Welcome, Guest. Please login or register.
Did you miss your activation email?
December 17, 2017, 02:38:18 pm
HomePredMockPollEVCalcAFEWIKIHelpLogin Register
News: Be sure to enable your "Ultimate Profile" for even more goodies on your profile page!

+  Atlas Forum
|-+  General Politics
| |-+  Political Geography & Demographics (Moderator: muon2)
| | |-+  PA Congressional Districts - 2020 Give and Take
« previous next »
Pages: [1] Print
Author Topic: PA Congressional Districts - 2020 Give and Take  (Read 528 times)
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« on: October 30, 2017, 08:37:45 am »
Ignore

Once whole-county districts with roughly balanced populations are established, they could be refined using a process called Give and Take.



In each county, a representative body of voters would be selected in a manner similar to juries, with one voter perhaps one per 10,000 persons, with a minimum of nine. Selection would be geographically stratified to be from throughout each county.

Give and Take would operate in rounds, with each district having one turn per round. A round is divided into two phases, (1) Take and (2) Give. Districts that have less than the ideal population would have a turn during the Take Phase, and take population from more populous neighbors during their turn. Districts that have a population greater than the ideal population would have a turn during the Give Phase, and give population to its five less populous neighbors during their turn.

During the Take Phase, the district with the least population (PA-7, Delaware County) would have the first turn.

PA-7    0.750
PA-8    0.831
PA-15   0.896
PA-16   0.917
PA-14   1.957
PA-18   0.960
PA-6    1.250
PA-1    2.110
PA-13   1.106
PA-11   1.051
PA-5    1.050
PA-17   1.041
PA-12   1.030
PA-4    1.028
PA-3    1.025


The redistricting jurors would choose to take population from a neighboring district with greater population (PA-1, Philadelphia); (PA-6, Chester); or (PA-13, Montgomery). The amount to be taken would be equal to:  (quota - population) / round_divisor.  Rather than making up the deficit in one fell swoop, it would be done gradually. The round_divisor for the first round would be 10, with subsequent rounds 9, 8, 7, ....  As the districts approach equality, we can take amounts that don't unbalance other districts unduly.

For the first round, the jurors would take 18,867 persons from a neighboring district and county. No more than one county may be divided between districts. If there are multiple counties along the boundary between districts, then the county would also be selected. Another county could not be selected, until all of the county is in one of the districts or the other.

The rules might provide that a district could take from one neighboring district; or from multiple districts. Jurors near Philadelphia might find they have a community of interest with the city, while those further out who might prefer to expand into Chester, and those along the Montco line might go in that direction. The persons moved would be based on the pro rata choices of the jurors. If there were one juror per 10,000 persons, Delaware would have 57 jurors, who would be making a collective decision.

After each turn, the district populations would be updated, and the districts reordered by population, which may change the turn order. During the Take Phase, the next turn will be the least populous of the districts that: (a) have a deficit; and (b) have not yet had a turn during the round. When all districts with a deficit have had a turn, the Take Phase is over.

Within the Take Phase, there would no reversals. That is, if District A takes from District B, then District B in its turn could not take back from District A. In addition, there may be cases where a taking district (i.e. one with a deficit) has no larger neighbors. In that case, they would pass.

After all districts with a deficit have had a turn, the Take Phase is over, and the Give Phase begins. The district with the largest surplus, in this case PA-6 (Chester, Berks) would have the first turn. They could give to any of their five neighbors with less population.

They must respect any county divisions that have already begun. If PA-16 had taken from Berks (rather than Chester), then PA-6 could give additional population from Berks, but not Chester. This is not a reversal, since it would be a continuation of the flow.

The amount to be given would be equal to:  (population - quota) / round_divisor.  Rather than eliminating the surplus in one fell swoop, it would be done gradually. The round_divisor for the first round would be 10, with subsequent rounds 9, 8, 7, ....  As the districts approach equality, we can give  amounts that don't unbalance other districts unduly.

As in the Take Phase, districts might be permitted to divide their giving among adjacent districts.

