Gerrymandering in the US

An intermission from the presidential horse race

For all the problems with US presidential elections, US congressional elections are arguably worse. The US president is at least usually in line with the popular vote; the US congress never is.

The two chambers of congress, the House of Representatives and the Senate, suffer from opposite problems. The House of Representatives has one member elected by each district, but the district lines are redrawn every ten years by politicians. Thus politicians can control their own reelection by the process called gerrymandering.

The Senate, on the other hand, never redraws its district lines. Instead, each state elects two senators. The state lines are the result of gerrymandering from a long time ago, but at least aren’t under the power of current politicians. Unfortunately, that means that the Senate is never proportional to the population sizes of the states, and heavily favors rural regions.

Although the Senate is blatantly unrepresentative of the US, the House arguably has it worse. Not to put too fine a point on it, but you can tell by how Republican the House is. Although Hillary is winning the popular vote and Democrats are likely to win Senate majority, the House will comfortably remain under Republican control.

Packing, Cracking, etc.

Clearly this gerrymandering process is fairly powerful, but how does it work? If you’d like a hands-on approach, The Redistricting Game is a browser-based game that walks you through it (unfortunately it’s Flash). Gerrymandering is also illustrated in this graphic from The Washington Post:

Title: Gerrymandering, explained. Three different ways to divide 50 people into five districts. Left: 50 people, 60% blue, 40% red. Middle left: 1. Perfect representation. 3 blue districts, 2 red districts, blue wins. Middle right: 2. Compact, but unfair. 5 blue districts, 0 red districts, blue wins. Right: 3. neither compact nor fair. 2 blue districts, 3 red districts, red wins.

Partisan gerrymandering consists of drawing district lines in such a way that the favored side gets more seats than they rightly deserve. This is usually done with a combination of packing and cracking. Cracking means splitting the opposing voters into several districts where they have slightly less than 50% of the vote. Packing means gathering the opposing voters into a few districts where they have a very large share of the vote. I indicate the packing and cracking in the image below:

A modified form of right half of the previous image. I indicate the districts where the losing side has 40% of the vote, and label this cracking. I also indicate districts where 90% of the voters are blue, and label this packing.

Not all gerrymandering is necessarily so easy to detect. Another very common form is bipartisan gerrymandering, which seeks not to favor one party over another, but to favor incumbent candidates. Both parties select districts such that their reelection is virtually guaranteed. In the Washington Post image, the middle left diagram is labeled “Perfect representation”, but it looks to me like it could be an example of bipartisan gerrymandering. After all, both blue and red representatives are guaranteed to win their respective districts.

Solutions, however imperfect

The problem with countering gerrymandering is that it’s hard to define what fair districts should look like.

One solution has been to look at gerrymandered districts, and observe that they often have funny shapes. Thus, in some US states, districts are required to be compact, i.e. not funny-shaped. While compactness enforces an arbitrary constraint that limits intentional gerrymandering, there’s no particular reason why compact districts should be fair either. For example, in the middle right WaPo image, a set of compact districts favors the blue party.

In practice, compactness appears to favor the majority party, as well as the less geographically clustered party (i.e. Republicans). Basically, Democrats “pack” themselves into cities, and “crack” themselves into rural regions.

Another idea is that fair districts should preserve regions that are geographically, demographically, and culturally distinct. Along those lines, the Voting Rights Act of 1965 prevents districts from needlessly splitting ethnic minorities. The rules on this are not especially clear, and there appears to be a history of Supreme Court cases about it.

There’s a more recent attempt to characterize gerrymandering, called the efficiency gap. Basically, they look at the difference in “wasted” votes for each party. If a party loses a district, then all votes for that party were wasted. If a party wins a district, all votes above and beyond 50% were wasted.

It would be interesting to see new laws use the efficiency gap, but I have a few complaints about it. First, it only addresses packing and cracking, and not any other kinds of gerrymandering (including things I didn’t even mention). Second, the least competitive districts probably have a lower voter turnout, which would bias the measure. Third, I find it mathematically weird that a party with 60% of the votes receives 70% of the districts when the efficiency gap is eliminated.

A few states, including my own, have implemented another solution: non-partisan redistricting commissions. Obviously this doesn’t address how the commission “should” draw the districts, but at least there’s no longer the obvious conflict of interest that occurs when politicians draw their own districts.

