Mid-2014 county population estimates out tomorrow, March 26

[muon2]

Here's a round of projections to the states of the western Great Plains. Neither of the Dakotas will rate a second seat any time soon. However, due to the oil boom the combined population is projected to be slightly less than that of WV in 2020. The other states in this set also project to keep their same CD count.

NE has one multicounty UCC for Omaha and its two counties project to have 1.18 CDs in 2020. To keep whole counties they are split between two counties keeping the UCC cover rule, but ignoring the pack rule given the lack of projection for the county subdivisions. The districts are within 0.5% of the quota, not that I expect that to hold up given the precision of these projections. It's just that easy to get close given the large number of counties with small populations.



KS also has one multicounty UCC for KC and its two counties project to have 1.055 CDs in 2020. As in NE I'll keep the cover rule with whole counties. Again with so many small counties all the districts are within 0.5% of the quota.



OK has two multicounty UCCs, one for OKC and the other for Tulsa. The three county Tulsa UCC
is projected to have 1.006 CDs in 2020 so it is left as a single district. The three county OKC UCC is projected to have 1.57 CDs in 2020 and Oklahoma county is only about 3% over the projected quota. The remaining cover of the OKC district is left slightly under population, so that the remaining districts are all within 1% of the projected quota.

[muon2]

I don't understand why you ignored the pack rule, but whatever. You can't sever Wyandotte from Johnson County.

Using the pack rule in places like KS and NE guarantees a county chop (in fact a macrochop). The pack penalty and county chop score the same so it should come down to the question of erosity. Calculating erosity in a macrochopped county requires knowledge of the populations of the county subdivisions, since the shape of the chop affects erosity. Thus without county subdivision projections I'll take the penalty point I know versus the point I don't.

Perhaps we can revisit this in June when subdivision projections are released to match the county data. It also requires producing the muni-level maps to assess erosity, which I don't have at present.

[muon2]

Anyone planning on cranking the NY county estimates? If NY loses a seat, and most of the population shortfall is north of Orange and Putnam Counties, I see my CD as the one of the chopping block. If fact, if one could email me the spreadsheet, that would be grand.

It's on my list, but it may be a few days.

[muon2]

Muon, when you do get a state done, is it ok if I put it onto a national map?  I have started one up.

Sure, but you'll have to figure out how you want to mark multi-district areas.

[muon2]

Since there was a request for NY, I'll post that. NY is projected to lose one seat in 2020, but it's on the bubble and would be the next state to get a seat by my projection. So I'll look at NY with both 26 and 27 CDs.

Both plans use 2014 estimates projected to 2020. Whole counties are used and UCC covers are preserved. Districts are within 5% of the quota except in the NYC UCC where the tolerance is 20%.

This is the 26 CD plan and the following areas have multiple districts:

Suffolk 2 (1.94)
Queens 5 (4.92)
Brooklyn 4 (4.19)
Manhattan 6 (5.93)
Buffalo 2 (1.98)


This is the 27 CD plan and the following areas have multiple districts:

Suffolk 2 (2.02)
Queens 5 (5.11)
NYC 10 (9.98)
Buffalo 2 (2.00)


The extra CD allows for a Poughkeepsie district that doesn't have to dip into Westchester and can include Columbia county for Torie.

[muon2]

Man, that 26-district NY might be less erose than the 27-district version, but I think that despite that the 27-district map is a better one, since it does a better job of keeping the North Country and Capital Region together (the Schenectady cut-out is still unfortunate though).

Also of course it doesn't matter much to separate out groups within NYC when VRA constraints will make mincemeat of most boundaries there.  

I'm curious about NJ.  Imagine there's not a whole lot of groups you can make happen; NJ just does not tend to play well with the numbers as they currently stand.  Hopefully at least you can still nest the Philly-oriented bits in approximately three districts, like is currently possible.

The Capital Region will have a population too large for one district and too small for a district made up of 3 of the 4 counties in the UCC. A map will either chop a county or fail to pack a CD within the UCC. Since I don't have the county subdivision projections, I'm going with whole counties using 5% as the cutoff to avoid a macrochop.

Without drilling into census tract ACS data I'm not sure that satisfying the VRA will need to hop many county lines. Brooklyn can nest two BVAP-majority CDs within, and current ACS data seems to suggest that will still be true in 2020. Queens-Nassau can create another black CD, and that will be the extent of the VRA for blacks. There may only be one line crossing to make a Latino CD joining Brooklyn and Queens. Bronx will have one Latino CD within and maybe one joining with Manhattan.

The few counties in NJ does make it tough at the UCC level. It would be better if it followed NECTA and worked from cities, towns and boroughs.

[muon2]

Muon, how would you see WA's districts coming together?

I will see what I can do this weekend.

