US House Redistricting: Michigan

[dpmapper]

Now, you cant be so sure about these things.  Democrats have never targetted these members.  People said that Allen Boyd, Melissa Bean, Jim Oberstar, Rick Boucher and Charlie Melancon and Bart Gordon(both of whom would have been defeated had they run for reelection) were invulnerable.  Democrats are likely to start targetting these Republicans in Obama districts and in a better Democratic year, they are likely to win some of these districts. 

Other than Oberstar (D+3 district), these are bad comparisons.  Bean was only a 3-termer, first elected against an old/weak incumbent and re-elected only in strong Dem years; the only reason her loss was an upset this year is because nobody thought Walsh was a strong candidate.  The 4 southerners that you mention represent districts that are R+6,+11,+12,+13, whereas you are complaining about districts that are roughly even, or maybe R+1 or 2.  With incumbents who have demonstrated an ability to win even through two straight Dem waves, you can take those chances.  (It's not as if *every* Dem in a D+3 or worse district went down this year - Minnesota still has Walz and Peterson, for instance.) 

[dpmapper]

Why bother?  Torie's last map is the way to go - Dems down to 4.5 districts.  I don't see how one can improve on it. 

[dpmapper]

Wouldn't MI-13 have to add that last precinct of Grosse Pointe Shores in Macomb County, due to the "minimize township splits" provision of the statute? 

[dpmapper]

Here's the map I think Torie has in mind for SE MI. This map has each district within 100 of the ideal population. It's what I've expected for some time with CD 13 (54.2% black VAP) and 14 (56.8% black VAP) each splitting into Oakland and Macomb to allow a district wholly contained in each of those counties. I use CD 8 from Livingston as the balancer per Torie's suggestion. I'll let him check this against his precinct tables for PVI.

It's a very sexy map. 

However, there are some minor issues, many of which have been discussed:

* Rogers has to move. 
* Given that you're getting rid of two Dem districts, McCotter and Rogers have districts that are probably too strong, esp. considering that they're fairly strong incumbents anyway. 
* Miller's district is only lean-GOP and Levin will probably hop over to try to challenge her, given that he has nowhere else to run.  She might not like that.
* Camp has to take the thumb, which means he has to get Bay County, and the leftovers from Genesee (since Kildee is taking Lansing).  That makes his district quite marginal (and he won't even be able to keep all of his home county, much less most of his current district). 
* Walberg has to take Monroe and the Washtenaw leftovers, plus (probably) parts of the Lansing area.  Quite possibly too much for him to handle. 

However, there is a solution - I noticed that the Genesee leftovers from Kildee will almost exactly match Livingston's population.  So swap them for Livingston:



Rogers doesn't have to move, and he can easily take the rest of Washtenaw as well as the rest of the Lansing area (which he currently represents anyway), now that Lansing proper is out of his district, saving Walberg.  Camp has more room to breathe.  Levin and Peters might fight over the bronze seat, leaving Miller free rein in Macomb, and whoever wins will still have a serious fight on their hands.  I can't run the intra-county numbers and might not have chosen which precincts to put where properly, but might this work just as well?  Or are the Flint suburbs too Democratic to add to northern/eastern Oakland? 

[dpmapper]

OK, after further review my map is not legal; the cycle of the Kildee/Rogers/Dingell/McCotter/bronze districts, each edge of which splits a county, could be rotated in some fashion (since none are VRA protected) so as to eliminate one county split.  Nuts. 

But I think some of Torie's assumptions are incorrect as well.  First off, since the statute's secondary guidelines say that avoiding county splits takes precedence over city/town splits, there is no "nose under the tent" argument - you can't create a county split in order to avoid a town split.  So you can't sneak a Macomb district into the Pointes just for that reason. 

More importantly, I think that you can't argue that the VRA districts demand a double crossing of the Oakland-Wayne border.  Yes, if one of the districts contains the Pointes, then the other district must go into Oakland; this is (probably) true.  But that just argues for using that district to be the sole population equalizer.  For instance, you could have 1 black district (including Pointes) plus one white district entirely within Wayne, use the 2nd black district to split the Oakland Wayne border, put one more district entirely within Oakland, and then go from there.  Unless there's a fortuitous way to somehow avoid yet one more county split by doing the double crossing, I think you have to avoid it. 

