Greetings from the future. I come with some very promising prognostications.
I wanted to give you a first look at a little project that I'm working on to measure the Drumpf effect. What you see above is data from contested special elections last year (in red) and this year (in blue). These are all state house and senate races, and there have only been a few dozen. Incidentally, if you want the complete data set (with links) here you go. I’ll keep the spreadsheet up to date moving forward (and if you have corrections or additions, let me know).
Essentially, I’m tracking all contested elections which have also been contested in recent years (no gimme seats are included). For some, I have data from the 2016 presidential election in that district as well, so we might reasonably guess that the expected outcome for a district is simply the average of the last time time the seat was up, and the presidential race. Here’s a quick summary:
State |
Race |
Result (D-R) |
Expected |
Drumpf Effect |
VA |
VA |
IA |
VA |
MN |
DE |
CT |
CT |
CT |
|
State Senate (22) |
State House (85) |
State House (89) |
State House (71) |
State House (32b) |
State Senate (10th) |
State Senate (2nd) |
State Senate (32nd) |
State House (115) |
|
-14.5% |
-5.8% |
45.0% |
83.3% |
-6.4% |
17.0% |
66.7% |
-1.6% |
20.0% |
|
-16.6% |
-22.8% |
35.0% |
75.9% |
-21.0% |
7.5% |
67.5% |
-28.0% |
38.0% |
|
2.1% |
17.0% |
10.0% |
7.4% |
14.6% |
9.5% |
-0.8% |
26.4% |
-18.0% |
|
|
|
|
|
|
Even scanning by eye, Dems seem to be outperforming expectations. The mean is a 7.6% improvement over expectations; the median, 9.5%, and a weighted mean (by number of voters): 6.2%
No matter how you slice it, Democrats seem to be doing much better in 2017 than you might naively supposed. The conventional wisdom says that Dems are supposed to suffer in low-turnout elections, and we can test this by looking at the contested races in 2016 (in red at the figure at the top). The mean and median in that case, both had the Republican outperforming by about 7 points. In other words, from 2016 to 2017, Dems saw their performance in special elections swing by about 15-16 points. Here’s that graphically:
The big points represent the median and uncertainty in the mean in the errorbars.
A swing of 15 points toward the Dems is a HUGE effect. It's the sort of thing that would easily win us the House, and keep losses in the senate to a minimum (maybe allowing us to pick up NV, AZ or possibly TX -- but let's not get greedy). It is a nightmare scenario for Republicans, and so far, the data suggests that that's where we're at.
Now here are the caveats:
- While Dems have dramatically overperformed expectations, no actual seats have been flipped yet. It's still early in the year, and hopefully Georgia's 6th, next month, will be an exception.
- It's early, and we're energized. This advantage may go down with time.
- Turnout in these races is awful. We're talking 10-20% in many cases (as was the case earlier this week in LA -- even though the good guys won, we need to get people in booths). We're not taking back the house with turnout like this. Vote for freaking dogcatcher if they are holding an election.
I'll update this as it progresses, and am looking to dialog with anyone who has statistical thoughts or questions. But in the meanwhile, I wanted to give a note of hope for the future.