Fair redistricting: Illinois

[muon2]

This is a plan I posted in Nov 2011,so I'll submit it here. It chops no counties except Jefferson/Louisville and was designed to have compact districts. The concept of UCCs didn't exist then so it has one extra cover chop. We also didn't have an erosity measure back then, but I think it should score well. The drf file was long ago deleted, so I've had to reconstruct the plan from scratch.

Edit - I looked at the impact of the UCC and it turns out that by shifting just 6 counties I can preserve the erosity, improve the inequality and keep the UCC whole. CD 6 uses the I-64 corridor to link Lexington to Ashland.

As a side note, if the chop into Jefferson comes from the south instead of the southeast, the PVI drops from D+5.7 to D+5.0, which just pushes it into the competitive range. I'm not doing that since the other CDs are so Pub, and shifting the chop costs a point of erosity.



CD 1: -1234; R+20.21
CD 2: +862; R+20.69
CD 3: -133; D+5.67
CD 4: +697; R+17.76
CD 5: -890; R+30.25
CD 6: +698; R+10.52

[muon2]

As a reminder, there are two UCCs in KY. The UCCs were created to recognize metro areas as community of interest and avoid fragmenting their population into adjacent rural areas.

The Louisville UCC consists of Jefferson, Bullitt, and Oldham; it is 1.21 times a CD so there should be two CDs covering the three counties and one CD entirely within to meet the UCC test.

The Covington UCC consists of Kenton, Boone, and Campbell; it is 0.51 of a CD and only needs one CD to cover it.

The Ashland UCC consists of Boyd and Greenup counties; it is 0.12 of a CD and only needs one CD to cover it.

[muon2]

I count three proposals for KY. Since all the plans are solid 5 R - 1 D, they will all get a SKEW of 1 and POLARIZATION of 12. I will treat jimrtex's plan for Jefferson the same as the other 2. The complete scores are:

Plan	-S-	-P-	-I-	-C-	-E-
Sol	1	12	4	4	50
muon2	1	12	6	1	48
jimrtex	1	12	10	1	52

SPICE scores would eliminate jimrtex by muon2.

[muon2]

Sol's map wins for Kentucky!

Were the results published somewhere?

would you like me to make a spreadsheet of all the state votes?

Yes.

kk i'll make it over the weekend (super busy this week )

What's the status of the KY results? I have more than a passing interest. I'm involved with some graduate level academic research on redistricting algorithms, and any feedback about which maps were preferred over others may find its way into that research. Even better would be comments about the maps, but that doesn't seem to be the way panelists want to react. Of course if people want to say why they voted for a particular plan, I'll dutifully make note.

On a side note: Shouldn't Sol be moved into the open Dem spot? Then a Pub or at least an indy would take the second R spot.

[muon2]

Sol's map wins for Kentucky!

Were the results published somewhere?

would you like me to make a spreadsheet of all the state votes?

Yes.

kk i'll make it over the weekend (super busy this week )

What's the status of the KY results? I have more than a passing interest. I'm involved with some graduate level academic research on redistricting algorithms, and any feedback about which maps were preferred over others may find its way into that research. Even better would be comments about the maps, but that doesn't seem to be the way panelists want to react. Of course if people want to say why they voted for a particular plan, I'll dutifully make note.

On a side note: Shouldn't Sol be moved into the open Dem spot? Then a Pub or at least an indy would take the second R spot.

The KY maps are all so similar, that it is hard to generate much passion about which map is best. The Sol and Muon2 maps appear to be  almost identical, except that Sol has more chops to get down inequality (not my bag, but that is a matter of taste), while the Jimrtex has some extra erosity that appears not to really be necessary (with that jut into Laurel County looking particularly unfortunate).

I tried to take a different approach for Lexington. As opposed to making it a cluster of counties surrounded by other districts, I ran it out to WV along the I 64 corridor. It turns out that had a positive effect in reducing erosity. However, one of my early observations is that people are most comfortable voting for district designs they are used to seeing in their respective states. The fact that other designs might perform better on objective metrics doesn't seem to convince voters.

