2020 Redistricting in Pennsylvania

[muon2]

Have you ever actually made a fair map in good faith?

What are you talking about?

Dude every single one of your fair maps are D gerrymanders, stop trying to deny it.

This particular court is overwhelmingly likely to draw a soft D gerrymander (in fact they already did for 2018-20), so it's appropriate for PA specifically.

This court may favor a soft D gerrymander, but they won't support more county chops then necessary. That was one of the clear directives to their special master. To draw 17 CDs one should only need 16 chops, though an approved plan might have 17 chops if it's easier to get population equality that way.

[muon2]

LOL ^ what a hack map.
That split of Berks/Pittsburgh and Alleghany is a meme.

Literally everyone but you agrees that there should be a SW westmoreland/Fayette/Greene/Washington district with 20k added.

Pittsburgh is split by the river except for about 40k of pop needed.

Berks is split so Chester and Delaware can stay whole.

How many chops are there in total? You didn't show county lines on the state map, and county chops are a criteria the court used with the special master.

For the record the number of chops in a county is the number of districts in that county above one. The term is an Atlas redistricting meme from the last redistricting cycle.

[muon2]

LOL ^ what a hack map.
That split of Berks/Pittsburgh and Alleghany is a meme.

Literally everyone but you agrees that there should be a SW westmoreland/Fayette/Greene/Washington district with 20k added.

Pittsburgh is split by the river except for about 40k of pop needed.

Berks is split so Chester and Delaware can stay whole.

How many chops are there in total? You didn't show county lines on the state map, and county chops are a criteria the court used with the special master.

Seven counties are part of more than one district.

Philly, MontCo, Berks, Monroe, Allegheny, Greene, and Beaver.

Sorry, I was late with my edit to define my terms. It's not the number of counties chopped that matters, it's the the total number of districts in excess of one in those counties that determines the number of chops.

[muon2]

LOL ^ what a hack map.
That split of Berks/Pittsburgh and Alleghany is a meme.

Literally everyone but you agrees that there should be a SW westmoreland/Fayette/Greene/Washington district with 20k added.

Pittsburgh is split by the river except for about 40k of pop needed.

Berks is split so Chester and Delaware can stay whole.

How many chops are there in total? You didn't show county lines on the state map, and county chops are a criteria the court used with the special master.

Seven counties are part of more than one district.

Philly, MontCo, Berks, Monroe, Allegheny, Greene, and Beaver.

Sorry, I was late with my edit to define my terms. It's not the number of counties chopped that matters, it's the the total number of districts in excess of one in those counties that determines the number of chops.

Is that the same criteria the court uses, or a personal preference?

The PA Supreme Court opinion specifically references the number of divisions within counties and municipalities, and they cite examples where some jurisdictions were divided over many districts. It wasn't just the number of counties that were split that mattered. The special master followed that guidance in drawing the replacement plan. His method was consistent with my definition of chops and seeking to minimize them.

[muon2]

LOL ^ what a hack map.
That split of Berks/Pittsburgh and Alleghany is a meme.

Literally everyone but you agrees that there should be a SW westmoreland/Fayette/Greene/Washington district with 20k added.

Pittsburgh is split by the river except for about 40k of pop needed.

Berks is split so Chester and Delaware can stay whole.

How many chops are there in total? You didn't show county lines on the state map, and county chops are a criteria the court used with the special master.

Seven counties are part of more than one district.

Philly, MontCo, Berks, Monroe, Allegheny, Greene, and Beaver.

Sorry, I was late with my edit to define my terms. It's not the number of counties chopped that matters, it's the the total number of districts in excess of one in those counties that determines the number of chops.

Is that the same criteria the court uses, or a personal preference?

The PA Supreme Court opinion specifically references the number of divisions within counties and municipalities, and they cite examples where some jurisdictions were divided over many districts. It wasn't just the number of counties that were split that mattered. The special master followed that guidance in drawing the replacement plan. His method was consistent with my definition of chops and seeking to minimize them.

The goal of my fair map was to keep metros and counties intact. A court drawn map may look different from mine, despite some commonalities with the current court drawn map. A D gerrymander or a court-speculative map would look different imo.

But that doesn't answer my question. I'm genuinely interested in the number of splits you used. If you have a version of your map with county lines, I can work it out myself. If you have a link to your map on DRA I can look it up there.

[muon2]

LOL ^ what a hack map.
That split of Berks/Pittsburgh and Alleghany is a meme.

