Fair redistricting: Texas

[Sol]

That map also chops a few munis in SE Mecklenburg.

[jimrtex]

Mecklenburg - Detail

The idea of following I-485 around makes sense if it starts around I-85 where it crosses the Catawba river from Gaston county. Connecting them across Lake Wylie without a bridge, seems akin to jumping across the Albemarle sound to connect Tyrrell to a district with Perquimans and Chowan county without including Washington county.

A disadvantage of using DRA, and not having time nor inclination to calculate PVI etc. for districts based on census data. Have you used the .csv option in DRA?

[jimrtex]

Better?



The north precinct line for 200 is not on I-85, but rather US 29-US-74, that also crosses the river. But I went ahead and included 224 which is mostly unincorporated. 230, 19, 87, and 88 cross I-485. If possible I would trim them on I-485, which would expand the neck somewhat.

The area along the Union line includs all of Mathews and Mint Hill in NC-3. I included 91 and 119 west of Mathews for population equality, and I already have cut Charlotte.

[Dr. MB]

Here's my plan.

SC-01: R+1
SC-02: R+16
SC-03: R+21
SC-04: R+16
SC-05: R+5
SC-06: D+13
SC-07: R+11

So four safe R districts, one lean R district, one swing district, and one safe D district.

Dist. 6 (the Columbia one) is plurality-black (48%).

There are 6 county splits.

[jimrtex]

I also redid the Guilford and Wake maps.



This places all of High Point and Jamestown in NC-7. Previously the northern part of High Point was omitted. If an attempt were to keep Greensboro and High Point together with Winston-Salem would require NC-7 to wrap around to the northern part of the county.



NC-7 includes Holly Springs. NC-9 includes Apex and Cary (and Raleigh and High Point). NC-10 includes Fuquay-Varina, Garner, Wendell, and Zebulon. The NC-10 portion now is in a single contiguous strip.



Right click for a larger map.

Updated partisanship is in red.

NC-1 (Appalachian, Asheville) +0.05%, R+7.71
NC-2 (Western, Unifour) +0.05%, R+20.60
NC-3 (Charlotte suburbs) -0.08%, R+14.55 R+13.81
NC-4 (Charlotte) -0.08%, D+16.43 D+15.41
NC-5 (Charlotte/Winston-Salem/Greensboro suburbs) +0.05%, R+18.66
NC-6 (Greensboro/Winston-Salem) +0.05%, D+6.63 D+5.98
NC-7 (South Central) +0.05%, R+9.42 R+9.03
NC-8 (North, Durham/Chapel Hill) -0.21%, D+12.69
NC-9 (Raleigh) +0.05%, D+9.11 D+9.27
NC-10 (Raleigh suburbs) +0.05%. R+9.89 R+9.74
NC-11 (Southeast, Fayetteville) +0.05%, D+0.25
NC-12 (Coast, Wilmington/New Bern) -0.04%, R+10.80
NC-13 (Albermarle, Rocky Mount) +0.05%, D+2.08

[GM Team Member and Senator WB]

Alright, here are my two submissions.

This one has a minority coalition district (Yellow).

01: -2,638, R+12.7
02: -614, R+18.77
03: -678, R+13.22
04: +420, R+7.4
05: (46.7% white) +160, D+10.45
06: +2,097, R+9.19
07: +1,255, R+4.87

5 county splits

And now, no majority-minority districts.

01: +867, R+12.62
02: -614, R+18.77
03: -678, R+13.22
04: +693, D+0.32
05: -227, D+0.99
06: -533, R+4.89
07: +494, R+7.47

6 County splits.

[muon2]

I don't intend this as a submission, but I thought it might be interesting to some on the thread. I posted this plan back in 2011 a few months after the new data was out. Both CD 6 and CD 7 are BVAP majority at 50.2% and 50.1% respectively. All CDs are within 100 of the population quota without splitting any voting districts.

[Torie]

Putting aside the mechanics of how a final map is chosen, how does the scoring for this exercise vary from the Muon2 rules, if at all?

