SHAUNA WILLIAMS v. REPRESENTATIVE DESTIN HALL (2025)

MEMORANDUM OPINION AND ORDER

TABLE OF CONTENTS

Page

I. PROCEDURAL BACKGROUND 1

II. STANDING 4

III. INTENTIONAL VOTE DILUTION 7

A. Legal Standards 7

1. Fourteenth and Fifteenth Amendments 8

2. Section 2 of the VRA 13

B. Findings of Fact 15

1. Historical Background 16

2. The 2023 Redistricting Process 22

a. Before Drafting 22

b. Senate and Congressional Criteria 24

c. Senate Plan 30

i. SD 1 and 2 31

ii. SD 8 33

d. Congressional Plan 34

i. CD 5, 6, and 10 36

ii. CD 12 and 14 37

e. Senate Bills 757 and 758 37

3. Prior Redistricting Processes 41

a. Timing and Transparency 41

b. Opportunity for Public Input 46

c. Consideration of Race and the VRA 47

4. Statistical Analysis of the 2022 and 2023 Congressional Maps 52

a. Plaintiffs’ Expert Dr. Rodden 52

i. Racial Sorting 54

ii. Implications 69

b. Critiques from Defendants’ Experts 76

c. Summary 88

5. Impact on Black Voters 89

6. Summary 94

C. Conclusions of Law 96

IV. RACIAL GERRYMANDERING 96

A. Legal Standard 96

B. Findings of Fact 99

C. Conclusions of Law 106

V. DISCRIMINATORY RESULTS 106

A. Legal Standard 106

B. Findings of Fact 112

1. First Gingles Precondition 112

2. Second Gingles Precondition 121

3. Third Gingles Precondition 122

4. Totality of the Circumstances 126

a. Extent of Racial and Political Polarization 127

b. Electoral Success of Black Candidates 138

c. Racial Appeals in Political Campaigns 142

d. History of Official Discrimination 152

e. Current Voting Practices 156

f. Effects of Discrimination that Hinder Political Participation 159

g. Responsiveness to Particularized Needs 167

h. Policy Underlying the State's Practice 170

5. Summary 172

C. Conclusions of Law 174

VI. CONCLUSION 175

In 2023, the North Carolina General Assembly redrew the State's districts for electing representatives to the state Senate and federal Congress. Plaintiffs seek to enjoin use of those district maps, claiming certain portions violate the Fourteenth and Fifteenth Amendments to the United States Constitution and Section 2 of the Voting Rights Act of 1965 (“VRA”), codified at 52 U.S.C. § 10301. This three-judge panel was convened to hear these consolidated cases under 28 U.S.C. § 2284. After conducting a six-day trial, we conclude that the General Assembly did not violate the Constitution or the VRA in its 2023 redistricting.

I. PROCEDURAL BACKGROUND

Plaintiffs in these consolidated cases sued in December 2023, claiming that North Carolina's new federal and state legislative districts were unlawful. Eighteen black and Hispanic individuals sued in case number 1:23-cv-1057 (“Williams Plaintiffs”). The North Carolina State Conference of the NAACP, Common Cause, and six black voters sued in case number 1:23-cv-1104 (“NAACP Plaintiffs”).¹ Plaintiffs initially presented broad challenges to the State's districting plans. (See Doc. 98 at 2–3.)² Over the course of litigation, however, they have amended their complaints to narrow the challenged areas. (See Docs. 104, 107.) This Court also granted summary judgment to Defendants on NAACP Plaintiffs’ malapportionment claims and on other claims in certain districts for lack of standing. (See Doc. 98.)

While these claims were pending, in October 2025, the General Assembly redrew two of North Carolina's fourteen congressional districts (“CD”): CD 1 and 3 in the eastern part of the State. See 2025 N.C. Sess. Laws 95. We granted Plaintiffs’ requests to file supplemental complaints challenging the 2025 revisions. (Doc. 179; see Docs. 180, 181.) Those supplemental claims remain pending before this Court. The present opinion resolves only Plaintiffs’ challenges to the General Assembly's 2023 redistricting. Importantly, however, Plaintiffs’ challenges to CD 1 as it was configured in 2023 have been superseded by the General Assembly's 2025 enactment. Therefore, this opinion does not address the lawfulness of CD 1 as it was previously configured in 2023.

Remaining before us as regards the 2023 redistricting are four sets of challenges to the state Senate and federal congressional plans. First, Plaintiffs contend that North Carolina intentionally diluted the voting strength of black voters in violation of the Fourteenth and Fifteenth Amendments.³ (Doc. 108 at Count I; Doc. 105 at Counts 5 and 9.) Plaintiffs allege this occurred in CD 5, 6, and 10 in the Piedmont Triad and in CD 14 near Charlotte. NAACP Plaintiffs also challenge Senate districts (“SD”) 1 and 2 in the northeast and SD 8 near Wilmington.⁴ Second, Plaintiffs claim that intentional vote dilution in the same areas, along with CD 12 near Charlotte, violated Section 2 of the VRA. (Doc. 108 at Count II; Doc. 105 at Counts 4 and 8.) Third, NAACP Plaintiffs argue the State racially gerrymandered a portion of SD 8 in violation of the Fourteenth Amendment. (Doc. 105 at Count 2.) Fourth, NAACP Plaintiffs contend SD 1 and 2 violate Section 2 of the VRA by creating discriminatory results, separate and apart from any discriminatory intent. (Doc. 105 at Count 1.)

This Court held a six-day bench trial in late June and early July 2025. At that trial, we heard from fourteen fact witnesses and twelve experts. We received testimony from two additional fact witnesses through transcript designations from one witness's deposition and the other's trial testimony in a related case. The parties submitted hundreds of exhibits into evidence, including forty expert reports and errata. After trial, we received 750 pages of posttrial briefing. Having thoroughly considered all these materials, we make the following findings of fact and conclusions of law.⁵

II. STANDING

We begin by shoring up a jurisdictional matter that we deferred at summary judgment, namely, whether Plaintiffs have standing to sue for each of their claims. “Article III of the Constitution restricts the power of federal courts to ‘Cases’ and ‘Controversies.’ ” Chafin v. Chafin, 568 U.S. 165, 171 (2013); see U.S. Const. art. III, § 2. A case or controversy exists “only when at least one plaintiff establishes that [it] has standing to sue.” Murthy v. Missouri, 144 S. Ct. 1972, 1985 (2024) (internal quotation marks and brackets omitted). Three elements form the “irreducible constitutional minimum of standing”: (1) a plaintiff must have “suffered an injury in fact” that is “concrete and particularized” and “actual or imminent”; (2) the injury must be “fairly traceable to the challenged action of the defendant”; and (3) the injury must be “likely” to be “redressed by a favorable decision.” Lujan v. Defs. of Wildlife, 504 U.S. 555, 560–561 (1992) (internal quotation marks, brackets, and ellipses omitted).

Plaintiffs in this case include individuals suing on their own behalf and two organizations suing in their members’ stead. Where, as here, an association sues on behalf of its members, it “must demonstrate that ‘(a) its members would otherwise have standing to sue in their own right; (b) the interests it seeks to protect are germane to the organization's purpose; and (c) neither the claim asserted nor the relief requested requires the participation of individual members in the lawsuit.’ ” Students for Fair Admissions, Inc. v. President & Fellows of Harvard Coll., 143 S. Ct. 2141, 2157 (2023) (quoting Hunt v. Wash. State Apple Advert. Comm'n, 432 U.S. 333, 343 (1977)).

Throughout this suit, Defendants have not challenged the NAACP's or Common Cause's ability to satisfy the second and third prongs of this standard. And Plaintiffs produced sufficient evidence at trial to meet these requirements. See Lujan, 504 U.S. at 561 (noting standing is “an indispensable part of the plaintiff's case, [and] each element must be supported in the same way as any other matter on which the plaintiff bears the burden of proof ․ at the successive stages of the litigation”). Deborah Maxwell, the President of the North Carolina NAACP, and Robert Phillips, the Executive Director of Common Cause North Carolina, both testified that their organizations exist to further causes germane to this action, including promoting minority voter registration and participation in the State's political processes. See Trial Tr. Vol. I at 11:5–12:1; Trial Tr. Vol. III at 603:11–604:22.⁶

That leaves the first prong. As we have noted throughout this litigation, each of the Plaintiffs’ alleged injuries arises from the boundaries and composition of the particular district in which the voter resides. (See Docs. 75, 98); Gill v. Whitford, 138 S. Ct. 1916, 1930 (2018) (“To the extent the plaintiffs’ alleged harm is the dilution of their votes, that injury is district specific ․ A plaintiff who complains of gerrymandering, but who does not live in a gerrymandered district, ‘assert[s] only a generalized grievance against governmental conduct of which he or she does not approve.’ ” (quoting United States v. Hays, 515 U.S. 737, 745 (1995))). Thus, to establish standing, Plaintiffs had to identify at least one allegedly injured individual Plaintiff or member of the NAACP or Common Cause who resides in each challenged district.

At summary judgment, we ruled that Plaintiffs had done so for each district before us now, save SD 8. As to that district, the parties alerted us that Hollis Briggs, the named Plaintiff residing there, had recently passed away. (See Doc. 98 at 12 n.2 (citing Doc. 96).) We deferred judgment on the matter but are now satisfied that Plaintiffs have proven standing for all remaining counts as to all districts not previously dismissed. At trial, Plaintiffs produced individual voting records, sworn declarations of residency, and testimony from Maxwell and Phillips confirming the membership of certain voters associated with the NAACP and Common Cause. See NAACPPX164–166; WX55–72; Trial Tr. Vol. I at 13:1–15:17; Trial Tr. Vol. III at 607:16–610:10. This evidence established that individual Plaintiffs reside in CD 6, 10, 12, and 14, and SD 1 and 2; that a standing member of Common Cause lives in CD 5; and that a standing member of the NAACP lives in SD 8. Because Plaintiffs have satisfied the requirements of Article III, we turn now to the merits of their claims.

III. INTENTIONAL VOTE DILUTION

A. Legal Standards

“Redistricting constitutes a traditional domain of state legislative authority,” and it is “an inescapably political enterprise.” Alexander v. S.C. State Conf. of the NAACP, 144 S. Ct. 1221, 1233 (2024). Certain constitutional and statutory provisions, however, impose constraints on that authority. Plaintiffs claim the North Carolina General Assembly violated the Fourteenth and Fifteenth Amendments and Section 2 of the VRA by intentionally discriminating against black voters in drawing state Senate and federal congressional districts. Although the constitutional and statutory claims differ in some respects, both require proof that the General Assembly acted with a discriminatory purpose. Because we find the absence of discriminatory purpose dispositive for both claims, we address them together.

1. Fourteenth and Fifteenth Amendments

“The Fourteenth Amendment was designed to eradicate race-based state action.” Alexander, 144 S. Ct. at 1236 (citing Students for Fair Admissions, 143 S. Ct. at 2141). The Equal Protection Clause is therefore implicated when, as here, a State is accused of “enact[ing] a particular voting scheme as a purposeful device ‘to minimize or cancel out the voting potential of racial or ethnic minorities.’ ” Miller v. Johnson, 515 U.S. 900, 911 (1995) (quoting Mobile v. Bolden, 446 U.S. 55, 66 (1980)). The Supreme Court has declined to say whether intentional vote dilution likewise violates the Fifteenth Amendment. See Reno v. Bossier Par. Sch. Bd., 528 U.S. 320, 334 n.3 (2000); Voinovich v. Quilter, 507 U.S. 146, 159 (1993). The Court of Appeals for the Fourth Circuit, however, has held that “[c]laims of racially discriminatory vote dilution exist under both the fifteenth amendment and the Equal Protection Clause of the fourteenth amendment” and has treated the claims as “essentially congruent.” Washington v. Finlay, 664 F.2d 913, 919 (4th Cir. 1981).

To succeed on a claim under either amendment, a plaintiff must show that the State acted with a “ ‘discriminatory purpose.’ ” Rogers v. Lodge, 458 U.S. 613, 617 (1982) (quoting Washington v. Davis, 426 U.S. 229, 240 (1976)); see also Bolden, 446 U.S. at 62–63; Gomillion v. Lightfoot, 364 U.S. 339, 342 (1960). A showing of disparate impact alone will not suffice. Rather, a challenger bringing a constitutional vote dilution claim “must show that the State's districting plan ‘has the purpose and effect’ of diluting the minority vote.” Alexander, 144 S. Ct. at 1252 (quoting Shaw v. Reno, 509 U.S. 630, 649 (1993) (Shaw I)).

In evaluating the State's intent, a court must begin with a “presumption of legislative good faith.” Abbott v. Perez, 138 S. Ct. 2305, 2324 (2018); see Miller, 515 U.S. at 915. As the Supreme Court has explained, the presumption of good faith furthers several constitutional interests. See Alexander, 144 S. Ct. at 1236. First, the presumption “reflects the Federal Judiciary's due respect for the judgment of state legislators, who are similarly bound by an oath to follow the Constitution.” Id. Second, “when a federal court finds that race drove a legislature's districting decisions, it is declaring that the legislature engaged in ‘offensive and demeaning’ conduct, that ‘bears an uncomfortable resemblance to political apartheid.’ ” Id. (first quoting Miller, 515 U.S. at 912; then quoting Shaw I, 509 U.S. at 647). The presumption of good faith prevents courts from being “quick to hurl such accusations at the political branches.” Id. And third, the presumption acknowledges that we “must be wary of plaintiffs who seek to transform federal courts into ‘weapons of political warfare’ that will deliver victories that eluded them ‘in the political arena.’ ” Id. (quoting Cooper v. Harris, 581 U.S. 285, 335 (2017) (Alito, J., concurring)).

Practically, “[t]his presumption of legislative good faith directs district courts to draw the inference that cuts in the legislature's favor when confronted with evidence that could plausibly support multiple conclusions.” Alexander, 144 S. Ct. at 1235–1236. This directive is particularly relevant “when race and partisan preference are highly correlated.” Id. at 1233. “When partisanship and race correlate, it naturally follows that a map that has been gerrymandered to achieve a partisan end can look very similar to a racially [motivated] map.” Id. at 1235. But “[a]s far as the Federal Constitution is concerned, a legislature may pursue partisan ends when it engages in redistricting.” Id. at 1233; see Rucho v. Common Cause, 139 S. Ct. 2484, 2497 (2019). Given the “complex interplay of forces that enter a legislature's redistricting calculus,” we must “exercise extraordinary caution in adjudicating claims that a State has drawn district lines on the basis of race.” Miller, 515 U.S. at 915–916. And we must be careful of attempts to “infer a racial motive” from “the racial effects of a political gerrymander in a jurisdiction in which race and partisan preference are very closely correlated.” Alexander, 144 S. Ct. at 1241–1242.

The plaintiff bears the “burden to overcome the presumption of legislative good faith” and show that the legislature “acted with invidious intent.” Abbott, 138 S. Ct. at 2325. The intent inquiry proceeds in two steps. First, the plaintiff must demonstrate that racial discrimination—here, the discriminatory purpose to dilute the voting potential of a racial minority—was a “ ‘substantial’ or ‘motivating’ factor behind enactment of the law.” Hunter v. Underwood, 471 U.S. 222, 228 (1985). The plaintiff need not prove that “the challenged action rested solely on racially discriminatory purposes.” Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252, 265 (1977). And direct evidence of discriminatory intent is not required. See Rogers, 458 U.S. at 618. Instead, it may “be inferred from the totality of the relevant facts.” Davis, 426 U.S. at 242.

Of course, “[p]roving the motivation behind official action is often a problematic undertaking.” Hunter, 471 U.S. at 228. “Determining whether invidious discriminatory purpose was a motivating factor demands a sensitive inquiry into such circumstantial and direct evidence of intent as may be available.” Arlington Heights, 429 U.S. at 266. Courts often consider the four evidentiary sources the Supreme Court identified in its Arlington Heights decision: “(1) historical background; (2) the specific sequence of events leading to the law's enactment, including any departures from the normal legislative process; (3) the law's legislative history; and (4) whether the law ‘bears more heavily on one race than another.’ ” N.C. State Conf. of the NAACP v. Raymond, 981 F.3d 295, 303 (4th Cir. 2020) (quoting Arlington Heights, 429 U.S. at 265–269); see also Brnovich v. Democratic Nat'l Comm., 141 S. Ct. 2321, 2349 (2021).

If the plaintiff proves discriminatory intent was a motivating factor, then at the second step of the inquiry “the burden shifts to the law's defenders to demonstrate that the law would have been enacted without” racial discrimination. Hunter, 471 U.S. at 228. Only at this point, after “there is a proof that a discriminatory purpose has been a motivating factor in the decision,” is judicial deference to the legislature “no longer justified.” Arlington Heights, 429 U.S. at 265–266. At this step, a court “must ‘scrutinize the legislature's actual nonracial motivations to determine whether they alone can justify the legislature's choices.’ ” Raymond, 981 F.3d at 303 (quoting N.C. State Conf. of NAACP v. McCrory, 831 F.3d 204, 221 (4th Cir. 2016)). If the State cannot make such a showing, then discriminatory purpose has been established.

An intentional vote dilution claim also requires proof that “the State's districting plan ‘has the ․ effect’ of diluting the minority vote.” Alexander, 144 S. Ct. at 1252 (quoting Shaw I, 509 U.S. at 649). To prove a dilutive effect, a plaintiff must “produce evidence to support findings that the political processes leading to nomination and election were not equally open to participation by the group in question—that its members had less opportunity than did other residents in the district to participate in the political processes and to elect legislators of their choice.” White v. Regester, 412 U.S. 755, 766 (1973); see, e.g., Rogers, 458 U.S. at 622–627. “[I]t is not enough that the racial group allegedly discriminated against has not had legislative seats in proportion to its voting potential.” White, 412 U.S. at 765–766.

2. Section 2 of the VRA

Congress enacted the Voting Rights Act of 1965 “in an effort to achieve at long last what the Fifteenth Amendment had sought to bring about 95 years earlier: an end to the denial of the right to vote based on race.” Brnovich, 141 S. Ct. at 2330. In its original form, Section 2 of the VRA closely tracked the language of the Fifteenth Amendment and consequently “had little independent force.” Allen v. Milligan, 143 S. Ct. 1487, 1499 (2023). The Supreme Court interpreted Section 2, like the Fifteenth Amendment, as “prohibit[ing] States from acting with ‘a racially discriminatory motivation’ or an ‘invidious purpose’ to discriminate,” rather than “prohibit[ing] laws that are discriminatory only in effect.” Id. (quoting Bolden, 446 U.S. at 61–65 (plurality opinion)). In response, Congress in 1982 amended Section 2 to prohibit voting practices “which result[ ] in a denial or abridgment” of a citizen's right to vote on account of race. 52 U.S.C. § 10301(a). After this amendment, discriminatory intent is no longer required for liability under Section 2. See Milligan, 143 S. Ct. at 1514; Shaw I, 509 U.S. at 641.

Nevertheless, a claim of intentional vote dilution may still proceed under Section 2 of the VRA. See Chisom v. Roemer, 501 U.S. 380, 394 n.21 (1991). Because a Section 2 vote dilution claim can be proven without a showing of intentional discrimination, “few voters have asserted intentional vote dilution claims since the 1982 amendments” to the VRA. Harding v. Cnty. of Dallas, 948 F.3d 302, 313 n.47 (5th Cir. 2020) (internal quotation marks omitted). Consequently, “[t]he role that § 2 ․ play[s] in intentional vote dilution claims as opposed to results-only claims is somewhat unsettled,” especially as it concerns “the level of proof of dilutive effects required in a § 2 intentional vote dilution claim.” Perez v. Abbott, 253 F. Supp. 3d 864, 942–943 (W.D. Tex. 2017); see also Backus v. South Carolina, 857 F. Supp. 2d 553, 567 (D.S.C. 2012), sum. aff'd, 568 U.S. 801 (2012); Cano v. Davis, 211 F. Supp. 2d 1208, 1249 (C.D. Cal. 2002), sum. aff'd, 537 U.S. 1100 (2003). For example, in Bartlett v. Strickland, the Supreme Court held that a legislature's failure to create or preserve crossover districts does not give rise to a cognizable Section 2 effects claim, but the Court left open whether proof that the legislature engaged in intentional discrimination when it drew lines to prevent the creation or preservation of crossover districts could form the basis of a Section 2 claim. 556 U.S. 1, 19–20 (2009).

Plaintiffs here have asserted claims of intentional vote dilution under Section 2. Because these claims, like their constitutional claims, require proof that the General Assembly acted with a discriminatory purpose, and ultimately fail due to the absence of such evidence, we need not decide whether, or how, intentional discrimination affects the dilutive effect analysis for a Section 2 claim.

B. Findings of Fact

After considering all the evidence, we find that Plaintiffs have failed to prove that the North Carolina General Assembly drew state Senate or federal congressional districts with the discriminatory purpose of minimizing or canceling out the voting potential of black North Carolinians. The racial data necessary to execute such a discriminatory purpose was not used in drafting the 2023 Senate or congressional plans. The circumstances surrounding the plans’ enactment and the resulting district configurations and composition are consistent with the General Assembly's non-racial motivations, which included traditional districting criteria, North Carolina law, and partisan performance. The evidence established the following.

1. Historical Background

Every ten years, the North Carolina General Assembly—which consists of the North Carolina Senate and the North Carolina House of Representatives—redraws the State's legislative and congressional districts based on the latest decennial census. See N.C. Const. art. II, §§ 3, 5. We begin by describing the redistricting cycle immediately preceding the one at issue here because it provides helpful context for understanding this case.

In the 2011 districting plans, the General Assembly followed a policy of creating majority-minority districts. Using racial data, the legislature created twenty-three majority-black state House districts and nine majority-black state Senate districts. Covington v. North Carolina, 316 F.R.D. 117, 134 (M.D.N.C. 2016), sum. aff'd, 581 U.S. 1015 (2017). A group of voters challenged twenty-eight of those districts, some of which covered parts of the State under scrutiny here. See id. at 128, 142, 147–148, 150–152, 156, 158–159, 163–164. The State defended the districts as an effort to comply with Section 2 of the VRA. A three-judge panel rejected the State's defense because the General Assembly did not have “a strong basis in evidence” for thinking that the VRA required majority-minority districts in those areas; in fact, evidence demonstrated that minority-preferred candidates “were already consistently winning” in the challenged areas without majority-minority districts. Id. at 167, 172–173. The panel ruled that the General Assembly's use of race to draw the districts violated the Equal Protection Clause. Id. at 124. But the court did not find that the legislature “acted in bad faith or with discriminatory intent.” Id. at 124 n.1.

Another lawsuit challenged two majority-minority congressional districts in the 2011 redistricting plan, which also included some counties at issue in this case. Harris v. McCrory, 159 F. Supp. 3d 600, 604 (M.D.N.C. 2016); see also Cooper, 581 U.S. at 323–326. A three-judge panel again held that the State's use of race to draw the districts violated the Equal Protection Clause. Harris, 159 F. Supp. 3d at 604. The panel rejected the State's defense that one of the districts—CD 1 in the northeast—was justified as an effort to comply with Section 2 of the VRA. To the contrary, the panel found “no evidence that the white majority votes sufficiently as a bloc to enable it usually to defeat the minority's preferred candidate” in CD 1. Id. at 624 (internal quotation marks and ellipsis omitted); see also id. at 606 (“For decades, African-Americans enjoyed tremendous success in electing their preferred candidates in former versions of CD 1 and CD 12 regardless of whether those districts contained a majority black voting age population.”). The Supreme Court agreed. Cooper, 581 U.S. at 298–299. As the Supreme Court explained, electoral history gave the State “no reason to think that the VRA required it” to create a majority-black district to avoid Section 2 liability. Id. at 303; see also id. at 304 (“North Carolina too far downplays the significance of a longtime pattern of white crossover voting in the area that would form the core of the redrawn District 1.”). The courts made no finding that the legislature acted in bad faith or with discriminatory intent. See Harris, 159 F. Supp. 3d at 604–605.

For the 2016 remedial congressional redistricting following Harris and the 2017 remedial legislative redistricting following Covington, the General Assembly implemented a policy forbidding consideration of race. JX315; JX321. Common Cause and other plaintiffs then sued, challenging the remedial congressional and legislative districts on a partisan gerrymandering theory. See Common Cause v. Rucho, 279 F. Supp. 3d 587, 605 (M.D.N.C. 2018); Common Cause v. Lewis, No. 18 CVS 014001, 2019 WL 4569584, at *1 (N.C. Super. Ct. Sept. 3, 2019). In Rucho, the United States Supreme Court held that “partisan gerrymandering claims present political questions beyond the reach of the federal courts,” and so dismissed the lawsuit challenging the State's congressional districts. Rucho, 139 S. Ct. at 2506–2507. The state court in Lewis, however, subsequently declared the partisan gerrymandering theory justiciable under the North Carolina Constitution, 2019 WL 4569584, at *124–129, and ordered the General Assembly to revise its state House and Senate districting maps without using “partisan considerations” or “election results data,” id. at *133, *135–137. The General Assembly then enacted remedial maps that were approved by the state court. See Common Cause v. Lewis, No. 18 CVS 014001, 2019 WL 13198027, at *1, 12 (N.C. Super. Ct. Oct. 28, 2019).

Following the 2020 census, North Carolina gained a federal congressional district due to a population increase. (See Doc. 148 at 9.) The North Carolina General Assembly began its redistricting process shortly after the release of updated data from the United States Census Bureau in August 2021. The Joint Redistricting Committees again implemented a policy forbidding the use of racial data in the construction or consideration of districts. JX224. The General Assembly passed its redistricting plans for congressional, state House, and state Senate districts in November 2021. (Doc. 148 at 11.)

A group of plaintiffs then challenged all three plans as partisan gerrymanders. See N.C. League of Conservation Voters, Inc. v. Hall, No. 21 CVS 015426, 2022 WL 124616, at *1–3 (N.C. Super. Ct. Jan. 11, 2022). The case made its way to the North Carolina Supreme Court, which struck down the 2021 plans, concluding they were unlawful partisan gerrymanders in violation of the North Carolina Constitution. Harper v. Hall, 868 S.E.2d 499, 559 (N.C. 2022) (Harper I). The court allowed the General Assembly to propose remedial plans but imposed metrics of partisan fairness that those plans had to satisfy to pass scrutiny. See, e.g., id. at 548–549, 559; (Doc. 148 at 11); Trial Tr. Vol. I at 109:14–110:13.