[dpmapper]

Are you saying Dingell's white CD in Wayne needs to take over McCotter's MI-11 territory in Wayne?  If so, where does MI-11 get its population back?  Won't it have to split another county?

Count your county splits.  You've got 4 districts hence 3 splits in Wayne, 2 splits in Oakland, 1 in Macomb, 1 in Washtenaw already, and 6 districts down.  Each new district you add will add another split (eg, the district that takes the rest of Washtenaw will likely terminate in a partial county somewhere) except for the last one. So that's 7 more splits, for a total of 14. 

You can definitely do it with 13 splits only; what I suggested involved one black district in Wayne, one white district in Wayne, and one black district spanning Wayne/Oakland.  That's 2 splits there.  Add one district entirely in Oakland.  Oakland is now split twice, since there are leftovers, and I have four splits for four districts down.  Each new district adds a new split where it terminates (save the last district) so this ends with 13 splits.   

[dpmapper]

Here's a map that gets it down to 12, but only because Upton's district fits perfectly in one set of counties.  (There are 2 county splits up north that aren't shown.)



It's a 5-Dem plan; I'm not sure there's a way to get it down to 4.5 without moving McCotter's district entirely out of Wayne, since if you soak up the Oakland blacks with a Wayne district, McCotter seemingly can't go anywhere other than Washtenaw, and if you don't, then you have to concede a district in Oakland. 

[dpmapper]

Yes, you have one less split because you excised Wastenaw from the map. Unless there is one county somewhere that has exactly 33,000 people or whatever, there will be another split elsewhere. I will engage in the exercise however, of seeing if another map results in fewer splits than mine. I tend to doubt it however.

Count the splits.  I've got one fewer than you; once you take care of Oakland and Wayne, the remaining county splits needed should be exactly one less than the number of districts you need to draw (at most).  It doesn't matter that Washtenaw isn't drawn yet. 

In fact, I just rejiggered some things so that there are 11 county splits.  If districts 1 and 2 take in everything north of the Clare-Isabella latitude line, plus everything on Lake Michigan from Ottawa north, plus Newaygo, Mecosta and Lake, that's exactly right.  

[dpmapper]

Tan is 52.2% VAP black.  Bronze is 50.2% VAP black. 




Here's the larger picture.  3 county splits in Wayne, 1 each in Oakland, Macomb, Monroe, Ingham, Eaton, Saginaw, Isabella, and one not seen in Traverse between CDs 1 and 2. 

[dpmapper]

OK, you have a chop in Monroe rather than Washtenaw. Other than that, if you select my plan A where there is no third CD in Oakland, we have the same number of chops in the Livingston, Washtenaw, Wayne, Oakland, Monroe and Macomb package.  

Which is your "plan A" map?  I too am lost amid this flurry of mapmaking.   

[dpmapper]

Torie, I count three CDs in Oakland on that map.  2 splits of Oakland, 1 in Macomb, 3 in Wayne, and 1 in Washtenaw = 7.  On my map I had 1-1-3-0 (plus one in Monroe) for 6. 

It looks like you are relying on the near equality of population for your CD 6. I get a value that is over by 299 persons. That isn't going to be exact enough for MI, so you will need at least one additional county split somewhere.

Ah, OK.  Didn't realize it needed to be that close.  We'll need 13 splits for my map, then (I had CDs 1 + 2 in a set of whole counties as well, but that was only good to within 100 or so.) 

[dpmapper]

The current map has 4 districts in Wayne and Oakland (3 chops each) plus 1 chop in each of Macomb, Washtenaw, Saginaw, Bay, Shiawasee, Allegan, Calhoun, and Kent, for a total of 14, for 15 districts.  This is normal - in general with x districts, one will need x-1 chops unless a numerical miracle occurs.   So with one fewer district this go-round, the norm should be 13 chops. 

I realize the GOP won't want a Dem district in Oakland, but how do you avoid that, while also protecting McCotter?  If he goes into Oakland, the black district can't, and vice versa.  Would a map like this be at all workable? 



