Redistricting with 2020 Population Estimates (and 2016/2018 Political Data)
       |           

Welcome, Guest. Please login or register.
Did you miss your activation email?
March 29, 2024, 06:52:06 AM
News: Election Simulator 2.0 Released. Senate/Gubernatorial maps, proportional electoral votes, and more - Read more

  Talk Elections
  General Politics
  Political Geography & Demographics (Moderators: muon2, 15 Down, 35 To Go)
  Redistricting with 2020 Population Estimates (and 2016/2018 Political Data)
« previous next »
Pages: [1] 2
Author Topic: Redistricting with 2020 Population Estimates (and 2016/2018 Political Data)  (Read 5538 times)
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« on: June 17, 2018, 12:49:29 PM »
« edited: November 19, 2018, 12:42:21 AM by Queen Pelosi, Regina of the House, Regnant of Amerittania 👁 »

In this thread I will look at 2020 redistricting using 2020 population estimates that I made, and also for some states using estimated precinct election results data for 2016 and 2018 elections.

The population estimates can be used directly in Dave's Redistricting App, and I will upload the files with the 2020 data in this thread so that other people can try out the 2020 data as well.

So far I have only done this for a few states. I probably won't bother to do this for literally every single state, but will at least do so for some of the more interesting states where population changes (and perhaps changes in control over redistricting) could make a significant difference. If anyone has a few particular requests for states, I will try to do those states.

The political data estimates so far only consist of Clinton-Trump precinct results for Illinois. Later I will probably add 2018 Senate Beto-Cruz precinct result estimates for TX, 2018 Governor Abrams-Kemp estimates for GA, and Clinton-Trump estimates for NY (maybe something else if I feel like it). For an explanation of how the precinct estimates are made, see this post.

For more 2020 population estimates for more states, see this thread by cvparty.


How I made these estimates

I took 2017 census population estimates and assumed that the same growth rate that has been estimated for 2011-2017 by the census will continue in 2018-2010 to calculate estimated population totals for 2020. These growth rates are then applied equally to the entire population within each county, across all precincts and to all racial groups. I did not do anything to update VAP data, and since this applies the same growth rate to all racial groups, it will not reflect greater-than-average minority population growth.

Projected 2020 race data for some states

However, for a few states I have made the extra effort to incorporate 2020 race data estimates. This is mostly in states where the Voting Rights Act is a particularly important consideration. I did this by combining the 5-year American Community Survey estimates of Race by Hispanic Origin for 2012-2016 with Census county total population estimates for 2017. The way I did this was to project the total population in each county as with other states, using the 2017 population estimates, and then assume that the distribution of growth by race and Hispanic origin within each county would match the distribution of growth from 2010 to the 2012-2016 ACS estimates (which on average can be thought of as 2014 estimates). The same overall growth rate for each racial group is applied to each precinct within each county, which means that population growth for each racial group is assigned within each county in proportion to the 2010 population for each racial group in that county.

So far, states with this race data are: Georgia



How to use these estimates

1) Find the folder where Dave's Redistricting App stores data on your computer. This may be located in a different place depending on your computer's operating system. If you are not sure where to find this, the help page here at DRA may help (look under "saved files"). For example, for me (using Windows 10) this folder is located here for Virginia:

C:\Users\Laptop\AppData\LocalLow\Microsoft\Silverlight\is\tdttpzju.v3n\j44hdtyb.0sj\1\s\mbcuc5y11hatden4rfah5rjgwlpl3hes2ogfqs2jqmkdx103loaaadaa\f\DavesRedistrictingFiles\Virginia

2) Back up your original DRA files. My 2020 population estimates will overwrite the original 2010 census data, so back up your files to make sure you can go back to the 2010 data.

3) Download my 2020 population estimates for the state you are interested in from the link below.

Iowa 2020 Population Estimates DRA File
Georgia 2020 Population Estimates DRA File
Texas 2020 Population Estimates DRA File
Virginia 2020 Population Estimates DRA File

4) When you download the file, you will need to rename it to remove a "_2020estimates" that I added to the filename. For example, for Virginia my file is named "vt51_data_2020estimates.csv". So rename that to "vt51_data.csv". Then copy it into the DRA folder for the relevant state to replace the original "vt51_data.csv" folder that DRA originally downloaded (again, be sure to back up your original file first before replacing it with my new data file).

5) If you have problems restoring the original 2010 census data later, you can always delete your entire DRA folder for that state, and then DRA will re-download the original 2010 census data files the next time you load the state.



Political Data

Illinois 2016 President Precinct Results Estimates DRA File

To use the Political Data, you just do the same thing as with the Population Estimates data - just overwrite your existing file that DRA saves to your computer.



I will keep this original post updated with links for data for new states as I add them.
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #1 on: June 17, 2018, 12:51:20 PM »

Now, I will post some maps I made using the data for Virginia. I was primarily interested in what a Democratic gerrymander might look like if the Supreme Court does not rule against partisan gerrymandering and if Democrats gain control over redistricting in Virginia, and also in what a neutral map might look like.

