US House Redistricting: Pennsylvania

[muon2]

It doesn't really matter. Murphy doesn't look like he's going to retire anytime soon.

Weren't we supposed to get a map yesterday?

Delayed till Wednesday.

They're checking numbers against Torie's.

[muon2]

A couple of quick observations:

The mappers went out of their way to avoid 3-way splits in the smaller counties. Other than Allegheny and Westmoreland, I didn't see any others in the north or west.

They want to eliminate Altmire and would rather face Critz. District 19 is eliminated and it is now 4. Changing numbers is a way to put an incumbent off balance, and Altmire comes up with a new one.

[muon2]

PA 06, 07, 08 and 16 have been drawn. The population for PA-07 is spot on, as is PA-08, but for inexplicable reasons for the moment, PA-06 is short 2,500 people, and PA-16 short 4,500.  All four have lower GOP PVI's than my projected map (understandable for PA-16 which takes Reading rather than PA-17, but no excuses for PA-06 and PA-07). The Lancaster legislators' stonewall hurt a bit.  PA-07 only got 32,000 residents out of Lancaster. The cherry picking of precincts in Montco, Chester, and Delaware was also a bit flawed in my opinion. Maybe 40 or 50 basis points were left on the table, maybe a tad more, due to that. PA-08 is down by 20 Pubbie basis points from my projected map.

I must say that PA-07 is indeed a horrendous monstrosity - just awful. A fair amount of that is again due to the Lancaster County stonewall, forcing PA-07 into Berks, and PA-16 in turn cutting more deeply into Chester with that ludicrous prong to the north in Chester.

MD-03 has a perfect partner in PA-07 as a poster child for redistricting reform. It looks like they were more concerned with holding PA-07 than with PA-06, based on the 2% boost it gets in comparison in your chart. Maybe there are other electoral factors at the local level.

[muon2]

I was surprised they kept Schuykill whole; all of Torie's maps had it split.

The GOP lives in terror of Holden is the only explanation. So many 60%-70% McCain precincts have gone to waste. Sad.

Or are they protecting him from a primary by maximizing his base? They may have decided some other Dem from that new district would be far worse than Holden.

[muon2]

I was surprised they kept Schuykill whole; all of Torie's maps had it split.

The GOP lives in terror of Holden is the only explanation. So many 60%-70% McCain precincts have gone to waste. Sad.

Or are they protecting him from a primary by maximizing his base? They may have decided some other Dem from that new district would be far worse than Holden.

I think they gave it to him to prevent Gerlach or Dent from having to pick it up but there was talk of a primary so this helps him in that area, too.

Yes, that is probably the reason, as you can see below. PA-17 just isn't that Dem. And maybe if Holden retires, the GOP hopes to get lucky, and have a redux of a Barletta-Kanjorski scenario. The area is probably trending a bit GOP anyway (outside Scranton which isn't). The other big news of course, is that the GOP chose to Pub up PA-15 much more than the published leaks, and Northhampton is stripped of a big slug of Dems in the Bethlehem suburbs in addition to Easton; the CD gets in exchange, not only a piece of Berks, but also the most GOP parts of Lebanon and Dauphin.  Maybe Dent told the map drawers that he doesn't plan to hang around in the House for the next decade.

Meanwhile we have trouble, trouble, right here in cartographic city! PA-17 is short by 13,000 people, while PA-11 has 5,400 too many.  I checked and checked the shapes of the CD, and just can't find any errors. I do see now that the 3 Dem pack CD's in the Philly area have 7,000 too many people (I summed the population deviations of the three CD's), which matches the 7,000 shortfall of PA-06 and PA-16, but having checked, and checked, and checked again, I just don't see any errors in the shape of my CD's as compared to the map. Everything matches exactly as far as I can see.

If anyone can identify the errors in my map, in particular PA-17, their prize will be to set my signature for a month (assuming that it is reasonably decorous of course; you don't want me to get my first infraction point do you?  )  Did the Pub operative who drew the map make some errors, or did I, that is the question.

So without further ado, here is the map for Eastern PA (with how the 3 Dem pack CD's, PA-01, 02 and 13 not done yet as to how they split their interior territory between them). I did a blowup of PA-17's perimeter so one can better match the published map.  If there is an error, it presumably is around Scranton or Wilkes-Barre. Again, I just don't see it, but maybe my old eyes are failing or something.

