NY with 26 CDs in 2020

[muon2]

Torie's map has some attractive features, but I want to follow up on train's observation on the Rochester UCC. The UCC consists of Monroe and Ontario, and my projection puts their combined population and 870K which is about 12% more than the population of one of 26 CDs. If the UCC expands at all this decade, Wayne is the most likely candidate since the Rochester UCC is already in that county and it wouldn't take much to link up the Palmyra and Newark urban clusters along NY-31.

With that in mind I intentionally put both Ontario and Wayne in the Syracuse district. I slightly overpopulated the Syracuse CD with the intention of taking population from Ontario in that district to beef up the slightly undersized Rochester CD. That way I could keep the UCC cover and pack rules intact there with a minimum of chops. The initial Torie plan would incur both a cover and pack penalty for Rochester.

Note that since a single chop of Ontario in my plan would bring both the Rochester and Syracuse CDs within 0.5% of the projected quota. Similarly chops of just Erie and Steuben are sufficient to bring the other western CDs within 0.5%. In the Hudson valley I selected the counties such that chops in Franklin and Sullivan are sufficient to bring those four CDs within the 0.5% margin. Of the aforementioned chops, only Erie is a macrochop.

[muon2]

I suspect Torie and I have reversed the roles we played earlier discussing IL when he complained about some liberties I took regarding the rules due to my preference in dealing with areas in Cook. Here I'm pushing to follow the rules, yet I sense from Torie a desire to recognize political expectations in NY.

Anyway, here is my attempt to provide more detail to the upstate districts. I preserve county subdivisions, but I'm not using estimates for the county subdivisions - instead I'm apportioning the county-level projections to the subdivisions on a more uniform basis. It's less accurate, but faster that way. All the districts should be within 0.5% when projected to 2020.

I preserve all UCC cover and pack rules, except the Albany pack. In exchange there are only 6 chops and none are macrochops except the mandatory chops of Erie and Westchester. None of my districts are unusually erose, so I expect a good score there. I don't see why one wants 9 chops (counting both fragments connecting Westchester to upstate) just to avoid the penalty for the Albany pack. Also, by placing Columbia with Dutchess and a piece of Ulster with Orange Torie's map incurs a pack penalty for the NYC UCC, which negates the avoidance of a penalty in the Albany UCC.

[muon2]

Westchester is growing at 0.6% per year so far this decade. Projecting out to 2020 adds 56 K in the county that aren't in DRA. Your CD 18 has Dutchess (295K) and Putnam (99K) leaving 385K from Westchester. 385K/1005K is 38%. 38% of 56K is 21K so with a uniform spread of the growth as I described, DRA should show CD 18 as 21K low with its 2010 data. DRA shows it 18.6K low for a difference of 2.4K. That's within 0.5% so my CD is ok for 2020.

Your chopped Delaware has become a bridge between whole counties which is forbidden under the rules.

Suffolk (projected 1516K) will be underpopulated for 2 CDs in 2020 by 43K, but it looks like you cut into it from Nassau.

You mention microchops, but I thought we abandoned them in part at your request.

[muon2]

When I add up all the projected county pops (20,269K) and divide by 26 I get 779.6K for the quota in 2020. That's the value I use to get the leftover population in Westchester. I get Dutchess+Putnam+Westchester is 159K short of 2 CDs, so that has to come out of the Bronx.

The two Buffalo UCC CDs as we both drew them project to be 9.4K short of the quota based on whole counties. That requires a chop out like mine into Steuben.

The UCC was originally designed to create a cover rule. It is only in the last few months that the notion of a pack requirement was put in to match. A pack rule by itself will tend to be a boon for the Pubs by driving the urban population into as few districts as possible. The cover rule can benefit either party depending on the size of the UCC. Without the cover rule I see no purpose to use UCCs at all.

[muon2]

When I add up all the projected county pops (20,269K) and divide by 26 I get 779.6K for the quota in 2020. That's the value I use to get the leftover population in Westchester. I get Dutchess+Putnam+Westchester is 159K short of 2 CDs, so that has to come out of the Bronx.

The two Buffalo UCC CDs as we both drew them project to be 9.4K short of the quota based on whole counties. That requires a chop out like mine into Steuben.