Livingston+ half of Oakland, and Lapeer + half of Oakland, might be lean GOP districts (assuming you do the division of Oakland correctly which I may not have)... I think?  One of the black districts takes the leftover bits in Southfield.  Together the two black districts take McCotter's worst areas (Redford, Wayne, parts of Westland) and Dingell's best areas (particularly Inkster, Taylor, Romulus, and Dearborn Heights) leaving them together with the whitest parts of Wayne, plus about 100K population from either Washtenaw or Monroe (more likely the latter).  I notice that Dingell only won his Wayne County portion by 12K votes in 2010, and lost Monroe, whereas McCotter won his Wayne County portion by 21K votes in 2010 (granted, it was an easy race this go-around).  I doubt McCotter likes the plan, but is this a toss-up district?  It's hard for me to tell. 

The GOP then concedes the two black districts, plus an Ann Arbor-Lansing district, plus the Flint-Saginaw district (which should ease any burdens on the other districts - Camp will be much happier not having to worry about Saginaw, for instance). 

[dpmapper]

I guess the "dp" in dpmapper stands for "Democrat Plan" or Democrat Planner" or Democratic Planner" doesn't it?  You are a pretty clever little adversary, I must admit. Well it is better to know now, rather than later. See you in court pal!  

Heh.  Not at all, I'm pretty conservative.  I'm just trying to game out what the scenarios are.  It's been fun arguing with you!   

[dpmapper]

I got it!  Having a black district taking Southfield cuts off McCotter from Oakland, but Dingell taking Ann Arbor cuts off McCotter from going south to Monroe.  The solution: the black district takes Ann Arbor!!!!!!!!!!!!!!



Black districts are at 51.4 and 51.5 VAP black.  The green district coming into Oakland from the north has all of the thumb counties as well, except for St. Clair.  Miller should be safe, as should the new 11th.  Only drawback is that Walberg and McCotter are in the same (green) district.  Is this a dealbreaker?  Can't be: Walberg just moves a few miles north into his new seat. 

[dpmapper]

Wait... does this still fulfill all the other requirements?

7 districts down, 7 splits made.  I'm fine there.  The black percentages are over 50.  You can't see the township lines but I've only split one town between each pair of districts (Detroit between the two black districts and Dingell's, Westland between Dingell and the 11th, one in Macomb, and Farmington Hills and Novi between the bronze district and the other two Oakland districts, respectively. 

The only question is, now that I've used the thumb plus Hillsdale and Lenawee to gain an extra district in the Detroit area, is there enough left to deal with whatever is leftover from the Flint pack?  I'm not sure; I think one of them might have to be swingy.  But I think I'm done for the time being.  Someone can take over from here. 

[dpmapper]

So I lied and decided to finish the job.  Here's the best I can do:





CD-02 takes Mt. Pleasant from Camp's district in red, which expands into Bay City and part of the thumb.  CD-03 (purple) takes Dem parts of Calhoun County (including Mark Schauer's home) away from Walberg, who otherwise would be in trouble since a finger goes up to Saginaw.  I think Upton should be OK in the southwest; his district is probably about the same in partisan balance.  If not, you could try to swap areas within Kent County with Amash, but there's not a lot of room to maneuver.  

CD-01 (blue) and CD-04 (red) can also swap some territory if it is necessary to balance their strength.  

I tried to put Kildee into Lansing, Flint, and Saginaw simultaneously, but doing so requires his district to be involved in 3 county cuts, which severely hamstrings one's options in designing the rest of the map (eg, a Holland-Saginaw district is then necessary) and in order to be legal some of the districts end up a bit too marginal.  

Here's Detroit:



I gave Rogers his area around Lansing back, since they love him there.  Also, I figured the light blue district could stand to come in a little bit more from the thumb.  I don't know what the partisan balance is but you can swap towns between the two if Rogers is too strong, for instance.  McCotter has options within Wayne County, depending on whether he'd prefer Westland/Garden City from his old district, or some slightly less Dem towns in SE Wayne County.  ETA: Ah, he definitely wants Grosse Ile!  So swap that in for more of Westland...

Tan district is 51% black VAP, bronze is 52.2%.  Neither of them ventures into Wayne County outside of Detroit (with one small exception of Northville, which might be necessary in order to not split the town; if it's not, McCotter gets a few hundredths of a percent back).  