Don’t forget to vote for members of congress!


  1. Jake Harban says

    Considering only technical workarounds and not the legal and constitutional issues in implementing them, one option would be to mandate that the party makeup of Congress match the popular vote.

    If the popular vote is 60% blue and 40% red, then the representatives must be 60% blue and 40% red. If the districts are drawn per example 2, then the blue party must choose two districts and order the (blue) winners to concede the races to their red opponents; if the districts are drawn per example 3, then the red party must choose a district it won and order the (red) winner of that district to concede the race to their blue opponent.

  2. says

    @Jake Harban,
    Yeah there are some democracies that work that way! I’m not familiar with that kind of system but hear there are issues. Usually none of the major parties achieve majority, and instead have to compete for the cooperation of third parties (sometimes by granting concessions that the majority of the country opposes). Not sure if that’s better.

  3. Jake Harban says

    I’m not sure they have the concede-races arrangement; from what I heard, the usual system is to have additional seats that don’t correspond to geographic districts that are awarded to make the proportions work out.

    That we only have two major parties is more a function of first-past-the-post voting than gerrymandering per se (although other countries with FPTP elections do have multiple major parties).

    Forcing major parties to make concessions a majority opposes to win the support of minor parties is not inherently unreasonable; most of our beliefs lack majority support. If a group consisting of 10% of the population elects 10% of the representatives to advance their 10 fringe beliefs, why not concede 1 of them to win that group’s support?

  4. wearsbellsonlegs says

    You can find two examples of proportional representation in Australia. One is the Australian Senate (federal upper house). It is nominally a states’ house, each state having twelve senators, six being elected each alternate federal election. To be elected, a candidate needs to get a quota of the votes (roughly one sixth of the total). Anything over the quota is passed on to the next member of that candidate’s party*. To decide the last seat distribution from the bottom is also needed, the votes for the candidate with the least votes being distributed to the voter’s next preference. This is an iterative process.

    Each state has the same number of senators, so our Senate suffers from the same problem you noted in the US senate.

    The system is indeed vulnerable to demands for concessions in some circumstances. If an individual or small party holds the balance of power, they have the government by the short and curlies.

    I have my doubts about how well most of us understand this system. Then there is Tasmania, where the lower house is elected by the Hare-Clark system – a system probably even less well understood.

    *We do have the option of being difficult and numbering the candidates according to our own preferences. Most don’t.

  5. Siobhan says

    Not sure if that’s better.

    It’s the system in all of the Scandinavian democracies. I’ve been watching Denmark in particular and one of the main consequences seems to be that change is much more incremental. On the one hand, that means progress is slow. On the other, it means progress doesn’t vanish in a puff of smoke after a single election, the way other systems cranking out a majority government might.

    One of my biggest fears in my upcoming elections is that our system will unfairly crank out a majority conservative government (the same way it unfairly cranked out a majority progressive government) and that the new government will repeal all the laws giving me human rights. That would be much harder to do under proportional representation.

    I think I would prefer the security of incremental, more-or-less always moving forward change than sporadic change where one election cycle I have rights and another election cycle I don’t.

  6. intransitive says

    With only two major political parties, gerrymandering in the US is simply a matter of grouping populations into wasted votes, as you describe them above. Voter disenfranchisement (i.e. republican efforts to prevent non-whites and poor people from voting) is gerrymandering that doesn’t even require redrawing lines.

    Multiparty democracies and “first past the post voting” don’t even require gerrymandering for popular parties to be shut out if voters are not concentrated in specific areas. Proportional represention and “levelling seats” (Scandinavia and Germany) render gerrymandering pointless because there’s no benefit to it.

    Even well intentioned division of electoral districts can have the effect of gerrymandering. Canada’s 338 electoral ridings for the House of Parliament have roughly 100,000 per riding in the four most populous provinces (BC, Alberta, Ontario and Quebec), where most of the power comes from. But distributing voters so evenly often requires cutting cities into sections and mixing urban with rural areas. A city might vote entirely for one or two parties yet have no representation because the rural areas dominated the vote, or vice versa.

Leave a Reply

Your email address will not be published. Required fields are marked *