Time for a quick update on the United States midterm elections, held two weeks ago – and particularly the House of Representatives, which I mentioned only briefly last time around.
The Senate is now final, apart from the runoff in Mississippi next week, which Republican Cindy Hyde-Smith will almost certainly win. The Democrats picked up two seats but dropped four of their own, producing an increased Republican majority of six, 53-47.
The House, however, is a better Democrat result than it looked at first. On the Friday after the election I estimated a gain of 35 seats, but now with everything just about decided, it seems to be a net gain of 39, for a healthy 234-201 majority.
FiveThirtyEight still has three seats on its doubtful list, but it’s hard to see any of them changing from here: California 21st and Georgia 7th look pretty solid for their Republican incumbents, while Utah 4th seems to have been won by its Democrat challenger. There may be a surprise there somewhere, but one seat either way won’t really change the overall narrative.
That doesn’t mean that counting has finished; the majority of states are still tallying votes, even though there are not enough of them to change anything. You can check the progress of the count at the spreadsheet maintained by the Cook Political Report.
As of this morning, it shows the Democrats leading in the popular vote overall by 8.66 million votes, or 53.0% to 45.2%. Factoring out the small number of “others”, that gives a two-party split of just on 54/46 – a swing of 4.5% from two years ago.
That’s a decent swing, but it’s not huge. For comparison, last time the Democrats won back the House, in 2006, it was with a swing of 5.5%, and when they lost it again in 2010 the swing was a whopping 9.0%.
It looks as if the Democrats did a very good job of making their gains in just the right places. In my preview, I said they would need a uniform swing of about 6.5% to win a majority, but they’ve done it with considerably less.
As always, there are some caveats on those figures. Firstly, late counting tends to favor the Democrats, because it comes disproportionately from big cities and big states (especially California); their overall vote may still tick upwards a bit.
Secondly, interpretation of the totals is made more difficult by a number of seats that are uncontested, and a much larger number that are only nominally contested: seats where incumbents are seen to be so safe that the other party puts in only a token effort. By my count, there were 158 seats (more than a third of the total) where one or other party had more than two-thirds of the vote.
Thirdly, there are a lot of places where specific local factors have an impact. One in particular to mention is Pennsylvania, where the district boundaries were redrawn this year by the state supreme court, after the Republican-controlled legislature refused to fix a blatant gerrymander. As a result, the parties won nine seats each there, a net Democrat gain of three.
Midterm elections have a poor record at predicting the following presidential election. The Republicans won big in 1994 and 2010, but each time their presidential candidate lost badly two years later.
On the other hand, the two parties start in different positions: the Democrats have an inbuilt disadvantage in midterms, because the electorate tends to skew older, richer and whiter. Turnout was well up this year (an increase of some 33 million on 2014), but still a lot less than in a presidential election year.
So the Democrats can reasonably take this result as a positive sign for 2020. But the presidential race remains wide open. Among other things, much will depend on what Donald Trump’s opponents do with their new House majority.
*UPDATE 23 November*
FiveThirtyEight now thinks the Democrat has a slight edge in California 21st, the only remaining doubtful seat. If that does come off, it’ll be a net gain of 40 seats, for a 35-seat majority, 235-200. It’s not expected to be finalised until next week – you can follow the counting here.