Using urban county clusters in MI

[muon2]

I know everyone was desperately awaiting my "fair" non-partisan Michigan map for the 2012-2020 cycle hewing to the Torie redistricting algorithms. Given who is in office, the Pubs picked up one seat for their gerrymandering efforts (MI-07 - the Pub gerrymander was all about containing Ann Arbor as it turned out). Well, MI-11 might have fallen, given the antics in that seat in 2012, but that seat would be a tough hold for the Dems if that happened. Have a nice day everyone.

cc: sbane

This plan does not seem to hew to your usual standards for fine maps. The slice through Kent stands out as particularly messy, and ideally the Bay City-Saginaw separation should be avoided. It get hard for me to craft an erosity measure for your plans as you seem to move further away from the use of microchops with each map. That moves the chop-erosity balance point away from a natural equilibrium.

[muon2]

Muon2, geography and population numbers drove me to the Kent County chop. It is a pretty clean chop actually, with most of the Grand Rapids metro area kept whole (yes, there is that one burb jut, but see if you can do better with the Kent County chop avoiding locality chops given the balance of my map (I suspect not)). The Bay City-Saginaw comment sounds more like a communities of interest issue, which I thought we had both jettisoned (unless perhaps it is an absolutely obvious one, and all other things are close to equal).

I seek out micro-chops, and there is one in this map (at the cost of a bit more erosity for MI-06), but not at the cost of substantially more erosity. I tried every county combo reasonably possible bearing in mind erosity issues seeking a mirco-chop for MI-01, but no dice. Once you are beyond a micro-chop, I still bear in mind the degree of the chop, but again, not at the cost of much more erosity. In the Detroit metro area, the counties are of such high population, that smaller chops are just not possible, and while maybe the chop could have been smaller if say, MI-05 picked up Huron County, and lost most of the Lepeer chop, or had a small chop into Genesee of Huron has more than MI-05's share of Lapeer), the extra erosity was just not worth it.

And yes, MI-12 is hideous, but the VRA made me do it, plus avoiding MI-11 chopping into Wayne County.

If you think there is a better map that I would like more (yes, me, not you, given my order of priorities, I'm all eyes).  I am not claiming that there is not a better map out there; I just couldn't "find" it if there is.

Below is an alternative chop for Kent, but as you can see from the population discrepancy, it will be at the cost of a locality chop, which I avoided.

I'll concede the Saginaw question. I thought Midland-Bay City-Saginaw was in the same metro area which you still used as a parameter. I checked the Census and they classify them as three separate MSAs.

What population tolerance are you using? The plan above is within the 1% range, which I thought was OK. Certainly a smaller range would be scored better, but that's a trade off with the chops and erosity.

If I use the 1% range I would put forward something like this, and I can microchop it down from there. It has only one county chop outside metro Detroit, and two muni chops in the metro, though perhaps I can improve that given more time for analysis. The two VRA CDs are both about 55% BVAP since I use the Pontiac-Romulus connection for CD-12. The partisan split is 3D, 4d, 2e, 4r, 1R where +1.5% is the tossup cutoff and +5.5% is the lean (lower case) upper bound.

[muon2]

Putting aside your Pontiac action for MI-12 (not acceptable), I think your map might have potential. I first went the Monroe route for MI-14 myself, but decided to go the more compact route (the Dems would howl that MI-14 was pushed close to the competitive zone, when it could have obtained its extra population by going into Washtenaw, making everything more compact rather than less. But maybe that led to the Kent chop; I don't know. And would you please put up the map again removing the coloration for the empty water "precincts?" It is hard to read your "ugly" map without all that colored water! Why don't you use the more updated version of Dave's utility, which makes for much more attractive maps? Thanks.

Or you could send me the data file, and I could generate the map doing it my way.

"The partisan split is 3D, 4d, 2e, 4r, 1R where +1.5% is the tossup cutoff and +5.5% is the lean (lower case) upper bound."  Was this just an FYI comment, or meant to be a factor in how you drew the map?

I'm not sure what your beef is with my CD-12. When I did a comparison to your version with a bounding-rectangle-type erosity test, like you favor, it comes out pretty much the same. Note that both of our CDs there have about the same length-width aspect ratio. Certainly they are not so different that it would generate a significant scoring difference? Or is it CoI that prevents you from making the leap?

I have a bookmark directly into the DRA app so I never noticed that he had a new link. In any case I loaded it from his page, and found that it is the same as my bookmark. The water comes from my search to fill in unassigned precincts and to make sure my whole county CDs don't have leftover pieces from another CD. The water around MI has to be assigned to make that work and it DRA leaves a few hundred souls in those water precincts. I suppose I could go back and unassign them all to make a pretty map after I've checked the numbers, but that will be a task when I've got more time on my hands. Numbers come first.