After each turn, the district populations would be updated, and the districts reordered by population, which may change the turn order. During the Give Phase, the next turn will be the most populous of the districts that: (a) have a surplus; and (b) have not yet had a turn during the round. When all districts with a surplus have had a turn, the Take Phase and the round is over. The next round will begin with another Take Phase.

As within the Give Phase, there may be no reversals. If District C gave to District D, District D on its turn may not give back to District C. Reversals between rounds are permitted.

The process would continue through multiple rounds until districts approached equality. Since the amount taken or given during a turn is proportional to the deviation (deficit or surplus), those with a larger deviation will change quicker converging the deviation range towards zero.
« Last Edit: November 20, 2017, 08:17:07 pm by muon2 »Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #1 on: November 15, 2017, 11:46:15 am »
Ignore

A common complaint about gerrymandering is that legislators choose their voters, rather than the other way around. But even if there is a redistricting commission, even one that is "independent", "citizen", and/or "bipartisan", there is no guarantee that it will act on the voters behalf. If there were evidence that such an appointed body can do such a superb job creating districts, and in effect choosing legislators, why not dispense with the legislature, and let the commission govern. Or if it is a computer program, then let the computer rule.

But if we believe that people should choose their representatives, why not let them choose their congressional districts.

We begin with the current map of Pennsylvania, organized in 106 county-CD parts: the portion of each county that is within a congressional district. There is no assumption that a CD will reform itself, but it might because the parts are adjacent, and they may in fact form a community of interest.



Each of the 106 parts form a proto-district. They will gradually coalesce from 106 to 88 to 74 to 62 to 51 to 42 to 35 to 29 to 24 to 20 to finally 17 districts of roughly equal population.

During each round, the least populous proto-district will vote to associate with an adjoining district. In effect, the voters of Cameron will be told that 4400 persons is not enough for a congressional district, and be asked which county they wish to join.

2020 county population projections are based on the 2016 estimates, projected forward linearly another 3.75 years. The county population is distributed among the county parts based on the 2010 census population. This assumes uniform change throughout a county. While this will produce small local errors, the overall placement of districts throughout the state should be correct.

When a proto-district that is smaller than a county is eliminated, it must join an adjacent district within the county. At that point they are fused, and will act as a single unit.

When a proto-district that is comprised of two or more counties is eliminated, the individual counties may independently choose their new district.

In reducing from 106 parts to 88 parts the following proto-districts will be eliminated:

Tioga-5 (portion of Tioga in PA-5).
Cameron
Crawford-5
Sullivan
Forest
Westmoreland-9
Huntingdon-9
Clarion-3
Fulton
Green-9
Lawrence-12
Potter
Carbon-12
Westmoreland-14
Montour
Somerset-9
Perry-11
Juniata

The largest part to be eliminated is Juniata, with a population of 25K (0.033 of a CD).
Logged
PNM
Mizzouian
YaBB God
*****
Posts: 2552
United States


View Profile
« Reply #2 on: November 15, 2017, 11:52:59 am »
Ignore

I would guess the PA Supreme Court would redraw the districts close to the county projections like you indicated. Chester-Berks is a good seat for a swing district. Northampton-Lehigh would be good for Dent's seat. Luzerne-Lackawanna-Monroe for Cartwright's seat. Delco for Meehan (no way he'd win now). Bucks-slice of Philly.
Logged

If Doug Jones can win Alabama, Phil Bredesen can win Tennessee and Beto O'Rourke can win Texas.

jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #3 on: November 15, 2017, 02:27:29 pm »
Ignore



The districts to be eliminated are in white. The 11 county-parts will be unified with a neighboring part within their county. In 9 instances, there are only two parts in the county, so this will unify the county. The two Westmoreland parts to be eliminated (tiny dots on the western edge of the county, where PA-9 and PA-14 enter the county will be merged with the larger parts in PA-12 and PA-18.



At this point the smallest counties could choose a neighbor they wished to join.

They might make the following choices. If anyone wants to represent the voice of Pennsylvanians they may.

