House forecast: Democrats will win 226 seats (and the House majority) while Republicans will win just 209 seats. A Democratic win of 202 seats and 260 seats is within the margin of error.
Senate forecast: Republicans will hold 51 seats (and maintain control of the Senate) next Congress while Democrats will hold just 49. Anything between Republicans holding 47 seats and 56 seats is within the margin of error.
Republicans are trying to hold the line in the House. They probably won’t be successful. The forecast that I created with Parker Quinn and Brice Acree currently gives Democrats a 226 to Republicans 209 edge after the midterm elections.
Yet, it’s not as if all hope is lost for Republicans at this point. Indeed, I would point out that it’s important to watch the margin of error. It still gives Republicans a decent chance of maintaining control.
Although we haven’t generally presented a percent probability, the model gives Democrats about a 72% chance of winning control of the House. That’s among the lower chances that most models give, though it’s in the general ballpark. You know who else had about a 72% chance of winning? Hillary Clinton in 2016.
Indeed, it’s not too difficult to paint a scenario in which the Republicans maintain control. Let’s say Republicans win all the seats where they are favored as well as the seats where Democratic candidate leads by about a point or less in our forecast. That would leave Republicans with 220 seats to Democrats with 215 seats. Now, that’s not the most likely scenario, but it’s on the table.
You could also imagine a situation in which the polling is off. Let’s say there’s a systematic polling error, and Democrats only win the House popular vote by 5 percentage points instead of the 8 or so they’re currently projected to win by. That too could keep Democrats out of the majority. In the 2014 midterm, such an error did occur.
Perhaps most intriguing is what happens if traditionally Republican voters who are thinking of voting Democratic decide in the final days of the campaign that they’re going to vote Republican after all. There’s little doubt that President Donald Trump is hoping that will happen given the rhetoric he’s used in the last week to rev up the base.
Right now, our forecast has 13 districts (including 9 with Republican incumbents running) where the Democrat is favored to win and the weighted average partisanship (i.e. our measure for how the district has voted in all sorts of elections over the last six year) is greater than 8 points for Republicans.
That means even in a national environment favoring Democrats, these are districts, on average we would think would go Republican. If the 11 of these districts that are currently projected to go Democratic by 4 points or less go the other way, Republicans would hold onto a bare majority.
Of course, it would be absolutely the wrong takeaway from 2016 to assume that forecasts are underestimating Republicans.
Rather, the correct takeaway is that forecast can be off in either direction: favoring Democrats or Republicans.
I’d actually argue it’s slightly more likely that our forecast is underestimating the Democrats. My former colleague Nate Silver at FiveThirtyEight has a “deluxe” forecast model that gives Democrats 230 or 231 seats, depending on whether you use “mean” or “median” forecast. In other words, looking at this data, chances are we’re on the low end of estimates.
Further, if you look at the polling this cycle, it’s actually the Democrats who have been outperforming their numbers, not the Republicans.
My guess though is that if our forecast is off by 9 seats and the side that is already favored does better (the Democrats), it won’t be remembered as infamously as an error in which the side that is the underdog (the Republicans) does better.