The political data was not a factor in drawing the map and was provided for a comparison to the data that you displayed. However, I suspect that any sound process should test whether some contending map has a significant bias in either the political polarization or skew. Polarization measures the extent that too many districts have become non-competitive (ala incumbent protection maps) and skew measures the bias from the expected political division. My observation is that all the Midwestern efforts for reform these last few years have insisted that there be some investigation after drawing to insure that a political party has not been unduly favored by the map.

edit: On a tangential question, does MI law permit the division of a political unit between two CDs to result in discontiguous pieces of a CD within a unit? For example, can a chop into a county from a CD grab two different townships on the county border that don't directly border on each other? I know we have discussed this in more general circumstances and viewed it as OK if it avoids muni chops, but I don't recall if that violates specific rules in MI.

[muon2]

My map accounts for all but one person, by getting rid of the water precincts.  Send me please your data file, and I will put up your map. I just can't read yours very well - at all.

I see that you simply grab the water areas when DRA puts a significant number of people there. Here's the list of "water people" in DRA 2.2 using 2010 voting districts:

LP:
Charlevoix - 657 (Beaver Island was placed in the not defined water area and you shade this on your map.)
UP:
Chippewa - 1 (This is the one you didn't shade. Why not treat it the same as Charlevoix?)
Luce juts out in a strange way, the coders didn't follow the shore, but part of the water is separate.

The larger issue for me is tracking down missing precincts since DRA doesn't have a whole county tool like mos redistricting packages. I use a spreadsheet to set up my apportionment regions which is the key to eliminating chops. When I have time I'll clean up the maps, but I'd prefer consistency, so I would be prone to uncolor Charlevoix (and Chippewa) after I've recorded the numbers.

[muon2]

Oh, I found a way to clean up the Wayne County situation, and get rid of that nasty chop of Livonia.
MI-11 moves 20 basis points more to the Dems in this iteration (by reducing Royal Oak to a micro-chop).

This is a reasonable way to deal with CD-12, but it also highlights some of the problems with most erosity or compactness measures. Let's compare it to my version that meets with your disfavor.



Your desired erosity has generally turned on the degree of "rectangularness" with squares being better than mere rectangles. To measure the rectangularness of a CD one generally draws some shape around it and then compares the areas. If the shape is a rectangle or other minimal convex polygon then my CD wins easily since your offering has that large expanse of Detroit sitting within the bounding area but not adding to the are of your district. My guess is that the areas of our districts is comparable, so the large the bounding area the worse the situation. You could come out better by using a bounding circle which is known to punish long thin districts compared to C-shaped ones, but that would hurt your first offering for CD-12 which also seemed like a reasonable shape. A perimeter test doesn't help your shape either since mine uses lots of straight lines and would score well.

"On a tangential question, does MI law permit the division of a political unit between two CDs to result in discontiguous pieces of a CD within a unit? For example, can a chop into a county from a CD grab two different townships on the county border that don't directly border on each other?"

Yes, I believe Michigan law permits that, but I try to avoid doing that unless it generates other good redistricting principles aspects. My new map of Wayne does that (MI-12 going back down into Redford), because it makes for cleaner lines (and avoids having long tails running west in Detroit even though the CD has more territory in Oakland to the north of the tail).

Oh, I managed to get rid of the Redford jut, and still keep MI-13 over 50% BVAP - barely. Whew!

A look at you last revision shows some improvement. The C-shape is less pronounced so the surrounded area of Detroit would be reduced as well. However, my eyeball guess is that the fairly direct shape of my CD-12 would still win a measure test like those I described above. Mine is really not a bad shape for a VRA district and if you divorced it from the map and embedded it in some other state it probably wouldn't bother you so much. You've spent a long time looking at situations in MI and that shows.

OTOH your arcing CD-12s make a good case for my use of local connectivity as a measure of erosity. In the simple case you can just count which jurisdictions in the chopped counties border on other jurisdictions. Now that big chunk of Detroit in CD-13 just counts as one unit with relatively few connections to units in your Wayne part of CD-12. It becomes competitive with my CD-12, as it should be, since I think it represents a reasonable alternative that a commission could consider.

[muon2]

This is a reasonable way to deal with CD-12, but it also highlights some of the problems with most erosity or compactness measures. Let's compare it to my version that meets with your disfavor.

The biggest problem with this map is that there are two districts that span Oakland and Macomb (9 and 11), as well as two that span Oakland and Wayne (9 and 12)- isn't that a no-no in Michigan?

I also prefer my map because it has no muni chops besides Detroit.

I forget some of the unique MI rules since they aren't part of the generic model. I've edited my map to get rid of the county spans. It reduces the chop count, too. And I edited out the water.



edit: CD-12 is 53.0% BVAP, CD-13 is 54.3% BVAP. No real struggle to reach 50% since Pontiac helps significantly.

[muon2]

I also prefer my map because it has no muni chops besides Detroit.

I forget some of the unique MI rules since they aren't part of the generic model. I've edited my map to get rid of the county spans. It reduces the chop count, too. And I edited out the water.



edit: CD-12 is 53.0% BVAP, CD-13 is 54.3% BVAP. No real struggle to reach 50% since Pontiac helps significantly.

