STATE OF NORTH CAROLINA WAKE COUNTY IN THE GENERAL COURT OF JUSTICE SUPERIOR COURT DIVISION 18 CVS 014001 COMMON CAUSE, et al. Plaintiffs, v. Representative DAVID R. LEWIS, in his official capacity as Senior Chairman of the House Select Committee on Redistricting, et al., Defendants. ) ) ) ) ) ) ) ) ) ) JUDGMENT TABLE OF CONTENTS INTRODUCTION ...........................................................................................................................5 FINDINGS OF FACT ...............................................................................................................11 A. B. Republicans Drew the 2017 Plans to Maximize Their Political Power ....................................................................................................................11 1. Republican Mapmakers Drew the 2011 Plans ............................... 11 2. The Covington Court Struck Down Certain 2011 Districts as Unconstitutional Racial Gerrymanders ..................................... 13 3. The General Assembly Enacted the 2017 Plans ............................ 14 4. The Covington Special Master Redrew Several Districts That Remained Racially Gerrymandered ....................................... 22 The 2017 Plans Were Designed Intentionally and Effectively to Maximize Republican Partisan Advantage on a Statewide Basis..........23 1. Legislative Defendants Admitted That They Were Drawing the 2017 Plans for Partisan Gain .................................... 23 2. Dr. Hofeller’s Files Establish That the Predominant Goal Was to Maximize Republican Partisan Advantage ...................... 24 3. Plaintiffs’ Experts Established that the Plans Are Extreme Partisan Gerrymanders Designed to Ensure Republican Control ............................................................................... 37 C. The 2017 Plans Were Designed Intentionally and Effectively to Maximize Republican Partisan Advantage Within Specific County Groupings ..........................................................................................................109 1. Senate County Groupings ................................................................. 109 2. House County Groupings .................................................................. 149 D. The 2017 Plans Protected the Republican Majorities in the 2018 Elections ............................................................................................................223 E. The 2017 Plans Harm the Organizational and Individual Plaintiffs ............................................................................................................224 F. 1. The 2017 Plans Harm the North Carolina Democratic Party ...................................................................................................... 224 2. The 2017 Plans Harm Common Cause .......................................... 230 3. The 2017 Plans Harm the Individual Plaintiffs .......................... 231 Defendants Offered No Meaningful Defense of the 2017 Plans............238 1. No Witness Denied That the Plans Are Intentional and Effective Partisan Gerrymanders ................................................... 238 2. Defendants’ Criticisms of Plaintiffs’ Experts Were Not Persuasive ............................................................................................ 239 3. Dr. Karen Owen’s Testimony on “Representation” and “Competitive Elections” and Representative John Bell’s Testimony on Competitive Districts Was Unpersuasive ........... 272 4. The Whole County Provision Did Not Prevent Systematic Gerrymandering of the Plans for Partisan Gain ......................... 277 5. Plaintiffs Do Not Seek Proportional Representation ................. 278 6. Legislative Defendants Did Not Seek to Comply with the VRA and Did Not Show Nonpartisan Plans Would Violate the VRA ................................................................................................. 279 7. Legislative Defendants, through Dr. Hofeller, substantially completed drafting the Enacted Maps in June 2017 ............................................................................................. 284 2 CONCLUSIONS OF LAW .....................................................................................................292 I. THE STANDING OF PLAINTIFFS ........................................................................292 A. The North Carolina Democratic Party Has Standing .............................293 B. Common Cause Has Standing ......................................................................295 C. The Standing of Individual Plaintiffs .........................................................296 II. THE 2017 PLANS VIOLATE THE NORTH CAROLINA CONSTITUTION’S FREE ELECTIONS CLAUSE ..................................................................................298 III. THE 2017 PLANS VIOLATE THE NORTH CAROLINA CONSTITUTION’S EQUAL PROTECTION CLAUSE............................................................................307 IV. A. North Carolina’s Equal Protection Clause Provides Greater Protection for Voting Rights Than its Federal Counterpart ......................................307 B. The 2017 Plans Were Created with the Intent to Discriminate Against Plaintiffs and Other Democratic Voters .....................................................309 C. The 2017 Plans Deprive Plaintiffs and Other Democratic Voters of Substantially Equal Voting Power and the Right to Vote on Equal Terms ..................................................................................................................312 D. The 2017 Plans Cannot be Justified by any Legitimate Governmental Interest ...............................................................................................................315 THE 2017 PLANS VIOLATE THE NORTH CAROLINA CONSTITUTION’S FREEDOM OF SPEECH AND FREEDOM OF ASSEMBLY CLAUSES ......317 A. North Carolina’s Constitution Protects the Rights of Free Speech and Assembly Independently from the Federal Constitution .......................318 B. Voting, Banding Together in a Political Party, and Spending on Elections Are Protected Expression and Association ..............................320 C. The 2017 Plans Burden Protected Expression and Association ...........322 1. The 2017 Plans Burden Protected Expression Based on Viewpoint by Making Democratic Votes Less Effective ............ 322 2. The 2017 Plans Burden Plaintiffs’ Ability to Associate ............. 326 3. The 2017 Plans Burden the NCDP’s Expression Through Financial Support for Candidates ................................................... 328 3 D. The 2017 Plans Fail Strict Scrutiny—and Indeed Any Scrutiny .........328 E. The 2017 Plans Impermissibly Retaliate Against Voters Based on Their Exercise of Protected Speech .............................................................329 V. PARTISAN GERRYMANDERING CLAIMS ARE JUSTICIABLE UNDER THE NORTH CAROLINA CONSTITUTION .......................................................331 VI. ANY LACHES DEFENSE LACKS MERIT ...........................................................342 VII. DEFENDANTS’ FEDERAL DEFENSES LACK MERIT ...................................343 A. The Covington Remedial Order Does Not Bar Changes to the 2017 Plans ..................................................................................................................343 B. There Is No Conflict with Federal Civil Rights Laws .............................344 C. Granting Relief Will Not Violate the Fundamental Right to Vote .......346 VIII. THE COURT WILL ENJOIN USE OF THE 2017 PLANS IN FUTURE ELECTIONS AND THE GENERAL ASSEMBLY IS TO IMMEDIATELY BEGIN THE PROCESS OF REDRAWING THE RELEVANT DISTRICTS ...................................................................................................................347 A. The Court Will Require the Redrawing of Specific County Groupings ..........................................................................................................347 B. The Court Will Require the Use of the Adopted Criteria, with certain exceptions, and Prohibit the Use of Other Criteria in Redrawing the Districts..............................................................................................................348 C. The Court Will Not Stay the Remedial Process Pending Appeal .........351 D. The Court Retains Discretion to Move the Primary Dates ....................352 DECREE ....................................................................................................................................352 4 The People of North Carolina have delegated, through the State’s Constitution, the drawing of the State’s legislative districts to the General Assembly. The delegation of this task, however, is not so unconstrained that legislative discretion is unfettered. Rather, the power entrusted by the People to the General Assembly to draw districts is constrained by other constitutional provisions that the People have also ordained. Some of these constitutional constraints are explicit—for example, the Whole County Provision of the Constitution limits a mapmaker’s discretion to traverse county boundaries. But other constitutional constraints that limit the legislative process of map drawing are not explicit or limited in applicability only to map drawing—some constraints apply to all acts of the General Assembly, and indeed all acts of government. These principles include the obligation that our