Computer lab v. Smoke-filled room: Gerrymandering case tests statisticians’ work

Mandatory Credit: Photo by (9111561k) People attend an anti-gerrymandering rally outside the Supreme Court after the Court heard oral arguments for the case Gill v. Whitford, in Washington, DC, USA, 03 October 2017. Anti-gerrymandering rally outside the Supreme Court, Washington, USA - 03 Oct 2017
Protesters gathered outside the Supreme Court on Oct. 3 while justices heard arguments in the case on gerrymandering, Gill v. Whitford.

WASHINGTON — Legions of activists, lawyers, and party operatives hung on every word of Supreme Court arguments this month over the dark arts of political gerrymandering. On the margins, another subgroup was equally riveted: statisticians.

Once and for all, after centuries of debate, some of these mathematical wonks believed they had developed a tool to statistically prove to the nine justices that an electoral map has been rigged — and it was being put to the test at the high court.

But the formula, which measures something called the “efficiency gap” in voting patterns, showed immediate signs of fraying under live fire in the Wisconsin case, with Chief Justice John Roberts calling it “sociological gobbledygook.”


The case is not closed, however: Roberts could end up in the minority on the potentially landmark case, Gill v. Whitford, which puts scientific theory and backroom politics on a collision course.

Get Ground Game in your inbox:
Daily updates and analysis on national politics from James Pindell.
Thank you for signing up! Sign up for more newsletters here

The business of district map-drawing frequently defies logic and common sense, as shown time and again by districts that snake, twist, and corkscrew through individual locales.

Plaintiffs in cases over the decades have argued countless times that geographic blocs of like-minded voters have been “packed’’ into congressional and state legislative districts, to weaken their power in other areas, or “cracked’’ across multiple districts, to diffuse their potency.

Some Supreme Court justices — especially the likely swing vote in this case, Justice Anthony Kennedy — have said previously they want an empirical standard to measure whether voting district maps are drawn to give one party a constitutionally unfair advantage.

So political scientist Eric McGhee came up with one, which was a central element in the legal challenge to Wisconsin’s state legislative districts. And on Oct. 3, he tucked himself into the back of the Supreme Court chamber to hear the justices of the nation’s highest court debate his work during oral arguments on that case.


In an interview later, McGhee, a research fellow at the Public Policy Institute of California, said he did not set out to make waves in American jurisprudence. He was just trying to have some nerdy, science-filled fun when he started crunching the numbers.

“For me, that was the exciting part: this political science debate,” said McGhee. “It never occurred to me that it would go beyond that.”

McGhee’s test measures each political party’s “wasted votes,’’ which are defined as the winning votes in an election beyond the amount needed for victory as well as all votes for the losing side. The formula calculates a statewide efficiency gap’ by dividing the net wasted votes by the total votes cast. A big efficiency gap is evidence of gerrymandered districts.

That’s the theory, anyway.

McGhee said his test was built to work with electorally competitive states like Wisconsin as well as uncompetitive places like the dependably blue Massachusetts. Based on his formula, Massachusetts is among 14 states with state legislative maps that are unfairly biased toward one party, raising the possibility of litigation in the state if the Supreme Court embraces the approach.


McGhee’s unpublished legal paper about the formula (which he wrote with University of Chicago professor Nicholas Stephanopoulos) made it into the hands of a New York University law professor advising a group of aggrieved Democratic activists from Wisconsin. The group, including William Whitford, who became the lead plaintiff in the federal lawsuit, believed their state’s electoral map was designed to entrench Republican power.

Wisconsin’s legal team on Tuesday urged the high court to maintain a hands-off approach, arguing that the state’s map reflects a natural geographic clustering of partisans rather than intentional bias.

In Wisconsin, Democratic voters are more heavily concentrated in urban areas, giving Republicans a natural advantage, said Rick Esenberg, the president of the Wisconsin Institute for Law and Liberty, a conservative nonprofit law firm. Esenberg said the efficiency gap fails to take into account this political geography, pointing only to nefarious political intent.

“I guess I can understand why he might call it gobbledygook,” Esenberg said of Roberts. “He’s on the Supreme Court — he gets to say things like this.”

The plaintiffs challenging Wisconsin’s map as unfairly favoring Republicans hope the justices will see the need for a science-based approach. As technology has advanced and computerized data have multiplied, it has become easier to both gerrymander a state to extreme degrees, as well as measure whether districts were drawn with bias, they argue.

While some liberal justices seemed intrigued by the math, Roberts was joined in skeptical questioning by Justice Neil Gorsuch, who compared the combined use of various partisan bias measurements to his steak rub recipe, and Justice Samuel Alito, who appeared to discount statistical and social science-based measurements.

“It’s full of questions,” Alito said of the efficiency gap test. “Mr. McGhee’s own amicus brief outlines numerous unanswered questions with this theory.”

Whitford, the lead plaintiff and a professor at the University of Wisconsin Law School, was himself taken aback by the doubt from conservative justices.

“To me, it was in a way the most startling thing about the hearing,” said Whitford. “That these three justices are more or less dismissing social science as something that might help the court. It sounds almost like they’re climate deniers.”

Other academics were equally surprised.

“It’s the kind of thing that I, as a mathematician, often get offended by,” said Mira Bernstein, a researcher at Tufts University. “That’s just not something he would ever say about a different branch of human knowledge.”

Bernstein is a member of the Boston-based Metric Geometry and Gerrymandering Group, a team of mathematicians that studies redistricting and trains people to be expert witnesses on it. While the efficiency gap metric was central to the plaintiffs’ case, other measurements for partisan bias were cited in oral arguments, sowing confusion in the court, she said.

Gary King, a Harvard University professor who developed one of the first of the statistical gerrymandering measurements in the 1980s, called the efficiency gap “a little sloppy.” But much more important than the specific test, King said, is convincing the justices to adopt a standard in which the same rules apply to Democrats and Republicans, no matter the state.

“It’s really clear, just looking at the data, that there really is partisan bias,” King said. “What we need is for the Supreme Court to stand up and make a decision.”

Julia Jacobs can be reached at Follow her on Twitter @juliarebeccaj.