Chops and Erosity - Mid Atlantic Madness

[muon2]

The precise lines in SE PA, were dictated by trying to minimize inter-county highway cuts, while avoiding subunit chops. Job one of course is to first identify the county seat, and work from there. I was unable to find a way for PA-08 to chop into Philly without a ward chop, so it chopped into Montco instead. And I needed to make sure PA-02 took in downtown Philly where the courthouse is, to avoid a highway cut from Montco by having the most direct highway able to go from Norristown to the Philly courthouse, without ever touching PA-01. This aspect of the game is the most time consuming. The way PA-16 juts into Chester County was no accident either. Finally, the cut into Montco by PA-06 was designed to avoid a traveling chop, which I think should be prohibited, and thus I won't due it.

I'd also opine that this "aspect of the game", such as it is, seems pretty artificial.  I haven't paid much attention to it with my own metro-area lines, and I am unlikely to see a particularly compelling reason to start doing so.

When there is no macrochop in a county there needs to be some modest constraints on the chops and erosity. That means assigning each prior link before the chop to one of the pieces created by the chop. Since the primary highway link exists as the proxy for an economic community of interest, it is least arbitrary of many choices to make that assignment. erosity is reduced when those links are kept inside a district. My observation is that if one is making reasonable choices (like the many Lapeer chops) they will tend to respect much of those primary links.

When there is a macrochop a different regime is present. Now all highway links across a county line count as proxies for the town-scale CoIs. It is still the case that erosity is reduced when those links are kept internal to a district. But in the macrochop case, there is potentially more than one link between two counties to consider.

[muon2]

So let me talk more about links using Torie's plan for Chester. First, it looks like Honey Brook twp along the Lancaster line is chopped. I'm not sure how Torie wants to fix that, but I'll treat it like it's all in CD 16.

Chester is macrochopped, so the county seats don't matter. Instead one looks at the townships on the perimeter that have state and federal highways.

Along the Lancaster border:
West Nottingham twp: PA 272
Lower Oxford twp: PA 472
Upper Oxford twp: PA 896
West Fallowfield twp: no connection
West Sadsbury twp: PA 41, PA 372, US 30 (three highways, but only counts as one link to the twp)
West Caln twp: PA 340
Honey Brook twp: US 322, PA 10 (These enter through two different precincts of Honey Brook. They count as one link if the twp is not chopped, but each would count separately if it is chopped like above. Note PA 10 )

Other than Honey Brook, Torie has kept all the links within CD-16, so they don't add to erosity. That chopped twp would add a point of erosity between CD-6 and 16.

That gives an example of how I build a map of all the links to determine erosity. It's time consuming for each macrochopped county. The good part is that if I save it, I don't have to do it again.

[muon2]

I'm not quite sure what is suggested then for UCCs.

It sounds like single county UCCs get no special treatment, and I lean that way, too.

It sounds like the new UCC pack rule is favored. I'm not sure it's as strong as some think, but it's worth continued consideration.

For the cover rule it sounds like Torie is suggesting looking at the cover of a UCC and if it exceeds the minimum and the number of county chops plus 1 in that UCC a penalty occurs. For example in my favorite guinea pig of Lansing, a cover of 2 would only get a penalty if the county chop count is 0, but not otherwise. I think this breaks down when one is able to exactly fit a number of districts into a subset of the counties in a UCC (like 4 CDs in Wayne+Macomb+St Clair), or if there is a double-spanning chop (perhaps for the VRA).

The simple cover rule still seems to me to be the strongest of the UCC rules, and I would not shy away from the double penalty it incurs (unless one wants to revisit microchops). Nonetheless, I'll keep working on the MI map set and try to put together a comprehensive study.

[muon2]

I can apply your suggestion just fine in 0 pack UCCs like Lansing. It's very hard to write a rule that applies in multi CD UCCs. You can look at a large county like Livingston and say that's bad. It becomes hard when there's a large UCC with a small excess and some small counties on its periphery. I would rather encourage a chop of a whole county rather than a chop into a larger county. If this doesn't make sense I'll find a state with an example (after I've finished scoring the MI maps on that thread).

Why not just have a double penalty for a chop into a county in a large UCC that also exceeds the cover minimum?

[muon2]

I can apply your suggestion just fine in 0 pack UCCs like Lansing. It's very hard to write a rule that applies in multi CD UCCs. You can look at a large county like Livingston and say that's bad. It becomes hard when there's a large UCC with a small excess and some small counties on its periphery. I would rather encourage a chop of a whole county rather thana chop into a larger county. If this ddoesn't make sense I'll find a state with an example.

