A Democratic Majority
Our 2018 House Predictions show the Democrats have a 95.3% chance of winning the House.
Here we present three models, each varying how gerrymandering will effect Democratic seat outcomes (scroll down for methodology).
The first model assumes Democrats need to win the House popular vote by 8.5 points to be considered even money for the House majority. In 2016, the decisive House district favored the Republican Party by 5.5 points -- model two uses that as the popular vote margin by which Democrats need to win to have a 50 percent chance of winning the House.
Lastly, we create a 2018 House prediction that assumes no gerrymandering effect outside of that baked into previous elections.
A Large Gerrymandering Effect
The below graph assumes the Democrats must win the House popular vote by 8.5 points to win 218 seats. Such a number reflects analyst estimates of Democratic competitiveness given high levels of geographic sorting, gerrymandering, and straight-ticket voting. The heavy line shows the mean prediction with shaded areas showing the 95 percent confidence interval based on the model’s standard deviation. 95.3% chance.
A Small Gerrymandering Effect
Below, we assume the Democrats must win the House popular vote by 5.5 points to win 218 seats. In 2016, the median House district favored Republicans by 5.5 points, so to be a tossup, Democrats would have to win nationally by that amount. 99.9% chance.
No Gerrymandering Effect
Here we assume no gerrymandering effect, the least likely of all outcomes. This model simply uses history as a guide and ignores the districting thicket. 100% chance.
How the 2018 House Predictions Work
The model regresses the number of Democratic House seats won in a given election on these three variables across 41 House elections:
- Forecasted congressional popular vote (drawn from adjusted generic congressional polls)
- The number of seats the Democratic Party won in the preceding House election
- Whether it's 1948 (an outlying year best explained by adding a dummy variable; see Cuzan and Bundrick)
Intuitively simple, it explains 95% of the Democratic seat variation across the elections with a RSME (root mean square error, or the model's error) of 6.921.
As mentioned above, the model assesses a large gerrymandering effect, a small one, and no redistricting effect.
The large gerrymandering effect adjusts the model such that Democrats have a 50 percent chance to win the House only when they win the popular vote by 8.5 points. It takes, on average, 20 seats from the Democrats. We vary this for each iteration (see below) to account for uncertainty around the assumed effect.
The small gerrymandering model assumes a 5.5 point margin of victory -- equivalent to the median House district's Republican tilt -- will give Democrats a 50/50 chance to take the House. This typically leads to the model taking 9.3 seats from the Democrats (also changed for each simulation).
The following steps outline how I generate the 2018 House predictions:
- Forecast numbers from each new generic Congress poll
- Run the model at least 40,000 times, with each iteration using a randomly selected point from a normal distribution centered around the mean forecasted popular congressional vote result
- Generate the mean seat graph
- Populate the histogram showing seat likelihood
Since one variable -- the popular House vote -- cannot be known ahead of time, the model must address the inherent uncertainty in the popular vote forecast (which can be best interpreted as a range of likely outcomes). To do so, each simulation uses a different popular vote number drawn from a normal distribution centered at the mean forecasted and with a standard deviation of the popular vote forecast's RSME. It does the same for gerrymandering effects, all normally sampled from a specified mean.
Those outputs, at least 40,000 of them generated with each new generic Congress poll (again, see the tracker), populate the seat distribution graph whereas the first shows the mean Democratic seat output as well as the margin of error from the model itself -- it does not include results expected to occur around 5 percent of the time.
The model does not look at individual races. Doing so proves harder given the local nature of many House races, the inconsistency of time-series data (redistricting renders unusable some invaluable variables), and other difficult to forecast variables (eg, straight-ticket voting and levels of geographic sorting that polarize districts and leave fewer and fewer uncompetitive). It's entirely possible that while Democrats currently have a large forecasted popular vote lead, a leading indicator of House seat gains, their actual benefits may be less than expected because geographic sorting has created too many safe seats unaffected even by a large swing.