The UCC was originally designed to create a cover rule. It is only in the last few months that the notion of a pack requirement was put in to match. A pack rule by itself will tend to be a boon for the Pubs by driving the urban population into as few districts as possible. The cover rule can benefit either party depending on the size of the UCC. Without the cover rule I see no purpose to use UCCs at all.

Below are my numbers FWIW. Did you use April 1, 2010 and April 1, 2020 as the bracket dates for the population change extrapolations?

Regarding your comments on the cover and pack rules for UCC's, sometimes the pack rule helps the Dems, as in Kansas certainly, and perhaps Nebraska. The pack rule would help the Pubs where a metro area as a whole is more Dem than the hinterlands. The Pubs used to chop Rochester and Columbus with regularity, and if that is used as the county chop without penalty, it would not hurt the map score that much. With the pack rule, ignoring the cover rule, can only cause mischief up to about half the population of a UCC. For myself, what is important is the size of the "violation" of the cover rule. Minor violations, like I did when NY-17 took in about 17,500 people in Ulster, I don't think violates the spirit of the rule to the extent it should be borne in mind. That is why I wanted to base the cover rule on macrochop increments, with a chop less than a microchip having no penalty. Anyway, I could also hew to the cover rule, by having NY-18 chop into Greene County, with NY-17 getting out of Ulster by chopping more deeply into Westchester County. But that makes for a much more problematical map, which will never, ever be drawn.

This assumes of course that my numbers are accurate, which perhaps they are not. If not, and yours are, I suspect that NY-17 need not chop into Ulster in all events, or it will be very close. This is all projection anyway. Given the 17K is based on what is going on in the state as a whole, and how much upstate is stagnant or losing population vis a vis how much the NYC area is growing, the odds are really about 50-50 at this point either way.

My main purpose in doing this actually, is to submit a 26 and 27 CD map to the local press, because it will be of interest that our CD is slated to undergo massive changes in the next census, due to the unusual shape of the state. Thus I was focusing on what would seem most likely to be drawn, absent the Dems entirely controlling the process, or some weirdo bipartisan gerrymander. What I think is most likely to be drawn, is my map (putting aside the Rochester issue), as amended by Train, to keep NY-18 on the east side of the Hudson River. That is much more likely than NY-18 crossing over into Orange County. NY has a long tradition of keeping Orange and Rockland Counties together, and it makes sense.

I take the 4/1/10 numbers and the 7/1/14 estimate and I calculate a constant growth rate for each county (not linear population change). I then project that to 4/1/20 using the calculated rate. I emailed you my population file a couple days ago. That has the numbers I'm using.

[muon2]

One problem with projections this far out is that projections based on the whole will differ from projections based on the sum. If I take the state estimate released for 2014 it equals the sum of the county estimates for 2014 as one would expect. The sum of my county projections to 2020 gives 20,269,018 (different because I round the estimate to 4.25 years to account for the 3 months from 4/1 to 7/1). But the state projection to 2020 using my same formula gives 20,300,327. Fortunately that's only a 0.15% discrepancy and the estimate shifts from year to year are much larger, so one can treat the sum of county projections as if they are the state projection.

[muon2]

I get your motivation now. Political realities are important, and it would be interesting to see how something like this fared on the Pareto test. Could it make the cut to go to the finals where it should fare well?

If the populations are equal then your chop of Ulster should be about 10K. The NYC UCC + Sullivan is only 3K over the population of 18 CDs. That's why it became the basis for my plan. Since Columbia projects to have 13K less than Sullivan, that leaves 10K to come from Ulster.

I still don't see how you chop into Suffolk when it is underpopulated for 2 CDs. The chop should be from Suffolk into Nassau.

[muon2]

I added NY-11 and NY-08 to the map, and tweaked the lines in NYC to follow ward boundaries where possible. Around NY-08 that was not possible to do entirely, since the ward boundaries there are so erose between my NY-08 and NY-10. NY-08 does not have the old PVI numbers on the chart, because the court did not draw a south Brooklyn seat in remotely the way that I drew it (the court screwed the Pubs out of a seat in south Brooklyn), so the comparison would be meaningless.

