Earlier in 2014, the Michigan Senate race was looking like it'd be a big source of trouble for the Democrats ... while at the same time, Democrats in less-blue states like Iowa and Alaska were holding their own just fine. The Democratic nominee, Rep. Gary Peters, was only well-known in the Detroit area, not statewide, while the Republican nominee, former Secretary of State Terri Lynn Land, had been elected statewide before, was fundraising well, and had enough family wealth to self-fund more if need be. In the months January through April in 2014, Land led
7 of the 11 polls released in this race.
Fast forward to today, where Iowa and Alaska are looking much worse, but the race in Michigan has gotten to the point where it's barely a competitive race any more. It might be nice to assume that Michiganders revolted against Land's unspeakably bad ads, but more likely they were affected by her unsteady public performances during the rollout phase, followed by her pretty much disappearing from sight and just letting her ads to the talking. More than anything, though, it may have simply been a matter of the bluish lean of the state finally taking hold, as Dem-leaning undecideds came off the fence and backed Peters.
Prior to this weekend, there had been 14 polls of the Michigan Senate race during the months of September and October, and Land led in a grand total of one of those polls. Peters led in every single other one, with margins as high as 13 in a Mitchell Research poll released last week. The one poll to give Land a lead during these months was YouGov, who gave Land a 1-point lead in early September ... and now that's gone too, as YouGov's return to the state this weekend found Peters leading too, 46-41. That (plus the Mitchell poll, and an EPIC-MRA poll putting him up by 9) pushed Gary Peters' odds of winning his race from 89 percent during our last look at the Election Outlook last Thursday, up to 95 percent. For perspective, that's now the same likelihood of victory as in the Minnesota Senate race, where virtually all pundits have regarded Al Franken as safe.
That Michigan poll was only the tip of the iceberg for YouGov. For many people, Sunday is the day to gorge yourself on NFL games ... but this Sunday, for people of a geekier persuasion, it was a day to gorge yourself on polls. YouGov, on behalf of the New York Times, released polls of every single state. You might think that with that many polls in the pipeline, though, that something dramatic would happen. However, no games were changed. Along with Michigan, Democratic odds improved somewhat only in Colorado and Iowa, but they didn't change so significantly as to boost the Dems' overall odds of retaining control of the Senate (which fell ever so slightly since last Thursday, from 34 percent to 33 percent, with the median outcome still being 48 Democratic seats).
YouGov's Colorado poll showed Dem incumbent Mark Udall leading GOP challenger Cory Gardner by a 48-45 margin. While that's in line with most of the polling of the Colorado race between March and early September, it's a good bit different from the most recent few polls; there were five polls in the last few weeks giving a small lead to Gardner. YouGov's previous Colorado poll, which also put Udall up by 3, came out slightly before the more recent spate of bad polls (where an 8-point Gardner lead particularly stood out, in Quinnipiac's most recent poll).
YouGov's near-identical results, bookending a month of bad polls for the Dems, leave you wondering whether Udall somehow faltered and then recovered, or the race never really changed and perceptions just got shaped by a coincidental string of inaccurate polls. At any rate, the new poll is enough to boost Udall's chances from 38 percent to back above water at 51 percent; the smoothed trendlines now point to Udall beating Gardner 50.1-49.9.
We'll continue our discussion over the fold:
Things also got slightly better in the pivotal race in Iowa, where YouGov found Democrat Bruce Braley leading GOPer Joni Ernst 44-43, the same small margin found by several Democratic pollsters last week. That was evened out, though, by a Marist/NBC poll that also came out on Sunday, which put Ernst ahead 46-44. Cumulatively, though, that made things a little less worse for Braley, and his odds of winning ticked up from 27 to 32 percent. Even taking the improvement in Colorado and Iowa into account, though, it doesn't matter, because the Republicans still have a clear path to 51 without either of those races or North Carolina: Just pick up the 'gimmee' states of Montana, South Dakota, and West Virginia, and then pick off three Dem incumbents, like Mark Begich, Mark Pryor, and Mary Landrieu, all of whom have trailed consistently in recent weeks. That's why the overall odds only barely changed.
Of course there's still one missing link in the no-Colorado-or-Iowa-needed plan, and that's getting Greg Orman in Kansas to caucus with the GOP. While that most recent round of YouGov polls found Orman and Pat Roberts tied, a Marist/NBC poll also released on Sunday gave Orman a 10-point lead over Roberts (48-38). With those kinds of numbers, our model now gives Orman a 90 percent chance of winning. However, in the majority of those scenarios, the Republicans also pick up control of the Senate, making it that much likelier that Orman (who has said he'd caucus with the party with a clear majority) caucuses with the GOP. That's why even though Orman has a strong chance of winning, there's still only a low chance (33 percent) of a "Democratic" "victory" in Kansas.
Finally, let's turn to the gubernatorial races. YouGov's commitment to cover every state means that they've also gotten results in the many states electing a governor this year. While we saw those numbers on Sunday (and many of the gubernatorial numbers provide good news for Democrats, at least by comparison to the Senate polls), these gubernatorial numbers haven't been formally released yet, so we haven't included them in our polling database or in the model for today. Once they're formally released in the coming days, they'll be added to the model.
Nevertheless, the few non-YouGov polls released in gubernatorial races in the last half-week turn out pretty well for the Democrats, whose chances of gaining gubernatorial seats has increased from 55 percent to 58 percent (with 22 seats the most common scenario, which would be a gain of 1 seat from now).
That's partly thanks to a Gravis poll in Kansas, where Democrat Paul Davis led GOP incumbent Sam Brownback by 8, increasing Dem odds there from 72 to 83 percent. It's also thanks to a continuing mini-surge for Charlie Crist in Florida, who's now up from 54 percent odds to 65 percent odds, based on a small Crist lead in a poll from Republican pollster 0ptimus. And it's thanks to a Rasmussen Reports poll that saw Dem incumbent John Hickenlooper with a decent 4-point lead in Colorado after trailing in several recent polls; that moved Hickenlooper back above water, from 41 to 60 percent.
The only downward move for a Democratic gubernatorial candidate was Mary Burke in Wisconsin, who fell from 54 percent odds to 33 percent odds. That's against the backdrop of not a new poll, but rather the deletion of an old poll. Gravis — already very high on our list of very sketchy pollsters — retroactively yanked a poll from last week that seemed too good to be true (or too bad to be true, for them, seeing as how they're partisan Republican pollsters), putting Burke up 5, offering no further explanation beyond "a skew in Milwaukee respondents." The disappearance of that poll means that Scott Walker has led in 4 of the last 5 polls in Wisconsin.