government provide all people with equal protection under law, that our government not restrict all peoples’ rights of association and political expression, and that our government allow for free elections. Plaintiffs in this case challenge the legislative districts enacted by the General Assembly in 2017 and assert that the General Assembly has exceeded the map drawing discretion afforded to it by the People by creating maps that impermissibly infringe upon the equal protection, speech, association, and free election rights of citizens. The People of North Carolina have also entrusted, through the State’s Constitution, the task of reviewing acts of other branches of government to the judicial branch. While it is solely the province of the General Assembly to make law reflecting the policy choices of the People, it is the province—and indeed the duty—of the courts of our State through judicial review to ensure that enacted law comports with the State’s Constitution. The Court cannot indiscriminately wield this power because the Court is also appropriately constrained by long-standing principles of law. Significantly, the Court must presume the constitutionality of acts of the General Assembly and must declare acts unconstitutional 5 only when such a conclusion is so clear that no reasonable doubt can arise or the statute cannot be upheld on any ground.1 The voters of this state, since 2011, have been subjected to a dizzying succession of litigation over North Carolina’s legislative and Congressional districts in state and federal courts. Today marks the third time this trial court has entered judgment. Two times, the North Carolina Supreme Court has spoken. Eight times, the United States Supreme Court has ruled. Yet, as we near the end of the decade, and with another decennial census and round of redistricting legislation ahead, the litigation rages on with little clarity or consensus. The conclusions of this Court today reflect the unanimous and best efforts of the undersigned trial judges—each hailing from different geographic regions and each with differing ideological and political outlooks—to apply core constitutional principles to this complex and divisive topic. We are aided by advances in data analytics that illuminate the evidence; we are aided by learned experts who inform our analysis; and, we are aided by skilled lawyers who have masterfully advanced the positions of their clients. But, at the end, we are guided, and must be guided, by what we conclude the North Carolina Constitution requires. The issue before the Court is distilled to simply this: whether the constitutional rights of North Carolina citizens are infringed when the General Assembly, for the purpose of retaining power, draws district maps with a predominant intent to favor voters aligned with one political party at the expense of other voters, and in fact achieves results that manifest this intent and cannot be explained by other non-partisan considerations. In this 1 “It is well settled in this State that the courts have the power, and it is their duty in proper cases, to declare an act of the General Assembly unconstitutional—but it must be plainly and clearly the case. If there is any reasonable doubt, it will be resolved in favor of the lawful exercise of their powers by the representatives of the people.” City of Asheville v. State, 369 N.C. 80, 87-88, 794 S.E.2d 759, 766 (2016) (quoting Glenn v. Bd. of Educ., 210 N.C. 525, 529-30, 187 S.E. 781, 784 (1936)); State ex rel. Martin v. Preston, 325 N.C. 438, 449, 385 S.E.2d 473, 478 (1989). 6 case, as is set out in detail below, the Court finds as fact that Plaintiffs have met their burden of proof on several critical points. Plaintiffs have established that:  the General Assembly, in enacting the 2017 legislative maps, had a partisan intent to create legislative districts that perpetuated a Republican-controlled General Assembly;  the General Assembly deployed this intent with surgical precision to carefully craft maps that grouped many voters into districts predominantly based upon partisan criteria by packing and cracking Democratic voters to dilute their collective voting strength, thereby creating partisan gerrymandered legislative maps;  the 2017 legislative maps throughout the state and on a district-by-district level, when compared on a district-by-district level to virtually all other possible maps that could be drawn with neutral, non-partisan criteria, are, in many instances, “extreme outliers” on a partisan scale to the advantage of the Republican party;  partisan intent predominated over all other redistricting criteria resulting in extreme partisan gerrymandered legislative maps; and,  the effect of these carefully crafted partisan maps is that, in all but the most unusual election scenarios, the Republican party will control a majority of both chambers of the General Assembly. In other words, the Court finds that in many election environments, it is the carefully crafted maps, and not the will of the voters, that dictate the election outcomes in a significant number of legislative districts and, ultimately, the majority control of the General Assembly. Faced with these facts, as proven by the evidence, the Court must now say whether this conduct violates the constitutional guarantees afforded to all citizens— 7 Democrats, Republicans, and others—of equal protection, the right to associate, to speak freely through voting, and to participate in free elections. Recently, the United States Supreme Court, in Rucho v. Common Cause, 139 S. Ct. 2484 (2019), held that even where enacted maps – i.e., North Carolina’s 2016 Congressional Map – were “blatant examples of partisanship driving districting decisions,” challenges of partisan gerrymandering were “beyond the reach of the federal courts” because the federal Constitution provides no “constitutional directive or legal standard” to guide the courts. Id. at 2507-08. However, the Supreme Court added that “our conclusion does not condone excessive partisan gerrymandering” and does not “condemn complaints about redistricting to echo into a void.” Id. at 2507. Rather, the Supreme Court observed that provisions of “state constitutions can provide standards and guidance for state courts to apply.” Id. The case before this Court asserts only North Carolina constitutional challenges to the enacted legislative maps. Hence, this Court considers whether the North Carolina Constitution provides the “standards and guidance” necessary to address extreme partisan gerrymandering. Of particular significance to this Court is Article I, § 10 of the North Carolina Constitution. This provision, originally enacted in 1776 and contained in the “Declaration of Rights” of our Constitution, simply states that “[a]ll elections shall be free.” The North Carolina Supreme Court has long and consistently held that “our government is founded on the will of the people,” that “their will is expressed by the ballot,” People ex rel. Van Bokkelen v. Canady, 73 N.C. 198, 220 (1875), and “the object of all elections is to ascertain, fairly and truthfully the will of the people,” Hill v. Skinner, 169 N.C. 405, 415, 86 S.E. 351, 356 (1915) (quotation omitted). The Court has also held that it is a “compelling interest” of the state “in having fair, honest elections.” State v. Petersilie, 334 N.C. 169, 184, 432 S.E.2d 8 832, 840 (1993). This Court concludes, for these and other reasons more fully set out below, that the Free Elections Clause of the North Carolina Constitution guarantees that all elections must be conducted freely and honestly to ascertain, fairly and truthfully, the will of the People and that this is a fundamental right of North Carolina citizens, a compelling governmental interest, and a cornerstone of our democratic form of government. Our understanding of the Free Elections Clause shapes the application of the Equal Protection Clause, N.C. Const. art. I, § 19, the Freedom of Speech Clause, id. at art. I, § 12, and the Freedom of Assembly Clause, id. at art. I, § 14, to instances of extreme partisan gerrymandering. In the context of the constitutional guarantee that elections must be conducted freely and honestly to ascertain, fairly and truthfully, the will of the People, these clauses provide significant constraints against governmental conduct that disfavors certain groups of voters or creates barriers to the free ascertainment and expression of the will of the People. Six years ago, this three-judge panel observed, perhaps presciently, the competing principles that are at the heart of the case before it today: “Political losses and partisan disadvantage are not the proper subject for judicial review, and those whose power or influence is stripped away by shifting political winds cannot seek a remedy from courts of law, but they must find relief from courts of public opinion in future elections.” Dickson v. Rucho, No. 11 CVS 16896 (N.C. Super Ct. July 8, 2013). This, the Court believes, is as true today as it was then. It is not the province of the Court to pick political winners or losers. It is, however, most certainly the province of the Court to ensure that “future elections” in the “courts of public opinion” are ones that freely and truthfully express the will of the People. All elections shall be free—without that guarantee, there is no remedy or relief at all. 9 This Court is acutely aware that the process employed by the General Assembly in crafting the 2017 