On remand from Harper I, the trial court approved the General Assembly's remedial state House and Senate plans but rejected its remedial congressional plan. See Harper v. Hall, 881 S.E.2d 156, 162 (N.C. 2022) (Harper II) (explaining the trial court's ruling). Court-appointed special masters subsequently drew a new remedial congressional map that satisfied strict partisan fairness metrics. See id. at 169. North Carolina conducted its 2022 congressional elections using that interim court-ordered map, while the state legislative elections were conducted using the remedial maps passed by the General Assembly. (Doc. 148 at 12.)

In the meantime, an appeal of the trial court's decision proceeded to the North Carolina Supreme Court. See Harper II, 881 S.E.2d 156. The court affirmed the trial court's holding that the General Assembly's congressional plan was an unconstitutional partisan gerrymander and affirmed the interim court-ordered congressional plan. Id. at 181. The Supreme Court also affirmed the trial court's holding that the state House plan was lawful but reversed the trial court's decision that the state Senate plan was constitutional, finding that it too was a partisan gerrymander. Id. The court therefore remanded the case for the trial court to “oversee the creation of” a modified state Senate plan. Id.

Before the case returned to the trial court, the General Assembly petitioned for rehearing, and in April 2023, the North Carolina Supreme Court reversed course. Granting the General Assembly's petition, the court withdrew Harper II, overruled Harper I, and held that partisan gerrymandering claims are “nonjusticiable, political questions under the North Carolina Constitution,” just as they are under the Federal Constitution. Harper v. Hall, 886 S.E.2d 393, 416 (N.C. 2023) (Harper III). Turning to the remedy, the court declined to reinstate the original 2021 plans, ruling instead that “[t]he General Assembly shall have the opportunity to enact a new set of legislative and congressional redistricting plans.” Id. at 446–448.

To the extent Plaintiffs appeal to this history for evidence of “official actions taken for invidious purposes,” we find their argument unavailing. Arlington Heights, 429 U.S. at 267. While courts found that the 2011 General Assembly configured districts predominantly based on race, those courts did not find that the legislature acted in bad faith or with discriminatory intent. See Covington, 316 F.R.D. at 129; Harris, 159 F. Supp. 3d at 604–605. And of course, “[a] legislature's past acts do not condemn the acts of a later legislature, which [courts] must presume acts in good faith.” Raymond, 981 F.3d at 298. That is especially true given the General Assembly's decision, after those rulings, to exclude race from its redistricting calculus.

2. The 2023 Redistricting Process

a. Before Drafting

After Harper III, the General Assembly began the process to construct a new set of maps. In August 2023, the North Carolina House Redistricting Committee, led by Representative Destin Hall, retained Blake Springhetti to draw its legislative districts and a congressional map, providing him with criteria to follow.⁷ (Doc. 148 at 2); see JX039. The North Carolina Senate also began drafting criteria for its redistricting plans around that time. (Doc. 148 at 2.) The Senate Redistricting and Elections Committee (“Senate Redistricting Committee” or “Committee”)—co-chaired by Senators Ralph Hise, Warren Daniel, and Paul Newton—opted not to retain a mapmaker. Instead, the chairs chose to draw the Senate and congressional maps themselves “with the technical support of partisan and non-partisan staff.” (Id.) Ultimately, the General Assembly chose to move forward with the Senate's version of the congressional map. That version did not include any input from Springhetti. Trial Tr. Vol. VI at 1251:1–2, 19–20.

At trial, Senator Hise testified in detail about the 2023 redistricting process and criteria as well as the drafting and configuration of the enacted maps. He was knowledgeable and credible; we credit his testimony. In addition to serving as a chairman of the Senate Redistricting Committee, Senator Hise also served as a co-chair of the General Assembly's Joint Elections and Redistricting Committee and described himself as “a primary drawer of both the Senate districts and Congressional districts.” Trial Tr. Vol. IV at 833:24–25. He joined the Senate Redistricting Committee in 2010 and also served as a Committee co-chair during the 2021 redistricting cycle. Id. at 870:2–22, 969:20–24. Since joining the Committee, he has participated in drawing dozens of draft district maps for North Carolina. Id. at 870:10–19.

According to Senator Hise, the Senate began drafting its maps in September or early October 2023. Id. at 834:9–14. He explained this timing, which Plaintiffs perceive as delayed. The North Carolina Supreme Court had empowered the legislature to redraw its maps in April, and the General Assembly aimed to complete the maps in time to adhere to candidate filing deadlines set by the North Carolina State Board of Elections. Id. at 834:16–835:4; see also (Doc. 162-2 at 135:20–137:2). Senator Hise explained that the lawmakers wanted to “finish predominantly” their biannual budget process before turning their focus to redistricting. Trial Tr. Vol. IV at 834:22–24; id. at 835:7–12; see also id. at 869:11–14 (explaining the “ideal situation” was “to have the budget behind us, done with that process and the entanglements that required, and be able to focus on the redistricting maps”). Accordingly, the General Assembly “set up the timeline to best accommodate those two goals.” Id. at 835:3–4; see id. at 915:21–25. The General Assembly passed an appropriations bill on September 22, 2023. See WX172.

On September 18, the Senate Redistricting Committee announced three public hearings to be held on redistricting. (Doc. 148 at 3.) The Committee settled on the following times and locations in coordination with Representative Hall: September 25 at the College of Albemarle in Elizabeth City; September 26 at Appalachian State's Hickory campus; and September 27 at the Legislative Office Building in Raleigh. (Id.) All meetings were set for 4:00 p.m. See JX066; JX068; JX070. Around September 23, the General Assembly also posted a public comment portal on its website. See (Doc. 148 at 3); JX065. Members of the public could upload comments and documents there, all of which were available to any Committee member. See Trial Tr. Vol. IV at 862:21–863:3.

b. Senate and Congressional Criteria

Written criteria guided the Senate Redistricting Committee chairs’ map drawing efforts. The Committee adhered to these criteria and enforced the criteria for all proposed amendments to the Senate and congressional maps.

Much of the 2023 Senate and congressional plan criteria were the same. Both sets of criteria required contiguity, encouraged compactness, and authorized respect for existing political subdivisions like county lines, municipal boundaries, and voting tabulation districts (“VTDs”). JX038; JX047. The Committee used two traditional methods, known as Reock and Polsby-Popper, to measure compactness. Trial Tr. Vol. IV at 838:4–13. Regarding respect for political subdivisions, the Committee also considered shared school areas, university campuses, military bases, and any evidence from public comments about other communities of interest. Id. at 837:21–838:2, 838:15–839:2. Many of these existing political subdivisions were visible in the legislature's mapping software.

Both the Senate and congressional plan criteria also expressly stated that “[d]ata identifying the race of individuals or voters shall not be used in the drafting of districts.” JX038; JX047. Racial data was not included in the legislature's mapping software. Consistent with the written criteria, racial data was not used in drawing districts for the 2023 Senate plan or the 2023 congressional plan. Trial Tr. Vol. IV at 839:13–15, 973:3–6. When amendments were offered in committee or on the Senate floor, the sponsor was asked whether data identifying the race of individuals or voters had been used in creating the amended districts. Id. at 845:22–846:5, 846:24–847:10. Ultimately, no amendments that used racial data were adopted.

Plaintiffs appear to contend that, even without using racial data, Senator Hise could nevertheless use race to discriminate against black voters when drawing district lines because he was already aware of the State's racial demographics. We are not convinced. On cross-examination, Senator Hise acknowledged his general awareness that some counties in northeastern North Carolina have a majority black population, although he could not identify which specific counties so qualified. Id. at 871:24–872:10. He also acknowledged that “[m]inority populations tend to be higher in urban areas than in rural areas” and expressed a general awareness that the town of Greenville in Pitt County has a higher minority population than the state average. Id. at 872:14–21, 874:5–18; but see id. at 830:7–8 (Senator Kandie Smith testifying that “in Pitt we don't have a very large minority population”). This high-level awareness of racial demographics would not enable him to make race-based decisions at the VTD level or otherwise engage in the type of precise discriminatory vote dilution that Plaintiffs allege. Moreover, even if Senator Hise or other legislators were aware of the State's existing racial composition in more granular detail, “there is nothing nefarious about [a map drawer's] awareness of the State's racial demographics.” Alexander, 144 S. Ct. at 1251; see id. at 1242 (“ ‘[R]edistricting legislatures will ․ almost always be aware of racial demographics.’ ” (quoting Miller, 515 U.S. at 916)). “[B]eing aware of racial considerations” is different from “being motivated by them.” Milligan, 143 S. Ct. at 1510 (internal quotation marks omitted). “The former”—which is all that was present here—“is permissible.” Id.

Political considerations were a different story. Both the Senate and congressional plan criteria permitted the General Assembly to consider “incumbent residence” and to “consider partisan advantage and incumbency protection in the application of its discretionary redistricting decisions.” JX038; JX047. Regarding incumbent residence, the Senate Redistricting Committee asked all members of the state legislature and all members of the congressional delegation to provide their addresses to the Committee. Trial Tr. Vol. IV at 840:17–19. Those addresses were added to the legislature's mapping software “as an overlay to the maps so both visually and by data you could see where any incumbent lived within the [S]tate.” Id. at 840:19–21. As for political data, election results from every statewide election for at least the last three election cycles were available for the drafters to use. Id. at 840:4–6, 960:5–25. As Senator Hise explained, they “had the ability to make either composites out of those or to use races individually” when reviewing how a given district performed politically. Id. at 840:6–7; see id. at 861:4–7. The chairs did consider political data in drawing the Senate and congressional maps. Id. at 840:2–4, 856:25–857:11.

Two other redistricting plan criteria were very important but differed somewhat between the Senate and congressional plans. The first was equal population between districts sufficient to comply with federal law. Both the Senate and congressional plan criteria required the Committee to use the 2020 federal decennial census data as the sole basis of population for establishing districts. JX038; JX047. The Senate plan criteria required that “any deviation from the ideal population for a legislative district shall be at or within plus or minus five percent,” based on rulings from the North Carolina Supreme Court. JX047. The congressional plan criteria, by contrast, required the number of people in each congressional district to be as equal as possible, reflecting the more stringent equal population requirement for districts electing federal representatives. JX038; see Trial Tr. Vol. IV at 855:6–7 (“[W]ith the congressional maps, equal population is almost a zero-variance concept.”).

Second, only the Senate plan criteria included a requirement to adhere to the county grouping and traversal rules established by the North Carolina Supreme Court. JX047. By way of background, the North Carolina Constitution commands that “[n]o county shall be divided in the formation” of a legislative district. N.C. Const. art. II, §§ 3(3), 5(3). To reconcile that “Whole County Provision” with the preeminent “one-person, one-vote” requirement of the Equal Protection Clause, the North Carolina Supreme Court has mandated a formula of county grouping and traversal rules crafted to comply with the state constitutional provision to the maximum extent possible under federal law. See Stephenson v. Bartlett, 562 S.E.2d 377, 396–398 (N.C. 2002) (Stephenson I); Stephenson v. Bartlett, 582 S.E.2d 247, 250–251 (N.C. 2003) (Stephenson II); Harper III, 886 S.E.2d at 421–422; see Pierce v. N.C. State Bd. of Elections, 94 F.4th 194, 228 (4th Cir. 2024) (“Redistricting plans can ‘depart from strict compliance’ with the Whole County Provision ‘only to the extent necessary to comply with federal law,’ like the VRA.” (quoting Stephenson I, 562 S.E.2d at 397)).

Under those rules, if a single county has sufficient population to be its own district or to contain multiple whole districts, that district or those districts must be created. Stephenson I, 562 S.E.2d at 397. If a county's population is too small to form a district alone, then the General Assembly complies with the Whole County Provision by “grouping the minimum number of whole, contiguous counties necessary to comply with the at or within plus or minus five percent” population standard and then draws districts within that grouping, crossing interior county lines “only to the extent necessary to comply with” the equal population standard. Id. After single county districts are formed, all two-county groupings must be created before three-county groupings can be formed, and so forth. Trial Tr. Vol. IV at 836:18–21. To comply with the county grouping and traversal rules of “Stephenson I and its progeny, the General Assembly uses an algorithm to determine county groupings ․ to minimize the number of counties traversed by district lines.” Pierce v. N.C. State Bd. of Elections, 713 F. Supp. 3d 195, 223 (E.D.N.C. 2024); see Trial Tr. Vol. II at 449:15–23; NAACPPX200 at 9 (“The county clusters are largely algorithmically determined through an optimization procedure outlined by the NC Supreme Court in Stephenson v. Bartlett.” (internal quotation marks omitted)).

c. Senate Plan

The 2023 Senate plan was predominantly drawn to satisfy the equal population and county grouping requirements. Following the Stephenson I formula, the Senate Redistricting Committee chairs began by drawing single-county districts out of all counties whose populations fell within a plus or minus five percent deviation from ideal equality. The drafters then created all groupings of two counties that complied with the required population deviations. They repeated this process incrementally until reaching a maximum grouping number of ten counties. Trial Tr. Vol. IV at 836:10–23. The chairs took efforts to comply with traversal rules, which limit the number of crossovers into a county when creating divisions within a grouping. Given the particularity of these parameters, Senator Hise testified that the enacted Senate plan was “predominately drawn by the county groupings” formed through this process. Id. at 837:6–11.

Plaintiffs challenge three Senate districts: SD 1, 2, and 8. We find no evidence of discriminatory intent in the creation of those districts.

i. SD 1 and 2

SD 1 and 2 are each composed entirely of whole counties. After implementing the equal population and county grouping and traversal criteria, the Committee chairs were left with eighteen counties in northeastern North Carolina that, following the other Senate plan criteria, could be grouped in two different ways to make an eight-county district and a ten-county district consisting of whole counties. Id. at 841:4–10. The chairs chose the same configuration chosen in 2021. JX223. That configuration grouped Northampton, Hertford, Bertie, Gates, Perquimans, Pasquotank, Camden, Currituck, Tyrrell, and Dare Counties in SD 1 and Warren, Halifax, Martin, Chowan, Washington, Hyde, Pamlico, and Carteret Counties in SD 2. JX202. Once this grouping was selected, there was no further discretion to alter the borders of SD 1 or 2; the contours were dictated by the Stephenson criteria.

The alternative county grouping that the chairs did not select had been used for the 2022 Senate map. See JX221. The chairs preferred the grouping they selected because they considered SD 1 more compact in that version, because it kept four of the five “fingerling” counties of northeastern North Carolina together as a community of interest, and because the counties of SD 1 were largely in the Virginia Beach media market. Trial Tr. Vol. IV at 841:13–842:5. The grouping that the chairs selected also “performed better for Republican candidates” than the alternative used in the 2022 Senate map. Id. at 912:9–12. We have seen no evidence that racial discrimination played any role in the chairs’ selection of the enacted grouping over the alternative.

Plaintiffs argue otherwise, claiming that SD 1 and 2 divide a community of interest in northeastern North Carolina that they call the “Black Belt.” Witnesses differed in their description of the counties that comprise the Black Belt. See, e.g., Trial Tr. Vol. VI at 1265:6–25 (defense expert Dr. Andrew Taylor testifying that he defined this “nebulous” area as consisting of 12 counties in northeast North Carolina, but that Plaintiffs’ expert, Dr. Joseph Bagley, defined it as containing 13); LDTX259 at 5 (Dr. Taylor listing Bertie, Chowan, Edgecombe, Gates, Halifax, Hertford, Martin, Northampton, Vance, Warren, Nash, and Washington Counties); NAACPPX180 at 4 (Dr. Bagley listing those counties as well as Pitt County); see also Trial Tr. Vol. IV at 829:11–17 (Senator Smith testifying that the Black Belt consists of Vance, Warren, Halifax, Northampton, Hertford, Bertie, and Washington Counties). But under any of those formulations, both available Stephenson grouping options for the eighteen counties in SD 1 and 2 would divide the Black Belt in one way or another. See JX202; JX221. We do not draw a negative inference from the chairs’ selection of the enacted option (which was also used in the 2021 Senate plan) over the alternative that was used in the 2022 Senate plan.

ii. SD 8

SD 8 includes Columbus and Brunswick Counties and part of New Hanover County. Following the equal population and Stephenson county grouping criteria, Columbus, Brunswick, and New Hanover Counties formed a three-county pod that must be split into two districts, SD 7 and 8. Trial Tr. Vol. IV at 842:16–21. The combined population of Columbus and Brunswick Counties was too small to comprise a district, whereas New Hanover County was too large to be a district because it exceeded the five percent population deviation threshold. Accordingly, the chairs had to move some population from New Hanover County into a district with Columbus and Brunswick Counties to create SD 8. Id. at 843:2–7.

To accomplish this goal, the drafters moved six New Hanover County precincts into SD 8. The contiguity criterion required that those precincts be reachable from the border of Brunswick County. Senator Hise testified that, following that requirement, the chairs identified the six contiguous precincts “that performed least for Republicans” according to the election results available in the mapping software and “moved them into District 8.” Id. at 843:9–13; cf. LDTX253 at 43 (map showing partisan lean of the precincts in SD 7 and 8). Racial data was not used in selecting those precincts. Trial Tr. Vol. IV at 844:3–6.

Other evidence confirms Senator Hise's unrebutted testimony that the six precincts were chosen because of their election performance. For example, the political data that the General Assembly used in drawing districts shows a 2.69 percentage point increase in the SD 7 vote for President Trump in the 2020 presidential election from the 2022 plan (49.02%) to the 2023 plan (51.71%). JX278 at 3; JX149 at 33. Senator Michael Lee, a Republican, won his election in SD 7 by a slim margin in 2022 but increased his margin of victory in 2024 under the new districting plan. JX365 at 2; JX368 at 2; see also Trial Tr. Vol. IV at 844:12–13 (Senator Hise testifying that SD 7 “is always a very close race” and “always a race that we invest tremendous amounts of [Republican] caucus funds into running”). A Republican also won SD 8 in 2022 and 2024, and in the former he ran uncontested. JX365 at 2; JX368 at 2. We credit Senator Hise's testimony about the nondiscriminatory reasons for the configuration of SD 8.

d. Congressional Plan

The 2023 congressional plan was drawn to comply with the congressional plan criteria. Beginning in the western part of North Carolina, the Senate Redistricting Committee chairs added whole counties into a district until they approached the population of an ideal district. Trial Tr. Vol. IV at 855:15–20. At that point, because of the stricter equal-population requirement for congressional districts, the drafters would divide a county and add population to the district until they reached “plus one or zero of the ideal district.” Id. at 855:20–23. They followed the same process starting in the eastern part of the State, moving toward the middle of the State from both directions. Ultimately, the drafters achieved population equality by creating eight congressional districts with “one person more” than the other six districts. Id. at 855:8–10. To minimize the number of counties that are divided while maintaining equal population, the drafters chose to divide more heavily populated counties multiple times. Id. at 856:7–15. The chairs did not start with the 2022 interim congressional map as a base for drawing the 2023 congressional plan because the 2022 interim map was imposed by state courts, not drawn by the General Assembly. As Senator Hise explained, “[w]e couldn't answer any of the questions as to how the former judges [who created the 2022 map] had put together the map or others, and it just didn't meet our criteria.” Id. at 857:16–858:2.

Plaintiffs challenge five congressional districts: CD 5, 6, 10, 12, and 14. We find no evidence of discriminatory intent in the creation of those districts.

i. CD 5, 6, and 10

CD 5, 6, and 10 converge in the Piedmont Triad region of North Carolina, which is defined by three major municipalities: Greensboro, High Point, and Winston-Salem. CD 5 includes nine whole counties—Alexander, Caldwell, Watauga, Ashe, Wilkes, Alleghany, Surry, Stokes, and Rockingham—and part of Guilford County, including the majority of Greensboro. See JX075; JX081 at 49; WX1 at 15 (showing municipal boundaries relative to congressional districts). CD 6 includes three whole counties—Rowan, Davie, and Davidson—and parts of Cabarrus County, Forsyth County, and Guilford County, including almost all of High Point. See JX075; JX081 at 49; WX1 at 15. CD 10 includes four whole counties—Lincoln, Catawba, Iredell, and Yadkin—and part of Forsyth County, including most of Winston-Salem. See JX075; JX081 at 58; WX1 at 15. Noting that “[t]he Triad is defined as three separate municipalities,” Senator Hise testified that the Committee chairs attempted to minimize splits within those municipalities but did not have a goal or criteria of pairing multiple municipalities within a single district. Trial Tr. Vol. IV at 861:14–862:3; but see id. at 952:15–955:7 (noting that portions of the municipalities had to be divided). Senator Hise also explained that, using data from three prior election cycles for multiple state races, the drafters attempted to draw all three districts so that they would perform for Republican candidates. Id. at 860:24–861:10. We credit his explanation.

ii. CD 12 and 14

CD 12 and 14 both contain parts of Mecklenburg County, which is too populous to be a whole-county congressional district. See JX075. The city of Charlotte, within Mecklenburg County, is itself too populous to be left whole in a single congressional district. When drafting CD 12 and 14, the Senate Redistricting Committee sought to keep as much of Charlotte as possible in one district, adjusting the remaining areas surrounding the city to balance the population of adjacent congressional districts as necessary. Trial Tr. Vol. IV at 859:13–25. Accordingly, CD 12 includes almost all of Charlotte and the majority of Mecklenburg County. See JX081 at 42, 46; WX1 at 20. CD 14 includes parts of Mecklenburg and Polk Counties and four whole counties—Rutherford, Burke, Cleveland, and Gaston. See JX075. As they were drawing the congressional plan, the drafters expected CD 14 to perform as a Republican district and CD 12 not to do so. Trial Tr. Vol. IV at 860:1–5; see also id. at 943:2–3 (“[W]e considered in every [congressional] district ․ whether it performed.”). We again credit Senator Hise's testimony.

e. Senate Bills 757 and 758

On October 18, 2023, the Senate Redistricting Committee chairs introduced Senate Bill 757, titled “Realign Congressional Districts 2023/CCJ-1,” and Senate Bill 758, titled “Realign NC Senate Districts 2023/SCJ-1,” to the full Committee.⁸ See JX099; JX106. The same day, the Committee posted the proposed maps on the General Assembly's public website, along with voluminous data files and statistical reports on their respective features. (Doc. 148 at 4–5.) For the first time in the redistricting process, the chairs authorized compilation of racial data about each proposed plan, including the percentages of minority voters in the proposed districts, which they also released to the public. See JX002 at 5:24–6:9; JX387; JX388.

The Committee met to debate S.B. 757 and S.B. 758 the next day. (Doc. 148 at 6.) During the meeting, the Committee chairs presented and explained the proposed maps and the criteria used to draft them. Regarding the prohibition on using racial data, Senator Hise explained that districts drawn “predominantly based upon race without a compelling interest ․ would be declared illegal racial gerrymanders.” JX002 at 4:20–25 (referencing Cooper, 581 U.S. 285; Covington, 316 F.R.D. 117). Acknowledging that the use of race may be justified by Section 2 of the VRA, Senator Hise opined that “[p]ast decisions and court records demonstrate that to this point nowhere in North Carolina can anyone provide evidence of the [VRA] preconditions.” Id. at 5:9–12. He advised the Committee, however, that the chairs would meet again the following week and would consider any evidence from Committee members or third parties providing “a strong basis in evidence that the [VRA] preconditions are present in a particular area of the [S]tate.” Id. at 6:25–7:3. “Only then [would] the chairs consider using race in amending the districts.” Id. at 7:3–4.

On October 22, members of the Committee and General Assembly leadership received a letter from the Southern Coalition for Social Justice. See NAACPPX047. The Coalition urged the General Assembly to conduct its own racially polarized voting analysis and presented preliminary findings from Dr. Kassra Oskooii claiming that the VRA racially polarized voting preconditions were established in the area covered by SD 1 and 2. The letter asserted that it was possible to draw majority-minority districts in several areas of North Carolina, but the Coalition did not request a majority-minority district be constructed or claim that black voters in northeastern North Carolina needed a majority-black district to elect their candidates of choice. Rather, the Coalition requested that the General Assembly use the county groupings from the 2022 Senate plan for that part of the State. See id.; Trial Tr. Vol. IV at 864:19–865:7.

The Committee chairs reviewed the Coalition's letter, and Senator Dan Blue entered it into the record during one of the Committee's meetings. See Trial Tr. Vol. IV at 905:2–906:19. However, the chairs “saw no reason to change” the maps or their process in response to the letter. Id. at 914:7–12; see also id. at 864:11–13 (testifying that the chairs did not receive any information suggesting there was legally significant racially polarized voting in any area of North Carolina).

During their debates, the Committee approved two amendments to the Senate map. (Doc. 148 at 7.) The sponsoring senators confirmed that their changes complied with the Committee's criteria, including the prohibition on using racial data to draw districts. Trial Tr. Vol. IV at 845:22–846:20; JX003 at 22:4–6. Three proposed amendments failed on the Senate floor. (Doc. 148 at 7–8.) Senator Blue offered two amendments affecting SD 1 and 2, among other parts of the northeast. See JX087; JX088. Those proposals were prepared with racial data and proposed districts that did not comply with the county-grouping requirements. See JX004 at 47:15–17, 50:8–13; Trial Tr. Vol. IV at 847:5–10.

The Committee approved one amendment to the congressional map, which related to military bases and affected Onslow, Cumberland, Sampson, and Robeson counties in CD 3, 7, and 8. (Doc. 148 at 7.) Otherwise, S.B. 757 remained the same throughout the enactment process.

After debates about the proposed Senate and congressional plans in the North Carolina House and Senate and their respective committees, the General Assembly passed S.B. 757 and S.B. 758 as amended on October 25, 2023.

