Ohio Redistricting Contest

[jimrtex]

Some groups involved with the previous 2009 Ohio redistricting competition are planning a new contest beginning this Friday (July 1) - though with no involvement of the Secretary of State.

Ohio Redistricting Contest

[jimrtex]

Some groups involved with the previous 2009 Ohio redistricting competition are planning a new contest beginning this Friday (July 1) - though with no involvement of the Secretary of State.

Ohio Redistricting Contest

The congressional districts in Ohio will most surely be drawn to benefit the Republicans as they will be drawn by the far-righters currently running the General Assembly.  However, I'm cautiously optimistic that SOS Husted will be able to insert at least some degree of fairness into Ohio's legislative redistricting.  He has been a vocal supporter of fairness in redistricting in the past and he has shown recently that he is willing to stand up to powerful people in his party when it comes to performing his job responsibly.

The constitutional requirements for legislative redistricting are pretty limiting (so far the contest rules only restate the constitution).

[jimrtex]

Yah, they want unsplit counties, I focussed on unsplit towns. 'Specially as Ohio has a number of them that cross county lines.

In the previous contest you could use split towns to game the rules.

If Springfield were split between Homer and Simpson counties, you could place the district boundary on the county line, dividing Springfield; or you could include all of Springfield in a district containing all of Homer County, and Simpson County would not be considered split.  Or you could include all of Springfield in with Simpson County and Homer County would not be split.

The net effect is that the two parts of Springfield could be considered counties for balancing population and not cause split counties, as well as other characteristics.  IIRC, Middletown barely extends into Warren County, so it was possible to have an extension of Warren County come into Butler, leaving the rest of Butler County in a different district.

County fragments were counted based on areas that were part of a county and part of a district, except when all of one or two districts were wholly contained in a county, the remnant would not be considered to form a county fragment.   So if one district was wholly in Hamilton County, and the remnant was in one district, that remnant was not a county fragment.

If you split a smaller county, it counted as two fragments.  But if you bring another district into the county, it only creates a 3rd fragment.  So there was an incentive to concentrate splits, particularly in counties with significant population, since you could manipulate the population used in other districts.  So Summit and Montgomery might be inviting targets, and possibly Cuyahoga and Franklin.

But they didn't really care about towns.  Once you had decided to split a county, you could draw the lines anywhere.  I did a lot of flipping of black precincts in Columbus to get one district to a 50-50 D:R which appeared to be counted as a tie for determining a fair outcome and it happened that 9.5 to 8.5 split was the closest to the statewide ratio.

The Columbus area was messy because of the townships that have been eviscerated by Columbus annexations.  The townships still exist and so have election precincts, but the election precincts are not self contiguous.  This is also true of Columbus itself, which looks like wards were defined assuming that Columbus had annexed all the holes, and precinct boundaries then jumped across the holes.

[jimrtex]

Haven't seen any discussion of Ohio yet.  I was looking at the state constitution's requirements for house and senate districts and I believe that they are impossible to fulfill. 

http://www.legislature.state.oh.us/constitution.cfm?Part=11

Any county which is over one house seat but less than one senate seat is required to be kept whole in one senate seat.  Same goes for the "leftovers" from any county that has more than one senate seat - they must be kept whole.  Cuyahoga County has 3 senate seats plus 2/3 of a seat.  Lake County is at 2/3 of a seat, so it can't take Cuyahoga's leftovers.  This means Lake's senate seat has to go into either Geauga or Ashtabula only. 

However, Portage, Mahoning, and Trumbull are all bigger than one house seat and smaller than one senate seat - yet they are all too big to combine two into one senate seat.  Thus Trumbull, being hemmed in by the other two counties, also has to have its senate seat go into Geauga or Ashtabula only.  However, there is not enough population in Lake + Geauga + Ashtabula + Trumbull to support two senate seats. 

So... anyone want to tell me how a state constitution is interpreted when its clauses come into conflict with one another and with reality? 

You have to draw the House seats first, and then create the senate seats.  The constitution does not require smaller counties to be kept whole, so Ashatabula may get split up in interesting ways.

Otherwise, I think the redistricting board could choose to have 2 senate districts go outside of Cuyahoga County or to split Lake County.

[jimrtex]

I suppose I could... but it would basically require the same sort of textual explanation to know what was going on.  And the point of my post was not to ask *whether* the constitution requires an impossibility, but to ask the legal question of what the requirements are if it does.  Does the Ohio Supreme Court get to decide what comes "closest" to fulfilling the requirements, does the entire clause become ignorable, do earlier sentences within the clause take precedence over later sentences? 

I would draw 11 whole house districts in Cuyahoga and 2 in Lake, and then draw two additional districts (Ashtabula .871, Lake .023, Geauga .081), and (Geauga .720, Cuyahoga .256).   The fractions are relative to the ideal district population of 116,530.

The 15 districts would average .9754 of the ideal population, or 1/2 of the allowed deviation.

When challenged, I would argue that § 11.08 only sets a minimum number of whole districts in Cuyahoga and Lake, and does not prevent creation of another.   If Cuyahoga had 1754 more people, it would be entitled to 11 whole districts plus a fraction; and there is nothing that says the whole districts use up the exact whole-district entitlement.

This plan would totally comply with the spirit of the constitution, since the remnant of the population of Lake and Cuyahoga counties are contained in a single house district.