Enacted House and Senate maps is a process that has been used for decades—albeit in less precise and granular detail—by Democrats and Republicans alike. However, long standing, and even widespread, historical practices do not immunize governmental action from constitutional scrutiny. See, e.g., Citizens United v. FEC, 558 U.S. 310, 365, 130 S. Ct. 876, 913 (2010); Reynolds v. Sims, 377 U.S. 533, 582, 84 S. Ct. 1362, 1392 (1964) (holding that malapportionment of state legislative districts violates the Equal Protection Clause, notwithstanding that malapportionment was widespread in the Nineteenth and early Twentieth Centuries). With this as our guide, this Court, in exercising its duty of reviewing acts of other branches of government to ensure that those governmental acts comport with the rights of North Carolina citizens guaranteed by the North Carolina Constitution, concludes that the 2017 Enacted House and Senate Maps are significantly tainted in that they unconstitutionally deprive every citizen of the right to elections for members of the General Assembly conducted freely and honestly to ascertain, fairly and truthfully, the will of the People. The Court bases this on the inescapable conclusion that the 2017 Enacted Maps, as drawn, do not permit voters to freely choose their representative, but rather representatives are choosing voters based upon sophisticated partisan sorting. It is not the free will of the People that is fairly ascertained through extreme partisan gerrymandering. Rather, it is the carefully crafted will of the map drawer that predominates. This Court further concludes that the 2017 Enacted Maps are tainted by an unconstitutional deprivation of all citizens’ rights to equal protection of law, freedom of speech, and freedom of assembly. These conclusions are more fully set out in the following Findings of Fact and Conclusions of Law. 10 FINDINGS OF FACT A. Republicans Drew the 2017 Plans to Maximize Their Political Power 1. 1. Republican Mapmakers Drew the 2011 Plans In the 2010 elections, as part of a national Republican effort to flip state legislative chambers in order to gain control of redistricting after the 2010 Census, Republicans won majorities in the North Carolina House of Representatives and the North Carolina Senate for the first time since 1870. PX587 ¶ 5; Tr. 867. 2. With their newfound control of both chambers of the General Assembly, Republican legislative leaders set out to redraw the boundaries of the State’s legislative districts. In North Carolina, legislative redistricting is performed exclusively by the General Assembly. The Governor cannot veto redistricting bills. N.C. Const. art. II, § 22(5)(b),(c). 3. Legislative Defendant Representative David Lewis and Senator Robert Rucho oversaw the drawing of the 2011 state House and state Senate plans (the “2011 Plans”). PX587 ¶ 8 (Leg. Defs.’ Responses to Requests for Admission); Tr. 95:17-21 (Sen. Blue). They hired Dr. Thomas Hofeller to draw the plans. Id. ¶ 7; Tr. 95:8-9. Dr. Hofeller and his team drew the plans at the North Carolina Republican Party’s headquarters in Raleigh using mapmaking software licensed by the North Carolina Republican Party. PX587 ¶¶ 10-11. 4. Legislative Defendants did not make Dr. Hofeller available to Democratic members of the General Assembly during the 2011 redistricting process, nor did Dr. Hofeller communicate with any Democratic members in developing the 2011 Plans. PX587 ¶¶ 12-13. No Democratic member of the General Assembly saw any part of any draft of the 2011 Plans before they were publicly released. Id. ¶ 14. 11 5. Legislative Defendants have stated in court filings that the 2011 Plans were “designed to ensure Republican majorities in the House and Senate.” PX575 at 55 (Defs.Appellees’ Br. on Remand, Dickson v. Rucho, No. 201PA12-3, 2015 WL 4456364 (N.C. July 13, 2015)); see id. at 16 (“Political considerations played a significant role in the enacted [2011] plans.”). Legislative Defendants asserted that they were “perfectly free” to engage in constitutional partisan gerrymandering, and that they did so in constructing the 2011 Plans. PX574 at 60 (Defs.-Appellees’ Br., Dickson v. Rucho, No. 201PA12-2, 2013 WL 6710857 (N.C. Dec. 9, 2013)). 6. To “ensure Republican majorities in the House and Senate,” PX575 at 55, Legislative Defendants and Dr. Hofeller used prior election results to construct the district boundaries to advantage Republicans. PX587 ¶¶ 6, 17. “[T]he recommendation of Tom Hofeller” was to “create a master database that would contain all [statewide] NC elections from the past decade . . . , each processed into a form that matches up with the 2010 VTD geography.” PX769 at 3 (Jan. 14, 2011 memorandum to Senator Rucho). Legislative Defendants obtained Census block-level election results from “all statewide election contests for each general election [from] 2004-2010.” PX760. 7. When reviewing the draft plans, all members of the General Assembly had access to a “Stat Pack” containing data on how the districts would perform using the results of prior statewide elections. Tr. 98:4-99:9 (Sen. Blue). Specifically, the Stat Pack showed the partisan vote share for each drafted district for each specific prior election. Id. Members of the General Assembly viewed the Stat Pack as containing “pretty reliable predictors of how [draft] districts would perform in the future based on how they performed in the past.” Tr. 99:6-9 (Sen. Blue). 12 8. In July 2011, the General Assembly enacted the 2011 Plans. N.C. Sess. Laws 2011-404 (House), 2011-402 (Senate). No Democrat voted for either plan, and only one Republican voted against them. PX587 ¶¶ 23-24. 9. In the 2012 elections, the parties’ vote shares for the House were nearly evenly split across the state, with Democrats receiving 48.4% of the two-party statewide vote. Joint Stipulation of Facts (“JSF”) ¶ 41. But Democrats won only 43 of 120 seats (36%). Id. ¶ 42. Republicans thus won a veto-proof majority in the state House—64% of the seats (77 of 120)—despite winning just a bare majority of the statewide vote. In the Senate, Democrats won nearly half of the statewide vote (48.8%) but won only 17 of 50 seats (34%). Id. ¶¶ 44-45. 10. In 2014, Republican candidates for the House won 54.4% of the statewide vote, and again won a super-majority of seats (74 of 120, or 61.6%). JSF ¶ 66. In the 2014 Senate elections, Republicans won 54.3% of statewide vote and 68% of the seats (34 of 50). Id. ¶ 66. 11. In 2016, Republicans again won 74 of 120 House seats, or 61.6%, this time with 52.6% of the statewide vote. Id. ¶ 66. In the 2016 Senate elections, Republicans won 55.9% of the statewide vote and 70% of the seats (35 of 50). Id. ¶ 66. 2. 12. The Covington Court Struck Down Certain 2011 Districts as Unconstitutional Racial Gerrymanders On May 19, 2015, a group of individual plaintiffs initiated a lawsuit— Covington v. North Carolina, No. 1:15-CV-00399 (M.D.N.C.)—against the State Board of Elections, Speaker Timothy Moore, President Pro Tempore Philip Berger, Chair of the Senate Redistricting Committee, Robert Rucho, and Chair of the House Redistricting Committee, David Lewis challenging 28 total House and Senate districts under the 2011 13 Plans as unconstitutional racial gerrymanders. This case was referenced at trial, the related briefs, and in these findings as the “Covington case” or “Covington litigation.” 13. On August 11, 2016, the federal district court ruled for the plaintiffs as to all of the challenged districts. Covington v. North Carolina, 316 F.R.D. 117 (M.D.N.C. 2016). The Covington court found that racial considerations rather than political considerations “played a primary role” with respect to the specific 28 “challenged districts” in Covington. 316 F.R.D. at 139. The Covington litigation did not involve any of the districts drawn in 2011 that are at issue in the present case. 14. Following appeal, on June 5, 2017, the U.S. Supreme Court summarily affirmed the district court’s decision invalidating the 28 challenged districts as racial gerrymanders. 137 S. Ct. 2211 (mem.). 15. The district court subsequently ordered briefing on whether to order enactment of remedial maps under a timeline that would enable special elections in 2017. Ultimately, the court declined to order special elections in 2017 and instead allowed a longer timeline for the General Assembly to enact remedial plans. Covington v. North Carolina, 267 F. Supp. 3d 664 (M.D.N.C. 2017). 3. 16. The General Assembly Enacted the 2017 Plans On June 30, 2017, Senator Berger appointed 15 senators—10 Republicans and 5 Democrats—to the Senate Committee on Redistricting. PX587 ¶ 44. Senator Hise was appointed Chair. Id. Also on June 30, 2017, Representative Moore appointed 41 House members—28 Republicans and 13 Democrats—to the House Select Committee on Redistricting. PX629 at 4-5. Representative Lewis was appointed Senior Chair. PX587 ¶ 45. 17. On July 26, 2017, the Senate Redistricting Committee and the House Select Committee on Redistricting met jointly (“Redistricting Committee”) for organizational and 14 informational purposes. Covington v. North Carolina, 1:15-cv-00399, ECF No. 184-7 at 3-4. At the meeting, Representative Lewis and Senator Hise stated that Republican leadership would again employ Dr. Hofeller to draw the new plans. PX601 at 23:3-6; see PX587 ¶¶ 4647. When Democratic Senator Van Duyn asked whether Dr. Hofeller would “be available to Democrats and maybe even the Black Caucus to consult,” Representative Lewis answered “no.” PX601 at 22:24-23:6. Representative Lewis explained that, “with the approval of the Speaker and the President Pro Tem of the Senate,” “Dr. Hofeller is working as a consultant to the Chairs,” i.e., as a consultant only to Legislative Defendants. Id. at 23:3-6; Tr. 101:618 (Sen. Blue). 