Draw It Yourself: A fair congressional map, Texas edition

Redistricting isn’t exciting, and that’s the way that politicians like it.  The tedious and precise drawing of district lines is often done with little to no public input and there’s almost never a public outcry or any political repercussions when a particularly nasty set of lines is proffered.

To wit, when Republicans controlled the line-drawing process in the Midwest and South after the 2010 Census, Democrats’ response from activists on the ground to the Justice Department was mostly indifference, even though the gerrymandered lines arising out of that round prevented Democrats from winning a majority in the House of Representatives in 2012.

But that’s changing.  As the age of Trump has piqued activist interest in downballot races, people have noticed things like Virginia Democrats falling short of a majority despite winning statewide by 9%.  And the Gill v. Whitford Supreme Court redistricting case made clear just how pernicious gerrymandering is, showing that the Wisconsin Republican Party devised a plan in total secret to keep 3/5 of the legislature under their control even as Democrats won a majority of votes.

In some states like California and Arizona, citizen commissions draw maps after multiple rounds of hearings and input.  In others like Minnesota and Colorado, judges had to justify why they drew lines the way they did and drew upon all sorts of evidence.  But for most people, their congressional districts are still drawn in a proverbial smoke-filled room.  So while it’s nice to say that states shouldn’t have gerrymandered lines, we don’t really know what citizens think non-gerrymandered lines would look like.

That’s where I come in.  Welcome to Draw It Yourself, a project where I, Robert Wheel, create a district map then send it out into the Internet for public comment. See my previous maps of North Carolina and Pennsylvania here.

I lay out my principles and criteria for drawing the lines as I did so you get insight into why the districts look the way they do.  But then you, the concerned citizen, get to tell me whether the line I’m drawing through your neighborhood should be drawn elsewhere instead, and why.  Ideally this iterative process can come up with nonpartisan maps that accurately reflect differing constituencies within states.  I’ll also break down what the political impact of the lines would be because, while the exercise is meant to be nonpartisan, the goal is to show how fairer districts lead to fairer elections.

So: let’s go to Texas.

The flag of Texas, depicted on a downtown wall in Pecos, the seat of Reeves County, Texas

No state has as fraught a history with redistricting as Texas.  In most states, the practice occurs once per decade after it receives Census results but Texas hasn’t had a map last more than ten years since their congressional delegation included Lyndon B. Johnson. Their original ‘90s map included some of the most baroque districts ever drawn, snaking from block to block to pick up minority voters.  They make today’s Pennsylvania 7th District look like a Mondrian.  Litigation led to multiple elections in some years, with Kevin Brady and Gene Fontenot facing off four separate times for the 8th district in 1996.

2002’s court drawn map was supposed to afford the state some breathing room. But after Tom DeLay engineered a Republican takeover of the legislature (which, in retrospect, would have eventually occurred without his intervention) it redrew lines so aggressively that Democrats fled to a motor inn in Oklahoma to avoid a cloture vote in the state senate.  Eventually this scheme did pass but some districts had to be redrawn again because of Voting Rights Act (VRA) concerns.

In 2011 the Republican-dominated state, as one would predict, tried to maximize conservative gains at the expense of minorities.  However, as a three-judge panel decided in Perez v. Abbott, the districts violated the Voting Rights Act in the Dallas-Fort Worth area and South Texas.  They only ruled that a few districts had to be changed, a ruling that has since been appealed to the Supreme Court, which may not deliver a ruling that can be implemented until the 2020 election.  And if the Court renders a favorable decision in Gill (reading oral argument tea leaves is an inexact science, but it does appear Kennedy is leaning toward siding with democracy) then it could force the three-judge panel to redraw the districts altogether.  Which begs the question: what would a fair map of Texas look like?

So let’s use the following criteria to draw a fair map in the state of Texas.

