Is it possible to ban gerrymandering?

[muon2]

First, there needs to be a set of clear geographic rules such as restrictions on county and municipal splits, beyond the minimum needed to meet federal law. Keeping split governments to one per pair of districts is probably as good a benchmark as any.

Second there needs to be a test to measure whether the map is too partisan. This has been the issue on which the court has yet to agree, though a clear majority believes that a test is desirable.

I would suggest that a simple test that would measure the PVI for the state and for each district. The state's PVI would determine the majority party for the state. Then count each district with a PVI over 5 as a whole seat for that party and over 1 as a half seat for the party. PVI's of 1 or 0 don't count for either side. Find the difference between the majority party's seats and the minority party's seats, divide by the total number of districts, and subtract the percentage PVI for the state. That's the measure of the map's partisan bias.

One expects any map to have some partisan impact, but not to dominate over other redistricting principles. To make a statistical argument that partisan gerrymandering has gone to far, I would multiply the partisan bias percentage by the number of districts in the state and square it. If that number is larger than the number of districts the partisan gerrymander is dominating other effects.

For example let's use the new WA congressional map. The state has a PVI of D+5. According to Krazen's post

PVIs from RRH.

1. D+1
2. D+7
3. R+3
4. R+13
5. R+6
6. D+5
7. D+29
8. R+3
9. D+16
10. D+5

There are 4 seats for the Dems (3+2 halves) and 3 seats for the GOP (2+2 halves) for a partisan difference of D+1. That translates to a partisan bias factor of 5% (1*10%-5%) which isn't much. The gerrymander test gives 0.5 squared or 0.25 which is far less than the 10 seats.

For IL the state PVI is D+8 and the result of the map is partisan difference of D+7.5 seats or a partisan bias factor of 34%. The gerrymander test gives 37 which is much larger than the 18 districts. It's fair to say that partisan concerns dominated the IL map, which is what the federal court said, but couldn't test.

As other PVI's come in I'd be happy to test them as well.

[muon2]

I don't believe in purposely gerrymandering a strong seat for a party. But I also don't believe in purposely gerrymandering swing seats.

But there should also be something to prevent a party from creating no swing seats, much like the current CA congressional map. Some number of swing seats insure that the legislature can change with the interests of the public.

[muon2]

I've dug up a couple more states and modified my rounding rules to be consistent. Most of the PVI's reported are in whole numbers so a D+5.2 = D+5. I'll round the PVI to the nearest percent before using it to determine the status of a district. That matters for instance in IL where 3 districts are between D+5.0 and D+5.5.

I can also measure the swing seat composition of a map with a similar technique. Add half the lean districts (PVI 2-5) to the even seats (PVI 0-1) to get a total. The swing seat factor (SSF) is the square of that total should be roughly equal to the number of districts in the state if the map has neither too few nor too many swing districts.

IL (18): Partisan Difference D+6, Partisan Bias 25.6%, PBF 21.2, Swing seats 4, SSF 16.
NJ (12): PD D+2.5, PBI 16.4%, PBF 3.9, Swing seats 2.5, SSF 6.
PA (18): PD R+4.5, PBI 27.0%, PBF 23.6, Swing seats 4.5, SSF 20.
WA (10): PD D+1, PBI 5.0%, PBF 0.3, Swing seats 3, SSF 9.

What this would suggest is that IL and PA are partisan gerrymanders, and NJ is fair but leans towards incumbent protection. WA passes this partisan test, though it could have other problems on purely geographical factors like county integrity.

[muon2]

I've dug up a couple more states and modified my rounding rules to be consistent. Most of the PVI's reported are in whole numbers so a D+5.2 = D+5. I'll round the PVI to the nearest percent before using it to determine the status of a district. That matters for instance in IL where 3 districts are between D+5.0 and D+5.5.

I can also measure the swing seat composition of a map with a similar technique. Add half the lean districts (PVI 2-5) to the even seats (PVI 0-1) to get a total. The swing seat factor (SSF) is the square of that total should be roughly equal to the number of districts in the state if the map has neither too few nor too many swing districts.