I was particularly interested in the question of how many Democratic seats NOVA could support, and what sorts of PVI they would have (whether they would be safe seats or not), following from a map that Mizzouian posted (an 8-3 Democratic gerrymander, but one where not all the Democratic seats are very safe and one of them is really a swing seat).

As it turns out, thanks to a combination of the population shift towards NOVA and also Republican self-packing in Appalachia (and to a lesser extent the Richmond exurbs), it is possible to make not just an 8-3 Democratic gerrymander, but actually a 9-2 Democratic gerrymander. This is a no-holds-barred gerrymander, similar to the sort of gerrymander drawn by Ohio Republicans and North Carolina Republicans. At least if one does not make an Arlington-Alexandria Democratic vote sink, NOVA can actually support about 4.5 Democratic districts.

This means that you can have 4 districts safely Democratic based only upon Dem votes in NOVA, but for a 5th district you need to combine parts of NOVA with some Democratic voting areas elsewhere - the only real candidates for that are either Charlottesville or the Richmond suburbs. The main difficulty in drawing a 9-2 Democratic map lies in combing the self-packing Republicans in Appalachia and the Shenandoah with the self-packing Republican cluster around Richmond, while also contiguously combining that 5th NOVA seat with either Charlottesville or with Richmond.

This is a problem that must be overcome, at least if one wants all the districts in NOVA to have solidly Democratic PVIs, because the GOP Richmond exurbs (places like 63% Trump Hanover county) are just a little bit too much for the NOVA districts to swallow without making at least one NOVA district competitive.

The other challenge is in making a safely Democratic seat centered on Charlottesville. It is pretty easy to make this a swing district, but much harder to make it safely Democratic. In order to be safely Democratic, this seat must either include some Democratic voters in NOVA or in the Richmond area, both of which threaten to block contiguity between the Republican Appalachia cluster and the Republican Richmond exurbs cluster.

The best configuration I have found to do this so far is this one:


9-2 Dem Gerrymander







VA-01 (blue) - D+6.7, 57.1%-42.1% Obama '08
VA-02 (dark green) - D+3.8, 55.7%-43.5% Obama '08
VA-03 (purple) - D+7.4, 59.0%-40.3% Obama '08
VA-04 (red) - D+7.5, 57.3%-42.0% Obama '08
VA-05 (yellow) - D+5.2, 57.2%-41.9% Obama '08
VA-06 (teal) - R+20.2, 65.4%-33.7% McCain '08
VA-07 (gray) - D+6.7, 57.2%-41.9% Obama '08
VA-08 (slate blue) - D+6.6, 57.1%-42.1% Obama '08
VA-09 (light blue) - R+22.6, 61.6%-37.0% McCain '08
VA-10 (pink) - D+6.9, 57.8%-41.3% Obama '08
VA-11 (light green) - D+6.4, 55.0%-44.2% Obama '08

The PVI of VA-02 is a bit lower than one might like for a Democratic gerrymander, but I don't think it is possible to do substantially better without reducing Black voters' control over VA-03. As drawn, VA-03 is 38.7% black and 51.1% white (based off of 2010 race data, so with 2020 data it would surely be majority-minority), which is about the same as the current VA-03.

The main point of potential controversy here is VA-04, which is centered more on Richmond than the actual current VA-04. I originally tried making VA-04 be a compact purely Richmond district, but the problem with that is it ends up at about D+12 and wastes Democratic votes - meanwhile that leaves VA-05 competitive, rather than fairly safe or at least strongly Dem leaning (D+5). So I extended an arm of VA-05 into Richmond to pick up the white liberal areas of central Richmond.

With 2010 race data, this VA-04 is 51.8% white, 34.7% black, 6.8% hispanic, 4.2% asian. So it would be majority-minority with 2020 data. By comparison, the current VA-04 is 51.5% white, 41.1% black. So this district has somewhat fewer, but basically the same overall minority population. Despite the slightly lower black population, black voters' control of the district may actually be increased, because the white liberal areas of Richmond are put in VA-05, so black voters may actually even make up a greater share of Democratic primary voters.

An alternative configuration is to have VA-05 include Charlottesville rather than part of Richmond, but then the problem is contiguity for VA-06. This also doesn't do anything to make the PVI of VA-05 more Democratic.



Alternative 9-2 Dem Gerrymander

If the reduction in the black population (and increase in the non-black minority population) in VA-04 is not acceptable to the VRA, I made an alternative 9-2 map. In this map, VA-04 is more similar to the current district, and rather than having the 5th NOVA seat combine Charlottesville with part of NOVA, the 5th NOVA seat combines part or the Richmond suburbs with NOVA. Meanwhile, VA-05 combines Charlottesville with another part of the Richmond suburbs.