3. Prior Redistricting Processes

Plaintiffs contend that the General Assembly's 2023 redistricting process differed from the process followed in prior years and that those differences are evidence of intent to discriminate on the basis of race. Though there were differences in the process, we find that they do not raise an inference of discriminatory intent. Cf. Rollerson v. Brazos River Navigation Dist. of Brazoria Cnty., 6 F.4th 633, 640 (5th Cir. 2021) (explaining that procedural irregularities are probative of invidious intent only if they “suggest[ ] the decision-makers were willing to deviate from established procedures in order to accomplish a discriminatory goal”). The 2023 redistricting was the General Assembly's third attempt at enacting new districts after the 2020 census and the first redistricting in which the legislature was free to consider political advantage in drawing districts. We find that the different process largely reflects the different context of this redistricting.

a. Timing and Transparency

According to Plaintiffs, the 2023 redistricting was conducted on a condensed timeline with less transparency than in prior years.

In some previous redistricting cycles, the General Assembly began drawing maps earlier in the calendar year and then debated for several weeks or months before enacting the new maps. See, e.g., JX263 (2011 process); Covington, 316 F.R.D. at 126–128 (same); JX265–267 (2001 process). The 2021 process was more compressed, with the redistricting committees first meeting in August to discuss plan criteria and then the maps being enacted on November 4. See Harper I, 868 S.E.2d at 511, 513. The 2023 process proceeded on a timeline more like 2021, with the Senate Redistricting Committee chairs developing plan criteria in the summer of 2023 and the plans enacted into law on October 25. (Doc. 148 at 2, 9.) But the public portion of the 2023 process was much quicker, with only eight days from the introduction of S.B. 757 and S.B. 758 to enactment. Though swift, that process proceeded as is typical for legislation in the General Assembly: redistricting bills were introduced and debated in committees in both chambers, amendments were considered and adopted, and then the bills were held for a vote on the floor of both chambers.

The “brevity of the legislative process” does not “give rise to an inference of bad faith—and certainly not an inference that is strong enough to overcome the presumption of legislative good faith.” Perez, 138 S. Ct. at 2328–2329. A “hurried pace, of course, [can] suggest[ ] an attempt to avoid in-depth scrutiny.” McCrory, 831 F.3d at 228. But Plaintiffs have offered no reason to believe that the speed of the 2023 redistricting process, or the decision to predominantly complete work on the budget before beginning in earnest to draft proposed districts, indicates an intent to discriminate on the basis of race. Cf. Raymond, 981 F.3d at 305–306 (rejecting any inference of bad faith from the General Assembly's process, which involved only five days of legislative debate).

Plaintiffs highlight, however, a connection between the budget the General Assembly passed in September 2023 and a reduction in the availability of redistricting records. Section 27.7(d) of the appropriations bill repealed N.C. Gen. Stat. § 120-133, which concerned “Redistricting Communications.” See WX172; Trial Tr. Vol. IV at 944:22–23. The repealed law had provided that “all drafting and information requests to legislative employees and documents prepared by legislative employees for legislators concerning redistricting the North Carolina General Assembly or the Congressional Districts are no longer confidential and become public records upon the act establishing the relevant district plan becoming law,” and it acknowledged that “[p]resent and former legislative employees may be required to disclose information otherwise protected by G.S. 120-132 concerning redistricting ․ upon the act establishing the relevant district plan becoming law.” N.C. Gen. Stat. § 120-133(a) (2023). By repealing this statute, the General Assembly changed the treatment of legislative employees’ redistricting materials and communications to conform to that of other types of legislation, over which legislators or employees may claim confidentiality.

The timing of this repeal is relevant. Because the repeal was included in the appropriations bill, its passage before much drafting had been done on the 2023 district plans could suggest a deliberate choice to prevent the automatic disclosure of legislative employees’ redistricting materials and communications in the 2023 process. See Trial Tr. Vol. IV at 834:22–24 (explaining the decision to begin drawing the maps after the budget process was largely complete). And some version of this statute had been on the books since 1983, so this change marked a departure from the process followed in multiple prior redistricting cycles. See N.C. Gen. Stat. § 120-129 (1983). Yet the General Assembly's desire to shield the confidentiality of legislators’ redistricting communications, while relevant to our analysis, is consistent with the legislature's avowed partisan motivation for enacting the 2023 Senate and congressional plans. And the presumption of legislative good faith counsels us to require much more before inferring that the legislature acted in bad faith by treating redistricting legislation the same as every other class of legislation considered by the General Assembly. See Alexander, 144 S. Ct. at 1235–1236; Perez, 138 S. Ct. at 2328–2329.

In a related vein, Plaintiffs argue that the 2023 redistricting process was less transparent than other recent processes. For example, in 2019 a court order required the General Assembly to conduct its “entire remedial process in full public view,” with maps drawn at public hearings on computers “visible to legislators and public observers.” Lewis, 2019 WL 4569584, at *134. In 2021, representative-drawn maps would be considered only if they were drawn at computers in a committee room at the legislature, and members of the public could observe the map drawing sessions both in person and via livestream. See JX239; Harper I, 868 S.E.2d at 512–513. By contrast, in 2023 the Senate Redistricting Committee chairs developed the plan criteria and drew the first drafts of the proposed Senate and congressional plans, which became public only when they introduced S.B. 757 and S.B. 758 to the full Committee.

While the map drawing procedures in 2019 and 2021 were undoubtedly more transparent, they cannot be considered “the normal legislative process” against which the 2023 process should be measured. Raymond, 981 F.3d at 303. The 2019 process was compelled by a state court to root out supposedly impermissible partisan considerations, see Lewis, 2019 WL 4569584, at *133–134, and the 2021 process was likewise conducted under the shadow of that ruling. Neither process represented the historical norm for the General Assembly; rather, the procedures used for those cycles were novel. See Trial Tr. Vol. IV at 814:13–16 (Senator Smith testifying that a procedure used in 2021 was never used before or since); cf. Vieth v. Jubelirer, 541 U.S. 267, 277 (2004) (plurality opinion) (“[P]artisan districting is a lawful and common practice.”). After winning the legal battle to consider politics in redistricting, the General Assembly in 2023 quite understandably elected to depart from the dictates of overruled state court decisions and return to a more traditional legislative process.

b. Opportunity for Public Input

Plaintiffs also observe that the General Assembly in 2023 offered fewer opportunities for public input compared to prior redistricting cycles. In 2023, the General Assembly held three public hearings, all conducted before the proposed maps or redistricting criteria were publicly released. Trial Tr. Vol. IV at 968:4–9; (Doc. 148 at 3–4). In previous years, the legislature held more public hearings, offered at more locations across the State, and conducted some hearings after the plan criteria and proposed maps had been made public. See JX258 (advertising dozens of meetings in 2011); JX257 (showing several meeting locations in 2016); JX247 (seven meetings in 2017); JX241 (thirteen meetings in 2021); Trial Tr. Vol. IV at 968:10–19. Several fact witnesses, including Senator Kandie Smith, testified that, in 2023, some members of the public were frustrated by the perceived brevity of the process. See Trial Tr. Vol. II at 343:21–344:2, 370:10–22; Trial Tr. Vol. III at 525:22–526:6; Trial Tr. Vol. IV at 809:17–24. Some also claimed the speed of the process prevented them from meaningfully participating. See, e.g., Trial Tr. Vol. II at 370:10–18.

Defendants emphasize that the 2023 hearings built on the record from the public hearings held in 2021 and 2022 because each of those redistricting efforts relied upon the same 2020 census data. Trial Tr. Vol. IV at 863:23–864:3; see id. at 864:3–6 (“[W]e had had multiple hearings when the decennial information came out; we had the hearings with regard to the redraw, and we did so once again when we were drawing for 2023.”). They also note that the General Assembly maintained the public comment portal from the release of the maps through enactment. Id. at 862:21–863:3.

We find that none of this evidence implies any bad faith or invidious intent on the General Assembly's behalf.

c. Consideration of Race and the VRA

One significant difference between the 2023 redistricting process and those preceding 2016 was the plan criterion forbidding the use of racial data in drafting districts. Plaintiffs argue that the General Assembly's race-blind approach, and its implications for the State's VRA compliance, demonstrates discriminatory intent. In support, Plaintiffs direct our attention to the districting plan criteria, the Committee chairs’ implementation of the prohibition on using racial data, and the Committee's decision not to conduct its own analysis of racially polarized voting.

Since 2016, the legislature's redistricting criteria have prohibited the use of “[d]ata identifying the race of individuals or voters” in the drafting of districts. JX321; JX315; JX224; JX038; JX047. Despite that restriction on the use of race, the 2021 criteria adopted by the joint redistricting committees went on to state that “[t]he Committees will draw districts that comply with the Voting Rights Act.” JX224. Like the criteria used before 2021, neither the Senate nor the congressional criteria in 2023 included that pledge. See JX038; JX047; JX315; JX321. But the Committee chairs who drafted the criteria and drew the districts in 2023 understood that they had to comply with the VRA whether it was mentioned in the criteria or not. Trial Tr. Vol. IV at 974:22–975:1. Indeed, in the Committee, Senator Hise acknowledged the legislature's obligation to ensure the “ultimate maps” complied with the VRA by making any necessary adjustments to the drafts “before the maps move forward.” JX002 at 9:4–11.

Implementing the prohibition on considering racial data, the Committee chairs did not use racial data in drawing any districts. They also asked senators proposing amendments to the maps to confirm that their revisions were drawn without using racial data. Trial Tr. Vol. IV at 845:25–846:5. Plaintiffs contend that the criterion forbidding use of racial data prevented serious consideration of an amendment from Senator Blue demonstrating the feasibility of a majority-minority district east of Raleigh. JX004 at 47:5–25; see Trial Tr. Vol. IV at 847:1–4. Senator Hise testified that inconsistency with plan criteria did “not supersede the committee's ability to adopt an amendment or the [legislative] body's ability to adopt an amendment,” as illustrated by the fact that Senator Blue's proposal was “put into the record” and the Senate conducted a vote on whether to table it. Trial Tr. Vol. IV at 973:5–10; see JX004 at 48:20–49:2. Ultimately, although the Senate Redistricting Committee acknowledged that race may be used to draw districts when necessary to comply with the VRA, the Senate rejected both proposed amendments that used racial data. See JX002 at 5:1–4.

While the Committee chairs acknowledged that race may be used to draw districts when the legislature has a strong basis in evidence to believe the VRA preconditions are satisfied, the Committee did not conduct its own analysis of those racial preconditions in 2023 and instead called for third parties to produce such evidence on a tight timeline before race-based districts would be considered. See JX002 at 6:16–7:6. From Senator Hise's perspective, prior racial polarization studies, including those conducted within the past two years, had not shown legally significant racially polarized voting in North Carolina, id. at 5:9–12, 12:16–21, 13:4–7, and there was no reason to “go[ ] through the same process again expecting a different result,” Trial Tr. Vol. IV at 898:10–11. Nor had any third party responded to his calls for evidence of legally significant racially polarized voting during the 2021 or 2022 redistricting processes. Id. at 892:21–893:2, 973:25–974:16. Without evidence establishing the VRA preconditions, the chairs rejected any use of race to draw districts in an effort “to protect the [S]tate from lawsuits alleging illegal racial gerrymandering.” JX002 at 5:12–17.

Plaintiffs contend that the General Assembly disregarded the VRA, and that its approach to VRA compliance is evidence of intent to dilute the voting strength of black North Carolinians. We do not agree that this evidence reveals discriminatory intent. “[A] legislature attempting to produce a lawful districting plan is vulnerable to ‘competing hazards of liability.’ ” Perez, 138 S. Ct. at 2315 (quoting Bush v. Vera, 517 U.S. 952, 977 (1996) (plurality opinion)). The Equal Protection Clause restricts consideration of race, subjecting race-based districts to strict scrutiny. Wis. Legislature v. Wis. Elections Comm'n, 142 S. Ct. 1245, 1248 (2022) (per curiam). At the same time, compliance with the VRA “often insists that districts be created precisely because of race.” Perez, 138 S. Ct. at 2314. A legislature risks an Equal Protection violation if it draws districts based on race and risks a Section 2 claim if it chooses not to consider race. But Plaintiffs have identified no court suggesting, as they would have it, that a legislature taking the latter path also risks a finding of intentional discrimination for its lack of racial intent.

“Facing this legal obstacle course,” id. at 2315, the General Assembly made the potentially risky choice not to independently assess the racial performance of its proposed maps for purposes of VRA compliance. The legislature chose that hazard in favor of steering clear of the Equal Protection Clause by creating electoral districts without regard for voters’ race. Whatever the flaws of the legislature's approach to the VRA, we do not infer racially discriminatory intent from the State's “laudable effort to disregard race altogether in the redistricting process.” Alexander, 144 S. Ct. at 1241 n.6. At bottom, Plaintiffs would have us conclude that the State's decision not to consider race in redistricting is proof of intent to discriminate on the basis of race. The evidence does not support such a finding. Moreover, such a conclusion would commit state legislatures to always consider race when drawing district lines, “unnecessarily infus[ing] race into virtually every redistricting” and carrying “ ‘us further from the goal of a political system in which race no longer matters—a goal that the Fourteenth and Fifteenth Amendments embody, and to which the Nation continues to aspire.’ ” Strickland, 556 U.S. at 21 (plurality opinion) (quoting Shaw I, 509 U.S. at 657).

4. Statistical Analysis of the 2022 and 2023 Congressional Maps

a. Plaintiffs’ Expert Dr. Rodden

Plaintiffs offered the expert testimony of Dr. Jonathan Rodden in support of their challenge to the federal congressional districts. Trial Tr. Vol. I at 26:3–5. Dr. Rodden examined racial results in North Carolina's 2023 congressional plan. This analysis proceeded in two steps. First, Dr. Rodden analyzed the distribution of racial groups across North Carolina and the extent to which the congressional districts sorted residents by race. WX1 at 2; see also Trial Tr. Vol. I at 29:23–30:8. Second, to the extent such sorting existed, he analyzed the effect it had on black and white residents “in terms of representation.” WX1 at 2; Trial Tr. Vol. I at 30:9–13.

At the completion of this analysis, Dr. Rodden concluded that:

In several of North Carolina's largest agglomerations, urban Black communities were extracted from their residential context and combined with districts dominated by distant rural White communities. This pattern was most evident in Winston-Salem, High Point, and Greensboro in the Piedmont Triad (District 6), as well as Gastonia (District 14) ․ In the Charlotte area, Black voters were moved from District 14 to District 12. These moves involved the creation of less compact districts and introduced additional county and municipal splits. The 2023 Plan disproportionately moved Black voters to new districts and moved them further from the median population center of their district, while often creating large differences between the demographics of the local community and that of the district. Such dispersion makes it difficult for communities to work together, gain the attention of representatives, and achieve shared policy goals.

WX1 at 36–37. To be clear, Dr. Rodden emphasized that his analysis was “descriptive,” not “causal.”⁹ Trial Tr. Vol. VI at 1478:4–5, 1479:2–9. In other words, he offered no opinion as to “the partisan objectives of the General Assembly in crafting the 2023 Congressional Plan.” Trial Tr. Vol. I at 94:1–9. He concluded only that there existed a statistical correlation between race and the placement of individuals and voting precincts into the challenged districts. Trial Tr. Vol. VI at 1478:10–12.

i. Racial Sorting

Dr. Rodden began by describing in broad terms the distribution of racial groups in the 2022 and 2023 congressional maps. He focused on two areas of interest: the Piedmont Triad and the Mecklenburg Area.

As previously mentioned, the Piedmont Triad region in north-central North Carolina encompasses three major cities: Greensboro (in Guilford County), Winston-Salem (in Forsyth County), and High Point (spanning Guilford, Davidson, Randolph, and Forsyth Counties). Trial Tr. Vol. I at 144:6–13. Each city has a substantial black population. Id. In Dr. Rodden's description of the 2022 plan, most of the Piedmont Triad, including the cities of Greensboro and High Point, fell within CD 6. WX1 at 9. The 2023 plan divided this population into four congressional districts: CD 5, 6, 9, and 10. Id. Each new district now encompasses both an area within the Piedmont Triad and other primarily rural communities. Id. This configuration change is shown below:

Tabular or graphical material not displayable at this time.

JX218; JX078. In the 2022 plan, the congressional districts largely followed county lines, with the exceptions of CD 5 and 6, which split Forsyth County and the municipalities of Winston-Salem and Walkertown.¹⁰ WX1 at 8. Dr. Rodden explained that the 2023 plan increased the number of political subdivision splits in the region, dividing Forsyth and Guilford Counties as well as the municipalities of Winston-Salem, High Point, and Greensboro. Id.

Dr. Rodden superimposed these district lines over a dot density map showing the racial geography of black voters in the Piedmont Triad, as seen below.

Tabular or graphical material not displayable at this time.

Id. at 14. The district boundaries in Figure 4 are represented by the green lines. Id. Each black dot represents thirty black voters, and each red dot represents thirty white voters. Id. Dr. Rodden opined that the map “demonstrates the extent to which Black neighborhoods of Winston-Salem were carved out of Forsyth County, and the correspondence of the district lines to the geography of race around High Point and Greensboro.” Id. at 13.

The Charlotte-Concord-Gastonia Metropolitan Statistical Area (“the Mecklenburg Area”) in south-central North Carolina encompasses the counties of Anson, Gaston, Iredell, Lincoln, Mecklenburg, Rowan, Union, and Cabarrus. Id. at 8. Under the 2022 plan, CD 14 included the southern portion of Mecklenburg County and the eastern portion of Gaston County. Id. CD 12 encompassed the northern part of Mecklenburg County and the western part of Cabarrus County. Id. The 2023 plan split Mecklenburg County twice and Cabarrus County once, moving portions to CD 6, 8, 12, and 14, as seen below:

Tabular or graphical material not displayable at this time.

JX218; JX078. The 2022 plan created three county splits (one in Mecklenburg, one in Gaston, and one in Cabarrus), while the 2023 plan also has three (two in Mecklenburg and one in Cabarrus). WX1 at 8. While the 2022 plan split three municipalities (Gastonia, Charlotte, and Matthews), Dr. Rodden opined that the 2023 plan created seven municipality splits (Concord, Kannapolis, Mint Hill, Matthews, Pineville, and Charlotte twice). Id.

As with the Piedmont Triad, Dr. Rodden prepared a dot density map showing the relationship between district lines and racial geography in the Mecklenburg Area:

Tabular or graphical material not displayable at this time.

Id. at 19.

To assess the significance of race in the construction of these districts, Dr. Rodden employed a county-level “envelope analysis.” Id. at 11. This is a descriptive methodology developed by Dr. Stephen Ansolabehere that is used to quantify the extent to which district boundaries split communities along racial lines. Trial Tr. Vol. I at 40:16–23, 42:17–20. In other words, the envelope analysis is designed to provide information about whether district lines have the effect of sorting voters on the basis of race. Id. at 42:23–25. The analysis begins by identifying the “envelope” of a district, which is comprised of all the counties that are fully contained or partially contained within it. WX1 at 11. These counties include the population from which a district could be drawn without crossing additional county lines. Id. This, Dr. Rodden testified, is “the starting point” because it is assumed that whole counties “belong together” within a district. Trial Tr. Vol. I at 41:14, 42:7–15. From there, he assessed the percentage of black and white voters within an envelope that were included in the ultimate district boundaries. WX1 at 11. Dr. Rodden wrote that “[i]f the lines were drawn without respect to race, one would expect the likelihood of inclusion to be roughly similar for White and Black voters.” Id. He did not, however, explain why, see id., and the Court finds no basis on which to adopt that assumption as valid. It does not account for other confounding variables given the fact that voters are not randomly distributed throughout the State by race.

As applied to the Piedmont Triad, Dr. Rodden found that white voters were more likely to be included in CD 6, while black voters were more likely to be excluded. Table 3 in Dr. Rodden's report details these results:

Tabular or graphical material not displayable at this time.

Id. at 12. The envelope for CD 6 includes the counties of Forsyth, Davidson, Davie, Rowan, Cabarrus, and Guilford. Id. Of this population, 47.6% is included within the boundaries of CD 6. Id. That number changes, however, when examining specific racial populations. 51.4% of white registered voters are included within the district, while only 38.2% of black registered voters are included—a difference of 13.2 percentage points. Id.

Dr. Rodden considered that this discrepancy might be explained by partisan affiliation, so he ran a similar analysis controlling for voters’ party registration. Trial Tr. Vol. I at 45:17–46:11. In other words, he looked at registered Democrats, Republicans, and unaffiliated voters as isolated data sets and assessed whether white or black voters within those groups were more or less likely to be drawn into a particular district. Id. at 46:12–20. The result of that analysis as to CD 6 is detailed below:

Tabular or graphical material not displayable at this time.

WX1 at 12. Within each partisan group, Dr. Rodden found the same evidence of racial sorting; white Democrats, white Republicans, and white unaffiliated voters were more likely to be placed into CD 6 than their black counterparts—42.9% of white Democrats compared to 37.7% of black Democrats, 55.6% of white Republicans compared to 43.4% of black Republicans, and 50.5% of white unaffiliated voters compared to 39.4% of black unaffiliated voters were placed in CD 6. Id. Though the discrepancy is highest for Republicans, the total number of black Republicans is much smaller relative to the other groups. Trial Tr. Vol. I at 104:2–6. Further, in this analysis, Dr. Rodden did not assess the geographic areas where black Republicans and white Republicans tend to live. Id. at 104:8–13. Put another way, the data does not account for the possibility that black Republicans and black Democrats tend to live in close physical proximity, which could account for a map drawn to maximize partisan advantage disproportionately affecting black voters, even when accounting for party registration. See id.

A similar trend is observed in CD 12 in the Mecklenburg Area. See WX1 at 17. In the CD 12 envelope, 63.7% of registered voters are grouped into the district. Id. However, 52.6% of white voters and 82% of black voters are so grouped. Id. The same pattern persists when controlling for party, as seen below:

Tabular or graphical material not displayable at this time.

The table shows that 59.7% of white Democrats compared to 82.9% of black Democrats, 45.7% of white Republicans compared to 79.5% of black Republicans, and 53.8% of white unaffiliated voters compared to 79.3% of black unaffiliated voters were placed in CD 12. Id. Dr. Rodden interprets this data as indicating that “District 12 was drawn to encompass a large share of the Black population in the envelope, regardless of party.” Id.

CD 14, which is partially in the Mecklenburg Area, corroborates this pattern:

Tabular or graphical material not displayable at this time.

Id. at 22. As seen above, white voters were more likely than black voters to be in CD 14. When controlling for party registration, 48.5% of white Democrats compared to 28.2% of black Democrats, 64.9% of white Republicans compared to 34.5% of black Republicans, and 54.3% of white unaffiliated voters compared to 29.8% of black unaffiliated voters were placed in CD 14. Id.

Next, Dr. Rodden conducted a core/in/out analysis. Id. at 15. This methodology “examine[s] all the voters that were in a given district in either the prior map or the new map, differentiating between three types of voters: 1) those that remained in the district across both the 2022 and 2023 plans—and thus make up the ‘core’ of the district; 2) the voters [who were] moved out of the district; and 3) the voters [who were] moved into the district.” Id. (capitalization altered). The analysis assumes that if the district boundaries were re-drawn without an eye towards race, the racial composition of the core should not differ much from the population of the prior district, and the composition of the voters who were moved into the district should generally correlate with the composition of the voters who were moved out of the district. Id. Again, no justification is given for this assumption. See id.

The core/in/out analysis for CD 6 is detailed below:

Tabular or graphical material not displayable at this time.

Id. at 16. As seen above, while the racial composition of the core is relatively consistent with the 2022 population, there is a dramatic difference in the racial composition of the voters who were moved into and out of the district. See id. The black share of voters who were moved out of the district in the 2023 plan was roughly seventeen points higher than the black share of voters who were moved into the district. Id. And of those who were moved into the district, 72.57% were white. Id.

As with the envelope analysis, Dr. Rodden controlled for party affiliation:

Tabular or graphical material not displayable at this time.

Id. Table 6 demonstrates that among Democrats and unaffiliated voters, the black share of voters who were moved out of the district is notably higher than those who were moved into the district. Id. For example, for Democrats, 60.66% of the voters who were moved out were black, while only 45.38% of the voters who were moved in were black. Id. Dr. Rodden notes that this trend, at first blush, appears less pronounced with Republicans; there is, after all, only a 1.44 percentage point difference between the black share of Republicans who were moved out of the district (2.43%) and the black share of Republicans who were moved into the district (0.99%). Id. He chalks this up to the low number of black Republican voters overall. Id. When examining the two figures as a quotient, the black share of Republicans moved out of the district is more than twice as high as the black share of Republicans moved into the district. Id.

Dr. Rodden also opined that the core/in/out analysis supported a finding of racial sorting in CD 12 and 14 in the Mecklenburg Area. Id. at 21, 23. Those results are seen below:

Tabular or graphical material not displayable at this time.

Id. at 21. Dr. Rodden emphasizes that CD 12 concentrates urban Charlotte black voters into a single district by drawing from the population of former CD 14. Id. Consistent with that conclusion, the percentage of black voters that make up the core of CD 12 is much higher than the percentage of white voters, and the black share of voters who were moved into the district is higher than the black share of those who were moved out of the district within each partisan group. Id.

As observed previously with the envelope analysis, this trend is reversed as to CD 14:

Tabular or graphical material not displayable at this time.

Id. at 23. Here, the black share of voters who were moved out of the district is higher than the black share of those who were moved in. Id. This trend holds true across the three partisan groups. Id.

In sum, Dr. Rodden concluded that the 2023 plan sorted residents according to their race, particularly in the Piedmont Triad, where black urban neighborhoods “were extracted from their immediate surroundings and combined with faraway rural White areas,” and in the Mecklenburg Area, “where Black voters were concentrated into [CD] 12.” Id. at 27.

ii. Implications

Having concluded that the 2023 plan sorted residents according to their race, Dr. Rodden sought to quantify the effect on voters. Id. at 27–28. He did so first using a “fractionalization” analysis, which calculates the probability that two randomly selected people from the same district under the 2022 plan will end up in the same district in the 2023 plan. Id. at 28; Trial Tr. Vol. I at 64:20–65:21, 66:14–19. Figure 10 from Dr. Rodden's expert report, shown below, presents the results of this analysis. WX1 at 28.

Tabular or graphical material not displayable at this time.

Id. As shown in Figure 10, the fractionalization statistic is represented on the vertical axis. Id. As this value approaches one, there is a higher probability that two randomly selected people who were in the same district under the 2022 plan will be in a different district under the 2023 plan. Id. Conversely, a value of zero would indicate that all voters continue to reside in the same district. Id. The horizontal axis represents the absolute difference in the black voting age population (“BVAP”) between the 2022 plan and the 2023 plan. Id.

