Poll re 2020 Ohio CD Map

[Torie]

To try to allay partisan concerns, in map 3 the GOP PVI for OH-02 is 2.2% based on 2008 figures, probably 1.0% or less now, and should be tilt to lean Dem by 2022 (it includes all of the city of Cincinnati), while in Map 1 and 2, for OH-01 the GOP PVI for the Cincinnati CD is minus 0.8%, and will no doubt be lean Dem in 2020, so the difference is not huge, and the partisan effects will be muted. The population projections have been refined having eye balled the 2020 census projections for the counties. Medina is a growth node by the way. Who knew?

Map 1

Map 2

Map 3
 

[Benj]

Map 1 is pretty easy. I'm not sure what you're trying to achieve with Map 2 that isn't achieved with Map 1, but the small towns-to-suburbs district along the Ohio River is bad. Map 3 is a mess in the Cincinnati area (connecting the city of Cincinnati to rural small towns while splitting it from its heavily minority suburbs? Really? Don't try to argue that's not partisanly motivated).

[Torie]

The maps speak for themselves. The factors to evaluate are keeping metro areas together, and minimizing county cuts and erosity.  Map I was criticized because the south Columbus CD went "all the way" to the Ohio River.  It doesn't particular bother me that much, but it bothers others. And in that map the Cincinnati suburban CD is kind of awkward, while in the other maps, there is more compactness. Also, in Map 3 makes OH-05 the most compact, and it includes quasi exurban, and becoming more so over time, Madison County, so all the Columbus metro area in in 3 CD's. So chopping Hamilton County has its offsetting benefits.

I gave the partisan stats. Yes in one the GOP numbers are better for Cincinnati, but by 2022, the Dems will probably win it either way. Bear in mind that the city of Cincinnati itself should trend heavily Dem as it gets more black, and has more singles and so forth which attends that. What was deleted from the CD in Hamilton County is pretty heavily Pub overall, just not as Pub as the average of Clemont and the 3 river counties.

[traininthedistance]

I'm not a huge fan of the northeastern corner you have for all three maps.  Districts 10 and 11 seem reasonable, but 12, 14, and 15 strike me as a very poor representation of how that region naturally divides.  In particular, I would lobby for an Akron-Canton district, instead of your District 14, which just screams gerrymander.

I think what I'd do is:

* Pull 15 out of Summit, and let it drop down closer to Youngstown and Cleveland (pushing 11 west, oh well)
* Have 14 withdraw from Akron and go south instead, possibly even eating into the northern portion of 13, which seems to me a better cultural fit.
* 12 becomes Summit, Stark, and whatver leftovers are necessary
* 10 can go south to Wayne, then, or further west pushing on 9.

I'd also consider a Dayton-Springfield district in the southwest.

[Torie]

The above sounds like a Dem gerrymander. Putting aside combining Akron and Canton (but then the Youngstown-Warren CD gets very erose as it will have to wander way down the Ohio River without the 243,000 folks it garners Summit County), there is nowhere for OH-15 to go without slamming into Warren, or diluting the black percentage in OH-11. Plus Akron and Youngstown/Warren have a lot in common, more so than with Canton, although Canton is closer. Combining Clark and Montgomery is what the Dem's would want to do in a Dem  gerrymander of course, but no map would support that combo that I can think of beyond Dem gerrymandering interests.  The population numbers just don't play out that way, without generating erosity. Concern about limiting erosity and county splits is what keeps one honest. Other than that, it's gamesmanship time, as each side tries to put lipstick on its favorite gerrymandering pigs.

If you draw your map, you will see what I mean.

[Miles]

I'm by no means an authority on OH, but I like Map I.

[traininthedistance]

I see what you're saying about CD-15, Torie, but

Plus Akron and Youngstown/Warren have a lot in common, more so than with Canton

is just totally, completely, in every way wrong.  Akron and Canton have a lot more in common with each other.  Take a look at the urbanization patterns, just for starters: Cleveland-Akron-Canton is basically one continuous conurbation, whereas there is a definite gap between them and Youngstown.  The Census even recognizes Cleveland-Akron as one CSA; they don't add in Canton but maybe they should.

