Volume 15: Gerrymandering
The Praying Mantis and Upside-Down Elephant
“It is the duty of every man, though he may have but one day to live, to devote that day to the good of his country.” – Elbridge Gerry
Gerrymandering has been around a long time. Controlling election maps is always a form of power. The term was coined in 1812. But in recent decades it has moved to the forefront in U.S. elections. What once was an art form is now a science – with profound implications for how our government is run.
It is easy to go blissfully through life, never thinking about the shape of the Congressional district you live in. But that shape determines your Representative, who makes all those critical decisions about your life. Even worse, you care about all the other districts: a Congressman from another state gets one vote on each bill, the same as yours. Who drew the lines for their district?
When politicians get to decide their own electorate, the possibility for conflict is obvious. Politics is a game of power, those who have it and those who want it. But we can hope that this power is used, at least occasionally, to fulfill the wishes of the people and improve their lives. Manipulating maps prevents citizens from having real input in the way they are governed.
What is gerrymandering?
How has it changed in the last twenty years?
How is this possibly legal?
What is gerrymandering?
Gerrymandering is the process of drawing district lines via any non-algorithmic process.[1] A gerrymander is the result of such a process. It takes just two ingredients to ensure gerrymandering: district-based representation and discretion in creating said districts. Any jurisdiction which has these two features will have discretion in the drawing of lines. Humans will impart their preferences when they do this drawing.[2]
Let’s go through a few examples, to see what is and is not gerrymandering:
In the United Kingdom, the House of Commons consists of 650 seats, each held by a single Member representing a specific geographic area. Each of these is drawn by a non-partisan Boundary Commission. This is gerrymandering.
The French legislature has the virtually unlimited ability to draw its own maps for single-representative districts. This is gerrymandering.
The United States Senate consists of two members from each State, whose boundaries are fixed.[3] This is not gerrymandering.
Israel elects 120 members to its parliament, the Knesset, based on nationwide proportional voting. In other words, if Party X receives 25% of the votes, it will receive 30 seats. This is not gerrymandering.
There is no right or “fair” way to draw district lines. As an example, consider a “state” with 60 Orange Party voters and 40 Purple Party voters, to be split into 10 districts. We could put 6 Orange and 4 Purple in each district, so each is similar to the overall population. But then the Orange Party would win all 10 seats, so that’s not representative.[4] There could be 6 districts of all Orange voters to go with 4 of all Purple, but then not a single seat will be competitive. You can split them up geographically, but how exactly? Put yourself in the seat of the decision maker and it is not at all simple.
When writing about gerrymandering, it seems to be obligatory to mention where the word comes from. Elbridge Gerry was the Governor of Massachusetts in 1812 when his party, the Republicans, took control of the state house.[5] They took this chance to draw state legislature districts to maximize their advantage. To do this, one seat in the Northeastern part of the state had a particularly odd shape. A local political cartoonist thought the district looked a bit like a salamander, and the portmanteau is still active 200 years later.[6]
It is interesting to look at other countries’ line-drawing methods, but we’ll have enough to do focusing on the U.S. House of Representatives. As a quick refresher: every ten years the U.S. does a census. Based on this, 435 House seats are split up between the states. States that gain or lose a seat will obviously have to create a new map. Other states also move their lines due to internal changes in population density. Helpfully, the U.S. Constitution tells states almost nothing about how to draw Congressional districts. It wasn’t until 1963 that the Supreme Court even clarified that districts must have roughly the same population.[7] The paragons of virtue who inhabit our nation’s fifty statehouses have been quite creative in deciding how they should gerrymander.[8]
We can use Ohio as an example of the redistricting process. Like 36 other states, Ohio’s Congressional map is created by its state legislature, the Ohio General Assembly. Both houses of the Assembly must agree on a map, which can then be vetoed by the Governor.[9] A party controlling both houses and the Governor’s Mansion can therefore do pretty much what it wants in Ohio. Republicans held the so-called “trifecta” in 2012. The map they created has 12 safe GOP seats of 16 total, despite the GOP winning just 51% of the total 2012 Congressional vote.[10] If a state is unable to agree on a map, either because of split control of the state government or dysfunction within a single party, generally a court is designated to impose its own map. This gives an incentive for the various stakeholders to compromise.[11]
There are six states that draw Congressional maps outside of their legislatures.[12] In these states, a commission of some sort decides the lines. The goal of redistricting commissions is to make the process less partisan and they have been largely successful. Most of these commissions were created via a citizen referendum process; in other words, legislators usually don’t voluntarily give up their map-drawing power.[13]
There is one more wrinkle in redistricting, but it’s a big one: The Voting Rights Act. Prior to its passage by Lyndon Johnson in 1965, equal protection of voting rights, theoretically guaranteed by the 14th Amendment, had no way to be enforced by Congress.[14] The VRA tried to remedy this in various ways, but today we care mostly about Section 2.