Cameron joins Elk (they don't want to be dominated by a much larger neighbor)
Cameron could seek to join any county adjoining Potter-Cameron. If for some reason they wanted to join Tioga, they could try to convince Potter to also join Tioga.

Sullivan joins Wyoming (size wins out over connection to Bradford)

Forest joins Elk (small counties join together)

Fulton joins Bedford (size and connectivity)

Potter joins McKean (prefers a northwest orientation)

Montour joins Northumberland-10 (want to join Northumberland, but chooses best connected part)

Juniata joins Miflin (keeps it west of the mountains)
« Last Edit: November 15, 2017, 09:18:46 pm by jimrtex »Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #4 on: November 16, 2017, 01:05:41 pm »
Ignore

This is after the first round has reduced 106 proto-districts to 88.
-Barr


The second round will eliminate 14 proto-districts, here shown in white.



Cambria-9 (will be fused with rest of county)

Lancaster-7 (will be fused with rest of county)

Wyoming (+Sullivan) both join Bradford in effort to create district in northeastern part of state not dominated by Wilkes-Barre Scranton.

Greene joins Fayette, as less Pittsburgh dominated.

Clarion joins Armstrong.

Warren joins McKean-Potter as consolidation of northwestern district continues.

Clinton joins McKean-Potter as consolidation of northwestern district continues.

Susquehanna joins Bradford in northeast.

Snyder joins Union, resisting consolidation in central part of state.

Elk(+Forest-Cameron) all join northwest consolidation.

Tioga joins Bradford

Jefferson joins Indiana

Huntingdon joins Mifflin-Juniata in center of state.

Perry joins Mifflin-Juniata in center of state.
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #5 on: November 16, 2017, 03:43:51 pm »
Ignore

After second round you can see coalescence beginning along the northern border.



Round 3 reduces the number of districts from 77 to 62. The eliminated districts are shown in white.unty



Northumberland-10(+Montour) Fused with the rest of Northumberland and Montour joins in.

Washington-9 (county is whole)

Dauphin-4 (merges with Dauphin-11)

Monroe-10 (county is whole)

Lackawanna-10 (county is whole)

Berks-7 and Berks-15 merge with Berks-6.

Wayne (joins northeastern district)

Venango (joins northwestern district)

Pike and Carbon (both join merged Monroe)

Bedford+Fulton (split with Bedford joining Somerset, and Fulton joining Franklin)
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #6 on: November 17, 2017, 12:40:34 pm »
Ignore

Counties outside the large city continue to aggregate, and four county splits are terminated. A recount shows that Fulton chose to join Bedford in combining with Somerset.



Round 4 reduces the number of districts from 62 to 51.



Lebanon-6 dissolved (county made whole)

Montgomery-2 and Montgomery-8 dissolved, reducing Montco to 3 districts.

Columbia and Union(+Snyder) join Northumberland(+Montour) to created mid-Susquehanna district.

Dauphin-15 dissolved, county, made whole.

Clearfield joins northwestern district.

Crawford joins Erie-5 on narrow vote.

Lawrence joins Mercer.

Luzerne-17 dissolved, county made whole.

Adams joins York.
« Last Edit: November 17, 2017, 02:16:24 pm by jimrtex »Logged
Brittain33
brittain33
Moderators
YaBB God
*****
Posts: 14932


View Profile
« Reply #7 on: November 17, 2017, 02:12:54 pm »
Ignore

What was the rationale for combining 3 of the chunks of Berks county but leaving the wriggly little worm separate for now?
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #8 on: November 17, 2017, 11:04:00 pm »
Ignore

What was the rationale for combining 3 of the chunks of Berks county but leaving the wriggly little worm separate for now?
Districts were eliminated based on population.

Berks-15 had a population equivalent to 0.080 districts, and was a connector across the northern part of the county between Allentown and Lebanon and Dauphin.

Berks-7 had a population equivalent to 0.077 districts, and was the last gasp of a district that snaked up from Delaware (County).

Berks-6 had a population equivalent to 0.233 districts and snaked up from Montgomery.