The PA GOP regional map does not agree with the statewide map. Compare Monroe in both. In the statewide map Stroudsburg and East Stroudsburg are entirely within PA-10 connected by a path through Stroud Twp. That area is in PA 17 in the regional map. The regional maps also use a different color scheme, which suggests they are from a different version. Since you followed the regional line in Monroe, I think you will find most of your lost population for PA 17 there.

edit: There's also a small piece to pick up Nazareth that is missing in your PA-17. Also, I see 17 stretch all the way up to Carbondale in the statewide map.

[muon2]

Muon2, my versions of the statewide map, and regional map, appear to match. Plus, PA-10 is already short a couple of thousand persons, so it has no surplus (unless it "lost" some territory per somebody's error in turn).  

Here is the statewide map:  

And the regional map:

And  a blowup of my map, which includes in PA-17 in Monroe everything that you described it seems:

So color me confused.  

Click on your statewide link. It isn't the same as your pic. I think it was changed overnight! Probably the regional map hasn't been updated yet.

[muon2]

Odd Muon2.  Anyhow,  my map has the version that gave PA-17 more territory, and it is still 13,000 short.

When I followed the "new" statewide map I got PA-17 within 2000. There may be other splits I haven't picked up yet.

[muon2]

What inspired that Dem map? The notion that since the 19th district is lost, the current 19th must be chopped and all other incumbents protected?

I want to know what the hell they were thinking by releasing it. Now they can't claim any moral high ground on gerrymandering.

I completely agree. They give the interested public no reason to prefer their version to the GOP offering. The OH Dems got good press out of the neutral plan they filed. It helps with their positioning on the referendum should it go forward.

[muon2]

Odd Muon2.  Anyhow,  my map has the version that gave PA-17 more territory, and it is still 13,000 short.

When I followed the "new" statewide map I got PA-17 within 2000. There may be other splits I haven't picked up yet.

The regional maps now appear to match the statewide.

[muon2]

Torie,

Here's my take on PA-17 from the GOP maps. You can see that with the Stroudsburg inclusion and by running up to Carbondale I get districts within 1000 of the ideal. Note that PA-17 also seems to have one precinct in the city of Bethlehem as well.

[muon2]

Torie,

Here's my take on PA-17 from the GOP maps. You can see that with the Stroudsburg inclusion and by running up to Carbondale I get districts within 1000 of the ideal. Note that PA-17 also seems to have one precinct in the city of Bethlehem as well.

Thanks Mike. Two precincts were taken in Bethlehem actually (Hispanics precincts, both of which voted 83.3% Obama).   That was an error I had made before; I was tricked by a very similar shape right next to it. A change in that area is that Nazareth was added to PA-17. Actually subtle changes to the map were made all over the place (including a revamp of the action around Reading, and doing cuts into municipalities to equalize population (PA-11 takes two in Harrisburg), and yes, they redid the layout of who gets what in the Scranton area in a rather major way). They also did the precinct cuts, since a few precincts don't match (including one in Tioga). Another looks to be in Lebanon City, to equalize out PA-11 which I still can't equalize. PA-17 now has 1,000 too many people in my map, but I suspect I see the precinct cut in Monroe.   

Isn't this fun?  Oh yes, at the moment in my map PA-06 is now short almost 11,000 people. I see a few precincts changes were made in the Philly suburbs - as well as precinct cuts. Yes, it's still morning in CA, but I need a drink! 

Here's my interpretation of SE PA. Some lines are pretty hard to discern. I do take 2 pcts out of Lebanon city, but my PA-06 is within 1K. See if you concur.

[muon2]

Torie,

Here's my take on PA-17 from the GOP maps. You can see that with the Stroudsburg inclusion and by running up to Carbondale I get districts within 1000 of the ideal. Note that PA-17 also seems to have one precinct in the city of Bethlehem as well.

Thanks Mike. Two precincts were taken in Bethlehem actually (Hispanics precincts, both of which voted 83.3% Obama).   That was an error I had made before; I was tricked by a very similar shape right next to it. A change in that area is that Nazareth was added to PA-17. Actually subtle changes to the map were made all over the place (including a revamp of the action around Reading, and doing cuts into municipalities to equalize population (PA-11 takes two in Harrisburg), and yes, they redid the layout of who gets what in the Scranton area in a rather major way). They also did the precinct cuts, since a few precincts don't match (including one in Tioga). Another looks to be in Lebanon City, to equalize out PA-11 which I still can't equalize. PA-17 now has 1,000 too many people in my map, but I suspect I see the precinct cut in Monroe.   

