Welcome to our Election Update for Thursday, Sept. 6!
After reaching a new high of 4 in 5 on Tuesday, Democrats’ chances of winning a House majority in our forecast have fallen back to Earth a bit. Recent national generic-ballot polls have been a bit less optimistic for Democrats, and as a result, their chance of taking control of the House have ebbed to 7 in 9 (or 77 percent) in our “Classic” model.1
Shifts of a few percentage points are a pretty commonplace occurrence for our forecast. But not all districts swing in tandem, even when the only new data available is national polls (as opposed to individual district polls). That’s because some districts are more “elastic” than others.
FiveThirtyEight editor-in-chief Nate Silver introduced the concept of an “elastic state” during the 2012 presidential campaign. A state’s elasticity is simply how sensitive it is to changes in the national political environment. A very elastic state is prone to big shifts in voter preferences, while inelastic states don’t blow as much with the political winds.
An elastic state isn’t necessarily a swing state, or vice versa. Think of the difference between a state that is decided by 1 percentage point every election (an inelastic swing state) and one that votes 10 points Democratic one year and 10 points Republican the next (an elastic swing state). In other words, elasticity helps us understand elections on a deeper level. Just knowing that both of those districts are competitive doesn’t tell you everything you need to know; for example, the two call for different campaign strategies (turnout in the former, persuasion in the latter).
Today, we’re excited to unveil not only an updated elasticity score for each state, but also, for the first time, the elasticity scores of all 435 congressional districts! These scores are derived from the 2016 version of the Cooperative Congressional Election Study, a massive, 60,000-plus person survey conducted by Harvard University in conjunction with YouGov. The scores work by modeling the likelihood of an individual voter having voted Democratic or Republican for Congress, based on a series of characteristics related to their demographic (race, religion, etc.) and political (Democrat, Republican, independent, liberal, conservative, etc.) identity. We then estimate how much that probability would change based on a shift in the national political environment. The principle is that voters at the extreme end of the spectrum — those who have close to a 0 percent or a 100 percent chance of voting for one of the parties, based on our analysis — don’t swing as much as those in the middle.
You can download the data, which our forecast uses to translate generic-congressional-ballot polling to individual districts, on GitHub via this link. However, here, at a glance, is the elasticity of every state (and the District of Columbia), plus the top 25 and bottom 25 congressional districts (higher scores are more elastic, lower scores are less).
|state||elasticity score||state||elasticity score|
|Most elastic||Least elastic|
|Michigan 5th||1.24||California 5th||0.83|
|Illinois 8th||1.22||Illinois 1st||0.83|
|Nevada 4th||1.22||New York 7th||0.82|
|Massachusetts 1st||1.22||Virginia 8th||0.82|
|Massachusetts 6th||1.21||California 15th||0.82|
|Massachusetts 2nd||1.21||California 28th||0.82|
|New York 21st||1.21||Georgia 10th||0.81|
|Florida 26th||1.20||Georgia 13th||0.81|
|Massachusetts 9th||1.20||Washington 7th||0.81|
|Florida 25th||1.20||California 37th||0.81|
|Minnesota 7th||1.19||Mississippi 3rd||0.80|
|New Hampshire 1st||1.19||New York 9th||0.80|
|Massachusetts 4th||1.18||New York 5th||0.79|
|California 26th||1.18||California 44th||0.79|
|Massachusetts 3rd||1.18||California 13th||0.79|
|Rhode Island 1st||1.17||Alabama 3rd||0.79|
|Illinois 12th||1.17||Alabama 6th||0.78|
|Texas 33rd||1.17||New York 13th||0.77|
|Iowa 2nd||1.17||Missouri 4th||0.77|
|Washington 5th||1.17||New York 15th||0.77|
|Utah 2nd||1.16||California 2nd||0.76|
|Alaska at large||1.16||New York 8th||0.74|
|Texas 29th||1.15||New York 14th||0.73|
|Maine 1st||1.15||Illinois 7th||0.72|
|Oregon 2nd||1.15||Pennsylvania 3rd||0.72|
Congratulations, Michigan’s 5th — you’re America’s most elastic congressional district! The Flint- and Saginaw-based district has an elasticity score of 1.24, which means that for every 1 percentage point the national political mood moves toward a party, the 5th District is expected to move 1.24 percentage points toward that party.2 In practice, that means the district votes differently from year to year and even within elections. For example, in 2016, it voted for Hillary Clinton for president 50 percent to 45 percent, according to Daily Kos Elections, but Democratic Rep. Daniel Kildee for Congress 61 percent to 35 percent. In the top 25 are also six Massachusetts districts and one each from Maine, New Hampshire and Rhode Island.
As a general principle, the swingiest districts tend to be those with lots of white voters who do not identify as evangelical Christians. (By contrast, white evangelical voters are overwhelmingly Republican, while nonwhite voters — with a few exceptions like Cuban-Americans in South Florida; note the presence of Florida’s 25th and 26th districts in the top 10 — are overwhelmingly Democratic.) These voters are plentiful in the Northeast, and in the Upper Midwest, where they were vital to President Trump winning states such as Ohio and districts such as Maine’s 2nd Congressional District.
On the other end of the spectrum, Pennsylvania’s 3rd District, covering downtown Philadelphia, is the most inelastic district the nation has to offer. That makes sense, given that it’s majority-African-American, a group that consistently votes for Democrats at rates around 90 percent. Seven other majority-minority districts in New York City likewise make the bottom 25. Two of the bottom 10 are in Alabama, where most voters are either African-American (reliably Democratic) or evangelical white (reliably Republican), making it very inelastic overall.
The list illustrates what I noted earlier: that competitive districts can be elastic or inelastic, and elastic districts can be competitive or uncompetitive. For example, Massachusetts’s 1st District is quite elastic (1.22), but it’s not closely pitted between Democrats and Republicans (according to FiveThirtyEight’s partisan lean metric,3 it’s 27 points more Democratic than the country),4 so when it bounces back and forth, it’s usually between mildly blue and super blue. And there are competitive districts up and down the elasticity scale: Nevada’s 4th District (rated as “likely D” by our Classic model) has an elasticity score of 1.22, Iowa’s 3rd District (“lean D”) has an elasticity score of 1.00 and Georgia’s 7th District (“lean R”) has an elasticity score of 0.85.
Keep these numbers in mind as the 2018 campaign goes on. Right now, we have Democrat Steven Horsford as a 5 in 6 favorite in Nevada’s 4th, but if the national environment sours for Democrats, Republican Cresent Hardy could make up ground in a hurry because the district is so elastic. Democrat Carolyn Bourdeaux in Georgia’s 7th, by contrast, who’s a 3 in 10 underdog, will probably need to rely on goosing turnout among her voters, in addition to a good national environment, because that district is so inelastic.