The skew using the 2008 numbers for the CD’s, but the 2012 numbers for the state PVI (10% Dem), is zero. Since NY trended a couple of points Dem in 2012, if the 2008 state PVI were used (about 8% Dem), the skew would be 1 in favor of the Dems.  It’s probably more skewed using the 2012 numbers for everything, given most areas of the state trended Dem, but that is not possible to calculate given Dave’s application uses the 2008 election numbers.  In Democratic states, the skew should typically be lower than a more Pub state, since the exponential rise in Dem seats as the PVI moves their way, would tend to offset the natural Pub geographic advantage that exists in most states when drawing CD lines hewing to jurisdictional boundaries.

To answer Train's question, I think the correct standard is to get up to 50% BVAP (but no more if securing a higher percentage involves racial gerrymandering, because as SCOTUS has just reminded us, that is illegal racial packing), if the minority community is contiguous, unless it costs them a seat within a contiguous area, and then you shave the percentage down to see if the second seat can be preserved. The subject CD probably only has maybe 12% HCVAP, so shaving it down much in a zone where the neighboring areas are not black minority friendly would be problematical anyway. Keeping the CD all in Queens will drop the BVAP percentage circa the 2010 census, down to about 46% FWIW. I suspect the percentage is lower now for the portion of the CD within Queens. My impression is that the black population change there is sluggish or declining. So it may arguably be necessary for the CD to jut more into Nassau which is where the black population is migrating (to the extent it is not decamping from the NYC area entirely). Also FWIW, the court when it drew the map for this CD, saw it my way.

 
 

It looks like you chop towns in Westchester. I assumed that villages were like in MI, and subsidiary to towns that are the appropriate entity to keep intact. Is your experience there telling you that towns are less important than villages?

Also, do you have HVAP and BVAP numbers for your NYC districts. I put together a plan minimizing borough chops and I want to see how it compares.

[muon2]

Everything south of North Castle and Greenburgh is a good fit for 16 CDs. You have chopped Greenburgh to peel off the river villages. I like the line keeping all of Greenburgh in the north. That leaves the question of where to get the extra pop for the Rockland/Orange CD. I'll look at the options.

[muon2]

FTR, I'm with train that the community board boundaries are the most accepted, though one could also use the neighborhood tabulation areas as a smaller division, and are built from census areas consistent with the 55 PUMAs in NYC.

Anyway, Given Torie's work I thought I would post my estimates of how big a 2020 CD in the NYC area should appear using the DRA 2010 data. To get the number I use 780K as the 2020 CD quota, then assume uniform growth within the county. This is the average target number one goes for with 26 CD plan for a CD entirely in that county. For a CD that spans counties the DRA size would be the weighted average.

Bronx 714K
Kings 701K
Nassau 755K
New York 725K
Queens 710K
(Richmond 762K)
Suffolk 768K
Westchester 737K

For example the Staten Island CD in 2020 has 480K in Richmond and 300K in Kings (2020 population). Averaging (480*762+300*701)/780 = 739. So that CD should be about 739K in population as seen on DRA.

[muon2]

I pulled the ACS racial and ethnic percentages for some of the NYC counties to look at how the VRA might affect city CDs in 2020. The table is based on the groupings used in DRA and uses the total population, not VAP. The first number is the percentage from 2011-2013 and the second is the change since 2008-2010.

Group	Bronx	Kings	Nassau	New York	Queens
White	10.6: -0.5	35.8: +0.1	63.8: -2.4	47.5: -0.4	26.7: -1.3
Black	29.7: -0.8	31.6: -0.8	10.8: +0.3	12.8: -0.2	17.5: -0.3
Asian	3.5: +0.0	11.0: +0.6	8.1: +0.5	11.3: +0.2	23.7: +0.5
Latino	54.3: +1.2	10.7: -0.1	15.3: +1.1	25.7: +0.1	27.8: +0.5

The decline of the black percentage in Kings and Queens indicates that to maintain a 50% BVAP CD, one needs a higher number using 2010 data in DRA. In particular, this suggests that the South Jamaica CD may have to reach into more of Nassau where the black percentage is growing.

The growth of the Latino population in Bronx and Queens indicates that a CD evaluated with 2010 data on DRA will likely have a higher HVAP in 2020 than shows up on the app. The slight decline in the Kings Latino fraction will make it harder to draw the traditional CD that loops down to Red Hook through Brooklyn. That suggests a CD linking Bushwick to Corona will work better, leaving two other Latino CDs in the Bronx plus the northern tip of Manhattan.

