The 2018 election is first national referendum we’ve had on Trump and the current iteration of the GOP, and it’ll be the closest thing we get to an up or down vote on him until November 2020. So I decided to do a slightly-different-than-normal pre-election overview of the fight for House control. Rather than go through the race district by district or state by state (because frankly even I get lost in those sometimes) I attacked it by dividing the districts into a few big batches (based on competitiveness), describing those batches very roughly and using them to come up with a projected final result.
After that I talked through some of the (big) sources of error and how to think about the results as they roll in.
(1) The Likeliest Outcome: Democrats Take The House
Thinking through 435 different races in different districts with unique sets of candidates is tough. Unlike a Senate or presidential election, you can’t count the truly marginal states on two hands. So I started to think through this by looking at handicapper ratings.
These handicappers (skip this paragraph if you’re not interested in methodology)—Cook Political Report, Inside Elections, and Sabato’s Crystal Ball—rate races as “Solid,” “Likely,” or “Leans” (and “Tilts” in Inside Elections’s case) towards one party or as a “Tossup” I translated those scores into numerical ratings from -3 to 3 (Solid as 3, Likely as 2, etc.) and then averaged them, giving me a sort of consensus for every district as of late last week. This is helpful because, despite the great work being done over at the Upshot, we don’t have polling data for every district. And these ratings (which use a combination of public and private data as well as interviews with candidates) give us information on under or un-polled elections.
So let’s just start off by giving Democrats every district that’s leans-likely or better (I used my scale as a rough guide, and these districts all were close to 1.5 or better for Democrats on the scale). That gives them 194 districts—almost exactly their current total. These districts are basically the Safe Democratic districts—blue places that almost assuredly aren’t going to flip. And the 194 number works neatly for our purposes: Democrats currently have 194 seats and the president’s party almost always loses seats, so it makes sense to start the Democrats out with 194.
After that, there are about 15 districts that I count as between tilts-leans and leans-likely (roughly those where the consensus among handicappers is that these districts are more “Lean Democratic” than “Tossup” or “Likely Democratic”). These districts include AZ-2, CA-49, CO-6, FL-27, IA-1, IL-6, KS-3, MI-11, MN-2, MN-3, NJ-11, NV-3, NV-4, PA-7, and VA-10. A lot of these districts (but not all) fit the sort of “suburban battleground” archetype—districts where Trump might have underperformed Romney and college-educated white voters might be energized to vote against Trump.
The Upshot’s polling has been mostly good for Democrats in these districts, so let’s just suppose that Democrats have a roughly 80 percent win rate in these districts and end up gaining 12 here. They might do better or worse (more on that in the error margin section), but 12 seems like a decent number to me.
That brings Democrats to roughly 206 seats heading into the swing-iest districts.
Which leads us to the 26 true toss-up districts. I defined these as districts where the handicapper consensus is roughly between “Tilts Republican” and “Tilts Democratic” Others might disagree with this list, but I have it as CA-10, CA-25, CA-39, CA-45, CA-48, FL-26, IA-3, KS-2, KY-6, ME-2, MI-8, MN-1, NC-9, NC-13, NJ-3, NJ-7, NM-2, NY-19, NY-22, OH-12, PA-1, TX-7, TX-32, UT-4, VA-7, and WA-8.
If Democrats win half of these districts, it’s game over. Democrats end up with 219 and have the majority.
I tend to think that these toss-ups will be somewhat better than 50/50 for Democrats and that they’ll be able to win some upsets in redder territory (e.g. there are about 55 more districts that I’d count as still technically on the playing field). When I factor that in and sub in some of my best guesses, the Democratic total ends up at about 228.
I don’t feel a great need to spell out exactly which districts Democrats or Republicans will/won’t win. Our district-by-district data is only so good. Everyone who predicts the House knows that some of their calls will be wrong but tries to make their errors cancel out and create a believable topline.
I think 228 Democratic seats is a sensible, most-likely-result projection.
(2) Error Margin: The Playing Field, Polarization and Polling Error
That being said, politics isn’t physics, and our tools for understanding it aren’t electron microscopes. There are many sources of errors and it’s possible for that projection to be off by a solid amount. Here are three of the big ones.
First, there’s polling error. Most of my analysis is (implicitly or explicitly) informed by district-level polling data as well as national polls (like the generic ballot). Polls are better than nothing (just imagine the bad punditry that partisan cheerleaders would produce if there were no numbers to anchor our discussions about elections), but they can miss the mark. Moreover, translating national polls (like the generic ballot) down into 435 district-level races is a tough analytical challenge, and even if the generic numbers are correct in the aggregate, the votes can end up distributed in funky ways.