[muon2]

I offered some muon rules scoring early on, but that has only figured in once when there was an unusually large number of submissions. I got a lol about their use from one of the panelists, so I'm probably not going to invest the time to do a calculation on plans other my own. As I noted on on of the other thread in the series, I am working with a research group and a master's student on some algorithm design related to the muon rules, so I may score some interesting maps to help with that.

[cvparty]

I got a lol about their use from one of the panelists, so I'm probably not going to invest the time to do a calculation on plans other my own.

omg no! i was saying LOL to solid who said “Sol is actually getting better scores with his very flawed plan than other plans that actually use sensible design methodology”

[jimrtex]

South Carolina has 10 RCOG's.



Central Midlands minus Fairfield and Newberry plus Calhoun has a population of 1.002 and keeps the Columbia UCC whole.

Berkeley-Charleston-Dorchester is coincident with the Charleston UCC and has a population of 1.006.

Waccamaw and Pee Dee minus Chesterfield has a population of 1.004.

The rest of the regions don't fit together particularly well, so I went directly to drawing districts.




The rest of the districts were fitted in. Greenville, Pickens and Oconee are equivalent to 0.975 districts, but this would require putting Spartanburg and Anderson in a very ugly district. This version cuts across the northern part of the county to link Spartanburg with Pickens and Oconee. It takes 27% of Greenville. The portion of Greenville plus Pickens in NC-1 is 32% of the Greenville UCC.



NC-1 includes Greer which sprawls across the Spartanburg-Greenville line, as well as Taylors CDP. NC-2 includes the near northern suburbs of Wade Hampton CDP, Sans Couci CDP, and Berea CDP as well as the city of Greenville proper and the southern part of the county. Keeping clear of Greenville explains the diagonal slope of the boundary.


SC-1 (Appalachian, Spartanburg) -0.65% R+21.24, A78, B14, H5, As2, O1.
SC-2 (Upcountry, Greenville) -0.92% R+13.98, A69, B22, H6, O1, As1.
SC-3 (Catawba-Santee-Lynches, Rock Hill) -0.29% R+5.90, A62, B31, H4, O1, As1.
SC-4 (Midlands, Columbia) +0.20% D+0.18 A58, B33, H5, As2, O2.
SC-5 (Pee Dee, Florence, Myrtle Beach) +0.41%, R+7.31 A62, B32, H4, O1, As1, AI 1
SC-6 (Ashley-Cooper&Atlantic, Charleston) +0.58%, R+3.81 A63, B27, H5, O2, As2.
SC-7 (Savannah, Aiken, Orangeburg, Hilton Head) +0.66%, R+4.42, A56, B35, H7, O1, As1.

Standard Deviation (0.58%)

One split county.

[True Federalist (진정한 연방 주의자)]

The RCOGs are invisible to the average citizen and are but one of multiple ways of grouping counties that exist in South Carolina.  The judicial circuits and the technical college system have far more impact. There probably other groupings that split the State without even counting special purpose multi-county areas that aren't established on a statewide basis.  One of the major weaknesses of South Carolina local government is the very real lack of coordination and high degree of geographic overlap.  (Which is also why school district and municipal boundaries will sometimes cross county lines.)

[True Federalist (진정한 연방 주의자)]

Here are my two maps.
Map 1

District	Deviation	PVI
1	29	R+3.87
2	-130	R+10.25
3	54	R+18.35
4	34	R+18.06
5	272	R+10.31
6	-180	D+13.61
7	-77	R+8.43

CD 6 is black plurality (49.6% B/42.9% W)

While Edisto Beach is part of Colleton County, the road to it is via Charleston County, so I included it in CD1 rather than CD 2 despite it adding another county chop to do so.
Other split counties include:
Chesterfield (CD's 5&7)
Dorchester (CD's 1&2)
Lexington (CD's 2&3)
Orangeburg (CD's 2&6)
Pickens (CD's 3&4)
Spartanburg (CD's 4&5)
Map 2

This is an attempt at a non-VRA Republican gerrymander that might be considered fair by other measures.

The justification for the four stringy districts is that CD 2 is the Savannah River corridor, CD 4 is the I-26 corridor, CD-5 is the I-77 corridor, and CD 7 is the coastal corridor.