18. In explaining the choice of Dr. Hofeller to draw the 2017 Plans, Representative Lewis stated that Dr. Hofeller was “very fluent in being able to help legislators translate their desires” into the district lines using “the [M]aptitude program.” PX590 at 36:17-19. 19. On August 4, 2017, at another joint meeting of the Redistricting Committees, Representative Lewis and Senator Hise advised Committee members that the Covington decision invalidating 28 districts on federal constitutional grounds had rendered a large number of additional districts invalid under the Whole County Provision of the North Carolina Constitution, and those districts would also have to be redrawn. PX602 at 2:1411:23. 20. At the same August 4, 2017, meeting, the Redistricting Committees allowed 31 citizens to speak for two minutes each. PX602 at 28:3-68:23. All speakers urged the members to adopt fair maps free of partisan bias. See id. 21. At another joint meeting on August 10, 2017, the House and Senate Redistricting Committees voted on criteria to govern the creation of the new plans. PX603 at 4:23-5:5. 15 22. Representative Lewis proposed as one criterion, “election data[:] Political consideration[s] and election results data may be used in drawing up legislative districts in the 2017 House and Senate plans.” PX603 at 132:10-13. Representative Lewis provided no further explanation or justification for this proposed criterion, stating only: “I believe this is pretty self-explanatory, and I would urge members to adopt the criteria.” Id. at 132:13-15. 23. Democratic members pressed Representative Lewis for details on how Dr. Hofeller would use elections data and for what purpose. Democratic Senator Ben Clark asked: “You’re going to collect the political data. What specifically would the Committee do with it?” PX603 at 135:11-13. Representative Lewis answered that “the Committee could look at the political data as evidence to how, perhaps, votes have been cast in the past.” Id. at 135:15-17. When Senator Clark inquired why the Committees would consider election results if not to predict future election outcomes, Representative Lewis stated only that “the consideration of political data in terms of election results is an established districting criteria, and it’s one that I propose that this committee use in drawing the map.” Id. at 141:12-16. 24. Representative Lewis had also stated that Dr. Hofeller used ten specific prior statewide elections in drawing the 2017 Plans: the 2010 U.S. Senate election, the 2012 elections for President, Governor, and Lieutenant Governor, the 2014 U.S. Senate election, and the 2016 elections for President, U.S. Senate, Governor, Lieutenant Governor, and Attorney General. PX603 at 137:22-138:3. 25. The House and Senate Redistricting Committees adopted Representative Lewis’s “election data” criterion on a straight party-line vote. PX603 at 141-48. 26. Senator Clark proposed an amendment that would prohibit the General Assembly from seeking to maintain or establish a partisan advantage for any party in redrawing the plans. PX603 at 166:9-167:3. Representative Lewis opposed the amendment, 16 stating he “would not advocate for [its] passage.” Id. at 167:10-11. The Redistricting Committees rejected Senator Clark’s proposal, again on a straight party-line vote. Id. at 168-74. 27. As explained in extensive detail below, Dr. Hofeller’s own files establish that he used prior elections results and partisanship formulas to draw district boundaries to maximize the number of seats that Republicans would win in the House and the Senate, and to ensure that Republicans would retain majorities in both chambers. PX123 at 48-76 (Chen Rebuttal Report); PX329 at 3-35 (Cooper Rebuttal Report); PX153, PX166; PX167; PX168; PX170; PX171; PX172; PX241; PX244; PX246; PX248; PX330; PX332; PX333; PX334; PX335; PX336; PX337; PX340; PX342; PX344; PX345; PX346; PX347; PX350; PX352; PX353; PX354; PX724; PX730; PX731; PX732; PX733; PX734; PX735; PX736; PX738; PX739; PX742; PX744; PX746; PX748; PX753; PX754; PX755; PX756. 28. As a further criterion, Representative Lewis proposed incumbency protection—namely that “reasonable efforts and political considerations may be used to avoid pairing incumbent members of the House or Senate with another incumbent in legislative districts drawn in 2017 House and Senate plans. The Committee may make reasonable efforts to ensure voters have a reasonable opportunity to elect non-paired incumbents of either party to a district in the 2017 House and Senate plans.” PX603 at 119:9-17. He clarified that the second sentence of this proposed criterion meant “simply” that “the map makers may take reasonable efforts not to pair incumbents unduly.” Id. at 122:16-18; see PX606 at 9:24-10:1 (Sen. Hise: “The Committee adopted criteria pledging to make reasonable efforts not to double-bunk incumbents.”). 29. The House and Senate Redistricting Committees adopted Representative Lewis’s incumbency-protection criterion, once more on a straight-party line vote. PX603 at 125-32. 17 30. The Redistricting Committees also adopted as criteria, yet again on straight party-line votes, that they (1) would make “reasonable efforts” to “improve the compactness of the current districts,” PX603 at 24:24-25:2; (2) would make “reasonable efforts” to “split fewer precincts” than under the 2011 Plans, id. at 79:8-12; and (3) “may consider municipal boundaries” in drawing the new districts, id. at 66:15-16; see id. at 98-104, 112-19 (adopting criteria). Representative Lewis clarified that these criteria meant “trying to keep towns, cities and precincts whole where possible.” PX607 at 10:5-6; see, e.g., PX603 at 66:22-23 (Rep. Lewis explaining that the Committees would “consider not dividing municipalities where possible”). 31. As a final criterion, Representative Lewis proposed prohibiting the consideration of racial data in drawing the new plans. PX603 at 148:11-15. 32. The full criteria adopted by the Committees for the 2017 Plans (the “Adopted Criteria”) read as follows: Equal Population. The Committees shall use the 2010 federal decennial census data as the sole basis of population for drawing legislative districts in the 2017 House and Senate plans. The number of persons in each legislative district shall comply with the +/- 5 percent population deviation standard established by Stephenson v. Bartlett, 355 N.C. 354, 562 S.E. 2d 377 (2002). Contiguity. Legislative districts shall be comprised of contiguous territory. Contiguity by water is sufficient. County Groupings and Traversals. The Committees shall draw legislative districts within county groupings as required by Stephenson v. Bartlett, 355 N.C. 354, 562 S.E. 2d 377 (2002) (Stephenson I), Stephenson v. Bartlett, 357 N.C. 301, 582 S.E.2d 247 (2003) (Stephenson II), Dickson v. Rucho, 367 N.C. 542, 766 S.E.2d 238 (2014) (Dickson I) and Dickson v. Rucho, 368 N.C. 481, 781 S.E.2d 460 (2015) (Dickson II). Within county groupings, county lines shall not be traversed except as authorized by Stephenson I, Stephenson II, Dickson I, and Dickson II. Compactness. The Committees shall make reasonable efforts to draw legislative districts in the 2017 House and Senate plans that improve the compactness of the current districts. In doing so, the Committees may use 18 as a guide the minimum Reock (“dispersion”) and Polsby-Popper (“perimeter”) scores identified by Richard H. Pildes and Richard G. Neimi in Expressive Harms, “Bizarre Districts,” and Voting Rights: Evaluating Election-District Appearances After Shaw v. Reno, 92 Mich. L. Rev. 483 (1993). Fewer Split Precincts. The Committees shall make reasonable efforts to draw legislative districts in the 2017 House and Senate plans that split fewer precincts than the current legislative redistricting plans. Municipal Boundaries. The Committees may consider municipal boundaries when drawing legislative districts in the 2017 House and Senate plans. Incumbency Protection. Reasonable efforts and political considerations may be used to avoid pairing incumbent members of the House or Senate with another incumbent in legislative districts drawn in the 2017 House and Senate plans. The Committees may make reasonable efforts to ensure voters have a reasonable opportunity to elect non-paired incumbents of either party to a district in the 2017 House and Senate plans. Election Data. Political considerations and election results data may be used in the drawing of legislative districts in the 2017 House and Senate plans. No Consideration of Racial Data. Data identifying the race of individuals or voters shall not be used in the drawing of legislative districts in the 2017 House and Senate plans. PX587 ¶ 53; LDTX007. 33. On August 11, 2017, Representative Lewis and Senator Hise notified Dr. Hofeller of the criteria adopted by the redistricting committees and “directed him to utilize those criteria when drawing districts in the 2017 plans.” PX629 at 7. The criteria were also placed on legislative websites for the public to view and comment. Covington v. North Carolina, 1:15-cv-00399, ECF No. 184-9 at 193. 