  1. Equal population.  Each district needs to have an equal population.  Because the program I use (the invaluable Dave’s Redistricting App, which allows citizens to perform tasks legislators normally hoard to themselves) does not let me split precincts, I can’t get to perfect equality, but each district in this map is within 100 of the ideal.  If it were adopted the court could shave some blocks off precincts with minimal effect on the districts’ voting behavior (as well as smooth some of the rougher edges like the eastern boundary of TX24). Also of note, the map uses 2008 voting precincts, so when I mention 2016 results usually I am referring to estimates that can’t take changed boundaries into account.  However, I doubt the cumulative effect of such estimates would be more than 1% of margin for any one seat.
  2. Abide by the VRA.  This is always a bit of a moving target as courts are constantly reevaluating what the VRA requires but there is some guidance from the court for the Texas map — South and West Texas need 7 districts that Latino voters can rely on to elect candidates of their choice (which, in practice, are Democrats).  Outside of South Texas most minorities are in urban areas so I tried to A) maintain existing VRA-compliant districts and B) cobble together new coalition minority districts where possible without crossing political boundaries. Nevertheless, the VRA can lead to some odd district shapes, like the pinwheel of TX18, but those are court-mandated.
  3. Respect political boundaries.  In Texas, city boundaries themselves are pretty baroque (Dallas and Houston are both quite the Rorschach test) so they’re less instructive, but county lines haven’t been tinkered with much over the past hundred years.  Often communities of interest overlap with county lines, but there’s usually no reason for two districts to break across multiple counties.  For example, in the current map, the 10th and 17th districts both include portions of rural Lee, suburban Bastrop, and urban Travis Counties.  In a fair map, the 10th is wholly within Travis County, while the 17th only takes in small cities and rural areas.  The line has to be drawn somewhere, but it shouldn’t be drawn in such a way that such disparate communities are thrown together for political purposes.
  4. Make districts as big and as small as possible.  This means keeping urban with urban counties, suburban with suburban and rural with rural. It means TX-5 loses its outpost in the City of Dallas and is centered on more rural areas east of the Metroplex.  There are only 2 districts taking in parts of Travis County instead of 5.  Even leaving the major metro areas, TX-4, TX-17 and TX-19 string small cities together so neighboring TX-1, TX-11, TX-13 can be as rural as possible.  It keeps communities of interest, people living similar lives, together.  The current TX-25 links central Austin, farmland and Fort Worth suburbs together.  What community interest is that?  And trying to give districts a character allows the congressional delegation to reflect the actual population of Texas.  I’m sure the state Republican party would love a delegation of 36 guys who live in suburban McMansions, but anyone who’s spent more than a few hours in Texas knows there’s more to the state than that.


Further to number 3, I’ve devised a metric called the County Breaks measure.  I tried to have district lines break through counties (instead of adhering to county lines) as few times as possible.  So, for example, Denton County had 662,614 people in 2010, which is less than the ideal congressional district size of 698,488.  So ideally it’d only be in one congressional district.  But neighboring Tarrant County had 1,809,034 people, which means at best it’d be in 3 congressional districts.  Accordingly, Denton would ideally be in 1 district while Tarrant would be in 3.

County breaks measure how many districts each county is “broken” into. You could draw 16 congressional districts entirely within one county, requiring at least 20 counties to break across multiple districts.  Accordingly, you’d expect a fair map to have around 20 “breaks” within counties in excess of the existing 16.  The map I provided has 19 (you can get less than 20 by putting the “leftovers” of two or more heavily populated counties into the same district like I did with TX-24), while the existing one has 37.  I’m not opposed to increasing the number of breaks after public input though – you’re a better judge of what’s a community of interest than some whiskey-swilling surveyor from the 19th Century.  His lines are merely a jumping-off point.

Texas map overview, methodology

So using that criteria, you get a map that looks like this:


Dallas-Fort Worth Metroplex


Greater Houston

Austin/San Antonio

As for methodology, first we’ll turn to Dallas-Fort Worth.  Dallas County can support three congressional districts entirely within its borders, and that’s what this map reflects.  TX-33 shifts so it’s entirely within Dallas County and remains over 60% Latino.  TX-30 moves its borders around a little bit, taking in the southern half of the county and gaining a bit in African-American population.  TX32 now takes in the northeast quadrant of Dallas County.  It’s also only 43.1% Anglo and swung from supporting McCain and Romney to backing Clinton (as it does in its current iteration).

Moving over to Tarrant County, TX-12 is now wholly within its borders and is 25% Anglo, 25% AA and 41% Latino.  TX-6 is also Tarrant-only, taking in the southern and western areas as well as much of Fort Worth.  TX-24 then takes in the remainder of Tarrant and Dallas.  Combined, the two counties support almost exactly six districts, but TX-24 does need to swipe a few of the precincts of Dallas City in Collin County to get to the right population.