A variant would keep Lake in 2 districts, and take the bare minimum from Cuyahoga to get Geauga + Ashtabula up to the equivalent of 0.95x2 population (about 26,000), and form the remainder of Cuyahoga into 11 districts.

Alternatively, I would argue that equal protection renders other restrictions unconstitutional and are severed to the extent necessary to create a plan.  I create three house districts from Ashtabula, Geauga, and Portage, and then split Portage between two senate districts.

This violates § 11.11, but only in dividing Trumbull, and the spirit of the constitution is to split smaller counties when necessary.

You  could create another senate district from Trumbull, the full district part of Portage, and a remainder of Stark, but that would require a Trumbull-Stark district.

So instead:

Geauga 801, Ashtabula 199 (with 2 Cuyahoga)

Ashtabula 672, Trumbull 328 (with 2 Lake)

Trumbull 1000,
Trumbull 477, Mahoning 523
Mahoning 1000

remnant Mahoning 526

Summit and Portage for 2 senate districts.

Stark for 1 senate district with a small remnant.

Lorain and Huron for 1 senate district.

Medina, Wayna, and Ashland for 1 senate district.

So basically constitution is largely ignored in Trumbull and Mahoning, but end up with Youngstown centered senate district.

[jimrtex]

You might come up with an example to help lesser minds ponder more concretely what you see as the conundrum dmapper.  Just a thought.

Ohio requires that when a county has more than enough population for one house district, that the whole number of districts be created in the county, with the remnant formed in one district that is combined to form a house district with other counties or parts of counties.

The above maps shows the population in terms of house districts.  So for example, Portage County would require one house district entirely in the county, with the remnant combined with other districts.  While it is desirable that smaller counties not be divided, it is not required, so that Geauga and Ashtabula can be hacked apart if necessary to equalize populations.  If you look at the existing map, you will see that is done quite regularly.  After we get outside the NE population concentration we will have a lot more flexibility since we can make fairly arbitrary splits of small counties.

There is a special rule that applies to counties like Columbiana.  If the population is equivalent to 0.90 to 1.10 districts, it may be formed into a single house district.  Otherwise, districts are allowed to vary from 0.95 t0 1.05 of the ideal population.

Senate districts are comprised of 3 house districts.  Counties with more than one district must be in a single senate district.  So Lorain, Medina, Lake, Portage, Trumbull, and Mahoning must be contained in one senate district.

For larger counties, such as Cuyahoga, Summit, and Stark, as many senate districts must be formed in the county, with the remainder in a single senate district.

It would be fairly easy to meet these rules with some counties.  For example, Summit could have 4 house districts, Portage one, and one district combining parts of the two counties.  These districts would only be 0.6% over the ideal population.   There could be one senate district comprised of 3 Summit house districts, and another comprised of the remaining Summit house district, the Portage district, and the Summit-Portage district.

Dmapper's conundrum relates to Cuyahoga and Lake counties.  Cuyahoga would be entitled to 11 house districts and Lake to 2 house districts.  Cuyahoga also would have 3 senate districts formed from 9 of its house districts.  A 4th senate district must be comprised of the other two house districts and one other house district.  Similarly the two Lake house districts must form a senate district with one other house district.

Lorain and Medina must be in a single senate district, so they can't be paired with Cuyahoga.  The remnant of Summit is larger than a house district.  Portage and Trumbull must be contained in a single senate district as well.  So the only source of the two house districts to be paired with Cuyahoga and Lake, are Geauga and Ashtabula, but they have insufficient population to form two house districts.

We can make the districts in Cuyahoga and Lake a little bit smaller than the ideal.  Then we can take the remainder of Cuyahoga, Lake, along with Geuaga and Ashtabula to form two house districts of sufficient population.  Overall, the 15 house districts would average 2.5% below the ideal.

If we could use the remnant from Portage, we would have quite close to an ideal population for 15 house districts.  But this would split Portage between two senate districts.  Moreover, it would cause problems with Trumbull which has too much population to be placed in the same senate district with Mahoning.

We could place Trumbull and Portage in the same senate district, but that would have a deviation of 6.3% for the senate district, and at least that for the 3 senate districts.

At this point, you might be able convince the SCOTUS that such deviation is necessary in order to comply with the other standards of the state constitution.  But you might not be able to convince the Ohio Supreme Court, since he 5% limit is in the state constitution, with a singular exception.  If the People wanted to allow a larger deviation in other cases, they would have placed it in the constitution.

At this point you have to start ignoring certain rules.  If you took the remnant of Portage and used it with Geauga and Ashtabula to form the two house districts that Lake and Cuyahoga need, then Trumbull, the whole Portage house district and the remnant of Stark could form a senate district.  But that would spit Portage between senate districts, and require a Trumbull-Stark house district.

So the best solution might be to carve a small part of Portage or Trumbull and use that with Geauga and Ashtabula.  Then form a senate district from Portage and Trumbull.  This will violate two provisions of the constitution.  It will split either Portage or Trumbull between three house districts, and also split that county between 2 senate districts.

But the argument will be made that equal protection trumps the county line provisions.  The Ohio AG will argue on behalf of the redistricting board that they made a rational and prudent reconciliation of the various provisions of the constitution.  Depending on the partisan composition of the Ohio Supreme Court, they will either approve or reject the map.

The final map then is:

Cuyahoga + Geauga: 12 house districts and 4 senate districts.    About 21,000 persons will be added to Geauga to form one house district, with the other 11 in Cuyahoga.