[muon2]

Sol's map wins for Kentucky!

Were the results published somewhere?

would you like me to make a spreadsheet of all the state votes?

Yes.

kk i'll make it over the weekend (super busy this week )

What's the status of the KY results? I have more than a passing interest. I'm involved with some graduate level academic research on redistricting algorithms, and any feedback about which maps were preferred over others may find its way into that research. Even better would be comments about the maps, but that doesn't seem to be the way panelists want to react. Of course if people want to say why they voted for a particular plan, I'll dutifully make note.

On a side note: Shouldn't Sol be moved into the open Dem spot? Then a Pub or at least an indy would take the second R spot.

I actually want the panelists to make comments, but it’s so difficult to have the others even just vote. I’d be happy to share my opinion though. I thought Sol’s grouping of Lexington and Frankfort made sense. I didn’t particularly like the way your map split that area among four districts. Also, Sol’s KY-05 and KY-01 reflected the ancestrally Dem parts of the state well imo. I look at pop density and voting trend maps primarily in forming my ideas about good district lines

About Sol’s role, there aren’t many Republicans so I’m just having Sol acting as a Republican. If a Republican does want to join then I would move Sol to a Dem. I know there might be concerns about partisan skew, but having 2 Ds and 2 Rs isn’t for them to want to gerrymander for their party, we’re all trying to follow the fair nonpartisan goal

Aren't you concerned that by considering voting trends you might fall prey to subtle gerrymanders? In many states that have enacted or considered neutral reforms, they explicitly require that the maps be made without considering election data to avoid that very thing. IA is a good example of an independent mapping process that bars the map drawers from considering election results.

As an aside, my wife has family in the Lexington area, so I am personally familiar with it. Frankfort isn't that closely affiliated with Lexington. As the state capital it's kind of a shared area by the major metros. I find that's often true when the capital is outside of one of the big cities in the state.

[muon2]

Here are my two submissions for IN. Both are from the same 2015 thread that included my CO map. Since PVI's weren't available then, I'll add them later this evening. The Indianapolis UCC (Marion , Hamilton,  Hendricks,  Johnson,  Madison,  Hancock) ideally is covered by 3 CDs with 2 packed into thsoe counties. The overall state PVI of R+9 would predict that a 3D-6R would be expected to avoid any skew.

muon2-A
There are no county chops except Marion, and there are no township chops in Marion. UCC cover rules are followed with one less pack than the maximum, so the chop score is 1. The erosity is 81, and the range is 0.89% for an inequality of 12. Politically the districts are 2D, 1e, 1r, 5R based on 2008.



CD 1: -3498; D+5.8
CD 2: -2685; R+8.2
CD 3: +2667; R+18
CD 4: -88; R+16
CD 5: -1451; R+18
CD 6: +2884; R+14
CD 7: -2859; D+13
CD 8: +2790; R+14
CD 9: +2244; R+12

2008 was an unusually good year for Dems in IN. CD 2 in this map was (e)ven R+0, but now has an uncompetitive PVI R+8. The lean Pub CD also shifts hard R, leaving this plan as a 2D, 7R. Of course when this map would have had to be designed those future shifts would be unknown.

muon2-B
All UCCs are maintained with minimal chops. Marion has two chops and a township microcrop, plus there is one other chop to equalize population and minimize erosity, putting the chop score at 4. The erosity is 66, and the range is 0.53% for an inequality of 8. The 2008 partisan breakdown is 1D, 1d, 1e, 3r, 3R.



CD 1: +1050; D+5.7
CD 2: -2252; R+8.1
CD 3: +1547; R+18
CD 4: -2011; R+11
CD 5: -326; R+5.6
CD 6: +173; R+18
CD 7: +841; R+0.7
CD 8: +746; R+16
CD 9: +236; R+11

The partisan shifts from 2008 to 2016 have even more effect on this plan. keep in mind these are shifts compared to the national averages, so that even as the presidential vote was shifting more Pub, the IN vote was shifting even more so. It also shows that though Indy is a huge Dem vote sink, trying to split that to get two CDs could potentially cost Dems a seat.

