Closed Jfer09 closed 1 year ago
In South Carolina, districts must:
https://redistricting.scsenate.gov/Documents/RedistrictingGuidelinesAdopted041311.pdf https://redistricting.schouse.gov/archives/2011/6334-1500-2011-Redistricting-Guidelines-(A0404871).pdf
We do not adhere to all criteria in the guidelines. We include the following constraints:
Data for South Carolina comes from the ALARM Project's 2020 Redistricting Data Files. <- not sure what I should put for 2010 because I couldn't find it in the ALARM website :(
No manual pre-processing decisions were necessary.
We sample 6,000 districting plans for South Carolina. No special techniques were needed to produce the sample.
R-hat values for summary statistics:
pop_overlap total_vap plan_dev comp_edge comp_polsby pop_white pop_black pop_hisp
1.0017184 1.0076736 1.0000837 1.0014655 1.0080891 1.0053784 1.0150619 1.0100801
pop_aian pop_asian pop_nhpi pop_other pop_two vap_white vap_black vap_hisp
0.9998978 1.0005043 1.0024173 1.0025930 1.0016946 1.0043995 1.0126980 1.0093582
vap_aian vap_asian vap_nhpi vap_other vap_two pre_16_rep_tru pre_16_dem_cli pre_20_rep_tru
1.0001556 1.0006002 1.0041020 1.0017029 1.0025194 1.0040227 1.0061493 1.0035632
pre_20_dem_bid uss_16_rep_sco uss_16_dem_dix uss_20_rep_gra uss_20_dem_har gov_18_dem_smi gov_18_rep_mcm atg_18_dem_ana
1.0013353 1.0022584 1.0117500 1.0052989 1.0006937 1.0032953 1.0096629 1.0029299
atg_18_rep_wil sos_18_dem_whi sos_18_rep_ham adv_16 adv_18 adv_20 arv_16 arv_18
1.0085896 1.0078122 1.0090309 1.0093736 1.0033046 1.0016536 1.0023676 1.0089577
arv_20 county_splits muni_splits ndv nrv ndshare e_dvs pr_dem
1.0043677 1.0082489 1.0058738 1.0012658 1.0053194 1.0049363 1.0049391 1.0009597
e_dem pbias egap
1.0021474 1.0082487 1.0003794
Sampling diagnostics for SMC run 1 of 2 (3,000 samples)
Eff. samples (%) Acc. rate Log wgt. sd Max. unique Est. k
Split 1 2,850 (95.0%) 11.1% 0.32 1,882 ( 99%) 9
Split 2 2,722 (90.7%) 15.1% 0.44 1,839 ( 97%) 6
Split 3 2,637 (87.9%) 20.0% 0.50 1,792 ( 94%) 4
Split 4 2,554 (85.1%) 14.7% 0.57 1,779 ( 94%) 5
Split 5 2,398 (79.9%) 18.4% 0.63 1,758 ( 93%) 3
Split 6 2,257 (75.2%) 8.2% 0.69 1,475 ( 78%) 2
Resample 1,186 (39.5%) NA% 1.22 1,454 ( 77%) NA
Sampling diagnostics for SMC run 2 of 2 (3,000 samples)
Eff. samples (%) Acc. rate Log wgt. sd Max. unique Est. k
Split 1 2,849 (95.0%) 14.4% 0.33 1,902 (100%) 7
Split 2 2,719 (90.6%) 18.5% 0.43 1,854 ( 98%) 5
Split 3 2,607 (86.9%) 20.3% 0.53 1,825 ( 96%) 4
Split 4 2,525 (84.2%) 23.0% 0.62 1,764 ( 93%) 3
Split 5 2,418 (80.6%) 24.3% 0.64 1,767 ( 93%) 2
Split 6 2,282 (76.1%) 8.2% 0.63 1,521 ( 80%) 2
Resample 1,241 (41.4%) NA% 1.16 1,453 ( 77%) NA
TODO
lines from the template code have been removedenforce_style()
to format my coderedist_map
and redist_plans
objects, and summary statistics) have been edited@kuriwaki
The performance plot is giving signs that the most Black district ("Ordered district 7") in a decent chunk of the plans are not performing, i.e. they are more Republican than Democrat. That might not be compliant, especially given that 2010 districts would have fallen under pre-clearance. So we may need to strengthen the constraints so we are left with all performant districts.
@Jfer09 can we try summary statistics like the below and see what the numbers are? You can check out how it was summarized in 2020 in the PR post, right under the performance plot.
In South Carolina, districts must:
https://redistricting.scsenate.gov/Documents/RedistrictingGuidelinesAdopted041311.pdf https://redistricting.schouse.gov/archives/2011/6334-1500-2011-Redistricting-Guidelines-(A0404871).pdf
We do not adhere to all criteria in the guidelines. We include the following constraints:
Data for South Carolina comes from the ALARM Project's 2020 Redistricting Data Files. <- not sure what I should put for 2010 because I couldn't find it in the ALARM website :(
No manual pre-processing decisions were necessary.
