Michigan: The urban-rural divide writ large

[Torie]

 
 

As some of you know, the Muon2 redistricting rules include a component that rewards drawing districts that are either nested within metro areas, or outside them, rather than including both. For larger metro areas, that metric tends to drive where the lines on the map go.

So, in drawing up Michigan after the 2020 census, putting aside exorcising gerrymandering, the CD’s tend to hew more closely to following the urban-rural divide than they do now.  And consequently, the partisan stats of the map below highlights well the urban cosmopolitans going one way, and the rest of the nation going another, at least when it comes to white voters.

The PVI chart below is organized based on the swing to Trump from Romney. Putting aside the Macomb County anomaly (big metro area white voters swinging to Trump to the same degree as their rural and smaller city compatriots), one can see that the swings to Trump that are smallest (or away from him), are within the zone of the big three metro areas (Detroit, Grand Rapids, and Lansing).  Everything outside that zone, swung massively to Trump (plus Macomb County).

As to the partisan effect of this map, as compared to the existing Pub gerrymander, it depends whether you think Trump is the future, Trump is half the future, or Trump is an anomaly, and in due course, his impact on politics will disappear as if he never existed.

If Trump is the future, then the Dems take the hit for the seat Michigan loses (the old MI-09, which disappeared, and is now the number for the old MI-14),  while the GOP drops MI-11 to the Dems, but picks up from them MI-05 (who knew that a Flint-Saginaw based CD would now be a Pub CD?!). 

If the future is Trump lite, with the PVI figures using the Cook method of average the PVI’s for the last two election cycles, then each party shares the loss of Michigan’s CD, each losing half a seat. MI-04 and 11 go swing from safe Pub (for a net loss of one seat), but for the Pubs MI-05 goes swing from Dem, leaving each party with a half seat loss (the Dems lose MI-09, with MI-05 going swing, but MI-11 and 4 go swing to them in exchange, also netting out to half a seat loss for the Dems). 

If Trump is an anomaly, and a mere vagrant on the waters of the public square, then the map does more what would be expected: the Pubs lose two seats (MI-04 as the Lansing seat is created for the Dems, rather than being gerrymandered away, plus MI-11 as that gerrymandered object d'art is also tossed out), for a net gain of 1 Dem seat as the the old Dem MI-09  disappears from the map.

You choose as to what the future may hold. My guess, is Trump lite, lite, when it comes to more upscale precincts that used to have a Pub lean, and Trump lite for the balance, leaving the Pubs with a true swing seat with MI-11, with MI-04 tilt Dem, but within striking distance, and MI-05 lean Dem (with the incumbent needing to work hard, and have political skill to avoid becoming vulnerable in what is potentially a highly volatile CD now).

[Torie]

That Detroit gets 2 CDs instead of 1 is my major problem with the Muon2 rules.  If I were to re-write the rules, cities would have to remain intact if their population merits it, and breaking cities up would be considered a macro chop of the first order.  Why?  Keeping cities intact stops the problem in states like Illinois where Chicago has a disproportionate number of seats through bacon strip districts.  With a population of 689,000 Detroit doesn't deserve 2 seats in Congress.

Detroit needs to be split to have two CD's that will elect black congresspersons.  After the next census, it is probably true that there are not two CD's out there with 50% BVAP (which would require the chop), but it may be that there are two CD's with 50% CBVAP. In any event, the political reality is that there need to be two black CD's.  Ideally, if more minority CD's are to be drawn than legally required while following the redistricting rules, it should require the consent of both parties. In this instance, such consent would probably be secured, because MI-11 is now a swing CD, and may slide towards a lean Dem CD. So it is not that costly for the Dems to avoid having the black politicians get angry at them.

[Torie]

"I would shift MI-2 eastward to include all of Saginaw and the thumb. The Michigan and Huron shoreline  are distinct and it is no longer possible to create two districts in Northern Michigan.

'This would mean coming north into Flint rather than west into Ann Arbor to get the extra half a district."

That sounds like you are reverting back to subjective COI criteria. 

[Torie]

It's all about chops (including chops of metro areas as defined) and erosity. Nothing else matters much, and if it did along the lines you suggest, we are back into the subjectivity quicksand pit, and the whole exercise falls apart.