The chart shows that the districts whose populations experienced the most fractionalization were CD 6, 12, and 14, along with the districts heavily affected by their redrawing, including CD 5, 9, 10, and 13. Id. Dr. Rodden observed that “[t]he upward sloping arrangement of data markers in Figure 10 reveals that where BVAP was increased or decreased most significantly, the district configuration was most radically altered.” Id.

Dr. Rodden conducted several additional analyses to examine other ways that the above-described sorting affected black and white voters. For instance, one aspect of representation depends on voters residing in relatively compact districts; geographically proximate communities often have similar interests or needs and can more effectively lobby for policies to further their shared interests when placed together in the same district. Id. at 27; Trial Tr. Vol. I at 63:15–64:17. So, voters whose communities are fragmented across different districts and placed with faraway areas can have their representation undermined. Trial Tr. Vol. I at 63:15–64:17.

Dr. Rodden found that, in the 2023 congressional plan, black voters were more likely to be assigned to non-compact districts and be placed further from what he determined was the center of their districts. WX1 at 28–29. For this analysis, he examined the distance and racial composition of every census block to the median population center of the district to which it is assigned. Id. at 29–30; Trial Tr. Vol. I at 67:15–68:3. This analysis demonstrated that, as the black population of a census block increases, so too does its distance from the center of its district. WX1 at 29–30; Trial Tr. Vol. I at 68:4–14.

Tabular or graphical material not displayable at this time.

WX1 at 30. The black line, which represents census blocks across North Carolina, shows that the average distance to the median population center of a district increases as the BVAP of the census block increases. Id. This trend is particularly pronounced in the Piedmont Triad. Id.

To further quantify the effect on voters, Dr. Rodden examined the concept of “dislocation.” Id. This analysis proceeded in two steps. First, it imagined that for each black North Carolina voter there was a bespoke congressional district comprised of their nearest 745,670 neighbors. Id. It counted the number of those individuals who were also black to assess the extent to which an individual lives in a neighborhood with other black residents. Id. Next, for each black voter, it counted the number of other black voters in the district to which they have been assigned and examined whether the black population share of the assigned district “is larger or smaller than that of the relevant geographical neighborhood.” Id. Dr. Rodden advised that “[t]he larger the difference between the racial composition of the neighborhood and that of the district, the more evidence [there is] that the districts were drawn in a way that disrupts local clusters of Black voters.” Id. at 30–31. His dislocation analysis of the 2022 and 2023 plans is shown below:

Tabular or graphical material not displayable at this time.

Id. at 32–33. Figure 12 serves to visualize where dislocation is most pronounced. Id. at 31. Dr. Rodden described the color gradient used in the figure as follows:

The colors correspond to the extent and direction of dislocation experienced by Black voters. Darker colors of green mean that Black voters have been placed in a district where the Black share is larger than that of their neighborhood. One might refer to this as ‘packing.’ Darker colors of red indicate that the individual is in a district where the minority population is lower than that of the neighborhood. One might refer to this as the ‘cracking’ or ‘diluting’ of Black voters. Lighter colors indicate that the racial composition of the district comes relatively close to the racial composition of the neighborhood.

Id. As demonstrated in the 2023 plan, in the Piedmont Triad, the orange shading indicates that black voters were consistently in districts with a higher white population than their surrounding community. Id. at 31, 33. According to Dr. Rodden, this indicated that the black population in the Piedmont Triad was “cracked.” See id. at 31. Conversely, in the Mecklenburg Area, the green shading in CD 12 indicated to Dr. Rodden that black voters were “packed,” i.e., placed in a district with a higher black population than their surrounding community. See id. at 31, 33. Dr. Rodden maintained that deviation in either direction from an imagined ideal is evidence that race correlates with an individual's placement into a district. Id. at 30–31. But he did not consider whether color-blind partisan sorting would have produced the same outcome.

b. Critiques from Defendants’ Experts

Defendants offered the expert testimony of Dr. Sean Trende and Dr. Michael Barber to respond to Dr. Rodden's report and to provide their own analysis of North Carolina's 2023 congressional plan.

One of their overarching critiques of Dr. Rodden's report was his use of the 2022 congressional plan as a comparison map. The core/in/out, fractionalization, and dislocation analyses each treat the 2022 plan, which was drawn by special masters at the direction of the North Carolina Supreme Court, as the comparison map against which to measure “changes” made in the 2023 plan. See, e.g., WX1 at 15, 28, 32–34. This, Dr. Barber argued, was erroneous because “it is entirely possible that the legislature gave no consideration at all to the 2022 map when they began drawing the 2023 [map].” LDTX253 at 29. He hypothesized that the legislature used the 2021 plan it had previously approved or even drew the map from a blank slate. Id. To illustrate how the starting map changes the core/in/out analysis, he examined how the core of CD 6 and CD 12 changes depending on whether one uses the 2022 plan or the 2021 plan as the comparison:

Tabular or graphical material not displayable at this time.

Id. at 32–33. The precincts with purple shading represent the “core” of the district. See id. When using the 2021 plan as the comparison map, the core of both CD 6 and CD 12 is significantly larger, which indicated to Dr. Barber that the legislature may have relied more on the 2021 map when drawing the 2023 plan. Id. at 30–31; Trial Tr. Vol. V at 1045:12–17.

The Court also notes that the 2022 plan was drawn following a decision by the North Carolina Supreme Court that required the special masters to achieve partisan fairness when drawing the map. See Harper I, 868 S.E.2d at 548–549, 559. This makes the 2022 plan a particularly inapposite comparison now that partisan gerrymandering is permitted under state law and is a stated motivator in the drawing of the 2023 plan. See Harper III, 886 S.E.2d at 449. Further, Dr. Rodden failed to consider that his methodologies, if applied to the 2022 plan, might flag race as a statistically significant factor in the drawing of that map; after all, as already described, an attempt to correct for partisan gerrymandering would almost invariably, considering North Carolina's political geography, move black and white voters in and out of certain districts at disproportionate rates. See LDTX253 at 7.

Dr. Trende and Dr. Barber also critiqued Dr. Rodden's use of the county envelope method, both generally as an analytical tool and as he applied it to the 2023 plan. They argued that it is “not a reliable way to determine if race was the predominant factor (or a factor at all) in the construction of the congressional district boundaries.” LDTX254 at 4.

Dr. Trende argued that the envelope method is unreliable because it does not have a testable null hypothesis—i.e., a theory about what a neutral, randomly drawn map would look like. Trial Tr. Vol. VI at 1337:25–1338:17. Dr. Rodden, as previously noted, posited that a neutral map would have an equal probability that white and black voters were included in a certain district. See WX1 at 11. Dr. Trende believes this is unfounded, particularly when considering the concept of clustering—the principle that people are not randomly distributed throughout the State. Trial Tr. Vol. VI at 1338:24–25, 1339:4–23. He argued that when there is evidence that members of a certain political party or race tend to “cluster” in the same geographic areas, one should not expect, using an envelope analysis, to find an equal probability that members of those groups are going to be included in a particular district. Id. at 1339:4–23. Because densely populated areas in North Carolina evidence such clustering and because redistricting requires legislatures to consider contiguity, Dr. Trende testified that he would expect that black and white voters were affected differently by any redistricting process, whether they were targeted on the basis of their race or not. Id. at 1340:7–15.

Concerning Dr. Rodden's application of the envelope analysis to the 2023 plan, Dr. Barber critiqued Dr. Rodden's use of voter registration files as a source of data. Trial Tr. Vol. V at 1034:23–1035:11. He reasoned that (1) this data does not appropriately reflect how the redistricting process occurs because legislatures must draw maps based on the entire population of a state, not just the registered voters; and (2) it does not account for the significant population of unaffiliated voters who otherwise exhibit partisan voter behavior. Id. at 1034:23–1035:21. To his mind, the more appropriate unit of analysis (rather than one registered voter) would be a voting precinct.¹¹ LDTX253 at 15.

Dr. Barber posited that one way to test the reliability of the envelope method is to look at a set of maps “in which the researcher is absolutely certain that race played no role in the drawing of the districts.” Id. at 25. The researcher will then run the same regression models that were used for a challenged map and see how many times the model flags race as a significant predictor of a district's boundaries, thus producing a “false positive.” Id. Here, Dr. Barber used a redistricting algorithm ¹² to generate 5,000 maps with fourteen congressional districts in North Carolina “that adhere to traditional redistricting criteria [such as] equal population, geographic compactness, avoidance of county divisions, and contiguity.” Id.; see also Trial Tr. Vol. V at 1036:22–25. The algorithms were provided with no racial data. LDTX253 at 25–26. Once the maps were generated, Dr. Barber computed each district's BVAP, partisan lean, and total population, and then conducted a precinct-level envelope analysis to see whether the BVAP of a precinct was flagged as a statistically significant predictor of whether it would be included in a certain district. Id. at 26. Figure 11, shown below, displays the results of this analysis:

Tabular or graphical material not displayable at this time.

Id. at 28. Despite being created without any racial (or partisan) data, every single map was flagged by the algorithm as containing at least two districts that were a potential racial gerrymander, and in a majority of the simulated maps at least half the districts were flagged as a potential racial gerrymander. Id. at 26, 28; see also Trial Tr. Vol. V at 1037:8–19 (discussing similar results in follow-up analysis incorporating data from 2024 statewide elections to calculate partisan lean); LDTX254 at 8. The envelope method most commonly flagged eight districts in Dr. Barber's first analysis, and it most commonly flagged seven districts in the second. LDTX253 at 26, 28; LDTX254 at 8. Dr. Barber concluded that these results demonstrate that the envelope method “generates numerous false positives by flagging districts as having race as a significant predictor in their shape even when [it is known] that race was not considered.” LDTX254 at 5. He therefore found that it was not a reliable method for determining whether the 2023 districts were drawn with an eye towards race. Id.

Dr. Barber hypothesized that these false positives were generated because race correlates with other factors such as population density, urban and rural distributions, and geography. Trial Tr. Vol. V at 1038:17–25. In other words, he believed the method incorrectly attributed to race what could be attributed to these other correlates. See id. To test this, he ran a second regression where he instructed the algorithm to have no regard for traditional redistricting criteria such as continuity, county boundaries, and population size. Id. at 1039:1–8. He then conducted an envelope analysis of those maps. Id. The result of this analysis is displayed below:

Tabular or graphical material not displayable at this time.

LDTX254 at 8. Under the new criteria, the envelope method flagged far fewer (often zero) districts as having race as a statistically significant predictor of its shape. Trial Tr. Vol. V at 1039:9–17. Dr. Barber noted that this was to be expected because the algorithm did not pay attention to race, but notably, it also did not pay attention to other redistricting criteria that often correlate with race. Id. So, when looking at the different results produced in both simulations, Dr. Barber concluded that the envelope method was unhelpful because it tends to falsely identify districts as potential gerrymanders even when districts are drawn using only traditional, race-neutral redistricting criteria. Id. at 1039:18–24.

Both experts also critiqued Dr. Rodden's use and application of the core/in/out method. LDTX253 at 29; LDTX266 at 37. Dr. Barber, as already noted, argued that the method is unreliable because “the map that is chosen as the comparison map will dramatically impact the conclusions [drawn] from this analysis.” LDTX253 at 30.

Dr. Barber also addressed Dr. Rodden's response that the 2023 plan treated black voters differently even as compared to the other draft maps considered by the legislature. LDTX254 at 26.

Tabular or graphical material not displayable at this time.

Id. at 31. In Figure 6, the left panel shows what percentage of the Piedmont Triad districts in both the proposed maps and the adopted one are comprised of black voters; the right panel shows the partisan lean of those districts. Id. at 30. Dr. Barber opined that the enacted map was not an outlier from the other proposed districts in terms of racial composition. Id.

For his part, Dr. Trende found that Dr. Rodden's attempt to analyze the 2023 plan's effect on black Democrats and black Republicans separately was unconvincing. LDTX266 at 66. He observed that there are very few black Republicans in North Carolina. For example, of the 1,015,031 registered voters in CD 6's envelope, only 5,718 (or 0.56%) are black Republicans. Id. He testified that black Republicans tend to live in the same precincts as black Democrats and black Independents. Trial Tr. Vol. VI at 1356:20–1357:6. Even if the legislature had wanted to harness black Republican voters for partisan advantage, it would be extraordinarily difficult to do so. LDTX266 at 67–68. After all, none of the precincts in North Carolina with a Republican majority has even 1% black Republican voters. Id. at 68.

Overall, Dr. Trende found that he could not disentangle race from politics. To illustrate this, he created several dot density maps that show the distribution of both black voters and Democratic voters across North Carolina.

Tabular or graphical material not displayable at this time.

See LDTX266 at 38–41. He found that “the concentrations of Democrats and the concentrations of Black residents of North Carolina, while not completely overlapping, are nevertheless consistent.” Id. at 39. So, to the extent that the black voter base was “cracked” in the Piedmont Triad and “packed” in CD 12, so too were Democratic voters. Trial Tr. Vol. VI at 1349:17–19.

c. Summary

The core dispute between the parties’ experts is whether the districts in the 2023 congressional plan can adequately be explained by something other than race. Using statistical analyses to distinguish between a map drawn to maximize partisan advantage and a map drawn to discriminate based on race is a difficult task considering that the overwhelming number of black voters in North Carolina are Democrats. Nearly any map that moves Democrats in or out of a district will disproportionately affect black North Carolinians, but that does not prove that the legislature acted with discriminatory intent. Correlation is not causation.

Dr. Rodden posits that his analyses can control for partisan affiliation and that they still show a correlation between race and placement in a district—especially in the Piedmont Triad and the Mecklenburg Area. WX1 at 36–37. However, the data do not support this conclusion. The envelope and core/in/out analyses attempt to control for partisan affiliation, but their conclusions rely on a remarkably small number of black Republicans that tend to live in the same regions as black Democrats. In any event, as Dr. Trende argued, these methodologies do not reveal the legislature's motivation. LDTX266 at 38. None of the above-described data can disprove—or is even inconsistent with—the General Assembly's stated motivation, which was to comply with traditional districting criteria and draw a map that benefitted Republicans. The data show, and Dr. Rodden does not dispute, that this was the result in the 2023 congressional plan. If Plaintiffs are to carry their burden and show racial motivation, they must do more. Yet Dr. Rodden has expressly disclaimed his capacity to show causation, and the Court, from his testimony and report, is unable to otherwise make that inference.

5. Impact on Black Voters

We turn next to Arlington Heights’ final factor and consider whether and to what extent the 2023 plan “bears more heavily” on black voters. Arlington Heights, 429 U.S. at 266 (internal quotation marks omitted). Without conducting a full dilutive effect analysis, we assess any disparate impact of the 2023 plan and whether it is evidence that the legislature “acted with invidious intent.” Abbott, 138 S. Ct. at 2325. After careful review of the evidence presented, we find that it is not.

As a threshold matter, it is undisputed that black-preferred candidates are less successful under the 2023 plan than under the 2022 plan. Plaintiffs’ experts Dr. Maxwell Palmer and Dr. Kassra Oskooii both used ecological inference models to identify cohesive racial voting blocks in North Carolina and the candidates that enjoyed “a significant majority” of those groups’ support. WX4 at 4; NAACPPX189 at 4. They found that black voters are highly cohesive; indeed, in forty-eight statewide elections from 2016 through 2022, an average of 97.6% of black voters supported the same candidate. WX4 at 4–5; Trial Tr. Vol. I at 160:16–21. Both experts assessed those candidates’ performance under the 2022 and 2023 plans. Using data from the above-described statewide elections, Dr. Palmer found that under the 2022 plan, black-preferred candidates, on average, won a majority of the vote in 6.2 congressional districts. WX4 at 7–8; Trial Tr. Vol. I at 169:8–14. Under the 2023 plan, that number dropped to 3.8 (a decrease of about 38.7%), inhibiting the ability of black-preferred candidates to win in CD 6 and CD 14. WX4 at 7–8; Trial Tr. Vol. I at 169:19–21. These trends persisted when considering data from the 2024 election cycle. There, black-preferred candidates under the 2022 plan won in an average of 6.8 congressional districts, while under the 2023 plan, they won in only 3.9. WX6 at 5–6; Trial Tr. Vol. I at 170:5–13.

Dr. Palmer also compared black and white voters’ respective ability to elect their preferred candidate in the 2020 and 2024 presidential elections. WX4 at 9; WX6 at 7. Using 2020 election data, he found that under the 2022 plan, 66.1% of black voters and 58.6% of white voters lived in a congressional district where their preferred candidate won a majority of the vote. WX4 at 9. Under the 2023 plan, the percentage of black voters living in such a district dropped to 43.3%, while the percentage of white voters increased to 69.5%. Id. These shifts are more pronounced when using data from the 2024 election. Under those parameters, the 2023 plan results in 30.1% of black voters and 72.8% of white voters living in a district where their preferred presidential candidate is elected. WX6 at 7.

Similar observations were made in the northeastern Senate districts. Dr. Oskooii found that under the 2023 plan, black-preferred candidates were unlikely to win in SD 1 and 2. NAACPPX208 at 20. Specifically, in SD 1, black-preferred candidates won zero statewide races from 2020 through 2024; in SD 2, they won zero statewide races in 2020 and 2022 but 7% of the races in 2024. Id. In the district-specific state Senate race, the black-preferred candidate lost in both districts in 2024. Id.

These statistical findings are undisputed; the parties contest only their implications. Plaintiffs argue that this evidence supports a finding that the 2023 plans were “enacted with the intent to discriminate on the basis of race” and thus “deny Black voters equal voting power in the challenged ․ districts.” (Doc. 165 at 272, 278.) We disagree.

As has been explained previously, a threshold difficulty in assessing the comparison data is Plaintiffs’ reliance on the 2022 map. At times, both Dr. Palmer and Dr. Oskooii compared the electoral outcomes of districts in the operative map against what was observed under the 2022 plan. See WX4 at 7; NAACPPX189 at 80, 89. Such reliance was in error. There is no indication that the General Assembly used the 2022 maps—which, again, were drawn by special masters at the direction of the North Carolina Supreme Court—when crafting the 2023 plans. See LDTX253 at 29. If anything, the record suggests that the General Assembly looked to its previously enacted 2021 map when beginning the process. Id. at 30–33; Trial Tr. Vol. V at 1045:12–17. It is unsurprising that a plan configured by a Republican-controlled legislature benefitted Republican electoral interests more than a plan configured by special masters to meet electoral fairness criteria. So we find that the severity of the impact on black-preferred candidates can, at least in part, be attributed to the General Assembly returning to its most recently enacted congressional map as a starting point.

All this notwithstanding, the 2023 plan has not so uniformly impacted the capacity of black voters in North Carolina to elect their preferred candidates as to otherwise support a viable intent claim. Among the six congressional districts Plaintiffs originally challenged, in the 2024 election black-preferred candidates were victorious in both CD 1 and CD 12. NAACPPX208 at 29; JX369 at 2, 3–4. Using the 2023 district boundaries for CD 1 and data from 64 statewide elections from 2016 through 2024, Dr. Oskooii found that black-preferred candidates would have won in 46 (or roughly 71.9%) of the races. NAACPPX208 at 29. As for Senate districts in the northeastern region, black-preferred candidates continue to win (and are predicted to continually win) in SD 5, while they are competitive in SD 11. Id. at 20. Despite Plaintiffs’ urging, we do not agree that a map where black-preferred candidates win one or two races in the Black Belt, as opposed to the historical rates of two or three, supports a finding of invidious intent. (Doc. 165 at 265–66.) More broadly, as discussed in detail below, see infra Section V(B)(4)(b), voters under the 2023 plan continue to elect black candidates at the federal and state level at rates at or above racial parity. In the 2024 general election, 28 out of 120 state representatives (23.3%), 10 out of 50 state senators (20%), and 3 out of 14 congressional representatives (21.4%) were black. These numbers are on par with the black population in North Carolina, which constitutes 22.46% of the total population and 21.37% of the voting-age population. JX149 at 8, 14.

Taken together, these results are entirely consistent with Defendants’ stated motivation in drawing the maps: to bolster the number of Republican representatives. Dr. Palmer himself admitted that in his analyses the electoral success of black-preferred candidates is, without exception, coextensive with the electoral success of Democrats. Trial Tr. Vol. I at 191:25–192:17. Put another way, when the black-preferred candidate lost, the Republican candidate won. See id. Without further credible evidence of unlawful motivation, this undercuts a theory that the maps were drawn to disadvantage black voters—separate and distinct from Democrats. These “racial effects of a political gerrymander in a jurisdiction in which race and partisan preference are very closely correlated” do not support a finding that the legislature acted with “a racial motive” when it drew the challenged districts. Alexander, 144 S. Ct. at 1241–1242.

6. Summary

In sum, Plaintiffs have not proven that the General Assembly enacted the 2023 state Senate or federal congressional map “ ‘as a purposeful device to minimize or cancel out the voting potential of’ ” black voters. Id. at 1252 (quoting Miller, 515 U.S. at 911). Plaintiffs have not proven that racial data was even available to the mapmakers when they drew the districts, much less that the mapmakers used racial data in drawing districts, much less that they used racial data to discriminate against black voters. We have seen no evidence that the General Assembly was motivated by a discriminatory purpose. Plaintiffs’ experts denied any effort to prove a discriminatory purpose, and their racial correlation analyses are equally consistent with, if not more consistent with, the General Assembly's avowed partisan motivation. Moreover, even if the 2023 maps had a racially disparate impact, and even if members of the General Assembly were aware of that impact, Plaintiffs still have not proven that the challenged districts were configured or adopted “at least in part ‘because of,’ not merely ‘in spite of,’ [their] adverse effects upon” black voters. Pers. Adm'r of Mass. v. Feeney, 442 U.S. 256, 279 (1979).

Plaintiffs have argued that the General Assembly's partisan motivation gave it reason to discriminate against black voters because they are reliable votes for Democratic candidates. The evidence does not bear out this hypothesis. Instead, we find the evidence more consistent with Defendants’ explanation: that the General Assembly, having persuaded both the United States Supreme Court and the North Carolina Supreme Court that neither the federal nor the state constitution limits the General Assembly's discretion to consider political factors when redistricting, did in fact work to effectuate its political goals with political data, as was its declared intent. Plaintiffs have provided no explanation why a mapmaker who wanted to produce safely Republican districts would use “racial data—with the associated legal risks—as a proxy for partisan data when he had access to refined, ․ precinct-level political data” reflecting the results of all statewide elections for the past three election cycles. Alexander, 144 S. Ct. at 1243. Plaintiffs have not overcome the presumption of legislative good faith and have not shown that racial discrimination was a substantial or motivating factor behind enactment of the 2023 redistricting maps.

C. Conclusions of Law

Because Plaintiffs have failed to prove that the General Assembly enacted the 2023 maps with a discriminatory purpose, they have not proven intentional vote dilution under the Fourteenth or Fifteenth Amendments or under Section 2 of the VRA. See id. at 1252; Chisom, 501 U.S. at 394 n.21.

IV. RACIAL GERRYMANDERING

A. Legal Standard

Racial gerrymandering presents a claim “ ‘analytically distinct’ from a vote dilution claim.” Miller, 515 U.S. at 911 (quoting Shaw I, 509 U.S. at 652). “Whereas a vote dilution claim alleges that the State has enacted a particular voting scheme as a purposeful device ‘to minimize or cancel out the voting potential of racial or ethnic minorities,’ ․ the essence of the [racial gerrymandering] claim ․ is that the State has used race as a basis for separating voters into districts.” Id. (quoting Bolden, 446 U.S. at 66). “The racial classification itself is the relevant harm” in a racial gerrymandering claim, “ ‘regardless of the motivations’ for the use of race.” Alexander, 144 S. Ct. at 1252 (quoting Shaw I, 509 U.S. at 645). Just as the Fourteenth Amendment forbids a State, absent extraordinary justification, to “segregate citizens on the basis of race in its public parks, buses, golf courses, beaches, and schools,” so also the State “may not separate its citizens into different voting districts on the basis of race.” Miller, 515 U.S. at 911 (internal citations omitted). In other words, the Fourteenth Amendment “prohibit[s] a State from engaging in a racial gerrymander unless it can satisfy strict scrutiny.” Alexander, 144 S. Ct. at 1233.

A claim of racial gerrymandering requires a plaintiff to show “that race was the ‘predominant factor motivating the legislature's decision to place a significant number of voters within or without a particular district.’ ” Id. at 1234 (quoting Miller, 515 U.S. at 916). “To make that showing, a plaintiff must prove that the State ‘subordinated’ race-neutral districting criteria such as compactness, contiguity, and core preservation to ‘racial considerations.’ ” Id. (quoting Miller, 515 U.S. at 916); see Bethune-Hill v. Va. State Bd. of Elections, 580 U.S. 178, 188–191 (2017). “Racial considerations predominate when ‘[r]ace was the criterion that, in the State's view, could not be compromised’ in the drawing of district lines.” Alexander, 144 S. Ct. at 1234 (quoting Shaw v. Hunt, 517 U.S. 899, 907 (1996) (Shaw II)); see also Miller, 515 U.S. at 913 (requiring a showing that “race for its own sake, and not other districting principles, was the legislature's dominant and controlling rationale in drawing its district lines”).

Racial predominance may be shown through a combination of direct and circumstantial evidence. Alexander, 144 S. Ct. at 1234. However, the Supreme Court has made clear that a plaintiff attempting to prove racial predominance with circumstantial evidence alone faces a tall task. Id. The Court has “never invalidated an electoral map in a case in which the plaintiff failed to adduce any direct evidence” but has, “at least in theory, kept the door open for those rare instances in which a district's shape is ‘so bizarre on its face that it discloses a racial design’ absent any alternative explanation.” Id. (quoting Miller, 515 U.S. at 914). As always, we begin “with a presumption that the legislature acted in good faith,” and a plaintiff bears the burden to overcome that presumption. Id. at 1233; see id. at 1235.

