search

alarm-redist / fifty-states

Redistricting analysis for all 50 U.S. states
https://alarm-redist.github.io/fifty-states/
Other
9 stars 7 forks source link

Re-run 2020 West Virginia Congressional Districts #104

Closed christopherkenny closed 2 years ago

christopherkenny commented 2 years ago

Redistricting requirements

In West Virginia, under the state constitution, districts must:

  1. be contiguous (I 1-4)
  2. have equal populations (I 1-4)
  3. be geographically compact (I 1-4)
  4. districts must be made of contiguous counties (I 1-4)

Interpretation of requirements

We enforce a maximum population deviation of 0.5%. We simulate at the county level.

Data Sources

Data for West Virginia comes from the The Upshot Presidential Precinct Map data and are joined with county level Census data.

Pre-processing Notes

Data is aggregated to the county level.

Simulation Notes

We sample 5,000 districting plans for West Virginia, across 4 independent runs of the SMC algorithm. No special techniques were needed to produce the sample.

Validation

validation_20220621_2356

SMC: 5,000 sampled plans of 2 districts on 55 units
`adapt_k_thresh`=0.985 • `seq_alpha`=0.5
`est_label_mult`=1 • `pop_temper`=0

Plan diversity 80% range: 0.097 to 0.434
✖ WARNING: Low plan diversity

R-hat values for summary statistics:
   pop_overlap      total_vap       plan_dev      comp_edge    comp_polsby      pop_white      pop_black       pop_hisp       pop_aian      pop_asian 
     1.0020773      0.9998947      0.9999536      1.0003087      1.0007092      1.0002150      1.0005952      1.0005577      1.0014897      1.0006485 
      pop_nhpi      pop_other        pop_two      vap_white      vap_black       vap_hisp       vap_aian      vap_asian       vap_nhpi      vap_other 
     1.0001644      1.0010625      1.0022005      1.0001193      1.0005887      1.0000335      1.0006007      1.0005678      1.0001469      1.0008277 
       vap_two pre_20_rep_tru pre_20_dem_bid         adv_20         arv_20            ndv            nrv        ndshare          e_dvs           egap 
     1.0016263      1.0009999      1.0011749      1.0011749      1.0009999      1.0011749      1.0009999      1.0006942      1.0006942      1.0014646 

Sampling diagnostics for SMC run 1 of 4 (1,250 samples)
         Eff. samples (%) Acc. rate Log wgt. sd  Max. unique Est. k 
Split 1     1,239 (99.1%)      2.8%         0.2   784 ( 99%)      4 
Resample    1,211 (96.8%)       NA%         0.2   778 ( 98%)     NA 

Sampling diagnostics for SMC run 2 of 4 (1,250 samples)
         Eff. samples (%) Acc. rate Log wgt. sd  Max. unique Est. k 
Split 1     1,239 (99.1%)      2.7%         0.2   783 ( 99%)      4 
Resample    1,209 (96.8%)       NA%         0.2   777 ( 98%)     NA 

Sampling diagnostics for SMC run 3 of 4 (1,250 samples)
         Eff. samples (%) Acc. rate Log wgt. sd  Max. unique Est. k 
Split 1     1,239 (99.1%)      2.7%         0.2   770 ( 97%)      4 
Resample    1,211 (96.9%)       NA%         0.2   783 ( 99%)     NA 

Sampling diagnostics for SMC run 4 of 4 (1,250 samples)
         Eff. samples (%) Acc. rate Log wgt. sd  Max. unique Est. k 
Split 1     1,239 (99.1%)      2.9%         0.2   806 (102%)      4 
Resample    1,209 (96.7%)       NA%         0.2   789 (100%)     NA 

•  Watch out for low effective samples, very low acceptance rates (less than 1%), large std. devs. of the log weights (more than 3 or so), and low numbers
of unique plans. R-hat values for summary statistics should be between 1 and 1.05.
• Low diversity: Check for potential bottlenecks. Increase the number of samples. Examine the diversity plot with `hist(plans_diversity(plans),
breaks=24)`. Consider weakening or removing constraints, or increasing the population tolerance. If the accpetance rate drops quickly in the final splits,
try increasing `pop_temper` by 0.01.

Checklist

@CoryMcCartan

Additional Notes:

Appears to me that the diversity is actually fine. This is a 55 county into 2 districts problem.