And, sure, my proposal would likely be a tiny bit more pro-Dem, but it wouldn't be universally so: CD-10 would go from marginal to lean Pub as compensation.

[Torie]

I have refined the maps with better population estimates, and have a new idea about the NE corner conundrum as an alternative to the conjoining of the uber rust belt cities. In due course, I will delete this poll, and put up another which will include the new NE corner alternative. Stay tuned.

[Benj]

The maps speak for themselves. The factors to evaluate are keeping metro areas together, and minimizing county cuts and erosity.  Map I was criticized because the south Columbus CD went "all the way" to the Ohio River.  It doesn't particular bother me that much, but it bothers others. And in that map the Cincinnati suburban CD is kind of awkward, while in the other maps, there is more compactness. Also, in Map 3 makes OH-05 the most compact, and it includes quasi exurban, and becoming more so over time, Madison County, so all the Columbus metro area in in 3 CD's. So chopping Hamilton County has its offsetting benefits.

I gave the partisan stats. Yes in one the GOP numbers are better for Cincinnati, but by 2022, the Dems will probably win it either way. Bear in mind that the city of Cincinnati itself should trend heavily Dem as it gets more black, and has more singles and so forth which attends that. What was deleted from the CD in Hamilton County is pretty heavily Pub overall, just not as Pub as the average of Clemont and the 3 river counties.

"It might vote Democratic 10 years from now" is your best defense to blatantly ripping apart communities of interest in the Cincinnati area? Frankly, the partisanship is even beside the point--you have provided no legitimate justification whatsoever for why Cincinnati should be combined with rural areas a hundred miles away with which it has literally nothing in common yet not be combined with its demographically similar suburbs that it directly abuts. Your proposal 3 is nothing but the worst of gerrymandering, and your attempts to justify it are exactly the sort of flimsy excuses provided by expert gerrymanderers like the Michigan GOP.

There are no offsetting benefits whatsoever. All of your district 5s are just a collection of unrelated rural towns plus some Columbus suburbs. Adding or subtracting some rural areas is pretty meaningless. If areas like Portsmouth were to go with some other district, it would be district 13--though stretching that further along the Ohio River has (not necessarily negative) knock-on effects elsewhere. Including Madison County or not is irrelevant. You would sacrifice the literally hundreds of thousands of connections in the Cincinnati area for a handful of exurban commuters near Columbus?

[JerryArkansas]

I like map 2.

[Torie]

Again guys, a couple of posters disliked the south Columbus CD going to the Ohio River. So I prepared a  couple of maps where it didn't go to the river. Sure Hamilton County in one map is split, but both the one without the city of Cinci is a suburban CD, and it reduces erosity by a bit over the one where Hamilton County is kept whole, and the suburban CD takes the river counties plus Warren County and a bit of Butler. My first map which has the south Columbus CD going to the river has the least erosity of all arguably (and is the one that I prefer pending revisions to reflect better population estimates), but whatever. The PVI swing one way or the other is just a couple of points, FWIW.

[CatoMinor]

In my, non-expert of Ohio opinion, map 2 is the finest of the 3. It appears the only substantial difference in these maps is that of the greater Cincinnati area. I am never a fan of breaking up cities and counties when they do not have to be, which would narrow it down to maps 1 and 2. The first map has more aesthetic appeal, but Hamilton and Dayton just seem to belong together, and who are we to break up true love? So map 2.

[muon2]

I'm not a huge fan of the northeastern corner you have for all three maps.  Districts 10 and 11 seem reasonable, but 12, 14, and 15 strike me as a very poor representation of how that region naturally divides.  In particular, I would lobby for an Akron-Canton district, instead of your District 14, which just screams gerrymander.

