North Carolina's ongoing partisan clash over gerrymandering has caused the courts to step in to break up the fight, or at least to calm the debacle until after the upcoming midterm elections. 

With the important Feb. 12 deadline to file paperwork for the 2018 election cycle quickly approaching, decisions and stays in two cases over the state's partisan gerrymandering have created a particularly bumpy and dramatic month—and Duke hasn't been far from the action. Jonathan Mattingly, chair of mathematics at Duke, has spearheaded Duke's Quantifying Gerrymandering group since the project began in 2013, and the group's work was featured in one of the cases.

“Our democracy is based on checks and balances. It is not a free for all. It is not the rule of the many," Mattingly said. "We have individual rights and [rights] protections, so maybe we need to put projections to make sure the will of the people—that their voice—is heard in spite of who might draw the districts."

The judicial timeline for the past month is a rather confusing one. The two cases affecting the maps are Common Cause v. Rucho, in which the plaintiffs argue that the congressional maps used in 2016 were unconstitutional because of partisan gerrymandering, and State of North Carolina v. Covington, in which plaintiffs argue that state Republicans racially gerrymandered state-level districts when drawing them in 2011. 

On Jan. 18, the Supreme Court granted a stay request on a federal district court’s order in Common Cause v. Rucho to redraw the 2016 congressional district map by Jan 24. The next day, a lower court in State of North Carolina v. Covington unanimously upheld its map drawn by a third-party expert called a Special Master for North Carolina’s General Assembly—after the 2016 legislature map was ruled to be racially gerrymandered.

Since 2013, Duke’s math department has been pioneering gerrymandering research, and now that work is being showcased in the two high-profile court cases. The group, which is part of Duke's Data+ program, has had a clear impact on how the Court views gerrymandering cases, Mattingly said. In addition to his involvement in the Data+ initiative, he was the author of a report for Common Cause v. Rucho and testified for about three hours as an expert witness. 

“It is refreshing that the [lower] court found the mathematical analysis compelling [in Common Cause v. Rucho]. The fact that the Supreme Court blocked and stayed the immediate action is disappointing but not completely surprising because [the Court] did the same thing in [Gill v. Whitford],” Mattingly said, referring to a case brought to the Supreme Court in 2017 centered around allegations of gerrymandering in Wisconsin.

The outcome of North Carolina’s 2012 congressional elections influenced the work of the Quantifying Gerrymandering project. Although the majority of North Carolinians voted for a Democratic candidate, only four Democrats were elected to hold a seat in the House of Representatives–leaving nine Republicans to prevail in the remaining nine districts. 

Acknowledging that “some people" would clearly argue that seven Democrats should have held the office, Mattingly said he was taken by the question of how many seats each party should hold based on election outcomes. 

“So, we started to ask that non-partisan question: how do we figure that out?" Mattingly said. "If we took the votes from the 2012 election for the House of Representatives, and we throw in a whole bunch of reasonable redistricting, then what would be the outcome of that election?"

The results showed huge variability based on the makeup of the districts.

"We saw a range from three Democrats to nine Democrats," he explained. "But, if you can have three to nine representatives elected with the exact same votes just by how you draw the districts, it seems like there is another layer that comes between that election and interpreting the results of that election.”

The code that was used for the project was written by two current Duke seniors, Hansung Kang and Justin Luo. Luo explained that all of the theoretical districts drawn by their work satisfied requirements such as equal population, compactness and contiguity.

“For the algorithm, we start from the given original redistricting, and for each step we take in the algorithm, we propose the switch of two small units called voting tabulation districts, which are like precincts," he said. "We proposed to switch them and with a certain probability based on the factors mentioned we switched them. We would get one redistricted district, and then another, and then another."

Over the past summer, Kang explained the group collaborated with retired judges who redrew districts as a third party for the Common Cause v. Rucho case. The group compared a sample distribution of 100,000 randomly generated North Carolina district maps to the ones used in 2012 and 2016, and to the one devised by the judges. The deviations between the 2012 and 2016 official maps and their model were found to be high, while the deviations between their work and the judges' were much lower.

Christy Graves, a Ph.D. candidate at Princeton University, was the undergraduate whose senior thesis acted as the foundation for Quantifying Gerrymandering. In an email, she concurred that the group’s findings displayed “pretty compelling evidence of excessive partisanship.” Graves recognized that the defense attempted to poke holes in the group’s code, but that the federal court rejected those complaints in their opinion.

“Although we ultimately find these objections either unfounded or insufficiently compelling to overcome the significant probative value of the analyses, these are fair criticisms," the court's opinion reads on page 70. 

Graves wrote that the defense will “certainly try to invalidate” the evidence presented by the plaintiff, but the three-judge panel ruled that the criticisms of the group’s methods were “not valid, or not substantial enough to undermine our results.”

Still, Duke’s involvement in groundbreaking research on gerrymandering is far from over. Mattingly confirmed that there will be a new Bass Connections project on gerrymandering next year, in addition to a new Data plus project this upcoming summer. 

Both Kang and Luo said that their involvement in Data+ provided them with new opportunities and a greater political awareness. Kang added that the program helped him form new relationships with “great mentors.”

“In general, [the project] has affected me to be more politically engaged with the political process,” he said.