The main benefit of this alternative map is that VA-04 is more similar to the current district, and there is no possibility of VRA issues. The main downside is that several of the Democratic districts are a bit less safe. The change in PVI relative to the first map is shown in parentheses:







VA-01 (blue) - D+5.7 (-1), 55.9%-42.1% Obama '08
VA-02 (dark green) - D+3.7 (-.1), 55.5%-43.7% Obama '08
VA-03 (purple) - D+6.4 (-1), 58.0%-41.3% Obama '08
VA-04 (red) - D+6.7 (-.8 ), 58.7%-40.5% Obama '08
VA-05 (yellow) - D+4.8 (-.4), 55.8%-43.3% Obama '08
VA-06 (teal) - R+20.8 (+.6), 62.3%-36.6% McCain '08
VA-07 (gray) - D+6.8 (+.1), 56.2%-43.0% Obama '08
VA-08 (slate blue) - D+6.7 (+.1), 56.9%-42.3% Obama '08
VA-09 (light blue) - R+18.8 (-2.8 ), 63.0%-35.8% McCain '08
VA-10 (pink) - D+6.6 (-.3), 57.5%-41.6% Obama '08
VA-11 (light green) - D+6.9 (+.5), 56.3%-42.9% Obama '08


As you can see, most of the Democratic districts are a fraction of a point to a point less Democratic. Either I just did a worse job in making the districts Democratic in this map, or else this configuration (NOVA to Richmond district rather than NOVA to Charlottesville) simply results in a less efficient distribution of votes. So personally I prefer the first configuration, as long as lowering the black population in VA-04 (but keeping the overall minority population the same) is acceptable for VRA purposes.



Next, here is an 8-3 Dem Gerrymander.

8-3 Dem Gerrymander







VA-01 (blue) - R+13.4, 58.0%-41.1% McCain '08
VA-02 (dark green) - D+3.7, 55.5%-43.7% Obama '08
VA-03 (purple) - D+6.4, 58.0%-41.3% Obama '08
VA-04 (red) - D+6.7, 58.7%-40.5% Obama '08
VA-05 (yellow) - D+8.1, 57.9%-41.3% Obama '08
VA-06 (teal) - R+19.1, 63.0%-36.0% McCain '08
VA-07 (gray) - D+9.3, 57.6%-41.6% Obama '08
VA-08 (slate blue) - D+9.0, 59.5%-39.7% Obama '08
VA-09 (light blue) - R+16.6, 56.7%-41.9% McCain '08
VA-10 (pink) - D+10.3, 60.0%-39.2% Obama '08
VA-11 (light green) - D+9.2, 57.6%-41.6% Obama '08

This actually makes VA-05 safer than VA-04, so I should probably switch some voters between those two districts.

In my opinion, this is just wasting votes. It makes the NOVA districts even safer, but it doesn't do anything to help the most vulnerable Democratic district (VA-02). It does make the map look a bit cleaner, but it still is obviously gerrymandered. It would still get just as bad press as a 9-2 map, but for worse results (from a D partisan perspective). So in my opinion, if you are going to gerrymander, you may as well do it properly and go for a 9-2 map.

With a 9-2 map, is it possible that Republicans might temporarily pick up 1 or 2 of the D+5 to D+7 districts in a really good GOP wave year? Sure. But if so, Republicans would be winning by enough nationally that they would definitely win control of Congress anyway, and the seats could be won back in the first non-GOP wave year.



Finally, here is a neutral non-partisan map. If the Supreme Court rules against partisan gerrymandering and we have a truly level playing field nationwide, we might hope to get a map that is something like this.

Non-Partisan map (leaning 6-5 with some competitive districts)



VA-01 (blue) - R+3.2, 51.0%-48.3% McCain '08
VA-02 (dark green) - D+3.2, 55.0%-44.1% Obama '08
VA-03 (purple) - D+7.2, 58.6%-40.7% Obama '08
VA-04 (red) - D+6.2, 56.8%-42.5% Obama '08
VA-05 (yellow) - R+5.2, 50.6%-48.5% McCain '08
VA-06 (teal) - R+20.3, 59.5%-39.2% McCain '08
VA-07 (gray) - R+2.7, 50.1%-49.1% Obama '08
VA-08 (slate blue) - D+21.7, 68.6%-30.6% Obama '08
VA-09 (light blue) - R+13.0, 56.6%-42.3% McCain '08
VA-10 (pink) - D+7.2, 56.5%-42.8% Obama '08
VA-11 (light green) - D+9.7, 57.2%-42.0% Obama '08

Unlike the Democratic gerrymanders above, this map is fairly well reflective of Virginia's actual politics. There are 6 districts that lean or are safe D, and 5 that lean or are safe R. On both sides, there are 1 or 2 competitive districts that could go the other way than their lean, especially VA-01, VA-02, and VA-07, but maybe also VA-04 and VA-05 in wave years.
Logged
America Needs a 13-6 Progressive SCOTUS
Solid4096
Junior Chimp
*****
Posts: 8,730


Political Matrix
E: -8.88, S: -8.51

P P P
Show only this user's posts in this thread
« Reply #2 on: June 17, 2018, 03:29:23 PM »

Nice work. Can you please do this for more states?
Logged
cvparty
Sr. Member
****
Posts: 2,100
United States


Show only this user's posts in this thread
« Reply #3 on: June 17, 2018, 05:52:40 PM »

I got one for Georgia. You can look at the deviations for how much each district is growing (hint: Atlanta is growing fast)
1st, 3rd, and 4th are competitive to some extent, with R+7, D+2, R+5 PVIs, respectively
The Cobb- and Gwinnett-based districts are currently D+3 but would likely become solid D by 2022

