Will the 2018 House Elections Be 2006 All Over Again?
Will Democrats win control of the House of Representatives in 2018?
That’s the million-dollar question for election wonks. If Democrats take the House, they will be able to block President Trump’s agenda in the same way House Republicans blocked President Obama’s after the 2010 elections. And since Democrats would have to pitch a near-perfect game to take control of the Senate (the map heavily favors Republicans), the House will likely remain the main event in national electoral politics for the next year and a half.
Most analysts take one of two routes to assess which party will likely win the lower chamber. Typically we either analyze generic ballot polling (surveys that ask voters which party they’ll vote for in their district in the upcoming House election) or assess, on a district-by-district basis, which seats are likely to flip from one party to another. There are advantages and disadvantages to each approach (we’ve done both at RCP and will continue to do so), but a slightly different method also bears consideration.
Specifically, I looked at win rates for both parties for various types of districts in past elections, applied those to the 2018 congressional map and attempted to figure out exactly what a Democratic win in 2018 might look like. I found that Democrats might be able to win by running a 2010 strategy while getting 2006 numbers in the battleground districts. By that I mean that Democrats could win the House by grabbing swing and light-red districts without reclaiming the conservative, ancestrally Democratic areas that padded their numbers in the not-so-distant past.
Running a 2010 Strategy With 2006 Win Rates
To understand exactly how this might work, we need to rewind about a decade. In 2006, George W. Bush's approval rating was being dragged down by the Iraq War, the aftermath of Hurricane Katrina and other problems, so voters gave Democrats control of the House. The shape of their majority was, at that time, familiar. They won almost every district that leaned far to the left in presidential elections, took seats in a large number of swing and light-red districts and held a few conservative, ancestrally Democratic districts where voters typically split their tickets between Republican presidential candidates and Democratic congressmen. In 2008 Democrats increased the size of their majority mostly by winning more swing and light-red districts, but in 2010 something important changed. Republicans took back the House in part by winning back swing districts, but also by beating Democratic incumbents in extremely red districts. Some of these Democrats voted for the Affordable Care Act and many fell victim to the nationalization of politics -- that is, voters identifying down-ballot candidates with their national party and voting accordingly.
That shift makes it more difficult for Democrats to return to their former glory in places like North Dakota, South Dakota, Mississippi, rural Georgia and other strongholds of the past and changes the playing field. In other words, a Democratic path to the majority in 2018 might look like the Republican path in 2010 -- winning every district that leans substantially towards their party on the presidential level, doing very well in swing districts and winning a large number of districts that lean slightly towards the other party while leaving the far reaches of the map more or less untouched.
But to get there, Democrats may need to run like it's 2006 again (sort of).
To reach this conclusion, I looked at the rate at which Democrats won districts with varying presidential leans in 2006 (measured as the weighted average of how much more Republican or Democratic a district voted than the nation, with three quarters of the weight given to the most recent election) and applied it to the 2016 presidential map. For example, I looked at the rate at which Democrats won districts that leaned two-to-four points more toward Republican presidential candidates than the nation in 2006, multiplied that by the number of districts in the 2016 map with the same partisanship and repeated that procedure for all districts. This is similar to the method Sean Trende used in 2013 to assess the effects of gerrymandering on our politics.
I then assumed Democrats wouldn't win any districts that leaned more than 10 points away from their party. The idea behind this step is simple -- if American elections are so polarized that certain districts are simply out of reach for Democrats, then their win rate there should be zero (note that the results of this calculation are somewhat sensitive to the ceiling chosen -- if all R+8, instead of R+10, districts are out of reach for Democrats, then the results below change).
The result? The 2006 Democrats (sans the most conservative members) barely won the chamber. You can vary the exact details of the 2018 calculation (e.g. calculate win rates for districts with leans that are within two points, three points, 1 ½ points, etc. of each other), but Democrats typically won around 220 seats -- just a couple more than needed to retake the House.
Performing the same procedure with 2008 results (applying win rates and eliminating Democratic wins in highly conservative districts), Democrats perform better, getting close to or above 230 seats, depending on how the calculation is tweaked. This makes sense -- some of Democrats’ gains between 2006 and 2008 were in the ideological “center” of the map, and Democrats won the House popular vote by a larger margin in 2008 than 2006.
These numbers clarify what a “2010 strategy with 2006 numbers” actually means. If Democrats can keep their safe districts and drive up their win rate in swing-to-right-leaning districts, they may be in a position to take back the chamber.
Is This Strategy Realistic? And How Will We Know If It’s Happening?
The main drawback of thinking about House races in this way is that it’s hard to quantitatively gauge whether 2018 will be like 2006. Conversely, if an analyst says that Democrats needs a certain percentage-point lead in the generic ballot polling to take the chamber, you can check that number against the RCP Average. Or if someone says that Democrats need a specific number of races to be rated as “Toss-Up” by some handicapper by a certain time, you can check if that benchmark is being met.
But there are some signs that could help election watchers gauge if we’re in 2006 again. Incumbency is a real advantage in congressional elections, so more GOP retirements would be a good sign for Democrats. A continued low job approval rating for Trump, coupled with poor GOP showings in generic ballot polls would also signal trouble for the GOP. Additionally, if races that we don’t currently expect to become competitive start to look that way (i.e. district-by-district polling shows tight races, parties unexpectedly start spending resources there, etc.) that might be a sign that it’s 2006 again.
In other words, if polling and other factors create this sort of “time travel,” that’s a good sign for the Democrats. Because if the calendar gets turned back, they will be at least competitive for control of the House.