Section 2 of the Voting Rights Act says that election maps are not permitted to dilute the voting power of minorities. What this means in practice is complex, as the courts continue to chase a formula to protect minority representation. However, an effective way for states to avoid challenges of racial dilution is by creating majority-minority districts. But these districts do not necessarily work to the advantage of said minorities. A constituent group, split up between several districts, may be able to affect elections in each; if in a single district, they often “waste” votes, winning the election there by huge margins.
An Aside: Texas 2003 Re-Redistricting
We don’t think of it today, but for over a century Texas was a Democratic bastion. It wasn’t until 2003 that Republicans won a majority in the state legislature.
The 2002 Texas Congressional map was already GOP-friendly, having been drawn by a GOP-majority commission. But with proper encouragement from national Republicans, they wanted to take their new-found power out for a spin. Despite the paint still being wet on the existing map, they wanted to try an unusual mid-decade gerrymander.
Things got weird when Texas Democrats attempted to prevent a quorum by not showing up to the statehouse. The GOP majority responded by sending the Texas Rangers to physically bring them to the Capitol. Fifty-two Texas House Democrats and then 11 Senate Democrats fled across state lines, outside the Rangers’ jurisdiction.
This game couldn’t continue indefinitely; the new map eventually passed, and the GOP gained several new seats. Since then, Texas maps have been challenged in court almost continually. They have been defeated several times as illegal racial gerrymanders.
But these cases take years to get through the court system. During that time, the map remains active.
How has it changed in the last twenty years?
Before we look at the recent history of gerrymandering, let’s talk a bit about strategy.
The core of partisan redistricting is “packing and cracking.” In order to maximize the number of seats for your party, you first “pack” as many of the other party’s likely voters into as few districts as possible. The remainder are then “cracked” in equal portions between the remaining districts; they are likely to form a significant but insufficient minority in each. Using this strategy, you can win most of the seats without winning most of the votes. Packing and cracking is easier when your political opponents live in concentrated areas. New York’s 15th Congressional District is an example of packing; José Serrano won a smooth 95.2% in his 2016 re-election. The high concentration of Serrano’s voters made this a “normal shape” district, but the 90% margin of victory means there are lot of “wasted” votes.
Another fun redistricting trick is to play with the personal homes of the opposing candidates. Maybe the district lines shift their abode into a neighboring district that your party safely controls. Maybe two Representatives of the other party are moved into the same district, guaranteeing only one can be elected. Maybe the poor fool’s home stays in the same district, but their base of support is moved next door. Politics ain’t beanbag.
This leads neatly to the start of our historical tale. In the olden days,[15] gerrymandering usually had the practical purpose of incumbent protection. Not that the world was any less partisan, because it wasn’t, but both sides did a bit of mutual back scratching. These men[16] chatted in the House Cloak Room and were shaved at the Capitol Barber.[17] They didn’t want to push too hard to kick out their friends out. Mutual non-aggression made sense for everybody.
Incumbent protection wasn’t the best kind of government, but it at least had some decent features. Because the seats were generally less partisan, when a long-time Representative retired, there was some hope of a competitive race to replace him. True partisan redistricting gained momentum in the 2002 cycle and had taken over completely ten years later.