Here's a question for traininthedistance, Torie, and others. train likes to avoid muni chops. So here's a variation of my above map that eliminates 2 muni chops in Oakland. The price is a county chop in Lapeer and increased erosity in CD 11 by elongation. Which should be preferred if either?

[muon2]

Avoiding muni chops by generating much more erosity does not appeal to me. A rule of thumb perhaps might be half (if that much) as much additional erosity as the limit for additional erosity to avoid county chops (putting aside the Michigan law issue). In Michigan, I guess you first minimize county chops, and then you get points for less erosity, and points, albeit fewer points as compared to erosity measures, for micro chops.

That's why I asked the question. The erosity didn't go up much, but there is the issue of two muni chops traded for a county chop. I'm ambivalent at the moment, so if you'd like to compare the two offerings and comment I'm interested.

It also doesn't help that I'm not getting any useful feedback on how to measure erosity. I hoped my example with your revised Wayne offerings would shed light on how to judge those. It goes to the issue of comparing your map to train's as well.

You must be logged in to read this quote.

I will consider the Pontiac issue in due time.

You must be logged in to read this quote.

The Census Bureau and Wikipedia list Beverly Hills as an incorporated village, and I understood incorporated munis of any flavor count as separate from the township for redistricting. However as I look at the legal description of the districts, it looks like incorporated villages are ignored. That's a bit different than other Midwestern states treatment of incorporated munis, but so it is.



The MI statute requires equality, "The constitutional guideline is that each congressional district shall achieve precise mathematical equality of population in each district." What is interesting is that it was clearly written based on understanding of SCOTUS rulings circa 1999. Today they could easily change it to the actual federal standard from Wesberry, "The Constitution requires that members of the House of Representatives be selected by districts composed, as nearly as is practicable, of equal population," and take advantage of rulings such as Tennant from last year.

edit: I found that incorporated villages are an in-between type of govt. They are still part of a township (or two or more) and don't appear as a Census minor civil division, but they are on the list of incorporated places. If a MI township is all villages or cities, the township is dissolved and all the villages in that township become cities. I'll rework my map with that in mind.

another edit: looking at the approved map it appears that there can be two districts that split the same two counties if one is VRA. See MI-11 and MI-14 in the current Congress.

[muon2]

With the new knowledge about the MI rules for townships I've revised my plan. It's prettier now, the water is cleared, the numbers match up with the approved CDs, and with the shift of Pontiac Torie should be well pleased.

The map begins with five apportionment regions (1, 2-3, 4, 6-7-8, 5-9-10-11-12-13). No region exceeds a deviation of 0.5%*sqrt(k) where k is the number of districts in a region, and the range between regions is 9189. The region chop score is 5.

To make districts no city or township is split except for Detroit. Microchops less than 0.5% of a CD are used to balance population, and only Ottawa outside of the three large counties has a chop larger than a microchop. The range is 3292 and the average deviation is 463. The district chop count is 8. The VRA districts 13 and 14 are 51.5% and 54.0% BVAP.

[muon2]

So with my revised map, let me address the OP in regards to MI.

I've used the Torie spreadsheet method to convert from DRA presidential numbers to PVI, and in parentheses put the actual PVI as calculated going into the 2012 election.

CD-1 R+3 (MI-1 R+4)
CD-2 R+3 (MI-2 R+7)
CD-3 R+6 (MI-3 R+5)
CD-4 D+1 (MI-4 R+4)
CD-5 D+5 (MI-5 D+10)
CD-6 D+0 (MI-6 R+1)
CD-7 D+7 (MI-7 R+3)
CD-8 D+5 (MI-8 R+2)
CD-9 D+2 (MI-9 D+5)
CD-10 R+2 (MI-10 R+5)
CD-11 R+1 (MI-11 R+4)
CD-12 D+2 (MI-12 D+14)
CD-13 D+28 (MI-13 D+34)
CD-14 D+27 (MI-14 D+27)

I applied the PVI shift to the actual vote results in 2012. The highlighted CDs would change from R to D, though CD-11 would have been very close. Conceding CD-11 the net shift due to redistricting would be 3 seats changing from 9R-5D to 6R-8D. Note that even in 2012 not all D+ CDs would have shifted since incumbency is a significant factor as well. In any case there would be a number of seats that would be much more competitive under my alternate map.

[muon2]

If Muon2's map has extra chops, macro or micro, his map is in trouble under Michigan law. As I said before, we are doing two things at once here. I don't mind an extra county chop if it is more than compensated by other benefits, mostly by reducing erosity, or say eliminating 3 muni chops, and so forth. But Michigan law precludes that.

In terms of Michigan law, if we assume that the microchops would be massaged to perfect equality using tools more precise than DRA later on (and if we assume that doing so negates the necessity of 5's two prongs into Oakland), then Muon's map would have to put the Grosse Pointes back in 14, adjust other things as necessary to account for that, and then throw in the tiniest of microchops to get into 1/4 up north.

My map would, obviously, just have to add in an equivalent series of microchops.

