The Elephant in the Voting Booth

Why Redistricting, Turnout, and “The Big Sort” Make These Midterms Tough for Dems


It’s September, and that means the midterm elections are just over a month away. These elections will decide the fate of Congress, which is currently divided — Democrats control the Senate and Republicans control the House. Perhaps you’ve heard the doom and gloom about the Democrats’ chances? If these developments confuse you, political data expert Amelia Showalter is here to answer your burning questions.


Q: I don’t get it. Why does nobody think Democrats can win back the House this year, and why are people now predicting they might lose the Senate? I heard that the demographic trends were all in the Democrats’ favor. There are more registered Democrats than registered Republicans. And didn’t I hear somewhere that more people actually voted for Democratic congressional candidates in 2012? If that’s true, how did Republicans still retain control of the House? And how can the Democrats have such poor prospects this year?

A: The answers are complex, but you’ll get a lot of mileage by understanding three peculiarities of American politics: redistricting, turnout disparity, and a demographic trend sometimes called “The Big Sort.” These trends can explain a lot about how Republicans can win majorities in Congress despite, yes, receiving fewer actual votes than Democrats nationwide.

Q: Okay, so what’s “The Big Sort” all about?

A: This term, coined by author Bill Bishop, refers to the fact that Americans are increasingly sorting themselves into concentrated areas of like-minded people. This doesn’t mean everyone in your neighborhood looks exactly the same. In fact, Bishop cites evidence that our country’s various racial and ethnic groups are actually becoming slightly more evenly distributed geographically (there’s still lots of segregation, of course, but it’s lessening just a little). But even though demographics often correlate with politics, demographic integration does not necessarily lead to political heterogeneity. Think, for instance, of Washington, DC. The city has become considerably more white in recent years, but no less Democratic.

Some of the Big Sort is due to people actually moving to places that align with their politics, and some of it due to people shifting their political beliefs to better match the people around them. In either case, we are getting increasingly concentrated pockets of Republican and Democratic voters in various parts of the country. Even as the national popular vote has stayed relatively close to 50/50 in recent elections, we have lots of areas that are completely safe for one party or the other.

But a weird thing has happened with these concentrated pockets. Democrats tend to be a little more concentrated than Republicans. That is, Democratic areas of the country (especially big urban centers) are often really REALLY Democratic. You’d think this would be good news for Democrats, but the flip side of this is that while Republican enclaves are not quite as concentratedly Republican, they are still largely unwinnable for Democrats.

This ends up mattering a lot to Congressional politics, because the members of the House of Representatives are all elected from single-member districts. That is, instead of a parliamentary system where seats are allocated to each party according to national popularity, we have 435 winner-take-all districts. If a Republican candidate wins her district by a few votes or a few hundred thousand, it doesn’t matter; Republicans get to count that whole district in their column.

Perhaps it’s best to illustrate this with a little thought experiment. Let’s remove ourselves from U.S. Politics for a moment and consider the made-up country of Rectangula. Let’s say that Rectangula has a stable population of 1000 people, with 100 people in each of ten national legislative districts. Let’s say that typically there are about 500 people who vote for the Red Party and 500 people who vote for the Blue Party, and initially those partisans are evenly distributed across all districts. Like this:

In this system, all districts are at the 50/50 mark between the two parties. This causes every district to be extremely competitive, as each party works hard to convince a few of its rival’s softer supporters to come over to the other side. Party control in Rectangula at large may shift back and forth between the Red and Blue parties, but it’s always close — just like the actual makeup of Rectangula.

Let’s suppose, though, that Rectangula undergoes a Big Sort. Blue Party voters decide they really like living on the eastern shore of Rectangula, whereas Red Party voters start coalescing into smaller enclaves scattered throughout the country. It might look like this:

Each of the ten districts still has 100 people, so there’s no need for redistricting. And the country as a whole still has about 500 Blue Party voters and 500 Red Party voters. But now there are two districts where the Blue Party gets around 80% of the vote, and eight districts where the Red Party gets 57–59% of the vote. The Blue Party is virtually assured of winning its base districts, which is nice. But it’s also pretty hard for the Blue Party to win anywhere else. Sure, they can occasionally work really hard to convince some red voters to switch side, and maybe win three seats instead of two. But they don’t have a reasonable chance at winning enough seats to control the Rectangulan national legislature. In this system, the Red Party usually wins eight of the ten districts (80%) despite enjoying only about 50% popularity.

Q: Okay, that makes sense. Thanks for the thinly-veiled representation of Republicans and Democrats in that example. But to get back to the real world, how does redistricting matter in all this?