I think what I'd do is:

* Pull 15 out of Summit, and let it drop down closer to Youngstown and Cleveland (pushing 11 west, oh well)
* Have 14 withdraw from Akron and go south instead, possibly even eating into the northern portion of 13, which seems to me a better cultural fit.
* 12 becomes Summit, Stark, and whatver leftovers are necessary
* 10 can go south to Wayne, then, or further west pushing on 9.

I'd also consider a Dayton-Springfield district in the southwest.

The above sounds like a Dem gerrymander. Putting aside combining Akron and Canton (but then the Youngstown-Warren CD gets very erose as it will have to wander way down the Ohio River without the 243,000 folks it garners Summit County), there is nowhere for OH-15 to go without slamming into Warren, or diluting the black percentage in OH-11. Plus Akron and Youngstown/Warren have a lot in common, more so than with Canton, although Canton is closer. Combining Clark and Montgomery is what the Dem's would want to do in a Dem  gerrymander of course, but no map would support that combo that I can think of beyond Dem gerrymandering interests.  The population numbers just don't play out that way, without generating erosity. Concern about limiting erosity and county splits is what keeps one honest. Other than that, it's gamesmanship time, as each side tries to put lipstick on its favorite gerrymandering pigs.

If you draw your map, you will see what I mean.

I also am not a fan of the NE corner. It looks too much like a Pub gerrymander. I decided to push county splits and erosity down to the point that only two counties are split other than the big three. All the districts get to within 1% of the projected ideal. I also avoid putting the southern district into either the Cinci or Columbus suburbs, which could answer all those concerns. I did split Cleve but it was to make a clean NE corner with a relatively low erosity and only a single split of Summit beyond the needed split of Cuyahoga.

I didn't use any partisan data to draw it, but looking at the DRA '08 numbers there are 5 districts with Obama ran 4% or more over his national numbers and 6 districts where Obama ran 4% or more under his national numbers. If I get a chance I'll look at the 2012 numbers for a better read on the partisan breakdown.

[traininthedistance]

By and large, I think the best attempt I've seen of the recent maps is jimrtex's one:

https://uselectionatlas.org/FORUM/index.php?topic=124180.msg3722805#msg3722805

Breaking the state into regions (Cincy/Dayton, Columbus, Toledo, NE, and the rural bits) and then balancing by population strikes me as the right approach; and both the regions he came up with, as well as the subdivisons of those regions, are pretty good.

I still don't like that Springfield gets stranded from Dayton/the southwest and stuck in with a Columbus outskirts district, but the math is pretty compelling that something in that area has to lose out, so I accept it.

The one modification I'd make to jimrtex's map is in the northeast: not the boundaries of the northeast region itself, but the division within it.  I think that, when you're looking at urban areas, you need to set aside a fetish for county boundaries, and look at urbanization patterns, municipal boundaries, and other such subregions.  And, while this has led me to advocate for an Akron/Canton district, there's actually one part of that map where such an approach is even more urgent: Youngstown/Warren.  So, with that in mind, here's roughly what I'd do in the Northeast:



I did not look at partisan data at all; I did try to make sure the Cleveland district was more overpopulated than the other four as well as less than 50 percent white, but that's it.  Besides that, I went by regions: Mahoning and Ohio Valleys, Akron/Canton, the Cleveland core, the western suburbs.  Ashtabula/eastern and southern suburbs is a bit of a hodgepodge, yes, but there is absolutely no way around that (and, yes, sending that district down to Warren would make it more of a hodgepodge from CoI standards).

[muon2]

By and large, I think the best attempt I've seen of the recent maps is jimrtex's one:

https://uselectionatlas.org/FORUM/index.php?topic=124180.msg3722805#msg3722805

Breaking the state into regions (Cincy/Dayton, Columbus, Toledo, NE, and the rural bits) and then balancing by population strikes me as the right approach; and both the regions he came up with, as well as the subdivisons of those regions, are pretty good.

I still don't like that Springfield gets stranded from Dayton/the southwest and stuck in with a Columbus outskirts district, but the math is pretty compelling that something in that area has to lose out, so I accept it.