[dpmapper]

You must be logged in to read this quote.

It is necessary.  First, it's necessary for Dingell to take the Pointes, lest the minority percentages drop too much.  Second, it's necessary because there isn't enough room in the two black districts to take all of Detroit anyhow.  Finally, and probably most importantly, it's necessary because there are 4 districts in Wayne, so there must be 3 splits somewhere.  My map puts two splits in Detroit and one in Westland; does that differ any from having one split in Dearborn, one in Westland, and one in Detroit? 

[dpmapper]

According to the legislation that enacted the 2001 map, there were 11 county splits. Here's the language from within that statute:

You must be logged in to read this quote.

Counting county splits in that map according to the above rules I get single splits in Bay, Saginaw, Kent, Allegan, Shiawassee, Calhoun, Washtenaw, and Macomb. That leaves three splits to get to 11, so the interpretation is that the current map counts two splits in Oakland and one in Wayne. Thus the the whole districts in those counties do not count as splits.

I'm not saying you're wrong, but this seems like an odd way to count splits.  If there are 3 counties in a state that go in a line, A next to B next to C, with populations 20-110-20 in that order, and there are to be three districts, you can do it two ways:

PLAN 1: 20 from A, 30 from B; 50 from B; 30 from B, 20 from C
PLAN 2: two districts entirely within B, one containing A, C, and a 10-person bridge from B

Are you telling me that PLAN 2 has no splits, whereas PLAN 1 does?  My method of counting splits makes much more sense.  

ETA: I should add that you can put the light blue district entirely within Oakland if you want, and give Rogers the two thumb counties in exchange, with little change in partisan balance (put Pontiac+W. Bloomfield in Rogers's district).  But it makes for an uglier map, in my opinion - a district from Lansing to Lake Huron? 

You must be logged in to read this quote.

Heh.  All our work for nothing, if so.    (But if you can get close to optimal without having to litigate this matter, so much the better, right?)  

[dpmapper]

OK, I guess you're saying that
"(i) Breaking a county line means assigning part of the population of a county to 1 or more counties in the formation of a district."
means, putting part of a county with a district that contains other counties.  (I couldn't figure out what it meant to "assign part of a county to other counties"...)

But if that's so, doesn't Oakland currently have 3 breaks, and Wayne 2?  

I still don't know what "including a district from 2 geographically-separate areas" means, though.  

I also don't understand how dividing Wayne into 2 whole districts plus two partial districts can constitute the same number of splits (1) as dividing Wayne into 2 whole districts plus one partial would. 

[dpmapper]

What I'm saying is that after reading the text that was part of the passed map, I looked at the splits and compared it to the reported number. From that comparison I can deduce the algorithm they used. I'll summarize how it apparently worked in 2001.

A county that is split in two counted as one split.

A county that is split in three would have counted as two splits. Think of the county as having been split twice from its original intact form.

A county split into n pieces would have counted as n-1 splits.

A county with one or more districts wholly contained and one other piece attached to other counties would have counted as one split. The lesser part crossed the county line to form that one split.

A county with one or more districts wholly contained and two other pieces attached to other districts counted as one split. This is somewhat counter-intuitive, but it compares to the case where a single county is split in two. The remainder fragment is treated as a county for splits, except that it would be one split even by itself.

A county with one or more districts wholly contained and three other pieces attached to other districts counted as two splits. This parallels the three-split case for a single county.

In general a county with N districts wholly contained and n pieces attached to other districts outside the county would have counted as n-1 splits, except when n = 1 and then it counts as 1 split.

Well, I guess I believe you.  But if so, that's a ridiculously stupid way of counting.  

Consider the scenario where one big county A (pop. 63) in the middle is attached to 3 smaller counties B/C/D (20 each) which are not adjacent to one another.  Dividing into three districts of 21-20 seems reasonable, but that has two splits.  

A different plan would have one whole district in A, 21 from A + all of B, and a district of D+C+a bridge via A.  This technically has one split, but I can't see any reason why this is preferable.  

