If you were only looking at Senate models — which is all that all the other 'competing' models look at — you'd rightly be concerned that the Democrats are on track for a terrible night on Election Day. All the models (even
Sam Wang's slightly-more optimistic model) currently predict that the Republicans are on track to control 51 or more Senate seats, enough for a majority.
Our model, for instance, currently predicts the Republicans are on track to hold 52 seats, which would be a 7-seat gain from their current 45; the Democrats retain control in only 34 out of 100 of our simulations. If that holds, that'd be even worse than
2010, where the Dems lost 6 Senate seats on Election Day (although the damage hits 7 if you throw the Massachusetts special election of early 2010 in there). In other words: #DEMSINDISARRAY!!!1!
Hold on there, though; there's a lot more to an election other than the bragging rights associated with whether it's Harry Reid or Mitch McConnell who gets the task of filibustering everything that moves in the Senate. For starters, there's also the matter of the gubernatorial races, which, as I mentioned, we also model, and nobody else does. Democratic gubernatorial losses also hit a net of 6 in 2010. What about this year? According to our model, the Democrats are on track to finish with a median of 23 gubernatorial seats, a net gain of 2 seats. If you squint closely at the histogram over the fold, you'll see that the modal result (the one that occurs in the most simulations) is 24 gubernatorial seats (which would be a gain of 3). And if you look at the individual races listed in the totem pole to the right, you'll see that the Democratic candidates are running above 50 percent odds in races that would pencil out to a net gain of 4 (Pennsylvania, Florida, Alaska, Kansas, and Maine, minus Arkansas). Throw in the still-promising races in Michigan and Wisconsin, and you've got Dems in ... um ... array?
Finally, there's the matter of the House, which is where the real apocalypse happened in 2010. There's no way to quantitatively model the House, at least not on a seat-by-seat basis; there just aren't enough polls to give us an adequate level of information. (There are some imprecise measures based simply on generic ballot polling and historical House elections; for instance, the current 44-44 tie in the generic ballot projects out to a 6-seat Republican gain according to Alan Abramowitz's system.)
So, where the House is concerned, we at Daily Kos Elections tend to take off our sabermetrics hats and put on our old-school, cigar-chewing scout hats, looking qualitatively at factors like fundraising and national committee ad reservations to see where the real battles are. By this point in the 2010 election, we were seeing a steady drumbeat of polls showing Democratic House incumbents losing, and freshman incumbents in defensive races getting triaged by the DCCC. At this point this year, though, we can only point to three Dem-held open seats that appear to have been written off (NC-07, NY-21, and UT-04), and the only triage decisions so far have come in long-shot offensive races.
There are still around a dozen Dem incumbents in pure Tossup races — and we won't be able to save them all, just given the law of averages — but most freshmen are in decent shape, and there are even a few GOP-held seats that either clearly lean the Dems' way (CA-31) or where the Dems are doubling down (IA-03 and NE-02). A net Republican gain in the House of 6 may well be the correct guess.
We'll discuss a mind-blowing analogy to this year's election, plus the specific changes to the model this week, over the fold:
Gubernatorial races histogram
A little-noticed column by Kyle Kondik of the Univ. of Virginia Center for Politics written in June of 2013 may turn out to be the most prescient thing written about the 2014 election. In it, he compared the 2014 election to the
1986 election, a momentous midterm but one that's starting to disappear down the memory hole. It was also an election during the second term of a president who was popular with his base but loathed by the opposition, against the backdrop of economic recovery and foreign policy uncertainty.
It was marked by two course corrections that seemed contradictory, on their face. The Democrats netted 8 seats in the Senate, retaking control of the Senate. The Republicans had swept into control of the Senate on Ronald Reagan's coattails in 1980; six years later, many of those Republican freshmen got bounced out (in the same way that the gains from the 2008 Senate elections look poised to rebound against the Dems this year).
Meanwhile, in the gubernatorial races — where the term is usually four years, not six — the Republicans netted 8 seats; the Democrats had gained significantly in the 1982 elections, which were a reaction to Reagan's first term (in the same way that the Republican gains in the 2010 election are poised to get washed out). In the House, it was largely a wash, with the Democrats gaining a ho-hum five seats; the Democrats had already slashed into the Republican share in the House in 1982 as well, so there wasn't much damage left to do in 1986, in much the same way that the Republican gains in 2010 mean there isn't much left for them to do in 2014. The 1986 election is by no means predictive, of course, but it seems to have a lot of eerie similarities in its converging patterns, where two smallish waves going in opposite directions crashed into each other.
As I mentioned earlier, the gubernatorial races continue to look better, with the Democrats now holding a 63 percent chance of gaining seats, up from
56 percent last week. That's mostly thanks to the latest wave of
YouGov polls, which were leaked more than a week ago but finally became official last Friday. This collection of polls included a number of unusually good results, including a 45-40 lead for Dem-friendly indie Bill Walker in Alaska, a 41-34 lead for David Ige in Hawaii, a 47-43 lead for Pat Quinn in Illinois, and a 47-41 lead for Martha Coakley in Massachusetts.
Throw in a 44-41 lead for Quinn from Republican pollster We Ask America in Illinois, and a 39-34 lead for Coakley from the Boston Globe, and you've got the recipe for positive movement in a lot of different gubernatorial races. Illinois advanced from 72 percent odds to 80 percent odds for Quinn in the last half-week; Hawaii went from 66 percent to 84 percent odds; Massachusetts went from 54 percent to 66 percent; and Alaska was the biggest gainer, moving from 39 percent to 62 percent.
Meanwhile, in the Senate, things got only incrementally better for the Dems; it was a slow polling weekend, and hardly anything came out, good or bad. The most momentous poll was probably the Selzer poll of Iowa on behalf of the Des Moines Register; this poll still had Joni Ernst leading Bruce Braley 47-46, but that's a big improvement over their previous 44-38 poll two weeks ago. That probably doesn't reflect anything changing about the race, just float within the margin of error, but it still boosts Braley's chances from 29 percent to 42 percent, an imminently winnable race.
In addition, a Landmark poll showing the Georgia Senate race tied 45-45 boosted Michelle Nunn's odds from 10 percent to 18 percent. Kay Hagan's odds in North Carolina went from 70 percent to 76 percent, thanks to two polls from Republican internal pollsters giving Hagan a small lead.