US House Redistricting: Ohio

[traininthedistance]

So here's my analysis of the competition quality of train's map. I've used his Obama number with the likely 3rd party vote to get an R fraction then adjusted it using the table above. Note the competition is about 1.5% more R on average than an equivalent PVI.

CD 1: 47.2% Likely D
CD 2: 62.6% Safe R
CD 3: 51.4% Lean R
CD 4: 65.7% Safe R
CD 5: 53.0% Likely R
CD 6: 54.1% Likely R
CD 7: 55.5% Strong R
CD 8: 68.2% Safe R
CD 9: 44.7% Strong D
CD 10: 47.2% Likely D
CD 11: 23.1% Safe D
CD 12: 63.7% Safe R
CD 13: 50.8% Tossup
CD 14: 48.1% Lean D
CD 15: 40.5% Safe D
CD 16: 47.0% Likely D

Competition scores (with high score for each category)
Fairness leans 3.4% more R than the state as a whole: 86.4 points (top score 99.6)
Competitiveness 3 highly competitive, 5 competitive, 2 somewhat competitive: 21 points (top score 33)
County splits 18: 32 points (top score 43)
Compactness: not scored

If I shifted them by 1.5% to get a PVI, then the fairness zooms to 99.2 points and the competitiveness rises to 22 points. The choice of election data really matters in assessing fairness (cf AZ).

Good to know.  FWIW, I don't think that it's actually good policy to artificially make districts as competitive as possible if other compelling factors lead to a certain number of non-competitive districts*. However, I could easily imagine setting the boundaries between 7 and 15 to make two lean-D districts instead of a safe D and a likely R; and there's always the option of giving up on the VRA in the northeast corner, which leads to all sorts of ripple effects. 

*I do think it's best to have some competitive districts in most states, though.  There are very few multi-district states where a truly fair districting would make all districts noncompetitive- the only examples I can think of are either small and heavily dominated by one party (Hawaii, Idaho), or feature polarized voting in the Deep South (I don't think anyone can seriously argue against a more-or-less guaranteed 3-1 delegation from Mississippi).

[muon2]

So here's my analysis of the competition quality of train's map. I've used his Obama number with the likely 3rd party vote to get an R fraction then adjusted it using the table above. Note the competition is about 1.5% more R on average than an equivalent PVI.

CD 1: 47.2% Likely D
CD 2: 62.6% Safe R
CD 3: 51.4% Lean R
CD 4: 65.7% Safe R
CD 5: 53.0% Likely R
CD 6: 54.1% Likely R
CD 7: 55.5% Strong R
CD 8: 68.2% Safe R
CD 9: 44.7% Strong D
CD 10: 47.2% Likely D
CD 11: 23.1% Safe D
CD 12: 63.7% Safe R
CD 13: 50.8% Tossup
CD 14: 48.1% Lean D
CD 15: 40.5% Safe D
CD 16: 47.0% Likely D

Competition scores (with high score for each category)
Fairness leans 3.4% more R than the state as a whole: 86.4 points (top score 99.6)
Competitiveness 3 highly competitive, 5 competitive, 2 somewhat competitive: 21 points (top score 33)
County splits 18: 32 points (top score 43)
Compactness: not scored

If I shifted them by 1.5% to get a PVI, then the fairness zooms to 99.2 points and the competitiveness rises to 22 points. The choice of election data really matters in assessing fairness (cf AZ).

Good to know.  FWIW, I don't think that it's actually good policy to artificially make districts as competitive as possible if other compelling factors lead to a certain number of non-competitive districts*. However, I could easily imagine setting the boundaries between 7 and 15 to make two lean-D districts instead of a safe D and a likely R; and there's always the option of giving up on the VRA in the northeast corner, which leads to all sorts of ripple effects. 

*I do think it's best to have some competitive districts in most states, though.  There are very few multi-district states where a truly fair districting would make all districts noncompetitive- the only examples I can think of are either small and heavily dominated by one party (Hawaii, Idaho), or feature polarized voting in the Deep South (I don't think anyone can seriously argue against a more-or-less guaranteed 3-1 delegation from Mississippi).

