United States | The mid-terms

We think the Democrats are favoured to take the House

But the contest is finely balanced

Listen to this story.
Enjoy more audio and podcasts on iOS or Android.

ALTHOUGH they lack the intense personal drama of a presidential race, America’s mid-term elections in November will be hugely important. Every seat in the House of Representatives is up for grabs, along with 35 out of 100 Senate seats. A Democratic takeover of either chamber would unleash a flurry of investigations into President Donald Trump and wreck his hopes of passing more conservative laws on a partisan basis. If the Democrats take the House, Mr Trump might also be impeached.

This year’s mid-term campaign is extraordinary in another way. It is expected to be closely fought. Thanks to Americans’ tendency to separate into like-minded communities and to deliberate gerrymandering, most individual House races are one-sided. Historically, control of the lower chamber has been a foregone conclusion as well. In every contest from 1954 to 1992, the Democrats won at least 232 seats, well above the 218 needed for a majority. Since 2002 the winning party has always claimed at least 229 seats, with the results predictable (mostly in the Republicans’ favour) months before election day.

By contrast, this year’s contest appears to be poised on a knife-edge. Because district lines favour the Republicans, most estimates suggest the Democrats could wrest control of the House if they manage to win 53-54.5% of the total votes cast for the two major parties. For most of 2018, pollsters have reported that about 53.5% of people who express an opinion intend to vote Democratic. The Republicans appear likely to keep control of the Senate, but a Democratic wave could also put it into play.

Partisans from both sides see encouraging auguries. Republicans cite the economy, which is entering its ninth consecutive year of expansion. Democrats point to their recent victories in special elections. They won a Senate seat in deeply conservative Alabama, a House district in Pennsylvania that Mr Trump had carried by 19 percentage points, and came stunningly close to winning control of the lower chamber of Virginia’s legislature. Democrats also note that Mr Trump is disliked by historical standards. Despite a recent improvement, his approval rating is much lower than his disapproval rating. Over the past half-century, this has previously occurred four times at this stage in a mid-term cycle (1982, 2006, 2010 and 2014). Each time, the unloved president’s party proceeded to suffer devastating losses in the autumn.

Some pundits think the House race is too close to handicap. But The Economist has tried to do so. We have created a statistical model to forecast votes for the House, which has been trained on every election cycle since 1942 and 6,500 historical district races. At the moment, we give the Democrats a two-in-three chance of capturing it.

Our model is explained fully on our website. It begins by calculating the probability that each party will win a specific share of the overall popular vote, drawing on opinion polls, special and off-year elections, the unemployment rate and the partisan polarisation of the electorate. Then it delves into individual district races, weighing things such as voting history, incumbency, fundraising and candidates’ ideological views. It runs thousands of simulations and reports the percentage in which one party holds at least 218 seats.

Spying a blue wave

The model makes the Democrats a clear but narrow favourite. It assumes that some historical patterns will persist. In the past, undecided voters have moved towards the opposition party as mid-term elections draw near. As a result, the model expects the Democrats’ lead in polls that ask about people’s broad preferences to grow from 6.4 percentage points now to 8.8 in November. It places a heavy weight on performance in special elections, particularly those for federal congressional seats held within a year of the vote, in which Democratic challengers have fared well.

The model expects Democratic candidates to be competitive in many ostensibly red seats with retiring Republican incumbents. Of the 240 districts won by Republicans at the most recent elections, incumbents in fully 41 have already announced they will not run for re-election in November. That is more than twice as many departures as the Democrats have suffered. It is far easier to pick off an open seat than it is to oust a sitting member of Congress. Furthermore, in February the Pennsylvania Supreme Court redrew the state’s congressional districts. The old map greatly favoured Republicans; the new one is expected not to.

Our model’s best guess is that the Democrats will win 222 seats, 27 more than they won at the most recent elections. That would give them a nine-seat majority. They obtain control of the House in 65% of simulations. This probability is slightly more auspicious for the Democrats than prevailing opinion suggests. Punters on PredictIt, a political betting market, put the Democrats’ odds at 59%. We will produce a forecast for the Senate races later this year.

When applied to past elections, the model’s predictions in late May have missed the final totals by an average of just seven seats. There is, of course, no guarantee that it will perform as well in 2018. A two-in-three chance measured almost six months before the election is hardly secure enough for the Democrats to rest easy. Because this year’s race is so close, it would take only a shift of 0.6 percentage points towards the Republicans in the model’s expectations of the national popular vote for it to anoint Mr Trump’s party as the favourite to retain control.

Such movement is possible. Mr Trump’s approval rating has already defied the historical pattern of declining sharply in the first half of a mid-term year. His party could have bottomed out unusually early in polls of party voting-intention as well, which would make it unlikely to lose additional ground as the model expects. The Democrats could nominate ideologically extreme or poorly qualified candidates, or be shut out altogether in some California districts because of that state’s odd primary rules. A dramatic event could occur. That is why we give the Republicans a one-in-three chance of hanging on.

As the race changes, so will our projections. The model will update every day to reflect new data from polls, special elections, economic indicators and primaries. You can follow its forecasts and learn more about our methodology here.

This article appeared in the United States section of the print edition under the headline "Our best guess"

The affair: Why corporate America loves Donald Trump

From the May 26th 2018 edition

Discover stories from this section and more in the list of contents

Explore the edition

Discover more

Do undocumented immigrants have the right to own guns?

A federal court in Chicago decides that some do. Republicans are outraged

Chicago wants to stop Glock pistols being turned into machineguns

The city is suing the manufacturer


Georgia’s black Republicans have a battle plan for 2024

Joe Biden will have to work harder to win the state’s black voters this year