[muon2]

The key question is WA is how much population will sit east of the Cascades. With the 2014 estimate projected out to 2020 the answer is about 110K over the population needed for 2 CDs. That's the population of Chelan+Douglas, so those counties have to attach to the west. It's also the population of Chelan+Kittitas, which is more compact to move west, but it splits the Wenatchee UCC.

The Seattle UCC (King+Pierce+Snohomish) will have about five and a quarter CDs, so keeping them together in a region of 6 CDs to gives this plan. The four single CDs are with 5% of the quota and the 6 CD Seattle region is 5.945 CDs in size.



In the above plan both the pack and cover of the Seattle UCC are preserved. King county will be with 5% of 3 CDs by itself. Since it is slightly over 3 CDs it can contribute the mountain area towards Stevens Pass to an Everett-Bellingham region large enough for 2 CDs since Snohomish is too large for a single CD. Pierce is also too large for a single CD and combine with the Olympic peninsula for another 2 CD region. That results in the following plan.

[muon2]

OR is currently forecast to still have 5 CDs after 2020. That fits very well with the current projection of Multnomah county to have almost exactly 20% of the state population. A 5 CD split can preserve the UCCs (3 CDs for the Portland UCC) and keep whole counties with less than 2.5% deviation from the quota.



However, OR is on the bubble to gain a 6th CD, so it's useful to look at that as well. Multnomah will be too large for one CD and going south with the new CD requires chops to counties and UCCs. If the counties along the Columbia are combined with Clackamas and the remainder of Multnomah, it is possible to create a 6 CD plan that only splits Multnomah, chops no UCCs, and stays with 2.5% of the quota.

[muon2]

OR is currently forecast to still have 5 CDs after 2020. That fits very well with the current projection of Multnomah county to have almost exactly 20% of the state population. A 5 CD split can preserve the UCCs (3 CDs for the Portland UCC) and keep whole counties with less than 2.5% deviation from the quota.



However, OR is on the bubble to gain a 6th CD, so it's useful to look at that as well. Multnomah will be too large for one CD and going south with the new CD requires chops to counties and UCCs. If the counties along the Columbia are combined with Clackamas and the remainder of Multnomah, it is possible to create a 6 CD plan that only splits Multnomah, chops no UCCs, and stays with 2.5% of the quota.

What's the partisanship for the 3 westernmost CDs on the 5 CD map?

I only used Census estimate data which doesn't have electoral results. If someone wants to pull up the Pres '08 and '12 D and R votes by county from Atlas, it would be straightforward to get a PVI.

[muon2]

Muon and Torie, is it alright if I use some of your estimates/maps on here as a baseline for drawing maps for my timeline?

Why not? It's all for Atlas, right?

[muon2]

While I'm looking at the West Coast, here's CA. CA is assumed to gain a CD to 54 in 2020. UCCs are generally preserved, though some splits are hard to avoid without sticking together large regions. Regions are consistent within 5% of a district population, though some multi-district regions may vary by more as a whole, though not by district. In particular Fresno is under as a region and San Diego is over. The numbers show the projected population for each region in terms of the CD quota.

North (1 - lilac): 0.983.
Sacto Valley (1 - purple): 0.975
Sacramento (3 - red): 2.979
Sonoma (1 - lime): 1.048
SF Bay (10 - green): 9.953
Stockton (1 - gray): 0.999
Modesto (1 - orange): 0.970
Fresno (2 - pink): 1.918
Monterey (1 - brown): 0.983
Bakersfield (2 - yellow): 2.043
Los Angeles (20 - beige): 20.031
San Diego (11 - gold): 11.119

[muon2]

It's easy to divide ID into two CDs using whole counties. The challenge is that one cannot both preserve UCCs and maintain state highway connectivity. The first plan preserves the Boise UCC, but requires connections using local roads into and through Boise county to connect the north and east parts of the state. The second plan preserves major connections, but must split the two counties of the UCC. In both plans the projected deviation is within 1000 persons.

[muon2]

Most of the growth in NV is in Clark, and by 2020 it is projected to be almost 3 CDs by itself. Nye is sufficient to bring it within 0.5% of the quota.




The growth in UT is faster than in NV and is spread among more population centers. This division preserves UCCs and stays within 0.5% of the quota. Salt Lake county has one whole CD within making up about 2/3 of its population. The remaining 1/3 (West Valley, West Jordan, Taylorsville, Magna, Kearns, and Copperton) attach to the rest of the yellow counties.

[muon2]

CO is projected to add a CD in 2020 bringing it to 8. That can potentially create some interesting shifts. Most of the growth is around Denver and the UCC would be projected to have just over 4 CDs. Without Broomfield it would be very close to 4 CDs, with Denver having one and the suburban counties taking the other three. Growth in Boulder/Larimer at a 1.5% to 2% annual pace causes CD 2 to contract and it should be able to fit entirely on the eastern slope of the Rockies. El Paso will be large enough for a single CD, and Pueblo would shift to the eastern CD.