Last night my task was to factor in my corrected understanding of cities vs villages. I also used that to reduce outstate chops beyond the micro size to just one in Ottawa. I'm now looking at the Grosse Point question, but I don't want to trade a fix there for the chop of Genesee (or other northern county) that train has in his map.

The macro chop counts are the same right now for our maps. Counting the number of districts in a county over one and ignoring microchops, I have 1 in Ottawa, 1 in Macomb, 3 in Oakland, and 3 in Wayne for 8. train has 1 in Kent, 1 in Monroe, 1 in Genesee, 1 in Macomb, 2 in Oakland, and 2 in Wayne, also a total of 8.

And yes, I expect that exact equality in a plan like mine is achieved by expanding or contracting the microchops as needed.

[muon2]

Last night my task was to factor in my corrected understanding of cities vs villages. I also used that to reduce outstate chops beyond the micro size to just one in Ottawa. I'm now looking at the Grosse Point question, but I don't want to trade a fix there for the chop of Genesee (or other northern county) that train has in his map.

I think the real trade-off here is whether you chop a northern county, or whether you decline to make an all-Macomb district when that county's population is more than large enough to support such a thing.  The Grosse Pointes are just an extra chop to be eliminated, no matter what.

"(with the caveat that a micro-micro chop would still be necessary to get into 1/4)"

What does the above mean?

Even with the microchops on Muon's and my map as it is, 1 and 4 are only microchopped with each other, but together are still about 100 persons over exact equality.  So an extra, unshown, microchop of particularly small size must be added.

Though as I noted before, MI could revisit its statute to comport with modern SCOTUS. Then things like microchops between regions of low deviation (such as 1 and 4) would be unnecessary.

[muon2]

Last night my task was to factor in my corrected understanding of cities vs villages. I also used that to reduce outstate chops beyond the micro size to just one in Ottawa. I'm now looking at the Grosse Point question, but I don't want to trade a fix there for the chop of Genesee (or other northern county) that train has in his map.

I think the real trade-off here is whether you chop a northern county, or whether you decline to make an all-Macomb district when that county's population is more than large enough to support such a thing.  The Grosse Pointes are just an extra chop to be eliminated, no matter what.

"(with the caveat that a micro-micro chop would still be necessary to get into 1/4)"

What does the above mean?

Even with the microchops on Muon's and my map as it is, 1 and 4 are only microchopped with each other, but together are still about 100 persons over exact equality.  So an extra, unshown, microchop of particularly small size must be added.

Thanks. I wonder if a twist can be made to Muon2's map counterclockwise, so that MI-09 gets shoved out of Wayne, and more into Oakland, getting rid of the micro-chops in Oakland, and only generating micro-chops going beyond Oakland, into Genesee or Livingston or whatever. In other words, can a macro chop be eliminated in Wayne without creating another macro-chop elsewhere?  If not, I don't see a problem with Muon2's little thrust into Wayne myself.

Well, my revised map now has one fewer macrochop (and one fewer chop total) than Muon's, so his map simply does not meet the legal standard.  Oh, and it's actually CD-10 that goes into Wayne, and oh, I just noticed, does so by way of a traveling chop.  That's multiple strikes right there- more chops than necessary, and a traveling one at that.  

I don't follow this concern about two county chops linked that you call a traveling chop. I don't see it barred when I read the redistricting statute. There's technically not a two districts spitting the same two counties restriction, and as I noted the current enacted plan with MI-11 and 14 though the VRA can be invoked as a defense. The actual statutory language is for district lines to break as few county boundaries as is reasonably possible (not the stricter as practicable), and that split must involve shifts of population between just the two affected districts.

One way to execute the statute is to approximate districts as whole counties. Then for each rough district make a shift of population between then to reduce their population inequality. This process continues pairwise until all the districts are equal. In the case of my above map Consider a starting point with CD 10 in St Clair and underpopulated, CD 9 in Macomb and overpopulated, CD 11 in Oakland overpopulated, and CD 14 overpopulated in Wayne. One pairwise shift in Macomb brings up the population of CD 10 but drops CD 9. The next pairwise shift in Oakland between CD 9 and 11 brings CD 9 up to full population. Then a pairwise shift between 10 and 14 in Wayne brings both to equality. No two pairs shifted more than once and each only in one county. Is there a reason that this is not consistent with the actual text of the statute?

[muon2]

Despite the preceding post justifying the legality of the previous map, I am interested in improving the scores for population inequality, chops, and erosity. I like train's use of southern Oakland to bolster the VRA districts so I'll start with that and see if I can avoid any chops outside the big three counties. If I just let CD 10 fill in the remainder of northern Oakland around CD 11, it's somewhat erose reaching down to gran Wixom and I'm not wild about such a large population deviation sticking out on CD 9. With a bit of rearrangement I found the following plan:



CD 11 is nicely compact along southern Oakland and has a deviation of +167. (DRA says +180 but I'll trust my spreadsheet addition of the actual 2010 census figures.)