There are many reasons why redistricting became more aggressively partisan in the last twenty years. Greater polarization in Congress created a motive. If your party is more likely to always vote with you, and the other party against, you care more about the raw political makeup of the Chamber. As we described above, the battle for gerrymandering is a battle in the statehouses; races for state legislatures have become far more nationalized. Which is just another way to say that more money has poured in. PACs, Super-PACs, Issue Advocacy Groups and scary Dark Money have all made their presence felt in previously obscure statehouse races. The parties themselves became more organized. The GOP especially yearned for the power available to those controlling the lines. Their REDMAP project raised previously unheard-of sums to specifically target the process of redistricting for the 2010 cycle.[18]
This came together in a perfect storm of the 2010 Congressional and Gubernatorial elections. The President’s political party usually loses seats in mid-term elections. The passage of the Affordable Care Act gave the Republicans a message (if not a policy) to run on. Big money, freed after the Citizens United decision earlier that year, made its presence felt across the political landscape as never before. The Republicans were focused on the redistricting prize, and organized to take advantage. The results were stunning:[19]
Partisan Control of redistricting: 2012 Cycle
As you can see, the GOP controlled redistricting for three times as many House seats as Democrats in this cycle.[20] How many seats does the GOP hold compared to a neutral map? Well, there have been a lot of studies, but these always involve a lot of assumptions.[21] For example, if a given party won 55% of the Congressional vote in a state, you would expect it to win more than 55% of the seats. How many more depends on how the partisan populations are clustered.[22] My non-scientific review of these studies shows that academics think the GOP currently gains almost twenty seats from partisan redistricting.
Let’s see what we can find without making any assumptions. In 2012, Democrats won 50.6% of the popular Congressional vote, but only 46.2% of the seats. This disparity of 4% was basically unchanged in the 2014 and 2016 elections.[23] Said differently, the Congressional seat in the middle of the partisan scale is 3% GOP-leaning compared to the national vote total. This means that Democrats would need to win the national House vote by around 6% to have a 50% chance to win a House majority. Now, this isn’t necessarily due to gerrymandering; clustering of Democrats in cities is not an effective strategy to maximize their seats. However, these assumption-free figures are consistent with the statistical methods mentioned above: 4% of 435 is close to 20 seats. Practically, this means that House control in two of the three most recent Congressional sessions been held by the line-drawers rather than the voters.[24]
How is this possibly legal?
Let’s go through the process again. State legislatures, flying well below the radar and susceptible to out-of-state money, are the primary force in redistricting. The party that controls each statehouse (usually) can draw lines to benefit their side, irrespective of the will of the voters. This is, somehow, considered OK.
Well, not quite. There are legal guidelines and limits on the drawing of maps. Unfortunately, the case law involved is probably the most complex topic we have written about to date.[25] Also, there are several major cases pending on the topic of redistricting. So, before you finish reading this piece it will already be out of date. But hopefully you’ll be able to understand what’s changed.
Redistricting jurisprudence began in earnest in 1962, with the case of Baker v. Carr. At issue in Baker was the fact that Tennessee hadn’t redrawn legislative districts in sixty years. Population movement in the interim had left districts representing vastly different numbers of citizens (i.e. 10-to-1). The plaintiff claimed that the situation was counter to the 14th Amendment’s guarantee of “equal protection of the law.” Tennessee argued that district lines were a purely political issue, outside of the Court’s purview.
Baker was one of the most challenging cases the Supreme Court ever faced. Inability to form a majority on any position caused the case to be re-argued. After nearly a year, Justice Brennan was able to cobble together a majority opinion to create a test for what types of issues are political; redistricting fell outside of the definition. The door to the courtroom was open for those wanting to challenge district maps.[26] Baker was shortly followed by Reynolds v. Sims, which firmly established the concept of “one person, one vote.”
With the right to oppose redistricting in court established, there quickly formed two separate but intertwined types of claims: racial and political gerrymandering.
