What does it take to take the House?

It's a heavy lift to win back the House, but I think we can do it. Check out my new State of the House tracker to see exactly how heavy a lift.

Let me go on the record and say that the state of representational democracy in this country is bullshit. Popular votes in the House race were roughly split, and yet Republicans control 55% of the seats – that's a 10% defiance of the popular will.  And this anti-democratic trend doesn't just hold in the House, but at the presidency (2 of the last 3 Republican "wins" were contests in which the Dem had the plurality of the votes) and in the senate (Democratic senators currently represent approximately 37 million more people than do Republicans, despite having a 2 seat "minority.")

We find ourselves on the cusp of a "blue wave," potentially one of historical significance, but will that be enough? There's been a lot of reporting about how much the Dems need and each one produces a breathless piece of clickbait at places like politico. Here's the Brennan Center for Justice saying that the Dems need to win by 11; Here's Nate Cohn at the Times insisting the number is a very precise 7.4 points; and there are lots more, all around the same level.

The truth is that under almost any reasonable assumptions, the necessary Democratic advantage isn't that tough to compute. The bad news is that it's about 10 points.  That's a lot, but with the insane enthusiasm gap that we've seen in both the polling on the generic ballot question and in the special elections, the number might be do-able.  Indeed, the prediction markets currently give about 2:1 odds of retaking the House in November,

A better question might be: Which factors (retirements, for instance) influence the House, and by what factor? Or: given how close the balance is, where should you put your resources?

The State of the House

I've created an awesome new widget to model every district (using the new PA numbers) under various factors: using a mix of the 2012 and 2016 presidential outcomes, or weighting by the 2016 Congressional outcomes.  I've connected it to a database which includes, among much else, all of the relevant electoral margins, and who has currently announced their retirements (and thus, who doesn't get an incumbency advantage).

  • Set the generic D-R ballot with the slider. It's set to a reasonable (but not overwhelming) +8 to begin with.
  • Historical results suggest that incumbents of both parties get a 7.5% built-in advantage.  While I generally wouldn't advise fiddling with this, you can adjust that assumption (for both 2016 and 2018) using the slider.
  • The database includes every announced retirement so far, but you can force all an assumption that no democrats or no republicans (or both) are running in the generals.  This will generally hurt that party.
  •  To estimate the neutral-environment lean of a district, we'll want to combine the historical performance of the district in 2016 (corrected with the incumbency) as well as performance in presidential elections.  The bottom slider tells you how much to weight 2016 (1 = weigh it exclusively) vs 2012 presidential results (corrected by the Democratic win in each: 2.1% in 2016, and 4.4% in 2012).  And the "weighting pres elections" slider tells you how much weight to give the presidential compared to the actual House results. 
  • You can search for an individual race or simply hover over the circles to see some stats for each.

Some Quick Results

Let's assume that you want to model districts evenly between past performance and presidential results (weighted 75% toward 2016, and 25% toward 2012):
  •  It takes a 10.5% national advantage for Dems to win the house (219 seats).
    • The median race (the official tipping point) is MN-08, currently held by retiring Democrat Nick Nolan.  Oddly, this seat was won by Trump by 15.5 and by Obama by 5.5. Nolan barely won in by less than a point in 2016.  This seat is actually a great example of how it's not absurd to assume that a lot of the marginal seats are winnable.
    • In and around this range, it turns out that each point of Dem advantage only correlates to about 2.5 seats.  Tihs makes sense, as it takes about 10 points to win an extra 25 seats.
  • Some specific Races (but don't get hung up on details):
    • PA-17: Conor Lamb is poised to win his new district (PA-17) by about 8% against  fellow incumbent Keith Rothfus (the redistricting means that there are a couple of incumbent vs. incumbent races).
    • WI-01: Unfortunately, Paul Ryan is poised to trounce "Iron Stache" Randy Bryce by about 16 points. Though perhaps this will be a race where concerted effort will make the difference.   Or we may need to content ourselves with simply taking his gavel.
    • CA-22: Traitor and Trump apologist Devin Nunes is also set to win the seat by around the same 16 point margin. 
  • If every congressperson decided to retire (or was successfully primaried), then the necessary margin shifts to about 5.5 points.  
    • In other words, only half the advantage is due to gerrymandering. The other half is due to incumbency advantage.
    •  For every 4 retirements (on average), this means that Dems can relax their advantage by about 0.1%.
 There is one important detail that isn't included here and doesn't seem to be included elsewhere.  Evidence from the Special Elections suggests that while Dems overall have had about a 11.5 advantage in a "neutral" district, the Trump Effect is stronger the more Republican the district is, which means that Dems are pretty consistently flipping 15 point R seats.  National D-R lean isn't just a single number – it really does affect different races in different ways.