We sample 6,000 districting plans across two independent runs of the SMC algorithm. We then remove all plans that do not contain any district that has both a BVAP of over 30% and an average vote share that is more Democratic than Republican. This removal occurs after verifying that such plans comprise less than 1% of the 6,000 plans. We then thin the sample down to exactly 5,000 plans. We also set the population tempering to 0.01 to avoid bottlenecks.
91% of the highest-BVAP district and 28.5% of the second highest-BVAP district are blue in this plot.
n_black_perf n
1 1 3996
2 2 997
3 3 7
Summary of Plan Object:
SMC: 6,000 sampled plans of 7 districts on 2,122 units
`adapt_k_thresh`=0.985 • `seq_alpha`=0.5
`est_label_mult`=1 • `pop_temper`=0.01
Plan diversity 80% range: 0.48 to 0.83
R-hat values for summary statistics:
pop_overlap total_vap plan_dev comp_edge comp_polsby pop_white
1.0120475 1.0042467 1.0016465 1.0014947 1.0304959 1.0015804
pop_black pop_hisp pop_aian pop_asian pop_nhpi pop_other
1.0050795 1.0005244 1.0076705 1.0002316 1.0035516 1.0016537
pop_two vap_white vap_black vap_hisp vap_aian vap_asian
1.0008869 1.0011788 1.0051134 1.0049364 1.0080004 1.0002082
vap_nhpi vap_other vap_two pre_16_rep_tru pre_16_dem_cli pre_20_rep_tru
1.0034503 1.0010427 0.9999715 1.0154680 1.0074157 1.0092683
pre_20_dem_bid uss_16_rep_sco uss_16_dem_dix uss_20_rep_gra uss_20_dem_har gov_18_dem_smi
1.0037990 1.0051767 1.0055020 1.0078760 1.0057119 1.0081538
gov_18_rep_mcm atg_18_dem_ana atg_18_rep_wil sos_18_dem_whi sos_18_rep_ham adv_16
1.0137191 1.0108854 1.0127145 1.0109889 1.0141728 1.0069088
adv_18 adv_20 arv_16 arv_18 arv_20 county_splits
1.0104412 1.0051354 1.0107766 1.0138879 1.0088795 1.0093744
muni_splits ndv nrv ndshare e_dvs pr_dem
1.0178542 1.0098929 1.0118179 1.0123676 1.0121183 1.0020453
e_dem pbias egap
1.0109351 1.0076293 1.0278558
Sampling diagnostics for SMC run 1 of 2 (3,000 samples)
Eff. samples (%) Acc. rate Log wgt. sd Max. unique Est. k
Split 1 1,569 (52.3%) 11.1% 0.32 1,882 ( 99%) 9
Split 2 1,998 (66.6%) 16.2% 0.90 1,490 ( 79%) 6
Split 3 2,064 (68.8%) 20.5% 0.70 1,636 ( 86%) 4
Split 4 2,094 (69.8%) 23.6% 0.70 1,726 ( 91%) 3
Split 5 1,953 (65.1%) 24.3% 0.71 1,679 ( 89%) 2
Split 6 1,678 (55.9%) 8.4% 0.81 1,436 ( 76%) 2
Resample 358 (11.9%) NA% 1.54 1,136 ( 60%) NA
Sampling diagnostics for SMC run 2 of 2 (3,000 samples)
Eff. samples (%) Acc. rate Log wgt. sd Max. unique Est. k
Split 1 1,595 (53.2%) 14.4% 0.33 1,902 (100%) 7
Split 2 1,971 (65.7%) 20.0% 0.88 1,513 ( 80%) 5
Split 3 2,024 (67.5%) 21.2% 0.72 1,631 ( 86%) 4
Split 4 2,093 (69.8%) 23.3% 0.71 1,681 ( 89%) 3
Split 5 2,107 (70.2%) 18.2% 0.71 1,681 ( 89%) 3
Split 6 1,654 (55.1%) 8.8% 0.74 1,473 ( 78%) 2
Resample 235 (7.8%) NA% 1.48 1,065 ( 56%) NA
TODO
lines from the template code have been removedenforce_style()
to format my coderedist_map
and redist_plans
objects, and summary statistics) have been edited@kuriwaki
Redistricting requirements
In South Carolina, districts must:
Interpretation of requirements
We enforce a maximum population deviation of 0.5%.
Data Sources
Data for South Carolina comes from the ALARM Project's 2020 Redistricting Data Files.
Pre-processing Notes
No manual pre-processing decisions were necessary.
Simulation Notes
We sample 6,000 districting plans for South Carolina. We impose a hinge constraint on the Black Voting Age Population so that it encourages district BVAP between 40 percent and 60 percent, but districts with BVAP of 30% or less are not penalized as much. We impose a municipality-split constraint to lower the number of municipality splits.
Validation
Checklist
TODO
lines from the template code have been removedenforce_style()
to format my coderedist_map
andredist_plans
objects, and summary statistics) have been edited@CoryMcCartan @christopherkenny