Skip to main content
Trump Probably Won’t Defy Midterm Gravity

The midterms will probably be rough for Donald Trump and Republicans next year.

Here’s the main reason we know that: They’re almost always rough on the president’s party. It’s one of the most consistent regularities in American politics. Usually it’s a question of whether the damage is pretty bad, really bad or catastrophic.

Here’s the other thing we know and which cases like the Georgia special election on Tuesday — and the one in Kansas last week — serve to demonstrate: Trump can’t defy gravity. These were bad results for Republicans. They’re consistent with what you’d expect from a public that would like to elect some Democrats to counterbalance an unpopular president and the Republicans’ hold on both branches of Congress.

You can debate exactly how bad these elections were for Republicans, of course. Each special election is subject to its own circumstances. But the narrative that “Democrats don’t have any wins yet” is dumb. Kansas’s 4th Congressional District, where Democrat James Thompson lost to Republican Ron Estes by just under 7 percentage points, is as red as Alabama. A Democrat coming close there is the sort of thing you’d see only in a really bad or perhaps even catastrophic midterm for the GOP.

The Georgia case is more ambiguous. Democrat Jon Ossoff — after winning a 48 percent plurality of the vote in the all-parties primary on Tuesday — is roughly even money to win the runoff against Republican Karen Handel. But the district has changed a lot, having gone from extremely red in 2012 to competitive in 2016, when Hillary Clinton lost to Trump by less than 2 percentage points there. I’m on the side that says the result on Tuesday is more consistent with a pretty bad outcome for Republicans in 2018, rather than a really bad one, although that might change depending on how the runoff goes.

What do I mean by pretty bad, really bad and catastrophic? Roughly, something like this:

  • I’d consider a pretty bad outcome for Republicans to be one in which they lost one to two dozen House seats next year, leaving them with only a narrow majority — or perhaps a razor-thin Democratic majority if Democrats played their cards just right and won a lot of the closest House races. Because the Senate map is so favorable for Republicans, the GOP might gain a seat or so on net there, but most of the vulnerable red-state Democratic incumbents would survive intact.
  • I’d consider a really bad outcome for Republicans to be one in which they lost 30 to 40 House seats along with control of the chamber. Republicans would hold the Senate but lose one or two seats.
  • I’d consider a catastrophic outcome for Republicans to be a 2010-style wave election in which they lost 50 or more House seats, Democrats either won the Senate or came really close, and Democrats won a large majority of purple-state gubernatorial races.

It may seem bold to lay out a menu of options like this. But if you look at the history of midterms, it isn’t. Take a look, for instance, at how the aggregate popular vote for the U.S. House has behaved in midterm elections dating back to 1922:

2014 Obama +5.7 -1.2 -6.9
2010 Obama +6.8 -10.6 -17.4
2006 W. Bush +2.6 -8.0 -10.6
2002 W. Bush +0.5 +4.8 +4.3
1998 Clinton +0.0 -1.1 -1.1
1994 Clinton +5.0 -7.1 -12.1
1990 H.W. Bush -7.7 -7.8 -0.1
1986 Reagan -5.1 -9.9 -4.8
1982 Reagan -2.6 -11.8 -9.2
1978 Carter +13.6 +8.9 -4.7
1974 Ford -5.6 -16.8 -11.2
1970 Nixon -1.7 -8.7 -7.0
1966 Johnson +14.7 +2.7 -12.0
1962 Kennedy +10.0 +5.3 -4.7
1958 Eisenhower -2.5 -12.4 -9.9
1954 Eisenhower -0.5 -5.5 -5.0
1950 Truman +7.2 +0.7 -6.5
1946 Truman +4.7 -8.5 -13.2
1942 Roosevelt +5.8 -3.8 -9.6
1938 Roosevelt +16.2 +1.6 -14.6
1934 Roosevelt +13.1 +11.7 -1.4
1930 Hoover +14.1 +8.0 -6.1
1926 Coolidge +15.1 +16.3 +1.2
1922 Harding +23.2 +7.0 -16.2
Average +5.5 -1.9 -7.5
Midterms almost always swing against the president’s party


On average, there was a 7.5-percentage-point swing against the president’s party at the midterms relative to the election two years earlier. Few presidents have been immune from this. And presidents who came into office with a big wave — as Barack Obama did in 2008 — saw their party suffer massive losses as the electorate reverted to the mean. But presidents who had a limited mandate to begin with, such as Jimmy Carter in 1978, also suffered. The president’s party gained ground only in 1926 and 2002.

I’ve framed these losses in terms of votes lost rather than seats lost because converting from one to the other is tricky. That there are fewer and fewer swing districts is an undoubtedly big help to Republicans. On average since 1922, the 7.5-point popular vote swing against the president’s party has translated into a loss of 29 seats in the House, but Democrats probably wouldn’t get quite so favorable an exchange rate. And there are other factors that go into the calculation, such as how strong the incumbency advantage is and how broadly the opposition party contests the map. It’s the sort of thing that you’d probably want a model to help figure out, and we don’t have a model yet.

What we do have, however, is an increasing amount of evidence that Trump is facing real and sustained resistance from the public. This evidence comes not only from the special election results so far, but also from his poor approval ratings, his difficulty in passing key pieces of legislation such as health care reform, and from Republicans’ problems in recruiting top-tier candidates for next year’s races.

And none of it should really be surprising. Trump was elected only narrowly — losing the popular vote to Clinton — and he hasn’t made much effort to reach out beyond his base. Some of his supporters were reluctant to begin with.

Being president is a tough job. Like many of his predecessors, Trump may get better at it as he goes along. But most of those predecessors suffered a substantial setback at the midterms first.

Nate Silver is the founder and editor in chief of FiveThirtyEight.