[muon2]

Packing urban areas together too heavily guarantees a polarized map with fewer meaningful contests. Having some competitive seats means that when there is a swing in the mood of the electorate, the results respond with a changed delegation. If every state in the country only grouped like socioeconomic areas to form districts, I suspect that wave elections would have very little impact on Congress.

[muon2]

Packing urban areas together too heavily guarantees a polarized map with fewer meaningful contests. Having some competitive seats means that when there is a swing in the mood of the electorate, the results respond with a changed delegation. If every state in the country only grouped like socioeconomic areas to form districts, I suspect that wave elections would have very little impact on Congress.

Although I like the sound of this idea, it doesn't make as much sense when it's put into place, tbh. How would we do this in Massachusetts? Oklahoma? West Virginia? Unless we gerrymandered these states to oblivion, we won't get even close to representational government. So if we do this in states like Indiana, attempting to create competitive seats for the sake of competitiveness, we are putting ourselves in a position where in order to be fair we have to do this everywhere. So while sure, it's good to have competitive districts, I'm not a huge fan of arbitrarily placing republican voters in a democratic district to spread out the potential democratic wins (or visa versa, obviously.)

I agree that drawing competitive districts solely for that reason generally results in some pretty bad districts. I'm advocating for competitive districts when they are a reasonable option.  If my choice was between two plans that were roughly equal in geographic characteristics, then I would prefer a plan that provides more districts that could change with the mood of the electorate. That's not going to happen in all states all the time.

If you look at my IN plans, both had more competitive districts in 2010 than they do today. That reflects changes in the electorate, but if all were D+10 and R+10 or higher, there would be little chance that they could become competitive over the course of a decade. IN could just as well shifted in ways that made their districts more competitive as in other states, but that it didn't happen doesn't mean it should be excluded at the beginning of the decade.

[muon2]

In 2011 OH held a redistricting competition with strict criteria to judge submitted plans. The criteria weren't the same as the muon rules. For instance they counted county fragments instead of chops, except that is the chop was only the part of a city that crossed the line it wasn't a chop. Whole districts within a county weren't fragments either, even if they included pieces of cities across the line to keep them whole. The competition used different geographical blocks than DRA and scored compactness based on the size of a bounding circle.

So I can't resist submitting that winning plan, slightly reworked to fit DRA and the recommendation that cities and towns not be chopped more than is reasonable. This plan keeps subunits whole, using wards within Columbus and Cleveland where they are chopped. The cities of  Crestline, Dublin, Westerville, Buckeye Lake, Reynoldsburg, Plain City and Middletown are kept intact as if they were all in one county. Using the competition measurements there are 8 fragments in 5 chopped counties.



CD 1: +782; D+5.0
CD 2: +2485; R+20
CD 3: +1389; R+2.4
CD 4: -3193; R+18
CD 5: -2730; R+0.8
CD 6: -271; R+13
CD 7: -3363; D+2.9
CD 8: +1756; R+18
CD 9: +1630; R+2.3
CD 10: -1033; R+2.7
CD 11: +2554; D+32; BVAP 46.8% (48% at the block group level, met the standard set by the Urban League for the competition)
CD 12: -419; R+16
CD 13: +2730; R+2.5
CD 14: -419; R+2.9
CD 15: -3269; D+9.7
CD 16: +201; D+1.3

It is interesting to see how the districts here have shifted since the competition. At that time, 8 of those were considered D districts: 1, 5, 7, 11, 13, 14, 15, 16, and 9 was rated a tossup. Also 11 had a PVI of 2.5 or less. Now only 5 rate D. It shows the dangers of trying to guess the partisan composition a decade out if the districts are designed to be competitive.

Here's the detail for the Columbus area illustrating cities that are kept together in a CD and not penalized with a chop/fragment. There are 7 in this view, including two from one city.

[muon2]

I'll use this opportunity to renew my call for comments on all maps from those voting, but even others are welcome to say what they like or not about any of the maps.