“[A] State's partisan-gerrymandering defense ․ raises ‘special challenges’ for plaintiffs.” Id. at 1235 (quoting Cooper, 581 U.S. at 308). When “race and partisan preference are highly correlated,” a plaintiff “must disentangle race and politics ․ to prove that the legislature was motivated by race as opposed to partisanship.” Id. at 1233; see Cooper, 581 U.S. at 308 (explaining that plaintiffs must “disentangle race from politics” to “prove that the former drove a district's lines”). “If either politics or race could explain a district's contours, the plaintiff has not cleared its bar.” Alexander, 144 S. Ct. at 1235.

An alternative map that shows how the State “ ‘could have achieved its legitimate political objectives’ ․ while producing ‘significantly greater racial balance’ ” can “go a long way toward helping plaintiffs disentangle race and politics.” Id. at 1249 (quoting Easley v. Cromartie, 532 U.S. 234, 258 (2001)); Cooper, 581 U.S. at 322 (explaining that only an alternative map can “carry the day” in a case with “meager direct evidence of a racial gerrymander”). In fact, the Supreme Court has instructed district courts to “draw an adverse inference from a plaintiff's failure to submit” an alternative map, interpreting that failure “as an implicit concession that the plaintiff cannot draw a map that undermines the legislature's defense that the districting lines were ‘based on a permissible, rather than a prohibited, ground.’ ” Alexander, 144 S. Ct. at 1250 (quoting Cooper, 581 U.S. at 317).

B. Findings of Fact

Plaintiffs have not proven that race was the predominant factor motivating the General Assembly to place a significant number of voters in or out of SD 8. Indeed, Plaintiffs have not shown that race was a factor in drawing SD 8 at all.

To begin, Plaintiffs have presented no direct evidence of racial predominance. The Senate redistricting criteria forbade use of racial data to draw districts, JX047, and Senator Hise—a Senate Redistricting Committee co-chair and “a primary drawer” of the Senate districts—testified that this directive was followed, Trial Tr. Vol. IV at 833:24–25, 839:13–15. Racial data was not used for any Senate districts, including SD 8.

The evidence instead shows that Senator Hise and the Committee chairs followed traditional, race-neutral districting criteria when drawing SD 8. As we have explained above, the Committee followed the equal population and Stephenson county grouping criteria, under which Columbus, Brunswick, and New Hanover Counties formed a three-county pod that had to be split into two districts, SD 7 and 8. Id. at 842:16–21. The combined population of Columbus and Brunswick Counties was too small to comprise a district, whereas New Hanover County was too large to be a district because it exceeded the five percent population deviation threshold. Accordingly, the chairs had to move some population from New Hanover County into a district with Columbus and Brunswick Counties to create SD 8. Id. at 843:3–7. The contiguity criterion required that the population be reachable from the border of Brunswick County. Following that requirement and Stephenson traversal rules, the chairs identified the six contiguous precincts “that performed least for Republicans” according to the election results available in the mapping software and “moved them into District 8.” Id. at 843:11–13; see id. at 836:24–837:1, 963:21–24. Racial data was not used in selecting those precincts. Id. at 844:3–6.

Plaintiffs have made no effort to demonstrate that SD 8 conflicts with traditional districting criteria. They compare the current SD 8 configuration to that of the 2022 Senate map. The 2022 map shows a similar jut into New Hanover County near the same area, but with a different shape. JX221. The compactness scores for SD 7 and 8 in the 2023 plan are comparable to their scores in the 2022 plan. See JX203 at 1; JX279 at 1. And both plans avoid pairing incumbents. Plaintiffs claim that the 2022 map shows SD 7 and 8 could be configured with lesser population deviations. In the 2022 map, SD 7 deviates -0.07% from the ideal population for a Senate district, while SD 8 deviates -2.11% from the ideal. See JX278 at 1. In the 2023 map, SD 7 deviates -4.94% from the ideal population and SD 8 deviates 2.76%. See JX149 at 1. Both maps fall within the “plus or minus five percent” deviation set by the 2023 Senate plan criteria, JX047, as required by North Carolina law for “compliance with federal ‘one-person, one-vote’ requirements,” Stephenson II, 582 S.E. 2d at 250. The 2022 map thus shows only that SD 8 could be drawn with a lesser population deviation, not that it conflicts with any traditional redistricting criteria.

The sole evidence Plaintiffs offer in their effort to connect race to the configuration of SD 8 is a choropleth map from their expert Anthony Fairfax showing a range of BVAPs for the jut into New Hanover County. See NAACPPX200 at A-14. Mr. Fairfax's map shows that, of the six New Hanover County precincts included in SD 8, one has a BVAP 50% or higher, three have BVAPs between 30% and 50%, and two have BVAPs 30% or lower. Id. Of the nearby New Hanover County precincts that were not included in SD 8, one has a BVAP between 30% and 50% and the others have BVAPs 30% or lower. Id. There is no evidence that Senator Hise or any of the mapmakers were aware of this racial information—or any racial information regarding this part of North Carolina—when drafting the Senate map.

Defendants offer their own map, a choropleth from their expert Dr. Michael Barber, showing the partisan lean of the SD 7 and 8 precincts. See LDTX253 at 43. The partisan lean metric in Dr. Barber's map was created using “an average of a number of actual election results.” Trial Tr. Vol. V at 1061:2–6. On the map, precincts are colored politically, with the darkest blue representing the highest performance for Democratic candidates. Id. at 1060:20–23. The map shows that the six New Hanover County precincts included in SD 8 are very deep blue, “indicating they're very Democratic leaning.” Id. at 1061:1, 7–10. Comparing his map to Mr. Fairfax's BVAP map, Dr. Barber testified that the six precincts are “uniformly very Democratic leaning, but they have a range of BVAP,” which “suggest[s] an effort to gather up the most Democratic precincts without really regard to the racial composition of those precincts.” Id. at 1062:22–1063:1. Dr. Barber's partisan choropleth strongly corroborates Senator Hise's unrebutted testimony that the drafters of the Senate map moved the six contiguous precincts “that performed least for Republicans” into SD 8, without regard for any racial considerations. Trial Tr. Vol. IV at 843:11–13.

Despite this evidence of partisan purpose in SD 8, Plaintiffs do not attempt to disentangle race from politics in any way. Most notably, they have not produced an alternative map showing that the General Assembly “could have achieved its legitimate political objectives in [SD 8] while producing significantly greater racial balance.” Alexander, 144 S. Ct. at 1249 (internal quotation marks omitted). The 2022 Senate map does not serve this purpose in the slightest degree. Plaintiffs provide no evidence that the 2022 configuration of SD 8 would achieve significantly greater racial balance. And Plaintiffs do not even claim, much less purport to demonstrate, that the 2022 configuration could achieve the legislature's legitimate political objectives. That failure of proof makes sense, given that the 2022 Senate plan was drawn to meet the partisan fairness metrics of Harper I.

Plaintiffs incorrectly argue that Defendants should be estopped from asserting a partisan gerrymandering defense because this Court previously ruled that partisanship did not predominate over all traditional districting criteria in the drafting of the 2023 Senate plan. At an earlier phase of this case, Defendants moved for summary judgment on Plaintiffs’ malapportionment claims. (Doc. 78.) We granted that motion because no reasonable factfinder could conclude that the Senate plan's “minor deviations from mathematical equality,” Gaffney v. Cummings, 412 U.S. 735, 745 (1973), reflected “the predominance of illegitimate reapportionment factors rather than the ‘legitimate considerations’ ” incident to “ ‘the effectuation of a rational state policy,’ ” Harris v. Ariz. Indep. Redistricting Comm'n, 578 U.S. 253, 258–259 (2016) (quoting Reynolds v. Sims, 377 U.S. 533, 579 (1964)). (See Doc. 98 at 22.) Even assuming that the pursuit of partisan advantage could be an illegitimate factor in that context,¹³ we found Plaintiffs had created no genuine issue for trial about whether that pursuit predominated over all traditional districting criteria to result in the minor deviations from mathematical equality about which Plaintiffs complained. (Doc. 98 at 22; Doc. 140 at 14.)

Our summary judgment decision did not, as Plaintiffs now claim, eliminate partisan advantage as a motivation for SD 8's configuration. And Defendants’ evidence about the political considerations involved in the boundary between SD 7 and 8 was the same at summary judgment as it was at trial. (See Doc. 98 at 28.) Defendants are not estopped from claiming a partisan motive behind the particular shape of SD 8's jut into New Hanover County.

Furthermore, even if we were to disregard Defendants’ partisan gerrymandering defense, Plaintiffs have not shown that race was a factor, much less the predominant factor, in drawing SD 8. Cf. Trial Tr. Vol. IV at 943:15–18 (Senator Hise testifying that “when you look at the list of the criteria, there are absolutes that we considered that were more important than partisan performance of a district”). Plaintiffs have no direct evidence of racial predominance. They have identified no manner in which SD 8 conflicts with traditional districting criteria or otherwise “subordinate[s] race-neutral districting criteria ․ to racial considerations.” Alexander, 144 S. Ct. at 1234 (internal quotation marks omitted). They have made no attempt to address “the design of the district as a whole.” Bethune-Hill, 580 U.S. at 192. And their map showing wide BVAP ranges for the six New Hanover County precincts within SD 8, while relevant, says little about the legislature's motivation for drawing the district boundary where it did. Put simply, this is not the “rare instance[ ]” in which “circumstantial evidence alone” proves racial predominance. Alexander, 144 S. Ct. at 1234.

C. Conclusions of Law

Because Plaintiffs have failed to prove “that race was the ‘predominant factor motivating’ ” the General Assembly's decision “ ‘to place a significant number of voters’ ” in or out of SD 8, they have not proven racial gerrymandering in violation of the Fourteenth Amendment. Id. (quoting Miller, 515 U.S. at 916).

V. DISCRIMINATORY RESULTS

A. Legal Standard

As we mentioned above, Congress amended Section 2 of the Voting Rights Act in 1982 “to prohibit legislation that results in the dilution of a minority group's voting strength, regardless of the legislature's intent.” Shaw I, 509 U.S. at 641. A Section 2 discriminatory results claim, therefore, “turns on the presence of discriminatory effects, not discriminatory intent.” Milligan, 143 S. Ct. at 1507. The amended Section 2 reads as follows:

(a) No voting qualification or prerequisite to voting or standard, practice, or procedure shall be imposed or applied by any State or political subdivision in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color ․ as provided in subsection (b).

(b) A violation of subsection (a) is established if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice. The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered: Provided, That nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population.

52 U.S.C. § 10301.

“The essence of a § 2 claim is that a certain electoral law, practice, or structure interacts with social and historical conditions to cause an inequality in the opportunities enjoyed by black and white voters to elect their preferred representatives.” Thornburg v. Gingles, 478 U.S. 30, 47 (1986). “That occurs where an ‘electoral structure operates to minimize or cancel out’ minority voters’ ‘ability to elect their preferred candidates.’ ” Milligan, 143 S. Ct. at 1503 (quoting Gingles, 478 U.S. at 48).

The Supreme Court articulated a framework for evaluating Section 2 claims in Thornburg v. Gingles, which it recently reaffirmed in Allen v. Milligan. “To succeed in proving a § 2 violation under Gingles, plaintiffs must satisfy three ‘preconditions.’ ” Milligan, 143 S. Ct. at 1503 (quoting Gingles, 478 U.S. at 50). First, the allegedly affected minority group must be “ ‘sufficiently large and geographically compact to constitute a majority in a reasonably configured district.’ ” Id. (brackets omitted) (quoting Wis. Legislature, 142 S. Ct. at 1248). Second, “ ‘the minority group must be able to show that it is politically cohesive.’ ” Id. (quoting Gingles, 478 U.S. at 51). Third, “the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it ․ usually to defeat the minority's preferred candidate.” Gingles, 478 U.S. at 51. Together, the three preconditions establish that (1) the minority group “ ‘has the potential to elect a representative of its own choice in some single-member district’ ”; (2) “a representative of the [the group's] choice would in fact be elected”; and (3) “ ‘the challenged districting thwarts a distinctive minority vote’ at least plausibly on account of race.” Milligan, 143 S. Ct. at 1503 (quoting Growe v. Emison, 507 U.S. 25, 40, 79 (1993)).

If these preconditions are met, then the plaintiff “must also show, under the ‘totality of the circumstances,’ that the political process is not ‘equally open’ to minority voters.” Id. (quoting Gingles, 478 U.S. at 45–46); see 52 U.S.C. § 10301(b). “A district is not equally open ․ when minority voters face—unlike their majority peers—bloc voting along racial lines, arising against the backdrop of substantial racial discrimination within the State, that renders a minority vote unequal to a vote by a nonminority voter.” Milligan, 143 S. Ct. at 1507; see Perez, 138 S. Ct. at 2331 (“[A] plaintiff ․ must then go on to prove that, under the totality of the circumstances, the district lines dilute the votes of the members of the minority group.”). At this stage, the court determines whether the possibility raised by the Gingles preconditions—“that the State's map has a disparate effect on account of race”—is in fact a reality. Milligan, 143 S. Ct. at 1507.

The totality “inquiry recognizes that application of the Gingles factors is ‘peculiarly dependent upon the facts of each case.’ ” Id. at 1503 (quoting Gingles, 478 U.S. at 79). It requires “ ‘an intensely local appraisal’ of the electoral mechanism at issue, as well as a ‘searching practical evaluation of the ‘past and present reality.’ ” Id. (quoting Gingles, 478 U.S. at 79); see also Johnson v. De Grandy, 512 U.S. 997, 1011 (1994) (“Gingles ․ clearly declined to hold [the preconditions] sufficient in combination, either in the sense that a court's examination of relevant circumstances was complete once the three factors were found to exist, or in the sense that the three in combination necessarily and in all circumstances demonstrated dilution.”). “[A]ny circumstance that has a logical bearing on whether voting is ‘equally open’ and affords equal ‘opportunity’ may be considered.” Brnovich, 141 S. Ct. at 2338 (quoting 52 U.S.C. § 10301(b)). For example, “the reason for polarized voting [can be] a critical factor in the totality analysis,” including evidence that “partisanship [is] the cause of the racially divergent voting.” United States v. Charleston Cnty., 365 F.3d 341, 347, 349 (4th Cir. 2004); see id. at 347–348 (“[T]he approach most faithful to the Supreme Court's case law is one that treats causation as irrelevant in the inquiry into the three Gingles preconditions, but relevant in the totality of circumstances inquiry.” (internal quotation marks omitted)).

The Supreme Court has directed lower courts to consider the so-called “Senate factors” when assessing the totality of the circumstances. Wis. Legislature, 142 S. Ct. at 1250; see id. at 1249. Derived from the Senate Report on the 1982 amendments to the VRA, the Senate factors are “neither comprehensive nor exclusive.” Gingles, 478 U.S. at 45; see id. (“[O]ther factors may also be relevant and may be considered.”). And “ ‘there is no requirement that any particular number of factors be proved, or that a majority of them point one way or the other.’ ” Id. (quoting S. Rep. No. 97-417, at 29 (1982)). The Senate factors include:

“1. the extent of any history of official discrimination in the state or political subdivision that touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process;

2. the extent to which voting in the elections of the state or political subdivision is racially polarized;

3. the extent to which the state or political subdivision has used unusually large election districts, majority vote requirements, anti-single shot provisions, or other voting practices or procedures that may enhance the opportunity for discrimination against the minority group;

4. if there is a candidate slating process, whether the members of the minority group have been denied access to that process;

5. the extent to which members of the minority group in the state or political subdivision bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process;

6. whether political campaigns have been characterized by overt or subtle racial appeals;

7. the extent to which members of the minority group have been elected to public office in the jurisdiction[;]

․

[8.] whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group[; and]

[9.] whether the policy underlying the state or political subdivision's use of such voting qualification, prerequisite to voting, or standard, practice or procedure is tenuous.”

Id. at 36–37 (quoting S. Rep. No. 97-417, at 28–29).

Section 2 does not “demand[ ] that where another majority-black district could be drawn, it must be drawn.” Milligan, 143 S. Ct. at 1506 (internal quotation marks omitted); see also Cooper, 581 U.S. at 305–306. And Congress specified that Section 2 does not establish a right to proportional representation. See 52 U.S.C. § 10301(b); Milligan, 143 S. Ct. at 1509 (“Forcing proportional representation is unlawful and inconsistent with this Court's approach to implementing § 2.”). Ultimately, Section 2's “exacting requirements ․ limit judicial intervention to ‘those instances of intensive racial politics’ where the ‘excessive role of race in the electoral process ․ denies minority voters equal opportunity to participate.’ ” Milligan, 143 S. Ct. at 1510 (brackets omitted) (quoting S. Rep. No. 97-417, at 33–34).

B. Findings of Fact

Plaintiffs claim that SD 1 and 2, as enacted by the North Carolina General Assembly, dilute the votes of black North Carolinians on account of race in violation of Section 2 of the VRA. After considering all the evidence, we find that Plaintiffs have failed to prove their claim. Specifically, Plaintiffs have not demonstrated that the boundaries of SD 1 and 2 interact with social and historical conditions to render the political process not equally open to black voters.

1. First Gingles Precondition

The first Gingles precondition requires that the minority group be “sufficiently large and geographically compact to constitute a majority” in a reasonably configured district. Gingles, 478 U.S. at 50; see Strickland, 556 U.S. at 19–20 (plurality opinion) (“[A] party asserting § 2 liability must show by a preponderance of the evidence that the minority population in the potential election district is greater than 50 percent.”). A district is reasonably configured “if it comports with traditional districting criteria.” Milligan, 143 S. Ct. at 1503.

Plaintiffs’ expert Anthony Fairfax produced two illustrative plans purporting to demonstrate that the General Assembly could draw majority-black districts in the northeastern part of the State. Mr. Fairfax's Illustrative Plan A redrew six of the enacted Senate districts—SD 1, 2, 3, 4, 5, and 11—leaving the other forty-four intact. NAACPPX182 at 53. The plan creates two majority-black districts in the northeast: Illustrative District 2 with a BVAP of 50.25% and Illustrative District 5 with a BVAP of 50.37%. Id. at 58. In Illustrative Plan B, Mr. Fairfax added a single majority-black district in the northeast, the same Illustrative District 2 from Plan A. Id. at 61. Illustrative Plan B retained forty-five of the fifty enacted Senate districts, including SD 5. Id. at 53, 61. Plaintiffs concede that SD 5, as drawn by the General Assembly, has a BVAP of 40.35% and usually elects black voters’ candidates of choice. See NAACPPX208 at 20.

The illustrative districts comport with redistricting criteria like contiguity and compactness. Illustrative District 2 comprises nine whole counties: Bertie, Chowan, Gates, Halifax, Hertford, Northampton, Vance, Warren, and Washington. NAACPPX182 at 54. Illustrative District 5 includes all of Edgecombe and Martin Counties plus part of Nash and Pitt Counties. Id. The illustrative districts are contiguous, and their population deviations fall below the State's 10% threshold. Illustrative District 2 is more compact than enacted SD 2, whereas Illustrative District 5 is less compact than enacted SD 5. Id. at 60. But both illustrative districts are more compact than the least compact districts in the enacted Senate plan. Id. Illustrative Plan B divides the same number of counties, VTDs, and census-designated places as the enacted plan, and one less landmark area. NAACPPX183 at 64. Illustrative Plan A splits the same number of census-designated places and VTDs but splits two more counties and four more landmark places than the enacted Senate plan. Id.

Defendants’ experts, Dr. Barber and Dr. Trende, criticize Mr. Fairfax's illustrative plans in three primary respects. First, Dr. Barber and Dr. Trende contest whether the illustrative districts are reasonably configured. They critique the overall compactness of the illustrative plans and the new county and municipality splits introduced by Illustrative District 5. See LDTX253 at 56–60; LDTX276 at 54–59. In particular, Dr. Trende highlights how Illustrative Plan A results in a three-way split of Pitt County and the city of Greenville, both of which are kept whole in the enacted plan. See NAACPPX187 at 567; LDTX276 at 55–59. In response, Plaintiffs observe that the enacted state House map divides Pitt County and Greenville each into two parts. See JX080; JX082 at 129. The enacted House map undoubtedly causes a less severe split to Greenville and Pitt County than Mr. Fairfax's Illustrative Plan A. See NAACPPX187 at 567; JX082 at 129. But the General Assembly's willingness to divide Greenville and Pitt County in the House plan somewhat blunts Defendants’ protest that Mr. Fairfax's plan cannot be reasonably configured because it splits those communities. While the enacted Senate plan divides fewer counties and landmark areas than Illustrative Plan A, the illustrative plan appears to be within the range of compactness and divisions the General Assembly judged acceptable in the 2023 plans. See NAACPPX183 at 64. Because the first Gingles precondition is not “a beauty contest between the plaintiffs’ maps and the State's,” we do not find this critique disqualifying. Milligan, 143 S. Ct. at 1505 (internal quotation marks omitted).

Second, Defendants and their experts criticize Mr. Fairfax for using race to draw his districts, contrary to the 2023 Senate redistricting criteria and the Equal Protection Clause. See LDTX253 at 60–61; LDTX276 at 55–62. Mr. Fairfax testified that he did not draw the illustrative districts with a racial target in mind, but we find that representation implausible. See Trial Tr. Vol. II at 488:15–17. After all, “the whole point of the enterprise” at the first Gingles precondition is to draw districts that achieve a majority BVAP. Milligan, 143 S. Ct. at 1512. That necessarily requires setting a racial goal and using racial data to achieve it, especially in an area like northeastern North Carolina, where the population for the most part is not compact. And Mr. Fairfax's illustrative districts barely reach majority-black status, with BVAPs of 50.25% and 50.37%. NAACPPX182 at 58. We doubt this was a happy coincidence. Nevertheless, the Supreme Court has instructed that drawing illustrative districts “with an express [racial] target in mind,” that is, “to show ․ that an additional majority-minority district could be drawn,” poses no problem for plaintiffs at the first Gingles precondition. Milligan, 143 S. Ct. at 1512. Unless and until the Supreme Court rules otherwise, Mr. Fairfax's use of race does not disqualify his illustrative districts from being reasonably configured.

Third, Dr. Barber observes that Mr. Fairfax's illustrative plans do not comply with North Carolina's county grouping requirements. LDTX253 at 47–49, 56–59. As we explained earlier, the North Carolina Supreme Court has reconciled the Whole County Provision of the state constitution, N.C. Const. art. II, § 3(3), with the preeminent “one-person, one-vote” requirement of the Equal Protection Clause by creating a formula of county grouping and traversal rules to comply with the state constitutional provision to the maximum extent possible under federal law. See Stephenson I, 562 S.E.2d at 396–398; Stephenson II, 582 S.E.2d at 250–251; Harper III, 886 S.E.2d at 421–422. Following the optimization procedure outlined by the state Supreme Court creates a set of county groupings that the General Assembly must honor. As Dr. Barber explains, even though Illustrative District 2 is composed of whole counties, it divides three different Senate county groupings. LDTX253 at 56. Similarly, Illustrative District 5 divides three Senate county groupings. Id. at 57. And these illustrative districts push other districts out of their county groupings, with the result that Illustrative Plan A divides six county groupings composing six districts and Illustrative Plan B causes five districts to break county groupings. Id. at 48–49. According to Defendants, this demonstrates that the illustrative districts are not reasonably configured.

However, as Plaintiffs correctly note, redistricting plans can “depart from strict compliance” with the county grouping rules “to the extent necessary to comply with federal law,” like the VRA. Stephenson I, 562 S.E.2d at 397. Indeed, Stephenson I directs that “legislative districts required by the VRA shall be formed prior to creation of non-VRA districts,” which then must be drawn in compliance with the Whole County Provision. Id. at 396–397; see Harper III, 886 S.E.2d at 444 (“[I]f Section 2 requires VRA districts, those districts must be drawn first so that the remaining non-VRA districts can be drawn in compliance with the [Whole County Provision].”). Plaintiffs offer Illustrative Districts 2 and 5 as majority-minority districts ostensibly required by the VRA. If they are correct, then the county grouping requirements would not forbid the creation of those districts.

Defendants further fault Mr. Fairfax because, after drawing his VRA districts, he did not create Stephenson-compliant groupings in the remaining areas of the map and draw his non-VRA districts from those groupings. See Trial Tr. Vol. II at 475:11–476:22. Instead, he redrew Illustrative Districts 1, 3, 4, and 11 around his proposed VRA districts without attempting to create Stephenson groupings and work within those parameters. Defendants contend that the General Assembly does not have the option of drawing non-VRA districts outside the bounds of the Stephenson formula. Defendants are correct, and we recognize that reconfiguring one or two districts in a plan can prompt a cascade of changes in neighboring districts. At the same time, the first Gingles precondition directs our attention to the hypothesized majority-minority districts. See Milligan, 143 S. Ct. at 1503; Perez, 138 S. Ct. at 2330–2331; Wis. Legislature, 142 S. Ct. at 1248; Gingles, 478 U.S. at 50. And while it is possible that applying the Stephenson criteria would make a hash of the remaining non-VRA districts in eastern North Carolina, the dearth of evidence leaves us unconvinced that Mr. Fairfax's failure to follow the Stephenson requirements for neighboring districts causes Illustrative District 2 or 5 not to be reasonably configured.

Mr. Fairfax's failure to run a county grouping algorithm or otherwise determine the available county groupings for the districts in his maps that are not majority-minority, however, gives us doubts about his Illustrative Plan B. That plan creates one majority-black district, Illustrative District 2, retains enacted SD 5, and alters the four surrounding districts. SD 5 is a crossover district with 40.35% BVAP that consistently performs for black voters. See NAACPPX208 at 20. Section 2 of the VRA “does not require crossover districts,” Strickland, 556 U.S. at 23, so SD 5 would be subject to regrouping under Illustrative Plan B, see Stephenson I, 562 S.E.2d at 396–397. It is possible that regrouping to accommodate Illustrative District 2 would dismantle SD 5; Plaintiffs offer no evidence on this point. If that is the case, then Illustrative Plan B would merely exchange one minority opportunity district (SD 5) for another (Illustrative District 2). But Plaintiffs cannot “establish a violation of § 2 of the VRA without showing that there is a ‘possibility of creating more than the existing number of reasonably compact’ opportunity districts.” Perez, 138 S. Ct. at 2331 (quoting LULAC, 548 U.S. at 430); see also Johnson, 512 U.S. at 1008. Indeed, “it is hard to see how” the enacted Senate plan dilutes the votes of minority voters compared to Illustrative Plan B “if th[at] alternative ․ would not enhance the ability of minority voters to elect the candidates of their choice.” Perez, 138 S. Ct. at 2332.