Literally everyone but you agrees that there should be a SW westmoreland/Fayette/Greene/Washington district with 20k added.

Pittsburgh is split by the river except for about 40k of pop needed.

Berks is split so Chester and Delaware can stay whole.

How many chops are there in total? You didn't show county lines on the state map, and county chops are a criteria the court used with the special master.

Seven counties are part of more than one district.

Philly, MontCo, Berks, Monroe, Allegheny, Greene, and Beaver.

Sorry, I was late with my edit to define my terms. It's not the number of counties chopped that matters, it's the the total number of districts in excess of one in those counties that determines the number of chops.

Is that the same criteria the court uses, or a personal preference?

The PA Supreme Court opinion specifically references the number of divisions within counties and municipalities, and they cite examples where some jurisdictions were divided over many districts. It wasn't just the number of counties that were split that mattered. The special master followed that guidance in drawing the replacement plan. His method was consistent with my definition of chops and seeking to minimize them.

The goal of my fair map was to keep metros and counties intact. A court drawn map may look different from mine, despite some commonalities with the current court drawn map. A D gerrymander or a court-speculative map would look different imo.

But that doesn't answer my question. I'm genuinely interested in the number of splits you used. If you have a version of your map with county lines, I can work it out myself. If you have a link to your map on DRA I can look it up there.

I did answer your question, eleven county splits. Some of those I can decrease even, to fit the criteria, if necessary.

The purpose of my map was not to emulate the court, but to draw something that adequately represents both sides of the divided government. As such, minimal change outside of the loss of a seat.

What you answered is the number of counties split. What PA also requires is a list of which districts split those counties. That becomes relevant if it is more than 2 districts in a county.

For example Monroe splits between 2 districts, 7 and 8, so it is just 1 chop. Philadelphia has at least 3 districts, and maybe 4 on your map: 2, 3, 4, and 6. That equals 3 chops if those 4 districts are in the county.

[muon2]

Here's part of the direction from the PA Supreme Court to the special master drawing the map.

c.A report detailing the number of counties split by each district and split in the plan as a whole.
d.A report detailing the number of municipalities split by each district and the plan as a whole.

Both c. and d. require two parts: a list of splits associated with each individual district and a list of splits for the state as a whole. That's equivalent to asking (as some other states do) which geographic units are split and by which districts.

Given the nature of the litigation that led to the current plan, I would expect that any plan created for the next decade will have those lists. Any plan next year is likely to be reviewed by a court, and since the Supreme Court asked for those lists, so will a lower court. Any plan that has more than the minimum necessary in those lists is going to have to justify them, or expect that litigation will focus on the excess splits in that plan.

[muon2]

Made some modest changes to 13, 14, 15, 16 and 17.

Much better IMO. I count the minimum of 16 chops in 13 counties with Allegheny, Berks, and Philadelphia with 2 chops and the rest with 1 chop. That aspect looks good from the criteria. Were you able to keep municipal chops to a minimum, too? I know they aren't always obvious unless you want to read the title of each precinct. May I ask what the biggest deviations were plus and minus from the ideal?

[muon2]

Anyway working with Sev's principles of a fair map I worked hard and finished this fair map. The court will likely not enact this though.

sarcasm

It looks like you have a CD with just Dauphin and York, but that would have too large a population deviation from the quota. The colors make it hard to tell, so is there a chop in Cumberland or somewhere else to bring the population of the Harrisburg CD up towards quota?

It looks like you should be able to shuffle population between the Scranton, Wilkes-Barre, and State College CDs to eliminate one of the chops (probably the one in Lycoming).

[muon2]

Made some modest changes to 13, 14, 15, 16 and 17.

Much better IMO. I count the minimum of 16 chops in 13 counties with Allegheny, Berks, and Philadelphia with 2 chops and the rest with 1 chop. That aspect looks good from the criteria. Were you able to keep municipal chops to a minimum, too? I know they aren't always obvious unless you want to read the title of each precinct. May I ask what the biggest deviations were plus and minus from the ideal?

https://davesredistricting.org/join/364fb604-2395-4513-8981-25ff353debed

Also, fun fact: the Scranton is district on my D gerrymander is four whole counties with no cuts or additional precincts, and has exact population. That obviously won't be the case when the real numbers are available, but hey.

Thanks. I see deviations from -1899 to +1465 for a range of 3364 or 0.45%.