[muon2]

NM is easy to divide into 3 CDs, but the traditional split violates the UCC pack and cover rules. This projection keeps whole counties within 0.5% of the quota and keeps the cover of the Albuquerque UCC.



AZ is not easy. Maricopa is projected to increase it's share of the state from 60% to 62% by 2020, and the UCC that includes Pinal, will go from 65.6% to over 68%. My current projections keep AZ at 9 CDs, so the Phoenix UCC will be about 110K too large for 6 CDs that could be packed there. The extra population would go to the 2 CDs covering the Tuscon UCC providing a connection from Yuma to Pima.



When the state estimate came out last Dec I noted that the population growth had picked up considerably in the last two years, perhaps reflecting the end of the slowdown from the Great Recession. If just the 2012 through 2014 estimates are used, AZ would be expected to gain a seat in 2020 to go to 10 CDs. I used that same set of estimate data to project the AZ counties to 2020. The Phoenix UCC is about 6.8 CDs and here the UCC pack and cover is maintained with 7 CDs. The northern CD is just over 5% over quota. The 2 CDs for Tuscon are just within 5%. There is no connection from Yuma to Pima, but shifting Cochise would allow a connecting chop of Maricopa at the cost of an extra cover.

[muon2]

Does AZ law (as opposed to your rules) actually require that CD's have road connections? It still seems silly to me to chop Maricopa for the sole reason of securing a road connection from Pima to Yuma within the CD.

These are the AZ requirements (from the commission page);

You must be logged in to read this quote.

They don't have a phrase requiring convenient transportation within a district (like WA). The current AZ-3 links the Hispanic parts of Yuma without roads through the Air Force range to the Hispanic parts of Tucson and some of the SW corner of Phoenix. That seems like the kind of jump-over gerrymanders you've tended to resist. I always thought that your suggestion of not crossing CA mountains (way back when) without a significant road was a good one and it seems it should apply in AZ, too.

[muon2]

Well the problem with AZ is that there are just not good alternatives, which is an example of why I tend to favor penalizing appending counties with a link, but not prohibiting it. So in AZ's case, you either chop Maricopa or not, to get to Yuma, and it makes no difference either way when scoring the map.

There are plenty of examples of chops forced due to the combination of county lines and current populations. AZ is no different. There are many partisan gerrymanders aided by contiguity where there is no connection, but not many cases like AZ. Furthermore, with so few counties and a final 0.5% deviation from quota extra chops are virtually inevitable in AZ. An exception to the connection rule here opens the door to far more mischief than the possibility of a forced chop.

[muon2]

TX is projected to gain 3 CDs for a total of 39 in 2020. The lack of county subdivisions and the impact of the VRA make it hard to project much below the scale of counties and UCCs. So, this plan keeps UCCs whole within regions that vary no more than 5% from the quota for their number of CDs.



UCCs (color) # of CDs
El Paso/Midland/Odessa (dark blue) 2
Lubbock/San Angelo (sky blue) 1
Amarillo/Wichita Falls (purple) 1
McAllen/Corpus Christi/Brownsville/Laredo (green) 3
San Antonio/Victoria (brown) 4
Austin (gray) 3
Waco (lilac) 1
Killeen/Abilene (beige) 1
Dallas (red) 10
Sherman/Texarkana (blue) 1
Tyler/Longview (orange) 1
Houston/College Station (yellow) 10
Beaumont (slate) 1

[muon2]

LA can be divided into 6 CDs within 3% of the quota while preserving the UCCs for both pack and cover. What is doesn't do is maintain a VRA district, which some here have questioned its necessity under section 2.



It's interesting to see what these projected districts look like with 2010 data.

CD 1 BVAP 36.1%, Obama 52.9%
CD 2 BVAP 23.7%, Obama 36.0%
CD 3 BVAP 22.9%, Obama 34.1%
CD 4 BVAP 31.8%, Obama 38.8%
CD 5 BVAP 33.0%, Obama 38.1%
CD 6 BVAP 32.5%, Obama 41.8%

The black population is so well distributed among the centers that make up the separate UCCs that none of the CDs breaks 40% BVAP but none are less than 20% BVAP. The state election Dem number in CD 1 on DRA is 57.4%, so it looks like a white Dem could win it, but a black Dem would be a toss up there due to racial bloc voting.

[muon2]

The Mid-South states of the Mississippi valley are generally straight forward to project and aren't expected to change in CD count.

AR stays at 4 CDs and with its many counties can be divided several ways. This plan starts with the current map, adjusts to whole counties and whole UCCs, then shifts few counties to bring the CDs to within 0.5% of the quota.



KY stays at 6 CDs. The Louisville UCC needs a cover of 2 which is done with the Covington CD. With whole counties the plan is within 5% of the quota. Shifting 32K out of Jefferson county puts all CDs within 0.5% of the quota.