Why not just have a double penalty for a chop into a county in a large UCC that also exceeds the cover minimum?

I understand your concern, but the additional potential erosity points would tend to discourage the chop (overwhelmingly so, if all intra-county cuts count, rather than just those cuts over the minimum as I propose). And you can sever the county without penalty, if both the cover and pack maximum/minimum are not violated (to deal with close cases, one might perhaps consider the pack/cover rule as not being violated if the amount of the otherwise pack CD outside the UCC, or the additional CD in the UCC, caused by the whole county severance, is less than a macrochop in size). Anything beyond a macrochop pad to me is just traducing the whole meaning of UCC's, and is a huge loophole.

You final point is about having two smaller chops into a UCC being penalized as opposed to one larger chop (the infamous Lansing example again). I don't think that should be penalized myself, beyond the chop penalty itself. That is protecting UCC's over other counties without good reason in my view.  

Actually my final point was to consider the distinction between small UCCs like Lansing and large UCCs greater than one CD. I can see the value of treating small UCCs like single county UCCs and your rule would be easy to implement for those. I think it is in this class that you have most observed the problems of overprotecting the UCC.

I think problems occur in larger UCCs where there will be a lot of combinations that make a rule for whole county chop vs chop in vs chop out hard to construct. For those I think the separate cover and pack penalties remain the best tool, and I think they would also best pass your public square test by their simplicity.

[muon2]

I can apply your suggestion just fine in 0 pack UCCs like Lansing. It's very hard to write a rule that applies in multi CD UCCs. You can look at a large county like Livingston and say that's bad. It becomes hard when there's a large UCC with a small excess and some small counties on its periphery. I would rather encourage a chop of a whole county rather thana chop into a larger county. If this ddoesn't make sense I'll find a state with an example.

Why not just have a double penalty for a chop into a county in a large UCC that also exceeds the cover minimum?

I understand your concern, but the additional potential erosity points would tend to discourage the chop (overwhelmingly so, if all intra-county cuts count, rather than just those cuts over the minimum as I propose). And you can sever the county without penalty, if both the cover and pack maximum/minimum are not violated (to deal with close cases, one might perhaps consider the pack/cover rule as not being violated if the amount of the otherwise pack CD outside the UCC, or the additional CD in the UCC, caused by the whole county severance, is less than a macrochop in size). Anything beyond a macrochop pad to me is just traducing the whole meaning of UCC's, and is a huge loophole.

You final point is about having two smaller chops into a UCC being penalized as opposed to one larger chop (the infamous Lansing example again). I don't think that should be penalized myself, beyond the chop penalty itself. That is protecting UCC's over other counties without good reason in my view.  

Actually my final point was to consider the distinction between small UCCs like Lansing and large UCCs greater than one CD. I can see the value of treating small UCCs like single county UCCs and your rule would be easy to implement for those. I think it is in this class that you have most observed the problems of overprotecting the UCC.

I think problems occur in larger UCCs where there will be a lot of combinations that make a rule for whole county chop vs chop in vs chop out hard to construct. For those I think the separate cover and pack penalties remain the best tool, and I think they would also best pass your public square test by their simplicity.

Actually I think my rule is simple - almost elegant really. Whether it has drawbacks in actual implementation remains to be seen with examples (which I can't envision at the moment, but no doubt your most active and creative mind is trying to do so ). In big UCC's, nest CD's per the pack rule, and if you sever off a county, if that sucks another CD in per the cover rule, that is penalized. If it doesn't, well something has to be cut off from the UCC, and if it is a whole county, that's great, and no penalty. Other than that, all chops are the same, everywhere, subject to triggering the intra-county erosity test, which while complicated, is necessary, to avoid mischief in how the chop in densely populated areas is effected. But with my unnecessary cuts only get a penalty rule, if you draw the chop cleanly, almost by definition you won't get hit too much with penalty points.

I'm afraid this is actually pretty opaque to me. I highlighted the part that I can't parse. The sucks in vs if it doesn't feels arbitrary, though I'm sure it's not supposed to be. What determines if a whole county chop or a partial chop due to a different district is the one that goes over the cover limit? I strongly support a rule that makes no priority among districts that cross beyond the UCC; they should all be measured the same.

I may need a stepwise procedure. I'm that kind of guy.

[muon2]

I was looking at the two plans for SE PA, and they seem to illustrate the trade of UCC penalties for erosity.