CD 9 now has a deviation +579. (DRA says -1027, but it clearly has a bug that links a precinct in New Baltimore with a disconnected one in Lenox Twp, again I'll trust the census.)

CD 10 has a reduced erosity and a corrected deviation of -1717 and its the only one that exceeds 1000 in the state.

I've also rearranged the VRA CDs to get more compact shapes. They have 51.5% and 54.8% BVAP for 13 and 14.

The overall inequality has dropped my range to 2476 with an average deviation of 427. The chop count is down to 7, and the erosity is reduced from my previous plan.

The downside is I've lost some of my most competitive CDs. CD 9 only shifts from D+2 to D+3, but CD 10 jumps from R+2 to R+6, and CD 11 goes from R+1 to D+3 and would have cemented that CD for the Dems in 2012.

[muon2]

To compare maps it's a useful exercise to push the county chops as low as possible. The theoretical chop minimum is equal to the number of districts that can be completely nested in a county. That is equal to 4 for MI. I couldn't get it to 4, but I could get it down to 5 with use of microchops. The range pushes out to 6599 and all CDs are with 0.5%. I would note that I could get rid of all the microchops but the one from CD 7 into Ingham by adding a chop that puts Holly twp in Oakland into CD 5. There is also one chop into Westland in Wayne, but it could be eliminated and a microchop of Rockwood used instead (I saw it after I printed the map).

One other issue that arises in the exercise is the minimum BVAP for the VRA. The SCOTUS standard for 50% only applies to the question of whether there is sufficient population to require that there is a district that can elect a candidate of the minority's choice. It doesn't apply to the district itself, and Dems in IL were successful in arguing that percentages in the upper 40% range where the white population was overwhelmingly Dem was sufficient to elect a candidate of choice for blacks. It didn't go to SCOTUS but it did pass the lower courts. The OH competition after consultation with the NAACP came to the same conclusion and allowed a CD of as low as 47% BVAP in Cuyahoga. I'm going to use their argument here and the CDs have 50.2% and 49.2% BVAP and I feel confident that statistical analysis of voting would conclude that blacks could elect candidates of choice in both CDs.

[muon2]

Well you are buying a legal challenge, Muon2, and might well lose here, because the 53% BVAP CD in existence now (just my recollection),  elected a white man. So with black Southfield right next door, what you suggest in the real world is just not going to happen in my opinion. It is almost entirely a theoretical exercise.

I'm just saying that the legal theory prevailed in court in IL, and Cook is not so different than Wayne and its immediate environs. I'm also noting that Pubs use the 50% standard mostly as a safe harbor when the NAACP is preparing help Dems challenge their map. The map has to follow federal law, and the court specifically does not say that 50% BVAP is necessary in a district designed to elect a black. This gets more complicated in areas where bloc voting is no longer prevalent and then no VRA district is mandated. For me this is a situation where there is some precedent to permit the district, so if the map scores well enough it should go to the commission. The commission should be the arbiter of whether there is too much risk of failure in court, not the mapping rules. IMO, the rules should only echo the VRA, not evolving interpretations of the VRA.

So, other than VRA issue, how should this map be viewed compared to my previous offering?

[muon2]

From an algorithm standpoint, your map is clearly high scoring, and probably the best of your lot, but I would lose your micro-chop and go for a macro-chop for MI-01 to make it less erose, and MI-04 looks unfortunate. Is there any way with another macro-chop to clean that CD up? I'm also uncomfortable with the quad-chop of Ingham County (and the cut-up of Lansing for which you should be docked), and again, would prefer another macro-chop to doing what you did. I think (or at least suspect) you give more weight to micro-chops than I think is either wise, or salable in the public square. The Detroit area portion of your map is just excellent however, and extremely high scoring - putting aside the near fatal VRA issue, in my opinion.

The Pubs of course would veto your map in a heartbeat, but that does not mean, again putting aside the VRA issue, that with a few tweaks, your basic design would not be one of the top five maps, as veto bait for the Pubs.

The Ingham chop actually follows the muni lines so East Lansing is in 4 and Lansing is in 8. The microchop from 5 could just as easily be in Livingston as in Ingham and the microchop from 7 could be in Washtenaw. A chop counts the same whether it is in an existing chopped county or not. Many systems (like used in the OH challenge) actually favor putting chops in an already chopped county. I prefer nesting them in corners because that generates less erosity than jutting in the middle of the county side.

The shape of 4 isn't bad from an erosity standpoint - its only when you see the CoIs that are split that one might raise a concern. I like to turn off muni and VTD lines and look at the new district by itself to see how it appears.

CD 1 is another case where the erosity measure makes a big difference. CD 1 is huge in area and the deviations from a straight line near Grand Traverse don't have much impact in a score based on geometric shapes. A boundary segment method like I've advocated does provide a way to weight that trade off. Are you willing to sign on yet?