34. Dr. Hofeller drew the 2017 Plans under the direction of Legislative Defendants and without consultation with any Democratic members. PX587 ¶¶ 48-51, 5556. Representative Lewis claimed that he “primarily . . . directed how the [House] map was produced,” and that he, Dr. Hofeller, and Republican Representative Nelson Dollar were 19 the only “three people” who had even “seen it prior to its public publication.” PX590 at 40:14-21. None of Legislative Defendants’ meetings with Dr. Hofeller about the 2017 redistricting were public. PX587 ¶ 51. Legislative Defendants did not make Dr. Hofeller available to Democratic members during the 2017 redistricting process, nor did Dr. Hofeller communicate with any Democratic members in developing the 2017 Plans. PX587 ¶¶ 4849; Tr. 126:16-18 (Sen. Blue). No Democratic member of the General Assembly saw any part of any draft of the 2017 Plans before they were publicly released. PX587 ¶ 50. 35. On August 19, 2017, the proposed 2017 House plan was released on the General Assembly website. PX629 at 7. The House Redistricting Committee made only minor adjustments to Dr. Hofeller’s draft, swapping precincts between a few districts. PX605 at 16:2-17:16. 36. On August 20, 2017, the proposed 2017 Senate plan was released on the General Assembly website. PX629 at 7. At a Senate Redistricting Committee hearing on August 24, 2017, Senator Van Duyn asked Senator Hise how prior elections data had been used in drawing the proposed maps. PX606 at 26:4-6. Senator Hise replied that the mapmaker, Dr. Hofeller, “did make partisan considerations when drawing particular districts.” Id. at 26:9-10. 37. The Senate Redistricting Committee adopted only two minor amendments to the district boundaries drawn by Dr. Hofeller. One change, proposed by Senator Clark, moved a small population from Senate District 19 to District 21. PX606 at 49:20-52:9. The other change, proposed by Democratic Senator Daniel Blue, swapped a few precincts between Senate Districts 14 and 15, two heavily Democratic districts in Wake County. Id. at 52:19-53:19. Senator Blue’s amendment passed by a unanimous vote. Id. at 67:13-19. 38. As in 2011, Stat Packs measuring the partisan performance of the draft districts under recent elections were made available to members of the Redistricting 20 Committees. Tr. 113:17-115:15 (Sen. Blue). The Stat Packs, released on August 21, 2017, see PX629 at 7, contained information for each proposed district based on the ten statewide elections that Representative Lewis had claimed would be used in drawing the 2017 Plans. PX591; PX597. 39. Following the public release of the draft House and Senate maps, Legislative Defendants held public meetings on August 22, 2017, in Raleigh and at six satellite locations across the state. PX607 at 7:22-8:11, 9:1-3. Many citizens spoke at the meetings and expressed grave concerns about the draft maps. As Senator Blue testified, “overwhelmingly they were saying that they wanted districts drawn that were not partisan in nature.” Tr. 105:8-12. 40. On August 24, 2017, the Senate Redistricting Committee adopted the Senate plan drawn by Dr. Hofeller with the minor modifications discussed above. PX606 at 131:1023. The next day, the House Redistricting Committee adopted Dr. Hofeller’s proposed House plan, also with the minor modifications discussed above. PX605 at 120:2-125:25. 41. During a Floor Session Hearing on August 28, 2017, Representative Lewis proposed an amendment to modify several House districts in Wake County. PX590 at 30:13-32:2. The amendment passed on a straight party-line vote. Id. at 31:18-32:2. 42. On August 31, 2017, the General Assembly passed the House plan (designated HB 927) and the Senate plan (designated SB 691), with only a few minor modifications from the versions passed by the Committees. PX629 at 8-9; see PX627 (HB 927); PX628 (SB 691). No Democratic Senator voted in favor of either plan. PX587 ¶ 71. The lone Democratic member of the House who voted for the plans was Representative William Brisson, who switched to become a Republican several months later. Id. 43. The 2017 Plans altered 79 House districts and 35 Senate districts from the 2011 Plans. JSF ¶¶ 169-70. 21 4. 44. The Covington Special Master Redrew Several Districts That Remained Racially Gerrymandered On September 15, 2017, the Covington plaintiffs filed an objection to the 2017 draft plans, alleging that Senate Districts 21 and 28 and House Districts 57 and 21 were still racial gerrymanders. Covington v. North Carolina, 283 F. Supp. 3d 410, 429 (M.D.N.C. 2018). The Covington Court agreed. Id. at 429-42. The court further held that the General Assembly’s changes to five House districts (36, 37, 40, 41, and 105) violated the North Carolina Constitution’s prohibition on mid-decade redistricting. Id. at 443-45. 45. The court appointed Dr. Nathaniel Persily as a Special Master to assist in redrawing the districts for which the court had sustained the plaintiffs’ objections. To cure the racially gerrymandered districts, the Special Master made adjustments to certain neighboring districts as well. Covington, ECF No. 220 at 46, 64. The court adopted the Special Master’s recommended changes to all of these districts. 283 F. Supp. 3d at 458. 46. The Special Master also restored the districts that the court had found were redrawn in violation of the ban on mid-decade redistricting to the 2011 versions of those districts. Covington, 283 F. Supp. 3d at 456-58. The court adopted these changes as well. Id. 47. On June 28, 2018, the U.S. Supreme Court affirmed the district court’s adoption of the Special Master’s remedial plans for House Districts 21 and 57 (and the adjoining districts, 22, 59, 61, and 62) and Senate Districts 21 and 28 (and the adjoining districts, 19, 24, and 27). North Carolina v. Covington, 138 S. Ct. 2548, 2553-54 (2018). But the U.S. Supreme Court reversed the district court’s adoption of the Special Master’s plans for the districts allegedly enacted in violation of the mid-decade redistricting prohibition, holding that the court’s remedial authority was limited to curing the racial gerrymanders and nothing more. Id. at 2554-55. 22 48. Ultimately, the Special Master’s Final Report altered the following districts: Senate Districts 19, 21, 24, 27, 28; House Districts 21, 22, 57, 59, 61. LDTX159. The Special Master also reviewed the 2017 Enacted Plan and chose to keep the General Assembly’s version of House Districts 58 and 60 in his recommended changes. Id. 49. Plaintiffs in this case do not challenge the following districts that were altered by the Covington Special Master: House Districts 21, 22, 57, 61, 62; Senate Districts 19, 21, 24, 28. B. The 2017 Plans Were Designed Intentionally and Effectively to Maximize Republican Partisan Advantage on a Statewide Basis 1. 50. Legislative Defendants Admitted That They Were Drawing the 2017 Plans for Partisan Gain At trial, there was little meaningful dispute that Legislative Defendants drew the 2017 Plans to advantage Republicans and reduce the effectiveness of Democratic votes. 51. The 2017 Adopted Criteria expressly provided for the use of “election data” in drawing the 2017 Plans. LDTX007. The Joint Select Committee on Redistricting considered results from 10 statewide elections, captured in Stat Packs available to legislators when they considered whether to adopt Dr. Hofeller’s draft House and Senate plans. Tr. 113:17-115:15. The Stat Packs demonstrated that, under those 10 statewide elections, Republicans would be expected to win between 72 and 82 seats in the House and between 31 and 35 seats in the Senate. PX591; PX597. In other words, Republicans would win a supermajority in both chambers of the General Assembly under each and every one of the 10 statewide elections used to evaluate the 2017 Plans (72 seats provides a supermajority in the House and 30 seats does in the Senate). 52. As Senator Blue testified, the election data used by Legislative Defendants— and in particular the performance of the proposed House and Senate plans under the range 23 of 10 prior statewide elections—revealed that the plans were “designed specifically to preserve the supermajority” that the Republican Party had gained under the 2011 Plans. Tr. 115:19-22. 53. At the Senate Redistricting Committee hearing on August 24, 2017, Senator Hise confirmed that the mapmaker, Dr. Hofeller, “did make partisan considerations when drawing particular districts” in 2017. PX606 at 26:9-10. And as discussed above, Legislative Defendants stated in prior court filings that the districts drawn in 2011 were “designed to ensure Republican majorities in the House and Senate.” PX575 at 16, 55 (Dickson v. Rucho, No. 201PA12-3, 2015 WL 4456364 (N.C. July 13, 2015)). 2. 54. Dr. Hofeller’s Files Establish That the Predominant Goal Was to Maximize Republican Partisan Advantage Files from Dr. Hofeller’s storage devices provide direct evidence of Dr. Hofeller’s predominant focus on maximizing Republican partisan advantage in creating the 2017 Plans. The Court specifically finds, based upon the direct and circumstantial evidence of record, that the partisan intent demonstrated in Dr. Hofeller’s files, as detailed below, is attributable to Legislative Defendants inasmuch that Dr. Hofeller, at all relevant times, worked under the direction of, and in concert with, Legislative Defendants. See, e.g., FOF § F.7. 