US House Redistricting: North Carolina

[jimrtex]

"The judges will see the maps for what they are, and what they are is an attempt to disenfranchise African Americans by segregating them and diminishing their voting rights and the influence of women in North Carolina," said Rep. Deborah Ross, D-Wake, as the House debated proposed House districts

lol? how did they do that?

In New Jersey it was claimed that districts represented by women were targeted for deletion. 

[jimrtex]

A court will let the legislature fix this technical error. You can bank on it.

Oh, I do suppose so. It's just, you know, funny. I mean, wtf do they have preclearance for?

Apparently, Texas.


Joe Hackney is making the hail mary argument.

Holder has really pissed me off by punting on NC, SC and LA.

We better end up with a court-drawn Texas, or something, for the sheer lack of effort tthe DOJ has invested in the other states.

Holder should have pushed for the creation of another black majority district in AL, SC, and LA.  It could easily be done.  

No! Quite the contrary. The 1st is good enough. The Republicans already tried to pack as many blacks into the 12th as they could without violating the Shaw ruling.

No, what the DOJ should be doing is requiring the creation of as many black majority districts as possible in these VRA states.  By having a bunch of 50%-55% black districts, it makes it much more difficult for Republicans to vote-sink.  

NC doesn't have statewide VRA coverage under Section V like LA, AL or SC.

For instance, making Mel Watt's district greater than 50% black VAP would be illegal.

It is covered by Section 2.

[jimrtex]

Oh look, isn't this nice. Another lawsuit challenging CDs 1 and 12.

The mapmakers relied on section 5 for their districts. Section 5 was overturned as far as applicability to NC after the maps were drawn and survived a challenge. Is the map now subject to a section 2 challenge having tried to meet the law as it existed at the time?

"The length of the district’s perimeter, according to the lawsuit, is 1,319 miles – 'almost precisely the distance from Chapel Hill to Austin, Texas.'"

I read this part as distance from Chappell Hill to Austin, Texas, and thought it can't be more than 100 miles (99.3 miles actually).  1319/100 = 4.19 pi, which is fairly close to 4 pi, the ratio of the circumference of circle squared to its area of a circle.   So how can they call this district non-compact?

[jimrtex]

Where did this 40% in an urbanized area county concept come from again, and how does one define "urbanized area?"

Since the late 19th century, the census bureau has distinguished between the urban and rural population.  The urban population lived in cities and towns of over 2500 persons.  The rural population included everyone else (farms, villages, and small towns).   I'm using cities, towns, and villages informally to indicate the size of the population concentration.  The census bureau used officially defined (ie incorporated) places.

Beginning in 1950, the census bureau began defining Urbanized Areas, which included cities of over 50,000  and adjoining towns and unincorporated areas.  This recognized suburban development that was not always formally organized into towns, but was clearly "urban" as opposed to "rural."

The census bureau also defined Metropolitan Statistical Areas which were groups of counties that either contained Urbanized Areas or were linked to them economically, generally by commuting, but in the earliest versions by such things as number of phone calls.

Over time, the census bureau refined their definition of Urbanized Areas to permit them to be based on groups of cities with a population greater than 50,000, as well as Census Designated Places, which have the density and appearance of cities, but without any formal legal status.

1990 was the first census where the entire country was delineated into census blocks, and in 2000 changes how the urban population was defined.  Urban Areas (proper noun) are areas of continuous higher density population, based on blocks without regard to cities or other legal entities that have a population of greater than 2500.   Thus areas adjacent to small towns may now be urban, while undeveloped land inside city boundaries may now be rural.  So the new definitions continued the century-old classification of urban and rural, but based it on actual development patterns.

Urban Areas with a population greater than 50,000 are classified as Urbanized Areas, to maintain continuity with previous definitions, while those with less than 50,000 are classified as Urban Clusters.

The new definitions recognized that higher density development can occur along highways, and permits uses of "jumps" and "hops" to connect urban territory.  But this also results in huge agglomerations of urban territory - such as from Springfield, MA, through Hartford and New Haven, CT, through New York City, Trenton, NJ, Philadelphia, PA to Wilmington, DE.

So the census bureau decided that Urbanized Areas would be divided based on 1990 Metropolitan Statistical Areas.  And since Metropolitan Areas are based on counties, the division was done at or near county lines.  No such protection is provided for Urban Clusters, so Urbanized Areas may eventually absorb nearby Urban Clusters.

For 2010, the 2000 Urbanized Areas were preserved, so where they abut, a boundary is devised.

The 2000 census also extended the concept of metropolitan areas to less populous areas.  The areas are called Core-Based Statistical Areas (CBSA).   CBSA are based on a counties associated with a core Urban Area either by containing the Urban Area, or having significant commuting ties to the counties.   If the core Urban Area is an Urbanized Area (ie more than 50,000 persons) then the CBSA is designated a Metropolitan Statistical Area.  If the core Urban Area is an Urban Cluster with between 10,000 and 50,000 persons it is designated a Micropolitan Statistical Area.