Berks-16 is the little worm, but includes the city of Reading, and has a population equivalent to 0.163 districts. The reason for the worm is to put all the Republicans in the Philadephia suburban district, while extracting the Democrats and adding them to a Lancaster based district.

So the two smallest pieces decided to join Berks-6 when they were eliminated. That was the only choice for Berks-15. Berks-7 could have joined Berks-16, since they were adjacent, but I decided that the voters would prefer to join up with the part of the county outside Reading.

Berks-7 will be eliminated in the next round.

In most of the other split counties, there were only two parts, and often one was significantly smaller. When it was eliminated, it unified the county.

In a couple of cases, a moderately large county (Norhampton and Cumberland) was divided in almost equal parts such that neither has been eliminated so far.
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #9 on: November 18, 2017, 12:13:53 pm »
Ignore

Number of split counties is reduced to 10.



Round 5 reduces the number of districts from 51 to 42



Armstrong(+Clarion) dissolve. Clarion decides to join the northwestern district, while Armstrong joins Westmoreland.

Chester-16 and Chester-7 merge.

Westmoreland-12 dissolves unifying county.

Lycoming joins northeastern district, with Williamsport becoming the largest city.

Berks-16 dissolves, completing reunification of county.

Blair joins counties to south as Altoona and Johnston join together to preserve a district based in area.

Cumberland-11 and Cumberland-4 unify county.

Indiana(+Jefferson) dissolve. Jefferson joins the northwestern district while Indiana joins Westmoreland.

Cambria joins Blair in joining district to south.
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #10 on: November 18, 2017, 08:22:08 pm »
Ignore

I've added populations for the districts, relative to the ideal population for 17 districts.

Of the 42 districts, 15 are single county, 17 county parts (7 counties), and 10 multi-county districts.



This is a detail of the Philadelphia Metro area.



The next round reduces 42 districts to 35 districts.



Montgomery-7 dissolves, and joins Montgomery-13.

Schuylkill and Lebanon dissolve and join Berks.

Northampton-7 and Northampton-15 join to unite the county.

Franklin is dissolved and joins (York+Adams)

Erie-3 is dissolved to unite the county (along with Crawford)

The district north of Harrisburg dissolves and the counties scatter.
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #11 on: November 19, 2017, 02:41:06 pm »
Ignore

35 districts, including 11 single county districts, 12 multi-county districts, and 12 county parts (5 counties)



The next round reduces 35 to 29 districts.



Beaver joins Washington.

Centre joins the northwestern district.

Fayette(+Greene). Fayette joins Westmoreland, and Greene joins Washington.

Delaware-1 unifies the county.

Butler joins Westmoreland.

Mercer(+Lawrence) join Erie.

Conceivably, Beaver, Butler, Mercer, and Lawrence might prefer to combine, but this is not permitted under my rules.

My original concept was for symmetric accretion. A state would be divided into a number of proto-districts equal to a power of two times the final number of districts. For example, Pennsylvania might be divided into 1024 x 17 = 17,408 districts of roughly 737 persons. The initial proto-districts might be formed automatically by apportionment to counties, then cities, towns, and wards, and automated districting such as by splitline. In each round, pairs of districts would join together reducing the number of districts by half.

The problem with this approach is that symmetric accretion requires agreement between the joining districts. If A chooses B, and B chooses A, they can join. But what if C wanted to also choose A? It might have to make a second choice, in a form of two-dimensional musical chairs. But if all of C's neighbors are taken, what then. Perhaps it can choose a more distant neighbor (i.e. non-adjacent) and then we can later patch things up.

An alternative approach could use census blocks as the starting point and apportion units to each block. There might initially be 1,048,576 x 17 units (1,048,576 = 220), or one unit for every 0.719 persons. In each round, districts with an odd magnitude would attempt to unite with a similar neighbor, and then a new apportionment with a doubled quota would be done. So a district with magnitude of 7 would join with a neighbor with magnitude of 3, and then be apportioned (7+3)/2 = 5 units. A district with an even magnitude, say 8, could remain unchanged and have its apportionment halved.