If you mean the Lackawanna/Luzerne/Monroe/Pike district, then yes I saw it. It's actually in my model plan. If a whole county CD comes with exactly the population quota, it would be hard for a court to reject it. Even some small deviation is permitted if you can make the case for strict adherence to other neutral standards and there was no better way to meet those standards. I'll be watching for such opportunities when the 2020 Census data is released.

[muon2]

Anyway working with Sev's principles of a fair map I worked hard and finished this fair map. The court will likely not enact this though.

sarcasm

It looks like you have a CD with just Dauphin and York, but that would have too large a population deviation from the quota. The colors make it hard to tell, so is there a chop in Cumberland or somewhere else to bring the population of the Harrisburg CD up towards quota?

It looks like you should be able to shuffle population between the Scranton, Wilkes-Barre, and State College CDs to eliminate one of the chops (probably the one in Lycoming).

There are a few towns in Cumberland I took.

Were you able to keep municipalities whole (except of course for Philly)?

I'll also ask the question I asked sev. What was your population range (difference between largest and smallest CDs)?

[muon2]

Anyway working with Sev's principles of a fair map I worked hard and finished this fair map. The court will likely not enact this though.

sarcasm

It looks like you have a CD with just Dauphin and York, but that would have too large a population deviation from the quota. The colors make it hard to tell, so is there a chop in Cumberland or somewhere else to bring the population of the Harrisburg CD up towards quota?

It looks like you should be able to shuffle population between the Scranton, Wilkes-Barre, and State College CDs to eliminate one of the chops (probably the one in Lycoming).

There are a few towns in Cumberland I took.

Were you able to keep municipalities whole (except of course for Philly)?

I'll also ask the question I asked sev. What was your population range (difference between largest and smallest CDs)?

1500 to -1200. Probably could reduce it a bit further. it a bit further but its good enough. Did try to keep cities whole as possible although DRA doesn't give all of them.

It sort of does give them all, but it requires work on the part of the user. The city lines on DRA isn't that helpful since it shows census designated places as well as incorporated cities. PA counts all municipalities (cities, boroughs, and townships) as items to avoid chopping, but it doesn't care about census designated places.

The best way to see what is a PA municipality with DRA is to look at the name of the voting district as you hover over it. If the municipality shows just the name, then it has only one voting district. If the name has the word DIST or VTD and a number or direction, then the municipality has multiple precincts, and to keep it intact, you have to find them all by name.

[muon2]

The problem is sometimes what is fair as opposed to what is neutral. People tend to think of fair in partisan terms, but that can run up against neutral mapping principles. I use the term skew to measure the lack of partisan fairness in a plan.

Skew in PA comes naturally from the city of Philly, especially with the necessary Black-majority CD. That VRA-mandated CD must be placed almost entirely in the city, and often is completely within the city. In fact, most neutral mapping efforts would put 2 CDs either entirely or mostly in the city for reasons of compactness or community of interest. Any plan that feathered more than 2 Philly CDs into the surrounding counties would smack of gerrymandering. And other than going into Bucks, those Philly CDs would have to feather out quite a ways to dilute the Dem pool in the city.

To show the impact look at the 2012/2016 PVI of the whole state: R+0.38, and without Philly R+4.95. That's a big shift. Moreover consider that the rest of the state would get 15 CDs. At R+5 the usual analysis (50%+ 2*PVI) would lead to the expectation that the Pubs would hold a PVI advantage in 60% of those seats, which is 9. So despite the effectively even statewide number Pubs should hold a 9-8 advantage in most scenarios.

[muon2]

The problem is sometimes what is fair as opposed to what is neutral. People tend to think of fair in partisan terms, but that can run up against neutral mapping principles. I use the term skew to measure the lack of partisan fairness in a plan.

Skew in PA comes naturally from the city of Philly, especially with the necessary Black-majority CD. That VRA-mandated CD must be placed almost entirely in the city, and often is completely within the city. In fact, most neutral mapping efforts would put 2 CDs either entirely or mostly in the city for reasons of compactness or community of interest. Any plan that feathered more than 2 Philly CDs into the surrounding counties would smack of gerrymandering. And other than going into Bucks, those Philly CDs would have to feather out quite a ways to dilute the Dem pool in the city.

To show the impact look at the 2012/2016 PVI of the whole state: R+0.38, and without Philly R+4.95. That's a big shift. Moreover consider that the rest of the state would get 15 CDs. At R+5 the usual analysis (50%+ 2*PVI) would lead to the expectation that the Pubs would hold a PVI advantage in 60% of those seats, which is 9. So despite the effectively even statewide number Pubs should hold a 9-8 advantage in most scenarios.

