Redistricting Washington with ten districts

[muon2]

Excellent maps. I guess coming from a Washingtonian perspective it is easier to understand a lot of the districting. Muon did a very good job, but the districts were very objectively created, and districts need to be someone subjectively made if they are to work.

That's an interesting thought. Many who support a truly independent redistricting procedure want to avoid subjective measures. The example that is frequently used is the Iowa process. There an independent agency runs a computer program based on clearly determined data to generate the maps. Without quite as much detail as used in IA, I've tried to emulate the basic features of their method.

[muon2]

The following is related to the discussion between Muon2 and myself in the Apportionment thread with regard to counties being linked or not, even if they share a border.

A single urban center was indicated for each county.  In counties which shared an urban center, the counties were treated as being merged: King-Pierce (Seattle); Chelan-Douglas (Wenatchee); and Benton-Franklin (Tri-Cities (Kennewick-Richland UA)).  In a few cases, the marker for the urban center was shifted.  The marker for Seattle is closer to Auburn to account for the spanning of the King-Pierce border; the marker for Adams County is nearer the center of the county, rather than in Othello in the panhandle in part to reflect the location of I-90.

A grid of lines indicating adjacent counties was then added.  This generally forms a triangular mesh.  A few extreme near-corners have been omitted (Douglas-Okanogan; Columbia-Franklin). 

It was then determined for each county-county link whether there was a reasonably direct surface transportation route between the urban centers for the counties, either by highway or ferry.  The route need not be limited to the two counties.  For example the route between Grant (Moses Lake) and Benton-Franklin (Tri-Cities) passes through the Adams panhandle.  On the other hand,  when the direct route is essentially formed by the direct routes to other counties, it was rejected.  For example the direct route from Grant (Moses Lake) to Yakima (Yakima) goes through Kittitas (Ellensburg).

County-County links that were rejected are marked with green lines.  Those that were accepted are shown as heavy red lines.  At this point, there could be a period of local consultation to consider changing the status of links.

The final map is a simplified graph, indicating county-county boundaries that may be crossed in drawing CDs.  The absence of a link between counties does not mean that they could not be included within the same CD (eg Whitman and Franklin) but simply that other counties must also be included.

This is an interesting idea, but I need to understand some of your non-connections.

Some land questions:

Why doesn't Skagit connect to Okanogan by way of hwy 20, or Snohomish to Chelan by way of US 2?

Why doesn't Okanagan connect to Grant on hwy 155 over the Grand Coulee Dam?

Some sea questions:

Doesn't the ferry from Edmonds to Kingston link Snohomish to Kitsap?

San Juan to Island uses the same ferry as San Juan to Skagit. Isn't that the direct route to Skagit and therefore excluded?

[muon2]

Haha, wow.

Kudos on creating a district that includes both Olympia and part of the Grand Coulee Dam.

Would this put Reichert in the same district as Inslee, though?

I think the that the Douglas-Grant border is actually west of the Grand Coulee.  80% of the population of Douglas County is in the Wenatchee area.  The other two CCDs in the county are Bridgeport and Waterville.  Waterville, the county seat is toward the western edge of the county, and Bridgeport is towards the western edge of the northern border.  Even in Grant County, there is less than 10% of the population in the northern tip.

I think that my map is quite responsible approach to meeting concerns of equal population, avoiding unnecessary county splitting, and providing reasonable district connectivity.

Current representatives or political results are illegitimate goals of redistricting, and would be outlawed by any sort of responsible redistricting legislation passed by Congress.

It is difficult to determine who lives in what district, it appears that many representatives have relocated to become more electable.  Inslee for example, relocated from Selah after being defeated.  Baird was associated with Pacific Lutheran in Tacoma before election, but now lives in Vancouver.  Smith was a Seattle prosecutor but now lives in Tacoma suburbs.

Anyhow, Inslee and Dick both live in Kitsap County.  Inslee would lose the portion of his district to the north of Seattle.  Dicks would lose the area to the west.  Smith represents a large chunk of southern King County which would be placed with Kitsap County, and Reichart may live in that area.

So Baird could take the Tacoma seat.  Inslee could still run in the Kitsap-King South district, though most of the voters would be new.  Since Dicks is from Bremerton, he could run as well.  Larsen could represent Snohomish.  This would open up the northwest seat, so Inslee could jump there.  Reichart represents more of the area to the east of Lake Washington, so even if he lives in Kent, he could run for many of the same voters.  This makes the Olympia-Rainier-Cascasdes seat the open seat.

On your map what deviation do you permit? From my projections it appears that you allow 1% from the ideal, whereas I've held to 0.5%.

[muon2]

This is an interesting idea, but I need to understand some of your non-connections.

He has too many connections, not too few. The two more northerly transcascadian links need to go, and I think the Jefferson-Island link needs to go as well.