As noted above Denton County in the north of the Metroplex is almost big enough to be its own district but it too needs help from Collin by taking in part of that county’s portion of Frisco.  Then TX-3 takes in all but the most rural remaining portions of Collin.  Moving back across the Metroplex, collar-shaped TX-25 takes in collar counties, while TX-4 takes in the northern exurbs and smaller cities in the Red River Valley in the state’s most Sooner saturated district.

Moving down to Houston, TX-18 remains a pinwheel in the center of Harris County taking in a 51% AA area.  TX-29 is to its east, 77% Latino with very low turnout.  TX-7 takes in a chunk of northwest Harris County that’s only 31.6% Anglo and, like TX-32, flipped from supporting McCain and Romney to backing Kerry.  TX-2 then forms a bracket around TX-7 taking in the more Anglo parts of inner Houston and TX-9 reaches deeper into diversifying Fort Bend County, going from 5% Anglo, 38% AA, 38% Latino and 12% Asian to 14% Anglo, 27% AA, 42% Latino and 14% Asian.  TX-36 has the eastern suburbs of Houston in Harris County, TX-8 takes in the northern edge of Harris and Montgomery and Liberty Counties to its north, TX-22 is centered on suburban western counties, and TX-14 comprises the oil-dependent coast, stretching from Beaumont through Galveston to Bay City (it may make sense to swap portions of inner Galveston for outer Brazoria to keep Houston suburban areas with each other, but this map doesn’t to minimize county breaks).

Then we turn our attention to the Austin area: Travis and Williamson Counties are big enough to support two seats wholly within them so TX-10 and TX-31 obliterate the pizza slices of districts surrounding the capitol meant to dilute the liberal city’s vote by appending it to conservative outlying areas.  TX-17 then takes in exurban Austin as well as Waco and the areas around Fort Hood.

The South and West of Texas need to be VRA compliant and have seven districts that give Latino voters the opportunity to select their own representatives.  This can be accomplished by locating two districts wholly within Bexar County (TX-20 and TX-35), having the massive TX-23 (light blue) reach up into Midland and Odessa and down into Laredo, keeping TX-16 (neon green) wholly within El Paso County (but taking in its more Anglo areas), and retaining the “tortilla strip” character of TX-15 (orange), TX-28 (pink-purple) and TX-34 (green), all of which are centered in the Brownsville-Harlingen-McAllen area but reach up to take in more rural and less Latino counties (including Nueces County, a particular issue in the Perez case).  All of these rural border districts have Latino populations of at least 78% and voted for Clinton by at least 16%, meeting the court’s requirements.

The new TX-21 takes in suburban areas around Austin and San Antonio; it’s another district that shifted to the left in 2016, but did not support Hillary Clinton.  Moving to the west of the state, mid-sized cities like Lubbock, Midland, Odessa and Abilene can create their own, relatively compact district (TX-19 — yellow-green).  This leaves TX-13 (tan) and TX-11 (chartreuse) to take in enormous swaths of the state with only one city of more than 100,000 people (Amarillo), giving rural voters the ability to choose their own representatives.  TX-1 (blue) and TX-27 (light green) perform similar function in the eastern half of the state, while TX-5 takes in a strip of mid-sized cities east of Dallas.

Partisan Effects

Anyway, now for the good stuff. If a map like this (or something similar) were adopted before the 2018 elections, what would the state congressional delegation look like?  Blake Farenthold, had he run, would have been doomed.  Mike McCaul and Pete Sessions would have to choose between forcing out another incumbent or facing the fights of their political life, though current targets John Culberson and Will Hurd would emerge safer.  Meanwhile, Latino voters would elect at least two new representatives.  In sum, Republicans could lose 3-6 seats.

However, just because Republicans would lose those seats doesn’t make a fair map a de facto Democratic gerrymander.  They currently control the delegation, 25-11.  So even under their worst case scenario, they’d still control the delegation, 19-17, while after a good election for them it’d be a 22-14 advantage.  For a state that leans Republican but was closer than expected in 2016, that seems right.