North Carolina 12th – 1992 version
Racial challenges to gerrymandering were greatly advanced, but also complicated, by the passage of the Voting Rights Act. Recall that the VRA made it illegal to dilute the voting rights of minorities, among other things. To prevent challenges under the VRA, states could usually “inoculate” themselves by creating the number of majority-minority districts deemed necessary by the Justice Department.[27] There have been many cases involving allegedly illegal racial gerrymanders, but we are going to focus on a single district: North Carolina’s 12th.[28]
The Supreme Court first looked at the 12th in 1993, in the case Shaw v. Reno.[29] North Carolina was in a tough position when conducting its 1990 redistricting process. The Justice Department required two African-American-majority districts, but it was honestly tricky to connect large minority populations across the central part of the state. So, the 12th had its first iteration, connecting several small urban areas by following Interstate Highway 85 – frequently little wider than the road itself. In Shaw, the Supreme Court set down a not-exactly-easy-to-follow guideline for the use of race in redistricting:
Usage of race in redistricting will be reviewed on a basis of strict scrutiny
However, satisfying the requirements of the VRA can create a compelling reason for state action
So, states were required to use race in redistricting, but not in the wrong way. Three years later, the Supreme Court had the 12th back on their docket, now under the name of Shaw v. Hunt. The State was now claiming that, while it did use race, compliance with the VRA was a compelling state interest, making this use legal. The Court agreed that VRA compliance was important, but said that the 12th wasn’t “narrowly tailored” to meet this goal;[30] they demanded new districts be drawn in time for the next election.
So, the redrawn district – still oddly shaped, but not to the previous extreme extent – was soon back in Court. Challengers[31] had claimed that the new map was still an impermissible racial gerrymander, and a lower court agreed. But this case had a fundamental difference from the Shaw era: North Carolina now explicitly said that the district was created with a partisan, rather than racial, intent. The Supreme Court, with Justice O’Connor shifting from the previous majority in Shaw, agreed with this line of reasoning and validated the new 12th District. The message was clear: have a partisan, rather than racial, basis for your map.[32]
Which begs the obvious question: are there rules to partisan gerrymanders, or do the mapmakers have carte blanche?
Just as with questions of racial gerrymandering, the Court first had to decide if accusations of partisan gerrymandering were something they were even allowed to decide. In Davis v. Bandemer, the Court decided that they were allowed to consider such a case. A mere twenty-three years after Baker. But there was one big problem. Remember that tricky guideline the Court uses for racial gerrymandering? Well, in Davis, the Court was unable to agree on any standard to decide if partisan gerrymander was permissible. In other words, they were against partisan gerrymandering, but didn’t know what it was.[33]
The last case in our list will be Vieth v. Jubelirer. Reaching the Court in 2003, this was a partisan challenge to the Republican-drawn Pennsylvania Congressional map. The logic of Davis was repeated in the Court’s decision – without a clear standard for partisan gerrymandering, there was nothing the Court could do. But, Justice Anthony Kennedy threw in a wrinkle.
In Vieth, four Supreme Court justices thought the case was non-decidable due to the lack of a standard.[34] Four other justices each presented a method by which the dispute could be resolved. Justice Kennedy did not like any of the proposed standards, concurring that the Court could do nothing in Vieth. However, he added a critical section to a fascinating concurring opinion:
“The plurality thinks I resolve this case with no standard, but that is wrong…If a subsidiary standard could show how an otherwise permissible classification, as applied, burdens representational rights, we could conclude that appellants’ evidence states a provable claim under the Fourteenth Amendment standard.” [35]
Justice Kennedy is still on the Court. With four other justices willing to establish a standard to place limits on partisan gerrymandering, he would form a majority.
Excessive partisan gerrymandering is bad. As we said at the beginning, there is no perfect map, and there will always be bias. But inability to achieve perfection does not mean that we must allow parties to do as they please, enshrining permanent majorities no matter the results of voting.
I’m sorry to keep picking on the Tar Heel State, but to see how gerrymandering overrules the will of the voters, we only need to look again at North Carolina. In the 2016 election, North Carolina had a nearly equal partisan split. At the Presidential level, the Republican candidate won by 3%, while a Democrat was elected to the offices of both Governor and Attorney General. But, recall that Republicans drew the maps for the North Carolina State Legislature. With the partisan breakdown roughly even, you might expect a small majority for either side. Instead, the GOP has a supermajority, large enough to override the Governor’s veto and enact nearly any legislation they want.
First, they attacked voting rights. Previously, each county’s election board consisted of one Democrat, one Republican and one member appointed by the Governor. This was bad: a Republican county might have a Democratically-controlled election board which could suppress the vote there.[36] The new method appeared fairer: the parties would alternate years to control the county boards. The only issue is that major elections occur only in even years; guess which party controls the boards in those years.
