An Extremely Detailed Ranking of the Most Gerrymandered States in America
Which state has America’s worst gerrymander? After a year of redistricting battles, we have an answer.
We’ve ranked all 44 states with more than one district by the fairness of their congressional maps. Many of the most notorious gerrymanders are ranked near the top of the list. But there are also some oft-derided maps that we judge to be fundamentally fine.
A gerrymander is as much art as science, with plenty of room for critics — or different statistical measures — to disagree about which map’s twisting lines are most offensive. But this list is not simply a matter of taste: We’ve based our ranking on the results of thousands of computer simulations that draw compact districts without any consideration of partisanship or race.
These simulations are a broadly accepted way to measure gerrymandering, including by courts. We compared each state’s map with the simulations, then sorted them according to two equally weighted measures: 1) How many more seats the map gives one party, over the simulated estimates; and 2) How aggressive the gerrymander is, meaning how close the map is to the most extreme gerrymander we can cobble together out of the simulations. (A more detailed description of our methodology is at the bottom.)
Because the simulations don’t take partisanship into account, some states are considered fair even if the map appears tilted toward one party or another.
Whether a map should ignore partisanship or reflect it is a hotly contested subject in redistricting. We accept both ways here: If a map does reflect the state’s partisan balance, even if it diverges from the simulations, we consider it fair, too. But the most gerrymandered states, as you’ll see below, fail no matter how we look at it.
2024 presidential election results from The New York Times; simulated district maps were either taken directly from Harvard’s ALARM project or generated with ALARM’s redistricting algorithms as a foundation. In particular we removed any race-based redistricting restraints, to reflect the Supreme Court’s decision in Louisiana v. Callais. We used 5,000 simulated maps per state. The map margins reflect 2024 presidential election results adjusted so that the national margin is even.
States were ranked using two criteria, Seats Gained and Aggressiveness.
The number of seats gained by state was calculated based on two different ways to measure it: simulations, which ignore partisanship, and the efficiency gap, which is all about partisanship.
The simulation-based measurement was calculated by simulating three kinds of elections on each state’s current map: a baseline, using the 2024 presidential election results adjusted so that the national margin is even; a “blue wave” six points to the left of the baseline; and a “red wave” six points to the right of the baseline. We tested each scenario thousands of times, where each district’s result varied randomly, typically by about five points. We used a weighted average (50 percent baseline, 25 percent for each wave) to combine all of these outcomes and calculate a real map’s expected seats. We then applied the same process to all 5,000 simulated maps per state. The seats gained measure is the difference between the average expected seats across the simulations, and those produced by the real map. A state can be fair according to this measure (with a low seats gained number) even if its map is tilted toward one party. That’s because what matters in the simulations isn’t how a state votes, but how different areas of a state vote (see Wisconsin and Massachusetts as prime examples).
Separately, we took partisan balance into account by calculating a state map’s so-called “efficiency gap.” The efficiency gap is a measure often used by academics (you can read more about it here) that measures whether one party won more seats than expected based on its share of the vote. The result is a percentage score for the difference in “wasted votes” that don’t contribute to winning a district. (Massachusetts, to use a previous example, has a high efficiency gap because more than a third of its residents vote Republican but none of those votes help make a G.O.P.-leaning district.) We then multiplied the efficiency gap percentage by the number of seats, and the difference between this and the expected number of seats in the simulations yielded this alternate seats gained measure.
For states with five or more congressional seats, the final seats gained number was the lesser of those two versions. If the efficiency gap measure was lower, and that measure was less than 0.1, we set it to 0.1. For states with fewer than five congressional seats — at which point the efficiency gap becomes more volatile — we just used the results of the simulated approach.
Aggressiveness was calculated based on the average of two measures: a comparison between the enacted map and a hypothetical extreme gerrymander derived from computer simulations, and an analysis of whether the enacted map was an outlier compared with computer simulations.
We first created the hypothetical extreme redistricting map by progressively selecting districts from among the simulations that did the most to pack the minority party’s voters together. Once we were unable to fit any more simulated districts, we split the state’s remaining votes evenly across the leftover districts. We measured aggressiveness in the direction of the party that mostly controlled redistricting in each state. When an independent or bipartisan commission controlled redistricting, we evaluated the extreme redistricting from each party’s perspective, and picked the lesser value. The breakdown of who controls each state is listed below:
Republicans: Alabama, Arkansas, Florida, Georgia, Indiana, Kansas, Kentucky, Louisiana, Mississippi, Missouri, Nebraska, North Carolina, Oklahoma, South Carolina, Tennessee, Texas, West Virginia
Democrats: California, Illinois, Maryland, Massachusetts, Nevada, New Mexico, Oregon, Rhode Island
Both: Arizona, Colorado, Connecticut, Hawaii, Idaho, Iowa, Maine, Michigan, Minnesota, Montana, New Hampshire, New Jersey, New York, Ohio, Pennsylvania, Utah, Virginia, Washington, Wisconsin
We separately calculated how unusual the current seat outcome was across the simulations, by measuring how many standard deviations the current map’s expected seats were from the average simulation. We also set a floor for both measures. Hawaii’s current map, for example, is not so far from the most extreme scenario favorable to Democrats, but the differences are so tiny it doesn’t make sense to call the enacted map aggressive. We combined both measurements into a value scaled from zero to 99.
Finally, the list was sorted by the product of seats gained and aggressiveness, and ranked from largest composite score to smallest.
Alaska, Delaware, North Dakota, South Dakota, Vermont and Wyoming were excluded from the ranking because they each have only one at-large congressional seat.
Related Knowledge Center: Voting rights
As Republicans move to restrict voting rights, and Donald Trump appears to be preparing for another confrontation about election results, a wide range of organizations are pushing back.
Top guidance in the run-up to Election Day
- Ben Wikler, a political organizer who was chairman of the Democratic Party of Wisconsin from 2019 to 2025 says, “The very most powerful thing you can do as a volunteer in politics is to talk to your actual friends and family who are fence-sitters.”
- He quotes Abraham Lincoln, who wrote in a letter to a campaign committee in 1840: “Keep a constant watch on the doubtful voters, and from time to time have them talked to by those in whom they have the most confidence.”
- The place to put our energy, Wikler says, is on the people whose votes will matter most because they live in a place with high-stakes, very close elections, as well as the people where we live.
- ACLU: Know your voting rights | How the ACLU is safeguarding the election
- Report intimidation, at the polls or elsewhere, immediately to the Election Protection Hotline at 1-866-OUR-VOTE or 1-888-VE-Y-VOTA (en Español).