Off the Kuff Rotating Header Image

Checking in on the Congressional forecast

Now that our November lineups are finalized, I thought I’d check in once again on the 2018 Congressional race forecast, from G. Elliott Morris of The Crosstab. I last wrote about this in December, at a time when the generic ballot preference was consistently showing a double-digit lead for Democrats. The polls are closer now but the Dems still have a sizable lead. Here’s how things project in Texas, according to this model:

Dist  Flip%  Margin  16 Marg  14 Marg
CD02  14.3%   -10.6    -18.6    -33.7
CD03   7.4%   -14.4    -25.1    -37.1
CD06  19.2%   - 8.7    -16.0    -21.3
CD07  49.1%   - 0.2    -11.5    -31.4
CD10  19.0%   - 7.5    -16.1    -22.6
CD14   5.5%   -13.8    -20.7    -22.8
CD17   4.6%   -14.7    -22.4    -28.9
CD21  19.3%   - 8.6    -18.6    -26.0
CD22  18.6%   - 7.7    -16.0    -33.3
CD23  86.8%     9.7    - 0.5    -15.5
CD24  26.1%   - 5.5    -16.4    -30.9
CD25  11.3%   -10.5    -21.1    -22.5
CD27   4.3%   -17.1    -23.6    -30.3
CD31  10.8%   -10.7    -19.5    -27.7
CD32  39.9%    -2.2    -12.1    -23.7

These data points are from Sunday; there are daily updates, which move things a bit one way or the other. “Flip% is the probability that the Democratic challenger will win that district. “Margin” is the difference between the projected Republican share of the vote and the projected Democratic share, so a positive number is a Democratic win and a negative number is a Republican win. (Obviously, that’s a point within a range, not a gospel truth, hence the Flip% probability.)

“16 Marg” and “14 Marg” are my additions, as earlier versions of this table had similar values. As with the Margin column it’s the difference between Republican and Democratic performance. However, while Margin compares Congressional candidate percentages, we can’t reliably do that for 2016 and 2014, since some of these races were unopposed. As is my custom, I used Court of Criminal Appeals races – CCA3 for 2014, CCA6 for 2016. This provides another illustration of my point from that post about the CD07 poll. You can’t have tighter Congressional races up and down the ballot and not have tighter statewide races. It may be that Morris’ model is wrong, and it may be that the totality of statewide polling data will make it clear that he’s being too bullish on the Dems. All I’m saying is that stuff like this has to be taken into account as well.

The differences in the margins fascinate me. For the 2014 to 2016 shift, most of that reflects the kind of turnout pattern we have been used to seeing in Presidential versus non-Presidential years lately. The effect is much more pronounced in urban areas, and in this case it was greatly enhanced by the Trump effect, with a side of demographic change and voter registration efforts. Projected shifts from 2016 to 2018 are nearly all about the national atmosphere. It’s kind of amazing to me that the district projected to be the most flippable outside the top three is CD24, which has gotten maybe one percent of the attention that even some of the second-tier districts have gotten. Maybe that’s a blind spot in reporting, and maybe it’s a non-optimized opportunity on the Dems’ part. CDs 06, 10, and 22 all had smaller 2016 margins than CD24, so maybe they’ll catch up when all is said and done.

I’ll check in on this again in August or so. In the meantime, here’s a story about G. Elliott Morris, the guy who’s doing these projections. One way or another, his work will be closely scrutinized on November 7.

Related Posts:


  1. asmith says:

    This is in line with what the experts are saying. Hurd is likely gone even in a smaller wave, Fletcher/Culberson is a tossup, and Allred/Sessions leans GOP but that’s right now. Federal Dems need to come out in droves and they just might.

    21st is in the next tier. 31st is worth watching.

  2. N.M. Horwitz says:

    George Elliott is great! I worked with him at The Daily Texan; I’ve been saying for some time to underestimate him at your own peril.