Logged
Lachi
lok1999
Sr. Member
****
Posts: 3,347
Australia


Political Matrix
E: -1.06, S: -3.02

Show only this user's posts in this thread
« Reply #4 on: June 17, 2018, 05:57:03 PM »

Holy crap, this is amazing.
Logged
America Needs a 13-6 Progressive SCOTUS
Solid4096
Junior Chimp
*****
Posts: 8,730


Political Matrix
E: -8.88, S: -8.51

P P P
Show only this user's posts in this thread
« Reply #5 on: June 17, 2018, 06:14:01 PM »

I would like to see Iowa in particular, because there is a relatively high chance in Iowa compared to other states that someone could put together what ends up being the real map (or at least extremely close to the real map) using these estimates.
Logged
Nyvin
Junior Chimp
*****
Posts: 7,623
United States


Show only this user's posts in this thread
« Reply #6 on: June 18, 2018, 12:22:23 PM »
« Edited: June 18, 2018, 12:40:19 PM by Nyvin »

I would like to see Iowa in particular, because there is a relatively high chance in Iowa compared to other states that someone could put together what ends up being the real map (or at least extremely close to the real map) using these estimates.

Here's what I made with 2017 estimates:



2012-2016 pvi's:

IA-1(Blue): R+3.28
IA-2(Green): D+4.78
IA-3(Purple): D+2.35
IA-4(Red): R+13.88

Really shows just how empty the western half of the state is without the Des Moines metro.  Also shows how "urban packing" can actually work to the Democrat's advantage in Iowa if the same principal is applied here in R gerrymandered states.

Here's the district's populations by 2017 estimates, obviously 1 is overpopulated and 3 is under populated (should be 786,428 per district), but the only real way to correct it would be a county split, which would mess up the estimates.

IA-1
TOTAL   793860
   
IA-2   
TOTAL   786530
   
IA-3   
TOTAL   776994
   
IA-4   
TOTAL   788884


https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk
Logged
Gass3268
Moderators
Atlas Star
*****
Posts: 27,478
United States


Show only this user's posts in this thread
« Reply #7 on: June 18, 2018, 12:26:36 PM »

I would like to see Iowa in particular, because there is a relatively high chance in Iowa compared to other states that someone could put together what ends up being the real map (or at least extremely close to the real map) using these estimates.

Here's what I made with 2017 estimates:



2012-2016 pvi's:

IA-1(Blue): R+3.28
IA-2(Green): D+4.78
IA-3(Purple): D+2.35
IA-4(Red): R+13.88

Really shows just how empty the western half of the state is without the Des Moines metro.  Also shows how "urban packing" can actually work to the Democrat's advantage in Iowa if the same principal is applied here in R gerrymandered states.

Here's the district's populations by 2017 estimates, obviously 1 is overpopulated and 2 is under populated (should be 786,428 per district), but the only real way to correct it would be county splits, which would mess up the estimates.

IA-1
TOTAL   793860
   
IA-2   
TOTAL   776377
   
IA-3   
TOTAL   787147
   
IA-4   
TOTAL   788884


https://factfinder.census.gov/faces/tableservices/jsf/pages/productview.xhtml?src=bkmk


I like this map, but IA-02 would probably violate their rules on compact districts.
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #8 on: June 18, 2018, 03:43:48 PM »

OK, I did Iowa. The original post is updated with a link to the Iowa 2020 population estimates DRA file. If you want to try using the data, as with the Virginia data you just need to download the file and overwrite your existing file in your Iowa DRA folder.



Here is a possible map using the 2020 estimates for Iowa which does not split any counties. It is pretty similar to the map Nyvin posted, but IA-01 an IA-02 are more compact. The population deviations are -249, 1865, -1841, and 226. You may be able to get the deviations down further without splitting counties with some alternative configurations, but the deviations are clearly well within the error of the estimates.



Since Iowa likes to avoid splitting counties, very slight error between population projections and the actual 2020 census data could lead to a significantly different map. But it does seem like it may be possible to have a neat compact Des Moines area map, 2 fairly compact districts in the east, and 1 in the west.

Iowa is another state where Republicans tend to be self-packed in the western part of the state. So it is pretty much inevitable that IA-04 will in some form or other be safe R, and the other 3 districts swing districts to some extent or other, unless maybe IA-04 is re-oriented in the northern part of the state instead of the west.

IA-01 is D + 0.4
IA-02 is D + 0.8
IA-03 is D + 2.3
IA-04 is R + 13.3
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #9 on: June 18, 2018, 04:04:00 PM »

Here's an alternate map for Iowa, pretty similar to the other one. The population deviations are all below 1000 on this one.



The districts look maybe slightly more compact overall, but less so the Des Moines district (IA-03) and the Des Moines IA southern suburbs/exurbs are split from IA-03.

IA-01 is D + 0.5
IA-02 is R + 0.3
IA-03 is D + 2.9
IA-04 is R + 12.9

Overall I would guess the 2020 map may look broadly similar to these two maps.
Logged
Nyvin
Junior Chimp
*****
Posts: 7,623
United States


Show only this user's posts in this thread
« Reply #10 on: June 18, 2018, 07:43:11 PM »
« Edited: June 18, 2018, 07:54:05 PM by Nyvin »



I like this map, but IA-02 would probably violate their rules on compact districts.