Even though Metropolitian Statistical Areas are based on Urbanized Areas, there is not a one-to-one relationship between Urbanized Areas and Metropolitan Statistical Areas.  A Metropolitan Statistical Area may include all or parts of multiple Urbanized Areas.  Some counties in a Metropolitan Statistical Area may be quite rural.  Part-time farmers may be willing to commute 50 miles, while a small town will provide some jobs that don't need commuting.  The rural area may thus have a greater percentage of its population commuting.

In defining Urban County Clusters, we start with Metropolitan Statistical Areas which are defined on both a structural basis - they contain Urbanized Areas and an economic basis - commuting links.  We want to isolate the urbanized core.

If a majority (50%) of the population of an county lives in urbanized areas, it clearly is part of that core.  Around large cities, close to 100% of the population lives in urbanized areas.  But in smaller cities it may be somewhat lower.  There will be a fringe that economically depends on the city, but is physically separate, perhaps in urban clusters that don't quite connect, in farm houses, even if they aren't farming, in small clusters or homes, or perhaps large acreage lots.

Counties with no population (0%) in urbanized areas don't qualify, since they are purely linked by commuting rather than settlement.

Urbanized Areas may often just have a small finger into a county, and don't really characterize the population settlement of the county.  But if 25,000 persons live in an Urbanized Area, that is half the population needed to be classified as an Urbanized Area as opposed to an Urban Cluster.

If the county is fairly rural, a very small population in an Urbanized Area can represent a noticeable share of the population.  The 12K persons in the Winston-Salem Urbanized Area that live in Stokes County represent 24% of the population.  So there needs to be a somewhat high threshold chosen to exclude relatively small incursions.  40% is a reasonable value..  Though a majority of the population doesn't live in Urbanized Areas, a dominate share does, and there are likely proximate areas that just miss being included in Urbanized Area.

This map illustrates the concepts (right click to view larger image).



Urban County Clusters are shown in the light tints, including three that extend into neighboring states (Charlotte; Myrtle Beach, SC; and Virginia Beach-Norfolk, VA).  Urbanized Areas within each are shown in a darker tone.  There is one Urbanized Area for each UCC, except that Charlotte UCC includes Charlotte UA, Concord UA, Gastonia UA, and Rock Hill, SC UA; and Greensboro-High Point UCC includes Greensboro UA and High Point UA.

The bright green areas (in North Carolina) are Urban Clusters (Urban Area with less than 50,000 population).  I didn't attempt to distinguish Urbanized Areas and Urbanized Clusters in other states, except in the Charlotte, Myrtle Beach, and Virginia Beach UCC.

Counties that are part of a Metropolitan Statistical Area, but not the corresponding UCC are shown in white.  The number is the population in thousands within Urbanized Areas, and the percentage of the county's population in Urbanized Areas.  The color of the numbers indicate the Metropolitan Statistical Area for the county.

[jimrtex]

Where did this 40% in an urbanized area county concept come from again, and how does one define "urbanized area?"

In the course of a couple of threads in July and August there was an emerging consensus that chopping a metro area should count as much as a chop of a county. There were a variety of Census Bureau definitions of metro areas and after looking at cases in a few states there was convergence on the concept of urban county clusters which are formed from metropolitan statistical ares. jimrtex has outlined the definition and I have now stickied the thread that he created showing all the qualifying UCCs in the US.

What I don't get specifically, is how you determine that an individual county is 40% or more "urbanized."  I don't think the census bureau chops counties that way, does it?

This can be loaded directly into a spreadsheet.

http://www2.census.gov/geo/ua/ua_county_rel_10.txt

http://www.census.gov/geo/maps-data/data/pdfs/rel/explanation_ua_county_rel_10.pdf

[jimrtex]

Where did this 40% in an urbanized area county concept come from again, and how does one define "urbanized area?"

In the course of a couple of threads in July and August there was an emerging consensus that chopping a metro area should count as much as a chop of a county. There were a variety of Census Bureau definitions of metro areas and after looking at cases in a few states there was convergence on the concept of urban county clusters which are formed from metropolitan statistical ares. jimrtex has outlined the definition and I have now stickied the thread that he created showing all the qualifying UCCs in the US.

What I don't get specifically, is how you determine that an individual county is 40% or more "urbanized."  I don't think the census bureau chops counties that way, does it?

The Census Bureau calculates for each county the fraction of population that is in urbanized areas, urban clusters (ie small cities) and in rural areas. That data was used to separate the mostly commuting counties from the more significant contributors to the urban area. The file is at http://www2.census.gov/geo/ua/PctUrbanRural_County.xls. The data from that file is the basis of the maps jimrtex put together here.