[muon2]

We've probably looked at more MI maps on this board than those of any other state. Many of the muon rules were refined looking at plans for MI. Urban county clusters gained prominence in 2013 as a tool to avoid splits of the Lansing metro counties, which when separated could result in good scores with few chops, but a mess in that area. Later they too were refined with other states.

In 2015 we returned to MI to do a detailed scoring of a number of plans. To fascilitate our exercise then I identified the county connections, UCCs (in pink with minimum cover size), subunits and connections in the big 3 Detroit counties, and subunits within Detroit.

[Img=left]https://uselectionatlas.org/FORUM/GALLERY/256_31_01_15_11_08_35.png[/img]

Here are the subunits of Detroit that would be used to compute chops and erosity. The neighborhood clusters have been matched to the extent possible to the DRA precincts. Connections are based on the actual neighborhood cluster boundaries.



NC 1: pop 66,076; BVAP 82.6%
NC 2: pop 81,321; BVAP 91.3%
NC 3: pop 79,779; BVAP 88.7%
NC 4: pop 61,346; BVAP 75.5%
NC 5: pop 74,720; BVAP 27.5%, HVAP 48.8%
NC 6: pop 58,410; BVAP 94.8%
NC 7: pop 88,225; BVAP 83.6%
NC 8: pop 72,117; BVAP 88.1%
NC 9: pop 78,604; BVAP 96.0%
NC 10: pop 53,179; BVAP 93.2%

[muon2]

In the aforementioned exercise there were a number of plans that had good scores. Unfortunately plans from Torie and traininthedistance were posted on Photobucket and aren't visible anymore. This was my best plan from that exercise, though the partisan caclulations do not reflect the latest PVI values.

Here's the rescore of muon2 B

Here's my plan to reduce chops in MI without relying on microchops. I used a UCC chop of whole county Livingston and a threshold of 47% BVAP for the Detroit CDs. This allowed the removal of one chop in Oakland and a chop outside of the Detroit UCC.

The Muskegon chop is a macrochop and the townships are used to determine cut links there. The other two outstate chops are small. The chop in Ionia isn't a microchop, but it could be if it were moved to the SW corner of Eaton. However if microchops get no advantage as county or UCC chops then to place it there would be counterproductive, despite a better shape.

The Detroit CDs are 48.3% and 47.5% BVAP for CDs 13 and 14 respectively. It's quite possible that the BCVAP in CD 13 is over 50% since there is a 7.5% HVAP population and a large Arab population which would have high non-citizen rates.

MI muon2 2015B

SKEW 1 (R) (5D, 2d, 2e, 4r, 1R) [3R in muon2 A2]
POLARIZATION 18 [14 in muon2 A2]
INEQUALITY 10 (range), 11 (ave dev) (range 5425, ave dev 1599) [11/13 in muon2 A2]
CHOP 8 raw (UC:9, UP:10, US:11) [9/10/12/13 in muon2 A2]
EROSITY 116 [119 in muon2 A2]

If the lowered BVAP is permissible, this beats muon2 A2 in all categories except polarization which is not used as a primary score.

[muon2]

Though I can't see traininthedistance's maps, I did build a plan that used some of his ideas on to top of one of my plans. I'll use that as my second submission. I'm quite curious how the currecnt audience sees these plans three years later.

After all that technical analysis here's a little amuse-bouche. I took my Detroit UCC splintering muon2 A and gave it a reworking. I decided to see how far I could push down the combined inequality and chop scores and bite the bullet on erosity. I also wanted to keep both 50%+ BVAP districts. I used train's pack of GR to eliminate any penalty there but I left in place the excess cover and one excess pack in Detroit at a cost of 3 points there. Here's what came out.