If you look at the entire Congress about 1/4 of the districts qualify as lean or likely. Rather than maximize competitive districts one can set a floor at 1/4 with the stipulation that they be evenly divided between the parties. Your plan easily exceeded that.

I would note that the good-government groups in OH are very, very big on competitiveness. In 2005 they had an amendment on the ballot that would have made competitiveness essentially the only criterion.

[they don't love you like i love you]

So I ended up redrawing all of Ohio's legislature for both houses. I started for basically two reasons, to get a feel for how the layout the state skews the partisan numbers, and to experiment with grouping three House seats into one Senate. The latter is actually a tad tricky, since you might have an area that's an obvious community of interest and has seven districts, but one has to be left out, and you end up with some awkward Senate districts. But those will come later. For now: House.



DISTRICT 1: WESTERN A CERTAIN OHIO COUNTY THAT IS NOTHING LIKE BAKERSFIELD, CALIFORNIA. O 26.2%.  96W.  This is actually the most conservative district in the state and the only one where McCain broke 70%. Most likely the same for Romney. Not much to see, Safe R.

DISTRICT 2: WEST CINCINNATI O 59.3%.  61.5W/33.1B. While more conservative than the rest of Cincinnati, this would take a severe drop in black turnout to ever be truly competitive. Likely D.

DISTRICT 3: CENTRAL CINCINNATI: O 85.2%. 50.4B/42.7W. Black majority and just barely in VAP. Obviously Safe D. Even the whites in this district seem to be pretty D.

DISTRICT 4: NORTH CINCINNATI: O 58.7%. 64W/31.6B. This is rather similar to 2 demographically, basically the same, Likely D. A bit more winnable of course.

DISTRICT 5: THE NORTH CENTRAL OF THIS CERTAIN COUNTY:  O 54.1%. 65.5W/26.4B. Probably Lean D. The demographics make it far more winnable in midterms.

DISTRICT 6: EAST CINCINNATI: O 44.6%. 90.6W/4.1B. While this is not as conservative as the demographics would imply in this area, it's still not winnable for any Democrat. Safe R.

DISTRICT 7: THE OTHER MIAMI AND INDIAN HILL: O 46.7%. 78.1W/14.8B. Kind of the same as above. While it's not overwhelmingly conservative, Democrats aren't winning any McCain seat in this region. Safe R.

DISTRICT 8: SOUTH CLERMONT: O 34.1%. 95.7W. Predictable district. Safe R.

DISTRICT 9: NORTH CLERMONT-BROWN: O 33.8%. 96.7 W. Another predictable one. Safe R.

DISTRICT 10: WEST BUTLER: O 40.9%. 89.7W/4.6B. While not as homogenous as you'd expect, this is still a Safe R seat.

DISTRICT 11: SOUTHEAST BUTLER: O 36.8%. 80.7W/8.8B/4.9H/4.3A. Despite the relative diversity, this is another Safe R seat.

DISTRICT 12: EAST CENTRAL BUTLER: O 36.9%. 87.8W/7B. Another generic Safe R seat.

DISTRICT 13: MASON AND LEBANON: O 33.4%. 88W/5.4A. Now it's easy to see where bandit's political outlook comes from. Safe R.

DISTRICT 14: REST OF WARREN AND BUTLER: O 30.2%. 93W/4B. I think this might be the second most R seat in the state. Safe R.

DISTRICT 15: PREBLE AND WEST MONTGOMERY: O 36.6%. 92.8W/4.4B. As expected, Safe R.

DISTRICT 16: WEST DAYTON: O 85.5%. 67B/29.5W. The black part of Dayton, and an effective Dem pack. Safe D.

DISTRICT 17: SOUTH MONTGOMERY: O 38.9%. 88.4W/4.9B. Another one of those generic Safe R seats.