Lake + Asthtabula + 20,000 from Trumbull: 3 house districts and one senate district.  Two house districts are in Lake, and the other is Ashtabula-Trumbull.  With this format we avoid splitting both Geauga and Ashtabula counties.

Remainder of Trumbull + Portage: 3 house districts and one senate district.  A house district in each county, plus one that crosses the county line.  This is where our constitutional violation lies.  The 20,000 people from Trumbull shifted north are in a different senate district, and Trumbull is split between 3 house districts.

Mahoning and Columbiana: 1 senate district, with 2 house districts in Mahoning and 1 in Columbiana.

Summit and Medina: 2 senate districts and 6 house districts.  One senate district and 4 house districts in Summit, one house district in Medina.

Lorain + Huron: 1 senate district and 3 House districts,  We could also use Ashland, or split Erie, though since a county split is not needed it might be unconstitutional.

Stark: 1 senate district and 3 house districts.  About 25,000 persons in Stark county added to a house district further south.

In SW Ohio:

Hamilton, Clermont, Brown: 3 senate districts.  7 house districts in Hamilton, one in Clermont, and one in Clermont-Brown.

Butler, Warren, Clinton, Fayette, Highland: 2 senate districts.  3 house and 1 senate in Butler, 1 house in Warren, 1 in Warren with a bit of Butler, and one house district in the 3 small counties, so the eastern senate district is a tiny bit of Butler, Warren and the three smaller counties.

Montgomery, Green: 2 senate districts, with 4 house districts in Montgomery, 1 in Greene, and one across the county line.

Franklin: 10 house districts, 3 senate districts.  The extra house district goes with two house districts outside Franklin.

Lucas: 3+ house districts.  One senate district, and total flexibility with the remnant.

[jimrtex]

jimrtex put it well.  I'd quibble with one thing, which is that Cuyahoga's excess could theoretically be paired with Medina.  If the first 10 house districts in Cuyahoga are drawn 5% larger than average, then they take up 10.5; the remaining ~.5 of a district can be paired with Medina's .5 excess.  Then one full Cuyahoga district, one full Medina district, and the Cuyahoga/Medina split district will make one Senate seat.  Not that this solves the Lake/Trumbull problem, though. 

If I were the GOP I'd go to the court *before* drawing any maps and ask them to decide which clauses take precedence, and how.  That way you get rid of all uncertainty and Democrats don't get to pick and choose their arguments. 

Cuyahoga + Medina is 12.469 house districts.  At some point, it may be questionable whether you may apportion 12 districts to an area that is entitled to essentially 12-1/2, just because the error is less than 5%.  You would be following the rules, but have lost sight of the underlying principle.

You could give Cuyahoga + Summit with a population of 15.643 house districts 15 house districts with an average deviation of 4.2% as well.

Republicans have a 5-2 majority on the redistricting board.  It is their responsibility to draw a plan.  Even if the Ohio Supreme Court were to overturn the plan they would have the responsibility to draw a new plan.  Any plaintiff would have to demonstrate that there is a constitutional alternative, or if there are none, that the plan by the redistricting board is irrational and arbitrary (or whatever other standard the court would apply).

§ 11.13 says that any reapportionment cases go directly to the Ohio Supreme Court, rather than a lower court and then appealed to the court.  They won't decide that any of the apportionment provisions are unconstitutional per se, but only when applied to the 2010 data.  The 3-way split of Portage or Trumbull is the least worst alternative, as far as complying with constitution.  If you split Trumbull, you avoid splitting Geauga, which is preferred by the constitution.

The only other alternative that I see would require a Trumbull-Stark house district.  And I suspect that a court would not approve a discontiguous district.

[jimrtex]

OK, here's a first draft of a GOP gerrymander in which the only constitutional violation is that Portage will be in two separate senate districts. 

Orange Cleveland blob is 6 house seats which will be turned into 2 senate seats, all safe Dem. 

Blue/Purple/Green are 3 house seats, to be turned into 1 senate seat, all various degrees of lean GOP.  Blue is 50.7-48.3 Obama, and basically did not change from its current incarnation.  Purple is 48.6-50.1 Obama-McCain.  Green is 52.0-46.5 Obama-McCain. 

Teal seat in east Cuyahoga is 58.2-41.0 Obama-McCain.  Whoever had that seat for the GOP is probably screwed; apologies (unless they lived in the part that got sliced off and put into the green seat).  The cyan seat that snakes from Cleveland down the west side of Akron is 50.0-48.7 O-M.  The bronze Akron suburbs seat is 47.1-51.7 O-M.  These three seats together (which are barely contiguous by going all the way around Akron!) form one senate seat which is a very slight GOP lean at 51.8% Obama - which might not be that different from the current Summit-only GOP-held senate seat but that's just a guess.  In the middle of Summit county in bluish gray is one senate seat to be divided into 3 house seats (probably 2 of which will be solid Dem and 1 at a very slight lean Dem - I think I can get one at 53.9% Obama). 

All of the aforementioned house seats are 4%+ above ideal. 

The aquamarine Geauga-Portage seat is safe GOP at 54.6% McCain.  It gets matched with the two Lake County districts (not drawn; probably one is lean GOP and one is a toss-up) for a lean GOP senate seat. 

The purple Geauga-Ashtabula seat is lean GOP at 51.2-46.9 O-M.  Probably a better seat for the GOP Ashtabula rep than his current one.  The grey Ashtabula-Trumbull district is actually a tossup at 52.7-45 O-M (currently held by a Democrat) since I was able to stuff the pink district with Niles, Warren and Sharon.  These three seats form one senate seat that will be safe D. 