District	Deviation	PVI
1	-267	R+3.62
2	-247	R+13.79
3	-215	R+18.57
4	608	R+10.63
5	471	R+0.41
6	-397	R+2.06
7	49	R+6.83

CD 6 has the highest minority percentage (54.7% W/39.8% B)

Split counties include:
Georgetown (CD's 1&7)
Horry (CD's 6&7)
Kershaw (CD's 5&6)
Lexington (CD's 2&4)
Laurens (CD's 3&4)
Richland (CD's 4&5)

I could stay within the 0.5% deviation limit and get rid of the chops of Georgetown and Kershaw.

[True Federalist (진정한 연방 주의자)]

The rest of the districts were fitted in. Greenville, Pickens and Oconee are equivalent to 0.975 districts, but this would require putting Spartanburg and Anderson in a very ugly district.

Not really.  But you did put yourself in a box by having a Lexington-Richland-Calhoun district which makes sense only if minimizing county chops are your primary criterion without considering whether counties go together.  Calhoun is much, much, much more connected to Orangeburg than Richland or Lexington.  Except for US 601 at the extreme eastern edge of the border, there is no way to get between Richland and Calhoun. Besides the RCOG, Calhoun and Orangeburg share a technical college, a hospital, and a judicial district.  It's primary reason for being made a county was that was back in the days when each county got a State Senator of its own.  Indeed, the highly arbitrary nature of many county borders in this State make it ludicrous to make minimizing county chops the primary criterion for district boundaries. Not that they shouldn't be considered, but don't dominate like they do in most other States.

[jimrtex]

The RCOGs are invisible to the average citizen and are but one of multiple ways of grouping counties that exist in South Carolina.  The judicial circuits and the technical college system have far more impact. There probably other groupings that split the State without even counting special purpose multi-county areas that aren't established on a statewide basis.  One of the major weaknesses of South Carolina local government is the very real lack of coordination and high degree of geographic overlap.  (Which is also why school district and municipal boundaries will sometimes cross county lines.)

The RCOG's match the Technical College districts with the exception of Newberry and Union counties, and that may represent a more modern sensibility as Union is drawn toward Charlotte, and Newberry is drawn towards Columbia (down the interstate). The Technical College districts are often smaller than the planning regions. Williamsburg has its own technical district - presumably because it doesn't fit with Florence, Myrtle Beach, or Charleston. It couldn't have its own RCOG so somebody chose an RCOG for it.

I don't see how in the name of Vic Ralls that you think that average citizens are familiar with circuit court districts.

[True Federalist (진정한 연방 주의자)]

The RCOGs are invisible to the average citizen and are but one of multiple ways of grouping counties that exist in South Carolina.  The judicial circuits and the technical college system have far more impact. There probably other groupings that split the State without even counting special purpose multi-county areas that aren't established on a statewide basis.  One of the major weaknesses of South Carolina local government is the very real lack of coordination and high degree of geographic overlap.  (Which is also why school district and municipal boundaries will sometimes cross county lines.)

The RCOG's match the Technical College districts with the exception of Laurens and Union counties, and that may represent a more modern sensibility as Union is drawn toward Charlotte, and Laurens is drawn towards Columbia (down the interstate). The Technical College districts are often smaller than the planning regions. Williamsburg has its own technical district - presumably because it doesn't fit with Florence, Myrtle Beach, or Charleston. It couldn't have its own RCOG so somebody chose an RCOG for it.

I don't see how in the name of Vic Ralls that you think that average citizens are familiar with circuit court districts.

More so than than the COG's, as at the very least they vote for the solicitor thereof, so they at least have a decent chance of knowing they exist and which one they are in.  You do need to double check on your county names, as that's Newberry, not Laurens that got included in the CMCOG. Both the technical colleges and the COG's got set up in the 60's.  The technical colleges were generally set up with more local input than the top-down COG's and not all at the same time.  For instance, the reason Williamsburg has its own technical college is that facility also doubles as a vocational education center for high school students which is what that county wanted.