I used this file, which is the relationship file between urban areas and counties, so for each county has the individual urban areas (plus rural remnant) for each county.

http://www2.census.gov/geo/ua/ua_county_rel_10.txt

[jimrtex]

Where did this 40% in an urbanized area county concept come from again, and how does one define "urbanized area?"

In the course of a couple of threads in July and August there was an emerging consensus that chopping a metro area should count as much as a chop of a county. There were a variety of Census Bureau definitions of metro areas and after looking at cases in a few states there was convergence on the concept of urban county clusters which are formed from metropolitan statistical ares. jimrtex has outlined the definition and I have now stickied the thread that he created showing all the qualifying UCCs in the US.

What I don't get specifically, is how you determine that an individual county is 40% or more "urbanized."  I don't think the census bureau chops counties that way, does it?

The Census Bureau calculates for each county the fraction of population that is in urbanized areas, urban clusters (ie small cities) and in rural areas. That data was used to separate the mostly commuting counties from the more significant contributors to the urban area. The file is at http://www2.census.gov/geo/ua/PctUrbanRural_County.xls. The data from that file is the basis of the maps jimrtex put together here.

I used this file, which is the relationship file between urban areas and counties, so for each county has the individual urban areas (plus rural remnant) for each county.

http://www2.census.gov/geo/ua/ua_county_rel_10.txt

Are there any places where the files would produce different results?

Under the final definition - probably not.  But I wouldn't have been able to understand the information or develop the definition without the relationship file.

[jimrtex]

This attempt involves the urban county cluster model. Urban clusters are made from counties in an MSA that have over 40% population in an urbanized area (or have 25K urbanized population). Minority clusters are contiguous counties that are over 40% BVAP. The map shows the UCCs in pink and the MCCs in green. The number in the circle is the minimum number of CDs it takes to cover the UCC.



For this plan UCCs and MCCs are each covered with the fewest number of CDs. Only Mecklenburg and Wake are chopped, and 4 microchops are used to keep all CDs within 0.5% of the population quota. There is no forced linking of the urban minority populations in Raleigh or Durham with CD 1, so it is left with only 40.3% BVAP. However, CDs 1, 3, 4 and 13 could be rearranged to provide a 50% BVAP CD without changing the rest of the map.

If I force a VRA CD unto the plan above, I get the following plan. CD 1 is at 50.3% BVAP. The other eastern CDs are adjusted to keep chops and erosity down.

Whole county-districts are intrinsically compact.

There is explanation for the tentacles into Raleigh and Durham other than race.  The second plan violates Equal Protection.

[jimrtex]

BTW Muon, your map of Charlotte needs to be fixed asap. Gaston county should not go all the way into the city.

The Gaston County line extends easterly from the South Carolina line for about 2 miles before turning to the north.  The area around the southeastern corner of Gaston County, on both sides of the county line is mostly outside the Charlotte and Gastonia urbanized areas.  The two urbanized areas abut further north.

[jimrtex]

The 1st district does not have to be drawn into Durham/Raleigh to be black plurality.  And it should not be.

That requires an extensive number of erose peninsulas into the center of every small city in eastern NC. Yuck.

Good to see that you are on erosity patrol these days, Mike.

And what are your thoughts on the VRA here? Is one mandated?

It fails the Gingles test, which may not even be applicable.

[jimrtex]

Charlotte:

Much better around Charlotte (though you sort of rely on a bridge since the border between Iredell and Lincoln is a reservoir with no crossing). Now if you can fix the crazy shape of CD-3 you'll be even better.

(Most of) the border between Iredell and Lincoln is within the Charlotte Urbanized Area.  The urbanized area cross below the dam and then extends northward along the western shore of Lake Norman.  There are some rules for enclosing indentions which bring that part of Lake Norman into the urbanized area.

In addition, the bridge on NC 150 is only about 4 miles north of the Catawba-Lincoln line with NC 150 going southwestward directly toward Lincoln.   That crossing and the crossing below the dam provide adequate connectivity.

And I'd view that map as placing the northern and western counties of the UCC (Iredell and Gaston) with a significant part of Mecklenburg, and then adding Lincoln.   The Mecklenburg portion of the district must be close to 300,000.  At around 30% of the county and 40% of the district, I don't think it can be considered a bridge at all.

[jimrtex]

I think that county chops should still penalized with large MSAs, although perhaps less so- this Mecklenburg example is one of many clear issues with that practice. In a way, in fact, it kind of does open the door to gerrymandering like-so:


which is obviously sub-optimal.

I would control gerrymandering within UCC by not permitting double-spanning, where two or more districts contain parts of pairs go counties.  That is, the district boundaries should be generally parallel to county boundaries, rather than perpendicular.  In addition, there should be no more counties split than is the minimum.

