The Texas Jigsaw Massacre
Governor Abbott has a brilliant plan. What could possibly go wrong?
Redistricting is supposed to happen only in the first year after each Census. Despite that, Donald Trump has induced Texas Governor Greg Abbott to use a special legislative session to potentially redraw his state’s Congressional map to squeeze out five more Republican seats. However, analysis by the Electoral Innovation Lab suggests that Trump's redistricting push could backfire spectacularly. Indeed, in 2026, such an aggressive map could end up giving Democrats 25 out of 38 seats.
Republicans, desperate to retain control of Congress after the 2026 election, are looking for ways to get more seats out of the national map. Generally speaking, midterm elections tend to go against the party of the president in power. Since 2004, the U.S. House has changed by a median of 13 seats in non-redistricting years, far more than the four seats currently needed to change partisan control, and far more than the benefits of any new redistricting plan in Texas.
The worst-case scenario for Republicans arises from the fact that they are currently unusually unpopular. In special elections, Democrats have overperformed their 2024 margins by 13 points on average. Districts that are drawn to be too close could in fact flip to be won by Democrats. This could lead to as few as 13 Republicans in the Texas delegation. That kind of accidental backfire effect is what my academic collaborator Bernie Grofman has called a dummymander. I'm not saying this is going to happen, but it is possible. As the financial saying goes: pigs get fat, but hogs get slaughtered.
Redistricting Texas is unlikely to help Republicans for a simple reason: the existing Congressional plan is already gerrymandered. An attempt to redraw it even more aggressively might cut margins too close. Under a pro-Democratic swing, up to a dozen safe Republican seats would be put at risk.
The Current State of Texas's Congressional Map
In 2022, Texas Republicans prepared a Congressional plan that gave them three to five more seats than neutral map-drawing principles would produce. The Princeton Gerrymandering Project gave this plan a failing grade.
(By the way, PGP has scored congressional and state legislative maps in all fifty states. Check it out, as well as a deep dive into how states get A’s and B’s, by Zachariah Sippy and me.)
The existing map generates 25 Republican and 13 Democratic seats. Trump and his allies might be heartened by the fact that Trump got more voters than Harris in 27 districts, with massive swings among Hispanic voters along the southern border and the Rio Grande Valley. However, these districts did still elect Democratic congressmen.
This plan is bombproof to swings in opinion in either direction. Special elections (i.e. held at times other than November) often presage what will happen in the next national election. In special elections since January 2025, Democrats have outperformed their past margins considerably, exceeding their 2024 Presidential vote margin by 13 points. But even with a 13-point swing in vote margins, not a single seat would change hands under the current map.
Mid-decade dreams
Unlike most other states, Texas law does not limit redistricting to the year after the Census, when every state must meet the federal requirement of maintaining the same number of people across all districts. But Texas can redistrict any time, a type of transgression they pioneered in 2003. So although their mid-decade redistricting is anti-democracy, it is not actually illegal. How very Texas of them.
But it might also not be a smart move. To understand why this might backfire, we analyzed what such aggressive redistricting would actually produce.
We can simulate thousands or even millions of hypothetical maps by computer. This defines a neutral baseline, which PGP uses to determine whether a specific candidate plan is an outlier. Here is how the current Texas map stacks up.
On average, simulations based on party-blind principles produce maps that elect 17 Democrats out of 38. But the current map produces only 13 Democratic seats based on 2020 voting patterns, and in real elections in 2024 that is exactly what happened.
We decided to see how Trump’s suggestion would play out, in the form of hypothetical new maps. Here I analyze maps drawn by Neil Dixit of Palo Alto, California. If you have your own extreme gerrymander, please be in touch!
How much juice is left in the lemon?
The basic question is whether, as demanded by Trump, a plan can be drawn that adds five safe Republican Congressional seats on top of the current plan.
To do this, Republicans will face three challenges: geographic, legal, and electoral.
The first challenge is geographic. Since the current enacted plan has 24 safe (and 1 competitive) Republican seats, the goal would presumably be to draw at least 29 safe Republican seats. This would require packing Democratic votes into 9 extremely safe districts and splitting the remaining Democratic votes into powerless minorities across many Republican districts. This requires finding five Democratic districts to flip. We found 1 such district in the Dallas-Fort Worth area, 2 districts in Houston, 1 district in San Antonio-Austin, and 1 district in the Rio Grande Valley.
This leads to a second, legal difficulty. Redrawing these districts would require violating the existing interpretation of the Voting Rights Act, which calls for the construction of districts that would allow Hispanic populations in the two urban areas to elect candidates of their choice. Ultimately, the Supreme Court would have to allow that. So one necessary component of this scheme is to eliminate some current Voting Rights Act and Fourteenth Amendment protections. Considering the nature of recent Supreme Court decisions on voting rights, this legal hurdle is probably surmountable.
But the third challenge is hardest: making the resulting districts safe for Republicans. The new districts will have to include suburban and exurban areas around the urban regions mentioned above. Those regions swing in their partisan preference more than the nation as a whole.
For example, see CD3 through CD21 in Neil’s hypothetical map.
In 2024, Trump won these 19 districts by 6 to 16 points, in part because of his increased appeal among Hispanic populations. The map above shows the 2020 Presidential vote, which comes closer to capturing how they would likely perform at the Congressional level. The pink and light blue shading indicates competitive districts. There are a lot of them.
A backfire effect of up to a dozen seats
If opinion swings against Republicans in 2026, many of these districts might go Democratic. For example, in the 2018 Senate race, Beto O'Rourke won more votes than Senator Ted Cruz in five of them. That would undo all the expected benefits of a new map, and Republicans would be back at square one.
In our hypothetical plan, up to 19 districts, fully half of the Texas map, would have races with margins within 7 percentage points or less. Defending these newly competitive districts would potentially cost tens of millions of dollars that could otherwise be spent elsewhere.
Should Democrats follow suit?
The threat of a new Texas map has triggered a tit-for-tat war with Democrats around the country. California Governor Gavin Newsom has floated the idea of redrawing the California map, and New Jersey Governor Phil Murphy has done the same.
However, mid-decade redrawing in California and New Jersey is against the law in those states. Worse yet, such repeal efforts might potentially also undo the progress that has been made in state legislative redistricting. Unlike Congressional gerrymanders, a legislative gerrymander in one state is not counterbalanced by a gerrymander in another state. In that regard, squeezing out a few seats might produce worse representation within each state.
References:
Current map: (DRA) (Excel spreadsheet exploring scenarios)
An add-five-seat gerrymander for Republicans: (DRA) (Excel) (the example discussed above)
One note: we did not search exhaustively.
A more general way to ask this question is to pose it as follows:
1) In 2026, the statewide vote could be as close as 50%-50%.
2) If the legislature packs D's in 9 districts that are 70%D/30%R, the remaining 29 districts must average 56%R/44%D.
3) To escape a dummymander, these must be cut precisely. This is possible, but challenging.
One way to accomplish this would be after drawing the packed Democratic districts, then drawing 29 R-dominated districts that are as close to 60%R, 40%D as possible using 2020 Presidential voting data.
If anyone would care to draw such a map, by all means please post it here.
Sam, I love the title of this essay! The lesson for Abbott here is Don't Mess with Texas