Here's a different scenario.  Consider the following plans:

Plan 1:
51 from A;
10 from A, 20 from B, 21 from C;
10 from A, 20 from D, 21 from C.  

Plan 2:
51 from A;
20 from A, 20 from B, 11 from C;
20 from D, 31 from C.  

Which is superior?  Clearly plan 2 is, but according to the formula you just laid out, both have exactly 2 splits.  

So the counting formula mandates an inferior (or at least, non-superior) plan in my first scenario, and fails to distinguish a superior plan in my second scenario.  Boo!

****************

Finally, what is the minimum number of splits according to the formula?  Ignoring VRA for a moment, I think the best you can do is the following:
2 whole districts in Wayne
1 whole in Oakland
1 whole in Macomb
1 with the leftovers from Wayne, Macomb, and some of the leftovers from Oakland

... leaving 9 to go, which will split 8 other counties between them.  Add in 1 split in Macomb, Oakland, and Wayne, and we're at 11.  Am I missing something?  

Let's say I shove the light blue district entirely into Oakland in my scenario.  Then Oakland has one split, Macomb has 1 split, and Wayne 2.  Washtenaw, Kent, Calhoun, Ingham, Saginaw, Isabella, and one between CD-1 and CD-4 are the others.  That's 11!  Ding ding ding, we have a winner!

[dpmapper]

I get that the lowest theoretical count comes from putting as many whole districts into counties that one can, and then use at least a two-way split for the remainder of those counties. In this case 4 districts would be entirely within one county, leaving 10 CDs. Those 10 require 9 county splits (n-1) at a minimum so 9 splits is the ideal.

OK, that seems right.  But it would require two separate districts having their border within the 20K or so extra people left out of the whole-Macomb district... and those two districts couldn't be in Oakland or Wayne!  You'd have to come in from the north.  Ai-yai-yai. 

In any case, do you agree that this method for counting splits is completely whacked? 

[I'd also point out that the theoretical minimum is 10 for the current 15 districts and it is not hit.  So who knows what is actually required?]

[dpmapper]

I've calculated that the district I've given McCotter gave Bush just over 53% of the 2-party vote in 2004.  He might be able to up that a tad by picking the right precincts to excise from Van Buren Township.  This might be half a point better than his current district, but I'm guessing the blue-collar whites in Monroe and southern Wayne are trending more GOP than the country as a whole is. 

Incidentally, does anyone know if there's a hippie commune in Van Buren Township or something?  One precinct (#3) apparently gave the Green Party 497 votes in 2004 (out of 1367 cast).  Is this just a typo?  (Bush got 349, Kerry 508, so 1367 does at least match the sum, and it is about the same number of votes as other precincts.) 

[dpmapper]

I redid the non-Detroit parts a bit:



I used CD-03 to give me another county split involving the Flint district, rather than soak up Battle Creek/Albion.  This lets Flint take Saginaw.  So CD-07 no longer looks like a game of Jenga, and no longer has to run into Saginaw, but it does have to move about 20K farther in towards Ann Arbor (there are some 65%+ Kerry precincts on the outskirts) and take back Battle Creek/Albion (and Mark Schauer).  It does get heavily GOP Barry County in the bargain; I haven't totalled everything but I think this makes it almost 55% McCainBush, compared to 53.7% previously.  

I think Camp will be a bit safer as well - his district goes up to southern Grand Traverse County and takes in less of Saginaw's Dem suburbs than it used to.  

The CD-08/CD-09 combo that splits Oakland takes in a bit more of the Lansing area, but is compensated by the black AnnArbor/Oakland/West Detroit district eating up Ferndale and a bit more of Farmington Hills.  One will have to juggle the split of Oakland carefully to make sure both districts are reasonably solid; I haven't done the number crunching yet.  

[In all of this, I'm assuming that the county splits metric is changed to something more sensible.]

[dpmapper]

You seem to have a spare Okemos precinct in the Lansing/Flint district.  Have you split it Okemos/East Lansing, then?  That would be... strange, but Okemos would probably like it.

Yeah, I know.  It's a precinct that has two disjoint parts, one of which is part of the East Lansing block.  I don't know what the rule is about those. 

[dpmapper]

That is a really hideous looking map! 

Um... thanks?