[muon2]

From an algorithm standpoint, your map is clearly high scoring, and probably the best of your lot, but I would lose your micro-chop and go for a macro-chop for MI-01 to make it less erose, and MI-04 looks unfortunate. Is there any way with another macro-chop to clean that CD up? I'm also uncomfortable with the quad-chop of Ingham County (and the cut-up of Lansing for which you should be docked), and again, would prefer another macro-chop to doing what you did. I think (or at least suspect) you give more weight to micro-chops than I think is either wise, or salable in the public square.

The Ingham chop actually follows the muni lines so East Lansing is in 4 and Lansing is in 8. The microchop from 5 could just as easily be in Livingston as in Ingham and the microchop from 7 could be in Washtenaw. A chop counts the same whether it is in an existing chopped county or not. Many systems (like used in the OH challenge) actually favor putting chops in an already chopped county. I prefer nesting them in corners because that generates less erosity than jutting in the middle of the county side.

A reasonable consideration when defining apportionment areas is to avoid splitting counties that share the same most significant urban area.

In Michigan this would include:

Detroit: Wayne, Oakland, and Macomb.
Lansing: Eaton, Ingham, Clinton.
South Bend, IN: Berrien, Cass.
Traverse City: Grand Traverse, Leelaneau

The only neutral definition of the urban area or MSA comes from the Census. Even with that defined, there are a number of ways to judge whether a split is significant or incidental. I agree that there should be consideration for regions that maintain metro areas, but given the subjective component, I think this is better left as a consideration for the commission selecting from the best maps.

[muon2]

The only neutral definition of the urban area or MSA comes from the Census. Even with that defined, there are a number of ways to judge whether a split is significant or incidental. I agree that there should be consideration for regions that maintain metro areas, but given the subjective component, I think this is better left as a consideration for the commission selecting from the best maps.

You can't see the county boundary within an urban area since there is continuous development.  A cut along a county boundary in such an area is indistinguishable from a county chop.

Perhaps not from a CoI perspective, but the county boundary is very clear electorally. Counties are fundamental political and electoral units. They're one of the primary identifiers used by members of the public. Few voters don't know what county they live in. I'd far rather preserve the county line than a changing urban area boundary.

[muon2]

The only neutral definition of the urban area or MSA comes from the Census. Even with that defined, there are a number of ways to judge whether a split is significant or incidental. I agree that there should be consideration for regions that maintain metro areas, but given the subjective component, I think this is better left as a consideration for the commission selecting from the best maps.

You can't see the county boundary within an urban area since there is continuous development.  A cut along a county boundary in such an area is indistinguishable from a county chop.

Perhaps not from a CoI perspective, but the county boundary is very clear electorally. Counties are fundamental political and electoral units. They're one of the primary identifiers used by members of the public. Few voters don't know what county they live in. I'd far rather preserve the county line than a changing urban area boundary.

I grew up pretty near the boundary of Essex and Passaic counties (specifically, in Bloomfield), and we would cross that line for shopping, errands, and such all the time, whereas traveling to west or south Essex was a rare thing.  We identified with Clifton and Little Falls and Passaic way more than we identified with Fairfield or Millburn or Irvington.  

I spent many weekends in the 80's visiting close friends in Montclair so I know what you are describing. But this is an example of a community of interest. Once one opens the door to this type of grouping as part of the rules, then there are so many subjective interpretations that it allows partisan desires to be masked as legitimate goals. The IL Dems passed two resolutions that extended hundreds of pages to outline exactly this type of CoI for their house and senate maps. They acknowledged that they were blatantly partisan, but used the CoI as cover.

You must be logged in to read this quote.

New Jersey is rather unique in the US with both strong counties and town subdivisions. In the Midwest, many voters do not know their township but certainly know their county. For most the township exists as a basis for the county to track assessments and to group voting precincts. Only in farm areas where the township is the primary provider of road maintenance does it take a more prominent role.

You must be logged in to read this quote.

But I am not proposing following the urban area boundary.  I am proposing not using county lines that significantly divide an urban area.  It is not the micro-chops that are being objected to.  It is the slice between Eaton and Ingham counties.

I see the use of apportionment regions as part of a process of decomposition with the state being divided into regions based on communities of interest which when done at a congressional district level includes areas of less-populated counties and metropolitan areas.

I think that by proposing the county lines not significantly divide an urban area then that is equivalent to saying that one use the core of a metro area as a boundary. To deal with geographic obstacles like mountains it made sense to disallow connections between counties that were otherwise contiguous. This proposal essentially requires that connections between counties for the same urban area not be broken. The best neutral version to apply this is to use the central counties of the MSA as defined by the Census Bureau. From a map construction view this is equivalent to treating the counties that can't be split as a single "super-county". When all such mandatory bindings are included it can be shown that the super-county is equivalent to the core of the MSA as defined above. Thus one would in effect be using that area as a boundary.

I view the apportionment region as fundamentally a tool to minimize chops while maintaining low population inequality. I'm not opposed to looking at the preservation of metro area as a legitimate goal of goon redistricting. However, I think that unless one wants to insist on super-counties to replace individual counties, this is a subjective exercise that should be left to the commission when comparing the top scoring plans.