[Torie]

"I would shift MI-2 eastward to include all of Saginaw and the thumb. The Michigan and Huron shoreline  are distinct and it is no longer possible to create two districts in Northern Michigan.

'This would mean coming north into Flint rather than west into Ann Arbor to get the extra half a district."

That sounds like you are reverting back to subjective COI criteria.  

We agree the Detroit UCC has the population for 5.5 districts. What is the difference between adding Genesee or adding Washtenaw, to get to 6?

It gets a lower Muon2 score in all probability. And putting aside that the map will probably have a higher erosity penalty score, and without seeing how it upsets the apple cart elsewhere, particularly with respect to the Lansing metro area, because adding Washtenaw is an exact fit generating no chop, while Genesee is not an exact fit and thus adds a chop, per se out of the box a chop will need to be lost elsewhere to make that up, meaning MI-05 will need to not have a chop, or MI-02 or MI-06 will need to have one less chop.

[Torie]

That Detroit gets 2 CDs instead of 1 is my major problem with the Muon2 rules.  If I were to re-write the rules, cities would have to remain intact if their population merits it, and breaking cities up would be considered a macro chop of the first order.  Why?  Keeping cities intact stops the problem in states like Illinois where Chicago has a disproportionate number of seats through bacon strip districts.  With a population of 689,000 Detroit doesn't deserve 2 seats in Congress.

A single district in Detroit would be like ~80% Black.   That wouldn't comply with the VRA.

If/when in the not-so-distant future the BVAP in SE Michigan falls below that which can support two CDs, the VRA won't be a barrier to the city Detroit getting the proportionate representation that it deserves, instead of the outsized representation that it doesn't.  Suburban interests deserve representation, too.

You still have the packing issue.  I am not sure there is an on point case on this. The precise legal issue is whether minority packing can be deemed legal where the rationale is that it keeps a subdivision whole, in this case Detroit.  In other words, say if Detroit were kept whole, the CBVAP of that CD is 70%, with the adjacent CD 15%, in lieu of say 50% CBVAP and 35%. In the context of a contiguous minority population crossing county or subdivision lines, even though it will entail additional chops, do you need to unpack a CD, in order to create a second adjacent CD with a substantial minority influence, that is the question. Maybe Muon2 has some knowledge on this matter. The law is in flux in this area. That much I know.

For the 2020 census, there probably will be enough contiguous blacks to still have 2 CD's able to elect candidates of their choice, perhaps 45% CBVAP in each CD. In that context, the odds are think are pretty high (at least more likely than not), that the courts will not be happy with a packed CD using the excuse that it keeps Detroit whole. That would certainly be the case, if it were possible to draw two 50% CBVAP CD's, but that will not be the case after the 2020 census. It will be closer to 45% CBVAP for each CD, maybe a tad higher, but not much.

[Torie]

That Detroit gets 2 CDs instead of 1 is my major problem with the Muon2 rules.  If I were to re-write the rules, cities would have to remain intact if their population merits it, and breaking cities up would be considered a macro chop of the first order.  Why?  Keeping cities intact stops the problem in states like Illinois where Chicago has a disproportionate number of seats through bacon strip districts.  With a population of 689,000 Detroit doesn't deserve 2 seats in Congress.

A single district in Detroit would be like ~80% Black.   That wouldn't comply with the VRA.

If/when in the not-so-distant future the BVAP in SE Michigan falls below that which can support two CDs, the VRA won't be a barrier to the city Detroit getting the proportionate representation that it deserves, instead of the outsized representation that it doesn't.  Suburban interests deserve representation, too.

You still have the packing issue.  I am not sure there is an on point case on this. The precise legal issue is whether minority packing can be deemed legal where the rationale is that it keeps a subdivision whole, in this case Detroit.  In other words, say if Detroit were kept whole, the CBVAP of that CD is 70%, with the adjacent CD 15%, in lieu of say 50% CBVAP and 35%. In the context of a contiguous minority population crossing county or subdivision lines, even though it will entail additional chops, do you need to unpack a CD, in order to create a second adjacent CD with a substantial minority influence, that is the question. Maybe Muon2 has some knowledge on this matter. The law is in flux in this area. That much I know.