Good point,  convinced me to redo it.

Here's a much improved version, aside from splitting up Linn and Johnson counties (yuck!).



2012-2016 PVI

IA-1: D+0.6
IA-2: D+1.26
IA-3: D+1.8
IA-4: R+13.88

District populations by 2017 estimates:

IA-1: 790699
IA-2: 780008
IA-3: 786801
IA-4: 788884

Tama County can be split to even out the populations of IA-1 and IA-2.  I don't think it gets much better than this for compactness.   Three swing seats too.
Logged
Nyvin
Junior Chimp
*****
Posts: 7,623
United States


Show only this user's posts in this thread
« Reply #11 on: June 18, 2018, 07:48:01 PM »

OK, I did Iowa. The original post is updated with a link to the Iowa 2020 population estimates DRA file. If you want to try using the data, as with the Virginia data you just need to download the file and overwrite your existing file in your Iowa DRA folder.



Here is a possible map using the 2020 estimates for Iowa which does not split any counties. It is pretty similar to the map Nyvin posted, but IA-01 an IA-02 are more compact. The population deviations are -249, 1865, -1841, and 226. You may be able to get the deviations down further without splitting counties with some alternative configurations, but the deviations are clearly well within the error of the estimates.



Since Iowa likes to avoid splitting counties, very slight error between population projections and the actual 2020 census data could lead to a significantly different map. But it does seem like it may be possible to have a neat compact Des Moines area map, 2 fairly compact districts in the east, and 1 in the west.

Iowa is another state where Republicans tend to be self-packed in the western part of the state. So it is pretty much inevitable that IA-04 will in some form or other be safe R, and the other 3 districts swing districts to some extent or other, unless maybe IA-04 is re-oriented in the northern part of the state instead of the west.

IA-01 is D + 0.4
IA-02 is D + 0.8
IA-03 is D + 2.3
IA-04 is R + 13.3

This look pretty good!   Your map pretty much confirms to me that it's near impossible to have a compact map and keep Linn and Johnson counties in the same district, lol.
Logged
America Needs a 13-6 Progressive SCOTUS
Solid4096
Junior Chimp
*****
Posts: 8,730


Political Matrix
E: -8.88, S: -8.51

P P P
Show only this user's posts in this thread
« Reply #12 on: June 18, 2018, 08:05:09 PM »

There is a misplaced precinct in that map in Dallas County, Iowa between Districts 3 and 4.
Logged
America Needs a 13-6 Progressive SCOTUS
Solid4096
Junior Chimp
*****
Posts: 8,730


Political Matrix
E: -8.88, S: -8.51

P P P
Show only this user's posts in this thread
« Reply #13 on: June 18, 2018, 09:29:08 PM »

Does anyone have a chart of state by state predicted populations for 2020 for excel?
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #14 on: June 18, 2018, 10:47:45 PM »

This look pretty good!   Your map pretty much confirms to me that it's near impossible to have a compact map and keep Linn and Johnson counties in the same district, lol.

This seems to be about the best that can be managed while keeping Johnson and Linn together. But it has a population deviation of a bit more than 2000 on IA-04, and Linn county is sticking out awkwardly.

Logged
Nyvin
Junior Chimp
*****
Posts: 7,623
United States


Show only this user's posts in this thread
« Reply #15 on: June 19, 2018, 11:25:07 AM »

This look pretty good!   Your map pretty much confirms to me that it's near impossible to have a compact map and keep Linn and Johnson counties in the same district, lol.

This seems to be about the best that can be managed while keeping Johnson and Linn together. But it has a population deviation of a bit more than 2000 on IA-04, and Linn county is sticking out awkwardly.



Nice,  what are the PVI's for that?  It looks like the map the Republicans would push for since it creates somewhat of a Dem Pack in the southeast district.
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #16 on: June 19, 2018, 09:30:24 PM »


Nice,  what are the PVI's for that?  It looks like the map the Republicans would push for since it creates somewhat of a Dem Pack in the southeast district.

IA-01 is D+5.0 (this is the one with Johnson and Linn, I accidentally switched IA-01 and IA-02)
IA-02 is R+1.7
IA-03 is R+0.3
IA-04 is R+13.5



Next up will be Georgia. For Georgia, I actually managed to get 2020 projections by race into the data as well as total population for Georgia, by combining the 2017 county level population estimates with race data for 2012-2016. Basically I used the 2017 data projected forward linearly to project the total population in each county in 2020, and used the ACS race/Hispanic origin data also projected forward linearly to determine the racial composition of that total growth. This methodology seems to match pretty closely with the more up to date estimates from the census for Georgia statewide. That makes it possible to actually look forward in a meaningful way to what districts might be like in the context of the VRA. This process is significantly more laborious in making the estimates though, so I don't know how many states I will do this for, but it can be done for states where race data is really needed because of the VRA.
Logged
muon2
Moderator
Atlas Icon
*****
Posts: 16,788


Show only this user's posts in this thread
« Reply #17 on: June 20, 2018, 02:19:48 PM »

I think it's useful to keep in mind how much the projections based on estimates will change as each new year comes out. Last year jimrtex launched a thread to look at a 2020 map of IA by projecting 2016 estimates to 2020. I used his data and type of methodology to make this map:



At that time the districts had minimal deviations (rounded to the nearest 0.1%): SE +0.1%, NE -0.1%, C 0.0%, W 0.0%.

