houses is 22, and the overall odds of the Democrats gaining gubernatorial seats in 57 percent.
The only big mover on the gubernatorial side is Rhode Island, in the woefully-underpolled race between Gina Raimondo and Allan Fung (a general election permutation that only one pollster tried out in the pre-primary polls). Raimondo's odds improved from 62 percent to 72 percent, thanks to a doubling in the number of polls of this race (from one, to two, with the previous one nearly a year old). And oddly, the boost in Democratic odds is thanks to a Republican internal poll, which showed a 42-42 tie. Once we've corrected for the fact that it's a partisan poll, that works out to a small Raimondo advantage.