Daily Kos Elections has been gradually rolling out more bells and whistles that go with our model that forecasts the 2016 election. Today, we’re adding some new details about the Senate side of the equation. Previously, all you could see was the projected number of Dem-held seats in the dashboard that’s at the top of your page. But now we’re adding some new graphics to our landing page, which will give you some of the same information that you have about the presidential race.
Now, on a daily basis, you’ll be able to see changes in the odds of the Dems’ overall chances to control the Senate. Today, those are odds are at 59 percent … favorable odds, but by no means a guarantee of success. When we last talked about the Daily Kos Elections model, those odds were 61 percent. As you can see, things are pretty stable right now; there weren’t a lot of new Senate polls last week, and almost all of the ones we did see fit the same pattern that we’ve been seeing before.
The other thing you’ll see on the landing page is the individual odds in each of the key Senate races, much as you see odds calculated for all of the presidential swing states. Currently, you’ll see that the Democratic challengers in Illinois, New Hampshire, and Pennsylvania (Rep. Tammy Duckworth, Gov. Maggie Hassan, and Katie McGinty, respectively) all are favored to win their races. Duckworth, for instance, wins slightly more than two out three of each of our simulations of the Illinois race. Those represent three of the four seats that we’ll need to bring the number of Dem-held seats (and seats held by Dem-caucusing independents) to 50, which, in the very likely event of a Tim Kaine vice presidency, would put the Democrats in control of the Senate.
You’re probably wondering though ... what’s the fourth seat? Well, there’s not one but two seats held by Republicans, where the Democratic candidate has gotten the race to the point where our chart no longer displays it as “competitive.” Both of those candidates are well-known ex-Senators looking to get their old jobs back: Russ Feingold in Wisconsin, who’s seeking a rematch with Ron Johnson, who defeated him against the backdrop of the 2010 wave, and Evan Bayh in Indiana.
(Bayh’s situation is kind of complicated: he retired at the last minute in 2010, eyeing the wave that might well have taken him down. But the Republican who replaced him, Dan Coats … who came out of retirement in 2010 to challenge Bayh … is re-retiring this year, and Bayh un-retired at the last minute in 2016 to get his old seat back under more favorable conditions. Bayh’s re-emergence was also, reportedly, thanks to considerable prodding from Chuck Schumer, who correctly understood that Bayh’s return would significantly boost Schumer’s odds of being the next majority leader.)
At any rate, our chart spans the range of races where the odds range between 10 and 90 percent. Feingold, however, is leading the hapless Johnson (who, given how blue Wisconsin is at the presidential level, would be facing a difficult race even if he’d created a moderate profile and regularly distanced himself from his party … but instead, has been one of the most conservative members of the Republican Senate caucus and comes up with facepalm-worthy public pronouncements on a regular basis) by a margin, according to our aggregate, of 49-40, and his odds of victory are over 90 percent.
Feingold, however, is walking a tightrope over a fiery pool of sharks riding motorcycles, compared to Bayh, whose odds of victory, according to our model, are somewhere north of 99 percent. Our polling aggregate puts Bayh ahead of Republican Rep. Todd Young 53-38. That margin was bolstered, in the last week, by two more internal polls giving Bayh huge margins over Young. As Dem internals, our model downweights them a bit, but Young didn’t (and probably couldn’t) offer anything up to counter those polls, and even after the downweighting the margins are still overwhelming.
If you look at the Daily Kos Elections qualitative ratings, though, you’ll notice that we’re still keeping the Indiana race at “Tossup.” That may seem a little dissonant with the model’s quantitative score, but the model only relies on the polls, rather than the intangibles associated with the race’s narrative. Maybe most importantly, Bayh’s been having some trouble making a good case for his Indiana residency when questions have come up about it (and considering how effectively that argument was used against his former colleague, Richard Lugar, that could potentially be a strong weapon for Young, or for the third-party groups doing Young’s negative-ad dirty work for him, this year).
For now, though, Bayh seems to have not just a huge name recognition advantage over Young but also a lot of residual goodwill from his previous tenure. And if the polls don’t change in the Indiana race, we’re likely to going to be moving the qualitative rating soon to match the model’s view of the race.
If you add up Bayh, Feingold, Duckworth, Hassan, and McGinty, that looks like it should take the Dems up to 51 seats, not 50. The problem here, though, is the race in Nevada to replace Democratic leader Harry Reid, who’s also retiring. Recruitment wasn’t the problem here: the Democratic candidate is former Attorney General Catherine Cortez Masto, probably the strongest candidate the Dems could ask for here. Masto, however, is trailing Republican Rep. Joe Heck by a 41-39 margin in our polling aggregate, and while the Nevada race is still the closest Senate race of any of the ones where the Dem is trailing, that’s still enough that Masto is winning only slightly more than one of out every three simulations.