Red Portage district is safe D, and will be combined with the two Mahoning house seats to form a safe D senate seat. 

From here on out things are pretty easy to draw.  One additional benefit is that the Dem-held senate seat based in Jefferson County might be able to be made swingier since Republican Carroll County is no longer tied to Mahoning. 

Final analysis: Dems go from 11 in Summit/Cuyahoga to 9.5-10, though, and lose one Senate seat.  They also get a weakened outer Trumbull house district and a weakened Jefferson Cty senate seat.  GOP trades their S+E Cuyahoga district for a swingier district that's strictly South Cuyahoga, and makes one of their Summit districts a toss-up/lean D (although how strong their previous Summit districts were, I have no idea), but otherwise is unaffected. 

Now someone is going to tell me that I've placed 4 GOP incumbents in the same district...

I think you may get challenged under the VRA for creating all of the Cuyahoga and Summit districts at the upper end of the population limit at 4.2% deviation, and you still have not avoided violating the Ohio Constitution.  The districts to the east are 2.8% under, so you have a concentrated 7% differential.   It may be particularly contentious if you appear to have packed blacks into overpopulated districts.

The Summit-Cuyahoga house district, and the eastern Summit house district are not compact (Ohio Constitution 11.07(A)).  I'm not sure about the Geauga-Portage district.  Could the town south of Ravenna be shifted to the salmon district?  While senate districts are not required to be compact under the Ohio Constitution, it could be an issue under a VRA challenge.

Do the Cuyahoga and Summit house districts comply with 11.07(B) and 11.07(C) regarding splitting of towns and cities?

What happens if you simply put the Summit senate district in Akron and the southern part of the county, and then push the green district northward, making the cyan a simpler cross-border district, and the bronze district to the east or south of that.

[jimrtex]

Well, OK, compactness... but that's entirely subjective, and there are quite a few current districts that are almost as bad.  Similarly VRA violations are subjective as well.  It is true that the Cleveland and Akron black districts are larger than ideal... but so are the white districts.  

I believe I followed the township/city rules to the letter, but I suppose there might some splits that one could avoid with more cleverness.  

Re: your suggestion, I think the green district becomes very marginal for the GOP if it gets pushed any farther north and it might endanger the senate seat.  Meanwhile, exchanging south Summit for the Cuyahoga Falls and Twinsburg areas would probably tip the border-crossing senate seat into lean D territory.  So if that really was required I think the GOP would look to go in another direction.  

So cyan is Walton Hills, Valley View, Cuyahoga Heights, Newburgh Heights, Brooklyn Heights, part of Parma, Brooklyn, Lindale, and Brecksvile?

Doesn't it also split New Franklin?  Aren't you permitted only one split per district, not one per district boundary?

If I can draw a plan that has the green district taking in all of Parma (and possibly Brooklyn and Lindale, though they could be shifted to the orange blob), and moving Seven Hills, Independence, Broadview Heights, and North Royalton, Strongsville or some combination thereof, don't I have to do so?  I have established feasibility if I can do so.

Do you have details of your Cuyahoga and Summit maps?

BTW, the redistricting contest is supposed to go live tomorrow.

[jimrtex]

The contest rules include the following paragraph:

You must be logged in to read this quote.

[jimrtex]

On a related topic, is it just me or does the DistrictBuilder software tool really suck?  It looks pretty but I hate having to wait 15 seconds every time I add an area to a district for it to recalculate the stats (which I don't even use most of the time).  I'd have entered already if they just used DRA, but I don't think wrangling with DistrictBuilder is worth anybody's time.  

It is terribly slow.  It appears that you might be able to assign new districts while it is calculating - but I'm not 100% sure, I seem to have managed to get a few districts ahead of myself.  EDIT while typing this reply, the statistics did catch up.

A bigger problem is that I seem to have got locked up.  I was adding districts according to the numbering scheme of the constitution.  So I did the 5 single county districts and then added all of Cuyahoga as District 6 (eventually it will be 6-16).   Then Franklin was 17, etc.

After Hamilton, Summit, Montgomery, Lucas, Stark, Butler, and Lorain, I tried to do Mahoning as 59 (though only the 15th district) and it shows the county as a red outline indicating it locked, plus all the other districts are eliminated, but can be restored with redo.  I can select Lake and it does the same thing.  I can't seem to get past that.

I'm in the process of assigning individual incomplete districts in each county, and have to through 43 Montgomery.

There is an export and import facility in DistrictBuilder which I think is a csv block assignment list.  Can DRA export such a file?

There is a webinar at 5 pm EDT today which you might want to sign up for.

[jimrtex]

On a related topic, is it just me or does the DistrictBuilder software tool really suck?  It looks pretty but I hate having to wait 15 seconds every time I add an area to a district for it to recalculate the stats (which I don't even use most of the time).  I'd have entered already if they just used DRA, but I don't think wrangling with DistrictBuilder is worth anybody's time. 

Webinar postoned/cancelled in part due to slowness of software. 

[jimrtex]

You heard it here first, folks! 

On a related topic, is it just me or does the DistrictBuilder software tool really suck?  It looks pretty but I hate having to wait 15 seconds every time I add an area to a district for it to recalculate the stats (which I don't even use most of the time).  I'd have entered already if they just used DRA, but I don't think wrangling with DistrictBuilder is worth anybody's time. 