Also, for Union County, it makes far more sense to be included with Spartanburg as it is for technical schools since it's in the same MSA.  Union isn't even in the Charlotte CSA.  The only reason I can think of for why the Appalachian Council of Governments was chosen to be those six counties is in its name as those are the seven counties included in the Federal Appalachian Regional Commission which was also set up in 1965. I wouldn't be surprised if the only reason the COG's got set up at all was because of Federal prompting.

[muon2]

Here's my real entry for SC. The main goal was to keep the CDs as close to whole county as possible while preserving UCC covers. CDs 5,6,7 are whole county. Only three counties are chopped between the other CDs and if the chops were removed to make them whole county, CDs 1,2,3,4 are each within 5% of the quota.

CD 1 is packed entirely within the Charleston UCC and would be a competitive district. CD 6 is D+6 with BVAP 38.7%, and would likely elect the candidate of the black minority by controlling the Dem primary. CD 2 is R+1 with 36.4% BVAP and is a swing district where the black minority would be able to control the Dem primary.



CD 1: -285; R+3.8
CD 2: +2103; R+1.0
CD 3: -3159; R+18
CD 4: -2152; R+19
CD 5: +454; R+14
CD 6: +312; D+6.4
CD 7: +2729; R+7.3

[jimrtex]

This provides more equal population (Standard Deviation 0.06%)



The major change is placing Newberry and McCormick in SC-7, and most of Greenwood in SC-2, making both more compact.

It would be possible to use this division of Greenwood, to permit a whole-county district of SC-1 of Greenville, Pickens, and Oconee. SC-2 would then be Cherokee, Spartanburg, Laurens, Greenwood (part), Abbeville, and Anderson. There is tradeoff between division of the larger county, vs. a major division of a smaller county. The city of Greenwood is in SC-2. I suspect that if people in Greenwood were car shopping or seeking entertainment they would head north.

There are also minor tweaks to Lexington (1401 persons, 1 precinct); Darlington (2068 persons 1 precinct) amd Dorchester (4341 persons, 3 precincts), representing a minimal 7810 victims of population equalization.

SC-1 Appalachian  -0.06%, R+21.22
SC-2 Upcountry -0.04%, R+14.05
SC-3 Catawba-Santee-Lynches 0.02%, R+5.96
SC-4 Midlands -0.01%, D+0.31
SC-5 Pee Dee 0.10%, R+7.25
SC-6 Ashley, Cooper & Atlantic -0.08%, R+3.81
SC-7 Savannah 0.07%, R+4.59

[True Federalist (진정한 연방 주의자)]

Here's my real entry for SC. The main goal was to keep the CDs as close to whole county as possible while preserving UCC covers. CDs 5,6,7 are whole county. Only three counties are chopped between the other CDs and if the chops were removed to make them whole county, CDs 1,2,3,4 are each within 5% of the quota.

A nice try, but that CD 6 of yours looks good only on a map that ignores roads.  Calhoun is not at all well-connected to Richland County thanks to the Congaree National Swamp. In my opinion, county government (and county borders) are just not important enough in this State to subordinate all other considerations to the holy grail of no county chops.

Oh, and jimrtex, there's no need to invent a new name for the Dorchester, Berkeley, and Charleston combo.  Trident works quite well and is used for far more than just the local technical college.

[jimrtex]

Oh, and jimrtex, there's no need to invent a new name for the Dorchester, Berkeley, and Charleston combo.  Trident works quite well and is used for far more than just the local technical college.

That's not a new name.

[jimrtex]

This an an alternative that keeps Greenville whole. SC-1 takes a nibble out of the corner of Anderson, which is Greenville suburbs. SC-7 takes part of Greenwood south of the city of Greenwood. In total about 30K persons are chopped.



Anderson



Greenwood



SC-1 (Appalachian, Greenville) -0.09%, R+18.77 A76, B14, H7, As2, O1.
SC-2 (Upcountry, Spartanburg, Anderson) -0.35% R+16.41, A71, B21, H5, O1, As1.
SC-3 (Catawba-Santee-Lynches, Rock Hill) -0.29% R+5.90, A62, B31, H4, O1, As1.
SC-4 (Midlands, Columbia) +0.20%, D+0.18 A58, B33, H5, As2, O2.
SC-5 (Pee Dee, Florence, Myrtle Beach) +0.41% R+7.30 A62, B32, H4, O1, As1, AI1.
SC-6 (Ashley-Cooper&Atlantic, Charleston) +0.58%, R+3.80, A63, B27, H5, O2, As 2.
SC-7 (Savannah, Aiken, Orangeburg, Hilton Head) -0.46%, R+4.48, A56, B35, H7, O1, As1.