MI muon2 2015C



SKEW 2 (R) (3D, 2d, 5e, 3r, 1R)
POLARIZATION 13
INEQUALITY 7 (ave dev 788, range 4217)
CHOP 12 (9 raw, 2 cover, 1 pack)
EROSITY 135

There are a number of interesting features here from a public policy perspective.
1. CDs 2 and 3 could have been maintained as whole counties with a pack penalty traded for the chop count. A whole county version would have reduced erosity, but average deviation was lower the way it's shown here.
2. The chop in Washtenaw is just enough to keep Milan in one CD. In the OH competition that would have been rewarded with no chop counted in that county.
3. The number of highly competitive districts (PVI=0 or 1) is up to 5 and the polarization is down to 13. In AZ increasing the number of competitive districts is a specific goal, though here it is only a consideration after the main scoring.
4. The shape of CD 14 is particularly erose, but linking the Grosse Pointes to Grosse Ile is a riverfront district and could be construed as a community of interest. There were districts in CA that seemed to use this type of logic.

[muon2]

Oconto may be in the Census MSA with Green Bay, but its urban population is small and it doesn't qualify as part of a UCC. There are three UCCs in WI: Milwaukee (Milwaukee, Waukesha, Washington, Ozaukee); Appleton (Outagamie, Calumet); Eau Claire (Eau Claire, Chippewa). The Milwaukee UCC can have 2 CDs packed within and a bit left over for a third. Munis matter in WI, so add a detail map for the Milwaukee area with city and town lines turned on to see if any munis are being chopped.

Also, here's a map of inter-county connections if you are trying to minimize erosity.

[muon2]

Here's my plan A. Back in 2012 I posted this map as a way to make compact districts out of whole counties (except for Milwaukee). The only chop is for Milwaukee, and there is a chop of the city of Milwaukee cutting off the sw peninsula. We didn't have UCCs in 2012, but this plan does comply with the UCC rules for Milwaukee since 2 CDs are nested entirely within the 4 county region and only one other CD covers the remaining county. However, the plan does chop the Appleton UCC into two parts following the county line.



CD 1: (-2794) R+1.0
CD 2: (+38) D+17
CD 3: (-533) R+1.9
CD 4: (+1374) D+22
CD 5: (+901) R+10
CD 6: (+1829) R+11
CD 7: (-671) R+5.8
CD 8: (-142) R+6.8

[muon2]

Here's my plan B. It maintains all the UCCs within the minimum number of districts. The only chop is in Milwaukee and se part of the state is identical to plan A. The erosity is a bit better that A, but ithas somewhat more inequality.



CD 1: (-2794) R+1.0
CD 2: (+2414) D+17
CD 3: (-1788) R+3.8
CD 4: (+1374) D+22
CD 5: (+901) R+10
CD 6: (-2317) R+7.1
CD 7: (+1252) R+5.1
CD 8: (+960) R+10

[muon2]

Curious Question Muon2, is it possible to have only a single County split, while also having a District entirely within Milwaukee County?

Probably not, since his one chop is butt ugly,  so if he could arrange the other counties is such a way that that the chop into Milwaukee county was far smaller, but that population shift would not cause a chop elsewhere (the odds of that sized number of folks being moved around, to accomplish a specific goal and not causing a chop elsewhere are low in any event), he would have done so.

It is possible with a chop of the city of Milwaukee creating a bridge between Waukesha and Ozaukee (now that bridge chops are legal in the rules with possible erosity penalties). I thought that was worse than chopping the part of the city that sticks out to the sw. It's also possible if one gives up the pack or cover rules for the Milwaukee UCC, or if point contiguity is permitted.

If I wanted really ugly, I could make the bridge chop include all of the black neighborhoods of Milwaukee. Even though the CD would have all of Waukesha and Ozaukee the PVI would only be R+0.9. The other CD would be entirely within Milwaukee county and have a PVI of D+10.

[muon2]

Please turn on city/town lines when making a zoom map of the Twin Cities. As someone who grew up there, I can tell you that they matter locally. They also matter if I try to score the submissions.