For the 2020 census, there probably will be enough contiguous blacks to still have 2 CD's able to elect candidates of their choice, perhaps 45% CBVAP in each CD. In that context, the odds are think are pretty high (at least more likely than not), that the courts will not be happy with a packed CD using the excuse that it keeps Detroit whole. That would certainly be the case, if it were possible to draw two 50% CBVAP CD's, but that will not be the case after the 2020 census. It will be closer to 45% CBVAP for each CD, maybe a tad higher, but not much.

The courts wouldn't just look at Detroit alone, there is also Black populations in places like Pontiac, Southfield, Oak Park, Inkster, River Rouge, etc.   Drawing two 50% BVAP districts in the greater Detroit area will continue to be easy as cake for the long term future.  

No, it's not actually. In fact, based on the 2010 census it is barely possible. Given the declining black population in the minority zone, and the declining Detroit population, along with some black dispersal into areas that are mostly white, by 2020, it might be around 45%-47% BVAP each for two CD's (the higher figure more along the lines of CBVAP), and maybe a bit less, unless you gerrymander to grab the non contiguous black population in Pontiac, which I doubt a court would require, although it would certainly tolerate that.

It is figures that are on the cusp like this, where there is no bright line 50% BVAP figure in play, but probably enough to elect two black congresspersons of the minority's choice, but with uncertainty as to just how low the figures can go, without putting that in play as well (if the two CD's don't divide the black population evenly, to avoid chops or whatever), that put the whole matter into the twilight zone of current court jurisprudence. Maybe some interesting court cases will ensue, alleging the Goldilocks solution has not been implemented. Then the parties argue what Goldilocks would want. Fun stuff!  

[Torie]

Jimrtex’s suggestions for a revised Michigan map are clearly a winner.  It loses two chops from my effort, including a macro chop in Wayne County, so it also has a much better erosity score (which it probably does even without the lost macro-chop).  I still take a pack penalty in the Grand Rapids urban cluster, to avoid a macro-chop.  I learned from Muon2 that in most cases, one takes a pack penalty to lose a macro-chop, because it tanks the erosity score.  There may be a higher scoring map that the uber computer can find, but I tend to doubt it. The population array worked to make this map a winner. When I did that rectangle in Oakland County for MI-11, and the population fit perfectly, I was just amazed. It was like winning a lottery ticket. Well done Jimrtex!

I also like the chaos that it does from a partisan standpoint, including a host of potentially marginal CD’s, but I digress. It also shows that while the Trump coalition may be disaster for Congressional Pubs in CA, the reverse is true in Michigan. With this map, the Dems are down to but two safe seats – the two black seats. Everywhere else, the Dems will have to sweat. But then, then there are seats the Pubs will need to sweat in too – like in four of the CD’s they currently hold rather safely. And that is the way it should be.  And oh yes, the incumbent politicians will hate this map.  Good! ☺

 
 
 

[Torie]

Based on the 2015 county census estimates and projecting the same rate of change forward through 4-1-2020, as we always do. Obviously the rate of change varies within counties, so that would affect any map. But that data is not available, or if it is, too much trouble to manipulate. Outside of the ultimate shape of the two black CD's, and to a lessor extent in Oakland County, the effect will be rather marginal in any event as it pertains to Michigan. In part that is due to just how rare the chops are.

[Torie]

Like the second revision a lot. If I had to guess, I'd say Dems win the 12th, 13th, 7th, 9th, 4th, and 8th. R's take the 1st, 2nd, 3rd, 5th, 6th, 10th, and 11th. But the only ones I'd call safe for either party would be the two Detroit seats for D's and the 2nd and 10th for R's.

When competitiveness takes first priority the map usually suffers as a whole (Arizona), but when a good map has the extra bonus of competitiveness it's like a cherry on top.

