With Election Day five days away, I thought I’d share how I’m spending my time to help elect Democrats. In selecting a place to contribute remotely (in-person canvassing is ideal, but I’m doing that for local candidates in San Francisco), I wanted to find a place where I could move House, Senate, and Governor races most efficiently with text banking, which I’ve found more effective and engaging for both parties than phone banking.
Thanks to FiveThirtyEight’s forecast, I could methodically pick such a place. By downloading and playing with their data, I scored states based on concentration of competitive House districts, per-voter Senate race competitiveness, and Governor race competitiveness. Each voter reached in this state can potentially swing the outcome of three important elections: two that affect federal legislation and checks on the President, and one that affects redistricting in 2020 and other matters like voting rights.
The state with the highest score that also has texting capabilities is Nevada, and I’ll start texting with them (between canvassing) tomorrow.
I used FiveThirtyEight’s House, Senate, and Governor models, each of which provides a probability of Democrats winning each race. Here’s the code that downloaded this data (I used the Classic forecast model as of November 1) and produced my final competitiveness score, a combination of House, Senate, and Governor competitiveness metrics.
The simplest step was the Senate. For this, the FiveThirtyEight team creates a Voter Power Index, which is “the relative likelihood that an individual voter in a state will determine the majority party.” North Dakota’s combination of a close election and low population (with 755,000 people, it’s the 4th-least-populous state) gives each voter there the greatest chance of flipping the Senate. North Dakota doesn’t have text banking, but this makes phone banking for incumbent Heidi Heitkamp a good use of time for Democrats most interested in flipping the Senate.
There’s no Voter Power Index for the House, so instead I assigned each race a competitiveness metric defined as distance between 0.5 and Democrats’ probability of winning.¹ I then aggregated this to state by taking the average of competitiveness scores across all the states’ districts. Montana and Alaska are the most competitive on this metric: each has a single House district, where the Democrat has a 36 and 29 percent chance of winning, respectively.
I computed Governor competitiveness the same way as for each House race: the distance between 0.5 and probability of winning. As with the Senate, states without a Governor race got a zero. Ohio and Georgia are the most competitive Governor races by this metric, with Democrats having a 50 and 43 percent chance of winning, respectively.
After normalizing each of these metrics to fall between zero and one, I averaged them to produce the competitiveness score.
Alaska turned up the top using this approach. While Alaska doesn’t have a Senate race, it makes up for it with its close House race and Governor race, where Democrat Mark Begich has a 34 percent chance of winning. Unfortunately they don’t seem to offer text banking, or even remote phone banking. If you have a way to help them out, I recommend doing so.
Close number 2 Nevada is competitive all around: Democrat Jacky Rosen has a 44 percent chance of ousting Republican incumbent Senator Dean Heller (and each voter has more power than anywhere outside North Dakota); Democrat Steve Sisolak has a 40 percent chance of beating Republican Adam Laxalt for the governorship; and in its four House races, Democrats have 5, 74, 83, and 99 percent respective chances of winning (not toss-ups, but still quite uncertain).
Click here to text bank for Nevada Democrats and potentially swing multiple races. Right now, they’re only taking sign-ups for Friday, but I expect other opportunities to pop up through the weekend.
If you want to do something today, you can sign up to text with Texas Democrats, where despite Republican Greg Abbott’s shoo-in for the governorship, Beto O’Rourke has a 21 percent chance of beating incumbent Senator Ted Cruz, and 10 of their 36 House seats are closer than “solid” (95 percent chance in either direction). I’ve been enjoying texting for them.
To reach a wider set of voters, I also recommend signing up with Resistance Labs’ Text Out The Vote initiative, which will be texting voters across the country for many Democratic campaigns. I texted with them for London Breed’s June mayoral campaign, and have found them exceptionally organized, for example using Slack to onboard and answer questions.²
If you’re reading this, chances are your Facebook friends have already seen many calls to vote, and they’re probably doing so. In these last few days, you can make a greater impact by reaching people who have heard less about the importance of voting, who have less information about candidates, and who could use help finding information on how to vote. Please join me in taking time to share your knowledge and time with these people. It could make the difference between more Republican control and a badly-needed balance of power.
 This implicitly assumes that Democrats have a 50–50 shot at the House. A potential improvement would be adjusting for the fact that Democrats have an 85 percent chance of winning the House. If Democrats are in the range of losing the House — for example due to changing conditions or systematic polling errors of a couple points — races with a 50 percent chance of winning would then have a lower win probability, so it might be better to focus on races near the median win probability (currently around 58 percent). A real House Voter Power Index would be ideal, so without it I made this simplifying choice to be consistent around 50 percent for both the House and Governor races.
 As much as I may disagree with their emphasis on passing California Proposition 10.