IL (18): Partisan Difference D+6, Partisan Bias 25.6%, PBF 21.2, Swing seats 4, SSF 16.
NJ (12): PD D+2.5, PBI 16.4%, PBF 3.9, Swing seats 2.5, SSF 6.
PA (18): PD R+4.5, PBI 27.0%, PBF 23.6, Swing seats 4.5, SSF 20.
WA (10): PD D+1, PBI 5.0%, PBF 0.3, Swing seats 3, SSF 9.

What this would suggest is that IL and PA are partisan gerrymanders, and NJ is fair but leans towards incumbent protection. WA passes this partisan test, though it could have other problems on purely geographical factors like county integrity.

How would this work with Massachusetts and its 0 swing seats?

As I understand it these are the Massachusetts PVIs.

1: D+14
2: D+12
3: D+8
4: D+11
5: D+15
6: D+6
7: D+30
8: D+8
9: D+7

Massachusetts has some seats which would be swing seats by PVI but are solid Dem because of the total lack of quality GOP candidates.

Actually none of the districts in MA would be swing seats based on my criteria, since the best is D+6. The formula would give a PBF of 63 and a SSF of 0. MA will always be a partisan map since there are no compact areas for a district with a majority of GOPs. However, the map could have been drawn with a couple of swing seats (MA-6 and 9 would probably be the easiest to shift) to lower the PBF and increase the SSF. It would still be partisan, but not such much incumbent protection.

[muon2]

The main thing is to take redistricting out of the hands of partisan legislatures and into the hands of procedurally neutral commissions of some sort. It's a basic democratic principle: the management of the election shouldn't be in the hands of one of the parties competing in it. I think this is much more important than developing an exact mathematical definition of a "fair" map.

I can understand the existence of criticisms of the various commission-drawn maps, but these criticisms only make sense against the background of a much higher expectation of fairness for the commissions. Any of AZ, WA or NJ would be considered at most a very mild gerrymander if they were instituted by a legislature of the party on whom they are supposed to confer advantage.

Better still is to have the commission sort from publicly generated maps. The MN contest drew 500 submissions after the OH contest had 100 entries. The public can draw good maps, and with some guidelines as to the goals from the commission then the effects of any commission bias can be reduced.

[muon2]

The main thing is to take redistricting out of the hands of partisan legislatures and into the hands of procedurally neutral commissions of some sort. It's a basic democratic principle: the management of the election shouldn't be in the hands of one of the parties competing in it. I think this is much more important than developing an exact mathematical definition of a "fair" map.

I can understand the existence of criticisms of the various commission-drawn maps, but these criticisms only make sense against the background of a much higher expectation of fairness for the commissions. Any of AZ, WA or NJ would be considered at most a very mild gerrymander if they were instituted by a legislature of the party on whom they are supposed to confer advantage.

Better still is to have the commission sort from publicly generated maps. The MN contest drew 500 submissions after the OH contest had 100 entries. The public can draw good maps, and with some guidelines as to the goals from the commission then the effects of any commission bias can be reduced.

Better still to have the public determine the maps.  It is legitimate for the public to have a bias and act upon that bias.

Agreed, but there is a role for an outside group to set clear standards so that public maps can be compared. A commission can perform that task.

[muon2]

Of course not.

You can craft an expansive set of rules restricting what maps are legal and what maps are not, but as long as there is more than one possible legal map that can be drawn, the politicians will choose that which they deem most beneficial. You can outsource the entire process to a commission, but in the end the commission will also draw the map they find most pleasing for whatever reason. Often this will result in a swing-seat gerrymander or some sort of bargain they deem "fair". But either way it's still drawn with a purpose of achieving some end. Gerrymandering is an unavoidable evil within our system.

I would not consider that to be gerrymandering with the usual definition. Gerrymandering means to manipulate the boundaries to favor one party or group (such as race). Drawing maps with rules such as county and municipality integrity and compactness are not gerrymandering. Is is possible to gerrymander given a set of rules - certainly and it happens all the time. However, the more specific the rules the less one can manipulate boundaries at will.

Furthermore if one sets rules in advance, then provide for a different map maker than ruler maker, it creates additional constraints on gerrymandering. It's like the classic solution to splitting a cupcake between two children; let one divide the cupcake and the other have first pick of the halves.