Yeah, more or less, although I think MI-08 is a tossup and very unpredictable (with the population trends presumably favoring the Pubs in this divided district as Genesee lags in population, while its chunk of quite Pub Oakland (60-40 Pub) has substantial growth going on (along with Livingston County), while MI-11 is only tilt Pub (and probably trending Dem a bit).  MI-07 is also trending Dem. Washtenaw is a growth hub, while the 5 Pub counties in it (to varying degrees) are all population losers (so sad!). In MI-04, there probably is no trend, and the Dems will need to be careful not to nominate some fashionable politically correct Lansing liberal (like the Dems did with Teachout in NY-19), that will repel folks outside of the cosmopolitan zone. Their margin for error in MI-04 is rather thin. The now retired Pub Mike Rogers if he ran in that seat (well he lived in Livingston but whatever) would hold it.

So 6 CD's to the Pubs, 5 to the Dems, and two tossups really, with a fair amount of unpredictability. Not bad from a political equality standpoint if Michigan is now a closely divided state politically.

Here by the way is another iteration of the same map, that loses a chop involving the Grand Rapids urban cluster, but at the cost of two additional erosity penalty points (two sensitive state highways are cut, that are avoided where one goes for two chops rather than one). I don't like it, because it "uglifies" the map, but it does make the pareto optimal frontier, given its fewer number of chops attended by more erosity penalty points. So the map is a finalist, with the ultimate map chosen a separate process. Note that Kent County cannot be chopped, because it is in the urban cluster, and MI-06 chopping into it would generate a cover penalty since MI-02 has already chopped into the urban cluster by taking Ottawa County. So that leaves Ionia County, with the attendant erosity penalty points and ugliness. One must choose.

[Torie]

Alas here is a road cut fail. It turns out that the pretty map (mine of course) generates a road cut between Allergan and Grand Rapids due to a lack of state highway going in a relatively straight diagonal line between the two cities (one goes SE to the freeway, and then north), while the ugly one does not.   So we are stuck with ugly.



[/URL]

[Torie]

Nobody has ever suggested that you're dumb, Muon2. Stubborn maybe, but not dumb. Jive aside, most excellent. That is the kind of exhibit that you just don't want, in an opposition brief.

Wait a minute. Grand that it is permitted (that's another issue), although that is not needed here. Are you saying that the ugly map gets on the pareto optimal frontier, while the pretty map does not, due to the extra erosity penalty point? (The balance of the map is identical.) If so, then you will have to face this exhibit. Maybe there is no escape from that. The highway paths are bizarre. But it is an issue.

[Torie]

I looked at my map, and your point is moot here. MI-179 runs from Bradley to Hastings so the Allegan fragment in the pretty map is regionally connected to Barry. Pretty wins.

You don't consider that there is a state highway connection to Kent County?  It needs to be the same highway number to count? Putting that aside, weird road connections can lead to weird maps.

[Torie]

No, unless I misunderstand something, in the pretty map, the state highway from the county seat of Allegan where it joins US 131 near Plainwell slips into MI-06 as one goes north on Hwy 131 for a bit before going back in MI-02 on its way to Kent County, while staying in Allegan County the whole way until it hits the Kent County line. Sure it is a circuitous route, that probably nobody would use that wants to go from Allegan to Grand Rapids, but rules are rules. So that is an extra highway cut no? In the ugly map, there is no wandering into MI-06.

[Torie]

No, unless I misunderstand something, in the pretty map, the state highway from the county seat of Allegan where it joins US 131 near Plainwell slips into MI-06 as one goes north on Hwy 131 for a bit before going back in MI-02 on its way to Kent County, while staying in Allegan County the whole way until it hits the Kent County line. Sure it is a circuitous route, that probably nobody would use that wants to go from Allegan to Grand Rapids, but rules are rules. So that is an extra highway cut no? In the ugly map, there is no wandering into MI-06.

Allegan is not a macrochop so all that matters is which district the road enters in the chopped county. Roads between towns within a county only come into play in a macrochop. In general, the number of links between counties doesn't change when a simple chop is made, all that changes is where those links between counties are assigned. Here a link to the relevant example.