The Legislature went further, attacking the power of the Governor. For the first time, Cabinet appointments would be subject to the Legislature’s approval. The number of jobs available for the Governor to appoint also dropped from 1,500 to 425. Say what you will about patronage, but the balance would continue to be held by political appointees from the previous Republican Governor. The partisan purpose was only an inch under the surface, with Legislators saying openly that the changes wouldn’t have happened had the Republican been re-elected.[37]
Politicians shouldn’t change the rules based on who is sitting in the seat. But it is especially egregious when the changes are clearly against the will of the voting public.[38] This is what happened in North Carolina, and it was due to gerrymandering.
But, for the first time in a long while, we can hope that partisan gerrymandering will be contained. After the 2010, Republicans controlled the redistricting process in Wisconsin and put in place a clear partisan gerrymander. It was challenged by twelve Wisconsin citizens, using a different legal theory than in Davis or Vieth. Critically, the plaintiffs also proposed a clear, mathematical standard for what constitutes a partisan gerrymander. Recall that – in theory – such a standard is the only element lacking for partisan gerrymandering to receive a proper Supreme Court treatment.
In Gill v. Whitford, the District Court for Western Wisconsin ruled the 2010 map as illegally partisan, based on this standard. The case will probably go to the Supreme Court, who will decide it in the next session.
Will Justice Anthony Kennedy follow his own logic from the Vieth case, finally creating some controls around partisan gerrymandering?
[1] This is my definition. I like it because it doesn’t consider the intent of those drawing the lines. I’ll use the term “redistricting” interchangeably for a change in tone.
[2] Gerrymandering is more powerful when the number of viable candidates in a jurisdiction is fewer. A system where the single leading vote-getter is elected regardless of vote share (called “first past the post”) is especially susceptible. Almost all US Congressional and State Legislative elections have only two viable candidates and are first past the post.
[3] Don’t get pedantic on me. I’m fully aware of a few outstanding state-versus-state border disputes. I suppose there are proposals out there, like the fragmentation of California or Texas, which could be considered Senate gerrymandering in the unlikely event of their occurrence.
[4] If voters are randomly assigned, you would also expect an outcome like this. So, random districting doesn’t work either.
[5] To confuse you a bit: Gerry’s Republicans are the ancestors of today’s Democrats.
[6] Gerry is another of those fascinating persons who we are forced to deal with briefly in a footnote. He was on the Massachusetts Committee of Public Safety in 1775, thus (in part) responsible for storing gunpowder at Concord, which the British fatefully decided to seize. He signed the Declaration of Independence, but the refused to sign the Constitution. Nevertheless, he was elected to the First Congress, where he was instrumental in passing the Bill of Rights. He got caught up in the ugly XYZ Affair, nearly leading to war with France and permanently tarnishing his reputation. After losing re-election for Governor in 1812, James Madison chose him as his Vice-Presidential Candidate to replace George Clinton, who had died in office. Gerry himself died in office in 1814. Today, there is a revisionist attempt to improve Gerry’s historical image. Specifically, there are claims (that I am not able to substantiate) that he was reluctant to approve that first gerrymander. If you want more on Gerry, check out his official Senate bio. Oh, and you say his name with a hard “G”, unlike his namesake map-drawing word.
[7] This is Baker v. Carr. We’ll come back to it later.
[8] OK, not fifty. A few states have never had more than one Representative, hence no ability to gerrymander at the Federal level. But these states still do an admirable job of gerrymandering their state houses.
[9] About half of the legislature-based states allow for Gubernatorial veto. Ballotpedia has a good overview by state. Loyola Law School has more details about the process in the states, including Ohio.
[10] Total Congressional vote is a flawed metric because (among other reasons) not every House seat is contested by both parties. When thinking about partisan gerrymandering advantages, it’s important to also look at Senatorial or Presidential races, even if what you care about is the House. This makes the Ohio partisan gerrymander look even worse, as the GOP lost both races in 2012, winning 44% and 47% of the vote, respectively.
[11] Nobody wants to see two representatives in long-held safe seats forced into a court-created battle royal.
[12] Arizona, California, Hawaii, Idaho, New Jersey and Washington.