The seven county region in SE PA almost precisely accommodates 7 CDs, so it's hard to imagine a top plan that doesn't take advantage of that. Since there are 7 CDs and no whole county subregions that form a whole number of CDs there should be at most 6 county chops, and both plans have that. The size of Lancaster, Chester, and Berks require that at least one of those counties gets a macrochop. Macrochops increase erosity both internally and on their perieters, so it's advisable to keep the macrochops internal to a region if lower erosity is a goal. Torie's plan does that by using Chester for the macrochop.

OTOH, the Philly UCC doesn't include Lancaster or Berks which are single county UCCs. The PHilly UCC has a cover of 6 and a pack of 5. Train's map adheres to that, but Torie's map increases the cover of Philly by 1. Using my MI model for scoring, neither would get a pack penalty, Torie's plan gets a cover penalty, and train's plan gets a single county UCC penalty.

I'm not a fan of the single county UCC penalty (though I'm tracking it since it's come up so many times), so let me ignore it for now. That leaves train's plan penalty-free, but with I suspect a higher erosity due to the shift of the macrochop out to Berks. Torie's plan is probably lower erosity when a detailed count is made. That actually seems like a reasonable trade to me, since this is the kind of flexibility I'd like to see. I'll put the task of a detailed score on my to do list to see how much trade there is.

[muon2]

Is there any way to draw it with only 3 districts in Philly?

Bucks plus Philly is more than 30K over the population of 3 CDs. If Bucks goes into Philly to gain the 80K it needs, then something else has to go into Philly, too. To get 3 CDs in Philly it is better to have Bucks get population from Montco and let Delco go into Philly alone.

Here's an example for SE PA that seeks to minimize erosity. It avoids the tri chop of Chester with a simple (non macro) chop of Berks. The cost is a penalty for both the cover and pack of the Philly UCC. But that's what the trade off is for.



Edit: The image has a precinct in Souderton, Montco that should be in CD 4. If it is swapped with the adjacent precinct for Telford then CD 4 is -1942 and CD 5 is +174 in population devaiation with no change to cut erosity.

[muon2]

Is there any way to draw it with only 3 districts in Philly?

Of course there is– but at the cost of an extra chop elsewhere (most likely Montgomery, as per Torie's map). 

Indeed. With seven CDs in seven whole counties there will be six chops. They can be placed in various counties but the total count is fixed.

[muon2]

Macrochops are defined by the whole county, not by the individual pieces. If the remainder after removing the largest district fragment is greater than 5% of a district then it's a macrochop. Based on that both Montco and Baltco are macrochopped and would have erosity measured based on county subdivisions as applied to all district fragments.

[muon2]

Train's map is very similar to one I posted a couple of years when train and Torie were both looking at MD. Note that this was before UCCs and with microchops, but some of the same problems show up with the MSAs here.

As I noted at the outset, the real issue comes to the existence of chops. One reason I start with apportionment regions is to identify where I can conserve on chops. In this case I looked at the regions with an eye towards MSA preservation, but I will suggest that there must be at least one CD and one region that spans the MSAs. The Washington MSA counties down the western shore but without Frederick are just barely above 3 CDs and with one microchop can be divided nicely within the 0.5% population limit. By putting all of Glen Burnie into the Baltimore city CD I was able to avoid any other chops for that CD or the Anne Arundel-Howard CD. Eliminating those two chops also has the effect of reducing the erosity by 1 point. I can probably clean up the line between CD 4 and 5 with some work, but the precincts aren't the best unit to match city lines. Edited to reflect a better line through PG that avoids city splits and keeps both CDs within 0.5%.



For the record the BVAPs for CD 4, 5, and 7 are 41.2% (plurality), 54.9% and 53.1%. The partisan distribution is 5D, 1e (D+0.1%), 2R for a polarization of 14 and a skew of 0 after accounting for the expected Dem lean.

On the same thread I also put up this plan that improves the UCC chop for Balto, but it would be disallowed since the ferry from Point Lookout is seasonal.

If the Smith Island ferry crossing counts as a connection, then it is possible to eliminate the microchop so that there are just the five main chops This substantially reduces the population inequality. In the plan below the range drops from 7066 to 3820 which corresponds to an inequality score change from 13 to 8. The increased erosity created by the split of the ES is compensated by reductions elsewhere so that the erosity is the same as for my previous plan adjusting for the ferry.

There is also a noticeable increase in competitiveness with the seats going 3D, 2d (D+1.9, D+5.2), 2e (D+0.1, R+1.2), 1R. That's a 4 point improvement though it does create a 1 point skew for the Dems.

[muon2]

I would think that LATFOR has the minority data from the ACS mapped to the census block group level for NYC as well as the rest of the state.