Standard deviation 0.37%.


Shifting 2068 persions, 1 precinct, from Darlington; 1401 persons, 1 precinct, from Lexington; 3368 persons, 3 precincts from Dorchester, and 2256 persons, one precinct from Greenwood reduces the standard deviation to 0.07%.

In total, 34,531 persons would be victimized.

[True Federalist (진정한 연방 주의자)]

Oh, and jimrtex, there's no need to invent a new name for the Dorchester, Berkeley, and Charleston combo.  Trident works quite well and is used for far more than just the local technical college.

That's not a new name.

Who uses it? Granted, I don't live in the Trident myself, but I find it hard to believe that anyone but a bureaucrat would willingly use such an ungainly appellation as Ashley-Cooper&Atlantic.

[muon2]

I showed that it was possible to draw two CDs with majority BVAP. Though the VRA is not a consideration, communities of interest are. The black population in southern SC would qualify as a community of interest.

The goal is also fair maps, and that can mean partisan fairness; certainly that's what's being litigated in the courts. SC is an R+8 PVI and that projects to an ideal delegation of 4R, 2D, and 1 even seat in PVI to have now skew. My submission had a delegation of 5R, 1D, and 1 even.

As an alternate plan B, I can add 3 county chops for a total of 6 and lose the Charleston UCC pack. The new CD 2 is D+4.7 with a 43.7% BVAP, so it moves the plan closer to partisan fairness (5D 2R) while providing an opportunity to elect the black candidate of choice.

Here's the original map and a zoom showing the modified CD 1 and CD 2. The Charleston chop follows I-26 into central Charleston. The Aiken chop is shifted and there is a minor adjustment of the Anderson chop to balance population. Revised stats follow the map.





CD 1: -647; R+7.9
CD 2: -858; D+4.7
CD 3: -1374; R+18
CD 4: -614; R+19
CD 5: +454; R+14
CD 6: +312; D+6.4
CD 7: +2729; R+7.3

[muon2]

Here's my real entry for SC. The main goal was to keep the CDs as close to whole county as possible while preserving UCC covers. CDs 5,6,7 are whole county. Only three counties are chopped between the other CDs and if the chops were removed to make them whole county, CDs 1,2,3,4 are each within 5% of the quota.

A nice try, but that CD 6 of yours looks good only on a map that ignores roads.  Calhoun is not at all well-connected to Richland County thanks to the Congaree National Swamp. In my opinion, county government (and county borders) are just not important enough in this State to subordinate all other considerations to the holy grail of no county chops.

US 601 meets my definition of a connection between Richland and Calhoun. Even if it is indirect it provides a path go go between county seats on numbered highways without going into a third county. You'll see in my alternate version I'm willing to trade county chops for representational fairness. I just want a defined metric to justify the chops.

[muon2]

Since SKEW predicts an ideal delegation of 4R, 2D, and 1e, I wanted to see what that might look like. This modification of my submitted plan has only 5 county chops, and all populations are within 1000 of the quota. (There is a small chop of CD 7 into Charleston that is hard to see.) CD 2 is D+2.0 (BVAP 37.5%), CD 6 is D+6.4 (BVAP 38.7%), and CD 7 is R+0.1 (BVAP 39.1%), giving the desired partisan division for a SKEW of 0.



SC poses some interesting questions about what makes a map fair. Is it geography alone in terms of definitions like chops and erosity? Is it adequate opportunity for minority representation? Is it partisan fairness? If they are all significant, what's the right balance between them?

If I knew whether partisan fairness or minority representation was more important, I'd know which of my alternate plans to submit as a second choice. More importantly, I'd have a sense of how to weight the variables in upcoming states.