The basic number of links is set by which counties are contiguous whether by land or by sea. The question then becomes which connections should be discarded. What I like is the idea for an orderly pruning of links based on some definable rules. In the end, it may be that links are removed based on a local determination, but that can't be generally defined.

[muon2]

To follow up, I can compare maps that permit or reject a northern transcascadian connection. Both plans impose a 0.5% deviation from the ideal. The map on the left restricted the connection to a path along the Columbia and requires a split of either Benton (my choice) or Yakima. The map on the right connects Whatcom to Okanogan and points east. It only requires splits of the three counties that must split: Snohomish, King and Pierce.

[muon2]

The second is the nice, but I take issue with the Lewis-Skamania link in the first.

But would mappers in WA ever consider the northern link across the Cascades?

[muon2]

The second is the nice, but I take issue with the Lewis-Skamania link in the first.

But would mappers in WA ever consider the northern link across the Cascades?

I honestly don't know.  I was incorrect about the Skamania link, though; I guess Wind River Road goes from Cougar to Carson.  It's still probably not realistic, though.  Northern Skamania County is virtually unpopulated, and most of it is federally protected forestland.  There is a town (Stabler) up there, if I recall correctly (Google Maps won't load) but it's tiny.

I'm not sure that the idea of a major trans-Cascadian link has ever been debated, but I get the impression a southern link would be heavily favored.

I took that second map and ran the 2004 presidential vote for the districts. I used the two party vote, and strong is over 55%, lean from 52-55%, and competitive from 50-52%.

CD 1 (Everett) Lean D 54%
CD 2 (Bellingham) Competitive R 51%
CD 3 (Vancouver) Lean R 54%
CD 4 (Yakima) Strong R 62%
CD 5 (Spokane) Strong R 59%
CD 6 (Olympia) Lean D 54%
CD 7 (Seattle) Strong D 80%
CD 8 (Bellevue) Competitive D 51%
CD 9 (Kent) Lean D 53%
CD 10 (Tacoma) Strong D

[muon2]

Kudos on creating a district that includes both Olympia and part of the Grand Coulee Dam.

Here is an alternate version.




It places Tacoma along with Lakewood, Fort Lewis, McChord AFB, and the Olympic Peninsula part of Pierce County in one district; and most of the Tacoma suburbs in the district crossing the Cascades.

For Tacoma-Olympia: Thurston 37%; Tacoma 33%; Pierce (non-Tacoma) 30%.

For Cascades-Mt Rainier: Pierce 58%, King 14%, Snohomish 6%, Transcascade 22%.

Positives: Eliminates Wenatchee-Olympia span.

Negative: Almost 50-50 split of Pierce County, making it the dominant county of two CDs, even though it should only have 1.2 CDs.  Aggressive split with the boundary cutting right along the eastern Tacoma city limits (In Seattle this is necessary because of the size of the city, which is about 90% of a CD.  Pierce County would be dominant, even though original effort was made to avoid placing the eastern area in a district that was 3/4 in King County.

Here's a version that follows your connectivity, maintains the 0.5% deviation, only splits the three big counties, and generally addresses the negatives.



Tacoma is now in a Pierce-only district so only the remainder of Pierce only makes up 20% of the Olympia district. The Everett district is 63% Snohomish and 37% Kitsap. The remainder of Snohomish is only 37% of the Bellingham district. So, King dominates 3 districts, and Pierce and Snohomish each dominates one district.

[muon2]

Here are some more details on my map.

CD 1 (Everett) 672.3 K. Includes Snohomish with roughly all the area on the SW side of the Snoholmie River (426.4 K) and Kitsap (245.9).

CD 2 (Bellingham) 672.7 K. Includes Whatcom (198.4 K), Skagit (124.1 K), San Juan (16.1 K), Island (88.1 K) and Snohomish north of the Snoholmie River and Skykomish valley (246.0 K).

CD 3 (Vancouver) 673.4 K. Includes Clark (459.8 K), Cowlitz (104.3 K), Klickitat (21.1 K), Lewis (76.7 K), and Skamania (11.5 K).

CD 4 (Yakima) 673.3 K.  Includes Yakima (239.6 K), Kittitas (39.7 K), Benton (170.6 K), Franklin (79.7 K), Walla Walla (59.3 K), Columbia (4.1 K), Garfield (2.1 K), Asotin (21.7 K), Adams (17.2 K), and Whitman (39.3 K).

CD 5 (Spokane) 669.2 K. Includes Spokane (464.9 K), Lincoln (10.5 K), Grant (87.8 K), Pend Oreille (13.7 K), Stevens (44.3 K), Ferry (7.7 K), and Okanogan (40.3 K).

