As we discussed on
Monday, the overall odds are somewhat against the Democrats being able to retain control of the Senate, but there are a number of credible pathways for them to get there. The most frequently-occurring pathways (when we run thousands of simulations) go through only a few races, though; they involve winning a mix-and-match assortment of three of four iffy races: Alaska, Colorado, Iowa, and Kansas.
Here's why any of those might happen:
Colorado: Polls in the last few cycles have tended to undersell Democratic odds in Colorado. The large majority of Senate polls in 2010 predicted that Republican Ken Buck would win (which didn't happen), and polls predicted a much narrower Barack Obama victory in 2012 than actually happened. That's probably because Colorado has larger shares of Hispanic and younger voters, who tend to be harder to reach (especially by pollsters who don't call cellphones) and who may not fit into a stringent likely-voter model. When pollsters take the time to make sure they're sampling the Latino population correctly — as a Wednesday poll from Strategies 360 did — they tend to find Mark Udall with a narrow lead.
Iowa: The polls seem likelier to be right in Iowa — a much whiter and older state, and one with a stable population instead of a lot of growth — than they are in Colorado. However, Iowa is the state with the most consistently close polling; most polls in the last few weeks have seen this as a one- or two-point race. Unfortunately, most of those one- or two-point leads have gone in Joni Ernst's favor, but a Wednesday poll by Loras College put Bruce Braley up by one. Good ground game isn't a magic bullet but can add a point or two to your total, so Dem GOTV efforts could get Braley narrowly over the top.
Kansas: Independent candidate Greg Orman seems likelier than not to win, although his lead is fairly small. The real question mark associated with Orman is whether he caucuses with the Democrats; he's said he would caucus with whoever is in the majority, but hasn't said what he'd do in a situation involving 49 Dems and 50 GOPers where the majority is entirely up to him. Considering what a beating he's taken from the NRSC lately, though, you've got to wonder how excited he'd be about breaking bread with them afterwards.
That leaves Alaska, which a week ago was looking the least promising of those four. Mark Begich had trailed in all nine polls from the previous month, by margins ranging from two to six. Then, odd stuff started happening. A poll by Hellenthal, a local pollster, over last weekend put Begich up 10; on its own, it screamed 'outlier,' but the pollster swore that Begich's unprecedented ground game efforts were truly changing the game. Then, on Monday, it stopped being an outlier; another poll by another local pollster, Ivan Moore, found Begich up 8.
Begich's opponent, Dan Sullivan, released an internal poll on Tuesday showing Sullivan leading by 4 to try to rebut those polls, but even that was quite a tell: it's an internal poll with a smaller margin for him than most of the public polls gave him in the preceding month. For purposes of our model, that pencils out to only a one-point Sullivan lead.
That isn't to say that we're confident that Begich has pulled into the lead here. For one thing, that's still not a lot of polls showing a lead; Alaska still gets polled a lot less than the other crucial Senate states. There were only two other Alaska polls in the last week: a Dem internal poll (on behalf of Senate Majority PAC) showing the race tied (showing a best-case scenario that's worse than the public pollsters), and a YouGov poll giving Sullivan a four-point lead. (Bear in mind, though, that the YouGov poll has a tiny Alaska sample, with a margin of error of 9 percent.) One other data point that might cool your enthusiasm: Harry Enten and Nate Silver found that Alaska polling tends to have a higher error rate than most other states, and — also unlike most other states, like Colorado above — those errors usually give an errant boost to the Democratic candidates.
We'll look other changes in the model this week, over the fold:
Despite the much improved situation in Alaska, the Democrats' overall odds in the
Senate actually fell a small amount since our last look at the Outlook model on Monday; odds of Democratic control fell from 34 percent back to 31 percent, though 48 remains the median number of Dem-held seats, as has been the case almost every day for the last month. For starters, Mark Begich's individual odds in Alaska still haven't gotten to the point where he's above water, rising from 22 percent last Thursday to 35 percent on Monday (taking the Hellenthal poll into account) to 44 percent now (also including the Ivan Moore poll). The trendlines still have to take into account the two internal polls, the YouGov poll, and all the older polls.
In addition, Democratic odds fell similar amounts in two other races. In Lousiana, Mary Landrieu's odds fell from 29 percent to 18 percent, mostly due to a new Suffolk poll that had her trailing Bill Cassidy by 7 points in a runoff. Also, while Jeanne Shaheen continues to have a distinct advantage overall, her odds fell from 73 percent to 65 percent following a New England College poll that put her one point behind Scott Brown. (It should be noted that NEC is the only public poll that has ever found her trailing, something they've done three times in a row now.)
On the gubernatorial side, there was a similarly small improvement in Democratic odds; the odds of Democrats gaining seats rose from 55 percent to 57 percent, with the median number of seats staying at 22 (which would be a one-seat improvement over current conditions). The biggest mover here was Illinois, where after falling slightly below water last week, Dem incumbent Pat Quinn rose back up from 48 percent odds to 75 percent odds. That's thanks to a Chicago Sun-Times poll and a YouGov poll, which put Quinn up 3 and 4 points, respectively.
Maine and Massachusetts also saw some improvement, bouncing back from unusually bad (and probably outlying) polls last week. In the Maine race, Dem odds moved from 24 percent to 48 percent, with three new polls (a tied race according to Pan Atlantic, a 2-point Dem lead according to YouGov, and a tied race according to a GOP internal from Magellan, which pencils out to a Dem advantage for purposes of the model) counteracting the older UNH poll that found Mike Michaud trailing by 10. And in Massachusetts, Martha Coakley's odds went from 35 percent to 45 percent, with a 4-point lead according to YouGov and a 1-point deficit according to MassInc, counteracting last week's 8-point deficit in a Boston Globe poll.
Finally, the worst polling news of the week was probably on Wednesday in Wisconsin's gubernatorial race, where Marquette Law's final poll of the cycle saw Scott Walker opening up a 7-point lead (after they found a tied race two weeks ago), which is actually the largest Walker lead anyone has seen since April. That pushed Dem odds in Wisconsin from 43 percent down to 24 percent.