Districts with a magnitude of one that did not find a partner would be eliminated and automatically joined to an adjacent district.

But this method has several problems. As more blocks are combined the magnitude of all districts   will converge towards one, which puts us back to the initial problem of finding pairwise combinations. In addition, pairwise combinations are not always the best. It is similar to the problem with nesting of legislative districts. You might be able to create very nice house districts, but some of them may have to combined in ugly senate districts.

Asymmetric accretion eliminates the problem of partners agreeing to merge, since the smaller districts gets to decide unilaterally who to join. In addition, when multi-county districts are dissolved, the individual counties can make independent decisions as to whom they join.

The reason I don't permit districts that are being dissolved to join together is based on an orderly sequence of elimination. In this instance, 35 districts were being reduced to 29 districts. If for example, Butler and Beaver joined together, then there would be 30 districts rather than 29. We could alternatively, then choose another district to eliminate. But in this case, the next target would be Washington, which might trigger other changes.

When county parts of similar size were merged (such as in Northampton or Cumberland), I did eliminate another district since I was keeping the fused district.

One solution might be to permit merging, but fix up the number of districts on later rounds. In this instance, if Butler and Beaver joined together, we would have 30 districts rather than the target of 29. The target for the next round is 24, so we could simply reduce from 30 to 24 on the next round.

Note: The targets were established based on the initial number of districts (106) and the final number (17). We want the reduction to be fairly gradual, and I have used 1/6 as a target. We can determine the number of rounds, r by solving:

1.2r = 106/17

And rounding to a whole number.

r = nint (ln(106/17) / ln(1.2) )

r = nint(10.04) or 10

The actual rate of decline, x, is then found by solving

xr = 106/17

x = eln(106/17) / r

x = 1.201

Applying this repeatedly as a divisor and rounding gives the sequence:

106, 88, 74, 61, 51, 42, 35, 29, 25*, 20, 17.

To make the number of eliminated districts monotonic, the 25 was adjusted to 24.

So if we permitted merging, we could simply retain the initial schedule, and reduce from 30 to 24, or recompute a new series from 30 to 17: 30, 25, 21, 17.

If we permitted mergers, we would first let adjoining districts that were concurrently being eliminated to join together. In this case, with the three adjoining districts (1) Beaver, (2) Butler, and (3) Mercer+Lawrence joining together, there could be choices among all three joining together, or of any three pairs.

Alternatively, we could automatically combine the districts, with an option to dissent.
« Last Edit: November 19, 2017, 11:06:59 pm by jimrtex »Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #12 on: November 20, 2017, 06:56:30 am »
Ignore

29 districts including 8 single county, 11 multi-county districts, and 10 county parts (4 counties)



The next round reduces 29 to 24 districts.



Lackawanna merges into Luzerne.

Chester-7 unifies county.

Allegheny-12 (northern suburbs) joins Allegheny-18 (southern suburbs)

Montgomery-6 unifies county.

Dauphin merges with Cumberland(+Mifflin)
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #13 on: November 20, 2017, 08:12:06 pm »
Ignore

24 districts, including 7 single county districts (all in southeast Pennsylvania), 12 multi-county districts, and 5 county parts in two counties (Philadelphia and Allegheny)



The next round reduces 24 districts to 20 districts.



Monroe(+Carbon+Pike) Monroe and Carbon join Lehigh, while Pike joins Scranton-Wilkes-Barre

Northampton joins Lehigh.

Northumberland dissolves. Northumberland, Montour, Columbia join Scranton-Wilkes-Barre, Union, Snyder, and Mifflin join northwestern district, and Juniata joins Harrisburg district.

Northeastern district dissolves. Four western counties join northwestern district, three eastern join Scranton-Wilkes-Barre

« Last Edit: November 20, 2017, 08:17:31 pm by muon2 »Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #14 on: November 21, 2017, 02:17:49 pm »
Ignore

20 districts, including 5 single county, 10 multi county, and 5 county parts.



The next round reduces 20 districts to the final target of 17.



