The Geeks Who Put a Stop to Pennsylvania's Partisan Gerrymandering

March 20, 2019 | Story | No Comments

The morning John Kennedy was set to testify last December, he woke up at 1:30 am, in an unfamiliar hotel room in Harrisburg, Pennsylvania, adrenaline coursing through his veins. He'd never gone to court before for anything serious, much less taken the stand.

Some time after sunrise, he headed to the courthouse, dressed in a gray Brooks Brothers suit, and spent the next several hours reviewing his notes and frantically pacing the halls. “I think I made a groove in the floor,” Kennedy says.

By 3:30 pm, it was finally time. Kennedy’s answers started off slowly, as he worked to steady his nerves. Then, about an hour into his testimony, Exhibit 81 flashed on a screen inside the courtroom. It was a map of part of Pennsylvania’s seventh congressional district, but it might as well have been a chalk outline of a body.

“It was like a crime scene,” explains Daniel Jacobson, an attorney for Arnold & Porter, which represented the League of Women Voters in its bid to overturn Pennsylvania’s 2011 electoral map, drawn by the state’s majority Republican General Assembly. The edges of the district skitter in all manner of unnatural directions, drawing comparisons to a sketch of Goofy kicking Donald Duck.

As an expert witness for the League of Women Voters and a political scientist at West Chester University, Kennedy’s job was to show how the state’s map had evolved over time, and to prove that the General Assembly had drawn it specifically to ensure that Republicans would always win the most seats in Congress.

“Mr. Kennedy, what is this?” asked John Freedman, Jacobson’s colleague, referring to the tiny, single point that connects one sprawling side of the district to the other. Or, if you like, where Goofy’s toe meets Donald’s rear.

“A steakhouse,” Kennedy answered, according to the court transcript. “Creed's Seafood Steaks in King of Prussia.”

The only thing holding the district together, in other words, was a single ritzy seafood joint.

“If you were in the courtroom, it was just devastating,” Jacobson says.

Districts like Pennsylvania’s seventh don’t get drawn that way by accident. They’re designed by dint of the centuries-old practice of gerrymandering, in which the party in power carves up the electoral map to their favor. The playbook is simple: Concentrate as many of your opponents’ votes into a handful of districts as you can, a tactic known as "packing." Then spread the remainder of those votes thinly across a whole lot of districts, known as “cracking.” If it works as intended, the opposition will win a few districts by a landslide, but never have enough votes in the rest to win the majority of seats. The age of computer-generated data splicing has made this strategy easier than ever.

Until recently, courts have only moved to stop gerrymandering based on race. But now, the law is taking a closer look at partisan gerrymandering, too. On Monday, the Pennsylvania Supreme Court issued a brand new congressional map to replace the one Kennedy testified about. The new map follows a landmark decision last month, in which the three Pennsylvania Supreme Court justices overruled a lower-court decision and found that Pennsylvania’s 2011 map did in fact violate the state constitution’s guarantee of “free and equal elections.” The court ordered the Pennsylvania General Assembly to submit a new map, with approval of Pennsylvania’s Democratic governor Tom Wolf. Following unsuccessful appeals by the General Assembly, the court drafted and approved its own map, which will now be in effect for the midterm elections in November, opening up a new field of opportunity for Democrats in the state.

On Tuesday morning, President Trump urged Republicans in the state to "challenge the new 'pushed' Congressional Map, all the way to the Supreme Court, if necessary. Your Original was correct!"

According to Jacobson, given the Supreme Court of the United States already declined to stay the Pennsylvania Supreme Court's decision, it's unlikely they'll take up the case. It's already agreed to hear four other gerrymandering cases this term, which may well re-write the rules on this twisted system nationwide.

The change that's already come to Pennsylvania may not have been possible without the research Kennedy and three other expert witnesses brought to light. They took the stand with a range of analyses, some based in complex quantitative theory, others, like Kennedy’s, based in pure cartography. But they all reached the same conclusion: Pennsylvania’s map had been so aggressively gerrymandered for partisan purposes that it silenced the voices of Democratic voters in the state. Here's how each came to that conclusion—and managed to convince the court.

The Only Bad Restaurant in Town

Carnegie Mellon mathematician Wes Pegden had already written an academic paper proving that the Pennsylvania map was drawn with partisan intent. His challenge in the courtroom was to convince a room full of non-mathematicians. So he came armed with an analogy.