DISTRICT 18: SOUTHEAST DAYTON AND KETTERING: O 47.4%. 90.7W. Not an overwhelmingly conservative seat, but Obama did outperform the generic D numbers here. It's probably at best Likely R.

DISTRICT 19: NORTHEAST DAYTON, VANDALIA AND HUBER HEIGHTS: O 47.2%. 82.6W/10.6B. Same as above basically....except the generic D and R numbers are pretty close here. This might be a bit more competitive even if McCain's numbers are slightly better. Lean R.

DISTRICT 20: MIAMI AND SOUTH SHELBY: O 33.7%. 95.3W. Very boring and homogenous rural seat. Safe R.

And more to come. So far though this is 15R-5D seats.

[jimrtex]

So I ended up redrawing all of Ohio's legislature for both houses. I started for basically two reasons, to get a feel for how the layout the state skews the partisan numbers, and to experiment with grouping three House seats into one Senate. The latter is actually a tad tricky, since you might have an area that's an obvious community of interest and has seven districts, but one has to be left out, and you end up with some awkward Senate districts. But those will come later. For now: House.

Just because Ohio doesn't follow their constitution is no reason for you to not do so.

[they don't love you like i love you]

I'm not familiar with the constitution of Ohio. I only looked for general CoI and population equality. That said though there are some districts I hate and had no choice but to draw this way, there are worse ones in the actual map.

More:



DISTRICT 21: SPRINGFIELD: O 49.7%. 86.1W/9.5B. You'd kind of expect this seat to be a more D on paper than the Obama/McCain or Romney numbers, but the generic D/R numbers are only 51/49. Definitely a true tossup.
DISTRICT 22: DARKE AND MERCER: O 29%. 97.7W. Ultraconservative rurals. Safe R.
DISTRICT 23: SIDNEY AND URBANA: O 36.4%. 94.9W. Another essentially rural and very conservative and homogenous seat. Ohio is full of boring ones like this. Safe R.
DISTRICT 24: LIMA: O 37.7%. 86.2W/10B. Considering that it's more urban and actually has a black population, it's kind of surprising this is still so conservative. Lima must be as awful a place as Glee being set there would imply. Safe R.
DISTRICT 25: PAULDING AND WERT-PUTNAM: O 36.1%. 94.3W. More boring conservative rurals. Safe R.
DISTRICT 26: WILLIAMS AND FULTON AND HENRY: O 43.8%. W94/H 4.3. Not as conservative as the rest and actually has a bit of a Hispanic population, based on the location I'm going to guess that the manufacturing industry isn't quite as dead here as elsewhere. Still it's probably at best Likely R, it'll take a real wave or scandal-plagued Republican for a Democrat to take this one.
DISTRICT 27: FINDLAY AND SOUTH WOOD: O 41.9%. 91.4W/4.8H. Kind of a combination of the last one and more those boring rurals. Still Safe R.
DISTRICT 28: BOWLING GREEN: O 52.2%. 91.4W/2.6B/3.3H/1.7A. Finally an interesting seat. Based around a big university and some of southwest Lucas County. Very variable on turnout. Obvious tossup.
DISTRICT 29: WEST LUCAS: O 51%. 88.9W/4.9B/2.5H/2.6A. Mostly blue collar Toledo suburbs, some affluent ones and exurbs to cancel it out. Generic D did way better here, but as is it's a tossup.
DISTRICT 30: SENECA AND SANDUSKY: O 48.6%. 92.1W/4.7H. I was going to name this "Seneca and Penn State Disgrace" but my last joke name didn't go over well and that's a bit touchier subject...regardless this district was quite narrowly for McCain, Romney did a tad better, and even the generic D/R numbers tilt Republican, so it's probably Lean R.

So that leaves us with 22R - 5D - 3 Tossup. It gets better in later parts of the state though.