You haven't drawn any districts in cities yet, have you?

Do you think it is feasible to have two minority-majority VAP senate districts in Cuyahoga County?   The Black VAP is 28%.  If we assume no race-based age differential, then that is 358,434 Blacks.  2 senate districts equals 6 house districts of 116,530.  358,434 / (116,530 x 6) = 51.2%   That is we have to have 6 house districts that average 51% Black, and 5 house districts that are 0% Black.  Moreover, two of the White house districts have to be contiguous with Geauga county.

I have 4 suburban districts from Parma west plus the southern tier of towns that are under 3% Black.  I have 3 eastern suburban districts that are 52%, 31%, and 10% Black.  So I think you would have to run 6 house districts clear across Cleveland east to west and into the eastern suburbs.

Have you tried to preserve existing districts that are in the 95% to 105% range?

[jimrtex]

Not really.  I had decided not to bother unless their software improves.  But running with your numbers, it does seem impossible.  If blacks total 307.2% of a house district in Cuyahoga then you need to "waste" at most 7% of a house district of blacks outside of the two senate seats.  If your 4 south/west suburban house seats are 3% black each, that's already 12% "wasted" with one more seat to go.  Will shrinking the VRA seats by 5% each be enough to make up for this?  Seems unlikely...

I think it may be barely possible, if you include Hispanics.

My 6 districts are:

West Cleveland: 19% Black, 15% Hispanic
East - Central Cleveland: 62/10
South - Southeast Cleveland: 63/4
Far East Cleveland - Euclid - East Cleveland: 67/1
Southeast Cuyahoga: 52/2
East Central Cuyahoga: 31/2

Southeast Cuyahoga wraps around the SE corner of Cleveland from Garfield Heights to Shaker Heights.  You could boost it a bit higher if you included Oakwood or Highland Hills, but that would split the other other eastern suburban district or require splitting towns or cities.

I drew the two eastern Cleveland district east west, to avoid packing the Southeast district.

If you average the VAP you can get two senate districts of:

Cleveland: 48/10
Cuyahoga East: 50/2

So depending on interpretation there are 2, 1, or 0.  Both would include 2 majority Black House districts, so that might count as a plus.

The rules appear to give some emphasis to 11.07(D) of the constitution.  That messes up my plan since the two western Cuyahoga districts are within 5% of the ideal, and both come into Cleveland.   11.07(B) may give a preference to splitting Cleveland, since its wards can be used as the equivalent of towns.

[jimrtex]

Can You See This?

I think this will let you see my House plan with the Cuyahoga districts and the 5 single county districts.  Does it?

There is an option to download a plan as a CSV file containing triplets of census blocks, districts, and (all values are 1).  So you may be able to create a plan in DRA and then upload it.

My house plan with just Cuyahoga and the 5 single-count districts was 42,501 lines, so a statewide plan is probably around 500,000 lines.

I think that it is definitely a substantive constitutional requirement to preserve existing districts when they are within 95% to 105% of the ideal.  § 11.07

There are 38 districts currently in this range.  In addition, Allen and Wood are existing single-county districts within the 90-100% range.  And it appears that all but a half dozen or so can be maintained.

[jimrtex]

Can You See This?

I think this will let you see my House plan with the Cuyahoga districts and the 5 single county districts.  Does it?

There is an option to download a plan as a CSV file containing triplets of census blocks, districts, and (all values are 1).  So you may be able to create a plan in DRA and then upload it.

My house plan with just Cuyahoga and the 5 single-count districts was 42,501 lines, so a statewide plan is probably around 500,000 lines.

I think that it is definitely a substantive constitutional requirement to preserve existing districts when they are within 95% to 105% of the ideal.  § 11.07

There are 38 districts currently in this range.  In addition, Allen and Wood are existing single-county districts within the 90-100% range.  And it appears that all but a half dozen or so can be maintained.

I can see the view just fine.

I'm puzzling over the rules minimal set of VRA districts. I can place the two black-majority Senate districts within Cuyahoga, but I can also place one entirely within Hamilton. I'm very close to getting one in Franklin as well. The black VAP gives 3.74 senate districts out of 33, so if possible one should try to have 3 or 4 black-majority districts for rough proportionality. Unfortunately I don't see how that fits in the rules, and it looks like it would be disadvantaged in the scoring.

Can you get the Hamilton senate district to be comprised of valid house districts?  The two westernmost suburban seats are within the 95%-105% range, and I think that Hamilton has to be paired with Clermont and Brown for the senate districts 7 in Hamilton and 2 in Clermont Brown.

So that gives one senate district with the two western house districts and one more

Then the 3 districts in Cincinnati and near northern suburbs.

And then one house district in the east that goes with Clermont-Brown.

Does that slide the VRA section too far east?

I found where the application does have an option to display total VAP, and so was able to calculate Black % VAP for my two Cuyahoga senate districts, which gives me 50.16% amd 48.05%.  I'm going have to redo the house districts to maintain the current house districts.  The two westernmost Cuyahoga districts qualify, but that leave Lakewood and a couple of others (maybe Brook Park and Brooklyn) in Cleveland districts.

[jimrtex]

Actually I bring the western Senate district along the river to get the numbers up for the black district in Hamilton. There are at least 2 reasonably compact black-majority house districts at that point.