Imagine, Pegden told the court, you’ve touched down in a new city and asked your taxi driver to drop you at any restaurant, something that would give you a sense of the local culinary scene. You give the cabbie a fat tip, go inside the restaurant, and have a terrible meal. Did the driver bring you to a bad restaurant on purpose? Or is it a true reflection of all of the restaurants in the city?

To answer that question, you could always sample every single restaurant, but that would take too long. A more efficient, but still effective option: test every restaurant immediately surrounding the bad one. If they're all bad, the driver really did pick a representative dining establishment. If they’re all really good? The driver screwed you over.

That's essentially how Pegden tested the Pennsylvania map. He developed a computer program that begins with the current Pennsylvania map, then, instead of drawing an entirely new map from scratch, it automatically makes tiny changes to the existing one to create 1 trillion slightly different maps. In the analogy, these trillion maps are the nearby restaurants. The system only draws districts that a court might accept, meaning they’re contiguous, reasonably shaped, and have similar population sizes, among other things.

Christopher Warshaw, George Washington University

Then, Pegden analyzed the partisan slant of each new map compared to the original, using a well-known metric called the median versus mean test. In this case, Pegden compared the Republican vote share in each of Pennsylvania's 18 districts. For each map, he calculated the difference between the median vote share across all the districts and the mean vote share across all of the districts. The bigger the difference, the more of an advantage the Republicans had in that map.

After conducting his trillion simulations, Pegden found that the 2011 Pennsylvania map exhibited more partisan bias than 99.999999 percent of maps he tested. In other words, making even the tiniest changes in almost any direction to the existing map chiseled away at the Republican advantage.

“You can almost hear the mapmakers saying, ‘No don’t do that. I wanted that right there just like that,’” Pegden says. “It gets at the basic question of what citizens, judges, and courts want to know: Did these people go into a room and design these maps to suit their purposes?”

Until now, researchers have struggled to find truly random maps to compare to gerrymandered maps; the number of possible maps is so astronomically high, it’s impossible to try them all. But Pegden’s theorem proves you don’t have to try every restaurant in town to know you got a raw deal. You just need to take a walk around the block.

The Bright Red Dot

Unlike Kennedy and Pegden, Jowei Chen was no witness-stand novice. The political scientist at the University of Michigan, Ann Arbor has provided expert testimony in a litany of redistricting cases, including in North Carolina, where judges relied heavily on Chen's testimony in their decision to overturn the existing map.

Like Pegden, Chen uses computer programs to simulate alternative maps. But instead of starting with the original map and making small changes, Chen’s program develops entirely new maps, based on a series of geographic constraints. The maps should be compact in shape, preserve county and municipal boundaries, and have equal populations. They’re drawn, in other words, in some magical world where partisanship doesn’t exist. The only goal, says Chen, is that these maps be “geographically normal.”

Chen generated 500 such maps for Pennsylvania, and analyzed each of them based on how many Republican seats they would yield. He also looked at how many counties and municipalities were split across districts, a practice the Pennsylvania constitution forbids "unless absolutely necessary." Keeping counties and municipalities together, the thinking goes, keeps communities together. He compared those figures to the disputed map, and presented the results to the court.

The following chart shows how many seats the simulated maps and the disputed map generated for Republicans.

Most of the maps gave Republicans nine seats. Just two percent gave them 10 seats. None even came close to the disputed map, which gives Republicans a whopping 13 seats.

The chart showing the number of split municipalities and counties paints a similarly compelling picture.

Chen used two other metrics to measure the disputed map’s compactness relative to the simulations. The first, called the Reock score, analyzes the ratio of the district’s area to the area of the smallest circle that can be drawn to completely contain it. A district that’s a perfect circle, in other words, would have a Reock score of one. The more distorted the district’s shape gets, the lower the score.

Chen also put the map up to the so-called Popper-Polsby test, which is the ratio of the district’s area to the area of a circle whose circumference is the same length as the district’s perimeter. Again, the lower number, the less compact the district.

Here’s how the disputed map fared on both tests against the simulations:

Chen conducted another simulation with an additional 500 maps, this time, requiring that none of them pit two incumbents against each other. The goal was to see if the General Assembly drew the original map this way not based on partisanship, but based on protecting incumbents. But the results were largely the same. On every metric, the disputed map was an outlier.

“These charts are what really resonated with the Pennsylvania Supreme Court justices,” says Jacobson. “You see 500 black dots. Then you see the actual plan. It’s way out in nowhere land.”

The results, Chen says, complemented Pegden's evidence perfectly. “It’s not a question of whose metrics and methods do you like better,” he says. “The point is: Here’s a diversity of methods, and they are leading us to the same answer. Maybe that tells us something.”

