As we now move within two weeks of Election Day, we look today at the third necessary component of a “wave election”, which is tilting the typical playing field for an election via turnout. Prior to this, we looked at the need for the Democrats to generate a vast playing field (check!), and we looked at the need for candidates to have the financial resources to compete late in the game (holy crap...check!).
Your party can have ample money to broadcast your message, and your party can manage to benefit from a ton of races that are within range of success, but those things alone will not bring the avalanche of seats falling on Election Day. The base has to get out, and your party has to pluck the bulk of the persuadable voters. How are the Democrats competing on that score? That is the question we will explore in this week’s edition of the House forecast.
But first, let’s crunch the numbers and see where the polling data takes us this week. It is a forecast that some will meet with increased pessimism, but above all else, it underscores just how narrow the distinctions are between an underwhelming election night for the Democrats and an utter tsunami.
THE HOUSE FORECAST: TWO WEEKS TO GO
A reminder, first, that last week we undertook a small methodological shift—to be considered for our “balance of power” forecast, a race (with the exception of the three races left for dead by the “incumbent party”—PA-05/PA-06/PA-14) has to have either been polled twice, or polled once by an independent outlet.
As was the case a week ago, the data this week is incrementally better for the GOP. Over the four week course of the forecast, this has been the trajectory of our forecast, with the net gain of Democratic seats bouncing around within a (relatively) narrow band):
WEEK # |
DATE |
DEMOCRATIC NET GAIN |
1 |
10/3 |
Democrats +27 |
2 |
10/10 |
Democrats +32 |
3 |
10/17 |
Democrats +27 |
4 |
10/24 |
Democrats +25 |
From the most likely to flip to the least likely to flip, here are the 33 races currently favored to change parties according to the polling averages on our House race pages. If these held according to form, the Democrats would net a gain of 25 seats, which would put them at a 220-215 majority (last week’s forecast was a Democratic gain of 27 seats):
POTENTIAL DEMOCRATIC PICKUPS (29 SEATS)
PA-05 (--); PA-06 (--); NJ-02 (current average: D+24); PA-17 (D+13); AZ-02 (D+11); CO-06 (D+10); PA-07 (D+10); NJ-11 (D+9); CA-49 (D+8); KS-03 (D+8); MN-03 (D+8); VA-10 (D+6); MN-02 (D+5); CA-10 (D+4); FL-27 (D+4); IL-06 (D+4); MI-11 (D+4); IA-01 (D+3); PA-01 (D+3); CA-25 (D+2); KS-02 (D+2); VA-05 (D+2); VA-07 (D+2); KY-06 (D+1); ME-02 (D+1); WI-01 (D+1); NC-02 (D+1); NJ-07 (D+0); TX-32 (D+0)
POTENTIAL REPUBLICAN PICKUPS (4 SEATS)
PA-14 (--); MN-08 (current average: R+11); AZ-01 (R+2); NV-04 (R+2)
So, shall we despair? After all, for two weeks in a row, the net number of projected seats gained has dropped!
The short answer is: no. As we noted last week as well, the field of potential contests is so heavily tilted against the Republicans that they need to hold onto a huge percentage of the close ones to retain their majority.
Let’s look at a metric that we've explored off-and-on since we started the DKE House forecast. I like the call it the “headwind/tailwind” model.
If the Democrats experience a slight headwind (which we will measure as 0-3 points in the Republicans’ favor), that would cost the Democrats the majority, as it would move a net of 13 seats into the GOP column (twelve seats listed above, plus the Democratic-held open seat in NV-03). That would leave the GOP with a narrowed majority of 228-207. At this point, this appears to be the “best case scenario” for the Republicans.
But if the Democrats benefit from even a slight tailwind (which we will likewise measure as 0-3 points in the Democrats’ favor), the shift is dramatic. Not only do the Democrats reclaim two of their “losses” above (AZ-01 and NV-04), but they’d also snag an additional 21 seats for a net gain of an additional 23 GOP-held seats. That would put the Democratic majority at 243 seats, an enormous 48-seat gain.