That's basically what my map is, although Lamb and Cartwright could conceivably win their Trump districts, and Fitzpatrick could potentially hold his Bucks Clinton district.

Although "fair" could mean other things, such as adequate minority representation, keeping metro areas whole, etc. Philly metro is essentially six districts +6000 so there's no reason to break it up in my mind.

The more compact Philly metro doesn't include Berks, but you are correct that adding that makes a nearly perfect fit for 6 CDs. With the 2010 Census and 18 CDs that same cluster with Berks was only 67 people over the quota for 6 CDs.

I also prefer to use no more than 4 CDs to cover the Pittsburgh metro, since it is just barely larger than 3 CDs. In addition there are 4 mid-sized 2-county metros I try to keep in 1 CD: Allentown (Lehigh and Northampton), Scranton (Lackawanna and Luzerne), and Harrisburg (Dauphin and Cumberland). I consider all of these to be communities of interest and separating them dilutes their voting strength just as chopping a mid-size county dilutes its voting strength.

[muon2]

Since everybody is in on it I guess I'll post my map. I tried putting Chesco with Delco since they're so similar demographically instead Lancaster or Berks (which just aren't similar to Chesco in any way). It actually works out really well and places together similar areas such as eastern Delco+West Philly (creates a second AA seat), NE Philly+Lower Bucks, Montco/Central Bucks, and Upper Bucks/Montco+Berks. There's an inevitable GOP tilt, but CDs 4/10/17 look like they'll be competitive

I have 2020 numbers but I'll save that for later

I like the shapes, but when two counties share the same two districts, it is always possible to shift population to reduce the number of chops. If the PA court looks at this coming cycle with the same methodology that was given to the special master, they'll call that type of sharing into question.

In your map both York and Cumberland have parts of 10 and 13. Both Bucks and Montgomery have parts of 1 and 4.

[muon2]

Since you have been kind enough to let me quiz you on your maps, it's only fair that you get to poke at mine. As with most of you this is based on DRA with the 2018 ACS data and is close to the 2016 estimates.

I started by setting as my primary goal the preservation of the metro communities of interest as quantified by the Urban County Clusters (UCCs). Counties were grouped around the UCCs to make regions within a half a percent of the population quota for a whole number of districts. The more regions, the fewer the number of county chops. This plan chops only 7 counties (Allegheny, Butler, Chester, Montgomery, Philadelphia, Schuylkill, and York) for a total of 11 chops. No municipalities (cities, boroughs, or townships) where chopped except Philadelphia which has the minimum number of 2 chops. A secondary consideration was to minimize the erosity when chopping counties.

The percentage deviation was set to comply with SCOTUS rulings on allowable population range - 0.79% in Karcher v Daggett (1983) and 0.78% in Tennant v Jefferson County (2012) - when strict neutral redistricting criteria are followed. This plan has a range of 0.69% while strictly preserving UCCs and municipalities and minimizing county chops. If challenged in court for population deviation the Tennant opinion suggests that an alternate plan would have to be presented that meets those strict criteria with a smaller population range to defeat this plan.

Staistically districts that are D+1, EVEN or R+1 (up to 1.49% including rounding) are equally likely to go for either party and I classify them as highly competitive. PVIs of 2 through 5 are competitive districts. PVI of 6 or higher are uncompetitive districts, though strong incumbents of the other party have been known to succeed in such districts. This plan has 8 uncompetitive R districts, 5 uncompetitive D districts, and 4 highly competitive districts (1, 5, 7, 8 ).

[muon2]

Since you have been kind enough to let me quiz you on your maps, it's only fair that you get to poke at mine.

This is a very good map. My only quibble is probably the 3-way split of York. I would instead put a bit of northwestern Lancaster County in the 10th district, as that area is more closely tied to Harrisburg than to the city of Lancaster.

My first thought is that district 9 kind of feels like a "leftovers" district. I did try some rotation around York and Lancaster to pair Cumberland with Dauphin but overall deemed it unnecessary. I also dislike the double cut into Montgomery, I'm sure it's for optimal population purposes but I blanket ban double cutting from my fair maps.

That's a tough call that I spent some time thinking about. Initially I had Lebanon with Lancaster which meant the Harrisburg CD picked up the city of York. The number of chops is the same either way. But it was an erose chop into York, and I could be more compact putting both chops in York county. Since compactness matters, too, I went in the direction of the double chop. As far as using part of Lancaster, the Fairview Park/Newberry corner of York is advertised as part of suburban Harrisburg these days, so that was another factor.