[muon2]

Here's may take on the NYC metro. I used my various projection tables to get populations appropriate to 2020. I avoided all extraneous county chops with the only exception for Nassau-Queens to accommodate the BVAP majority CD. Within NYC I used community district boundaries when not otherwise constrained by VRA districts or population.

The 6 VRA CDs have the following VAPs using 2010 data.

CD 5: BVAP 50.8%
CD 8: BVAP 53.1%
CD 9: BVAP 52.8%

CD 7: HVAP 55.0%
CD 14: HVAP 58.1%
CD 15: HVAP 56.5%

Because the black population in Brooklyn is growing at a much slower rate than the overall borough population I estimate that to keep CD 8 and 9 over 50% they will need to show at least 52% BVAP in the 2010 numbers. The difference in Queens isn't as much, but the South Jamaica CD will need at least 50.5% BVAP in 2010 numbers to stay at 50% in 2020. That leads me to link all the way out to Hempstead to get sufficient black population. Because of that connection, I decided chopping towns and keeping villages intact was preferable to the nasty erosity of trying to wrap CD 4 all the way around the CD 5 peninsula into Nassau.

The one nasty piece of erosity I have left is CD 6. The shape of CD 7 and borough constraints forced me into the link through LaGuardia along the Grand Central Pkwy to connect Astoria to Forest Hills. I like my Bronx CDs, which I think track the neighborhoods quite well (though I may be out of date in my understanding), but it pins CD 6. I'm open to ideas about all that. I'm also curious to get train's take on my splits around his home.

At some point I'll post my revised upstate with an Albany pack for Torie but no extra Westchester chop across the Hudson. In any case, I presume one can marry this to any plan for the upstate CDs.

[muon2]

I like much of the new Torie concept as it mirrors my own. I take it you agree with the needed erosity in CD 6. I am concerned with the BVAP majority CDs. I suspect that 2 of the three won't be over 50% in 2020.

Because the black population in Brooklyn is growing at a much slower rate than the overall borough population I estimate that to keep CD 8 and 9 over 50% they will need to show at least 52% BVAP in the 2010 numbers. The difference in Queens isn't as much, but the South Jamaica CD will need at least 50.5% BVAP in 2010 numbers to stay at 50% in 2020. That leads me to link all the way out to Hempstead to get sufficient black population. Because of that connection, I decided chopping towns and keeping villages intact was preferable to the nasty erosity of trying to wrap CD 4 all the way around the CD 5 peninsula into Nassau.

[muon2]

I'm assuming that this plan is designed to potentially show people off-Atlas what might come to pass in 2020. It's entirely possible that a Dem legislature/gov would consider CDs with BVAPs under 50% that would still be likely to elect the black candidate of choice. I'm less sure that a court would make that decision, nor am I certain a neutral commission would either when a 50%+1 option exists.

I think it's much easier to say this is what a BVAP 50% CD would look like in 2020, then to explain why a sub 50%, but high 40's CD is likely to be drawn. If that's the case subtract 2.0% from the BVAP in Kings and 0.5% from the BVAP in Queens when describing the districts.

[muon2]

Defining shape will be a challenge here. There's really only a one town hop in my version of CD 5 and it gets to 50.8% with 2010 numbers - about 50.3% with 2020 estimates. Since this isn't adhering to any specific rules about hops, the shape is certainly no worse than the Queens-based Hispanic CD. I agree that a sub-50% district will elect a candidate of choice in that area, but would anyone but the Dems draw it?

I also return to my question about the goal of this plan. Given the shape of the aforementioned Queens Hispanic CD, is it appropriate to then say one doesn't like the potential shape of a 50% BVAP CD so the percentages were reduced? I think there would be merit in pointing out what could come to pass if the 50% BVAP standard is used. It seemed like pointing out the possibility for Hudson's CD was a goal of the description Upstate, so why not push that same line in NYC?