In Cuyahoga my two Senate seats are 50.7 and 51.4% VAP. I haven't worked out the number of house seats yet. I kept municipal boundaries intact, but it leaves me with two county fragments. To meet the VRA I have to create fragments of the county or multiple municipalities. I assume that the VRA trumps the OH Constitution, but I'm not sure what the better solution is for the map.

Does the VRA require creation of majority-minority districts if that would violate neutral districting standards that are intended to produce compact and stable districts?

[jimrtex]

Federal district court decision following 2001 legislative redistricting

[jimrtex]

Video includes Franklin County and explains how townships are protected at the expense of Columbus

[jimrtex]

Federal district court decision following 2001 legislative redistricting

The key part of the decision (as noted by one of the concurring opinions) was on the ability to meet the Gingles test. In particular the court found that the plaintiffs could not produce a map with a majority-black-VAP district that complied with the OH Constitution as a remedy for the challenged districts. If my map is both constitutional and has the extra majority-black district(s) then that defect of the plaintiffs would be overcome.

I'd interpret the constitutional requirements as:

(1) § 11.07(D) keep existing districts if feasible and consistent with § 11.03.   § 11.03 is the equal population requirement, including an explicit requirement of districts being in the range of 95% to 105% of the ideal size.  But this does not appear to require that those districts be the best in terms of building blocks.  Of the 40 districts in the 95% to 105% range, it would appear that it is feasible to maintain all but about a half dozen where that would force another district to violate the equal population standard.

(2) § 11.07(A), § 11.07(A) and § 11.09 create single-county districts.   It appears reasonable to create all 5 of these districts.  Collectively their population is equivalent to 4.965 districts so there is little population imbalance. 

It could be reasonable to split off a piece of Wood to go with Lucas, which would allow Erie and Ottawa to be paired and better population equality.  The alternative is to keep Wood whole with a 7.7% deviation, and combine Lucas, Erie, and Ottawa for 5 districts with an average deviation of -3.8%.   The difference is over 10%.

Lucas+Wood would be -2.7%, and Erie-Ottawa would be 1.7%.

(3) § 11.07 (A) Build districts from whole counties.  This is not possible in the larger counties.

(4) § 11.07 (b) Build districts from counties, townships, municipalities, and city wards.  In many cases cities and townships have been conformed.  So if we include all of a city, we are also including its coterminous township. 

But larger cities can not be formed into single districts.  This requires building districts from city wards.  But city wards tend to have equal population, which may make construction of districts from city wards alone difficult if not impossible.  No current Cleveland districts are entirely in the city.  The average population of Cleveland ward is equivalent to about 0.180 house districts.  5 wards equals 0.90, 6 wards equals 1.08.  It may be possible to get get inside the 0.95 to 1.05 limits based on population changes since 2000 (this is a silly provision since the ward boundaries will probably also be updated based pn the 2010 census.

[jimrtex]

Video includes Franklin County and explains how townships are protected at the expense of Columbus

Nice. I was in the room for that particular presentation. I suspect I'm in one of the Q & A segments.

I was interested in the grid maps, but couldn't find a definition of the algorithm.

[jimrtex]

Jimrtex, are you saying that maintaining existing conforming districts supercedes building districts from whole counties, etc? ThaT doesn't seem to match past practice. I could see recognizing districts that currently exist and can't be improved, but if one must keep all the conforming districts it would be a strong constraint. In any caSe that would work against the compedtition goals.

11.07(D) says "In making a new apportionment, district boundaries established by the preceding apportionment shall be adopted to the extent reasonably consistent with the requirements of section 3 of this Article." 

11.03 says "The population of each house of representatives district shall be substantially equal to the ratio of representation in the house of representatives, as provided in section 2 of this Article, and in no event shall any house of representatives district contain a population of less than ninety-five per cent nor more than one hundred five per cent of the ratio of representation in the house of representatives, except in those instances where reasonable effort is made to avoid dividing a county in accordance with section 9 of this Article. "

11.09 is the special 90-110% rule for single-district counties.

So even though 11.07(D) is juxtaposed among the rules for using whole counties, etc.  The basis for continuation is solely population.  I think it would be pretty hard to claim that a district in the range of 95% to 105% was not "substantially equal" and was merely allowed.  That is too narrow of a splitting of 11.03.   It is not districts per se that are continued, but district boundaries.  There are a number of in-range districts that box in out-of-range districts (see Lorain and Mahoning counties).  In those cases the district boundaries would need to be changed.  I would not interpret 11.07(D) so narrowly as requiring continuation of parts of a district's boundaries.

But there are a number of in-range districts along the Indiana border where counties are split between two in-range districts.  It could be considered "reasonably consistent" with "substantial equality" to shift townships between the two districts to achieve greater equality.  My personal belief is that once you have split a county, you have sailed from the 5% safe harbor.

The competition goals are somewhat contradictory.  One is to demonstrate that "fairer" districts can be created.  But another is to produce districts that can be presented to the LAB (for legislative districts) or the legislature for congressional districts.  If the legislative districts don't comply with the constitution, the LAB would likely disregard them.

The rules (Page 6, #5 says that existing in-range districts should be kept.  It is unclear what "other legal requirements" means.  You possibly could argue that if a district on one side of township split is in-range, but that on the other side is out-of-range, that in drawing the new out-of-range district, you may not maintain the existing township split (if it can be avoided).

But the rule is contradictory to the scoring rules, since it is likely to preserve some existing county splits.  There was a proposed constitutional amendment which would have added "fairness" criteria, but would have removed the provision regarding existing districts.