Isn't this fun?  Oh yes, at the moment in my map PA-06 is now short almost 11,000 people. I see a few precincts changes were made in the Philly suburbs - as well as precinct cuts. Yes, it's still morning in CA, but I need a drink! 

Here's my interpretation of SE PA. Some lines are pretty hard to discern. I do take 2 pcts out of Lebanon city, but my PA-06 is within 1K. See if you concur.

Nothing was attached.  Maybe you just decided to nuke SE PA (hey you're a physicist so you have the power man) since the SE PA map is such an excrescence.

take 2 ...

[muon2]

Can't can change PA-07 from that drab gray to darker blue or cyan or something muon2?  Think contrast my man. Opposites attract.  It's really bad where the grey appends PA-13's salmon color.

Oh, and a zoom would be nice too. Yes, I know I'm demanding. I have a license to be as it were.  

The zoom is there. Right click then view image works for me.

[muon2]

I've updated my map based on the Senate release of the municipal splits. Here's my map of the SE region with that update.

[muon2]

My God. They'd better take that map to the courts.

Yes, but given the decision in IL on the challenge there, on what ground will the Dems prevail? Here's a key part of this weeks ruling against the GOP on the Dems map in IL:

You must be logged in to read this quote.

Until a political party can show the court a clear defensible standard as to when partisan mapping goes too far, the courts will continue to reject the claims of partisan gerrymandering. The court would clearly like someone to find that standard, but all attempts have generally resulted in an arbitrary threshold that the court cannot justify.

[muon2]

When the 7th district looks like a fish having sex with a cat (they aren't even connected, mind you), it's a problem. When a tilt D state has a 2:1 GOP advantage in Congress, it's a problem.

I agree that the map is over the top. However, so is the IL Dem gerrymander. The federal court in IL couldn't find a reason to overturn the IL map, despite their concerns about it. I haven't heard a non-arbitrary standard that either party can state that will allow a court to toss a map like the GOP gerrymander in PA.

[muon2]

When the 7th district looks like a fish having sex with a cat (they aren't even connected, mind you), it's a problem. When a tilt D state has a 2:1 GOP advantage in Congress, it's a problem.

I agree that the map is over the top. However, so is the IL Dem gerrymander. The federal court in IL couldn't find a reason to overturn the IL map, despite their concerns about it. I haven't heard a non-arbitrary standard that either party can state that will allow a court to toss a map like the GOP gerrymander in PA.

The Illinois decision is notable not for its result, but it appears the plaintiffs actually game up with a standard to evaluate gerrymandering.

As the Committee frames
its issue, proving the required discriminatory effect would require a showing of three things: (1) that
the Adopted Map increases the number of districts that favor Democrats by at least 10 percent
according to an accepted measure of partisan voting, which the Committee proposes as PVI; (2) that
the Adopted Map keeps at least 10 percent more constituents of Democratic incumbents in the same
district as their representative than it does constituents of Republican incumbents; and (3) that at
least one of the districts created with the intent to advantage Democrats is among the districts that
contributes to the proof of elements 1 and 2.




It's both useless and unworkable, but at least it is numerical rather than 'I simply don't like it'.

Correct, so the challenge to the PA Dems is to come up with something better.

[muon2]

When the 7th district looks like a fish having sex with a cat (they aren't even connected, mind you), it's a problem. When a tilt D state has a 2:1 GOP advantage in Congress, it's a problem.

I've thought about this and after you remove the 2 or 3 districts in Philadelphia and the Pittsburgh district, there really isn't much left for Dems to have too many more districts.  I mean those districts are like beteen 90-10 to 75-25 for the Dems while the GOP ones are like 60-40 (on average), therefore, the numbers seem more out of whack than they are for Congressional reistricting.  What map would you propose that is actually clean?

Here is the neutral map I proposed last month. I used Torie's method for converting 2008 presidential numbers to PVI, and then count PVI's within 1% as even and within 5% as lean. This gives me 8 R - 1 lean R (3) - 2 even (6, 8 ) - 2 lean D (11, 15)  - 5 D. The concentration of Ds in Philly does tilt the map more R than D, with an effective partisan index of 56.6% R when the statewide index is 48.5% R.