There has been a movement since we started these forecasts at the beginning of the month, and it is to the benefit of the Democrats. In our inaugural forecast, the Democrats had a total of 11 “projected pickups” where the projected margin of victory was a single point or less. In the current forecast, that number was down to just six races.
Meanwhile, in the initial forecast on October 3, there were six “Republican holds” where the projected margin of victory was a single point or less. Today, that number has bumped up to ten races. In other words, even though the DKE House forecast has seen a slight erosion in the “net number of pickups”, the underlying numbers have changed very little, and the positive end of the range for Democrats has actually improved markedly.
ANATOMY OF A WAVE, PART THREE: CHANGING THE ELECTORATE
Late last week, New York Times political analyst Nate Cohn, who has been shepherding an ambitious polling project powered by a partnership between the NYT and Siena, wrote a fascinating article about how their data shows that this midterm cycle is not behaving like its predecessors:
A wide range of evidence indicates that Democratic voters are poised to vote in numbers unseen in a midterm election in at least a decade.
Democrats have largely erased the turnout deficit that hobbled them during the Obama presidency, according to results from more than 50 New York Times Upshot/Siena College polls of the most competitive House battleground districts.
Any election victory is driven by two factors: either managing to maximize turnout, or managing to persuade the (admittedly thinning) reservoir of voters that are truly persuadable. Cohn’s article argued that Democrats might be managing to accomplish the former.
Across the Times/Siena polls, Republicans have a six-point lead among voters who turned out in 2014. But Democrats counter with a 10-point advantage among voters who didn’t turn out in that election. Those voters are poised to represent more than one-third of the electorate, enough to essentially eliminate the Republican turnout advantage of the last decade.
In this cycle, we have already seen this phenomenon take place. As I wrote in January, the Virginia landslide for Democrats last year was driven in no small part by a turnout surge in the counties and municipalities where Democrats normally are heavily preferred. In Virginia, the 2017 gubernatorial turnout increased by 16.7% over the turnout in the prior gubernatorial election in 2013. But, in heavily Democratic counties, the turnout increase was even more pronounced: Charlottesville was up 31.6%, while Alexandria, Norfolk, and Richmond also saw increases north of 20 percent.
Now, all we have to go on right now is polling data and (very) early statistics on early voting. Trying to draw sweeping conclusions from early voting data, as we learned in 2016, can be a fool’s errand and an invitation to disillusionment.
Sure, the first day early voting totals in Travis County (Texas) this week was not only more than double of the early vote in 2014, it was actually incrementally higher than 2016. And it’s hard to see that as an equivocal sign, given that the county is the one truly deep-blue large county in Texas (Travis, home to Austin, went 66-27 for Hillary Clinton in 2016).
That said, there are enormous caveats in looking at early voting. For one thing, it is difficult to know for sure whether a huge early vote (or vote-by-mail) for your party is a sign of genuine and lasting enthusiasm, or merely a cannibalizing of certain voters who would’ve voted on Election Day anyway. For another, knowing the partisan identification of early voters does not always portend how they’ll vote. For example, no shortage of Democrats believe (and hope!) that a number of suburban women who are registered Republicans will come their way as a reaction to Trump.
There are a couple of analysts who I think it is safe to say have become trusted resources on the matter: I would cite Jon Ralston in Nevada and Steve Schale in Florida as excellent resources on reading the tea leaves in their respective (and critically important!) states.
But, for now, watch the polls. If the pollster is modeling their results on the assumption of a 2014-esque turnout, that seems like it would be well off the fairway. Whatever else we know about 2018, it seems to be a mortal lock that it will not be a 2014 turnout. Turnout looks like it will be considerably higher than that. However, don’t assume that just because the amount of voters turning out will be considerably higher than 2014 that this means the composition of this electorate will be dramatically different than 2014. One of the mistakes that optimistic (and sometimes overly so) Democrats are making are mistaking clear Democratic enthusiasm this cycle with an absence of Republican enthusiasm. To generate the kind of wave that will create a solid Democratic House majority (and put Senate seats and governorships into play), Democrats will need not just an enormous turnout, they will also need to do something they haven't done in several cycles—seize a sizable majority of the “persuadable” vote.
Can the Democrats do that? That will be the topic of our forecast next week.