[13] Between the time I decided to use Ohio as my example and when I finished this paragraph, an Ohio ballot measure to move to a commission passed a critical step.
[14] If you want to know more about the inability to prevent voter discrimination based on race, read the introduction to Master of the Senate, by Robert Caro.
[15] Like, say the 1990s.
[16] Yes, men. As recently as 1993, men made up 93% of the House.
[17] Actually, the Cloak Rooms are partisan, but you get the idea.
[18] You can read more about REDMAP here or many other places. None of this is to claim that Democrats don’t also take advantage of redistricting when they are able. Just that, at least in 2010, they didn’t do so very effectively.
[19] From Ballotpedia. States not listed had either i) split control, ii) a bi-partisan commission or iii) a single district. The word “control” is debatable; for example, Missouri’s map was passed by the GOP-controlled legislature over the Democratic governor’s veto. However, they had to provide concessions to do so; I didn’t include it here. Washington redistricts via a commission, but said commission had a clear GOP lean; again, not included here.
[20] It’s actually even more lopsided. Massachusetts and Connecticut are Democratic enough where there is little potential for partisan gerrymandering. The brilliant maps installed by Democrats in Arkansas and West Virginia resulted in those two states having precisely zero Democratic Representatives. Such maps are sometimes called “dummy-manders.” Only in Illinois and Maryland did the Democrats gain due to their maps. Most analyses think they gained one House seat in each.
[21] A few studies on the effects of gerrymandering for you.
[22] To take an extreme example, if a state voted 80% for Party A, you would expect them to win close to 100% of the seats. This is because, unless there was “extreme clustering”, Party B just doesn’t have enough voters to win a majority in any individual district.
[23] From National Committee for an Effective Congress.
[24] We can confidently say that in the 114th Congress (2015-2017), the Republican Party would have controlled the House in any remotely neutral map.
[25] Recall that we’ve previously discussed algebraic topology and gotten super in depth in the theory of health insurance markets. But without Laura Cohen, a brilliant rising 3L at the University of Michigan Law School, even I would have been completely lost here.
[26] C-SPAN lists Baker as one of its twelve landmark Supreme Court cases. Chief Justice Earl Warren called Baker the most important case of his career; he was on the court for Brown v. Board of Education, Gideon v. Wainwright and Miranda v. Arizona.
[27] Another simplification. Due to lengthy histories of racist voting policies, most of the states in question were required by the VRA to obtain approval from the DoJ for any changes to their conduct of elections (called “pre-clearance”). So, de facto, they had to create the number of majority-minority seats they were told.
[28] No idea why, but the Minnesota State Senate has a really nice overview of the travails of the North Carolina 12th District.
[29] Actually, there was an even earlier case, but it was not important to this narrative.
[30] Strict scrutiny is applied in cases involving use of race in state action. To survive a strict scrutiny review, there must be a compelling governmental interest and the action must be narrowly tailored to meet that interest. This is a really tough standard, so when the Court uses strict scrutiny, the underlying state action is usually reversed.
[31] Interestingly, the challengers claiming racial gerrymandering were not always of the race allegedly impacted by the alleged gerrymander.
[32] Justice Thomas wrote the opinion, and his frustration at having to deal with this district is not well hidden. It begins, “This is the third time in six years that litigation over North Carolina’s Twelfth Congressional District has come before this Court.” Oh, and North Carolina again had to defend District 12 before the Supreme Court just this year. It was again struck down as an illegal racial gerrymander.
[33] In a strange twist, this was a challenge to a gerrymander created by the Democratic-controlled Indiana State Legislature. Indiana’s Democrats were supported in briefs by the National Republican Committee and opposed by the Democratic Party of California.
[34] As a reminder, the Supreme Court has nine members; in usual conditions, it takes five justices to create a decision of binding precedent.
[35] Kennedy, J., concurring in the judgment of the Supreme Court of the United States. From Cornell Legal Information Institute.
[36] Republicans did this in the 2016 election, when they controlled the Governorship. They cut the number voting precincts and schedule for early voting in Democratic counties whose election boards they controlled. Funny enough, Republican areas were not so impacted.
[37] For more on the actions of the North Carolina legislature, see Slate or The Atlantic.
[38] After all - said GOP Governor had just been rejected for re-election.