If you suddenly awoke, Rip Van Winkle-style, from a multi-year sleep in the middle of a presidential year, and you found out that Kansas was about to vote for the Democratic candidate and Massachusetts was about to vote for the Republican candidate, you might think the world had gone mad. Even at the Senate level, it's improbable (though the mercifully-short reign of Scott Brown reminds us that anything is possible). At the gubernatorial level, though, where there's a largely different set of issues, all bets are off.
Also, recent history tells us it's not unusual; less than a decade ago, Kansas and Massachusetts were led by Kathleen Sebelius and Mitt Romney, respectively. Well, history may have a chance of repeating itself this year: Democratic challenger Paul Davis has led Republican incumbent Sam Brownback in most Kansas polls all year. And now all of a sudden, Massachusetts Democratic nominee Martha Coakley is running only narrowly ahead of, and occasionally behind, Republican nominee Charlie Baker. (Four polls of this race have been released this week; Coakley led by 3 and 1 in two of them, while Baker led by 1 in the other two.)
In the space since our last Election Outlook update on Monday, Davis's odds of winning have risen from 69 percent to 72 percent, while Coakley's odds of winning have fallen from 84 percent to 69 percent (after starting at 97 percent odds) ... meaning Davis has (ever so slightly) better odds of winning.
Unlike her negligent 2010 loss to Brown, however, Coakley doesn't seem to be doing anything especially wrong this time, other than simply being a not-so-personable technocrat. Instead, part of what's happening is that Coakley got roughed up in her primary — she usually led by at least high single digits, if not better, in general election polls prior to the primary, but it seems like some of rival Steven Grossman's supporters have gravitated toward Baker since then — and also she's at a distinct financial disadvantage to Baker.
We'll talk about the bigger gubernatorial picture, as well as the Senate map, over the fold:
Overall, Democratic odds of gaining gubernatorial seats improved despite Coakley's slide; the odds of Dems gaining governorships went up from 52 percent on Monday to 55 percent now. (The median number of seats holds steady at 22, a small improvement over the current 21.) More so than Kansas, the gain is pushed along by Alaska (where Bill Walker's odds increased from 51 percent to 64 percent, thanks to a 5-point lead in a new Rasmussen Reports poll) and Wisconsin (where Mary Burke's odds increased from 44 percent to 54 percent, thanks to a 5-point lead in a Gravis poll that more than canceled out a 5-point Scott Walker lead in Wednesday's Marquette Univ. poll). Most notably, though, may be Florida, where Charlie Crist is actually above water for the first time since we started running the model; thanks mostly to a 6-point lead in a new SurveyUSA poll, Crist's odds shot up from 38 percent to 54 percent.
And here's one final example of what a topsy-turvy year it is in some places; the Democrats now also have better odds in the Oklahoma gubernatorial race than they do in Ohio. Granted, the odds still aren't very good, but in the wake of Democratic candidate Joe Dorman unveiling an internal poll showing him down just 2 points to GOP incumbent Mary Fallin, Dorman's odds improved from less than 1 percent to 5 percent, giving him much better odds than Democrats in much swingier states like not just Ohio but also New Mexico, Nevada, and Iowa.
The Senate side of the equation is a bit less fun to talk about, partly because the Democrats continue to be on track to lose control of the Senate but also because it's settled down into a much more predictable pattern than the constantly-surprising gubernatorial map. There are now absolutely no Senate races where the Democratic candidate has odds between 40 percent and 80 percent; everything is either better than that (North Carolina, Michigan) or worse (Arkansas, Louisiana, and even Alaska, Colorado, and Iowa).
The good news is that Democratic odds of controlling the Senate ticked up a tiny amount since Monday's all-time low, from 33 percent to 34 percent. The median number of Dem-held seats continues to be 48, while the modal number (i.e. the one that occurs most frequently in all the simulations that we run) is down to 47, which would result in an 8-seat loss. Very little has happened this week to change the numbers; Mark Udall's odds in Colorado improved slightly, from 33 percent to 38 percent, thanks to a Rasmussen Reports poll which had Udall down by only 1 point. That was counteracted, though, by a decline in Bruce Braley's odds in Iowa from 42 percent to 27 percent, thanks largely to this weekend's Selzer poll that put him down by 6 points against Joni Ernst.
The big mover on the Senate map in this period, in a bit of a surprise, is Kentucky. That seems largely due to the internal poll from the Mellman Group that the Allison Lundergan Grimes campaign released on Wednesday, showing a 2-point lead over Republican incumbent Mitch McConnell. Even though that gets downweighted in the model for being a partisan poll, that's still enough to boost Grimes's odds from 17 percent to 29 percent. That means, temporarily at least, that Grimes is better-positioned than red-state incumbents Mark Begich, Mary Landrieu, or Mark Pryor.
One thing the model can't take in to account, though, is how far back from 50 Grimes still is. If the poll gave her a 49-47 lead, that'd be one thing, because she'd be inches from her target ... but the poll has her leading 42-40, leaving you to wonder where she scrapes up another 8 percent from a pool of undecideds that's likely to break against her. (There's no discussion of that in this poll, but previous PPP polls have shown the undecideds in Kentucky heavily tilted toward registered Republicans and/or 2012 Mitt Romney voters.) So, that object in McConnell's mirror may appear larger than it actually is.
Nevertheless, the small surge by Grimes, combined with the faltering overall Dem odds, increases the likelihood of a very unusual possibility, from the almost-impossible to the somewhat-plausible. What are the odds of the Republicans taking control of the Senate, but Mitch McConnell losing nonetheless? According to the model, higher than you'd think: It now happens in 8.4 percent of all simulations. While losing control of the Senate would no doubt suck, the sheer popcorn-chewing satisfaction of watching the Senate session open with an all-out battle royale for the majority leader slot between John Cornyn and some member of the wingnut brigade (Ted Cruz? Mike Lee?) would take away some of the sting.