SMU team generated 1.5M district maps, but none were as gerrymandered as Texas’ pick

A group of SMU mathematicians and one philosopher are using math to measure how biased voting district maps are in Texas.

The group, called Math for Unbiased Maps in Texas, uses software that generates millions of maps to compare with proposed district maps, creating an opportunity for greater accountability in redistricting.

Every 10 years, after the U.S. census, states use data on race, Hispanic origin and the voting-age population to redraw the boundaries of their congressional and state legislative districts.

In Texas, Republicans control the redistricting process because of their majorities in the House and the Senate.

Legislators use gerrymandering, or the practice of dividing election districts in a way that gives one political party an advantage, to manipulate these maps. This practice has been used by U.S. politicians for about as long as our country has existed.

But just how biased have modern-day maps become in the state of Texas? The map that was approved last October is so highly biased, it is quite literally off the charts, according to the SMU findings.

The open-source software that the SMU researchers use helps them generate millions of maps that follow state guidelines for drawing districts.

The software allowed the researchers to answer a simple question: “If you didn’t try to design [maps] to maximize Democratic seats or Republican seats, if you just pick them randomly to satisfy the law – what would you get?” said Andrea Barreiro, associate professor of mathematics at SMU.

Using this large set of randomly generated maps, the group established a baseline for what a typical map that follows state guidelines looks like. With this baseline, the researchers were then able to measure how far from the baseline a proposed map was – and, therefore, how biased it was.

“If you have something that’s way outside of [the baseline], then there must have been some design goal that pushed it away from all these randomly generated maps, and that’s what we would call a biased map,” said Scott Norris, SMU associate professor of mathematics.

As soon as Texas’ first proposed congressional maps were made available on the Capitol Data Portal in late September, the SMU team got to work analyzing how the maps fared.

As different maps were proposed, the team generated over a million maps to create an unbiased baseline, offering key measures that are commonly used by political scientists to assess gerrymandering.

They completed this analysis for 59 proposed congressional maps. (The SMU team also shared their analysis for proposed Texas House, Senate and city council maps.)

Of the 1.5 million maps that the team generated and analyzed to compare to the final proposed Texas congressional district map, not a single baseline map showed levels of bias as high.

The proposed map was more biased than every single map their software had generated, the SMU researchers showed.