55. Plaintiffs obtained this evidence through a subpoena to Dr. Hofeller’s daughter. PX676; PX781 (S. Hofeller deposition). Plaintiffs issued the subpoena to Ms. Hofeller on February 13, 2019 and provided notice to all other parties the same day. PX676. After no party objected to the subpoena, on March 13, 2019, Ms. Hofeller produced 22 electronic storage devices that had belonged to her father and that her mother gave her after Dr. Hofeller’s death. PX781 at 1-43. The Hofeller files admitted into evidence at trial 24 all came from these storage devices. PX123 at 2, 39, 48 (Chen Rebuttal Report); PX329 at 3-4 (Cooper Rebuttal Report).2 56. This Court granted Plaintiffs’ pretrial motion in limine to admit the relevant files from Dr. Hofeller’s storage devices, finding sufficient evidence of authenticity and chain of custody. As the Court suggested in its pretrial ruling, and now holds, these files are public records pursuant to N.C. Gen. Stat. § 120-133(a) and Dr. Hofeller’s contract with the General Assembly to draw the 2017 Plans. PX641. The Court denied Legislative Defendants’ motion in limine to exclude the Hofeller files based on purported misconduct by Plaintiffs or their counsel. 57. Dr. Hofeller maintained two folders related to the 2017 redistricting, titled “NC 2017 Redistricting” and “2017 Redistricting.” Tr. 449:20-450:5. Plaintiffs’ expert Dr. Chen reviewed the entire contents of these two folders and found that, other than verifying that draft districts met the equal population and county grouping requirements, the files exhibited a consistent focus on partisan considerations. PX123 at 76 (Chen Rebuttal Report); Tr. 450:6-13. Among the hundreds of files in these two folders, there were a “few files” that report on VTD and county splits, “[b]ut beyond these few files,” these hundreds of files focused overwhelmingly on each party’s expected vote share in the draft districts and on the identities and party affiliations of the incumbent members in each district. PX123 at 76 (Chen Rebuttal Report). The fact that these folders focused overwhelmingly on partisan considerations is persuasive evidence that partisan intent predominated in the drawing of the 2017 Plans. The Court at trial allowed the parties to admit expert reports as “corroborative evidence”—i.e., as evidence that “tends to add weight or credibility” to the experts’ testimony. State v. Garcell, 363 N.C. 10, 40, 678 S.E.2d 618, 637 (2009); see Tr. 537:8-538:7. 2 25 a. 58. Dr. Hofeller’s partisanship formulas The specific contents of the two folders confirm Dr. Hofeller’s focus on Republican partisan advantage. In the folders, Dr. Hofeller had three partisanship formulas. First, as reflected in a Microsoft Word document titled “FORMULA FOR POLITICAL ANALYSIS OF LEGISLATIVE DISTRICTS,” Dr. Hofeller used a formula that measured the average Republican vote share in each VTD across nine statewide elections from 2008 to 2014. Tr. 450:24-451:15; PX123 at 49-52 (Chen Rebuttal Report). These nine elections were different from the ten elections Representative Lewis claimed would be used. Tr. 451:20-452:6. Dr. Hofeller used this partisanship formula based on 2008-2014 elections to measure the partisanship of his draft districts through at least July 2017, Tr. 452:7-10, by which point he had already substantially completed drawing preliminary drafts for most of the final districts, FOF § F.7. Plaintiffs’ Exhibit 153 is a screenshot of Dr. Hofeller’s Microsoft Word document containing this partisanship formula: 59. Dr. Hofeller’s second partisanship formula was based on the ten statewide elections from 2010-2016 that Representative Lewis claimed would be used in 2017. Tr. 452:12-453:21. Dr. Hofeller did not employ this formula, however, in the Excel worksheets where he analyzed the partisanship of his draft districts. Tr. 453:12-17. 26 60. Dr. Hofeller’s final partisanship formula, titled “Off Year,” was based on the results of statewide elections during non-Presidential election years, namely 2010 and 2014. Tr. 453:22-454:9; PX123 at 65 (Chen Rebuttal Report). It is apparent that Dr. Hofeller used this formula to evaluate how his districts might perform in non-Presidential years. Tr. 454:10-17. 61. Dr. Hofeller’s “NC 2017 Redistricting” and “2017 Redistricting” folders contain numerous Microsoft Excel spreadsheets analyzing partisan considerations, using his partisanship formulas, for the draft House and Senate plans that he was developing and modifying from November 2016 through June 2017. See PX123 at 53-64 (Chen Rebuttal Report). 62. First, Dr. Hofeller placed a special focus on how many of his draft House and Senate districts had an average Republican vote share of 53% or higher using his partisanship formulas. For instance, in a spreadsheet last modified on November 26, 2016, analyzing a draft Senate plan, Dr. Hofeller wrote “23 Under 53%” at the bottom to indicate the number of draft districts for which Democrats had less than a 53% vote share and Republicans had a 53% or higher vote share. Tr. 456:14-20; PX248 at 2. In other words, as shown in Plaintiffs’ Exhibit 248 below, Dr. Hofeller projected that 27 of the 50 districts in this draft Senate plan would have a Republican vote share at or above 53%. 27 28 63. In subsequent June 2017 spreadsheets analyzing draft House and Senate plans, Dr. Hofeller color-coded the districts to differentiate between districts that had slightly-under and slightly-over a 53% expected Republican vote share. Dr. Hofeller shaded the “Avg R” column yellow for draft districts with an expected Republican vote share of 5053%, and shaded cells in the column a peach color for districts with an expected Republican vote share of 53-55%. Tr. 460:6-461:8, 464:19-465:11; PX244; PX241; PX246; PX123 at 66 (Chen Rebuttal Report). 64. Dr. Hofeller stratified all of the Republican-leaning districts in his draft House and Senate plans using highly granular gradations. Tr. 461:1-8, 463:6-25, 465:16466:20; PX241 at 3; PX244 at 2; PX246 at 3. As illustrated in Plaintiffs’ Exhibits 244 below, Dr. Hofeller counted how many districts in each draft House and Senate plan had between a 50-53%, 53-55%, 55-60%, 60-65%, and 65%-100% expected Republican vote share. Id. In contrast, Dr. Hofeller did not analyze Democratic-leaning districts with such granularity. Whereas Dr. Hofeller analyzed the Republican-leaning districts in five different bands, he analyzed Democratic-leaning districts in just two bands of 0-45% Republican vote share and 45-50% Republican vote share. Tr. 466:1-20; PX241 at 3; PX244 at 2; PX246 at 3. 29 30 65. The Court finds that Dr. Hofeller’s granular sorting and analysis of Republican-leaning districts—and his particular emphasis on districts with an over-53% expected Republican vote share—provide substantial evidence of the partisan intent and effects of the 2017 plans. The evidence establishes that Dr. Hofeller drew the 2017 Plans very precisely to create as many “safe” Republican districts as possible, so that Republicans would maintain their supermajorities, or at least majorities even in a strong election year for Democrats. Tr. 456:21-457:25. For instance, Dr. Hofeller’s June 13, 2017, spreadsheet above estimated that 28 of 50 draft Senate districts had an expected Republican vote share above 53%, PX244 at 2, and Dr. Hofeller’s June 14, 2017 spreadsheet for a draft House map estimated that 74 of 120 districts in the draft House plan had an expected Republican vote share above 53%, PX246 at 3. The Court is persuaded that Dr. Hofeller drew the maps with an intent to preserve Republicans’ control of the House and Senate. 31 66. As further evidence of partisan intent, using his partisanship formula, Dr. Hofeller calculated the difference in the Republican vote share between the new draft version of each district and the prior 2011 version of that district, showing precisely how his draft plans would alter the partisanship of each district. Tr. 459:8-460:5; PX241; PX244; PX246; PX248. 67. Dr. Hofeller’s spreadsheets also highlighted in yellow many of North Carolina’s largest and most-Democratic counties, such as Wake, Mecklenburg, Cumberland, Forsyth, and Guilford Counties. Tr. 461:9-462:2, 468:9-20; PX244; PX246. As Dr. Chen explained, the spreadsheets show Dr. Hofeller’s specific focus on trying to “squeeze out” as many Republican-leaning districts as he could in these counties. Id. 68. For both his draft House and Senate plans, Dr. Hofeller analyzed what he described as “Pressure Points for GOP Incumbents.” Tr. 462:3-463:5, 467:7-468:8; PX244 at 2; PX246 at 2. He analyzed draft districts that could create concerns or vulnerabilities for Republican incumbents. Id. Dr. Chen did not find any comparable analysis by Dr. Hofeller of “pressure points” for Democratic incumbents. Id. Dr. Hofeller’s spreadsheets contradict Legislative Defendants’ contention at trial that the 2017 Plans sought to place all incumbents in politically favorable districts. It is clear from Dr. Hofeller’s files that the mapmaker predominantly focused on benefitting and electorally protecting Republican incumbents and not Democratic incumbents. 69. Dr. Hofeller’s spreadsheets also reveal that he evaluated the partisanship of draft maps created by Campbell University Law students at an exercise by Common Cause. In 2017, Common Cause invited two Campbell Law students to draw new legislative maps without using political data. Bob Phillips, the Executive Director of Common Cause North Carolina, testified that the purpose of the exercise was to raise awareness and show how a nonpartisan redistricting process could occur. Tr. 53:17-54:14. 