[Benj]

So I ended up redrawing all of Ohio's legislature for both houses. I started for basically two reasons, to get a feel for how the layout the state skews the partisan numbers, and to experiment with grouping three House seats into one Senate. The latter is actually a tad tricky, since you might have an area that's an obvious community of interest and has seven districts, but one has to be left out, and you end up with some awkward Senate districts. But those will come later. For now: House.

Just because Ohio doesn't follow their constitution is no reason for you to not do so.

The Ohio Constitution rules are terrible and internally inconsistent anyway. Not worth trying to follow.

[jimrtex]

So I ended up redrawing all of Ohio's legislature for both houses. I started for basically two reasons, to get a feel for how the layout the state skews the partisan numbers, and to experiment with grouping three House seats into one Senate. The latter is actually a tad tricky, since you might have an area that's an obvious community of interest and has seven districts, but one has to be left out, and you end up with some awkward Senate districts. But those will come later. For now: House.

Just because Ohio doesn't follow their constitution is no reason for you to not do so.

The Ohio Constitution rules are terrible and internally inconsistent anyway. Not worth trying to follow.

Could you be more specific?

[jimrtex]

I'm not familiar with the constitution of Ohio. I only looked for general CoI and population equality. That said though there are some districts I hate and had no choice but to draw this way, there are worse ones in the actual map.

Minimum requirement is that for a county entitled to N+ representatives, N districts must be wholly contained within the county, with only one district crossing county boundaries.

Counties entitled to close to one representative (and the Ohio Constitution provides for a 10% leeway for these counties should be in one district).  For 2010 this applies to Wood, Richland, and Columbiana counties.

The other provisions say that to the extent feasible counties should not be divided.  The Ohio redistricting board has interpreted "not feasible" to mean "couldn't be bothered", when in fact it is demonstrably quite feasible to have very few extraneous county splits.

[Torie]

Ohio is set to lose a CD in 2020, and so I prepared a little mappie of what a map at that time might look like, which hews to appropriate non-partisan redistricting principles. I anticipated population changes. Can you guess which CD I projected to grow the most in population, and by a rather substantial amount over the balance of the CD's?  The population range runs from 80,000 over the average population per the 2010 population figures for one CD to 155,000 short for the CD I project will increase in population most robustly.  Obviously, everything is guesstimates.

I dumped the concept of Cleveland snaking to Akron. The VRA does not require that (they are not one community of interest), and I don't think playing that game is defensible in any non-partisan good government type map.

[Brittain33]

Can you guess which CD I projected to grow the most in population, and by a rather substantial amount over the balance of the CD's? 

OH-12?

[Torie]

Can you guess which CD I projected to grow the most in population, and by a rather substantial amount over the balance of the CD's?

OH-12?

Bingo.  My map maker has now labeled it OH-07, but I know which one you mean. Columbus is the growth epicenter, and it's moving north and east mostly I think.

[TJ in Oregon]

I like it Torie. You did a good job of keeping communities of interest intact (and you drew the Cleveland that really should have been drawn this time from that perspective). I don't think a VRA seat would be possible in 2020 for Cleveland. You'd have to link it to Columbus by that point.

The only unfortunate issues to me are the NE corner seat and the 5th that stretches from Columbus to the Ohio River.

I'd guess that map would be something like:
8 R (2,3,4,5,7,8,12,13)
4 D (6,9,11,14)
3 Tossup (1,10,15)

Bingo.  My map maker has now labeled it OH-07, but I know which one you mean. Columbus is the growth epicenter, and it's moving north and east mostly I think.

I agree here too: Columbus is clearly the metro area expected to grow the most over the next decade. To some extent, the population growth missing from NE Ohio recently is living in Columbus.

The north and east sides have the widest selection of growing suburbs at the moment, so the seat you chose is quite likely to be the fastest growing. Interestingly, Licking County was the one that swung most toward Obama in 2012 despite him losing it by a larger margin than he did in 2008 just from the population growth. By 2020, suburbia might stretch all the way to Newark and Delaware City.