Philadelphia-13 reduces to 2

Washington(+Beaver+Greene) dissolves, joins Pittsburgh suburbs.

Southwest(including Cambria and Blair) dissolves. Four western counties join Westmoreland district, while Fulton joins York and Huntingdon goes north.
« Last Edit: November 21, 2017, 02:37:14 pm by jimrtex »Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #15 on: November 22, 2017, 10:07:37 am »
Ignore

This represent the 17 districts, which would then be modified by a process of give and take to balance population.



The districts are not well placed for simple balancing. The five western districts have a population equivalent to 5.746 districts, or have the population for about 3/4 of an extra district. The seven southeastern districts (Philadelphia metro, plus Lancaster) have a population equivalent to 6.223 districts, or are short if population by about 3/4 of a district. Essentially, one of the southeastern districts would have to move to the west.

A structural problem with county-based districts in Pennsylvania is the cluster of counties in southeast Pennsylvania that are smaller than a district, but larger than half a district: Bucks, Delaware, Chester, and Lancaster. Philadelphia and Montgomery have very small surpluses.

An alternative approach would be to continue to reduce the number of districts (areas) to a smaller number, say 3 or 4. At that point, districts could be apportioned to each area, and whole county adjustments made to equalize the populations.

The process would then be repeated within each area to produce final areas. The use of existing district-parts is probably not a desirable starting point if we wish to avoid assumptions of what the final districts look like. We certainly would not want to assume the districts should look like the current districts. I used, the current county-parts because I had just discovered that the shapefiles were available, and I wanted a simple way to divide Philadelphia and Allegheny.

We could divide larger counties into sub-units. A threshold of 0.300 might be used (chunks of that size can be aggregated if they are relatively few in number). A target size of 0.150 could determine the number of chunks that a county were consolidated into. The consolidation would be done independently for each county.

When creating districts, chunks within a county would probably consolidate, but it would be not be required. The largest counties would consolidate in a couple of parts. Refinement of districts could occur at the chunk level.
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #16 on: November 22, 2017, 12:04:10 pm »
Ignore

This shows the 62 county subdivisions (townships and boroughs) in Montco, with a graduated.



This highlights the ten that would eliminated in the first round (9 actually)

Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #17 on: November 22, 2017, 04:40:00 pm »
Ignore

Ten districts to be eliminated in 2nd Round (53 to 43)

Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #18 on: November 22, 2017, 06:55:07 pm »
Ignore

A political map after the 2nd Round, with 43 districts.



The 3rd Round reduces the number of districts from 43 to 36. The 7 to eliminated are highlighted in white. This completes elimination of the smaller boroughs, with some consolidation of townships in the upper county.

Logged
muon2
Moderator
YaBB God
*****
Posts: 13179


View Profile
« Reply #19 on: November 23, 2017, 12:16:40 am »

Before you jumped to Montco, it seemed like there might have been more that could be done out west. Shouldn't the next step be to shift excess population between districts in the west to better equalize their population. For example shift Venango, Clarion, Forest, Warren, Elk, and McKean to the Erie district. Move Greene to the Westmorland district. Then move a bunch of counties from the Westmoreland district to the York district. Finally York and Adams would shift to more eastern districts.
Logged



Great American Eclipse seconds before totality showing Baily's Beads.
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #20 on: November 23, 2017, 01:44:23 pm »
Ignore

Before you jumped to Montco, it seemed like there might have been more that could be done out west. Shouldn't the next step be to shift excess population between districts in the west to better equalize their population. For example shift Venango, Clarion, Forest, Warren, Elk, and McKean to the Erie district. Move Greene to the Westmorland district. Then move a bunch of counties from the Westmoreland district to the York district. Finally York and Adams would shift to more eastern districts.
You might recall in the original thread (Congressional Districts - 2016 ACS estimates) I had began by assigning each county to the district with the largest share of its population. This resulted in excess districts, because counties were split apart. So I placed the isolated counties in adjacent districts, and then eliminated two districts to get to 17 districts (15 actually, with 2 double districts because of Allegheny and Philadelphia). I then did further adjustments to get closer.