A: Redistricting can sometimes compound the effects of the Big Sort, and it is indeed a factor in skewing Congressional results towards Republicans, at least at this point in time. Redistricting happens after each decennial Census, and the redistricting process is often controlled by state legislatures. Republicans, who had already been doing reasonably well at the legislative level in the mid-2000s, did fantastically well in 2010 elections (amid an anti-Obama backlash and the usual midterm apathy from Democrats). This gave Republicans control over the pre-2012 redistricting process in many states. Not surprisingly, this led to many instances of gerrymandering and Democratic incumbents being drawn out of their districts. (Lest anyone accuse me of excessive partisanship, Democrats have also done all these things at various points in history – we just didn’t have as many opportunities this time around).

But it’s important to remember that gerrymandering and other shenanigans can only happen when the population is unevenly distributed (because of things like the Big Sort). After all, if the left- and right-leaning population were distributed in a totally uniform fashion throughout the country, it would be nearly impossible to draw districts that pack Democrats into some districts and Republicans others. With complete heterogeneity, we’d have a whole lot of competitive districts. This is why some argue that the Big Sort is a more important phenomenon than redistricting. (It’s also why Senate elections — which are based on state lines and therefore aren’t subject redistricting — are still subject to some of the same pressures as House elections.)

In practice, redistricting and the Big Sort are inextricably linked. Let’s return to our Rectangula example, where the Big Sort has given the Red Party solid control in the national legislature. After the Rectangulan Census, the Red Party redraws the district lines like this:

Each district still has 100 people. But now there is one district where 100% of the voters lean towards the Blue Party, six districts where the Red Party still enjoys a 57% advantage versus the Blue Party’s 43%, and three closer districts where the Red Party typically gets around 52% or 53% of the vote. It’s a little easier for the Blue Party to make a play for these three districts, but they are still definitely at a disadvantage. All things being equal, the Red Party now typically wins nine out of ten seats (90%), even though the overall Rectangulan public is still a 50/50 split between Red and Blue voters.

Of course, things in the United States are not this extreme. But it helps explain why redistricting is a big part of why Democrats aren’t expected to regain control of Congress until after the next Census (and even then, there are no guarantees – a lot depends on who wins big in 2020).

Q: Okay, what about turnout?

A: If you go purely by party registration, there are more Democrats in this country than Republicans. And when people poll the whole voting-age population, Democrats and their policies tend to do quite well. But voting is not compulsory in this country – and lots of people aren’t even registered to vote. Even in Presidential election years, less than 60% percent of the voting-age population actually votes.

But in midterm elections (i.e. 2014, 2010, 2006, etc), turnout is even lower. And unfortunately a lot of the groups that Democrats depend on (racial and ethnic minorities, young people, unmarried women, etc) are the most likely to stay home during midterm elections, relative to Presidential years. The categories that Republicans tend to do best in (old, white, married, etc) are also the categories that reliably turn out at high rates, even in midterm elections. This doesn’t mean Republicans always win the midterms; Democrats did quite well in 2006 amid an anti-Bush backlash. And Democrats have been making great strides in the science of turnout in recent Presidential elections. But there is a fair amount of skepticism that Democrats, even with our mobilization prowess, will be able to overcome all the countervailing forces against us. (Personally, I must recuse myself from speculation – in my career in politics I’ve worked with too many smart progressive operatives to count us out.)

Q: I still don’t get it. Everyone I know is a Democrat! How can the Republicans be doing so well?

A: This is exactly what happens when people cluster into like-minded groups! Let’s say you’re a young, unmarried woman living in a large city in a liberal state. Sure, you may think your set of friends is pretty diverse. But thanks to the Big Sort phenomenon, you may only rarely encounter people whose political opinions are distinct from yours. And since you and your like-minded friends are clustered rather densely into your urban district, it was easy to draw you into a highly Democratic district (and to draw even more safe Republican districts out in the suburbs and rural areas).

Q: That’s not true. My district is actually very competitive.

A: Even within competitive districts there can still be concentrated pockets of Democratic and Republican voters. Your own particular social networks can be quite politically skewed even if your overall district isn’t. But, hey, you’re now a highly sought-after voter this year! Congrats! And if you’re in a state with a competitive Senate race this year, you are also a prime target. Just make sure you’re registered to vote, and be sure to mark Tuesday, November 4th on your calendar. There’s evidence that when people make concrete plans to vote (learning the location of their polling place, blocking off some time on their calendars, etc.) they actually follow through on those plans and vote at a higher rate.

Let’s make this the midterm election where all voices are heard in equal strength — including young people, unmarried women, racial and ethnic minorities, and everyone else in this big crazy-quilt of a country we live in. Vote!

Amelia Showalter is a data expert and political consultant who served as Director of Digital Analytics on the 2012 Obama campaign. She lives in Washington, DC, and tweets at @ameliashowalter.