I'd have to imagine though that at some point people will start to realize how much available land there is on the south side close to downtown that is pretty desolate and undeveloped. Columbus is a weird city in that you can exit downtown to the north or east and be in developed areas for close to an hour's drive, but leave southwest and be in a corn field in 10 minutes.

[Torie]

THe NE corner seat is still kind of trapped if you want to keep the black percentage up in OH-11 and keep Warren and Youngstown and Akron together which makes sense. If you want to keep the Cincinnati metro seats compact, the population numbers kind of force the seat going from the south Columbus burbs to the Ohio River.  It is either that or a long snake along the Ohio River, and keeping erosity down (along of course with minimizing county chops) is a major consideration. OH-04 has to squeeze that seat (my OH-05) the way it does to get its population up. The three river counties east of Clermont are always been kind of lost children that don't fit anywhere, and continue to be. They are what's left after creating compact communities of interest CD's elsewhere (plus the trapped OH-15 CD, but bear in mind that north Summit County is really part of the Cleveland suburban/exurban ring anyway, which OH-15 shares with OH-10.  It just has to reach deeper into Summit and take all the northern and western areas in order to get rid of the Trumbell and Portage County chops, which should be jettisoned in a non partisan map. It also looks a bit ugly because OH-11 takes most of Solon, but Solon is the only place left within reach of OH-11 within Cuyahoga County that has some blacks in it, with potential for more down the line.

The political numbers are below. Presumably by 2020, OH-01 will be clearly lean Dem, and OH-10 more Dem than now, maybe a tossup, maybe tilt Dem. All it needs to flip it is some blacks living right next door moving in to some of the almost totally white areas. Ditto OH-15 for that matter.

Given the map, and the population changes, it is interesting that from the 2011 Pub gerry however, that the GOP just loses the 16th seat, OH-01 goes marginal to tilt Dem (even with 2008 election numbers, but tilt now), and OH-10 and OH-15 get a bit more marginal. OH-13 is marginal too (the McCain percentage overstates Pub local strength), but that is the one seat that might have potential to trend Pub over time.

[Benj]

I'd try connecting Akron to Canton and putting Columbiana County in with Mahoning and Trumbull, to the extent possible. Akron-Canton is a clear community of interest (they share an airport, family from Massilon would definitely consider Akron part of their area), as is the Pennsylvania border area (Lisbon and Columbiana city are clearly connected to Youngstown). Otherwise looks good, though.

[Torie]

I'd try connecting Akron to Canton and putting Columbiana County in with Mahoning and Trumbull, to the extent possible. Akron-Canton is a clear community of interest (they share an airport, family from Massilon would definitely consider Akron part of their area), as is the Pennsylvania border area (Lisbon and Columbiana city are clearly connected to Youngstown). Otherwise looks good, though.

OH-14's share of Summit has 140,000 more people than  Columbiana, so that dog won't hunt.

[TJ in Oregon]

Actually Torie, I think your OH-15 will see more blacks moving in than your OH-10. By 2020, I'd suspect Solon will be close to a majority black and Lyndhurst will look like South Euclid in its racial mix. Lyndhurst and South Euclid share a school district, so based on the way white flight works, I think it will be a major factor. On the west side, white flight still hasn't happened from the city itself yet for the most part (though it is now happening in a number of places in part as a consequence for gentrification of the near west side and downtown area). But the west side of Cleveland itself likely won't yet be majority black (or close even) by 2020. In addition, the few places that would seem somewhat likely to have some blacks moving in (maybe Berea since its a college town and somewhat "hip", or Lakewood, Brook Park, or Brooklyn) are in your OH-11. Parma and Parma Heights may see some racial changes, but it will be minimal because they have a history of locally being known as racist suburbs, although it's not quite how it used to be. Otherwise, the western suburbs are close to 0% black. I think Rocky River for instance is 1.0% black. Many of the working class parts of that area are ConservaDems and another decade will likely erode their strength some more.