[muon2]

My understanding of Gingles is that if you don't have 50% in an area, no VRA district is required. So presumably if a black VRA district is required in Queens in addition to the two in Brooklyn, then there is a finding that a suitably "compact" area exists with the population of a CD that is at least 50% for a single minority (compact often loosely construed). My observation is that once there is that finding, 50% BVAP becomes something of a safe harbor. When a map goes below 50% BVAP and is challenged in court then the parties trot out their dueling experts to determine whether or not the sub-50% district can in fact perform for the minority. When courts are called upon to draw the map and a VRA district is required, they tend to go for the safe harbor and avoid that type of expert analysis.

I'm not saying that either of us would like such a district. I am saying that the track record of court-drawn VRA district tends towards meeting the population goals of the Gingles test. I am also saying that if you want to show your plan to NY bloggers (or other interested parties), then it is easiest to justify a district that is at 50%, since we haven't actually done the ecological inference or other techniques needed to demonstrate minority performance. I also think that if one is showing the plan to other parties, then there is utility in pointing out that if one wants to maintain 50% it will require a CD more strangely shaped than the area is used to.

[muon2]

Yeah, I'm quite confident that you could just let the southern Queens district sit somewhere in the mid-40s, entirely within Queens, and it would still be secure for Meeks/kosher for the VRA– use the current Illinois districts as precedent (IIRC a lot of the non-white non-black vote around the fringes of that area is first-gen immigrants anyway, so it might even remain 50%+1 BCVAP).  At some point I will get on making a map that does just that.

If it turns out to be 50%+ BCVAP then the problem would be solved. If all parties to a contest agree on some reduced standard, then that's fine, too. In IL the use of sub-50% won in the 7th circuit, but only after competing testimony from experts. A neutral projected map should not rely on a successful expert presentation IMO, despite the unpleasant result. Which goes to my hanging question as to whether this is being designed to show beyond Atlas.

BTW train, did you have any feedback on my partition of Brooklyn?

[muon2]

I'm sold on the BCVAP argument. I did some checking and found the following VAP citizenship rates by race/ethnicity using the 2013 3-year ACS:

Bronx: WCVAP/WVAP 90%, BCVAP/BVAP 81%, ACVAP/AVAP 62%, HCVAP/HVAP 72%.
Kings: WCVAP/WVAP 89%, BCVAP/BVAP 82%, ACVAP/AVAP 62%, HCVAP/HVAP 68%.
Queens: WCVAP/WVAP 90%, BCVAP/BVAP 85%, ACVAP/AVAP 61%, HCVAP/HVAP 61%.

If I apply these numbers to Torie's CD-05 (assuming the difference is WVAP) then 50.0% BVAP rises to 54.2% BCVAP. That's well above any number needed to accommodate the slower growth of the black population.

I can apply this to the Hispanic CDs as well. Your Queens CD (Torie 12, muon2 07) is 50.6% HCVAP. Our South Bronx CD 15 is 51.5% HCVAP; it's good, too.

This raises the question as to whether the 2020 Census will ask the citizenship question since it will be required for redistricting. If they don't one can't get an exact number and it is delayed as well.

[muon2]

Earlier I posted numbers by county for the growth rates by race/ethnicity, and for the CVAP/VAP ratios. I put those together and built a spreadsheet calculator to get projected 2020 CVAPs. I input the 6 VAP numbers from DRA and it gives me either the BCVAP of HCVAp for the district adjusted for the county. I assume that the Native VAP population is 100% citizens and the Other population has the same citizenship and growth rates as the Asian population.

I used that tool to adjust my map and determine the projected 2020 CVAPs. It turns out that one can put the Jamaica CD entirely within Queens and just break 50% BCVAP. Who knew? In any case it allows me to construct a map with the minimum number of county chops and to follow community district lines within the boroughs when possible. Here are the relevant 2010 VAPs and 2020 CVAPs for the VRA CDs.

CD 05: BVAP 46.3%, BCVAP 50.0%
CD 08: BVAP 52.9%, BCVAP 52.4%
CD 09: BVAP 52.8%, BCVAP 50.9%

CD 07: HVAP 54.9%, HCVAP 52.3%
CD 14: HVAP 58.1%, HCVAP 59.3%
CD 15: HVAP 56.5%, HCVAP 57.8%

For the record the 2010 AVAP in CD 6 is 31.1% which projects to a 28.6% ACVAP in 2020.