Th apportionment regions in this map are
Region A (1 district: CD 1, deviation -0.11%)
Region B (1 district: CD 2, deviation -0.35%)
Region C (1 district: CD 3, deviation +0.39%)
Region D (1 district: CD 6, deviation +0.34%)
Region E (3 districts: CD 4, 7, 8, deviation -0.57% is within 0.5%*sqrt(3))
Region F (3 districts: CD 5, 9, 11, deviation +0.19%)
Region G (4 districts: CD 10, 12, 13, 14, deviation +0.10%)

As an aside, none of this discussion would have any affect on this map. The apportionment region in question includes CDs 4, 7, and 8. That region has the entire Lansing MSA within it. The issue of the urban area split happened when the region was divided into CDs.

As another aside, the central counties of the Detroit MSA are listed as Wayne, Macomb, and Oakland, with three other counties listed as outlying in the MSA.My map divides these into two separate apportionment regions, but still keeps a sense of the overall metro area (only outlying Livingston of the MSA is left out) in those two regions. Doing so explicitly eliminates a county chop and the metro area is going to be split among multiple CDs anyway. It seems to me that the map rules should allow this.

[muon2]

I had trouble following the partisan balance formula Muon2 put up (at the end of it, it seemed to not make sense to me - perhaps he can put up an example of how the math works). I also have an issue with MI-04 in his map with the jaggedness of its north-south lines, and wish the erosity formula took that into account (even if MI-04 had straight lines, and was more of a rectangle, it to me is pushing the limits of elongation - Train with one of his map had MI-02 too elongated as well for me, although the lines were straight enough). Ditto with MI-01. I would prefer a macro chop to the jagged lines in the south. One macro chop for nice long straight lines as opposed  to some jagged affair, just seems more sensible to me. A bit of jaggedness to find a micro chop is fine. Reaching for it does not appeal to me. I would prefer a formula that recognized that as it were.

Start by finding the state PVI, which is D+3.8 for MI. Multiply by 14 CDs and double it to get an advantage of 1 seat. Since the seat gained is matched by a seat lost, the expected skew is 2 for the Dems.

Using caps for PVI 6+ and smalls for PVI 2-5 the current map is 4D, 1d, 1e, 7r, 1R. The raw skew is 5-8 = -3 for the Dems. The skew score is the difference between the expected and raw skew and equals 5.

My map with the questionable CD 4 has a breakdown of 5D, 2d, 1e, 6r. The raw skew is 7-6=1 for the Dems and the skew score equals 1. Even though the plans have different skews they both have a polarization of 18 (2*5+8).

[muon2]

We can simplify the map by using urbanized areas only, which have populations greater than 50,000 persons.  This does not change the multi-county regions, but reduces the clutter.  Urban area with less than 50,000 (ie urban clusters) are less than 7% of a congressional district, and concern about dividing them in a county-based system is overly constraining.  Urban clusters were not defined before the 2000 census, and would not have been available for the 2000+ redistricting (this delay is a liability in any redistricting scheme that depends on current census delineations).

This seems like an excellent map to use for discussion. There are two points I would launch from here. Let me first affirm that this is the most recent table of county-based areas at the Census. Please point me at some other table if this isn't the right one to use for reference.

My first issue is the linking. Niles-Benton Harbor is a separate MSA but in the same CSA as Cass, so here they are linked, though I know from experience BH is not that strongly linked to SB. Van Buren is in the same MSA as Kalamazoo, but is outlying so it is not linked. Battle Creek is a central county MSA in the same CSA as K-zoo and is very strongly linked to K-zoo. Either Cass-Berrien and Kalamazoo-Calhoun are both grouped together or both stand alone. I can't see how to resolve this with one pair one way, but the other treated differently. I personally would treat each county grouped CBSA/MSA independently and not use the CSA-level lumps as they make apportionment regions with 0.5% district limits a task that begins to fail to craft useful results.

The second is the question of scale. Clearly, the western and SE groups in the map cannot each be placed into a single CD. If this is CD based, then there needs to be some definition of how much of the cluster falls into one or more CDs. If the map is only used at the level of region formation, then my issue is resolved, but there comes a separate question about when and whether a very large county cluster, such as the SE, can be treated as two separate regions that together span the larger area.

[muon2]

My concept is simple: Groups of counties that share urbanized area(s), and the delineation is straightforward using census data.

Metropolitan Statistical Areas based on commuting patterns are more problematic.   The threshold for the amount of commuting is arbitrary.  Is a county with 24% commuting more integrated than a county with 26% commuting?  As a suburban area becomes more developed, the number of commuters may increase, but the percentage decrease.  First there are workers to build the homes, and then grocery stores and malls, dentists, and auto dealers.   And then officer parks.  Fewer jobs are in manufacturing and the CBD.

In 2000, the Michigan planning authorities noted that Grand Rapids could be anywhere from a 1-county to a 7-county metropolitan area based on small percentage shifts in commuting.  The outlying counties shifted between 2000 and 2010.