After considering it some, I'd guess OH-1 would be lean Dem, OH-10 would be lean R, and OH-15 a true toss-up (moving left since it's R+3 at the moment). OH-10 could be made lean D if desired by switching out some suburbs in favor of some white liberal places like Lakewood or Berea, assuming they're still largely white by then, or by putting Lorain and Elyria in it instead of so much of Medina County. OH-1 is a map-drawing problem for the Republicans since the white flight reinforces their exurban strongholds that are safe anyway in expense of Cincinnati. There's not much that can be done about that. Of course, if the GOP figured out how to get 30-40% of the black vote, Cinci would be theirs indefinitely.

[Torie]

Moving right along, I "solved" the issue of OH-05 going from Columbus to the Ohio River.

From the map, one can guess which CD gave me the most headaches to draw. So many psephological "walls," so little time.

[jimrtex]

I dumped the concept of Cleveland snaking to Akron. The VRA does not require that (they are not one community of interest), and I don't think playing that game is defensible in any non-partisan good government type map.

I agree with TJ on the Columbus-Ohio River district.

I'd take Franklin and all surrounding counties and see how close that is to 3 districts.

Then I'd create a Dayton-Springfield district.   And then take a 3 district area in the southwest and give the Cincinnati area 3 districts, even if you have to extend outward quite a way.

I do like OH-15 better in your second map.  If you come into eastern Cuyahoga, there is no reason to not also come into the county from the south as well.  The northern couple of tiers of townships in Summit County are very much Cleveland suburbs.

You are likely correct on Akron-Youngstown.  If you take Cuyahoga and its neighbors and then extend outward until you have the population for 4 districts, you probably can't include both Stark and Mahoning, so Canton gets left out of the NE districts.

[muon2]

Moving right along, I "solved" the issue of OH-05 going from Columbus to the Ohio River.

From the map, one can guess which CD gave me the most headaches to draw. So many psephological "walls," so little time.

CD 7 just looks nasty. What's the redistricting principle there?

[Torie]

It's called gerrymandering Mike. Desperate measures needed to be resorted to, to make it something other than a dummymander, res ipsa loquitur.  

[muon2]

It's called gerrymandering Mike. Desperate measures needed to be resorted to, to make it something other than a dummymander, res ipsa loquitur. 

I was taking this preamble as applying to both your first as well as second offerings in this series.

Ohio is set to lose a CD in 2020, and so I prepared a little mappie of what a map at that time might look like, which hews to appropriate non-partisan redistricting principles. I anticipated population changes. Can you guess which CD I projected to grow the most in population, and by a rather substantial amount over the balance of the CD's?  The population range runs from 80,000 over the average population per the 2010 population figures for one CD to 155,000 short for the CD I project will increase in population most robustly.  Obviously, everything is guesstimates.

[Torie]

Your assumption was erroneous.

[krazen1211]

If he is still around, John Boehner's district cannot and will not be messed with in such a manner.

[jimrtex]

These maps are based on 2012 census estimates projected forward to 2020, assuming constant same growth rate as 2010 to 2012,



Cincinnati does not come close to supporting 3 districts.  So you have a choice between Hamilton and a Butler-Warren-Clermont quarter-donut, or splitting Hamilton.  But splitting Hamilton may split Cincinnati.  So the first is the preferred option.  We need to trim a bit off the southwest area, so Springfield is gone, and the Dayton seat will be Montgomery-Greene.

Clearly we have to go west from Columbus to  get enough for a 3rd district.

In the northeast, Clark and Mahoning makes about the excess above 4 seats, so in the Northeast there will be a 5th seat that is Canton-Youngstown-Steubonville.

[jimrtex]

These maps are based on 2012 census estimates projected forward to 2020, assuming constant same growth rate as 2010 to 2012,

Balancing the population we get the following map.



Most regions are within 1%, which is more than close enough this far out.

The lavender region can expand on its ends to take in the excess from the Yellow (Greater Toled) and Blue (southeast regions).