[jimrtex]

The rules (Page 6, #5 says that existing in-range districts should be kept.  It is unclear what "other legal requirements" means.  You possibly could argue that if a district on one side of township split is in-range, but that on the other side is out-of-range, that in drawing the new out-of-range district, you may not maintain the existing township split (if it can be avoided).

This is similar to what I was thinking. I was willing to maintain any district that was within range and consisted of whole counties, or of whole towns for those districts entirely within a county. District 86 seems to be the only multi-county district in this category. Districts 16 and 18 in Cuyahoga, and district 54 in Butler fit the multi-town within one county description.

Senate districts 10 and 20 would meet the requirement, but I don't see any constitutional provision to maintain senate districts that already meet population requirements.

Edit: On closer inspection I see that North Olmstead has a slight bit that crosses into district 18. Perhaps its due to recent annexation. In any case it removes 16 and 18 from my list.

Edit 2: The split in Lake also meets the test. These are current districts 62 and 63. As to my previous edit, though I may not need to technically keep 16 and 18, I'm leaning towards their inclusion after shifting the boundary to reflect the tiny change in N. Olmstead.

My inclination would be to keep:

3 Wayne County
16,18 Cuyahoga
21,26,27 Franklin
29,30 Hamilton
36,37,38 Montgomery with the intent that there will be a shared house district with Greene
42 Summit
50 Stark the population is right for 2 districts inside.
53,54 Butler with the intent there will be a shared district with Warren
56, but not 58 Lorain   58 has 57 blocked in, plus Lorain+Huron equals 3 districts, which means chopping off the piece of Seneca.
Neither 59 or 61 Mahoning, et al.   59 has 60 blocked in, and when 60 expands, 59 will have to take the remnant of Mahoning.
62,63 Lake
The following 3 pairs split a county (Defiance, Auglaize, and Darke).   Since both side of the county split are in range, there is no need to redraw either district.
74+75
76+78
77+79
Not 81.  Counties to north need all of Ottawa, and change of 58 will allow creation of Sandusky-Seneca district.
82+83 split of Marion County can be kept
Not 84, splits of Greene and Clark County are wrong.
86 OK
None of 87,88,89 Split of Clermont will have to redone which will force elimination of split of Adams, which will force change in Lawrence split.
93+94 split of Washington will have to be redone which will in turn trigger a redo of the Muskingum split.
Not 98 Geuaga + a bit of Cuyahoga.   Initially, I was inclined to keep this, since it would permit keeping both Geauga and Ashtabula whole.  But I decided that it was a bit dubious to give Cuyahoga 11 whole districts and part of another, when it doesn't quite have enough for 11 districts.  Since Ohio anticipates splitting counties, it doesn't seem proper to treat it more like an apportionment.

Looking at whether it would be possible to work around the other districts, keeping 86 makes forces a bunch of additional splits.  Switching Adams for Pike allows creation of 8 whole county districts in southern Ohio (counting the Fairfield district and the Fairfield remnant as whole counties).

Delaware and Licking would make a good pairing for a senate district, but it doesn't work with the counties to the north.   I had originally anticipated two split counties, including one in Marion.  This would force elimination of 83, but would allow keeping of 82.  But since it is on the low side (95.7%) it would make sense to redo the boundary.

That reduces my list to

3 Wayne County
16,18 Cuyahoga
21,26,27 Franklin
29,30 Hamilton
36,37,38 Montgomery
42 Summit
50 Stark
53,54 Butler
56 Lorain.
62,63 Lake
74+75
76+78
77+79

Before OMOV, Ohio never (or rarely) changed its legislative districts since it did fractional apportionments of representatives and senates.  After the initial senate districts were defined the last county in Ohio was created from parts of counties that were in different senate districts.  Even though the senate districts were made up of whole counties, those senate districts split the new county.  So I think a strong case can be made foe keeping districts if at all possible.

[jimrtex]

On a related topic, is it just me or does the DistrictBuilder software tool really suck?  It looks pretty but I hate having to wait 15 seconds every time I add an area to a district for it to recalculate the stats (which I don't even use most of the time).  I'd have entered already if they just used DRA, but I don't think wrangling with DistrictBuilder is worth anybody's time. 

They seem to have improved it some.

You don't have to wait for the statistics to update before you start adding more, though you do have to wait for the district to update before you make another change.

[jimrtex]

I see that jimrtex has posted a ward file for Cleveland on district builder. Are those the boundaries that should be used to meet the constitutional requirements?

Those match the boundaries on the Cuyahoga BOE web site (as far as I am able to do it).  If you notice, they have the wrong population (each has the total population of Cleveland).

What I was trying to do was renumber them.  I had drawn all the wards.  I would lock all the "districts", and then unlock one, desassign it geography, and then select Cleveland.  When you select an area that is partially locked, then only the unlocked area is selected.  I then assigned that to a new district with the correct number.  The assignment seems to work fine, and it calculates contiguity and compactness, but it gets the population for the entire selected area.

I am concerned that if you have created a district and locked it, and then are drawing the district on the other side of the boundary by lassoing across the boundary, that it may be calculating the wrong population, by a tens.  I don't see anything substantively different between selecting an entire city with most of it locked, and selecting a dozen blocks, with a few of them locked, other than the magnitude, and the selection tool.