Commuting data is based on the American Community Survey (ACS), so there is possibility of sampling error.  There may also be response error.  Workers may not know their county of employment.  Question 30 of the ACS questionaire is quite elaborate as it seeks the information in multiple ways, so that the census bureau can determine where employment occurred.  A worker may work at multiple job sites during any week, and may have been employed in a different state or even country than where they resided.

Inclusion of outlying counties may be useful for a subjective COI analysis, but not for use as part of a rule set.

I understand your rationale for using urbanized areas instead of MSAs. However, I'm not sure I follow the application in the linking example that I used above. The separation of Benton Harbor-St Joseph UA from the South Bend UA is substantial compared to that between the Kalamazoo and Battle Creek UAs. The South Bend portion in Berrien seems quite small, so I can't follow why BH-SB is linked when KZ-BC is not.

Also if counties are the coarse building block, which I think they should be, then I would need something that clearly maps the UAs and their fractional population into counties. I haven't found that list on the Census site yet. I used MSAs in my earlier comments in large part because of their clear delineation by county.

[muon2]

From the steps that jimrtex describes, I don't see that we differ all that much. There is one very important caveat to that statement, however. The SCOTUS rulings on population equality are critical. Ranges in excess of 1% will probably not be tolerated even with a strict set of rules, so capping deviations at 0.5% becomes a necessary constraint. That 0.5% constraint then forms the natural division between a chop and a microchop. Minimizing chops also becomes an important factor, since only by holding to a consistent goal can any population deviation be allowed. I'm willing to trade chops for erosity, but a careful balance needs to be established to justify the rules and that's a discussion that is evolving more among the MD maps.

My first response to MI maps in this thread produced a plan that I contend is quite like jimrtex's. The impact of my 0.5% condition forced the western region to reflect two CDs, which were resolved with a chop in Ottawa. To compare with jimrtex's maps I've started with his version and I've shown the chops in the big three SE counties as pentagons. Technically you can view that area including the big three as a single region of seven CDs (7.004 to be more precise) not unlike jimrtex's description. The southern CD is the one anomaly in the map at 0.987 of a CD and outside the 0.5% tolerance. But it is close enough that it can be adjusted by making three microchops shown as triangles.



As a thought experiment to push the envelope I found an older file I had worked with in pre-microchop days, but while I was using the 0.5% maximum deviation. The key elements were that it found two separate CDs in the west, one of which was quite compact, and found that the SE region had a nice subset of three counties that were almost exactly 4 CDs (4.001).

I then found that the three CD region for Pontiac-Flint-Saginaw (3.002) could be split with Flint-Saginaw having the same 0.987 that the Ann Arbor CD had in my first plan. If the microchops worked there, they would work here, too. Similarly the middle southern stretch included a set of 7 contiguous counties in the very south that came in at 0.991 and could be also be fixed with microchops. That left only one chop to place outside of the big three, and Ingham is the only place it can be made with one chop as the map shows. Again, it's hard to say this doesn't follow jimrtex's process, except for the desire to keep groups of counties within the 0.5% mandate and use microchops of 0.5% where possible.



There must be firm constraints or it is easy to show that one can draw a fairly partisan map despite the intent of the rules. That's the exact problem in MI. For three cycles the state used the Apol standards in neutral hands and got neutral maps. Those rules were codified in statute in the late 90's thinking that all maps would come out looking like Apol's. The strong partisan lean of the last two sets of MI maps shows the problem of not sufficiently constraining the mappers. Yet, overly constrained maps can result in a plan that would not pass local sensibility. The trick here is to allow multiple maps to pass through a strict test.

The type of thought experiment I provided here is to show what might or might not pass a test so that the right balance of a few select maps can move forward with some confidence that politically drawn offerings are unlikely to be in the selection. Local decisions can then provide the insight so that if both my maps above were to pass, the second might fail in the commission due to the carve up of the Lansing area, and that's OK by me. The question is should the second map be knocked out in the first step, and if so why?

[muon2]

This is a final version.



If there are VRA concerns swap Dearborn, Dearborn Heights, and Inkster for more of northwestern Detroit (probably to around Highland Park.).

There could also be a 3-way rotation of Plymouth and Northville to Wayne South-Monroe, suburbs south of Dearborn and Detroit to Detroit-Wayne East (The Detroit district includes Detroit east of the Southfield Freeway, the Grosse Pointe's, Hamtramck, Highland Park, and River Rouge), and more of Detroit to Wayne West-Oakland South.  It is probably too much of a stretch to include Livonia as well.

It's an attractive map, but I count 9 chops in 7 counties, and lower chop counts are definitely possible. I doesn't look like it's the least erose map possible either. We've seen other versions that also seem to maintain as much of urban areas intact in districts, too.

In order to have any population deviations at all one must show that this is the best implementation of a set of uniformly applied principles. Your steps are clear, but the rules that drove your choices on individual county swaps and chops are not. Can you codify principles that would result in this map?