Seat-by-seat breakdown, likely outcomes

TX-1: Brian Babin (Safe R)

Babin’s old TX-36 is now entirely within greater Houston, but he still gets put into a rural district with no incumbent, as current TX01 rep Louie Gohmert’s home is now in TX-5. Considering he doesn’t have to worry about Houston Republicans ganging up on him in a primary now, Babin comes out a winner under this map.

TX-2: No incumbent (Safe R)

John Culberson lives outside this district but seems the most likely candidate here, as other Houston-area Republicans get their own seats.  That’s a relief for him, because his old district narrowly voted for Hillary Clinton, he is facing his first tough re-election campaign and he has been implicated in an insider trading scandal.  The Democrats fighting to take him on would most likely run in the now vacant TX-7 (though the most hyped – Jason Westin, Alex Triantaphyllis, Lizzie Pannill Fletcher and Laura Moser – would live in TX02 as well), letting Culberson continue to avoid breaking a sweat.

TX-3: No incumbent (Likely R)

Taking in most of Collin County, TX-3 is currently represented by Sam Johnson, the oldest Republican in the House, who’s retiring after this term.  Its partisan composition wouldn’t change much under this map – it’s still a solidly Romney seat that was slightly less solidly Trump.  State Senator Van Taylor appears to have little competition in the open seat, even from other Republicans.  Though one of the Democratic candidates is named Sam Johnson, so perhaps he can pull a Distinguished Gentleman.

TX-4: No Incumbent (Safe R)

This district remains fairly similar to its current version, but John Ratcliffe’s house is drawn out of its new iteration. Still, with no other obvious candidate he shouldn’t have any issue running here from outside the district (congressmen don’t need to live in the district they run in — they only need to live in the state).

TX-5: Louie Gohmert (Safe R)

Gohmert’s strident right-wing views, anathema to the nation as a whole, seem to reflect his constituents’ wishes. Luckily for him, his district number changes but the nature of the district remains similar: centered on Tyler, with no other incumbent congressmen, and heavily Republican.

TX-6: Kay Granger (Safe R)

Tarrant County is trending Democratic, but it’s still one of the few counties with more than one million people to vote for Trump.  The current 6th is an open seat that could be winnable by Democrats if they pull the political equivalent of an inside straight, but if the relatively moderate Granger (who didn’t even endorse Donald Trump) is the Republican standard bearer it’s hard to see how they’d lose it.

TX-7: John Culberson (Leans D)

Hillary Clinton narrowly won the current 7th, but in its new iteration it would’ve backed her 52-44%.  As noted above, Culberson and his most hyped Democratic opponents live outside this district’s borders (as do most local elected Democrats) so this race could turn into a free-for-all.  Owing to Clinton’s performance here, the seat leans Democratic but the historic Republican strength here cannot be ignored.

TX-8: Kevin Brady (Safe R)

Brady has committed a few sins of Republican apostasy (like voicing tepid support for gun control but never really acting on it) so he faced the closest call of his two-decade House career in 2016 where he only won 53% in the Republican primary (thankfully not against Gene Fontenot).  But whoever wins the primary for this seat in 2018 would defeat any Democratic challenger.

TX-9: Al Green (Safe D)

Green’s district gets a little more Anglo, Latino and Asian at the expense of its African-American population as it moves deeper into Fort Bend.  Green might not like that, but he’s still locally popular and being the first Democrat to call for Donald Trump’s impeachment should help him beat back any primary challengers.

TX-10: Lloyd Doggett (Safe D)

Doggett is state Republicans’ bete noir: an Anglo Democrat that they can’t displace despite targeting him in the last 3 rounds of redistricting.  Under this map, he gets a seat similar to the one he represented before 2004: wholly in Travis County, centered on its east end.  Doggett could hold onto this district until he retires, after which this 38% Anglo seat might elect a minority to represent it.

TX-11: No incumbent (Safe R)

This rural district, one of the biggest in the country by land area, has no incumbent but either of current TX-11 and TX-19 incumbents Jodey Arrington or Mike Conaway could easily move here and run.  As neither would actually have to move to this district to run here it should be easy for the two of them to divvy these seats up, but whoever takes the plunge could face a rural-based opponent in the primary.

TX-12: Marc Veasey (Safe D)

Creating a Latino VRA district in Dallas County leaves a coalition district in Tarrant County where Anglos are only 25% of the population, the third biggest ethnic group after Latinos and African-Americans.  It’s similar enough to Veasey’s old district that he shouldn’t have any issues winning re-election.