So with your Hennepin-Anoka-Ramsey district, you could not have another district connect Hennepin and Anoka, nor Anoka and Ramsey, nor Ramsey and Hennepin.

[jimrtex]

Charlotte:

Much better around Charlotte (though you sort of rely on a bridge since the border between Iredell and Lincoln is a reservoir with no crossing). Now if you can fix the crazy shape of CD-3 you'll be even better.

(Most of) the border between Iredell and Lincoln is within the Charlotte Urbanized Area.  The urbanized area cross below the dam and then extends northward along the western shore of Lake Norman.  There are some rules for enclosing indentions which bring that part of Lake Norman into the urbanized area.

In addition, the bridge on NC 150 is only about 4 miles north of the Catawba-Lincoln line with NC 150 going southwestward directly toward Lincoln.   That crossing and the crossing below the dam provide adequate connectivity.

And I'd view that map as placing the northern and western counties of the UCC (Iredell and Gaston) with a significant part of Mecklenburg, and then adding Lincoln.   The Mecklenburg portion of the district must be close to 300,000.  At around 30% of the county and 40% of the district, I don't think it can be considered a bridge at all.

That's why I said "sort of".

In the strictest sense a bridge refers to a county fragment use to link two discontiguous whole counties that cannot otherwise be linked using whole counties in the district. While Iredell and Lincoln are contiguous in a geographic sense, their border is short and entirely in Lake Norman. I don't need to rehash our debates about using roads outside of the counties in question to determine connectivity; we just won't agree.

Within the UCC, Gaston is only connected to Mecklenburg, and Union is only slightly connected to Cabarrus.   While Iredell is connected to Rowan and Cabarrus, the functional connection is with Mecklenburg.

Functionally the urbanized area is like a starfish with western (Gastonia), northern, northeastern (Concord), southeastern, southern (Rock Hill, SC) arms.  To avoid using Mecklenburg as a "bridge", it would have to be split in 3 parts.

Taking a large chunk of the body and connecting to two arms is not really a bridge.  Mecklenburg must be split, and one district is wholly within the county, and the remnant is self contained and of substantial population, larger than either Iredell or Gaston.

I would make a distrinction between county fragments necessary for gross population equality (ie within multi-district regions), and those small chops that might be necessary to placate the SCOTUS (interregional).  Of course those bridges would not be possible under a system that first defines the regions, because the chop is necessary to make the region contiguous.

Here is the statement of our disagreement with regard to county contiguity.

We agree that regions and districts should be comprised of contiguous whole counties or parts of counties.  We agree that superficial contiguity should not be permitted.

Point contiguity is not permitted.  Contiguity across large bodies of waters is not permitted, unless there is a bridge or ferry that traverses the boundary.  For example Richmond and Queens, nor Nassau and Westchester in New York are not considered to be contiguous.  Also where there are substantial barriers to travel, such as the northern Cascades.

Where we disagree:

I would restrict near point contiguity, and I have provided a workable formula.  I would permit direct travel between counties, so long as it is not substantially through the main portion of oher counties.  Since districts are comprised of counties, contiguity is most significant, road connectivity is secondary.  I do not regard rivers, or reservoirs to be barriers, particularly if there is a nearby way to cross them.  If simplified boundary length is used for measuring compactness, regions that connect through narrow isthmuses will be heavily penalized. 

You would require true road without any incidental travel outside the two counties.

[jimrtex]

I think that the split of Wake is a chop, and after all our discussion on chops in other states, there was no good rationale to create a benefit for a single county CD. So, like the Lumbee issue, that would be a local preference if the map is suitably fair to make it to the final set. As it is, you are creating a lot of erosity in CD 2 to have your preference in CD 13. OTOH there is a benefit to avoiding extra chops to a UCC. That puts a premium of keeping Durham and Orange in the same CD, with at most a split for a VRA district.

So, do you think CD1 could go into Durham city as long as the rest of the county was with Orange?

I like the 45% BVAP threshold, but to get it, I have to go into either Durham or Raleigh.

I agree that's what is needed, and then the question is how little erosity and how few chops are needed to meet the threshold. However, once you have to send a prong into either of those cities the question again arises as to whether the minority area is sufficiently compact as to require section 2 protection. This was the great debate we had over AL maps and whether the prong into Birmingham from the Black Belt was compact enough to trigger the VRA.

If the trigger isn't met I can go back to a map like this with 40.3% BVAP in CD 1 with D+2 that keeps the Minority County Cluster intact (counties over 40% BVAP).

Your CD 1 was:

Senate 2008: Hagan 59.0% Dole 39.0%
Senate 2010: Marshall 52.3% Burr 46.2%
Governor 2008: Perdue 62.3% McCrory 36.0%
Governor 2012: Dalton 54.3% McCrory 44.3%
President 2008: Obama 55.6% McCain 43.8%
President 2012: Obama 55.7% Romney 43,5%