Note: The original post analyzed the congressional plan proposed in Senate Bill 756; we have updated the analysis to include the plan proposed in Senate Bill 757.
We have completed an initial analysis of the redistricting plans recently proposed by the NC Legislature on Wednesday, October 18. The legislature has proposed changing the maps used to elect the North Carolina congressional delegation to the US House of Representatives and the maps used to elect the NC House and the NC Senate members of the NC Legislature.
We use the same methodology as described in our report here. The nonpartisan ensembles of maps used in the below analysis are the primary ensembles previously presented to the court in Harper v Hall/Moore in 2021.
A first indication of gerrymandering in the new Congressional maps comes from comparing it with the remedial map from 2022. We do this by using the votes from the 2022 US Senate race which has a statewide partisan vote share of 48.35% preferring the Democrats. Under these votes, both proposed maps (from SB756 and SB757) would elect 11 Republicans and 3 Democrats to the US House of Representatives while the remedial map it replaces would have elected 7 Republicans and 7 Democrats.
One of the basic principles of a functioning democracy is that when the electorate changes its political opinion this change should be reflected in the elections outcome. The proposed congressional maps are both highly non-responsive to changes in the opinion of the electorate. We find that over a number of recent historical elections and other possible political environments, the newly proposed congressional map from SB756 reliably elected 11 Republicans and 3 Democrats despite the elections having wildly different votes. Similarly, the map proposed in SB757 reliably elected between 10 and 11 Republicans and between 3 and 4 Democrats. Furthermore, even under dramatic voting swings, the newly proposed congressional plans consistently elect more Republicans than would be expected from a non-partisan process.
To see this, we examine voting patterns between 45% and 55% Democratic statewide vote shares based on the 2022 US Senate election and the 2020 Presidential election and reproduce the movies described here. We see that the newly proposed maps are significantly less responsive to the changing will of the people as expressed in their votes than the 2022 remedial map they would replace. Over all of the historical elections considered in the plot below, the proposed map from SB756 always elects 11 Republicans and 3 Democrats and the proposed map from SB757 elects 10 or 11 Republicans and 3 or 4 Democrats. In contrast, the remedial map that is likely to be replaced and which was used in the 2022 elections varies from electing 9 Republicans and 5 Democrats to electing 7 Republicans and 7 Democrats.
Furthermore, we observe that the newly proposed plans are even less responsive than the originally overturned plan from 2021 and lean more Republican than either the overturned plan drawn by the legislature in 2021 and the remedial plan that will likely be replaced from 2022. The collection of histograms over historic elections displays the rigidity of both plans. It also displays how the plan from SB756 leads to an extra Republican seat when comparing to the overturned 2021 map. The difference between the overturned 2021 map and the plan proposed in SB757 is not evident in the collection of histograms, however, the difference can be observed in both the videos and also in the rank-ordered districts that we present at the end of this post and which are explained in this past post. The key take away is that there are three Democrats residing in packed districts that will reliably vote Democratic, whereas the 4th most Democratic seat is even less reliably Democratic in the newly proposed plan from SB757 than it was under the overturned 2021 plan.
The two videos below show how the remedial map that was used in the 2022 election responds to elections swinging from preferring Republicans to elections preferring Democrats while the newly proposed maps fail to change the seats elected as the election votes swing from 45% to 53% Democratic vote share.
In addition to the analysis in the Congressional maps, we also present an analysis of the state house and state senate maps. We reproduce the animated effect using the Presidential 2020 results. In both chambers, we find that the proposed plans are even more extreme than the originally enacted 2021 maps. Additionally, we find that new proposed maps are less responsive to changes in the votes than the remedial maps used in the 2022 elections.
Both the Senate and House maps under-elect Democrats as one moves to more balanced elections with Republican statewide vote fractions near 50%. This has important implications for the preservation of the super-majority in the chamber. Under the newly proposed Senate maps, the Republicans may reasonably expect to obtain a super majority, even when the statewide Democratic vote share is over 50%.
The figures below show the number of Democrats elected to the NC Senate and House using a number of different historical elections which have a wide range of state-wide Democratic vote share. As one moves into the elections the ensemble shows that the Republicans would typically fail to obtain a super majority, the proposed map significantly under-elects Democrats.
In the Senate, using the AG16, the AG20, the GV16, the USS20, the CL20, and the PR20 elections, Democrats would typically break the supermajority using the maps in the ensemble or using the remedial map from 2022. Yet the newly proposed map would preserve the Republican super-majority. In these elections, the statewide Democratic vote share ranges from 48.9% to 50.2%.
In the House, using CL20 and USS20 election data, the Democrats would typically break the supermajority using the maps in the ensemble or using the remedial map from 2022. Yet the newly proposed map preserves the super-majority. In the more democratic-leaning elections the ensemble and the remedial map from 2022 would typically give control of the chamber to the Democrats but the newly proposed map leaves the Republicans with a sizable majority.
As stated above, we conclude by returning to the congressional maps and compare the 4th most Democratic seat between the plan from SB757 and the overturned plan from 2021. We observe that under both the Presidential 2020 votes and the US Senate 2022 votes, the 4th most Democratic seat is more Republican under the plan from SB757 than it is under the overturned 2021 plan. We have also repeated the analysis for the votes under the 2020 Attorney General election and the 2016 Presidential election and find the same conclusion: Under the 2020 Attorney General election, the fourth most Democratic seat would yield a 53.6% of the vote going to the Democratic candidate under the overturned results, while giving a narrower margin of 52.8% to the Democratic candidate under the plan proposed in SB757. Similarly, under the 2016 Presidential election, the fourth most Democratic seat would yield a 53.0% of the vote going to the Democratic candidate under the overturned results, while giving only a narrower margin of 52.4% to the Democratic candidate under the plan proposed in SB757.
Jonathan Mattingly
Greg Herschlag
The data used in this analysis can be found here. In addition to this, we have imputed the 2022 voting data down to the census block level and report those numbers here.
