Voting Rights Data Institute (2018)

"The whole point is you’re taking these issues away from democracy and you’re throwing them into the courts pursuant to, and it may be simply my educational background, but [what] I can only describe as sociological gobbledygook.”

    - Chief Justice John Roberts of the Supreme Court of the United States in Gill v. Whitford (2017)

The Metric Geometry and Gerrymandering Group led by Moon Duchin at Tufts and Justin Solomon at MIT assembled 52 graduate and undergraduate students in the summer of 2018 to work on a variety of mathematical and statistical projects concerning redistricting practices in the United States.

The US, accompanied by Canada, the UK, and France, is rare among advanced democracies in that it uses a majoritarian system to elect representatives, the districts of which are redrawn every ten years following the publication of new census information. Legal mechanisms exist to achieve an ideal of population equality between districts to equalize the weight of votes across districts and the Voting Rights Act protects sufficiently compact and cohesive minority groups from being denied representation. Aside from this, states are given discretion in how they design and enforce rules about redistricting. In all but a few states, incumbents are left to redraw the borders and often do so to their own advantage by packing voters of opposing parties into one or a few districts while dispersing the rest of the opposing voters across districts, strategically diluting their voting power. The US is alone in allowing state legislatures to carry out the redistricting process and gerrymandering cases are frequently litigated while they are infrequent in Canada and the UK where independent redistricting or "boundary delimitation" commissions carry out the process of creating new plans, headed by a judge and requiring passage by the legislature without being designed by it.

Some of the materials we produced during the summer can be found here and on the gerrymandr github page though I worked primarily on developing data sets containing information on how the redistricting process is carried out internationally and across state lines.

One important product was a collection of simulations on elections in each state using the method of markov chain monte carlo (MCMC) with 2010 census data and various kinds of election data. Information on how to install and use the software developed by the group, RunDMCMC, can be found at gerrydata.org along with materials from the institute and other tools that allow you to create your own redistricting plans, test their mathematical properties, and use the method of ecological inference to see whether a district is in compliance with section II of the Voting Rights Act. The chain method works by taking a set of redistricting rules as constraints, such as district compactness levels and equal population requirements, and then randomly generates up to billions of plausible alternative redistrcting plans that satisfy the given constraints and then bases results off of census population, spatial, and voting data. With these sample maps, metrics can be generated to evaluate deviations from partisan symmetry and provide an estimate of the likelihood that a map would have been generated if it had not been done to confer partisan advantage.

The following picture shows the voter tabulation districts (VTDs) of New York, some of which are merely water, do not contain people, and do not map nicely into congressional districts:

See a feature on the Voting Rights Data Institute here.