If Trump’s approval rating stayed at 40 percent in 2018, Republicans would be expected to lose the national House popular vote by 10 percentage points (the GOP won it by 1 point in 2016). That’s a shift of 11 percentage points from 2016. That size loss would probably be big enough that Republicans would lose control of the House, gerrymandering and urban packing notwithstanding. But that projection comes with a margin of error of about +/-8 percentage points. That could mean an even a larger win for Democrats, which wouldn’t be too surprising given the enthusiasm we’ve seen from Democrats in marches and town hall meetings. Or, a much smaller one.
Recent midterm elections have been tough on Democrats. Compared to years with a presidential election, the electorate in midterm cycles has been older and whiter. Both of those blocs lean Republican. So surveys of all adults — such as Gallup’s — could be underestimating Trump’s popularity with the 2018 midterm electorate (if it looks like recent midterm electorates). Additionally, Congress’s approval rating seems to be up. And while presidential approval ratings are a larger driver of vote choice, congressional approval ratings also matter in midterm elections.
So if Democrats win the national House vote by a margin in the low- to mid-single digits, that may not be enough to take back the House. The median congressional district was 5.5 percentage points more Republican-leaning in the presidential race than the nation as a whole in 2016, meaning Democrats are essentially spotting the GOP 5.5 points in the battle for control of the House. And even that may be underestimating Republicans ability to win a majority of seats without a majority of the vote. Since 2012 (or when most states instituted the current House district lines), Republicans have won, on average, 51 percent of the two-party3 House vote and 55 percent of House seats. If that difference holds for 2018, Democrats would need to win the House popular vote by about 8 percentage points to win half the House seats.