Cleveland had 21 wards, and that is what the census bureau has for VTDs (actually they have the election precincts, which are numbered with the ward number and the letter.  If you recall from the previous contest, Ohio did not define VTD's for the 2000 census, and the people doing the contest ended up having to define them.  So it may be that they tried to do a better job.  The Cuyahoga BOE maps don't always match census blocks, so it appears that they still haven't got the counties to conform to census geography or talk the census bureau into making adjustments for the counties (this happens in areas where streets don't completely cut through a block, and so the census bureau tends to not like to use projections as block boundaries.

Fairly recently 2008(?), Cleveland reduced the size of its city council to 19, and drew the new wards.  So I'm pretty sure these wards were in place at the time of the 2010 census (see
§ 11.06 of Ohio Constitution, though it is possible the legislature has directed otherwise).  If you look at the ward maps, or alternatively, the house district maps, you will see that they are not 100% in conformance.  But it looks like the new districts were drawn in an attempt to match up.  Given that they are so new, they are not very equal in size.  This may be due to an impossible attempt to match 19 wards up to districts drawn to match 21 wards (there are no house districts entirely in Cleveland, so they each have just a few wards).  There are two house districts that include disjoint areas of Cleveland.  One uses Bratenahl as a connector, and the other uses Cuyahoga Heights and Brooklyn Heights.  There was perhaps an attempt to keep more black-majority wards, as they are generally less populated.

The charter also, now (2010(?), requires the size of the city council to vary based on the population, with a reduction of 2 wards for every 50,000 loss in population.  Cleveland is now in the 17-ward bracket (375,000 to 425,000, with a 2000 population of 396,000.   So they will have to further redraw the ward boundaries.   The 5 wards along the shoreline out towards Euclid will have to be reduced to 4, likely eliminating Ward 9 which is the lowest black VAP% in eastern Cleveland and has Case Western Reserve at its southern tip.  The other ward to go would be Ward 12 which is the ward that wraps around Cuyahoga Heights with the eastern half distributed towards 5 wards in southeast Cleveland, and the western half shifted toward the west.  The 3 central wards (3, 14, and 13) are right about on target for a 17-ward apportionment, so there will have to be an eastward shift of the wards as the population is gradually absorbed going west.

It really doesn't make sense to use wards as building blocks for districts.  It doesn't necessarily simplify administration of elections, since the wards change, so that there is not a total correspondence.    In addition, since wards have quite similar populations it may be difficult to combine in districts.  With an average population equivalent to 0.180 house districts, 5 wards equals 0.900 house districts, while 6 wards equal 1.080 house districts.  This helps explains while all the house districts end up going outside Cleveland to pick up towns, since these are of varying sizes.  It would be better to let larger cities (50,000+?) to delineate neighborhoods in the range of 5,000 to 25,000 which would be used for constructing districts.

I'm not sure that it is possible to draw two black majority VAP senate districts in Cuyahoga County.  If you take the entire black VAP and divide by 6/11 of the total VAP (6/11 is the share of the population contained in 6 of 11 house districts in the county), you get 52.0%.  That is, if the age distribution of the black population is the same as the total population and you totally disregarded geography, you get only get two senate districts with 52.% black VAP.  If the other 5 house districts went above 2.5% black it would be too much, and places like Solon, Lakewood, and Pepper Pike have 10%, 7%, and 6% VAP%, respectively.

The adult population relative to the total population has some variance, but not that much.   The wards with a high percentage of adults are Ward 3 including downtown; Ward 9 which includes Case Western Reserve, and Ward 19 on the far west, south of Lakewood.

The least adult is Ward 5 east of downtown, and which appears to have lots of housing projects.  The other young wards mostly in the western part of the city and appear to be associated with a Hispanic population.  If an area has a relatively low share of adults, then you can use the children to meet OMOV, and need fewer adults to get to 50% VAP.

I have 4 majority black VAP, 65.4%, 53,5%, 63.2%, 62.0% which are mainly east-west in a stack, with each extending into the suburbs.  I then have a north-south suburban district at 35.6%, which is trio-ed with the northern two districts, for a senate district at 51.5%.

The two southern districts are then trio-ed with a western district.  But that district is 12.7% black VAP, making the senate district 45.4% BVAP.  

I made one ward swap to make the western districts more equal in population (+2.0% and +3.2%, vs +0.7% and 4.5%)   If I switch those, then I can get the senate district up to 46.7% BVAP.  But, I really don't like pushing a district so close to 5.0% deviation unless I can avoid splitting something.  Or I could draw a different house district in the west running from west of downtown southwestward towards the airport, and get the senate district up to 48.0% BVAP.  But that would require redoing all the western suburban districts since it would isolate Lakewood.  And it would require another city be split, probably Parma.

The plan I have now has a Hispanic VAP% of 7.5%, so I am over 50% minority VAP.  Does that count for anything under the VRA?  Whether it counts for scoring in the contest is another question.

It might be feasible to get 5 majority black VAP house districts, in the mid-50s.  That would make the senate differential a bit worse.

I am inclined to leave it as drawn.  There are 7 wards that are over 85 BVAP%, so I think I did a reasonable job of avoiding a pack.

I just realized that since they have partisan data, they have to have mapping of blocks to election precincts.  The instructional video for Michigan said that they were using census tracts for one of the layers.  So what are the advantages of census blocks vs. VTDs, particularly since you can avoid making assumptions about weighting of voters, and the wards are comprised of blocks.

PS the North Olmsted annexation appears to have occurred in 2000, just a few months after the census.

390351109011007