32 70. Emails introduced at trial reveal that, in late June 2017, an aide to Legislative Defendants asked the General Assembly’s legislative services office for copies of the “block assignments files” for the simulated maps created by the Campbell Law students. PX757. Common Cause had the Campbell Law students create the maps using the General Assembly’s public computer because it had Maptitude installed on it. Tr. 55:18-56:17. Within roughly a week, Dr. Hofeller had created Excel spreadsheets analyzing the partisanship of the Campbell Law students’ simulated districts. Tr. 471:6-472:15; PX167; PX170; PX123 at 70-75 (Chen Rebuttal Report). In spreadsheets last modified on July 5 and 8, 2017, Dr. Hofeller scored every one of the Campbell Law students’ House and Senate districts using his partisanship formula derived from the 2008-2014 statewide elections. Id. Dr. Hofeller then evaluated, for every district, whether Republicans could obtain a “Better Possible” district than the version the Campbell Law students had drawn, with Dr. Hofeller writing “No,” “Yes,” or “Little” for each district. Tr. 473:8-474:6; PX168; PX123 at 70-71 (Chen Rebuttal Report). 71. The final enacted 2017 House plan contains two county groupings, with four districts in total, that match the districts in those county groupings drawn by the Campbell Law students. Tr. 474:7-475:23; PX123 at 71. Those two groupings—Nash-Franklin and Granville-Person-Vance-Warren—are two small groupings for which there are a very limited number of ways to draw the groupings, and the Campbell Law students happened to draw these groupings in the way that is most favorable to Republicans. Id. 72. Dr. Chen thus concluded that Dr. Hofeller evaluated the partisanship of all of the Campbell Law students’ districts and then included in the 2017 maps four districts for which the students happened to draw the districts in the way maximally favorable to Republicans. Id. The Court agrees with Dr. Chen’s assessment, which went unrebutted by Legislative Defendants at trial. 33 b. 73. Dr. Hofeller’s Maptitude files Dr. Hofeller’s Maptitude files from his storage devices further demonstrate that partisanship considerations were “front and center” in his drafting of the relevant districts in both 2011 and 2017. Tr. 944:5-15, 968:4-5 (Dr. Cooper). The Maptitude files remove any doubt that Dr. Hofeller “was clearly working with partisan data on the same maps at the same time that he [was] drawing lines for our state,” all to maximize Republican partisan advantage. Tr. 945:4-11. 74. As Dr. Cooper explained, the Maptitude files indicate that Dr. Hofeller used partisanship formulas, along with multiple color-coding systems to visually depict partisanship on his draft maps, in order to deliberately pack and crack Democratic voters into particular districts with precision. Tr. 939:1-940:12, 944:9-945:8; PX329 at 3-4 (Cooper Rebuttal Report). 75. In the “NC Senate J-24” Maptitude file last modified in July 2017, Dr. Hofeller calculated the Republican vote share for each North Carolina VTD based on his formula using nine statewide elections from 2008-2014. PX330; Tr. 939:9-940:2, 942:22943:2; PX565. Dr. Hofeller then color-coded the VTDs on the “Map” window based on this partisanship formula, using more granular stratifications for competitive and Republicanleaning VTDs than for Democratic-leaning VTDs, just as he had done in his Excel spreadsheets assessing district-wide partisanship. Tr. 944:16-21. Dr. Hofeller used a “traffic light” color-coding scheme, in which he shaded Democratic-leaning VTDs pink and red, Republican-leaning VTDs green, and more competitive VTDs yellow. Tr. 940:23-941:4. Plaintiffs’ Exhibit 335 below is one example of Dr. Hofeller’s use of this color-coding scheme. As is apparent in the example below and discussed in more detail with respect to additional county groupings discussed below, Dr. Hofeller drew district boundaries based on this colorcoded partisanship data with remarkable precision. 34 76. Dr. Hofeller used the same partisanship formula in his Maptitude files containing draft 2017 House districts. Tr. 979:6-19; PX337; PX329 at 13 (Cooper Rebuttal Report). Dr. Hofeller also employed a color-coding system to visually represent the partisanship scores for each VTD in his 2017 House plan, but with the more familiar red coloring for Republican-leaning VTDs, blue for Democratic-leaning VTDs, and yellow and green for more competitive VTDs. Tr. 979:20-980:19; PX329 at 13 (Cooper Rebuttal Report). For example, Dr. Hofeller’s Maptitude file labeled “NC House J-25,” which he created on June 26, 2017, and last modified on August 7, 2017, depicted boundaries (in red) of House Districts 8, 9, and 12 in the Pitt-Lenoir House county grouping. Tr. 981:2-5; PX340; PX562. Plaintiffs’ Exhibit 340 below shows that Dr. Hofeller used his color-coding system to pack the bluest VTDs in Pitt County into House District 8. Tr. 982:1-7, 983:5984:7; PX340; PX329 at 16 (Cooper Rebuttal Report). 35 77. Dr. Hofeller similarly used a partisanship formula and color-coding scheme in drawing the districts at issue in this case enacted in 2011 and kept unchanged in 2017. Tr. 991:9-992:6, 994:4-996:11; PX347; PX350; PX352; PX329 at 23, 27, 30 (Cooper Rebuttal Report). For example, Dr. Hofeller’s Maptitude file titled “NC House w New Raleigh - June 28,” which was last modified on June 30, 2011, contained Dr. Hofeller’s drafts of the 2011 House districts at issue in this case. Tr. 995:20-997:11; PX329 at 30-35; PX564. There, Dr. Hofeller scored the partisanship of each VTD using the results of the 2008 Presidential election and then colored each VTD based on those results, with Democratic-leaning VTDs shaded blue, Republican-leaning VTDs shaded red, and competitive VTDs shaded yellow and tan. Id. Plaintiffs’ Exhibit 353 below is an example of Dr. Hofeller’s use of this partisanship data to draw the 2011 House districts—in this example, to crack Democratic voters across House Districts 55, 68, and 69. 36 78. Legislative Defendants offered no additional files from Dr. Hofeller’s storage devices to rebut Dr. Chen’s and Dr. Cooper’s analyses. They offered no plausible alternative explanation of Dr. Hofeller’s intent as he drew the State’s House and Senate districts in 2011 and 2017. 3. 79. Plaintiffs’ Experts Established that the Plans Are Extreme Partisan Gerrymanders Designed to Ensure Republican Control The analysis and conclusions of Plaintiffs’ experts further establish that the 2017 Plans are extreme partisan outliers intentionally and carefully designed to maximize Republican advantage and to ensure Republican majorities in both chambers of the General Assembly. Three of Plaintiffs’ experts—Drs. Chen, Mattingly, and Pegden—employed computer simulations to generate alternative House and Senate plans to serve as a baseline for comparison to each enacted plan. Even though these experts employed different 37 methodologies, each expert found that the enacted plans are extreme outliers that could only have resulted from an intentional effort to secure Republican advantage on a statewide basis. Plaintiffs’ fourth expert, Dr. Christopher Cooper, explained how this gerrymandering was carried out across the State. The Court gives great weight to the analysis and conclusions, to the extent set forth below, of each of Plaintiffs’ experts individually, and the Court finds that the consistent findings of each of these experts, using different methodologies, powerfully reinforce that the 2017 Plans are extreme, intentional, and effective partisan gerrymanders. a. 80. Dr. Jowei Chen Plaintiffs’ expert Jowei Chen, Ph.D., is an Associate Professor in the Department of Political Science at the University of Michigan, Ann Arbor. Tr. 237:6-9. Dr. Chen has extensive experience in redistricting matters. Tr. 238:2-239:3 (Dr. Chen). By the admission of Intervenor Defendants’ own expert, Dr. Chen is one of the “foremost political science scholars on the question of political geography” and how it can impact the partisan composition of a legislative body. Tr. 2220:14-18 (Dr. Barber). Dr. Chen also helped pioneer the methodology of using computer simulations to evaluate the partisan bias of a redistricting plan, and he has published four peer-reviewed articles employing this approach since 2013. Tr. 240:1-241:2; PX2. The Court accepted Dr. Chen in this case as an expert in redistricting, political geography, and geographic information systems (“GIS”). Tr. 245:4-8. 81. Dr. Chen has presented expert testimony regarding his simulation methodology in numerous prior partisan gerrymandering lawsuits, and his analysis has been consistently credited and relied upon by the courts in these cases. Tr. 241:15-242:19; see League of Women Voters v. Commonwealth, 178 A.3d 737, 818 (Pa. 2018) (finding “Dr. Chen’s expert testimony” to be “[p]erhaps the most compelling evidence” in invalidating 38 Pennsylvania’s congressional plan as an unconstitutional partisan gerrymander); Raleigh Wake Citizens Ass’n v. Wake Cty. Bd. of Elecs., 827 F.3d 333, 344 (4th Cir. 2016) (“[T]he district court clearly and reversibly erred in rejecting Dr. Chen’s expert testimony.”); League of Women Voters of Mich. v. Benson, 373 F. Supp. 3d 867, 907 (E.D. Mich. 2019) (“[T]he Court has determined that Dr. Chen’s data and expert findings are reliable.”); Common Cause v. Rucho, 279 F. Supp. 3d 587, 666 (M.D.N.C.), vacated on other grounds, 138 S. Ct. 2679 (2018) (“Dr. Mattingly’s and Dr. Chen’s simulation analyses not only evidence the General Assembly’s discriminatory intent, but also provide evidence of the 2016 Plan’s discriminatory effects.”); City of Greensboro v. Guilford Cty. Bd. of Elecs., 251 F. Supp. 3d 935, 943 (M.D.N.C. 2017) (relying upon the “computer simulations by Dr. Jowei Chen” to find impermissible partisan intent). 