What is Extreme Gerrymandering?

Understanding how extreme partisan gerrymandering works.

September 5, 2017
history of gerrymandering

Figure 1:  Tisdale, Elkanah. “The Gerry-Mander,” Boston Gazette, March 26, 1812.

What is extreme gerrymandering?

Gerrymandering is the term used in the United States to describe the intentional manipulation of district boundaries to discriminate against a group of voters on the basis of their politics or race.

The term dates to 1812 when Massachusetts Governor Elbridge Gerry signed into law a redistricting plan that included a district that many thought looked like a salamander, leading opponents to nickname the district after him.  

But while the term has become a synonym for redistricting abuses, it actually covers a wide variety of sins, not all of which are related. 
For example, one form of gerrymandering involves making a district super safe for an incumbent. Likewise, sometimes districts are drawn so a powerful lawmaker’s brother-in-law or another favored candidate can successfully run for office. These types of gerrymanders – which often occur through bipartisan collusion between political parties – can be harmful to democracy by pre-determining outcomes and depriving voters of a meaningful choice. 
But Gill v. Whitford involves another variant of political gerrymandering that is even more pernicious. 
In the type of extreme gerrymandering being challenged in Whitford (and in partisan gerrymandering cases in North Carolina and Pennsylvania), a political party uses its control of the process to artfully craft maps that lock in an outsized share of seats. For an example of this type of gerrymandering at play, take a look at North Carolina. At a statewide level, North Carolina is a robust democracy with highly contested elections for everything from president to state auditor. In some cases, Democrats win, in other cases Republicans do. But when it comes to its state legislature and congressional delegations, the exact opposite is true. In the state’s congressional delegation, for example, Republicans enjoy a safe and durable 10-3 advantage. 
Gerrymanders of this sort are harmful of not only because they bake in results but because they result in maps that are deeply unrepresentative. John Adams famously wrote in 1788 that the House of Representatives – and by extension state legislatures - should be a “exact portrait” and “miniature” of the people as a whole. That doesn’t happen when district boundaries are manipulated in this way. 
How does extreme gerrymandering work?
People often associate gerrymandering with the creation of super-safe districts that a party wins by overwhelmingly large margins. But, in fact, making districts too safe makes it hard to do an extreme gerrymander. Rather, the goal of a party seeking to use an extreme gerrymander to grab a disproportionate share of seats is to spread its supporters out among districts, letting it win a larger number of seats. 
To illustrate how this works, consider a simplified hypothetical state with four districts and slightly more Democrats than Republicans. You could draw the districts in such a way that the parties split the seats.

But, alternatively, you also could draw the districts in a way that Democrats end up with all four seats - essentially trading super-safe districts for more seats.

Of course, there is a danger for gerrymanderers. The trick is not to spread your voters out so much that districts become vulnerable to flipping to the other party in the normal give and take of electoral politics.

Fortunately for gerrymanderers – and unfortunately for the rest of us – this is becoming easier to do with “Big Data” and advancements in technology. If it works, gerrymanderers will have succeeded in creating the worst of all worlds – a map that both is uncompetitive and skewed in favor of one party.

How can you tell when a map is an extreme gerrymander?

Although people often focus on individual districts when they talk about partisan gerrymandering, the extreme gerrymandering being challenged in Wisconsin and other states involves looking at the map as a whole to gauge whether a map results in an unfair allocation of seats between parties.

To figure that out, one starting point are mathematical measures like the efficiency gap or mean-median difference. These and other similar statistical measures are powerful diagnostic tools that can help courts identify when a particular distribution of seats is statistically unlikely to be random – or at least when a map needs to be looked at more closely. Computer simulations that create hundreds and even thousands of random maps also can help point to situations where something is likely amiss.

But statistical bias is not the end of the inquiry.

Rather, just as a doctor trying to make a diagnosis of a disease will look at test results in tandem with the results of a physical examination and questions to the patient about lifestyle and family medical history, so courts can confirm the results of statistical tests with other evidence. This evidence includes things like floor statements by lawmakers, emails, and odd district shapes that can’t be explained by neutral considerations.

Likewise, a state whose map is being challenged can present evidence to show that the bias in a map was due to things other than the intent to maximize partisan advantage.

Doesn’t the fact that Democrats are concentrated in cities explain why Republicans get more seats?

No. It’s true that big cities like New York and Los Angeles are heavily Democratic. And it’s true that residential patterns - or so-called clustering - may have some impact on the number of seats each party has. But there still are plenty of opportunities to gerrymander in areas outside big cities. In fact, Princeton professor Sam Wang shows how easily this is possible when there are significant pockets of both parties outside the big cities if there is a relatively even spread of partisans in those areas:

Figure 2: Sam Wang (@SamWangPhD). 2017. “This population can give a 5-1 split...or 4-2 the opposite way...all w/nice boundaries. Need more than maps to solve.” Twitter, August 25, 2017, 1:53pm.https://twitter.com/SamWangPhD/status/901155535541723137.

In fact, it’s notable that extreme gerrymanders occur not in deeply red or deeply blue states but in battleground states like Wisconsin, Michigan, North Carolina, and Pennsylvania, that aren't starkly clustered but that just happened to be controlled by a single party at the time of redistricting. To be sure, the cities in those states are fairly to heavily Democratic, but, as a precinct level map will show, these battleground states also have a lot of Democrats in suburbs, college towns, and rural areas. Given this comparatively even spread of Republicans and Democrats, it absolutely matters how you draw districts. Draw a district heading one direction, you end up with a Democratic or at least a competitive district. Draw it in the other direction, and you have two safe R seats.  

A great real world example is in Bell County, Texas, where the town of Killeen currently is split between two state house districts. 

Figure 3: Texas Legislative Council, “State House Maps, Plan H358” (2015), http://www.tlc.state.tx.us/redist/districts/house.html.

If instead Killeen were kept together in one district, it would be a very competitive district for Democrats because of the presence of a large military base (and lots of minority voters connected with the military).  

Figure 4: Texas Legislative Council, “State House Maps, Plan H302” (2013), http://gis1.tlc.state.tx.us/?PlanHeader=PLANh302.

And that example is in a party of central Texas where the concentration of Republicans is much denser than in places like small-town Wisconsin and Michigan. In those Midwestern battleground states, the geography is even more favorable for creating Democratic, or at least highly competitive, districts. (It would be different if every census block outside of the big cities were uniformly, say, 56% / 44% Republican over Democrat, but that isn't the case, not even in Texas.)

One final point: It’s telling that the problem of high partisan bias is closely correlated with single-party control of the redistricting process. Contrast that with states that where commissions, split control legislatures, and courts draw maps, which have much lower - and much less durable - levels of bias. California had high levels of bias in the 1990s when Democrats controlled drawing of the maps. It has negligible levels today.