We’ve seen enough polls in the last week to know that Hillary Clinton won a decisive victory in the first debate of the presidential general election. Polls that asked specifically about the debate gave her the win, almost always by double-digit margins. Unfortunately, though, our model doesn’t care about polls about how who won the debate, any more than it cares about national polls.
The good news is that in the state polls that we’ve seen, that had field periods after the debate, the numbers have looked pretty good. A series of polls from PPP of the “firewall” states gave Clinton leads of 6 in Colorado, Pennsylvania, and Virginia (along with leads of 2 in Florida and North Carolina). That didn’t represent a major improvement over the current averages in Pennsylvania and Virginia, though it did help shore the average up in Colorado (which had seen a trio of polls with Donald Trump leads in the previous weeks … though two of that three was dubious pollsters Gravis and Emerson).
And just as importantly, a grab-bag of other pollsters put up some other positive numbers: Clinton +7 in New Hampshire from MassInc, Clinton +4 in Florida from Mason-Dixon, and perhaps the biggest number, Clinton +6 in Nevada (a state that’s been woefully underpolled this year, and one where this year’s polls have shown a big falloff from a strong Barack Obama performance from 2012) from Suffolk.
The Suffolk poll in Nevada wasn’t enough to bring Clinton’s odds in the Silver State up to 50-50 (they’re still at 30 percent, though with a sharply upward trajectory), but the new Florida polls were enough to bring Clinton’s odds there up to 53 percent. That helps to bring Clinton’s overall odds to 72 percent. That’s still a far cry from where we were when we went public with the model (90 percent in mid-August), though, in retrospect, that was still at the peak of her convention bounce … even though we waited, to avoid starting at an artificial high water mark like that (it turns out her convention bounce lasted a lot longer than usual, perhaps having gotten conflated with the Khan family story). But it’s the highest we’ve seen since Sept. 14, which is right when the first polls after the media’s fanning of Pneumonia-ghazi and Deplorable-gate started to show up.
If anything’s a little odd right now, it’s simply a question of “where are all the polls?” It’s not your imagination that there are fewer polls this year compared with the 2008 or 2012 cycles … or at least the traditional individual-state polls using call centers. (There certainly has been an increase in online non-probability samples, like the national tracking polls from NBC/Survey Monkey and Ipsos/Reuters.) That’s largely a factor of newspapers, who traditionally have been the funding source for much of the state-level polling, just not having the financial resources to commission as many polls as they used to. And that’s why you’re seeing less of certain once-prolific pollsters like SurveyUSA and Mason-Dixon this year; they don’t do freebie polls to promote their internal polling operations (like PPP) or to promote their university brand (like Quinnipiac), and their clientele is mostly papers and TV stations.
In particular, it’s odd that we’ve seem so little polling after a big event, in the form of the debates. Beyond what I mentioned above, there’s really been a few other random polls in the more marginally competitive states in the last week. For instance, this weekend saw one with a 7-point Clinton lead in Michigan, though that didn’t get anywhere near the attention that a poll of New Mexico from the Albuquerque Journal.
That poll had only a 4-point lead for Clinton (which, as you may have noticed, pushed New Mexico from dark-blue down to medium-blue on our map, though that’s because her odds there fell to “only” 95 percent), though what really got the attention was that Gary Johnson was polling at 24 percent in his home state. Even taking his residence into account, though, it seems highly unlikely Johnson is able to sustain that level of support through Election Day; with the exception of Ross Perot’s 1992 run, third-party candidates tend to fall way short of their poll numbers once the rubber hits the road (and Perot in 1992 was fundamentally different because he was able to participate in the debates).
It’s possible that there is a huge wave of polls all ready to drop on Monday (in which case you might already be seeing them and wondering why I’m complaining about the dearth of new polls). But it’s also possible that isn’t happening, and that points to another potential problem that all modelers, not just us, have to grapple with: effective aggregating requires lots of polls in order to smooth out all the noise, so fewer polls means that outliers or simply poorly executed polls are likelier to have a bigger impact on the big picture.
Finally, let’s turn briefly to the Senate side. The odds of the Democrats hitting 50 seats in the Senate were at 56 percent when we last checked on Thursday … and those are about the same odds today, at 55. Unfortunately, none of the new polls that we’ve seen since then had any game-changing numbers, even the Nevada poll from Suffolk that gave Clinton a good-sized lead. That same poll gave Republican Joe Heck a 3-point lead over Democrat Catherine Cortez Masto. The good news about that poll is that the topline was only 38-35, meaning that the candidates still aren’t very well-known (despite the election being a month away) and that Masto has room to grow if she can consolidate Clinton voters.
The best news on the Senate front in the last week, though, was probably the MassInc poll of New Hampshire, which had Democratic challenger Maggie Hassan leading Republican Kelly Ayotte by 2. That helped push that race back into almost 50-50 territory (48 percent for Hassan, specifically), and also pushing it past North Carolina (at 41 percent odds for Deborah Ross) back into the “pivot point” slot on the Senate totem pole. In other words, if we’re going to win Illinois, Indiana, Pennsylvania, and Wisconsin, but lose the currently Dem-held Nevada, the Dems need a fifth seat somewhere else to hit 50, and New Hampshire once again provides the best odds of that.
Frustrated you don't live in a swing state? No matter where you live, MoveOn has a great way for you to help their on-the-ground efforts to defeat Donald Trump and take back the Senate. Click here to volunteer.