US House Redistricting: Texas

[muon2]

While watching the hoops his weekend, I thought I'd apply my algorithm from CA and NY to TX to compare to the various official maps. As with the other states I start with a regional division of whole counties representing an area with a whole number of districts. However, TX has lots of counties and that allows for more regions with tighter tolerance to the ideal. In this case I required all the regions to be within 0.1% of the ideal district population (ie within 698). This is the map.



The regions (district count) and deviations from ideal are:
El Paso region (2 CDs) +133
Lubbock region (1 CD) +24
Amarillo region (1 CD) -160
Brownsville region (3 CDs) -368
San Antonio region (4 CDs) -422
Waco region (1 CD) +167
Austin region (3 CDs) -130
Dallas region (9 CDs) +33
Sugar Land region (1 CD) -182
Houston region (8 CDs) +262
Beaumont region (3 CDs) +644

I'd start from the following map, which are the state planning regions.



I'd let the citizens of each county decide whether they wanted to switch to another apportionment region.  The regions are somewhat imposed, and are centered on cities.  Some of the outer counties like Erath, Walker, Llano, and Matagorda would probably want to switch.   This might be on the ballot in November of the -0 year.

The next step would be to adjust the areas so that they are close to an integer multiple of a number of districts.  The 19 areas with a population less than 1.5 districts, have a total population equivalent to 10.36 representatives.

I would dissolve one at a time, until the number of such areas equals the total number of representatives apportioned to them (eg at least 9 would be dissolved).   Redistricting juries in each county would choose where their county is switched to.  So the counties in the Concho Valley (San Angelo) region would choose first.

There would eventually be 15 or so regions which would each be apportioned an integer number of districts.  Each region would have a surplus or a deficit.

Regions with the deficits would select areas from adjacent regions with a surplus or smaller deficit.  Regions with surpluses would release areas to adjacent regions with a deficit or smaller surplus.

Then regions apportioned multiple districts would be divided.

What fascinates me is that I can picture my regions emerging from yours through the process you describe. I not saying that those areas would choose to join, but my regions look like they were made by combining the ones in your map, and then adjusting a few counties to get them close to a whole number of districts.

[muon2]

https://docs.google.com/file/d/0BxeOfQQnUr_gOEx1X0dxbllTS3VhVmxtRk9aSjJ6QQ/edit

Ultimately, the court decided that - on an interim basis - the loss of CD-25 was offset for the district’s Hispanic population by inclusion of most of the Hispanic population in the new CD-35 and that the impact on the district’s African-American population was offset by the creation of the new CD-33 in North Texas.

I don't interpret it that way.

23 is kept at benchmark performance;

33 is to address issues of fragmentation of the minority population in the DFW area, and the court doesn't believe a compact Hispanic CVAP majority district can be drawn.

35 addresses statewide retrogression claims, and 25 is not protected.   And because it is not possible to create 8 compact Hispanic districts in south Texas (maybe the Supreme Court will decide that 15, 23, 25, 27, 28, 33, 34, and 35 are all non-compact),

Is the court going to use CVAP? I noticed that submitted documents typically reference Spanish Surname Registered Voters, rather than CVAP.

I think that SSVR is used to estimate registration of Hispanic voters since Texas not use race as a voter qualification.  I think that in general that they are using reconstructed election results, as the actual test.  It gets really messy when you start trying to figure out whether you not only have to choose enough minority voters, and factor in whether they vote the right way or, vote at all even though eligible, and whether other voters vote the same way.

In any case I didn't see a CVAP analysis from any of the parties like I did in CA. CVAP also is an estimate since it isn't from 2010 and is a sample. I assumed it wasn't relevant in the TX circuit. It's not used in IL under the 7th circuit.

The interesting feature of SSVR, is when a district is over 50% SSVR and votes solidly GOP. I would conclude that either there isn't much polarized voting compared to other areas of the state. My example is CD 2 in my map - 50.2% SSRV and 60%+ GOP.

The DC circuit rejected the use of CVAP for Section 5 purposes.   The USDOJ claims you can't determine "ability" to elect without checking voting results.  This was the interpretation of the Florida Supreme Court of its constitutional amendments (rather than simply saying districts must conform to federal law, Florida incorporated the standards of the VRA into their constitution).

That seems to put SSRV on a stronger footing than CVAP as a proxy for voting strength. Voting analysis can follow. An ecological analysis is probably needed to discern whether the minority engages in sufficient bloc voting to meet the Gingles test for section 2 districts.

[muon2]

There are some implicit pairing of regions which will get a lot closer to integer apportionment.

Panhandle-Nortex + South Plains-West Central

I might end up shifting Wise to the northern area, and using the surplus from the southern district below.

El Paso + Trans Pecos-Permian Basin-Concho Valley

Will need a shift from El Paso County

My version that used the middle Rio Grande instead of the Concho Valley was to get a 50%+ SSRV district. If that's not a factor, I like your compactness.

You must be logged in to read this quote.

Which is pretty much what I did for my East TX region with 3 CDs.

You must be logged in to read this quote.

But you can't deny that the pairing gets to almost exactly 3 CDs. Once those are paired, are the fajita strips really better than my Laredo-Corpus link?

You must be logged in to read this quote.

I found it much easier to cut Ellis and Navarro out. The remaining population is very close to 9 CDs, and it facilitates drawing VRA districts in Dallas County (in case that matters here). Those excess counties on the south then move to ...

You must be logged in to read this quote.

I also had 5 CDs in Harris, and I also used the Galveston-Brazoria and Montgomery-Liberty links. though I came in from College Station instead of Fort Bend, in part because I could so precisely make a CD with Fort Bend and other counties outside of Harris. I think your region will force a split of Fort Bend.

[muon2]

As I read the opinion I was struck at the number of times the decision pointed out the differences between sec 5 and sec 2 so they could come to their conclusions. That may also be something that gives SCOTUS pause. I also found the dissent on CD 25 far more compelling than the majority from a methodological viewpoint.