The problem in Nevada, as we discussed last week, is that we aren’t seeing a lot of polls of Nevada, and in the ones we are seeing, Hillary Clinton isn’t leading by that much, so the coattails aren’t enough to pull Masto ahead of Heck. It may simply come down to the fact that two of the last four polls of the Silver State were from Rasmussen, and even non-Rasmussen pollsters (especially ones who call only landlines and poll only in English) tend to have trouble reaching and correctly estimating Nevada’s large Hispanic population.
In fact, most pundits seem to err on the side of expecting Masto will win, and while we at Daily Kos Elections have this race at “Tossup” in our qualitative ratings, if forced to choose, I’d probably give the edge to Masto as well. As it stands, though, the majority of our model simulations check this one off as a loss for the Democrats, taking the total seats down to 50 instead of 51.
You might also be noticing one other high-profile race missing from our chart of key races: the one in Arizona, where Democratic Rep. Ann Kirkpatrick is taking on incumbent Senator John McCain. (Or, potentially, ex-state Sen. Kelli Ward, if she defeats McCain in the GOP primary, which looks extremely unlikely.) Here, Kirkpatrick’s odds have fallen below 10 percent. Our aggregate is currently finding good news! for John McCain, in the form of a 47-37 lead overall. That’s almost entirely thanks to one new poll that came out last week from CNN, which gave McCain a 52-39 lead; the CNN poll is inconsistent with the other recent polls we’ve seen of the race, most of which have given McCain a lead in the low single digits. But without a lot of other polls in the pile, it still has a big impact.
Somewhat surprisingly, the sleeper race in Georgia (where Democratic businessman Jim Barksdale is up against low-key Republican incumbent Johnny Isakson) now offers better odds than Arizona. That will likely change in coming weeks as we get more polls in Arizona to balance out the CNN one, but for now the little-known Barksdale seems to be holding onto Clinton’s coattails in Georgia, enough to get up to only a 45-39 deficit in our aggregate. If Clinton continues to gear up in Georgia with more ad and GOTV spending, that may help boost Barksdale even further … but then, Arizona is also the other likeliest state for Clinton to expand into, which could just as easily bolster Kirkpatrick against McCain.
Speaking of which, let’s turn briefly to the presidential race. Hillary Clinton’s likelihood of winning is currently at 82 percent, which is down a bit from the 88 percent figure we gave you last Monday. Hopefully, this shouldn’t be causing you too much alarm; you might remember that in both the 2008 and 2012 elections, quantitative models weren’t giving Barack Obama odds anywhere near that high at this point in the year (in fact, we hadn’t even gotten to the GOP convention at this point in the 2008 cycle). 82 percent is still a thoroughly dominant performance, and, as we discussed last week, the remaining 18 percent is more about allowing for the potential room for things to go wrong in the remaining two months (and the negative weight exerted by the “fundamentals”) than about any specific polling weakness right now.
Still, the drop from 88 to 82 can’t only be attributed to normal fluctuation within the model the way that, say, a drop from 88 to 86 could be. It was a relatively quiet week in polling … pollsters are probably taking a bit of a breather before the post-Labor Day onslaught … but a few polls in a few key states seem to have tipped the balance slightly. Part of the problem was in Florida, where we saw a couple polls with small Donald Trump leads (though one of the polls was from a Republican pollster), which pushed the Florida overall odds down, though Clinton’s chances there remain over 50 percent. Also problematic was the same CNN sample in Arizona that we mentioned before, where the presidential portion clocked in at Trump +5, taking Clinton’s currently narrow path in Arizona and making it even narrower.
You might also have noticed that, since last week, Clinton’s odds in Iowa and Nevada fell from slightly better than 50 percent to slightly below 50. That may seem peculiar, since in both states, there were no new polls in the last week. Bear in mind, though, that our model does link the states together somewhat, instead of just allowing them to each exist in their own universe independent of each other. In other words, slightly diminished odds in Florida and Arizona will carry over to all the other swing states and depress the odds there as well by a fractional amount. In most states, that wouldn’t be more than a blip, but when Iowa and Nevada were only slightly above water before, the ripples are enough to be make the difference in those two states.
Iowa and Nevada are just so much garnish this year, though (as are Florida and Arizona, as well). The most important thing to keep in mind is that as long as the Colorado/New Hampshire/Pennsylvania/Virginia parlay is solidly in place, that (along with all the usual blue states) takes Clinton past 270 right there. Clinton’s odds are over 75 percent in all four of those states … and in even in the simulations where one or more of those states falls through, there’s often a victory in Ohio or Florida (both of which have better than 50-50 odds) ready to fill in that gap.