[muon2]

I would think that LATFOR has the minority data from the ACS mapped to the census block group level for NYC as well as the rest of the state.

Why would one think otherwise?  I am not sure why you are mentioning this.  Did you notice my question about how subdivision chops are treated, and how it varies between ordinary chops and macro-chops?

I was only mentioning it in the context of knowing more about how to draw your NY-07.

[muon2]

For NYC neighborhood mapping, should one use the NYC map, or the “official” neighborhood tabulation map. The latter has smaller hoods, and election result data (which is very convenient).  Are smaller hoods better or larger ones better, that is the question.  Some of the NYC hoods in the first map like Bensonhurst, have very large populations, and others wander around quite a bit, making things unwieldy.

One way to think about the ideal size of neighborhoods is to think about the point at which they can't be macrochopped by a single chop. If the threshold for a CD macrochop is 5% of the quota, then a unit which is 10% of the quota can't be macrochopped by one chop since one piece will always be under 5%. For NY in 2010 that number is 71,771. So subunits with populations around that size will generally work well. The second map is a better fit by that measure.

[muon2]

For NYC neighborhood mapping, should one use the NYC map, or the “official” neighborhood tabulation map. The latter has smaller hoods, and election result data (which is very convenient).  Are smaller hoods better or larger ones better, that is the question.  Some of the NYC hoods in the first map like Bensonhurst, have very large populations, and others wander around quite a bit, making things unwieldy.

One way to think about the ideal size of neighborhoods is to think about the point at which they can't be macrochopped by a single chop. If the threshold for a CD macrochop is 5% of the quota, then a unit which is 10% of the quota can't be macrochopped by one chop since one piece will always be under 5%. For NY in 2010 that number is 71,771. So subunits with populations around that size will generally work well. The second map is a better fit by that measure.

There are 59 community districts (the first map), and 188 official tabulation districts (the second map). So in 2010, that was an average of 138,814 residents for the 59 districts, and 43,564 for the 138 districts. I assume one goes smaller until perhaps the size gets down to less than half of the 5% quota perhaps (unless the other option is more than twice the quota perhaps? What is the metric for choosing, taking into account how erose or wandering the shape of the districts are presumably?

At twice the macrochop threshold or less, a single chop cannot create a macrochop. That number was 72K in 2010. The community districts are almost double that, so as one created VRA districts in NYC there would likely be a number of macrochops. That would mean identifying the subunits of the community districts in order to determine erosity. The NTAs are on average well under 72K, so there would be little concern about chopping them and creating a macrochop.

[muon2]

I have read Tennant and I don't find anything there that suggests a state can have a range substantially beyond the "minor" variation of 0.79% without something more compelling than the whole counties and minimal population shift in WV. That 0.79% refers back to a number discussed in Karcher v Daggett. As the case notes,

You must be logged in to read this quote.

Unlike WV, NY does not have a long history of completely preserving counties and they could not use that argument from Tennant (The court found that WV had never in its history chopped a county). A case could be made that preserving towns and cities while chopping some minimal set of counties would better balance those interests. I don't see the case that NY could make to have a range in excess of 1%.

That said, we are only looking at projected estimates for 2020. A larger range makes sense in that context. But it should be understood that as the estimates get closer to Census day and the accuracy of the projection increases, the range should drop accordingly.

[muon2]

We've had this debate before. Fundamentally I think that as the range gets up over 1% it starts to look like the substantially equal standard which is applied to the states. The standard for CDs uses the equal as practicable standard. If words mean anything then "substantially equal" is different than "as equal as practicable", and I would expect the court to find that there is a difference. We know that substantially equal can mean up to a 10% range. If the range of a congressional plan approaches 10% and is upheld then there is no difference that I can see.

[muon2]

I don't find the history of NY CDs compelling before the OMOV rulings of the 1960's. I'm also not seeing the case and 5 justices that would reinterpret Wesberry to match Reynolds during the upcoming cycle.

[muon2]

I don't find the history of NY CDs compelling before the OMOV rulings of the 1960's. I'm also not seeing the case and 5 justices that would reinterpret Wesberry to match Reynolds during the upcoming cycle.

Why is this better?



And why doesn't it beat anything that can be produced under your rules?

It might depending on the specific data.

There is a natural trade off of chops for population inequality, and as more geographic units are available the lower the theoretical inequality. Analysis of the 2010 data gave rise to this graph and table from the muon rules thread.

You must be logged in to read this quote.