So it’s not hard to imagine a scenario in which the polls systematically underestimate the GOP by a reasonable amount, allowing them to hold almost all the districts that lean even a bit towards them, do well in the toss-ups and retain the House despite losing seats. Alternatively, the polls could underestimate the Democrats, allowing them to clean up in the marginal districts, reach into redder districts and beat the 228 seats projection by quite a lot.
Second, there’s polarization. One of the biggest analytical questions of this election is how well Democrats will do in pretty-red-but-not-safe territory. There are a lot of light-to-medium red districts where Republicans are favored to hold on. If polarization sets in hard, regular Republicans show up and the GOP holds the line there, they can at least put a ceiling on Democratic gains in those districts and possibly (if they also do well enough in the toss-ups) hold the chamber.
Third, there’s the flipside of polarization—playing field size. If the opposite happens (that is, the GOP shows weakness in the toss-ups but also allows Democrats to score some upsets in tougher races) then the playing field could expand in a way that seriously harms Republicans. The name of the game for the GOP is damage control. So if Democrats force them into a large playing field (i.e. if some of the underdog House candidates who are raising lots of cash are actually able to score upsets) things could turn out pretty disastrously for this iteration of the Republican party.
There’s a lot of error here, and everything from a GOP hold to a Democratic landslide is still on the table. That being said, the best single guess is that Democrats take the House but don’t end up with an overwhelmingly huge House majority. But if something else happens (including either a Democratic landslide or a GOP hold) you shouldn’t be shocked.
(3) This Isn’t All About the Suburbs
The final seat count isn’t the only interesting/important part of this election. These elections give us a window into public opinion in key segments of the country. A lot of media attention has been focused on the suburbs, and on now-swing-y, traditionally Republican, college-educated white voters who may go Democrat this time. That’s not a crazy thing to focus on—those voters are plentiful in a lot of the most competitive districts. But it’s not the only thing that’s happening
One way the GOP could hold down Democratic gains (aside from doing well in the suburbs) would be to do well in diverse and Trump-ier areas of the country. Trump won MN-1, MN-8, KY-6, IA-3, ME-2, OH-12, NY-22, and a many other areas that are stereotypically Trump-ier than the garden variety suburban district. Moreover, Republicans have been doing reasonably well in some racially diverse districts (e.g. FL-15, FL-26, FL-27, TX-23). And a number of the more Trump-y districts also feature a substantial number of Obama-to-Trump blue collar GOP converts.
Put simply, a win or loss for either sides is going to end up looking complex. The basic storyline of this election can’t just be about the suburbs because there are too many battleground districts that don’t fit into that mold.
(4) What a Win/Loss Actually Means
I’m not sure how significant the House election is going to be from a policy perspective. If Democrats take the House but Republicans hold the Senate (which seems like the likeliest outcome) then there’s probably going to be a lot of legislative gridlock. If Democrats take both chambers, President Trump will likely veto a lot of the more liberal things they’d try to pass. And if Republicans hold both chambers, their margin in the House will almost assuredly be lower (though their Senate margin may be greater). It may end up being tough to pass legislation through a House with fewer GOP members (remember the stop-and-start nature of the Obamacare Repeal and Replace effort). The midterms may matter for the Russia investigation, but I think it’s a lot harder to prognosticate around that whole mess of issues than many believe.
But this election matters because public opinion matters. The things that Americans feel and believe matter.
As I said up top, this election is the first national referendum on the Trump presidency. And right now, the most likely outcome seems to be a well below average—but not historically bad—verdict. A Republican hold (which would qualify as a good midterm in terms of seat count, but could look different once we look at final House popular vote) and a Democratic landslide are still on the table. But if the expected outcome happens (Republicans losing 30-something seats), it’d qualify as a not-great-but-not-catastrophic result.
And over the next few years, this midterm is going to be a big part of the discussion of how “normal” Trump is. If the GOP does reasonably well in this midterm (that includes a bad-but-still-normal loss) and Trump wins re-election, then I tend to think that some of the weirdness of the Trump Era will be regarded as normal parts of How Things Work Now. If Trump gets a truly terrible midterm verdict and then either loses re-election or is removed from office, then politicians might (at least in the short term) try to be less Trump-ian.