Political Silencing

Another question before the court was whether the partisan map actually impacted representation in Congress. After all, just because most Pennsylvania representatives are Republicans doesn't mean they'll always vote with Republicans. But Christopher Warshaw, a political scientist at George Washington University, showed mathematically that the Republican advantage also meant that the state's Democrats had little chance of having their voices heard in DC.

To assess the map’s partisan nature, Warshaw used a metric called the efficiency gap, which researchers at the University of Chicago Law School and the Public Policy Institute of California devised in 2015. It measures the number of votes that each party “wastes” in a given election to gauge how packed and cracked its districts are. Every vote a party gets in a district that it loses counts as wasted. In districts the party takes, any vote over the total needed to win is considered waste as well.

“You want to get as many seats in a legislature with as few votes as possible,” Warshaw explains. “You want to get zero votes in the districts you lose.”

To determine Pennsylvania's efficiency gap, Warshaw calculated the difference between each party’s wasted votes and divided it by the number of total votes cast in the election. He found that the 2011 map not only gave Republicans a bigger advantage in Pennsylvania than they had before redistricting; it gave them an advantage like few the country has ever seen. “It is really one of the most extreme partisan gerrymanders in modern American history,” Warshaw says.

Warshaw analyzed the average efficiency gap in states with more than six representatives between 1972 and 2016, and found that the vast majority have historically had an efficiency gap hovering around zero.

He also found, however, that since 2010, the last year before districts were redrawn, maps have become increasingly skewed toward Republicans, as the party dominates state legislatures and governorships across the country.

Even so, the slide toward Republican advantage has been far more drastic in Pennsylvania. In 2012, Republican candidates won only 49 percent of the congressional vote in Pennsylvania, but gained 72 percent of the seats.

Finally, Warshaw deployed a commonly used model called the DW-Nominate score to show how partisanship has changed in Congress over time. This score ranks members of Congress on a scale from -1, being the most liberal, to +1, being the most conservative. As the chart shows, both parties have been creeping toward their respective poles steadily over time.

Warshaw doesn’t try to prove that gerrymandering created that partisanship in Congress. His point is merely that in Pennsylvania, where more Democratic votes are wasted, it becomes almost impossible for Democrats to see issues they support turn into federal policy. This degrades trust in government and in elections.

“Representative democracy should be largely responsive to what voters want, and if it’s not, it calls into question democratic bona fides,” says Warshaw. In societies where elections shut one entire subset out of power, he says, “all kinds of bad things can happen.”

“Ultimately, people think why are we even having elections?” Warshaw says. "There’s nothing inevitable about democracy.”

The Evolution of Maps

Though by far the least technical expert in the case, John Kennedy was perhaps the most compelling. In preparation for his nerve-wracking two hours on the stand, Kennedy, an expert in Pennsylvania elections, dug through decades of old maps dating back to the 1960s to assess how the shape of districts and their partisan outcomes have evolved over time.

He methodically walked through how Pennsylvania’s first congressional district, comprising much of Philadelphia, has been packed with Democrats, while Democrats in Harrisburg have been cracked between the fourth and eleventh congressional districts, creating Republican majorities in both places.

But it was the seventh congressional district—and the single seafood restaurant holding it together like a piece of Scotch tape—that clinched it. He showed the court how the district had morphed from a squarish shape to today's sprawling, cartoonish scene. “How do you justify the seventh congressional district?” Kennedy says. “It’s absurd.”

Where Chen and Pegden laid out the mathematical proof of partisanship, and Warshaw demonstrated how that partisanship translates to policy, Kennedy showed in the starkest terms just how obviously gerrymandered these maps looked even to the untrained eye.

As gerrymandering cases proliferate across the country, there’s been some talk in research circles of the need for one true metric to measure it. Overturning Pennsylvania's gerrymandered map, though, required detailed analysis from all angles. “Metrics are just evidence," says Jacobson. "It’s always helpful to have more evidence not less.”

In the Pennsylvania case, Judge P. Kevin Brobson of the Commonwealth Court agreed that Republicans had obviously and intentionally given themselves an advantage, but stopped short of saying they had violated the state’s constitution. In January, the Supreme Court disagreed, striking down the old Pennsylvania map.

In a matter of months, Pennsylvanians will head to the polls once more to elect 18 representatives to Congress, based on an entirely new electoral map that leans far less in one party’s favor. For Kennedy, an academic who spends most of his time studying history, it’s been a rare opportunity to make history, instead.