I don't know if this was noticed before, but the seven SE counties are almost exactly the size of 7 CDs.

My (2nd and 3rd) maps on page 15 use that.

Very good. I assume that if I want to sustain the deviation greater than 1 in those districts, I need to show that I had followed a dictum to keep as many whole counties or districts within a county as possible. Hence our differences in how to split those seven counties. Other than Philly and Montco which must have splits after creating as many districts entirely within, I only split Chester.

In the meantime I've completed the rest of the state based on the principle of county integrity.

I found that six districts fit almost exactly in the western half of the state, plus Tioga and Bradford, form six districts with a deficit of only 372 persons total. Allegheny and Butler together are within 0.3% of two districts. The other four districts are each within 1.4% of the ideal population if rounded to whole counties. Other than Allegheny, only two counties are divided to bring the districts to with 100 persons. As in the SE no more than one township is divided between any two districts.

The remaining five districts in the east would be within 2.8% of the ideal size if rounded to whole counties. Three of the districts (11, 15, and 17) are within 0.9%. Three counties are divided to bring these five districts to with 100 persons, and only one county subdivision is split between any two districts. CD 10 isn't very pretty, but both CD 11 and 15 are very close to exact with the three counties that make up each district and the western division constrains the rest of CD 10.



For those interested in the political breakdown - here are the 2008 stats:

CD 1: Obama 89.9% - 9.7%
CD 2: Obama 79.7% - 19.7%
CD 3: Obama 51.0% - 47.6%
CD 4: McCain 54.9% - 44.1%
CD 5: McCain 55.1% - 43.9%
CD 6: Obama 53.5% - 45.3%
CD 7: Obama 61.4% - 37.7%
CD 8: Obama 53.7% - 45.2%
CD 9: McCain 58.6% - 40.3%
CD 10: McCain 55.7% - 43.0%
CD 11: Obama 57.6% - 41.4%
CD 12: McCain 56.2% - 42.5%
CD 13: Obama 58.8% - 40.3%
CD 14: Obama 64.0% - 35.0%
CD 15: Obama 55.2% - 44.8%
CD 16: McCain 52.1% - 47.1%
CD 17: McCain 53.3% - 45.6%
CD 18: McCain 54.0% - 44.7%

[muon2]

Isn't my matrix chart just awesome Mike?  I should copyright it! 

The scientist in me is curious as to how accurate the 2008 to PVI conversion is in general. Have you tested it on a state with both available?

[muon2]

Isn't my matrix chart just awesome Mike?  I should copyright it! 

The scientist in me is curious as to how accurate the 2008 to PVI conversion is in general. Have you tested it on a state with both available?

The other part of "both" being what?  The short answer is no, but I do adjust on a state by state basis, and sometimes for particular regions, looking at a wider range of data (e.g., AZ of course, and I did it with OH-06 as well as a mental exercise). I don't trust state office races for this exercise by the way. My bragging about the chart was more about its design.  I like good design. It is in my genes. My Dad when he was in advertising was awesome at it.

Oh, I also look at the trends from 2004 to 2008, and ask myself why. Well the GOP trend in SW PA hold, or substantially hold,  for example, is a question I ask myself.

I decided to make a quick check against my neutral map. In particular, my CD 9 and CD 11 are each almost exactly 3 counties and I can look up the county totals from our favorite Atlas. Using the conventional PVI they are D+4.5 and R+14.0. From the Torie approximation they are D+4.5 and R+13.0. The difference reflects the shifts from 2004 to 2008 in the baseline.

[muon2]

Altmire loves the Pub gerrymander as well, and is lobbying for it, which is understandable, since he will most likely put Critz away with this map, and has a fighting chance of getting re-elected. So now we have the "traitorous quartet."

Oh, I see Phil has already noted it. The gladiators in PA are not very disciplined are they?  

Altmire sounds like Holden redux.

[muon2]

Do they need a two thirds majority or what was that about needing votes?

Also, is there, like, a party breakdown?

It didn't need two thirds. They were just worried about it not passing because quite a few Republicans were opposed. I don't have a party breakdown yet.

What was behind the GOP opposition?