CD 6 (Olympia) 672.4 K. Includes Thurston (252.8 K), Clallam (74.4 K), Jefferson (31.2 K), Mason (60.3 K), Grays Harbor (74.4 K), Pacific (22.2 K), Wahkiakum (4.1 K), Pierce from Ft Lewis to south of Puyallup to east of Sumner (138 K), and King along the White River from Enumclaw east (15 K).

CD 7 (Seattle) 672.8 K. Includes King west of Lake Washington with Seattle (594.2 K), Shoreline (51.7 K), and the Lake Forest Park area (26.9 K).

CD 8 (Bellevue) 672.6 K. Includes King east of Lake Washington, Renton, and Kent (522 K), Snohomish from Monroe up the Skyhomish valley (39 K), Chelan (73.8 K), and Douglas (37.8 K).

CD 9 (Kent) 673.0 K. Includes King south of Seattle and from Renton and Auburn to Vashon Island.

CD 10 (Tacoma) 671.5 K. Includes Pierce from the Lower Peninsula east through Puyallup and Sumner.

[muon2]

Here are some more details on my map.

How did you make your population projections?

I used the 2006 Census county estimates:

  proj2010 = base2000 * (est2006 / base2000)**(10./6.25)

Where

  base2000 is the April 1, 2000 estimates base, same as census expect where corrected;
  est2006 is the July 1, 2006 estimate issued in 2007; and
  proj2010 is the April 1, 2010 projection.

My county projections seem to be slightly higher than yours, though they must be generally proportional given that we are coming up the same relative deviation for all districts.

I use the same formula. Here's what I get for an example in Benton County, WA.

Census Base (C): 142,478
2006 Estimate (E): 159,463
estimate period (n): 6.25
total period (N): 10.0

Projection (P = C*[E/C]^[N/n] ): 170,611

Do you get a different result?

[muon2]

I don't think East King and SE Snohomish counties should be thrown together with chunks of Eastern Washington like that... It doesn't really make sense. It doesn't even go along I-90. And even then, with "connections" so important, it is worth mentioning that many of those mountain highways are closed during the winter. Even I-90 is closed right now. And even then, except for some ski areas, there's almost nothing between North Bend and Cle Elum/Roslyn...

Skamania-Klickitat is the only appropriate East-West crossing, IMO.

Eastern Washington looks pretty messy in muon's latest map...

I think it's useful to see some of the difficulties WA will face making a map with 10 CDs. The key is that with 10 CDs about 135 K population from east of the Cascades will have to join with one or more western districts. A 9 CD split (which is still my projected number) works almost perfectly to separate the state along the Cascades.

With 10 CDs the natural tendency is to maintain a connection along the Columbia. To get 135K in the south, one option is to split the Kennewick/Richland/Pasco area. That's the option I used in my first map. Otherwise, to get 135 K requires about 114 K from Yakima County, and that probably would require splitting the city of Yakima itself.


A second attempt uses the northern edge to link a district. There are a number of variations, but invariably the district will link Bellingham to either Coulee Dam, Wenatchee or both. As pointed out by jimrtex, it's not necessarily a great connection through the Cascades national park.


The third way goes across the central Cascades. This can use either Stevens Pass to Wenatchee or Snoqualmie Pass to Ellensburg. Either of these will link Wenatchee or Yakima to suburban Seattle. My version below includes a mix of population from the east in both the south and central corridors. It is possible to make it central only as well by extending the Yakima district well to the west.

[muon2]

The third way goes across the central Cascades. This can use either Stevens Pass to Wenatchee or Snoqualmie Pass to Ellensburg. Either of these will link Wenatchee or Yakima to suburban Seattle. My version below includes a mix of population from the east in both the south and central corridors. It is possible to make it central only as well by extending the Yakima district well to the west.

Arguably, Skamania and Klickitat are best placed in a western district due to their links along the Columbia to Vancouver.

Besides US 2 over Stevens Pass Stevens Pass web cam there is the main northern US transcontinental railroad route.  Chelan County was set off from Kittitas County in the 1899 because it was so difficult to travel to the county seat in Ellensburg in winter.  Travellers had to travel via Seattle or Spokane by railroad.City of Wenatchee - Great Northern Railroad

I don't think there would be any toleration of any east-west connections except for the Columbia route in the south, the passes are just too vulnerable to the weather and having an entire mountain range make a 50 mile barrier between two sides of a congressional district would just be too much.

The question then is what to attach to Vancouver. Would there be more tolerance to split the city of Yakima, or to split the tri-city region with Kennewick separated from the cities across the rivers?

BTW, there are similar difficulties for a 6-CD Oregon. The east side is too big for a single CD, and part has to go to the west side. From the way projection look, either the Medford area gets split, or a Portland-area Clackamas district extends east past The Dalles.