TX-13: Mac Thornberry (Safe R)

The 13th gets a bit more rural under this map, which suits Clarendon (population: 2,026) native Thornberry just fine, as he still represents one of the most Republican areas of the country.

TX-14: No incumbent (Safe R)

The 14th’s current representative, Randy Weber, would get thrown into TX-22 with Pete Olson.  But he shouldn’t have much trouble winning in TX-14; it would’ve given Trump around 60% and while it contains some ancestrally Democratic areas it still should be safe for Weber as he represents much of the area already.

TX-15: Vicente Gonzalez (Safe D)

Under this map the Rio Grande Valley Districts (TX-15, TX-23, TX-28 and TX-34) all have similar Democratic performance and there are three existing representatives (Vicente Gonzalez, Filamon Vela and Henry Cuellar) to divvy them up.  TX-28 is left open and Gonzalez could always opt to run there, but most likely he’d stay in his home district, which is most similar to the one he represented before and is slightly more Democratic.

TX-16: No incumbent (Safe D)

With Beto O’Rourke running for Senate, this mostly unchanged El Paso district would send a different Democrat to Congress next year.

TX-17: No incumbent (Safe R)

As Travis County is now split between two instead of five districts its liberal voting power is no longer diluted.   This is bad news for McCaul and Carter, who love living in the area but are less enthused about representing its voters and would be thrown in a district that gave Hillary Clinton a sizable margin.  While McCaul hasn’t represented this area before he has enough money and clout to scare away challengers from either party, including the 76-year-old Carter.  However, McCaul could try to put some of his $300m net worth (really, his wife’s, as he married the heiress to the ClearChannel fortune) to holding down the more marginal TX-31.

TX-18: Sheila Jackson Lee (Safe D)

Jackson Lee is abrasive enough that she’s never totally safe from an intraparty challenger.  But she’s not Cynthia McKinney yet, so she shouldn’t have any issue winning re-election in the most heavily African-American district in Texas.  And unless she’s got a freezer full of cash she won’t face any issue winning the general.

TX-19: Mike Conaway and Jodey Arrington (Safe R)

As noted above, throwing the mid-sized cities in West Texas together also throws the congressmen for TX-11 and TX-19 together.  Whoever does win the primary would continue representing one of the most Republican areas of the country.  Arrington probably has an edge in such a matchup because Conaway is caught up in the same insider trading scandal as Culberson.

TX-20: Joaquin Castro and Will Hurd (Safe D)

Hurd would be screwed under this map had Lamar Smith from the neighboring 21st District not retired.  While his house is in the 20th, he could easily run in the 21st.  Meanwhile Castro gets to continue representing a safe, Bexar-based district.

TX-21: Lamar Smith (Safe R)

In a delightful postmodern irony, the fervently anti-science Smith is in charge of the House Science Committee.  His current district is trending Democratic and tech executive Joe Kosper (military vet, Texas accent, wears a Longhorns hat with a blazer) seems like an ideal opponent.  But Kosper lives in Austin and the new TX-21 would only include the city’s more Republican southern suburbs.  It also voted for Trump by close to a 59-36 margin, so if Hurd does run here he’ll no doubt face opposition from the right in the primary (he currently represents a district that voted for Clinton so has had to tack to the middle on occasion).

TX-22: Pete Olson and Randy Weber (Safe R)

As noted above, Weber would probably move to his old district, leaving Olson to continue representing the suburbs west of Houston in a district that got more Republican under this map.

TX-23: Henry Cuellar (Safe D)

Henry Cuellar would face a decision here.  TX-23 contains his home and voted for Clinton 57-39.  But he lives just a few blocks from its border with TX-28 and his more conservative tendencies (he voted for Kate’s Law and the 20-week abortion ban) mean he might feel more comfortable in the Valley’s most conservative district (though one that still supported Clinton by 16%).  Also, former USDA official Judy Canales lives here and is already running for TX-23 – it’d be easy for her to win a primary from the left.  Former Rep. Pete Gallego could also reconsider his decision not to run for Congress in 2018, especially if his home is moved to a district that voted for Clinton by 18%.