That was my starting point for the give and take process. While one could think of the give and take process as one applied by an outside agency (legislature, commission, or a computer), I think of it as a process by which the voters define their districts.

My thesis is that voters can self-organize into districts that represent their collective self-interest. But there is no reason that this process should be limited to the refinement stage, but could begin with no predefined districts.

Using the county-parts from the existing districts was an expedient. I knew they would fuse except in Philadelphia and Allegheny, and I had discovered that the map function in American Fact Finder could generate shapefiles. Starting from primitives such as townships and boroughs and wards or neighborhoods in Philadelphia would have been too tedious. The current exploration of Montco is to see what a preliminary local clustering would look like.

Your suggestions are what a top-down procedure might do. But under give and take, the voters of the Erie district might prefer to add Beaver. I may be stuck in the 15th Century, but I for some reason think the York/Lancaster border is inviolate. This might also be a perception that one could end up badly divided, and the other whole.

There are several ways forward.

(1) The Westmoreland district has more than twice the population of Chester. So split the Westmoreland district (Butler, Armstrong, Westmoreland 0.799, and the eastern 6 counties 0.780. While Chester would probably join Delaware. This gets the districts more in the correct places, but might be considered a cheat. You might have imbalance, but not one where a district was twice as large or could be so evenly split.

(2) Do accelerated give and take using whole counties, where a county could be shifted if it improved equality. Lancaster could add Lebanon, or Berks could force its way out through Northumberland. (2) could be done after (1).

(3) Use chunks in the larger counties for step (2). Using a threshold of 0.300, would result in 16 counties being divided into chunk. 10 are in southeastern Pennsylvania, the band of five counties from Northampton to York, plus the 5 Philadelphia metro counties; Luzerne (Scranton); Dauphin and Cumberland (Harrisburg); Allegheny and Westmoreland (Pittsburgh); and Erie (Erie).

(4) Continue the consolidation process to four areas covering the state. Apportion representatives to the areas, and adjust the areas to balance population using give and take. Then restart the process within those areas.
Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #21 on: November 23, 2017, 10:34:43 pm »
Ignore

After the 3rd Round with 36 districts.



4th round reduces from 36 to 30 districts. However, I've decided to permit mergers, in this case Pennsburg borough with Upper Hanover township. This will result in 31 districts, which will be caught up with in the fourth round.

Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #22 on: November 23, 2017, 10:59:08 pm »
Ignore

After the 4th round with 31 districts.



5th round reduces from 31 to 25 districts.



After the 5th round with 25 districts.



6th round reduces from 25 to 21 districts.

« Last Edit: November 25, 2017, 02:37:21 pm by jimrtex »Logged
jimrtex
YaBB God
*****
Posts: 8443
Marshall Islands


View Profile
« Reply #23 on: November 25, 2017, 03:17:06 pm »
Ignore

After the 6th Round with 21 districts.



7th Round reduces from 21 to 17 districts.



After 7th Round with 17 districts.



8th Round reduces from 17 to 14 districts (Horsham and Upper Dublin merge, leaving 15 districts)



After 8th Round with 15 districts.



9th Round reduces from 15 to 12 districts.



After 9th Round with 12 districts.



10th Round reduces from 12 to 10 districts.



There are several possibilities how the district comprised of Upper Salford, Lower Salford, Salford, and Franconia might go. If they go south and split, it makes long skinny districts. If they don't split, then it produces a panhandle.

My current inclination is for the three Salford's to go north, and Franconia to go south. Lower Salford could also go west.
Logged
Bad Apple!!
omegascarlet
YaBB God
*****
Posts: 3216


View Profile
« Reply #24 on: November 28, 2017, 12:57:14 pm »
Ignore

This still happening?
Logged

I live in Arizona. The DC avatar just looks nice.
Pages: [1] Print 
« previous next »
Jump to:  


Login with username, password and session length

Logout

Powered by SMF 1.1.21 | SMF © 2015, Simple Machines