:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/51612663/GettyImages-619466308.0.jpg)
With the presidential election eight days away and October surprises dropping right and left, the polls have tightened just enough to give Democrats some anxiety — though not yet enough to give Donald Trump a good chance of victory.
Hillary Clinton has led Trump in the vast majority of national polls conducted over pretty much any timespan of this campaign, and continues to do so going into its final week.
Yet more and more polls are popping up lately showing her with leads in the low single digits, rather than the bigger leads she tended to get earlier in October. As of Monday afternoon, Clinton’s lead over Trump in a four-way race was down to just 2.4 percentage points in the RealClearPolitics average. The HuffPost Pollster average, which works in newer data more gradually, has her still up by 5.7 percent, though.
Furthermore, Clinton’s easiest path to top 270 electoral votes — winning the solidly blue states plus the six “lean Democrat” states of Michigan, Wisconsin, Virginia, Colorado, Pennsylvania, and New Hampshire — still seems pretty safe for her, since Trump hasn’t led a single poll in any of those states since September or earlier.
Now, hardly any polls have been conducted entirely after FBI Director Jim Comey’s revelation on Friday that the FBI was examining new Clinton-related emails. But the couple of national polls that were — from Morning Consult and YouGov — found little to no change. Early voting numbers didn’t seem to be affected either, as Michael McDonald writes.
Yet it’s also worth noting that polls can be wrong! Though we tend to remember the 2012 presidential polls as getting the outcome right, in fact poll averages underestimated Obama’s popular vote by 2 to 3 percentage points. So if further data corroborates that Clinton’s lead truly is down to just a couple of points, Democrats shouldn’t feel all that comfortable.
What’s going in national polling
Clinton’s lead in national polling averages has clearly shrunk from where it had been in mid-October, when she was up by 7 points or so. But different outlets disagree on just how much it has shrunk. As of Monday afternoon, the RealClearPolitics average had her ahead by just 2.4 points when Gary Johnson and Jill Stein are included, and HuffPost Pollster had her ahead by 5.7 percentage points.
The two outlets have a few differences in methodology, but one big one is that RealClearPolitics swings more quickly in response to new polls (since it drops older polls out of the average entirely, while HuffPost gives them less weighting gradually). Newer polls have shown Clinton ahead but generally with a lead in the low single digits, and we haven’t had a reliable poll showing her with a double-digit lead for a while.
In all the major models, too, Hillary Clinton’s chance of victory remains at 87 percent or above — except for FiveThirtyEight, whose poll-only model gives her a still solid but less impressive 76 percent chance of victory.
All year, the FiveThirtyEight model has tended to react more dramatically to short-term changes in national polls than the other models (which, depending on whom you ask, could be either a feature or a bug). But this isn’t all that dramatic a discrepancy, and the big picture is that the models essentially reflect what the polls are saying.
What’s going on in state polling
Since August, it’s appeared that Clinton’s easiest path to top 270 electoral votes has been through winning solidly Democratic states as well as the six “lean Democratic” states of Michigan, Wisconsin, Virginia, Colorado, Pennsylvania, and New Hampshire.
And that still looks to be the case. All along, Clinton has led the vast majority of polls in these six states. In five of them, Trump hasn’t led even a single poll since at least July. In the case of Wisconsin and Michigan, he hasn’t led a single poll all year.
Yet we haven’t gotten all too many polls of these states since the national numbers started to tighten again. And campaign observers did note that the Clinton campaign began buying its first ads in Wisconsin last week, which conceivably could indicate some tightening there. (However, the ad buy could also be an attempt to help out the Democratic Senate candidate, Russ Feingold, by padding Clinton’s margin of victory.)
The best state poll news for Trump lately has been a few new polls showing him once again leading in Florida, which is an absolute must-win state for him. However, there have been other polls showing Clinton still holding on to a narrow lead.
And again, winning Florida wouldn’t be enough for Trump. He also likely needs North Carolina, and out of 15 pollsters who have polled the state in October, only one — a Republican firm — has shown him ahead there.
What’s going on in early voting
The effect of any other last-minute “November surprises” may be blunted somewhat by the fact that more than 22 million people have already voted. And about 10 million of those votes are in swing states, according to statistics tracked by Michael McDonald at US Election Project.
Indeed, votes equivalent to a third or more of total ballots cast in 2012 have already been cast in Florida, North Carolina, Nevada, Arizona, and Colorado, either through in-person early voting or by mail.
As to what early voting has shown us so far — well, as McDonald has observed, “early voting statistics can be a bit of a Rorschach Test that entices partisans to see victory for their favored candidate.”
But there have been some interesting analyses from Dave Wasserman (who suggests Hispanic turnout is up and black turnout is down from 2012), Steve Schale (who writes that the Florida data suggests the state could be very close), and Jon Ralston (who suggests “Trump almost surely is losing Nevada”).
What is clear, though, is that if there is a truly last-minute “November surprise,” then it would likely have its effect concentrated in states where early voting doesn’t exist or tends to have a low amount of participation, like Pennsylvania, New Hampshire, and Virginia.