Plaintiffs have not shown that Illustrative Plan B would create an additional minority opportunity district in northeastern North Carolina, as opposed to moving the location of that district. For that reason, we conclude that only Illustrative Plan A can carry Plaintiffs’ burden to show the possibility of increasing the existing number of black opportunity districts.

Illustrative Plan A has its own problems. To begin, Illustrative Plan A would move some black residents of Pitt County out of SD 5, which is a performing crossover district, into Illustrative District 3, which would not likely afford them the opportunity they currently have. See NAACPPX182 at 53; LDTX276 at 55; LDTX254 at 48. Illustrative Plan A would also reconfigure SD 11 in a way that appears likely to provide less minority opportunity. See LDTX254 at 48; NAACPPX208 at 21.

Most troubling, Illustrative District 5 within Plan A packs more black population into a district that is already performing for black voters. As discussed, enacted SD 5 has a 40.35% BVAP and consistently elects black voters’ candidates of choice. See NAACPPX208 at 20. Illustrative District 5 would reconfigure this district in order to raise its BVAP to 50.37%. NAACPPX182 at 58. If the General Assembly were to take this step, it would surely face an Equal Protection lawsuit alleging that the legislature created a majority-minority district where one was not necessary for black voters to elect their candidate of choice. See Cooper, 581 U.S. at 305–306 (chastising North Carolina because its belief that Section 2 “cannot be satisfied by crossover districts” was “at war with [the Supreme Court's] § 2 jurisprudence”). But if the first Gingles precondition asks only whether the minority population “is sufficiently large and geographically compact” to be packed into a majority-minority district, then Plaintiffs have shown that it is. Gingles, 478 U.S. at 50.

Accordingly, although we acknowledge the practical problems with Plaintiffs’ Illustrative Plan A and its unsuitability for use as an actual legislative redistricting plan, we find that Illustrative Plan A demonstrates that “the minority group ․ is sufficiently large and geographically compact to constitute a majority” in two single-member districts in northeastern North Carolina. Id. at 50; see Milligan, 143 S. Ct. at 1503.

2. Second Gingles Precondition

The second Gingles precondition requires Plaintiffs to show that black voters are “ ‘politically cohesive.’ ” Milligan, 143 S. Ct. at 1503 (quoting Gingles, 478 U.S. at 51). Plaintiffs may show political cohesion by “showing that a significant number of minority group members usually vote for the same candidates.” Gingles, 478 U.S. at 56.

Plaintiffs’ expert Dr. Kassra Oskooii used ecological inference methods to measure racially polarized voting. See NAACPPX189 at 4. Based on results from the past five election cycles (2016 through 2024), Dr. Oskooii found that average black cohesion estimates are above 95% across enacted SD 1, 2, 5, and 11, as well as in Plaintiffs’ illustrative districts. Id. at 3–4; see also NAACPPX208 at 3–4. His findings were unrebutted, and Defendants do not appear to contest this precondition. We find that black voters in this area of North Carolina are politically cohesive.

3. Third Gingles Precondition

The third Gingles precondition requires Plaintiffs to demonstrate “that the white majority votes sufficiently as a bloc to enable it ․ usually to defeat the majority's preferred candidate.” Gingles, 478 U.S. at 51. “[T]he usual predictability of the majority's success distinguishes structural dilution from the mere loss of an occasional election.” Id. “[A] general finding regarding the existence of any racially polarized voting, no matter the level, is not enough.” Covington, 316 F.R.D. at 167. Instead, Plaintiffs must address the relevant local question of whether, in the particular district, white bloc voting is operating at such a level that it would “generally minimize or cancel black voters’ ability to elect representatives of their choice.” Gingles, 478 U.S. at 56 (internal quotation marks and citation omitted); Cooper, 581 U.S. at 304 n.5; see also Covington, 316 F.R.D. at 168 (explaining that the third Gingles precondition concerns the “effect of bloc voting on electoral outcomes”). Without “ ‘significant white bloc voting it cannot be said that the ability of minority voters to elect their chosen representatives is inferior to that of white voters.’ ” Voinovich, 507 U.S. at 158 (quoting Gingles, 478 U.S. at 49 n.15). “[I]n general, a white bloc vote that normally will defeat the combined strength of minority support plus white ‘crossover’ votes rises to the level of legally significant white bloc voting.” Gingles, 478 U.S. at 56. Importantly, at the third Gingles precondition, we consider the “degree” of white bloc voting, not “why” it occurs, which becomes relevant “in the totality of circumstances inquiry.” Charleston Cnty., 365 F.3d at 348 (internal quotation marks omitted).

Plaintiffs’ expert Dr. Oskooii applied the same ecological inference methods to measure white voter cohesion in what he termed the northeastern districts. Using 64 contested statewide (i.e., exogenous) elections from the past five election cycles (2016 to 2024), Dr. Oskooii analyzed white bloc voting in enacted SD 1, 2, 5, and 11, as well as in Plaintiffs’ illustrative districts. He found that average estimates of white bloc voting against black-preferred candidates range from percentages in the mid-70s to high-80s across the enacted districts he analyzed and the models he applied. NAACPPX208 at 3–4; see NAACPPX189 at 4. For example, in the 2024 elections, Dr. Oskooii found white bloc voting averages of 77.77% in SD 1 and 80.23% in SD 2. NAACPPX208 at 10. Dr. Oskooii also looked at the 2024 election results for the Senate contests within those Senate districts, i.e., endogenous election results. He calculated white bloc voting against the black-preferred candidate at 79% in SD 1 and 82.1% in SD 2. Id. at 16–17.

In addition to these calculations, Dr. Oskooii also conducted an electoral performance analysis to measure whether the white bloc voting he detected is sufficient to defeat black-preferred candidates in the districts he considered. In SD 1, he found that black-preferred candidates did not overcome white bloc voting in any of the 42 statewide elections across 2024, 2022, and 2020. NAACPPX208 at 20. In his analysis of SD 2, black-preferred candidates won 7% of statewide elections in 2024 and no contests in 2022 or 2020. Id. In 2016, black-preferred candidates won 33% of statewide elections in SD 1 and 28% in SD 2. Id. And in both SD 1 and 2, black-preferred candidates won 50% of elections in 2018, reflecting two three-candidate contests (out of four total elections that year) in which white voters split their vote between the two losing candidates. Id.; see NAACPPX189 at 22.

Defendants’ expert Dr. John Alford did not dispute the reliability of Dr. Oskooii's methodology or the results of his analysis. Trial Tr. Vol. V at 1222:14–1223:1, 1226:12–23. Dr. Alford instead highlighted the degree of crossover voting revealed in Dr. Oskooii's election performance data.¹⁴ For example, analyzing the 49 statewide elections (from 2016 through 2022) in Dr. Oskooii's initial report, Dr. Alford observed that SD 5, with a 40% BVAP, elected the black-preferred candidate in every single election and SD 11, with a 37% BVAP, elected the black-preferred candidate in 65% of elections. LDTX242 at 34. In the same region, CD 1, with a 40% BVAP, elected the black-preferred candidate in 82% of elections. Id. at 35. Further, according to Dr. Oskooii's analysis, Plaintiffs’ Illustrative Districts 2 and 5, with BVAPs of 50.25% and 50.37%, respectively, would elect the black-preferred candidate in every single election, with win margins ranging from nine to thirty-two percentage points. Id. at 34; NAACPPX208 at 21; NAACPPX210 at 7; Trial Tr. Vol. II at 408:12–409:24. As Dr. Alford put it, those are not “equal-opportunity district[s],” they are “guaranteed-performance district[s].” Trial Tr. Vol. V at 1190:12–18.

Dr. Alford also conducted a district effectiveness analysis to determine the proportion of BVAP necessary to provide an equal opportunity to elect black voters’ candidate of choice. See LDTX244 at 13. He estimated the so-called “BVAP needed to win” based on the 2024 elections, using estimates regarding turnout, black support for the black-preferred candidates, and crossover support for those candidates. Id. Applying this methodology, Dr. Alford determined that majority-black districts are not necessary to provide black voters with an equal opportunity to elect their candidates of choice. Id. at 13–14. Across the districts analyzed, the BVAP needed to win is in the high 30% to mid-40% range. Id. at 14. In SD 1 and 2, Dr. Alford estimated a 41% and 42% BVAP needed to win, respectively. Id.

We find that white bloc voting exists in SD 1 and 2 and that, given the BVAP of those districts, bloc voting is sufficient usually to defeat the black-preferred candidate. The evidence clearly shows that, because of white crossover voting, a majority-black district is not necessary to elect black-preferred candidates in this part of the State. Nevertheless, as currently constructed, SD 1 and 2 have BVAPs of 29.49% and 30.01%, respectively, which is not sufficient, even with significant white crossover voting, to prevent the black-preferred candidates in those districts from usually suffering defeat. Thus, we find that Plaintiffs have demonstrated the third Gingles precondition, although the extent of white crossover voting, and the reason for these patterns, will factor into our totality of the circumstances analysis.

4. Totality of the Circumstances

We turn then to consider the “touchstone” of Section 2, Brnovich, 141 S. Ct. at 2338, whether “the totality of circumstances” shows that voting is “not equally open” to minority voters, 52 U.S.C. § 10301(b). Having shown that “it is possible that the State's map has a disparate effect on account of race,” Plaintiffs now must prove “that possibility is reality.” Milligan, 143 S. Ct. at 1507; see also Johnson, 512 U.S. at 1011 (“[C]ourts must ․ examine other evidence in the totality of circumstances,” because Gingles “clearly declined to hold the [three preconditions] sufficient” to “prove a § 2 claim.”); Charleston Cnty., 365 F.3d at 348 (“[C]learing the Gingles hurdles ․ is not sufficient to establish an actual violation.”). The totality inquiry is “ ‘an intensely local appraisal’ ” and “a ‘searching practical evaluation’ ” that is “ ‘peculiarly dependent upon the facts of each case.’ ” Milligan, 143 S. Ct. at 1503 (quoting Gingles, 478 U.S. at 79). We may consider “any circumstance that has a logical bearing on whether voting is ‘equally open’ and affords equal ‘opportunity,’ ” Brnovich, 141 S. Ct. at 2338 (quoting 52 U.S.C. § 10301(b)), including the Senate factors, but we may not treat those factors as “a mechanical ‘point counting’ device,” S. Rep. No. 97-417, at 30.

a. Extent of Racial and Political Polarization

We begin with the extent and cause of racially polarized voting. The second Senate factor evaluates the extent of racially polarized voting. See Gingles, 478 U.S. at 37. “[C]rossover voting patterns and ․ effective crossover districts” can be evidence “of equal political opportunity under the § 2 totality-of-the-circumstances analysis.” Strickland, 556 U.S. at 24. In communities where “minority voters do receive substantial support from white voters,” it “would be exceedingly difficult for plaintiffs to show that they were effectively excluded from fair access to the political process under the results test.” S. Rep. No. 97-417, at 33.

In addition to the extent of polarized voting, “ ‘the reason for polarized voting is a critical factor in the totality analysis,’ including evidence that ‘partisanship [is] the cause of the racially divergent voting.’ ” Pierce, 97 F.4th at 223 (emphasis added) (quoting Charleston Cnty., 365 F.3d at 347, 349). Section 2 concerns dilutive effects “on account of race or color,” 52 U.S.C. § 10301(a), and it is not offended when “minority-preferred candidates have suffered electoral defeat on account of partisan politics, not race or color,” Charleston Cnty., 365 F.3d at 349.

We find these factors highly probative on the facts of this case. The evidence shows significant crossover voting to support black-preferred candidates and that politics, not race, best explains the polarized voting Plaintiffs have identified.

Beginning with crossover voting, the analyses of Plaintiffs’ experts Dr. Oskooii and Dr. Palmer show substantial levels of crossover voting across the State. Dr. Palmer's analysis of statewide voting across the 2016 to 2022 elections shows an average level of white crossover voting of 28.2% and as high as 35%. Trial Tr. Vol. I at 181:7–24; WX4 at 4–5. In the 2024 elections, Dr. Palmer's analysis of statewide voting shows an average level of white crossover voting of 29.6% and as high as 41.7%. Trial Tr. Vol. I at 182:17–183:1; WX6 at 3. These levels of crossover voting are among the highest Dr. Palmer has ever seen in any State. Trial Tr. Vol. I at 183:2–20. Estimates for northeastern North Carolina also show meaningful white crossover voting. Across SD 1, 2, 5, and 11 in the 2024 exogenous elections, data show an average level of white crossover voting around 23% and as high as 40.2%. See LDTX244 at 3.

Dr. Oskooii's performance analysis confirms the extent of white crossover voting in this area. His analysis shows that, because of crossover voting, SD 5, at 40% BVAP, elects the black-preferred candidate in 100% of contests and SD 11, at 37% BVAP, elects the black-preferred candidate in 65% of elections analyzed from 2016 through 2022. See LDTX242 at 34; Trial Tr. Vol. V at 1190:7–11. Adding the 2024 election results does not change this conclusion. SD 5 performed for black-preferred candidates in 100% of contests and SD 11 performed in nearly 60% of elections from 2016 to 2024. See NAACPPX208 at 20. Similarly, CD 1 in this region, at 40% BVAP, elected the black-preferred candidate in 82% of elections. LDTX242 at 35.

Victories for black-preferred candidates in crossover districts illustrate the significance of crossover voting. Plaintiffs’ evidence shows that the black-preferred candidate is uniformly the Democratic candidate. In 2024, Senator Kandie Smith, a black Democrat, won her Senate race in SD 5, which has a BVAP of 40.3%. NAACPPX205 at 14. Congressman Don Davis, also a black Democrat, won CD 1, where the BVAP was 40.4%.¹⁵ Id. at 16. Democrats also won in five state House districts in this region (HD 8, 23, 24, 27, 32), three of which are crossover districts with BVAPs between 37.14% and 43.42%. JX082 at 14; JX367 at 2, 5, 6.

The evidence demonstrates that majority-black districts are not necessary to elect black-preferred Democratic candidates, including black-preferred Democrats who are themselves black. Dr. Alford compiled a table listing the 26 black state House members elected in 2022: 25 black Democrats and one black Republican. LDTX242 at 32. Only three of the black members were elected from majority-black districts, while 17 (including 16 black Democrats) were elected from districts with BVAPs below 40%. LDTX242 at 32; LDTX243 at 2. Five black Democrats were elected from districts where the BVAP is less than 25%. LDTX242 at 32; LDTX243 at 2. This is strong evidence of meaningful crossover voting, showing that minority voters in North Carolina receive substantial support from white voters.

Turning to the cause of racially polarized voting, Defendants presented evidence that the polarization is best explained by partisan motivations, not racial ones. None of Plaintiffs’ experts showed otherwise.

Defendants’ expert Dr. Alford reviewed the evidence provided by Dr. Oskooii and Dr. Palmer and concluded it was “clear evidence of party polarization” rather than racial polarization. Trial Tr. Vol. V at 1151:3–18, 1160:9–13; see also LDTX242 at 36. Dr. Alford hypothesized that, if voters were responding to a candidate's race, we would expect to see black voters demonstrating stronger support for black candidates and white voters showing stronger support for white candidates. LDTX242 at 7. But in practice, Dr. Alford found that voters supported candidates of their preferred political party regardless of their race, with black voters favoring Democrats and white voters backing Republicans. Dr. Alford's analysis shows that voting behavior is not, on the part of either white or black voters, responsive to the race of the candidates in the way it is strongly responsive to the party affiliation of the candidates. Trial Tr. Vol. V at 1160:13–18.

For his initial report, Dr. Alford relied on the data and analyses provided by Dr. Oskooii and Dr. Palmer. See LDTX242 at 6–28. Dr. Alford focused on Dr. Oskooii's analysis of forty-nine statewide contests between 2016 and 2022 in four state Senate districts (SD 1, 2, 5, and 11), Plaintiffs’ Illustrative Districts 2 and 5, twelve state House districts, and five congressional districts (CD 1, 5, 6, 9, and 10), and on Dr. Palmer's analysis of forty-eight statewide contests between 2016 and 2022 in five separate regions. Id. Dr. Alford found that these contests were consistent with a polarized response to party affiliation, but not to the race of the candidate. Id. at 7, 11, 16, 22, 28. In all of these contests, black voters were highly supportive of the Democratic candidates, and white voters typically gave majority support to the Republican candidate. Id. at 7, 11, 16, 22, 28; see, e.g., Trial Tr. Vol. V at 1181:8–19. The race of the candidate, however, did not have a polarizing impact on vote choice. Black voters were no more supportive of black Democrats than they were of white Democrats, and white voters were no more likely to oppose a black Democrat than they were a white Democrat. LDTX242 at 35–36. Across all these years, there was a “pattern of clearly polarized voting by party, but no indication of any significant difference in voter behavior based on the racial status of the contest.” Id. at 7, 28; see also Trial Tr. Vol. V at 1186:5–14.

For example, in the 2022 elections, white voters were on average equally supportive of the black Democratic candidates in SD 1 (23%) and SD 11 (17%) as they were of the white Democratic candidates, who also received an average of 23% and 17% of the white vote, respectively. LDTX242 at 11. In SD 2 and SD 5, white voters were actually more supportive of the black Democratic candidates (at 19% and 26%, respectively) than they were of white Democratic candidates (at 18% and 25%, respectively). Id. White crossover voting for black Democrats was slightly lower in the 2016 contests, but that slight difference does not appear in the 2018, 2020, or 2022 contests. Id. at 7; Trial Tr. Vol. V at 1185:11–19.

Significantly, the pattern was noticeably different in the one race where candidates ran without a party affiliation on the ballot, the 2016 Supreme Court race. Without the partisan designation on the ballot, black and white voters demonstrated less polarized voting. Black support for the black candidate was significantly lower than the typical level of support for either the white or black Democratic candidates in the partisan elections analyzed. Trial Tr. Vol. V at 1183:17–20; LDTX242 at 8. White voters, meanwhile, split their support nearly evenly between the black candidate and the white candidate, with very high levels of white crossover voting for the black candidate. Trial Tr. Vol. V at 1183:20–25; LDTX242 at 8. In fact, a majority of white voters (52%) in SD 1 voted for the black candidate, as did 44% of white voters in SD 2. LDTX242 at 8.

These results show that the race of the candidate has no impact beyond the impact of party, particularly for white voters. If party were not driving voting behavior, then the results of the 2016 Supreme Court race should be the same as those in partisan contests. But if party is the driving factor and it is removed, different voting behavior would result, and that is what occurred in the 2016 Supreme Court contest. Without a party cue, the same levels of cohesion are not present, especially among white voters, as in contests where party affiliation is on the ballot and where voting is more polarized. Trial Tr. Vol. V at 1184:1–8, 1184:16–22. This election is probative evidence that divergence in candidate preference by race is a partisan phenomenon, not a racial phenomenon.

Dr. Alford's analysis of congressional districts (CD 1, 5, 6, 9, and 10) showed the same pattern of partisan polarization rather than racial polarization. LDTX242 at 22–26; Trial Tr. Vol. V at 1186:23–1187:3. Black voters were highly supportive of Democratic candidates, and white voters were supportive of Republican candidates, regardless of the candidate's race. LDTX242 at 23–26. For example, levels of white crossover voting for Democratic candidates were roughly equal in contests where the Democratic candidate was white as in contests where the Democratic candidate was black. Id. at 23–26. And in 2022, white voters in CD 5 and CD 6 were more supportive of black Democratic candidates (at 26% and 22%, respectively) than they were of white Democratic candidates (at 25% and 21%, respectively). LDTX242 at 26.

The 2024 elections confirmed Dr. Alford's conclusion that the elections demonstrated a pattern of partisan, rather than racial, polarization. Across all the 2024 elections, black voters were no more supportive of the black Democrat than they were of the white Democrat in any areas, and white voters were no more likely to oppose a black Democrat than they were a white Democrat. LDTX251 at 7. In the state Senate districts, black voters in all fourteen statewide contests consistently gave high levels of support to the Democratic candidate and white voters gave high levels of support to the Republican candidate. LDTX244 at 2–3. This is consistent with a polarized response based on party affiliation of the candidate, whereas the race of the candidates did not appear to have a polarizing impact on voters’ behavior. Id. at 3. In all four Senate districts studied, less than one percentage point separated black voters’ support for black Democratic candidates versus white Democratic candidates and white voters’ support for white Democratic candidates versus black Democratic candidates. LDTX244 at 3; LDTX251 at 3; Trial Tr. Vol. V at 1192:2–14. The same is true for Dr. Alford's analysis of the congressional districts analyzed by Dr. Oskooii and Dr. Palmer, except that in Dr. Palmer's CD 14 region, black voters were three percentage points more favorable to white Democratic candidates. LDTX244 at 8–9; LDTX251 at 4, 6; Trial Tr. Vol. V at 1192:16–18.

Plaintiffs’ experts did not rebut Dr. Alford's opinions on this point or provide any empirical refutation to his analysis. Dr. Oskooii noted the limits of Dr. Alford's analysis for proving causation or disentangling race from any individual's party preference.¹⁶ Trial Tr. Vol. II at 393:2–14. Even accepting those limits, we find Dr. Alford's analysis well supported and persuasive.

Plaintiffs argue that the 2024 gubernatorial race reveals a shortcoming in Dr. Alford's analysis. In that election, white Republicans gave considerably less support to Mark Robinson, a black Republican candidate for governor, than they did to white Republicans on average in other 2024 statewide contests. See LDTX244 at 3, 5, 8. We do not find this single election disruptive to Dr. Alford's analysis. Plaintiffs fail to account for the highly publicized implosion of Robinson's campaign after a string of scandalous allegations, including allegations of sexual misconduct and disturbing rhetoric. As Dr. Alford explains, these accusations and Robinson's campaign meltdown were not “obscure ‘insider politics’ event[s].” LDTX244 at 12. They were “a highly visible news story covered and commented on extensively” in North Carolina and “even in the national media.” Id. Dr. Alford opines that the “highly publicized downturn” in Robinson's campaign was a “special circumstance that makes this election less useful when evaluating voting patterns in North Carolina politics.” Id.

We agree with Dr. Alford. North Carolina voters knew Robinson was black when they elected him to be Lieutenant Governor in 2020, and he received white voter support in that election on par with the average white support for white Republican candidates that year. Id. at 10; see LDTX242 at 10. And in the 2024 primary, Republicans nominated Robinson, with nearly 65% of the vote, over two white Republican candidates. LDTX244 at 10. The drop in support for Robinson after salacious allegations surfaced and his campaign publicly collapsed does not cause us to doubt Dr. Alford's analysis or conclusions.

On the whole, the evidence in this case demonstrates polarization on the basis of political party, not polarization on the basis of race. Plaintiffs have not proven the existence of white bloc voting against black candidates or against the candidates black voters prefer, as opposed to Republican bloc voting against Democrats of all stripes. White Republicans vote Republican, and white Democrats vote Democrat, in what the racial lens of Section 2 jurisprudence calls “crossover voting for the black-preferred candidate.” White and black Democrats are similarly situated with regard to their ability to influence elections—both will be able to elect their candidates of choice if they are in majority-Democratic districts, but will be unable to do so in majority-Republican districts. Trial Tr. Vol. V at 1151:13–17. The evidence convincingly shows that North Carolina, including the northeastern part of the State, is not a place “where racial politics ․ dominate the electoral process,” S. Rep. No. 97-417, at 33, even though the partisan preferences of racial groups “diverge,” Gingles, 478 U.S. at 100 (O'Connor, concurring in the judgment).

b. Electoral Success of Black Candidates

Another factor we find significant is the extent to which minority group members have been elected to public office, the seventh Senate factor. See Gingles, 478 U.S. at 37; see also id. at 40, 74–75 (assessing elections both statewide and in the challenged districts). Section 2 expressly authorizes considering this factor. 52 U.S.C. § 10301(b). If “no members of a minority group” or only “a few” have been elected to office “over an extended period of time,” that fact is probative. S. Rep. No. 97-417, at 29 n.115. But Congress has specifically instructed that Section 2 does not “establish[ ] a right to have members of a protected class elected in numbers equal to their proportion in the population.” 52 U.S.C. § 10301(b).

In the 2024 general election, North Carolina voters elected 28 black representatives out of 120 (23.3%) and 10 black senators out of 50 (20%) to the General Assembly. Those 38 black members constitute 22.4% of that 170-member body. WX11 at 6. North Carolina's population is 22.46% black and its voting-age population is 21.37% black. JX149 at 8, 14. Thus, in the words of Plaintiffs’ expert Dr. Chris Clark, the black representation ratio for the North Carolina General Assembly is “above parity.” Trial Tr. Vol. II at 284:3–18. The current seat share and representation ratio is a dramatic increase from the four black state legislators in office in 1981 and even the nineteen black state legislators serving a decade later in 1991. JX370 at 2, 4–5.

Most of these black senators and representatives were elected from districts that are not majority black. Only one of the ten Senate districts and only four of the 28 House districts that elected black members to the General Assembly have a majority-black voting age population. JX149 at 13–14; JX082 at 21–23; JX372. The other nine senators and 24 representatives were elected from districts that are not majority black and that have a wide range of racial compositions.¹⁷

Black members of the General Assembly hold important leadership roles in their respective chambers. Representative Robert Reives is the House Democratic Leader, and Senator Sydney Batch is the Senate Democratic Leader. Before Senator Batch, Senator Dan Blue was the Senate Democratic Leader from 2014 to 2024. And from 1991 to 1994, Senator Blue was Speaker of the House.

Looking at the northeastern area of the State that is the focus of this claim, black candidates in 2024 won elections in SD 5, HD 8, HD 23, HD 24, and HD 27. NAACPPX205 at 14–15. This was not an aberration. Since the 1980s, voters in this area have consistently elected black representatives and senators to the General Assembly. See JX370.