We looked at use of the table in the MI exercise a few years ago. Summing the CHOP and INEQUALITY scores gave a good measure of that trade off and how well optimized it was. For example, this was my upstate NY plan from last year's estimates (2015 vintage) projected to 2020. The upstate CDs (19-26) have 4 county chops, 1 UCC pack penalty, and a range of 7401 for an INEQUALITY of 13. The total with CHOP is 18 and is paired with an erosity of 44.

This might be even better politically, and it's hard to see any serious complaints from the fruited plain. It keeps chops the same and reduces erosity by 3, so it is certainly better by the basic rules. Two points of the erosity reduction is in a more compact Buffalo-Niagara Falls CD, the other point comes from a reduction of 3 from the Hudson CD with an increase of 2 from the North Country. Everything still projects within 0.5% of the quota for 2020.

[muon2]

As I noted earlier, the precision of the estimates doesn't really work well in trying to project what counties can be grouped together within a 0.5% deviation from the quota. Let me use my map from last year as an example. It used the 2015 county estimates and projected them to 2020 using the same rate of change that occurred between 2010 and 2015. Without changing boundaries, I'll do the same calculation with 2016 estimates.

CD	2020 (2015)	%deviation	2020 (2016)	%deviation
quota (thousands)	776.6	0.00%	768.3	0.00%
19	774.5	-0.28%	771.8	0.45%
20	780.3	0.47%	777.0	1.13%
21	779.6	0.38%	775.7	0.97%
22	777.5	0.11%	765.6	-0.35%
23	773.3	-0.43%	768.8	0.06%
24	775.4	-0.16%	770.9	0.33%
25	772.9	-0.48%	771.0	0.35%
26	777.6	0.13%	774.3	0.78%

The estimates for NY state as a whole dipped lowering the projected quota for 2020 by 8.3K. Even so, the projection has a greater impact on NYC so effectively upstate is now projected to be too big for 8 CDs. The changes from 2015 to 2016 are reflected in projections that go anywhere from a 1.9K drop in CD 25 to a 10.9K drop in CD 22.

The upshot is that projecting whole county groups within 0.5% isn't possible with this level of accuracy in the estimates. Even projections to within 1% would fail as seen in the example CD 20. At best this table suggests that 2% deviations could be projected, but that wouldn't help determine chops if SCOTUS doesn't want to consider those deviations minor in the sense used in Tennant.

[muon2]

If the goal is to share maps that show what might be in 2020, and those maps are based on minimizing chops and erosity, then I'm saying those maps may have little to do with what will really be possible after Census data comes out. A 1% shift in the overall deviation due to differences in projection vs future reality can substantially change the groupings that minimize chops.

[muon2]

So with the idea that the sets of numbers are independent, here's what I might draw for upstate NY using 2016 county estimates projected to 2020. CDs are all within 0.5% of the quota and chops are minimized and placed to minimize erosity. CD 18 (with Hudson) is drawn to avoid a macrochop since it would only be about 14K overpopulation with whole counties.

[muon2]

So with the idea that the sets of numbers are independent, here's what I might draw for upstate NY using 2016 county estimates projected to 2020. CDs are all within 0.5% of the quota and chops are minimized and placed to minimize erosity. CD 18 (with Hudson) is drawn to avoid a macrochop since it would only be about 14K overpopulation with whole counties.

You are using a target deviation range of 1%, where the SCOTUS has explicitly refused to set a de minimis standard. Your plan has a deviation range about 8 times as large as my last plan. What state interest justifies this large deviation?

I set an upper limit (not a target) of 1% range to be consistent with described minor deviations. I find a state interest in minimizing divisions of political subdivisions including counties and towns. I find a state interest in making districts compact in terms of the road connections between political subdivisions. I find a state interest in minimizing population balanced by those other interests using a consistently applied measure of that balance.

[muon2]

That "speech" by Muon2 has a nice cadence. He must be a politician or something.

But what of the map? Did the rules work on this year's data set well enough to satisfy the fruited plain?

[muon2]

That "speech" by Muon2 has a nice cadence. He must be a politician or something.

But what of the map? Did the rules work on this year's data set well enough to satisfy the fruited plain?

I quite like your map. The major flaw is the carve out of Schenectady from the Albany CD, but life is not perfect. But the CD that Columbia County is parked in is far superior to mine, which had a carve out of Hudson, that would freak folks out. although the folks involved are so few, that the freak out would not have much punch. Faso will send you an Xmas card for your map. You saved his butt.

The Albany UCC is like Grand Rapids, a UCC too large for one CD and requiring a macrochop to satisfy the pack rule. Shifting the macrochop to a UCC pack penalty greatly reduces erosity in both cases.