The intersection of data analysis and voters’ rights was before the U.S. Supreme Court last week as it took up a case exploring the constitutional limits of partisan gerrymandering. On Oct. 3, the court heard oral arguments in the case of Gill v. Whitford, a challenge to Wisconsin’s 2011 legislative redistricting plan. The lawsuit claims that Republican legislators used sophisticated computer and data modeling tools to create legislative districts that were designed to give the party an overwhelming majority of seats in the state legislature. Opponents assert that Wisconsin’s political geography, and not data analytics, was the driving force behind the plan.Gerrymandering Defined
Gerrymandering, or the political process of arranging legislative districts to the advantage of a specific political, ethnic, racial, or other defined group, has a very long history in the U.S. The term itself dates back to an 1812 effort by Massachusetts governor Elbridge Gerry to align legislative districts to benefit his political party. One such district reportedly was so oddly shaped that a local newspaper said it looked like a salamander, and the name “gerrymander” was born. Gerrymandering occurs in response to the census, which is conducted every 10 years. Among other things, the census requires the redrawing of congressional and state legislative district lines in response to population changes and redistributions. This is often a political process whereby the party in power attempts to draw the lines in a way to ensure that it keeps its power.
Classic gerrymandering occurs by a combination of “packing” and “cracking,” which means that a small number of districts are “packed” with overly large numbers of a target group, and the remaining members of the group are “cracked” among the remaining districts. As an example, say a region containing 10 legislative districts has an overall voting population that’s split roughly 50-50 between the Tory party and the Whig party. To get a Tory advantage, the party might draw the lines of the districts so that three of them have a “packed” 80% majority of Whigs, virtually ensuring that the Whigs would overwhelmingly win those districts. However, because there is now a minority of Whigs remaining among the other seven districts, those district lines could be “cracked” so that each district has a Whig minority. The result is that in a region with a 50-50 split, the Tory party, which drew the lines, would be expected to win seven of the 10 districts.
Previous Gerrymandering Cases
Following the 2010 election, Wisconsin, a state with a roughly equal overall balance between Democratic and Republican voters, found itself with a Republican governor and a Republican majority in both the state senate and legislative assembly. As required by its constitution, after that year’s census, the Republican-controlled state legislature drew a new map for electing representatives to the assembly. A Republican-dominated team was put together to create and test various computer-generated maps. In constructing them, the team used extensive census, demographic, and voting data analysis and statistical modeling to develop what came to be known as a “composite partisan score” for each proposed map.
A spreadsheet of various maps as well as the composite partisan scores for each map were prepared and presented to Republican legislative leaders. They selected a final map, which projected that the Republicans would “expect to win” 59 of the 99 assembly seats. The subsequent elections of 2012 and 2014 proved the accuracy of the data and statistical analysis. In the 2012 election, the Republican Party received 48.6% of the statewide vote for the assembly, but won 60 of the 99 seats. In 2014, it received 52% of the vote, but won 63 of 99 seats.
To be clear, gerrymandering is not just alleged to have been done by Republicans. A separate case out of Maryland is challenging a Democratic redistricting plan that allegedly parlayed a 2-to-1 statewide Democratic voter advantage into a 7-to-1 congressional representative advantage.
A number of Democratic voters in Wisconsin complained that the redistricting violated their First Amendment rights to free speech and assembly by essentially penalizing them for their political beliefs and that it also violated their Fourteenth Amendment right to equal protection. In November 2016, a three-judge panel agreed with them and found the redistricting plan to be unconstitutional. The appeal to the Supreme Court followed.
The Appeal’s Goal
The Supreme Court has dealt with gerrymandering before. In a series of decisions over the last 2 decades, it has severely restricted the use of racial characteristics as part of gerrymandering. However, it has been more reluctant to restrict gerrymandering for purely partisan purposes. In a divided opinion in 2003, the court rejected a challenge to a Pennsylvania gerrymandering case, holding that it was too difficult to come up with a “workable standard” that could determine when a proposed redistricting plan crossed the line from acceptable political considerations (such as preserving existing city, county, or district boundaries or the existing political “geography” of the state, such as differences among urban, suburban, and rural voters) to unconstitutional partisanship.
The challengers in the Wisconsin case are basing their argument in part on the development of newer, more precise data and statistical models that can provide a workable standard for evaluating gerrymandering claims. They point to a measurable “efficiency gap” that can determine whether a particular plan is too partisan to be constitutional. The efficiency gap is based on the presence of “wasted” votes in elections. Under this analysis, every losing vote plus every vote in excess of the minimum needed to win is essentially wasted. In a balanced system, the wasted votes would favor neither party. A modest value of -3 or +5, in which one party has a 3% disadvantage to or a 5% advantage over the other party in the number of wasted votes, might be seen as the result of acceptable political considerations.
An analysis of previous statewide elections between 1970 and 2000 shows an average efficiency gap of -1.5. After the 2000 census and a new, predominantly Republican plan, the average efficiency gap increased to -7.6. By 2012, it was -13.3. The soundness of the statistics and the ability to quantify partisanship based on actual election results, the challengers argue, constitute a workable standard.
In opposition, the state of Wisconsin asserts that its redistricting map is based on acceptable nonpartisan considerations, primarily the concentration of Democratic voters in urban areas, and that the efficiency gap still cannot be a workable standard as it does not measure the reasoning or justification for any gap—and that it will result in near-constant litigation.
The Supreme Court’s Decision
In oral arguments on Oct. 3, much of the focus was on Justice Anthony Kennedy. Kennedy is usually included with the conservative side of the court, but has often been a swing vote, siding with either liberal or conservative judges depending on the issue. In the 2003 decision, Kennedy voted to reject the challenge to the Pennsylvania gerrymandering case, but critically stated that he would “not foreclose all possibility of judicial relief if some limited and precise rationale were found. …” Much of the oral arguments seemed to be directed toward convincing Kennedy that the efficiency gap methodology was the “limited and precise rationale” that he was looking for.
The outcome of the case will be critical and involve data and statistical analysis no matter which way it is decided. If the Wisconsin redistricting plan is found to be unconstitutional, the efficiency gap will become a key metric in future redistricting plans. If it is upheld, then most commentators expect the use of statistical and data analysis to rapidly become more of a fixture in what may be increasingly partisan redistricting plans. Given that the next round of redistricting plans will occur just after the 2020 census—and after the 2020 election—the timing and outcome of this decision will impact American politics for years to come. The ruling is expected this winter or early next spring.