Turning to North Carolina's representation in the U.S. Congress, there are currently three black members out of fourteen in the State's congressional delegation, or 21.4%. Dr. Clark calculated a 0.97 black representation ratio for North Carolina's congressional delegation, just short of parity with the State's black population. WX8 at 38. Using the figure for the State's black voting-age population—21.37%—parity would be achieved. See Trial Tr. Vol. II at 263:16–22. Each of those members won reelection in 2024. Congresswoman Alma Adams, a black Democrat, currently represents CD 12; she was first elected in 2014. JX369 at 3–4; LDTX259 at 30; LDTX261 at 19. Before Congresswoman Adams, Melvin L. Watt, another black Democrat, served CD 12 from 1993 until 2014. In CD 4, Congresswoman Valerie Foushee, a black Democrat, retained her seat in the 2024 election; she was first elected in 2022. JX369 at 2; Trial Tr. Vol. II at 275:8–11.

Congressman Don Davis, a black Democrat, currently represents CD 1, located in northeastern North Carolina. JX369 at 2; LDTX259 at 30. Congressman Davis was first elected in 2022 and was successfully re-elected in 2024. Before 2022, G.K. Butterfield, another black Democrat, had represented CD 1 from 2004 to 2022. LDTX261 at 23. In fact, CD 1 has been represented by a black congressperson since 1992. LDTX261 at 20, 23, 25.

Black North Carolinians have also been elected to local offices throughout the State. Plaintiffs’ expert Dr. Joseph Bagley reports, based on a website, that there are 581 black elected officials at the local level. NAACPPX181 at 47. That includes many in the State's northeastern counties. Id. And several Plaintiffs and black fact witnesses have held or currently hold public office in the State. For example, Courtney Patterson has served on the Board of Elections for Lenoir County for fourteen years and is the current chair of the Board. Trial Tr. Vol. III at 592:5–10. Plaintiff Earl Jones served on the Greensboro City Council for eighteen years and was the North Carolina State House District 60 Representative for eight years. Trial Tr. Vol. I at 135:21–24, 143:16–17. Plaintiff Calvin Jones served on the Warren County School Board for over fifteen years. Trial Tr. Vol. II at 349:24–350:3.

As for statewide positions in North Carolina's executive and judicial branches, black North Carolinians have also had some success. Seven black jurists have served on the Supreme Court of North Carolina, including current Justice Anita Earls. WX8 at 34; Trial Tr. Vol. II at 271:21–272:16. Black jurists have also served on the North Carolina Court of Appeals. In the executive branch, Ralph Campbell was elected State Auditor in 1992 and served until 2005. NAACPPX181 at 45. North Carolinians elected a black lieutenant governor, Mark Robinson, in 2020. And in 2024, North Carolina elected a black superintendent of public instruction, Maurice Greene. Plaintiffs point out that North Carolina is not one of the five States to have ever elected a black governor nor one of the ten States that have elected the only fourteen black Americans ever to serve in the U.S. Senate.

Black candidates are regularly elected in North Carolina, including in the northeastern part of the State. Although it is not the benchmark for Section 2, North Carolina has even achieved proportionality in the General Assembly and its U.S. congressional delegation. These facts undercut Plaintiffs’ claim of unequal opportunity.

c. Racial Appeals in Political Campaigns

The sixth Senate factor asks whether political campaigns “ ‘have been characterized by overt or subtle racial appeals.’ ” Gingles, 478 U.S. at 37 (quoting S. Rep. No. 97-417, at 29).

Plaintiffs offered testimony from Drs. Clark and Bagley. At the threshold, their methodologies strike us as somewhat rudimentary and outside their areas of expertise. Dr. Clark conducted “targeted searches” in the Raleigh-based News and Observer (“N&O”) newspaper. WX8 at 21. He did not log the search terms used or report the full results of his searches. Trial Tr. Vol. II at 290:23–25, 291:8–10. And, perplexingly, Dr. Clark explained his decision not to consider any other newspapers by stating, “I figured -- I don't know if [N&O] has the largest circulation, but it's a large paper, and they cover statewide politics.” Id. at 290:18–22. He also watched Senator Ted Budd's campaign advertisements and debate on YouTube. Dr. Bagley, for his part, cited various news articles but did not record or explain his methodology. He did not apply a consistent set of search terms or parameters across an identified set of databases or otherwise conduct a search that could be replicated. Trial Tr. Vol. III at 695:11–696:1. The experts’ ad hoc approaches do not facilitate an assessment of whether racial appeals occur regularly or otherwise “ ‘characterize[ ]’ ” North Carolina political campaigns. Gingles, 478 U.S. at 37 (quoting S. Rep. No. 97-417, at 29). Nevertheless, we have carefully considered the examples they offer.

Dr. Clark opined that the use of overt racial appeals in North Carolina politics dated back to the late nineteenth century, citing an example from an 1898 N&O article and one from a 1950 campaign for U.S. Senate. WX8 at 22. Both Drs. Clark and Bagley mentioned the well-known “hands” advertisement run by former Senator Jesse Helms in his 1990 campaign against his black Democratic opponent, warning about the effect of “racial quotas.” See WX177.

Moving into the 21st century, Dr. Clark cited a 2010 advertisement used during white Democrat Hugh Holliman's campaign for the North Carolina House of Representatives. WX8 at 24. The state Republican Party mailed a flier with an image of two death row inmates—one black man and one white man. See WX178. The mailer criticized Holliman's support of the North Carolina Racial Justice Act, warning that death row inmates “could leave prison early and move in next door” because of the law. Id. Despite the prominent featuring of both a black and a white inmate, Dr. Clark described the advertisement as “play[ing] on stereotypes” and “on the idea of Black criminality.” Trial Tr. Vol. II at 255:15–20. We do not find his opinion persuasive.

Considering the same time period, Dr. Bagley cited an advertisement run by the North Carolina Republican Party in 2008 opposing two white candidates in the Democratic primary for governor. NAACPPX181 at 39. According to Dr. Bagley's description of the video, it featured images of presidential candidate Barack Obama with Reverend Jeremiah Wright in “vaguely African garb” and “reminded viewers that Wright was Obama's spiritual mentor.” Id. Then a clip of Wright cursing America played and the narrator stated that the North Carolina candidates endorsed then-candidate Obama, remarking, “They should know better. He's just too extreme for North Carolina.” Id. We doubt Dr. Bagley's characterization of this advertisement as a racial appeal, but even if we were to accept his classification, it would not materially alter our evaluation of this factor.

We acknowledge Dr. Bagley's more recent examples of racial appeals in North Carolina but find them largely unpersuasive. See generally id. at 40–43. Dr. Bagley noted that Russell Walker, a 2018 candidate for the General Assembly, published a website with comments supporting white supremacy. Id. at 40. As Dr. Bagley confirmed at trial, the North Carolina Republican Party withdrew its support for Walker and he lost the general election by a margin of 25 points. Trial Tr. Vol. III at 698:23–700:6. Dr. Bagley also referenced social media posts by Republican Party executive Dallas Woodhouse, who Dr. Bagley incorrectly claimed ran for the state Supreme Court in 2018.¹⁸ According to Dr. Bagley's report, Woodhouse criticized Democratic candidate Justice Anita Earls for supporting the Racial Justice Act, claiming she viewed it as “racist” to prosecute certain death row inmates. NAACPPX181 at 40. Dr. Bagley confirmed at trial that Woodhouse did not run for office and Justice Earls won that race. Trial Tr. Vol. III at 700:11–24. In a similar vein, Dr. Bagley cited a 2019 speech from North Carolina gubernatorial candidate Dan Forest, in which he warned about “diversity and multiculturalism” caused by “a lack of assimilation” and “identity politics” that divide the Nation. NAACPPX181 at 40. Forest also lost his election. Trial Tr. Vol. III at 701:11–12. And Dr. Bagley quoted op-eds reposted on social media in 2024 by Michele Morrow, a candidate for superintendent of public instruction, criticizing “DEI” and “race based discipline.” NAACPPX205 at 7–8; Trial Tr. Vol. III at 704:3–14. Morrow, a white Republican, lost that election to Maurice Green, a black Democrat.¹⁹ Trial Tr. Vol. III at 704:19–20.

After the 2024 election, Dr. Bagley supplemented his report to discuss racial comments made by black Republican Mark Robinson, which surfaced during his gubernatorial campaign. See NAACPPX205 at 9–10. As Dr. Bagley confirmed at trial, however, Robinson did not make those comments during his 2024 campaign and he was not the person who brought those prior comments to light. Trial Tr. Vol. III at 705:2–15. In fact, as Dr. Bagley correctly observes, Robinson's comments were used “against the candidate,” NAACPPX205 at 9, and he lost the election, Trial Tr. Vol. III at 706:1–2. This is not an example of a candidate using a racial appeal in an effort to win votes; quite the opposite, it is an instance of a candidate's former racial comments coming back to hurt him with North Carolina voters.

Both experts analyzed white Republican Ted Budd's 2022 campaign for United States Senate against black Democrat Cheri Beasley, the former Chief Justice of the North Carolina Supreme Court. The experts opined that Budd's campaign made implicit racial appeals by painting Beasley as soft on crime because “Black candidates being soft on crime is [a] racial trope.” NAACPPX181 at 41. They cited an advertisement featuring the brother of a white state trooper killed by a black man who Beasley represented when she was a public defender. See WX8 at 26; NAACPPX181 at 40–41. Another advertisement labeled Beasley as “dangerously liberal” because of a judicial decision rendered when she was a Supreme Court justice and featured a mug shot of the black offender from that case. NAACPPX490; see Trial Tr. Vol. IV at 750:3–24. A third advertisement discussed by Plaintiffs’ experts switched between photos of children and photos of Beasley while the text and narration recounted that Beasley's judicial rulings had released sexual predators. NAACPPX489. Dr. Bagley describes this ad as featuring only “white children,” NAACPPX181 at 41, but we have viewed it and can plainly see it features children of all races, see NAACPPX489. Also on the topic of crime, Dr. Bagley quoted Budd as saying Beasley had been “encouraging [riots] all across our state and all across our country in 2020” and noting that he had been endorsed by the North Carolina State Troopers Association. NAACPPX181 at 41. Similarly, Dr. Clark opined that Budd made racial appeals by campaigning on the issues of immigration and crime. WX8 at 26–27; see also WX181.

Plaintiffs’ expert Dr. LaFleur Stephens-Dougan explained this central premise of Plaintiffs’ presentation: that references to legitimate policy issues are actually “racial dog whistles.” NAACPPX195 at 19–30. To begin, drawing largely on outside scholarship, Dr. Stephens-Dougan assessed racial attitudes and politics in the “American South.” Id. at 2. Dr. Stephens-Dougan had not previously studied race and partisanship in North Carolina. Trial Tr. Vol. IV at 759:11–14. Her analysis for this case relied on aggregated data from a sample of eleven “southern States,” which she acknowledged she could not disaggregate to reach “meaningful conclusions” about North Carolina. NAACPPX195 at 5; Trial Tr. Vol. IV at 756:13–758:24. Dr. Stephens-Dougan asserted that “on average Southern whites are racially resentful.” NAACPPX195 at 6. She based this conclusion on a political survey known as the “racial resentment scale,” although on cross-examination she acknowledged debates about whether that survey actually measures principled conservative views rather than racial attitudes. Trial Tr. Vol. IV at 780:5–9.

Building from this premise, Dr. Stephens-Dougan opined that “racially coded language” is used in American politics “to activate or prime voters’ racial predispositions.” NAACPPX195 at 21. And she claimed that, in turn, attitudes toward “racialized issues” drove support for either the Democratic or Republican party. Id. at 2–3, 40; see Trial Tr. Vol. IV at 743:21–744:2. Dr. Stephens-Dougan identified the following as racialized issues: crime, illegal immigration, welfare, Medicaid, affordable housing, affirmative action, the death penalty, and diversity, equity, and inclusion (DEI) programs.²⁰ NAACPPX195 at 21; Trial Tr. Vol. IV at 766:24–768:4. She explained that campaign references to these political issues could therefore operate as implicit racial appeals, or “racial dog whistles,” by activating negative racial predispositions even without a person's knowledge. Trial Tr. Vol. IV at 763:22–765:8. For example, Dr. Stephens-Dougan testified that “use of the phrase ‘illegal immigration’ ” is “a racial dog whistle.” Id. at 764:4–7.

At trial, however, Dr. Stephens-Dougan's framework proved unworkable. When asked whether there was “a way for politicians to discuss the issues of crime and Medicaid without being accused of racial appeals,” Dr. Stephens-Dougan replied, “[y]es,” that a politician could make “a positive racial appeal” like “call[ing] for Medicaid for all, for example.” Trial Tr. Vol. IV at 794:14–23. That contradictory response reveals that, in her opinion, many legitimate political issues are always racial appeals, either positive or negative. We reject that approach, which would render this Senate factor useless. Accordingly, we decline to classify the race-neutral policy positions in the record as probative evidence that the State's campaigns are generally characterized by implicit racial appeals.

We also cannot accept Dr. Clark's determination that Budd's alignment with President Donald Trump was a racial appeal by extension. In his report, Dr. Clark stated that “President Trump has made many racially insensitive remarks in recent years” and reasoned that “a candidate in North Carolina purposefully aligning with Trump is, by extension, a racial appeal.” WX8 at 29. Yet at trial Dr. Clark declared that he was “not at all” saying that “anytime someone aligns with Donald Trump it is, by definition, a racial appeal.” Trial Tr. Vol. II at 261:18–21, 299:10–15. When confronted with this contradiction on cross-examination, Dr. Clark declined to retract either statement. Id. at 299:10–301:2. We find his competing testimony difficult to reconcile and give it no weight. As Defendants’ expert Dr. Taylor testified, and Dr. Clark admitted, it is a common practice for candidates to affiliate with the candidate at the top of their party's ticket, particularly when that candidate enjoys nationwide recognition. Id. at 301:13–16; Trial Tr. Vol. VI at 1286:8–11.

Relatedly, we do not find remarks made by President Trump, a national figure from a distant State, probative on the question of racial appeals in North Carolina campaigns. For example, Dr. Clark cited a comment by President Trump at a campaign event in North Carolina with Budd and other candidates, but there is no evidence that Budd adopted or aligned himself with that remark. WX8 at 30–32. We find these comments to be more nationally focused with limited import in evaluating North Carolina campaigning.

On cross-examination, Dr. Bagley confirmed that his reports did not identify any racial appeals in North Carolina campaigns for the state Senate or federal House of Representatives, the offices at issue in this case, in any year. Trial Tr. Vol. III at 696:23–697:4. Nor did he identify any racial appeals for any North Carolina campaigns in 2012, 2014, or 2016. Id. at 696:11–19. He also confirmed that, over the sixteen years he analyzed, thousands of election contests occurred in North Carolina. Id. at 697:14–698:14. This absence of evidence is probative.

On the whole, we find that racial appeals do not characterize North Carolina political campaigns. Plaintiffs have presented evidence of some racial appeals, including in recent years. Yet those isolated incidents do not characterize North Carolina politics, and many of the recent appeals Plaintiffs cite were not successful with the State's voters. The climate of North Carolina political campaigns does not support Plaintiffs’ claim of dilution.

d. History of Official Discrimination

The first Senate factor concerns “ ‘the extent of any history of official discrimination in the state or political subdivision that touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process.’ ” Gingles, 478 U.S. at 36–37 (quoting S. Rep. No. 97-417, at 28). We must consider history in the totality of circumstances in order to evaluate whether the boundaries of SD 1 and 2 “ ‘interact[ ] with social and historical conditions to cause an inequality in the opportunities enjoyed by black and white voters.’ ” Milligan, 143 S. Ct. at 1503 (quoting Gingles, 478 U.S. at 47). In making this assessment, we agree with the parties that recent history is more probative for understanding whether the 2023 Senate districts are discriminatory in effect. See Pierce, 97 F.4th at 221.

This Court is well aware that “North Carolina has a long history of race discrimination generally and race-based vote suppression in particular.” McCrory, 831 F.3d at 223. Plaintiffs offered expert testimony from Dr. James Leloudis and Dr. Bagley, both of whom provided reports on North Carolina's voting rights history. Dr. Leloudis traced the State's history beginning with the Civil War and Reconstruction. See WX13 at 8–16. His report documented North Carolina's use of poll taxes, literacy tests, and similar measures restricting the franchise during the nineteenth and early twentieth centuries. See id. at 25–31. Both experts recount the use of at-large electoral schemes and anti-single-shot laws in elections around the middle of the twentieth century. See id. at 37–62; NAACPPX181 at 9. When Congress enacted the Voting Rights Act in 1965, thirty-nine of North Carolina's one hundred counties were covered jurisdictions, meaning that any changes to their electoral systems required preclearance from the Department of Justice. See Shelby Cnty. v. Holder, 570 U.S. 529, 537 (2013) (citing 28 C.F.R. pt. 51 app. (2012)). Dr. Bagley represents that, from 1970 through 2012, the Department of Justice entered sixty objections to proposed electoral changes in North Carolina, including twenty-six objections concerning counties or municipalities in eastern North Carolina. NAACPPX181 at 11–27; see id. at 13. Around the same time, private plaintiffs successfully challenged at-large schemes and single-member systems for town and city councils, county commissions, and school boards. Id. at 12; see also McCrory, 831 F.3d at 224. We accept the experts’ findings that the State has a long and cyclical history of minority voter suppression over the nineteenth and twentieth centuries.

As for this century, Plaintiffs have identified one case in which a court found that the General Assembly acted with discriminatory intent.²¹ In 2013, the General Assembly passed an omnibus election law that restricted or eliminated certain voting mechanisms and included a voter-identification requirement. The Fourth Circuit struck down the law as racially motivated, though it expressly disclaimed the suggestion that “any member of the General Assembly harbored racial hatred or animosity toward any minority group.” McCrory, 831 F.3d at 233; see id. at 223 (acknowledging “the obvious political dynamics”). Drs. Leloudis and Bagley reference Senate Bill 824, a subsequent law mandating voter identification, which the General Assembly enacted in 2018. See WX13 at 83; NAACPPX181 at 27. But the Fourth Circuit and the North Carolina Supreme Court have both reversed findings that the State enacted Senate Bill 824 for discriminatory purposes. See Raymond, 981 F.3d at 311; Holmes v. Moore, 886 S.E.2d 120, 144 (N.C. 2023). We do not accept the attempts of Plaintiffs’ experts to impugn those rulings.

Plaintiffs also cite the litigation following the General Assembly's decision to draw numerous majority-minority districts in the 2011 districting plan. As we previously discussed, courts ruled that the legislature's use of race to draw those districts, both legislative and congressional, violated the Equal Protection Clause. See Covington, 316 F.R.D. at 124, sum. aff'd, 581 U.S. 1015; Harris, 159 F. Supp. 3d at 604, aff'd sub. nom., 581 U.S. at 290. But no court found that the General Assembly “acted in bad faith or with discriminatory intent” in drawing the majority-minority districts.²² Covington, 316 F.R.D. at 124 n.1; see Harris, 159 F. Supp. 3d at 604–605. And, after those rulings, the General Assembly adopted a policy of not considering race in redistricting, marking an important break with the past.

e. Current Voting Practices

Another relevant factor in our analysis is whether any current voting practices or procedures “ ‘enhance the opportunity for discrimination against the minority group.’ ” Gingles, 478 U.S. at 37 (quoting S. Rep. No. 97-417, at 29); see Pierce, 97 F.4th at 220–221. Plaintiffs have not made such a showing.

North Carolina uses typical voting practices and procedures. The State does not employ “ ‘unusually large election districts, majority vote requirements, [or] anti-single shot provisions,’ ” the practices identified in the third Senate factor.²³ Gingles, 478 U.S. at 37 (quoting S. Rep. No. 97-417, at 29). The General Assembly does not have cumulative voting or at-large legislative districts. Trial Tr. Vol. III at 681:25–682:4; Trial Tr. Vol. VI at 1267:15–25.

Plaintiffs introduced evidence about the operation of North Carolina's voter ID requirement. Courtney Patterson, Vice President of Plaintiff North Carolina NAACP, testified about one person who he thought was turned away from the polls incorrectly on voter ID grounds. Trial Tr. Vol. III at 593:10–596:19. Plaintiffs’ expert Dr. Bagley asserted that, in the 2024 general election, black voters were more likely to have their ballots rejected than white voters because of the ID requirement. See NAACPPX205 at 3. The data on which Dr. Bagley relied, however, shows that over 99.9995% of black voters and over 99.9998% of white voters had their ballots counted in the 2024 general election. See Trial Tr. Vol. III at 685:12–686:9. We reject Dr. Bagley's opinion that a ballot acceptance rate of over 99.99% for both groups nevertheless demonstrates a “significant” “disparate impact” on black voters. Id. at 686:10–19. Further, Dr. Bagley's speculation that the voter ID law may have dissuaded some black citizens from going to the polls is not based on any analysis or study of actual facts and we do not credit it. Id. at 688:9–689:7.

Regarding measures to make voting convenient and accessible, Defendants’ expert Dr. Andrew Taylor reported that, according to the “cost of voting index,” North Carolina ranked as the 24th easiest State in which to vote in 2022, tied with Rhode Island. LDTX259 at 8. North Carolina has early voting, N.C. Gen. Stat. § 163-166.40, same-day registration during early voting, id. § 163-82.6B, pre-registration, id. § 163-82.6, online voter registration, id. §§ 163-82.6, 163-82.19, no excuse absentee voting, id. § 163-226, and curbside voting, Trial Tr. Vol. II at 312:25–313:6. North Carolina also generally allows citizens to register to vote up until 25 days before an election, giving its citizens five more days to register to vote than federal law requires. See N.C. Gen. Stat. § 163-82.6(d)(1)–(3) (allowing North Carolinians to register to vote up until 25 days before an election); 52 U.S.C. § 20507(a)(1)(A)–(D) (requiring States to hold voter registration for federal elections open until at least 30 days before the election).

Dr. Taylor also analyzed election day polling places for the 2022 general election and early voting locations for the 2022 and 2024 general elections, concluding that counties in the area that Plaintiffs consider the Black Belt had a larger share of polling places than their voting age population. For example, in the 2022 general election, the North Carolina Board of Elections placed 5.05% of its polling places in Black Belt counties, even though those counties constitute only 3.61% of the State's voting age population. LDTX259 at 11. Sixteen of the twenty-two early voting sites in those counties for the 2022 general election were in census tracts where the black population exceeded the black population in the county as a whole. Id. at 12–13. And for the 2024 general election, the Board of Elections placed 6.24% of its early voting sites in those same northeastern counties, far exceeding their 3.61% share of the State's voting age population. Id. at 12.

Plaintiffs did not cast any doubt on this evidence. Plaintiff Mitzi Reynolds Turner testified that a “lack of resources” makes voting less accessible for people in her community, but the anecdotal hindrances she identified occur in poor or rural communities of all races. Trial Tr. Vol. III at 579:24–581:11. Dr. Bagley did not review the number of polling places or early voting sites in North Carolina. Id. at 682:10–13. He relied solely on a 2018 report that he later admitted does not reflect the current status of early voting sites in the State. NAACPPX199 at 6; see Trial Tr. Vol. III at 682:20–683:1. He did not analyze the effect of the availability of early voting or absentee voting on black voters in any of his reports. Trial Tr. Vol. III at 684:10–13.

Based on this evidence, we find that North Carolina does not employ voting practices or procedures that enhance the opportunity for discrimination. To the contrary, the State has undertaken many measures to facilitate minority voting, including specifically in the northeastern counties relevant to our analysis of SD 1 and 2.

f. Effects of Discrimination that Hinder Political Participation

The fifth Senate factor addresses the extent to which minority group members “ ‘bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process.’ ” Gingles, 478 U.S. at 37 (quoting S. Rep. No. 97-417, at 29). According to the Senate Report from which the Supreme Court gleaned this factor, “disproportionate educational[,] employment, income level and living conditions arising from past discrimination tend to depress minority political participation,” and when plaintiffs demonstrate these conditions they “need not prove any further causal nexus between their disparate socio-economic status and the depressed level of political participation.” S. Rep. No. 97-417, at 29 n.114. Plaintiffs have not endeavored to prove a causal link between past discrimination and the socioeconomic conditions they identify; they rely instead on the differences between averages for white and black North Carolinians to infer a connection. The parties have presented evidence regarding disparities between races within the State, as well as disparities over time and in relation to national averages.

Education: Plaintiffs’ expert Dr. Clark identified disparities between black and white North Carolinians using data from the U.S. Census Bureau's American Community Survey (ACS) estimates.²⁴ See WX8 at 5 & n.1. Dr. Clark presented one-year estimates for individuals 25 years and older from 2010 to 2022, excluding 2020 due to data-collection limitations. Beginning with educational outcomes, Dr. Clark found that, in 2022, 89.1% of North Carolina's black population had at least a high school diploma, compared to 93.1% of the white population and 90.2% of the total population. Id. at 7. The gap between black and white North Carolinians has decreased over time.

Looking at the same data, Defendants’ expert Dr. Taylor calculated the proportion of the black population with a high school diploma as a proportion of the white population with the same. See LDTX259 at 16–18; Trial Tr. Vol. VI at 1269:10–24. He then compared that figure to the nationwide equivalent. Dr. Taylor found that the proportion of black North Carolinians over 25 with a high school diploma in 2022 was “very close to the level of whites” at a ratio of 0.957, or over 95%. LDTX259 at 18. This ratio surpassed the nationwide average. Id.

To get a better sense of North Carolina high school graduates in particular, Dr. Taylor analyzed high school graduation rates from the North Carolina Department of Public Instruction. Id. at 19. That data showed that black and white students in the State's northeastern counties have graduated high school at very similar rates for over a decade. Id. at 20.

Dr. Clark found a larger disparity for college degrees, with 25.6% of black individuals over 25 having a bachelor's or more, compared to 39.8% of the white population and a relatively consistent gap over time. WX8 at 7–8. Using different ACS data, Dr. Bagley reported that 23% of black North Carolinians hold a bachelor's degree or higher, compared to 36% of white individuals.²⁵ NAACPPX181 at 32. Dr. Taylor also acknowledged the disparity but noted that the black college graduate rate as a proportion of the white college graduate rate was on par with the national average. LDTX259 at 20–21.