I agree, Montgomery+Greene make the best CoI, and they are too big to add Clark as well. One doesn't have to put Clark(Springfield) with the Columbus burbs, however. I think it's worth considering Springfield with the other small cities in western OH like Lima and Findlay. It is tempting to put Montgomery with Warren like in map 3, just because they are likely to be just the right size for a district. But that tends to make more of a mess with the other Cinci/Dayton districts.

One other thing I tried to avoid was to use a split county piece to connect two whole counties. This came out of the extensive CA discussions, and it seems like a good rule that can suppress certain types of gerrymandering. It does prevent districts that link suburban counties on opposite sides of the central city county through a bridge that wraps around the city.

You must be logged in to read this quote.

I like the fact that you keep Trumbull and Mahoning together. They are a natural CoI like Montgomery/Greene. I disagree that Ashtabula with Trumbull is any worse than with Lake. Lake is clearly part of the Cleveland suburban ring. Ashtabula is a largely rural county and is not in either the Cleveland or Youngstown metro area by census definition (Astabula is its own micropolitan area). I don't see any good reason to make a district erose or to split the county just to keep part of Ashtabula with the Cleveland metro.

[traininthedistance]

I like the fact that you keep Trumbull and Mahoning together. They are a natural CoI like Montgomery/Greene. I disagree that Ashtabula with Trumbull is any worse than with Lake. Lake is clearly part of the Cleveland suburban ring. Ashtabula is a largely rural county and is not in either the Cleveland or Youngstown metro area by census definition (Astabula is its own micropolitan area). I don't see any good reason to make a district erose or to split the county just to keep part of Ashtabula with the Cleveland metro.

I definitely agree that Ashtabula is its own thing that doesn't really fit perfectly anywhere.  It could just as easily be a part of the Trumbull/Mahoning district as with the eastern Cleveland burbs district, depending on how the population math shakes out by 2020.

[Benj]

I like that map a lot and had begun drafting something very similar myself. I agree that Ashtabula County basically goes where you need it. It has little in common with any other county. Draw the map and put Ashtabula wherever you need to based on how the rest of NE Ohio shakes out. Splitting Ashtabula to balance population isn't really a problem, though; southern Ashtabula is very rural and has nothing in common with the areas along the lake, really. It's important to recognize that county lines poorly reflect communities of interest.

[muon2]

I like that map a lot and had begun drafting something very similar myself. I agree that Ashtabula County basically goes where you need it. It has little in common with any other county. Draw the map and put Ashtabula wherever you need to based on how the rest of NE Ohio shakes out. Splitting Ashtabula to balance population isn't really a problem, though; southern Ashtabula is very rural and has nothing in common with the areas along the lake, really. It's important to recognize that county lines poorly reflect communities of interest.

It may be true that county lines don't reflect real CoI's but they are a well-established proxy for a CoI. Even the Census uses them as building blocks to define metropolitan areas. Using them as a proxy is important since it provides an objective criterion for judging a map.

[Benj]

I like that map a lot and had begun drafting something very similar myself. I agree that Ashtabula County basically goes where you need it. It has little in common with any other county. Draw the map and put Ashtabula wherever you need to based on how the rest of NE Ohio shakes out. Splitting Ashtabula to balance population isn't really a problem, though; southern Ashtabula is very rural and has nothing in common with the areas along the lake, really. It's important to recognize that county lines poorly reflect communities of interest.

It may be true that county lines don't reflect real CoI's but they are a well-established proxy for a CoI. Even the Census uses them as building blocks to define metropolitan areas. Using them as a proxy is important since it provides an objective criterion for judging a map.

Not really. County lines are a crutch, not an objective criterion. They're for people who aren't willing to be intellectually serious enough to actually try to examine the settlement patterns of a given area. Maybe courts are lazy, but we should demand better.