I updated my spreadsheet to use the 2017 estimates that came out earlier this year. The deviations are now SE +0.2%, NE +0.2%, C+0.3%, W -0.7%. That means the change from last year would be SE +0.1%, NE +0.3%, C +0.3%, W -0.7%. Each 0.1% is about 800 people, so in just one year the changes were enough to cause the 2020 projection for the western CD to dip by over 5000 people compared to the 2020 projection based on the prior year. Think about what deviation in the 2020 projections will be after 3 more years.

IA has very strict rules governing their maps. Based on the number of counties it is statistically likely that no CD will deviate more than 100 persons from the quota even with whole counties. That means it is highly unlikely that any of us will correctly guess today the map that IA will adopt in 2021.

IA is made difficult because it is so thoroughly driven by the cold calculus of geography and population. In states that have humans chop counties to exact equality it is far more likely that we might get close if we can get inside the heads of those who will draw the map.
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #18 on: June 22, 2018, 12:33:06 PM »

The original post is updated with a link to the DRA file for Georgia projections for 2020. This includes projections for Race and Hispanic Origin for 2020, not just total population as previous states so far have had.



Now for some maps made with the data. If the Georgia GOP retains full control of redistricting, they will presumably draw a map similar to the current map, but probably will try to split up the Atlanta suburbs a bit more to try to dilute Democratic voters and keep all the districts such as GA-06 and GA-07 safe Republican. Although possibly, they may be forced to concede another majority minority district or two because of minority population growth and the VRA.

However, if Stacey Abrams wins the Georgia gubernatorial race, we would get something different - a court drawn map after she vetoes the Republican gerrymander.

Court Drawn Map #1 (GOP Leaning Semi-Least-Change)

Here is one hypothetical court drawn map. This map is generally similar to the pre-existing gerrymander drawn by the Georgia GOP, but districts are sucked into the Atlanta area due to population growth, and the districts are generally a bit more compact and follow county lines more. Since it generally follows the current Republican gerrymander (maybe not exactly a "least change" map, but something like it), this is probably about the best court-drawn map that the GOP could hope for. The North Atlanta suburbs districts (GA-06, GA-07, and GA-11) all have Republican PVIs, but Clinton probably won most and quite possibly all of them. So those may all be swing districts for the moment.



GA-01: R+6.9; 57.9% White, 30.7% Black, 46.1% Obama '08, 53.4% McCain '08
GA-02: D+3.7; 40.1% White, 50.1% Black, 55.5% Obama '08, 44.0% McCain '08
GA-03: R+20.2; 70.7% White, 20.6% Black, 33.3% Obama '08, 65.9% McCain '08
GA-04: D+30.9; 25.5% White, 58.8% Black, 80.3% Obama '08, 19.1% McCain '08
GA-05: D+30.1; 24.9% White, 64.0% Black, 79.4% Obama '08, 20.0% McCain '08
GA-06: R+7.0; 55.2% White, 14.2% Black, 43.1% Obama '08, 56.1% McCain '08
GA-07: R+2.2; 36.6% White, 30.3% Black, 44.6% Obama '08, 54.6% McCain '08
GA-08: R+19.7; 60.7% White, 28.0% Black, 34.4% Obama '08, 64.9% McCain '08
GA-09: R+31.4; 79.1% White, 4.4% Black, 22.2% Obama '08, 76.7% McCain '08
GA-10: R+22.1; 66.9% White, 14.3% Black, 32.3% Obama '08, 66.7% McCain '08
GA-11: R+3.9; 50.3% White, 28.1% Black, 45.1% Obama '08, 54.0% McCain '08
GA-12: R+6.5; 54.4% White, 37.2% Black, 46.4% Obama '08, 53.1% McCain '08
GA-13: D+7.4; 31.3% White, 51.8% Black, 56.4% Obama '08, 43.0% McCain '08
GA-14: R+26.9; 78.4% White, 8.6% Black, 28.1% Obama '08, 70.8% McCain '08



Court Drawn Map #2 (Dem Leaning Not Based on Current Map)

Next, here is another hypothetical court-drawn map. This one does NOT follow the current GOP gerrymander as a starting point, although it does keep GA-02 the same (presuming it will probably stay about the same regardless because of the VRA).

Whereas the previous map is what Republicans might hope for from a court drawn map, this is more like what Democrats could hope for from a court drawn map. GA-12 is redrawn to combine Savannah and Augusta, which makes it a narrowly majority-minory (but plurality white) district more similar to the old GA-12. Atlanta is partially unpacked, but DeKalb County (which has almost perfect population for a district) remains packed. As with the other court-drawn map, GA-06, GA-07, and GA-11 all have somewhat Republican PVIs, but Clinton probably won them and they are winnable for Democrats. In addition to that, however, GA-10 also becomes fairly competitive, combining Athens with parts of the Atlanta suburbs.