So you don't need a state or federal highway that covers the entire distance within the chopped county? Once a state highway enters the chopped county, it no longer matters if you have to use a county highway within the CD to get to the county seat (because the available state or federal highways wander outside the CD)?  If so, that might allow for some pretty nasty chops into a county, as long as no macro chop is involved. Just be careful to avoid a chop on the county line that cuts off the state highway entry point.

[Torie]

So is my statement true?  Some of the confusion I think arises from the other discussion about what counties are connected for purposes of being allowed in the same CD (where this issue I know came up). But here, we are just chatting about whether or not an erosity penalty point is in play.

[Torie]

So is my statement true?  Some of the confusion I think arises from the other discussion about what counties are connected for purposes of being allowed in the same CD (where this issue I know came up). But here, we are just chatting about whether or not an erosity penalty point is in play.

The statement that careful selection of where a district cuts into a county can be used to lower an erosity score is true. There are consequences to any of the possible interpretations of a county chop like we are discussing. I thought that on the balance we generally agreed that the consequences were worse with other interpretations.

On the question of which counties are regionally connected I posted this map two years ago during our analysis of different MI plans looking for the Pareto frontier. The only change since then is that local connections can be used to connect counties, but with an erosity penalty.

OK, thanks for the discussion. It is good that you have carefully thought all of these things through, balancing the pros and the cons, after receiving input from others. Now please do the same on the Pub health care bill discussion to which I invited you. Be brave!  

[Torie]

With reference to the pretty map versus the ugly map, they are tied for chops and erosity. So then one turns to partisan aspects, but that is tied too. And then there is equality. If the ugly map is more equal, that wins, but the maps are tied there too (the 5th and the 8th CD's define the extremes).  So do you flip a coin?  Was it finally resolved how equality plays into the formula for picking maps, or is that still up in the air?  I recall Muon2 wanted to give it more importance, and I was resisting. I think partisan balance is more important.

[Torie]

With reference to the pretty map versus the ugly map, they are tied for chops and erosity. So then one turns to partisan aspects, but that is tied too. And then there is equality. If the ugly map is more equal, that wins, but the maps are tied there too (the 5th and the 8th CD's define the extremes).  So do you flip a coin?  Was it finally resolved how equality plays into the formula for picking maps, or is that still up in the air?  I recall Muon2 wanted to give it more importance, and I was resisting. I think partisan balance is more important.

I assume that this is hypothetical since I showed the pretty map has a lower erosity due to the existence of MI-179. When we (Torie, jimrtex, traininthedistance, muon2) drew multiple MI maps with 2010 data we looked at the role inequality could play. We scored maps in different ways, including use of populaton as a tie breaker. Our conclusion was that the best use was to add the INEQUALITY score to the CHOP score as one axis of the Pareto test. This the table used to determine the score.

The INEQUALITY score for a plan is found by taking the range for a plan and comparing it to the table below.

Range	Inequality
0-1	0
2-10	1
11-100	2
101-400	3
401-900	4
901-1600	5
1601-2400	6
2401-3200	7
3201-4000	8
4001-4800	9
4801-5600	10
5601-6300	11
6301-7000	12
7001-7700	13

This creates some incentive for lower inequality, but any lowering must be substantial enough to offset any extra chops.

So if one ends up with maximum inequality, that adds 13 chops? That is huge. Did you test out this schedule? It seems to me you have baked into the cake, maps that have a lot more chops. I don't think that is good policy. Am I missing something?

[Torie]

OK.  We shall see what the monte carlo run looks like in due course.

[Torie]

What is the penalty for a chop of a subdivision again, and how does it vary depending on whether the county is macro-chopped or not? It's confusing to me, because I think subdivision chops are all erosity penalty points, rather than chop penalty points, even though one pretends otherwise with macro-chopped counties, that each of its subdivisions are counties. (I do remember ala Warren that if you macrochop a subdivision, than the precincts generate erosity points, but that is not my question here.) And I assume with respect to subdivision chops, one gets no reward at all for respecting village lines with the chopped subdivision. Is that correct?

[Torie]

Torie you might be overestimating the willingness of Trump voters in Michigan to vote Republican in House races.

Not really. How Trump Democrats vote down the road is heavily dependent on events. And predicting how things will evolve with Trump is close to a fool's errand. He can turn on a dime.