[traininthedistance]

I like that map a lot and had begun drafting something very similar myself. I agree that Ashtabula County basically goes where you need it. It has little in common with any other county. Draw the map and put Ashtabula wherever you need to based on how the rest of NE Ohio shakes out. Splitting Ashtabula to balance population isn't really a problem, though; southern Ashtabula is very rural and has nothing in common with the areas along the lake, really. It's important to recognize that county lines poorly reflect communities of interest.

It may be true that county lines don't reflect real CoI's but they are a well-established proxy for a CoI. Even the Census uses them as building blocks to define metropolitan areas. Using them as a proxy is important since it provides an objective criterion for judging a map.

Not really. County lines are a crutch, not an objective criterion. They're for people who aren't willing to be intellectually serious enough to actually try to examine the settlement patterns of a given area. Maybe courts are lazy, but we should demand better.

In rural areas, county lines are actually meaningful (since people out there identify more strongly with their county government) and should be given deference when possible.  In metro areas, not so much.  I would love to see the Census try to expand their NECTA areas (or a similar municipality-based metro area measurement) to other parts of the country.  They certainly have the ability to do so in the Mid-Atlantic and Midwest, at the very least.

[muon2]

I like that map a lot and had begun drafting something very similar myself. I agree that Ashtabula County basically goes where you need it. It has little in common with any other county. Draw the map and put Ashtabula wherever you need to based on how the rest of NE Ohio shakes out. Splitting Ashtabula to balance population isn't really a problem, though; southern Ashtabula is very rural and has nothing in common with the areas along the lake, really. It's important to recognize that county lines poorly reflect communities of interest.

It may be true that county lines don't reflect real CoI's but they are a well-established proxy for a CoI. Even the Census uses them as building blocks to define metropolitan areas. Using them as a proxy is important since it provides an objective criterion for judging a map.

Not really. County lines are a crutch, not an objective criterion. They're for people who aren't willing to be intellectually serious enough to actually try to examine the settlement patterns of a given area. Maybe courts are lazy, but we should demand better.

Settlement patterns are hardly objective. One expert will cite a density measure while another may prefer socioeconomic data. In court they can battle back and forth, but it's unlikely that they will undermine a politically gerrymandered map.

County lines are recognized by state and federal courts as objective, so much so that the consistent use of whole counties can forestall other legal challenges to a map. This was the case most recently in WV. There are no competing experts when it comes to use of counties which makes them objective. Adherence to strictly measurable standards works against political gerrymandering since the most successful gerrymanders need a free hand to draw the map.

I'm not sure why you think holding to whole counties is so easy. Given the need for equal population, avoiding county splits makes the problem much harder than if one can split counties without limit. It can be shown mathematically that with a very large set of blocks or precincts it is much easier to reach a particular population standard than with a relatively small number of counties.

[TJ in Oregon]

My preference of the maps is 3>1>2.

The main problem with the first is the Columbus-to-Ohio River CD that doesn't really gel, a nice rurban thing spanning all divisions of media marker, culture, and COI.

The third one has the flaw of cutting Cincinnati in half, which is pretty bad in itself, although Cincinnati clearly does warrant multiple districts, so it isn't that horrible, at least from a COI standpoint. If we weren't putting aside partisan considerations, this would be a little worse since it clearly favors the Republicans. But we aren't considering that.

I like the second the least and the reason for that is the treatment of the Dayton district. The section of Butler County it gets contains far more areas that would identify with Cincinnati, not Dayton.

[TJ in Oregon]

The county lines in NE OH did a great job of respecting communities of interest a hundred years ago when they were drawn. Suburban growth has changed things substantially in a number of places, perhaps the most notable in Ohio being the intersection of Cuyahoga, Geauga, Summit, and Portage. Solon, Twinsburg, Aurora, and Bainbridge Townships all have a ton more in common with each other than any does with Cleveland, Akron, Chardon, or Ravenna.

That being said, when a court is considering maps from a COI argument, county lines are a reasonable standard to use, simply because they are objective. County lines are not subject to argument, like say whether Twinsburg is more a Cleveland suburb or an Akron suburb.