GA-01: R+18.8; 62.5% White, 25.8% Black, 35.4% Obama '08, 64.0% McCain '08
GA-02: D+3.7; 40.1% White, 50.1% Black, 55.5% Obama '08, 44.0% McCain '08
GA-03: R+19.2; 69.7% White, 20.7% Black, 34.1% Obama '08, 64.9% McCain '08
GA-04: D+4.8; 29.1% White, 53.3% Black, 78.9% Obama '08, 20.5% McCain '08
GA-05: D+29.2; 41.9% White, 47.2% Black, 54% Obama '08, 45.4% McCain '08
GA-06: D+6.9; 52.4% White, 27.4% Black, 59.2% Obama '08, 40.1% McCain '08
GA-07: R+4.2; 35.9% White, 23.9% Black, 42.9% Obama '08, 56.2% McCain '08
GA-08: R+15.6; 60.7% White, 29.9% Black, 37.6% Obama '08, 61.7% McCain '08
GA-09: R+29.8; 71.5% White, 5.8% Black, 22.7% Obama '08, 76.4% McCain '08
GA-10: R+3.7; 50.5% White, 35.2% Black, 46.3% Obama '08, 52.9% McCain '08
GA-11: R+3.9; 50.5% White, 27.9% Black, 45.0% Obama '08, 54.1% McCain '08
GA-12: D+2.5; 49.6% White, 40.3% Black, 54.6% Obama '08, 44.9% McCain '08
GA-13: D+4.6; 37.2% White, 48.5% Black, 55.2% Obama '08, 44.2% McCain '08
GA-14: R+31.3; 79.8% White, 5.5% Black, 25.1% Obama '08, 73.8% McCain '08



9-5 Democratic Gerrymander Map

Finally, here is a Democratic gerrymander. No, this is not going to happen (there is no way Georgia Democrats will win control of the state legislature in 2018), but it is interesting to see as a what-if.

It seems possible in a Democratic gerrymander to make 9 safely Democratic seats to 5 Republican vote sinks. As compared to the court drawn maps, the Democratic votes in Atlanta are used more efficiently to make 7 Democratic seats, while GA-02 and GA-12 are strengthened and made more safe D.



Interestingly, all the Democratic and Republican districts are in contiguous blobs with the single exception of GA-12:





GA-01: R+21.2; 66.4% White, 21.5% Black, 32.4% Obama '08, 67.0% McCain '08
GA-02: D+6.7; 37.5% White, 53.7% Black, 59.3% Obama '08, 40.2% McCain '08
GA-03: R+28.9; 81.0% White, 7.0% Black, 26.6% Obama '08, 72.3% McCain '08
GA-04: D+7.7; 41.2% White, 50.3% Black, 58.9% Obama '08, 40.5% McCain '08
GA-05: D+7.6; 37.6% White, 50.8% Black, 56.4% Obama '08, 43.0% McCain '08
GA-06: D+6.3; 55.8% White, 19.6% Black, 58.2% Obama '08, 41.0% McCain '08
GA-07: D+6.2; 30.9% White, 26.9% Black, 54.8% Obama '08, 44.3% McCain '08
GA-08: R+19.1; 63.1% White, 26.0% Black, 33.8% Obama '08, 65.6% McCain '08
GA-09: R+24.4; 72.7% White, 9.6% Black, 27.4% Obama '08, 71.7% McCain '08
GA-10: D+6.1; 40.7% White, 38.3% Black, 55.2% Obama '08, 44.0% McCain '08
GA-11: D+6.4; 41.5% White, 37.6% Black, 55.4% Obama '08, 43.8% McCain '08
GA-12: D+6.7; 45.8% White, 45.2% Black, 58.5% Obama '08, 41.0% McCain '08
GA-13: D+7.5; 35.8% White, 49.3% Black, 57.9% Obama '08, 41.5% McCain '08
GA-14: R+33.9; 82.0% White, 5.1% Black, 21.5% Obama '08, 77.4% McCain '08

GA-09's shape is probably unnecessary - I only drew it that way in the vague hope that maybe if the North Atlanta suburbs keep trending D, it could some day (maybe in 2028 or 2030) becomes somewhat competitive. I doubt that would really happen, but it would be interesting to see the Clinton-Trump numbers in this district to judge if that is at all within the realm of possibility.
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #19 on: June 22, 2018, 01:10:46 PM »

IA is made difficult because it is so thoroughly driven by the cold calculus of geography and population. In states that have humans chop counties to exact equality it is far more likely that we might get close if we can get inside the heads of those who will draw the map.

Agreed. A small amount of error can make a big difference in Iowa. Not only are there changes in each year's population estimates, but there are also always some surprises when the actual census data comes out, in comparison to the estimates.
Logged
muon2
Moderator
Atlas Icon
*****
Posts: 16,788


Show only this user's posts in this thread
« Reply #20 on: June 22, 2018, 01:31:22 PM »
« Edited: June 22, 2018, 01:35:34 PM by muon2 »

For example, I took the 7/1/2017 estimates for IA and projected the populations forward to 4/1/2020. I started with the 4-corner model as used for the current map. I'm sure this isn't the best solution to the state requirements, and based on the inaccuracy of the estimates won't be the actual map, but it is whole county and provides all CDs within 100 of the quota.