TN stays at 9 CDs and Shelby will be a 200K too large for a CD, but the remainder can still easily be BVAP majority. The Nashville UCC will be just over 2 CDs, and this plan preserves the UCC cover, but not the pack which would require at least two chops. All other UCCs are preserved within 1 CD. All Cds are within 0.5% of teh quota except for Nasville which is 1.1% over quota.

[muon2]

MS will still have 4 CDs in 2020 and I expect will still have a mandate for a BVAP majority CD. There are many whole county combinations that would suffice, but it is useful to see both the UCCs and MCCs (minority county clusters). Minority county clusters are assembled from contiguous counties that have more than 40% BVAP (2010 data) and are used to preserve communities of interest based on a minority population much like UCCs preserve urbanized area communities of interest.

In the map below the three pink circles are UCCs, all less than one CD in size. The green counties are those that meet the criteria for MCCs, and there are three that contain more than one county.



Projecting to 2020 it looks likely that a BVAP majority CD cannot be constructed with the whole Jackson UCC, unlike in 2010 where it could just barely. The Jackson UCC overlaps the Mississippi river MCC and one of those would have to be chopped anyway, so the Jackson UCC is chopped to preserve the VRA CD with the whole MCC. Elsewhere all other UCCs and MCCs are maintained and CDs are within 0.5% of the projected quota.

[muon2]

AL is a little trickier. We had a lot of debate about how one might draw rational CDs in 2010, and this map shows the three multi-county UCCs and the large MCC. Because the MCC stretches all the way across the state it has to be chopped. The overlap with the Montgomery UCC pulls the BVAP% of any district containing both way down.



If I assume 7 CDs in 2020 based on my current projection, then a simple plan can be put together that keeps whole counties, cover rules for UCCs and all CDs within 2% of the quota. There is only a minimal chop of the MCC as needed to connect the SE CD. The BVAP of the remaining MCC CD is just over 40%.



To get an CD with a BVAP over 50% without a lot of chops, then the CD links Birmingham city to the MCC through Tuscaloosa. The UCC cover rules are all still in place and there is still only one chop of the MCC as well as a 2% tolerance for the quota.



In my 2020 apportionment projection AL-7 was 435th seat, and alternate models have it losing a seat. If a plan is needed with 56 CDs including a VRA cmpiant on, it is probably impossible to avoid linking to Birmingham. As above, the tolerance in 2% and the UCC covers are preserved.

[muon2]

Two years ago in the thread fleshing out the definition of UCCs we spent a lot of time looking at AL. We generally agreed that the Black Belt was a CoI that should not be unduly chopped, yet we also wanted to see what areas might be regarded as contiguous and reveal any gaps. The notion that one should not be forced to bridge across Tuscaloosa if other alternatives were available was important to you. You agreed the 40% BVAP threshold was a good one to distinguish which counties should be clustered. That gave rise to the AL MCC. For these rural counties and directly adjacent urban counties like Montgomery I don't think the packing problem would come into play. When I did MS the other day I replicated the process used two years ago.

In the first map I only claimed to preserve the UCC cover, not the pack. Since the other districts only hewed to a 2% variation and I could keep whole counties I did so for Birmingham as well, since a UCC pack violation counts the same as a county chop and I don't have reasonable county subdivisions for Jefferson or Shelby.

I would also note that in the first map I could chop the Montgomery UCC and add Tuscaloosa to up the BVAP%. With the declining rural population it won't be as effective in 2020 as it was in 2010 to make a high-40's BVAP CD with whole counties. So for purposes of illustration I left it as a low 40's CD keeping the UCC intact.

[muon2]

The UCC rules exist to identify a community of interest that spans counties and is large enough to influence elections. By invoking UCCs one can get plans that avoid the dilution of that CoI's power, and for large UCCs avoid giving them excessive power. I contend that a minority of sufficient concentration spread over a number of counties represents an equivalent CoI, and a case should be made that redistricting should avoid dilution of their voting power as well. Personally I think using an equivalent approach helps make the case under the VRA that race is not a predominant factor, but one of a number of equivalent factors including county chops, UCCs and erosity. Of course one must avoid packing a minority under the VRA and that in my mind is a separate determination viewed when looking at the totality of a plan.

In the case of a 2020 CD plan in AL the issue will not be one of packing. If the current polarized voting persists I think that a neutrally-drawn CD (ie with the kind of rules we use) can't be drawn in a way that it is likely to elect the candidate of choice of the black minority. The CDs will just be too large, especially if the number drops to 6. With 6 CDs it isn't hard to draw a whole county plan where all 6 CDs vote Pub, and what if the AL legislature does that?. This may well test your assertion that the VRA doesn't require the bridge to Birmingham.