TX-24: Kenny Marchant and Pete Sessions (Safe R)

On paper, Marchant is the big winner of the three Dallas County Republicans, as he’s represented the most of the GOP district in the inner Metroplex.  Sessions seems to be relishing his fight for the 32nd district next year so it seems likely he’d run there under this map as well, unless Marchant retires.

TX-25: Roger Williams and John Ratcliffe (Safe R)

Thanks to the imaginations of Republican line drawers, much of Austin is represented by a Republican from the Fort Worth suburbs.  This changes under the new map, but Williams still gets a safe Republican seat to run in, and unless John Ratcliffe wants to primary him he shouldn’t have an issue retaining it.  Williams actually rents an apartment in Austin so he can live in his current district, but his primary residence is in the new TX-25 as well, so he’d welcome the opportunity to stay closer to home.

TX-26: Michael Burgess (Safe R)

TX-26 still takes in all of Denton County and wouldn’t be much of a change for Burgess.  There’s Democratic growth in the college town of Denton proper, but this is still a Republican district.

TX-27: Bill Flores (Safe R)

Flores’s home is moved from TX-17 to TX-27, but parts of this rural district are still familiar to him.

TX-28: No incumbent (Safe D)

See TX-23. The issue is which district Cuellar wants to run in.  If he runs in TX-23, then State Representatives Robert Guerra, Sergio Munoz or Ryan Guillen might emerge from the primary here.

TX-29: No incumbent (Safe D)

If Houston area Latinos voted in higher numbers you could draw two VRA districts for them entirely in Harris County. But for now the community is well served by only having one majority-Latino district (with TX-7 representing a coalition opportunity for them).  Retiring Rep. Gene Green would still likely be replaced by State Senator Sylvia Garcia.

TX-30: Eddie Bernice Johnson (Safe D)

Johnson seems like she’ll retire but hasn’t announced anything yet.  Regardless, a Democrat will represent this 45% (up from 43%) AA district come 2018.

TX-31: Michael McCaul and John Carter (Tossup)

This district narrowly went for Obama, but gave Clinton a healthy 9% margin.  As noted above, at least one of McCaul and Carter would seem likely to run in TX-17 instead.  If McCaul decides to stay and fight in a fair district, he might win: Republican congressmen Ed Royce, Mike Coffman and Erik Paulsen all have similar lifetime ACU scores and represent similarly Democratic leaning districts.  But McCaul would be new to much of the district and hasn’t had to run a contested general election before; he’d also face Kosper, current TX-31 candidate MC Hegar, or a more established Democrat like State Sen. Kirk Watson or State Rep. Donna Howard.  It’s a Tossup if he runs here, Leans Democrat if he doesn’t.

TX-32: No Incumbent (Tossup)

The old TX-32 loses its Collin County portion to TX-3 and some of its more Anglo precincts to TX-24 in exchange for some urban parts of TX-5, shifting it from a 2-point Clinton edge closer to a 10-point one.  As noted with TX-31, other Republicans represent similarly Democratic-leaning districts and Sessions appears to be ready to go down swinging in order to hold a tough district.  And while the Clinton topline is discouraging for Republicans, this district still voted for John McCain and Mitt Romney.  Game on.

TX-33: No incumbent (Safe D)

This map shows that Dallas County can support a compact 60% Latino district entirely within its borders (line drawers can get that percentage up if they want to draw a proboscis further into TX-30).  This is also a 60% Obama district, and Clinton did even better than that.  Current TX-32 candidate Ed Meier lives here, but presumably he’d face a Latino candidate in the primary (former TX-33 candidate Domingo Garcia would salivate upon seeing this map).

TX-34: Filemon Vela (Safe D)

Vela’s home base, Cameron County, is entirely within one district and he could continue his political career here with ease.

TX-35 No Incumbent (Safe D)

Current TX-23 candidates Jay Hulings and Gina Ortiz Jones are drawn out of the sprawling seat, but they’d have the chance to represent a much more compact and Democratic seat based entirely in Bexar County.  That said, neither actually lives in this district, so other candidates may emerge.

TX-36: No Incumbent (Safe R)

Rep. Ted Poe announced his retirement in November, leaving this district without a Republican incumbent.  While this is the type of suburban area that should be shifting toward Democrats, it’s still heavily Republican and will vote for a Republican for the foreseeable future.

What do you think of this map? Let us know on Twitter @50StatesofBlue or Facebook!