Logged
America Needs a 13-6 Progressive SCOTUS
Solid4096
Junior Chimp
*****
Posts: 8,730


Political Matrix
E: -8.88, S: -8.51

P P P
Show only this user's posts in this thread
« Reply #21 on: June 23, 2018, 06:33:56 PM »

Dallas County, Iowa has had some major population growth.
Logged
muon2
Moderator
Atlas Icon
*****
Posts: 16,788


Show only this user's posts in this thread
« Reply #22 on: June 24, 2018, 11:46:31 AM »

Dallas County, Iowa has had some major population growth.

Of course it has. Dallas county has been averaging 3.9% growth per year so far this decade. That's why I don't use the current estimate, I forecast a projection to 2020 based on that rate of growth. For example the 7/1/17 estimate is 87,235 but I use a 2020 population of 96,896.

The question here is whether the rate of population growth in Dallas county will be faster or slower over the next 3 years than it has been over the previous 7. Are you suggesting that the growth rate will be significantly higher in the last years of this decade than it currently is?
Logged
America Needs a 13-6 Progressive SCOTUS
Solid4096
Junior Chimp
*****
Posts: 8,730


Political Matrix
E: -8.88, S: -8.51

P P P
Show only this user's posts in this thread
« Reply #23 on: June 24, 2018, 12:28:20 PM »

Dallas County, Iowa has had some major population growth.

Of course it has. Dallas county has been averaging 3.9% growth per year so far this decade. That's why I don't use the current estimate, I forecast a projection to 2020 based on that rate of growth. For example the 7/1/17 estimate is 87,235 but I use a 2020 population of 96,896.

The question here is whether the rate of population growth in Dallas county will be faster or slower over the next 3 years than it has been over the previous 7. Are you suggesting that the growth rate will be significantly higher in the last years of this decade than it currently is?

I was just noting how surprising it is that the County has grown so much. It has almost caught up to Story County in population based on these 2020 estimates, even as Story County has also been growing.
Logged
Former Dean Phillips Supporters for Haley (I guess???!?) 👁️
The Impartial Spectator
Junior Chimp
*****
Posts: 7,761


Show only this user's posts in this thread
« Reply #24 on: June 24, 2018, 02:24:18 PM »

Dallas County, Iowa has had some major population growth.

Of course it has. Dallas county has been averaging 3.9% growth per year so far this decade. That's why I don't use the current estimate, I forecast a projection to 2020 based on that rate of growth. For example the 7/1/17 estimate is 87,235 but I use a 2020 population of 96,896.

The question here is whether the rate of population growth in Dallas county will be faster or slower over the next 3 years than it has been over the previous 7. Are you suggesting that the growth rate will be significantly higher in the last years of this decade than it currently is?

The methodology that I used to make my estimates is slightly different than what you used. This makes only a small difference, but it does make some difference - for example, the number I have in my DRA files for Dallas County IA is 96,278 instead of 96,896. Your method will estimate slightly higher population gains (or losses) than mine.

In the case of Iowa, that small difference of 618 people is enough to change the map, if they are trying to limit population deviations below e.g. 100 people.

What I did was to apply the growth rate for each additional year to the 2010 base estimate. What it sounds like you did was to apply the growth rate for 2018 to the 2017 base population estimate, the growth rate for 2019 to the 2018 base population estimate, etc.

I am not sure which is better; your method is more strictly mathematically adhering to a constant growth rate, in the sense in which a constant growth rate is used for calculating things like interest payments, whereas my method only has a constant growth rate with reference to a constant base population (the 2010 estimates). This means that my method adds an equal number of people each year, whereas your method adds more people each year (assuming positive growth).

The best case to be made for my method is that there is an overall downward trend in population growth rates in general across the country. I don't know this for sure, but it seems to me that when urban/suburban counties are growing, they don't very often (at least in the current stage of development) start off by adding a smaller number of people and then a larger number of people (although that could be the case for some rural counties that are becoming exurban).

The larger a county gets, the more its growth rate tends to slow down, for various reasons. First of all, population growth is declining in general across the US and across the world. Second, in the case of large metro areas, after growth has gone on for a while, there tends to be less land available in an urbanizing county, and growth moves a bit more to other counties, which then become suburban/exurban counties.

In truth, I would guess your method may be accurate for some counties/states, and mine for others. I would guess mine is probably better for most large urban counties, whereas yours may be better for counties that are in the early stages of suburbanization/urbanization.

The wild card is population loss - your method implies that if a county is losing population, it will lose a greater absolute # of people per year. That in turn implies, at least if one considers no in-migration, that a greater percentage of the existing population must be dying or moving away every year. But who knows, that could actually be the more accurate way to model some places like Detroit and to a lesser extent Cleveland.
Logged
Pages: [1] 2  
« previous next »
Jump to:  


Login with username, password and session length

Terms of Service - DMCA Agent and Policy - Privacy Policy and Cookies

Powered by SMF 1.1.21 | SMF © 2015, Simple Machines

Page created in 0.372 seconds with 12 queries.