[Torie]

Well, my work is done I think. Below I believe is the best possible map that can be drawn. Well, one possible alternative is to have OH-07 kick OH-05 out of Franklin County entirely. I will try that later to see how the erosity looks.

I consider keeping metro areas together as the most compelling factor, and then after that minimizing erosity, although some additional erosity is OK if it reduces county splits (and secondarily the size of the split). Minimizing erosity is a good discipline. As you draw the lines say in the county chop, you always try to minimize erosity except where it involves a locality split, and that you might tolerate a tad more. Ditto when you add counties.  You add the counties that minimize erosity, where you can.

Then I minimize county splits, and then try to minimize the size of the split (keep as high a percentage of a county in one CD as opposed to the other, as you can, but I accept a sizable split if necessary to hew to more important factors. Thus Allen County (Lima) has a sizable chop, as does Fairfield County (and Tuscarawas too), but that is necessary to keep erosity down, and thus acceptable. I also try to avoid a rural county being an after thought addition to urban counties where I can, without adding a county chop, which was one reason, but not the only reason, that I removed Preble County from the Dayton CD. (In that regard, I was pleased to kick Knox County out of OH-07.) Then I try to minimize local splits (and will accept some additional intra-county erosity where counties are split to avoid or minimize locality splits - thus the lines are as they are at the north end of OH-12 in Summit County). I think the map below follows this formula. This formula works well in states like Ohio that don't have mountains and/or huge rivers around that can dictate more erosity than would otherwise be appropriate.

Oh, and in drawing the Cleveland CD, I tried to keep the black percentage up as much as possible, without getting ridiculous about it.

It was an interesting exercise. I measure erosity by my eye, and just "know" when it is too much, but it would be good to have an actual formula. I played and played with OH-04 and OH-08 to get the erosity down (it was tough), and finally decided upon the Allen County chop to get it done. You could have a much smaller chop up in Paulding County, but it just makes both CD's too erose, particularly OH-04, which then goes from Pauling along the Indiana line on the north end, down to the north line of Butler county to the south, and that is just too much. No bueno. I then twisted a couple of wheels to get rid of 3 micro chops as the final icing on the cake.

Finally the map is based on the census projections as to the population in 2020. Any deviations from those numbers should not be more than 5K per district, although the allocation of population in Cuyahoga was a bit of a guesstimate. I found the population projection for the City of Columbus, so that one was easy (it goes up 9%, while the whole county goes up 11%, so that means the burbs in Franklin go up by about 15%, and the map reflects that.

And now - the map!  Yes, adding Ashtabula to the Youngstown/Warren CD was the rabbit out of the hat, that made the Akron/Summit County/erosity/county tri-chop issue largely go away. What do you all think? No TJ, appending the Ohio River Counties to the Cinci burbs is just too costly. Among other things, it messes up the Dayton CD, as well as increases erosity. It also means three rural Counties become dominated as an afterthought in an urban CD, which violates one of the parameters outlined above. It simply is a dog in short which won't hunt. Sorry.

If we all agree that the map is inspired genius (putting aside the Franklin County alternative possibility discussed above), then I will put up the partisan stats. Thank you.

[Benj]

Why not put Lorain and Elyria in OH-10 and Wood County (and maybe some of Seneca and/or Huron for population) in OH-09 (and rearrange OH-08 to fit)? Probably requires a few other counties moved around, but there are Toledo suburbs in Wood County that belong with Lucas, and Lorain and Elyria, while not exactly Cleveland suburbs themselves, have much closer ties with the Cleveland suburbs than with Toledo (and certainly much closer ties with the Cleveland suburbs than Mount Vernon does!). Might need to rearrange the OH-08/OH-04 boundary a bit for neatness, too, but that border is pretty arbitrary.

Also, I guess you can't fit all of New Philadelphia in the Akron-Canton district? Because that would be neater than going into Carroll County.