[muon2]

Since you have been kind enough to let me quiz you on your maps, it's only fair that you get to poke at mine.

This is a very good map. My only quibble is probably the 3-way split of York. I would instead put a bit of northwestern Lancaster County in the 10th district, as that area is more closely tied to Harrisburg than to the city of Lancaster.

My first thought is that district 9 kind of feels like a "leftovers" district. I did try some rotation around York and Lancaster to pair Cumberland with Dauphin but overall deemed it unnecessary. I also dislike the double cut into Montgomery, I'm sure it's for optimal population purposes but I blanket ban double cutting from my fair maps.

That's a tough call that I spent some time thinking about. Initially I had Lebanon with Lancaster which meant the Harrisburg CD picked up the city of York. The number of chops is the same either way. But it was an erose chop into York, and I could be more compact putting both chops in York county. Since compactness matters, too, I went in the direction of the double chop. As far as using part of Lancaster, the Fairview Park/Newberry corner of York is advertised as part of suburban Harrisburg these days, so that was another factor.

I guess my point is that keeping counties together doesn't mean very much if you're picking and choosing from opposite ends (MontCo).

I put one district entirely within MontCo. Nesting districts entirely within a county when possible is a big issue with some neutral mapping schemes, though I can be flexible on that. I wasn't going to chop municipal Philly more than necessary, since that creates a potential legal weakness. Given the cluster around Philly, I didn't see a better option that moved the second Montco chop to another county.

[muon2]

Since you have been kind enough to let me quiz you on your maps, it's only fair that you get to poke at mine.

This is a very good map. My only quibble is probably the 3-way split of York. I would instead put a bit of northwestern Lancaster County in the 10th district, as that area is more closely tied to Harrisburg than to the city of Lancaster.

My first thought is that district 9 kind of feels like a "leftovers" district. I did try some rotation around York and Lancaster to pair Cumberland with Dauphin but overall deemed it unnecessary. I also dislike the double cut into Montgomery, I'm sure it's for optimal population purposes but I blanket ban double cutting from my fair maps.

That's a tough call that I spent some time thinking about. Initially I had Lebanon with Lancaster which meant the Harrisburg CD picked up the city of York. The number of chops is the same either way. But it was an erose chop into York, and I could be more compact putting both chops in York county. Since compactness matters, too, I went in the direction of the double chop. As far as using part of Lancaster, the Fairview Park/Newberry corner of York is advertised as part of suburban Harrisburg these days, so that was another factor.

I guess my point is that keeping counties together doesn't mean very much if you're picking and choosing from opposite ends (MontCo).

I put one district entirely within MontCo. Nesting districts entirely within a county when possible is a big issue with some neutral mapping schemes, though I can be flexible on that. I wasn't going to chop municipal Philly more than necessary, since that creates a potential legal weakness. Given the cluster around Philly, I didn't see a better option that moved the second Montco chop to another county.

What I'm talking about is district 1 cutting into both Northern MontCo and southern MontCo. The county cut should be contiguous.

To a lesser extent, I also would try to make the district 5 MontCo cut to an area bordering both Berks and Chester.

I looked at both those issues. I couldn't easily find one that didn't split municipalities and kept the districts within the population quotas. If you can see one, I'd gladly adjust my plan.

[muon2]

FYI you can easily nest two districts in Dauphin, York, Cumberland, and Lancaster.

I saw that, but if Berks goes with the Philly metro, then there's the unfortunate connection of Lebanon to Schuylkill to points north. It looked even worse in terms of compactness than the CD 9 on my map, so I dropped the idea.

[muon2]

I’m not going to lean too much on the choice of one word as I say something the poster is already aware of, but it’s always good to remember that the “natural” skew of large cities and “self-packing” are the result of decades of federal, state, local, and business policies that heavily restricted where people of color could live to a small number of communities, so a neutral political process that perpetuates the impact of that ghettoization by wasting tens of thousands of votes in compact but lopsidedly uncompetitive districts, reducing political power for the minority group per individual, is not neutral at all.

I would suggest that the VRA is a factor as well. Without the VRA one might more easily entertain a gerrymander that feathers those inner city minority areas out into districts that include a lot of the suburbs and exurbs. But that will potentially dilute their votes to the point that they cannot get the candidate of their choice, just the party of their choice. A strong Dem party can in turn protect the minority candidates in the primary, but relying on a party to protect the interests of the minority isn't the point of the VRA.