What 2018 Tells Us About 2020

.

The 2018 election is over, so naturally the world of elections journalism is about to pivot to 2020 coverage. Soon you’ll start seeing “Power Rankings” for Democratic presidential candidates, chatter about a GOP challenger to Trump (though I think the results of this election discouraged that) and all of the stuff you expect/crave/dread from our two-year presidential cycle.

But if you jump into that too quickly, you can leave important insights on the table. I think that digging deep into the 2018 election results is going to give us valuable information about where our politics is and where it’s going. So I decided to take a first pass at the House data—basically to dive into it without much of an agenda and see what I came up with. And I had three major takeaways.

House Results: In Electoral Settings, Trump is Normal

Most journalists and analysts judge midterm House results on how many seats a party lost or gained. I don’t have a problem with that metric (I’ve used it a lot in the past) but I’ve been wondering whether it truly captures everything. So I decided to grab some data from a couple other measures and try to get a more holistic view (note that this table shows data in terms of the president’s party)

Year Popular Vote Win/Loss Seats After Last Election Seats After Midterm Seat Loss Percent of Members Gained/Lost
1994 -7.1 258 204 -54 -20.93%
1998 -1.1 206 211 5 2.43%
2002 4.8 221 229 8 3.62%
2006 -8 232 202 -30 -12.93%
2010 -6.8 256 193 -63 -24.61%
2014 -5.7 201 188 -13 -6.47%
2018 -6.9 241 206 -35 -14.52%

This table shows the House popular vote, the seat totals before and after the election, the raw seat loss, and the seat loss as a percentage of what they won last time. The numbers are all in terms of the president’s party. So you see 1994, 2006, 2010, 2014 and 2018 look really bad for the president’s party with more positive results only in 1998 (mid-Lewinsky Republican overreach) and 2002 (post-9/11 George W. Bush) .

This context basically suggests that the 2018 House Democrats did about as well as the 2006 Democrats. Their popular vote totals were similar, their seat gains weren’t far off that year. The 1994 and 2010 Republicans flipped more seats than the 2018 Democrats did, but they won the House popular vote by a similar margin (some don’t like this metric, but I tend to think it’s both flawed and better at capturing overall public opinion than most of our other metrics).

To me, this signals that Republicans had a bad-but-not-unheard-of year. Which, in turn, suggests that Trump is neither magical nor invincible. When he does unpopular things (e.g. push an unpopular healthcare bill, tweet way too much, etc.), his poll ratings take a hit and his party suffers. Trump’s Republicans don’t always over-perform their polls (though I want to have a more nuanced discussion of what that looked like on the Senate side in a later piece). And if we treat Trump as a basically normal president when we’re doing elections analyses, we’re more likely to get to the right result. Economic and big picture political indicators (e.g. Hillary Clinton was running for a third Democratic term) suggested that 2016 would be close, and it was. And I suspect that traditional indicators will tell us useful things about Trump’s chances when the time comes in 2019 and 2020.

Put simply, this is a bad-but-normal result that suggests that the normal, high-level patterns that have governed elections in the past still also apply to Trump.

Democratic Gains: NBD for the presidential elections, but maybe matters for the next House election

In the past, I’ve argued that midterm results don’t do much to shape the outcome of general election outcomes. I think that’s basically right on the presidential level.

house_potus_relationship_2.jpg


I published this graphic a few weeks ago, and each point is a midterm. The horizontal location is the midterm result (GOP House seat loss) and the vertical location is the popular vote in the next election.

There’s no pattern here. Typically, the midterm doesn’t tell us too much about what’s going to happen in the next election. So this race doesn’t prove that Trump will be in trouble next time around, and it doesn’t prove that he’ll rebound either.

But the results are slightly more informative when we think about future House elections. Democrats failed to win by a landslide margin, and that means that the House majority will likely be at least up for grabs again in 2020. The installation of new Democratic incumbents in marginal House districts will likely erode the GOP’s structural advantage a little bit (e.g. a Democratic Congressman who gets a three point incumbency bonus could turn a very lightly red district into a purple one), but the Republicans will probably maintain much of that advantage. So there’s a decent chance that the House will be competitive and could easily flip back to Republicans next time around.

The New House Majority

And the contours of this win suggest that, to some degree, we’ll be seeing more of the same from our politics:

polit_post_house_18.jpg


This scatter plot shows the politics of the different districts. Democrats did well in a lot of district where Clinton made gains compared to Obama in 2012 (e.g. districts with a lot of traditionally Republican suburbanites who disliked Trump). But Democrats also won in some districts where his margin was better than Mitt Romney’s.

The real pattern is that these districts hug the center line and the region just a bit to the right of it—that is, places where that Trump won by a bit or didn’t lose in a landslide. This suggests that the sort of battering ram pattern described before (that is, that parties tend to win the House by grabbing marginal districts rather than building broad bases in areas that vote differently at the top of the ticket) is to some degree showing up again.

In my view, these political patterns seems stronger than the basic demographic patterns

demog_post_house_18.jpg


Many of the districts that Democrats won fit into the mold of the suburban, well-educated, traditionally Republican district. But that narrative hides some of the variation, and even displays like this hide some of the other ways these results vary (e.g. region, size of metro area for suburban areas, etc.)

There are some Democratic members who need tons of votes from very red areas in order to keep their seats. And they’ll make future Speaker Nancy Pelosi’s life more difficult. But my guess is that this version of the House Democrats may be a bit more uniformly liberal than past iterations of the party. And that might lead to a rougher, more contentious landscape on Capitol Hill (if you can believe that that’s possible).

Related Content

Related Content