MN has 3 UCCs to track. The Minneapolis UCC (Hennepin, Ramsey, Dakota, Anoka, Scott, Wright, Carver, Sherbune) has about 4.6 CDs of population. That means the ideal plan from a UCC perspective only covers those counties with 5 CDs while putting 4 completely within that part of the metro. The other two UCCs are much smaller: St Cloud (Stearns, Benton) and Mankato (Blue Earth, Nicollet), and ideal those pairs should not be split.

[muon2]

Here's my plan A submission, originally posted in 2013. Counties matter in the rural parts of the state and munis matter in the Twin Cities metro. This plan chops no counties except in the metro and chops no munis within a metro county. It gives up a UCC pack point to put St Cloud in the same CD as Sherburne county with includes part of the St Cloud city limits though it is in a different UCC due to suburbs on the east side of the county.




CD 1: (-1015) R+5.0
CD 2: (+513) R+3.3
CD 3: (+126) D+0.1
CD 4: (-1169) D+12
CD 5: (+632) D+25
CD 6: (+944) R+7.6
CD 7: (+191) R+13
CD 8: (-225) R+3.5

[muon2]

The goal in my plan B, also from 2013, was to separate St Cloud from the Twin Cities. Otherwise this plan also chops no counties outside the TC metro and no munis within a metro county. As discussed with other states this plan includes a CD that exceed 0.5%, but the overall range is less than 1%. Keeping Ramsey intact creates more skew for this plant than my plan A.




CD 1: (-2899) R+5.6
CD 2: (+3555) R+2.5
CD 3: (-432) R+1.6
CD 4: (-62) D+14
CD 5: (+632) D+25
CD 6: (-16) R+8.6
CD 7: (-2689) R+13
CD 8: (+1908) R+2.7

[muon2]

Please turn on city/town lines when making a zoom map of the Twin Cities. As someone who grew up there, I can tell you that they matter locally. They also matter if I try to score the submissions.


MN has 3 UCCs to track. The Minneapolis UCC (Hennepin, Ramsey, Dakota, Anoka, Scott, Wright, Carver, Sherbune) has about 4.6 CDs of population. That means the ideal plan from a UCC perspective only covers those counties with 5 CDs while putting 4 completely within that part of the metro. The other two UCCs are much smaller: St Cloud (Stearns, Benton) and Mankato (Blue Earth, Nicollet), and ideal those pairs should not be split.

I think it is very justified to have these 2 Counties split if you want to use the river as a boundary line between 2 Congressional Districts.

I'll let you argue that point with BRTD. I think he would confirm how closely linked North Mankato is to Mankato. In any case splitting them counts the same as any other chop.

[muon2]

Muon and Australia swing voter, could you make a 7 seats MN map please? As it seems a likely scenario.

We often engage in projected map-making when new census estimates come out. One of those exercises involving MN with 7 CDs for 2020 was in 2015. My plan is here. One can't just use DRA since it has 2010 data, and many of the counties and cities are projected to change relative to each other. In 2009 DRA came out with 2008 estimates and perhaps they'll do that again next year with 2018 estimates. I used a spreadsheet with the projections and then used DRA just as a drawing tool.

[muon2]

Muh CounTIes

If you lived in one of those MN counties, particularly the smaller ones, I think you would have a different attitude. On Solid's list, I've lived in two, plus one not on the list. They mattered to the residents, sometimes more than the municipality.

Do counties matter so little in FL? Are there no shared services at the county level?

[muon2]

Here's my plan A submission, originally posted in 2013. Counties matter in the rural parts of the state and munis matter in the Twin Cities metro. This plan chops no counties except in the metro and chops no munis within a metro county. It gives up a UCC pack point to put St Cloud in the same CD as Sherburne county with includes part of the St Cloud city limits though it is in a different UCC due to suburbs on the east side of the county.




CD 1: (-1015) R+5.0
CD 2: (+513) R+3.3
CD 3: (+126) D+0.1
CD 4: (-1169) D+12
CD 5: (+632) D+25
CD 6: (+944) R+7.6
CD 7: (+191) R+13
CD 8: (-225) R+3.5

I feel like putting Eden Prairie and St. Cloud in one district would cause rioting!

That was why I came up with a plan B back then.