Employment and Income: To evaluate economic outcomes, Dr. Clark considered unemployment rates, average incomes, and the percentage of the population below the poverty level. In 2022, 6% of black North Carolinians were unemployed, compared to 2.9% of white individuals and 3.8% of the State's total population. WX8 at 9–10. The disparity decreased from 2010 to 2022. By contrast, the difference in median household income between black and white households has become more pronounced over time. By 2022, black households earned a median $50,059 per household, while white households made a median of $75,197 and the total population median was $67,481. Id. at 12. In the same year, 18.7% of black North Carolinians lived below the poverty line, compared to 9.6% of white individuals and 12.8% of the total population. Id. at 13. Using different ACS data, Dr. Bagley similarly observed that black individuals are twice as likely to experience poverty as white individuals. NAACPPX181 at 32.

Dr. Taylor's figures confirmed similar disparities. According to his calculations, black unemployment, as a proportion of white unemployment, increased from 2010 to 2022. LDTX259 at 23. The black poverty rate as a proportion of the white poverty rate decreased over time but remained elevated. Id. at 25. And the black median household income was only 66.6% of the white median household income, although Dr. Taylor noted that this proportion was better than the national average. Id. at 23–24.

Health: Turning to health outcomes, Dr. Clark concluded that more black individuals over 18 years old and, separately, over 65, were disabled in 2022 (17.6% and 35.9%) compared to their white counterparts (16.9% and 32.7%) and the total population (16.3% and 33.3%). WX8 at 14–15. On both metrics, the gap decreased over time. Black North Carolinians were also more likely to be uninsured than white individuals, at a rate of 9.1% compared to 6.6%, but slightly less likely to be uninsured compared to the total population (9.3%).²⁶ Id. at 15–16. That gap also narrowed over time. Dr. Taylor confirmed the insurance disparity but noted that the black uninsurance rate was lower in North Carolina than in the United States for each year analyzed. LDTX259 at 26–27. He did not evaluate disability rates other than to point out the improvement in black disability figures over time, both in an absolute sense and by comparison to the white disability rate.

Political Participation: Turning last to political participation, Dr. Clark presented voter turnout data as a percentage of registered voters. WX8 at 19. He represented that, on average, over elections from 2010 to 2022, black voter turnout was 7.1 percentage points lower than white voter turnout (5.3 points lower in presidential election years) and 4.1 percentage points lower (3.2 points lower in presidential election years) than the statewide average for all registered voters. Id. at 19–20. Dr. Bagley also presented turnout figures as a percentage of registered voters. He represented that, in the 2024 elections, black voter turnout was 12.2 percentage points lower than white voter turnout statewide. NAACPPX205 at 4–5. In four Black Belt counties, the turnout gap was less than the statewide figure, and in eight it was greater. Id.

Dr. Taylor evaluated voter turnout compared to the State's black and white voting age populations, not only registered voters of each race. LDTX259 at 10. Looking at those figures, he found that the black turnout rate exceeded the white turnout rate in 2014 and 2016, remained very close in 2018 and 2020, and dropped to 83.6% of the white rate in 2022. Id. at 10–11. In every year, the black rate as a proportion of the white rate was greater in North Carolina than it was nationally. Id.

In Sum: Plaintiffs have demonstrated disparities between black and white North Carolinians in education, employment, and health. Although those disparities generally have improved over time (except for income), lingering differences remain in most areas, with the exception of the consistently similar white and black graduation rates in the northeastern counties.

We are less confident that Plaintiffs have shown depressed political participation. Dr. Taylor's measure of voter turnout as a percentage of the voting-age population appears to provide a more accurate reflection of participation among black and white individuals who are eligible to vote than does Dr. Clark's and Dr. Bagley's measure of voter turnout as a percentage of registered voters of each race. Dr. Taylor's evidence shows that black voter turnout was better than or on par with white voter turnout for four of the five election cycles analyzed (2014 through 2022). That is not depressed turnout. Plaintiffs do not call that evidence into question or offer any explanation for how it can be reconciled with their claim of reduced political participation.²⁷ Nor has any party attempted to reconcile the two turnout metrics or explain their differing results.

Accepting both measures of turnout, we find that Plaintiffs have made a meager showing of lower political participation among black North Carolinians. While the Senate Report instructs that plaintiffs need not prove a causal nexus between “disparate socio-economic status and the depressed level of political participation,” both of those conditions must be “shown.” S. Rep. No. 97-417, at 29 n.114. The evidence shows that, typically, a lower proportion of black registered voters cast a vote than the proportion of white registered voters who do so but that a similar proportion of black and white voting-age populations vote in a given election. The lower black participation Dr. Taylor observed in 2022 does not illustrate that socioeconomic disparities hinder black voters’ ability to participate, because black participation approximated white participation in the preceding years.

Accordingly, in assessing the totality of circumstances, we will consider the continuing disparities in education, employment, and health that Plaintiffs have demonstrated and the modest extent to which they have hindered black voters’ “ ‘ability to participate effectively in the political process.’ ” Gingles, 478 U.S. at 37 (quoting S. Rep. No. 97-417, at 29).

g. Responsiveness to Particularized Needs

The eighth Senate factor evaluates “ ‘whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group.’ ” Id. (quoting S. Rep. No. 97-417, at 29). This has not been shown.

Neither Plaintiffs nor Defendants cited data or studies analyzing the particularized needs of black North Carolinians or polling about their support or opposition to particular State policies. Dr. Bagley made no attempt to show that any of the bills he mentioned related to a particularized minority need or were supported or opposed by the majority of black North Carolinians.²⁸ Dr. Stephens-Dougan's testimony about racialized issues did not address the particularized needs of black constituents. And Dr. Taylor's data about the responsiveness of North Carolina elected officials to their constituents as a whole also did not answer the particularized question posed.

The anecdotal evidence we do have does not show a significant lack of responsiveness. In his report, Dr. Bagley mentioned bills about “critical race theory” that black lawmakers opposed, and at trial he confirmed that the General Assembly did not pass those bills. NAACPPX181 at 48; Trial Tr. Vol. III at 707:8–17. Dr. Bagley and other witnesses also discussed the expansion of Medicaid, which he said black North Carolinians “fought for,” and which the General Assembly enacted in 2023. NAACPPX181 at 49. Senator Smith, who represents SD 5, testified that healthcare was an important issue for her black constituents and that she worked together with legislative leaders to obtain $215 million in funding in the 2021 budget for the Brody School of Medicine in Pitt County. Trial Tr. Vol. IV at 803:20–23, 818:16–819:8. Senator Smith similarly noted that the General Assembly approved $175 million in appropriations for SD 5 in the State's 2023 budget. Id. at 820:5–12. We also heard about appropriations in the General Assembly's 2023 budget for historically black colleges and universities (“HBCUs”) and black heritage and advancement organizations. See LDTX259 at 32–33. Several witnesses mentioned public education. For example, Dr. Bagley described a long-running lawsuit brought by five school districts in low-wealth counties, including Vance and Halifax, against the State in pursuit of more school funding. NAACPPX181 at 49. Dr. Taylor responded with 2024 data showing that all of the county public schools in what Dr. Bagley terms the Black Belt receive low wealth supplemental funding from the State, in addition to the small county supplemental funding from the State that most of those counties also receive. LDTX259 at 35–36; Trial Tr. Vol. VI at 1282:8–1284:8. Dr. Taylor also mentioned the NC Promise program, which reduces the in-state tuition at four public universities to $500 per semester; two of the four are HBCUs. Trial Tr. Vol. VI at 1275:5–11.

Fact witnesses presented varied opinions about their individual legislators at the state and congressional level; some found them accessible, others did not. Senator Smith remarked that she had received complaints from black North Carolinians about other legislators’ failures to visit constituents or return calls. Trial Tr. Vol. IV at 802:18–803:4. She testified that the Legislative Black Caucus represents “all 100 counties” and so responds to constituent questions and concerns no matter their district. Id. at 829:18–23.

On the whole, the evidence in this case does not show a “ ‘significant lack of responsiveness’ ” on the part of North Carolina elected officials “ ‘to the particularized needs’ ” of black North Carolinians. Gingles, 478 U.S. at 37 (quoting S. Rep. No. 97-417, at 29).

h. Policy Underlying the State's Practice

The ninth Senate factor asks whether the policy underlying the State's use of the challenged “ ‘standard, practice, or procedure is tenuous.’ ” Id. (quoting S. Rep. No. 97-417, at 29); cf. Brnovich, 141 S. Ct. at 2339–2340 (considering “the strength of the state interests served” and “the reason for the rule”). We find the reasons for the configuration of SD 1 and 2 legitimate and weighty, not tenuous or pretextual.

As we found earlier when resolving Plaintiffs’ intentional discrimination claims, SD 1 and 2 were configured to comply with the 2023 Senate plan criteria. Adhering to those criteria, the General Assembly considered equal population, county grouping and traversal rules under the Stephenson cases, compactness, contiguity, and respect for political subdivisions. See JX047. The General Assembly also took political considerations and incumbent residence into account. See id. Those are all permissible districting considerations. See, e.g., Rucho, 139 S. Ct. at 2497; Harris, 578 U.S. at 258. Senator Hise explained that, after implementing the equal population and county grouping and traversal criteria, the Senate Redistricting Committee was left with only two ways to group the counties of SD 1 and 2 consistent with Stephenson. Trial Tr. Vol. IV at 841:4–10. The Committee chairs preferred the chosen grouping because they considered SD 1 more compact in that version, because it kept four of the five “fingerling” counties of northeastern North Carolina together as a community of interest, and because the counties of SD 1 were largely in the Virginia Beach media market. Id. at 841:13–842:5. That grouping also “performed better for Republican candidates” than the alternative. Id. at 912:9–12. The grouping left no further discretion to alter the borders of SD 1 or 2; their contours were dictated by the Stephenson criteria.

The General Assembly's adherence to state law is not tenuous. The North Carolina Constitution has long prohibited dividing counties in the formation of Senate districts, and the North Carolina Supreme Court has articulated detailed rules for complying with that dictate, which the General Assembly implements through objective, non-arbitrary means. See N.C. Const. art. II § 3(3); Stephenson II, 582 S.E.2d at 249–251; Stephenson I, 562 S.E.2d at 396–398.

Likewise, the State's interest in complying with federal constitutional principles against racial gerrymandering is not tenuous. See Pierce, 97 F.4th at 221 n.10 (“Equally baseless is Plaintiffs’ assertion, under the ninth factor, that North Carolina's interests in complying with federal constitutional prohibitions against racial gerrymandering and state constitutional prohibitions against splitting counties are illegitimate considerations.”). The 2023 Senate plan criteria forbade the use of racial data in drafting the Senate map, and Senator Hise credibly testified that this policy was followed. See JX047; Trial Tr. Vol. IV at 839:13–15. Plaintiffs believe that the VRA required race-based districting in this part of the State. But they certainly cannot contend that the General Assembly's desire to adhere to the Equal Protection Clause was a tenuous basis for making “the laudable effort to disregard race altogether in the redistricting process.” Alexander, 144 S. Ct. at 1240 n.6. Indeed, it is difficult to imagine a more compelling justification for state action in this context than a desire to comply with multiple court rulings condemning the race-based districting choice demanded in this case. See, e.g., Covington, 316 F.R.D. at 124, aff'd, 581 U.S. 1015; Harris, 159 F. Supp. 3d at 604, aff'd, 581 U.S. at 290.

5. Summary

Upon considering the totality of circumstances, we find that Plaintiffs have failed to prove that the boundaries of SD 1 and 2 have “a disparate effect on account of race.” Milligan, 143 S. Ct. at 1507. To the contrary, the evidence convincingly demonstrates that “voting is ‘equally open’ and affords equal ‘opportunity’ ” to black and white North Carolinians alike. Brnovich, 141 S. Ct. at 2338 (quoting 52 U.S.C. § 10301(b)).

Black voters in northeastern North Carolina and throughout the State have successfully elected candidates of their choice, both black and white. They have done so most frequently by forming coalitions with voters from other racial and ethnic groups, having no need to be a majority within a single district in order to elect candidates of their choice. Black members of the General Assembly are above parity with the black percentage of the statewide population and hold important leadership roles in that body. The State's congressional delegation is also at or near parity with the statewide black population. The most recent election for all these offices occurred under the 2023 maps, including the Senate map at issue here, which were drawn without considering race.

All of this accords with the evidence on polarization, which demonstrates that voters in North Carolina and in the northeast are acutely polarized on the basis of political party, not on the basis of race. Indeed, black and white Democrats are similarly situated in their ability to influence elections: both will be able to elect candidates of choice if they are in majority-Democratic districts, but will be unable to do so in majority-Republican districts, regardless of the district's racial composition. This comports with the evidence showing meaningful white crossover voting to support black-preferred Democrats.

Nor do elections in North Carolina occur “against the backdrop of substantial racial discrimination.” Milligan, 143 S. Ct. at 1507. While North Carolina's history of race discrimination appears to have lingering effects as suggested by disparities in education, employment, and health, the State has undertaken measures to facilitate minority voting, including specifically in the northeastern counties relevant here. Current voting practices do not enhance the opportunity for discrimination, and racial appeals do not characterize North Carolina political campaigns.

At bottom, considering all the circumstances, this is not an “instance[ ] of intensive racial politics.” Milligan, 143 S. Ct. at 1510 (internal quotation marks omitted). There has been no showing that race plays an “excessive role ․ in the electoral process” as to “den[y] minority voters equal opportunity to participate.” Id. (internal quotation marks omitted). Accordingly, Section 2's “exacting requirements” foreclose “judicial intervention” in the State's redistricting decision. Id.

C. Conclusions of Law

Because Plaintiffs have failed to prove that the boundaries of SD 1 and 2 in the 2023 Senate map have a discriminatory effect, they have not proven their claim under Section 2 of the VRA. See 52 U.S.C. § 10301; Milligan, 143 S. Ct. at 1503.

VI. CONCLUSION

Having considered the entire record, judged the credibility of each witness, and weighed all the evidence, the Court finds that Plaintiffs have failed to prove any of their claims against Defendants regarding the 2023 redistricting.

The parties have requested this Court to resolve the claims against the 2023 plan utilizing the record created at trial, notwithstanding the pending claims based on the 2025 adjustments to CD 1 and CD 3 by S.B. 249. (Doc. 172 at 8; Doc. 173 at 8; Doc. 174 at 10; see Doc. 171 at 1-2 (State Board Defendants, taking no position on S.B. 249's effects but noting that the State Board would need to receive any new maps by December 1, 2025, to accommodate the March 2026 primary).) NAACP Plaintiffs have requested that the Court enter partial final judgment on all challenges to the 2023 plan, except for those relating to CD 1, which, along with the new challenges to CD 3, are the subject of the supplemental complaints. (Doc. 174 at 10.) NAACP Plaintiffs contend that the “appropriate course” is for the Court to hold Plaintiffs’ challenges to CD 1 in abeyance because they “will once again be ripe for consideration if and when the 2025 configurations” are struck down. (Id.) At the hearing on Plaintiffs’ pending motions for preliminary injunctions regarding S.B. 249, Williams Plaintiffs and Defendants agreed with NAACP Plaintiffs that partial final judgment is appropriate.

Federal Rule of Civil Procedure 54(b) governs partial final judgment:

[w]hen an action presents more than one claim for relief — whether as a claim, counterclaim, crossclaim, or third-party claim — or when multiple parties are involved, the court may direct entry of a final judgment as to one or more, but fewer than all, claims or parties only if the court expressly determines that there is no just reason for delay.

The Fourth Circuit requires this Court to follow a two-step inquiry when entering a judgment under Rule 54: (1) whether the judgment it is asked to enter is final; and (2) whether there is no just reason for delay in the entry of judgment. MCI Constructors, LLC v. City of Greensboro, 610 F.3d 849, 855 (4th Cir. 2010) (providing five factors the court should consider in determining whether there exists no just reason for delay under Rule 54). The five factors enumerated by the Fourth Circuit are:

(1) the relationship between the adjudicated and unadjudicated claims; (2) the possibility that the need for review might or might not be mooted by future developments in the district court; (3) the possibility that the reviewing court might be obliged to consider the same issue a second time; (4) the presence or absence of a claim or counterclaim which could result in a set-off against the judgment sought to be made final; (5) miscellaneous factors such as delay, economic and solvency considerations, shortening the time of trial, frivolity of competing claims, expense, and the like.

Id. (footnote omitted) (quoting Braswell Shipyards, Inc. v. Beazer E., Inc., 2 F.3d 1331, 1335-1336 (4th Cir. 1993)); see also Fox v. Baltimore City Police Dep't, 201 F.3d 526, 531 (4th Cir. 2000) (noting there are “broad sets of considerations” in deciding whether to enter judgment under Rule 54(b)).

We have weighed these factors and conclude they support the entry of partial final judgment here. Although the claims of the supplemental amended complaints may rely on evidence presented at trial, they are based on the General Assembly's changes to CD 1 and CD 3 under S.B. 249. (See Docs. 180, 181.) No party suggests, nor do we conclude, that we cannot resolve the claims relating to the 2023 plan at this time. We also perceive no future development that could affect the Court's decision or require us to revisit an issue we have decided. Most importantly, the State Board stated it would need to receive the maps for the March 2026 primary by December 1, 2025 (Doc. 171 at 2), and the candidate filing deadline is December 19, 2025, see N.C. Gen. Stat. § 163-106.2. Based on the foregoing, the Court finds no just reason for delay in the entry of final judgment on all claims except those related to S.B. 249 and the new CD 1 and CD 3. As for Plaintiffs’ challenges to the 2023 plan for CD 1, we reserve ruling pending the outcome of the claims in the supplemental complaints.

Accordingly, IT IS ORDERED that all remaining claims in Plaintiffs’ First Amended Complaint in Case No. 1:23-cv-1104 (Doc. 105) and all remaining claims in Plaintiffs’ Second Amended Complaint in Case No. 1:23-cv-1057 (Doc. 108), except for those relating to CD 1 on which the Court reserves ruling, are DISMISSED WITH PREJUDICE. In both cases, Plaintiffs’ supplemental complaints regarding the 2025 redistricting (Docs. 180, 181) remain pending. A partial final judgment in conformance with this order will be entered simultaneously.

FOOTNOTES

1.   Eight voters originally sued in case number 1:23-cv-1104, but the First Amended Complaint named only six.

2.   Pursuant to the Court's order consolidating these cases, all citations are to case number 1:23-cv-1057 unless otherwise noted. (Doc. 34.)

3.   Williams Plaintiffs also alleged that the congressional plan negatively impacted Hispanic voters. (See generally Doc. 108.) Williams Plaintiffs did not present evidence to support these claims at trial nor brief them in their trial materials. We therefore grant judgment for Defendants on Counts I and II in case number 1:23-cv-1057 to the extent those claims concern Hispanic voters.

4.   At the close of Plaintiffs’ case, NAACP Plaintiffs acknowledged they had not advanced evidence to support their claims of vote dilution in SD 40 or 41 under the Constitution (Count 5) or the VRA (Count 4). We granted Defendants’ motion for judgment on these claims under Federal Rule of Civil Procedure 52(c). Trial Tr. Vol. V at 1009:13–1025:21.

5.   The evidence from Williams Plaintiffs and NAACP Plaintiffs overlaps to some extent, although their claims differ. We have considered all applicable evidence as to every claim before us, regardless of which party offered it. Accordingly, from this point forward in the opinion, we refer to “Plaintiffs” generally.

6.   All trial transcripts can be found on the docket. (See Docs. 152–155, 157–158.)

7.   Among other things, those criteria forbade the use of “[d]ata identifying the race of individuals or voters” in constructing or considering districts and authorized consideration of certain 2020 and 2022 election results. JX039.

8.   The chairs also introduced Senate Bill 756, titled “Congressional Districts 2023/CBP-5,” which was an alternative congressional plan. JX100. That plan was not adopted and the parties make no arguments about it, so we do not discuss it further.

9.   Dr. Rodden described the scope of his conclusions in a colloquy with a member of this Court:Q: I have a question. I'm trying to make sure I understand what's correlation, what's causation, what's statistically significant correlation and when it crosses over to sufficient as a matter of social science to move to the point where you think we should be thinking about causation as opposed to correlation. So some of these regression analyses come back as statistically significant. Are they sufficiently statistically significant to cross to causation from correlation?A: This is something where I believe—I'd like to believe that Dr. Barber and Dr. Trende would provide the exact same answer to you, which is that this analysis does not allow—even if it's highly significant—Q: So it doesn't purport to do it?A: Right, it's not a causal analysis. And that's why I use the word “descriptive” often ․ because that's really what this is. And I don't want to make any claims that just because they ran a regression, I now have a causal story where I can say that race causes this outcome. The best we can do is say that race is correlated with the placement of either VTDs or individuals in these districts. And then we can go a step further, and we can run a regression and see that when we control for party, it's still significant ․ But that doesn't mean that we can say that race causes the outcome.․Q: I just want to make sure I understand the scope of your claim, which seems to be more modest is that once we control, we still have some statistically significant remaining correlation, that there's never a claim that that rises to causation or is demonstrative of a gerrymander; it's merely that the correlation remains?A: Yes.Trial Tr. Vol. VI at 1477:15–1480:7.

10.   We note that the 2022 plan also divided the municipality of High Point, which extends into multiple counties.

11.   In response, Dr. Rodden conducted envelope analyses for CD 6, 12, and 14 using voting precincts as the unit of analysis. WX2 at 4–6. These tests produced similar results to those tests previously discussed. See id.

12.   Dr. Barber used an algorithm called “redist.” LDTX253 at 25 n.10.

13.   But see Rucho, 139 S. Ct. at 2503 (“A permissible intent—securing partisan advantage—does not become constitutionally impermissible, like racial discrimination, when that permissible intent ‘predominates.’ ”).

14.   We also heard testimony from Plaintiffs’ expert Dr. Maxwell Palmer that, statewide, the 2024 election featured average white crossover voting of 29.6% and as high as 41.7%, among the highest he had ever seen in any State. Trial Tr. Vol. I at 182:17–183:20; see WX6 at 3.

15.   Before Congressman Davis was first elected in 2022, CD 1 was represented by Congressman G.K. Butterfield, another black Democrat, from 2004 to 2022. (Doc. 146-1 at 22:6–9.) In fact, CD 1 has been represented by a black congressperson since 1992. (Doc. 146-1 at 22:13–18.) During his tenure, Congressman Butterfield won elections in CD 1 by significant margins, including when the district was 44.5% BVAP. (Doc. 146-1 at 29:1–17.)

16.   Plaintiffs’ expert Dr. LaFleur Stephens-Dougan indicated racial attitudes drove political polarization in the American South. For the reasons stated below in our analysis of the sixth Senate factor, we do not find her testimony persuasive.

17.   We reject Dr. Clark's description of some of these districts as “majority-minority,” which he admitted at trial did not mean majority-black. Trial Tr. Vol. II at 284:25–285:11.

18.   We note that Dr. Bagley's reports were riddled with errors on this topic and others, which he confirmed at trial. These have necessarily impacted our credibility assessment.

19.   Dr. Bagley also cited a 2022 television advertisement for white Republican Bill Ward that depicted black Democrat Howard Hunter as “guzzling gas on your dime.” NAACPPX181 at 42. The video, which made no reference to race, noted that Hunter and other “Maserati driving politicians” had co-sponsored a bill raising reimbursement rates for state legislators. Id. We do not consider this a racial appeal. As Dr. Bagley agreed at trial, connecting Hunter's wealth with voters’ tax dollars could resonate with voters of any race. Trial Tr. Vol. III at 702:18–21.

20.   Dr. Stephens-Dougan's list of racialized issues was not specific to North Carolina. After seeing on cross-examination that white and black North Carolinians responded similarly to a survey question about immigration, she acknowledged that “there is a possibility that they could not have a divide on an issue that's racialized.” Trial Tr. Vol. IV at 771:11–12. This casts further doubt on Dr. Stephens-Dougan's methodology for identifying racialized issues.

21.   Plaintiffs also cite North Carolina A. Philip Randolph Institute v. North Carolina Board of Elections, 730 F. Supp. 3d 185 (M.D.N.C. 2024), aff'd, 155 F.4th 298 (4th Cir. 2025). Although that decision issued recently, it concerned the intent of “[t]he 1899 General Assembly,” which was “the legislature responsible for the operative form of the Challenged Statute.” 155 F.4th at 312.

22.   The Department of Justice also precleared the 2011 maps as not retrogressive under Section 5 of the VRA, which was still in effect. See Covington, 316 F.R.D. at 127.

23.   The fourth Senate factor is not relevant because North Carolina does not use a candidate slating process. See Gingles, 478 U.S. at 37.

24.   The data in the ACS one-year and five-year surveys are estimates based on responses from nationwide questionnaires that are imputed onto large populations. At trial, Defendants highlighted the limitations of this data. See, e.g., Trial Tr. Vol. II at 316:21–323:21. Among other things, ACS one-year estimates, like those Dr. Clark used, come with wider confidence intervals and, accordingly, less certainty about the values reported. See LDTX259 at 14 n.19.

25.   Dr. Bagley's analysis of the various statistics discussed in this section was less thorough than that of Dr. Clark or Dr. Taylor. Dr. Bagley also appeared less acquainted with the data and amended his initial report to correct percentages and the source of his data. See, e.g., Trial Tr. Vol. III at 690:19–693:8. We give his report and testimony less weight than the other experts.

26.   Dr. Bagley reported uninsurance rates of 11% for black individuals, compared to 9.1% for white respondents. NAACPPX181 at 32.

27.   We give no weight to Dr. Bagley's unsupported assertion that black North Carolinians are “less likely to be able to take time off to vote, to get to the polls, to contribute to a political campaign, or to run for office.” NAACPPX181 at 31. Dr. Bagley did not actually analyze black voters’ ability to do any of these things, nor did he cite any study doing so or linking these activities with any data point that he did analyze. See id.; Trial Tr. Vol. III at 690:1–18.

28.   In their posttrial brief, Plaintiffs similarly list various bills proposed in the General Assembly, on which they apparently would have us assume that black North Carolinians share a uniform view. Without evidence establishing black constituents’ positions on these subjects, much less whether the bills succeeded, we find this evidence unhelpful. Cf. Miller, 515 U.S. at 912 (rejecting the “demeaning assumption that voters of a particular race, because of their race, think alike” (internal quotation marks omitted)).

Allison J. Rushing United States Circuit Judge, Richard E. Myers, II Chief United States District Judge, Thomas D. Schroeder United States District Judge