82. Using his simulation methodology, Dr. Chen analyzed whether partisan intent predominated in the drawing of the 2017 Plans and subordinated the traditional nonpartisan districting principles of compactness and avoiding the splitting of municipalities and VTDs. Tr. 245:13-17, 248:6-18. Dr. Chen further analyzed the effects of the 2017 Plans on the number of Democratic-leaning House and Senate districts statewide. Tr. 247:6-10. 83. Based on his analysis, Dr. Chen concluded that partisan intent predominated over the traditional districting criteria in drawing the current House and Senate districts, that the Republican advantage under the 2017 Plans cannot be explained by North Carolina’s political geography, and that the effect of the 2017 Plans is to produce fewer Democratic-leaning districts than would exist if the map-drawing process had followed traditional districting principles. Tr. 246:18-22, 247:12-18, 248:20-249:1; PX1 at 3-4 (Chen Report). With respect to the effects in particular, Dr. Chen found that the gap between the enacted 2017 Plans and the nonpartisan simulated plans in terms of Democratic-leaning 39 districts gets wider in electoral environments more favorable to Democrats, and is widest around the point when Democrats would win majorities in the House or Senate under the simulated nonpartisan plans. Tr. 247:25-248:3, 296:7-24, 330:17-23. The Court gives great weight to Dr. Chen’s findings and, to the extent set forth below, adopts his conclusions. 84. In what Dr. Chen described as his Simulation Set 1, Dr. Chen programmed his algorithm to follow the traditional districting principles embodied within the Adopted Criteria. Tr. 281:12-16. In addition to following the equal population and contiguity requirements, as well as conforming to the same county groupings and number of county traversals that exist under the 2017 Plans, Dr. Chen programmed his algorithm to prioritize the traditional districting principles set forth in the Adopted Criteria of compactness, avoiding splitting municipalities, and avoiding splitting VTDs. Tr. 251:18259:10; PX1 at 10-18 (Chen report). 85. Dr. Chen explained that, other than the county traversals requirement, his algorithm did not attempt to “maximize or optimize” any one criterion. Tr. 262:24-263:3. Rather, the algorithm equally weighted the criteria of compactness, avoiding splitting municipalities, and avoiding splitting VTDs. Tr. 263:4-12. In creating districts within each county grouping, the algorithm considered thousands of random iterations, measuring for each proposed iteration whether the change would make the districts in the grouping better or worse on net across these three criteria. Tr. 261:18-263:19. The algorithm accepted a change only if it would improve the districts across these three criteria on net. Id. 86. In his Simulation Set 1, Dr. Chen ran the algorithm 1,000 times for each House county grouping and 1,000 times for each Senate county grouping, producing 1,000 unique statewide maps for both the House and the Senate. Tr. 263:23-264:16. 87. Beginning with the House, Dr. Chen compared the 1,000 simulated plans in his House Simulation Set 1 to the enacted 2017 House plan along a number of measures. 40 First, Dr. Chen compared the number of municipalities that the simulated and enacted plans split. The enacted House plan splits 79 municipalities. Tr. 266:22-269:15; PX1 at 38, 41 (Chen Report). The 1,000 plans in House Simulation Set 1 split a range of only 38 to 55 municipalities, with most splitting just 43 to 48 municipalities. Id. From this, Dr. Chen concluded with over 99.9% statistical certainty that the enacted House plan subordinates the traditional districting criterion of following municipal boundaries, and splits substantially more municipalities than would be split if the map-drawing process had prioritized, and not subordinated, this traditional districting principle. Tr. 269:21-270:4; PX1 at 38 (Chen Report). 88. Plaintiffs’ Exhibit 15 depicts the number of municipalities split under the enacted plan and the 1,000 simulations in House Simulation Set 1: 89. The Court finds that the enacted House plan subordinates to partisanship the traditional districting principle of avoiding the unnecessary splitting of municipalities. 41 The Court finds that the current House plan splits substantially more municipalities than would be split if the map-drawing process had not subordinated to partisanship this traditional districting principle. 90. Dr. Chen also compared the number of VTDs split in the enacted 2017 House plan and the 1,000 simulations in House Simulation Set 1. Dr. Chen found that, while the simulated House plans split between 6 and 18 VTDs, the enacted House plan splits 48 VTDs, more than four times as many as the vast majority of the simulations. Tr. 270:6271:3; PX1 at 38, 42 (Chen Report). From this, Dr. Chen concluded with over 99.9% statistical certainty that the enacted House plan subordinates the traditional districting criterion of following VTD boundaries, and splits far more VTDs than is reasonably necessary. Tr. 271:5-12. 91. Plaintiffs’ Exhibit 16 depicts the number of VTDs split under the enacted House plan and the 1,000 simulations in House Simulation Set 1: 42 92. The Court finds that the enacted House plan subordinates to partisanship the traditional districting principle of avoiding the unnecessary splitting of VTDs. The Court finds that the current House plan splits substantially more VTDs than would be split if the map-drawing process had not subordinated to partisanship this traditional districting principle. 93. Dr. Chen found the enacted House plan is also less compact than all 1,000 of his simulations in House Simulation Set 1. Dr. Chen employed the measures of compactness set forth in the Adopted Criteria, known as Reock and Polsby-Popper scores. Tr. 271:16-273:15; PX1 at 38 (Chen Report). For both measures, a higher score indicates that a plan’s districts are more compact. Id. Dr. Chen found that, as measured by both Reock and Polsby-Popper scores, the compactness of the enacted House plan is outside the range of scores produced by the 1,000 simulated House plans. Id. From this, Dr. Chen concluded with over 99% statistical certainty that the enacted House plan subordinates the traditional districting criterion of compactness, and that the current districts are less compact than they would be under a map-drawing process that prioritizes and follows the traditional districting criteria. Tr. 273:18-274:4. 43 94. Plaintiffs’ Exhibit 14 depicts the compactness of the enacted House plan and the 1,000 simulations in House Simulation Set 1: 95. The Court finds that the enacted House plan subordinates to partisanship the traditional districting principle of compactness. The Court finds that the current House districts are less compact than they would be under a map-drawing process that had not subordinated to partisanship this traditional districting criteria. 96. To compare the partisanship of his simulated plans to the enacted House and Senate plans, Dr. Chen used Census Block-level election results from recent statewide elections in North Carolina. Tr. 274:5-275:20; PX1 at 19-20 (Chen Report). For most of his analysis, Dr. Chen used the following ten statewide elections: 2010 U.S. Senate, 2012 U.S. President, 2012 Governor, 2012 Lieutenant Governor, 2014 U.S. Senate, 2016 U.S. President, 2016 U.S. Senate, 2016 Governor, 2016 Lieutenant Governor, and 2016 Attorney 44 General. Id. Dr. Chen provided several reasons for his choice of these ten statewide elections. 97. First, Representative Lewis indicated at an August 10, 2017, hearing that these ten statewide elections would be the elections that the Joint Redistricting Committees would use to evaluate the 2017 Plans. Tr. 275:8-11; PX1 at 20 (Chen Report). 98. Second, Dr. Chen testified that it is well-accepted in academic literature and in redistricting practice that statewide elections, rather than legislative elections, provide the best basis for measuring the partisanship of a district and for comparing the partisanship of districts across alternative possible plans. Tr. 276:3-27:18; PX1 at 19-20 (Chen Report). Dr. Chen explained that legislative elections, such as state House and state Senate elections, do not provide a sound basis for measuring the partisanship of Census Blocks and districts because the results of legislative elections can be skewed by various factors. Id. For instance, if districts are gerrymandered or otherwise uncompetitive, the results of the legislative elections can be biased by the district boundaries in a way that they would not be under an alternative plan. Id. As Dr. Chen noted, the General Assembly did not have Dr. Hofeller use legislative elections to measure partisanship in drawing the 2017 Plans. Tr. 277:9-14. 99. Third, Dr. Chen testified he did not use party registration to measure the partisanship of districts because it is well-known in academic literature and in the redistricting community that party registration is not a reliable indicator of actual partisan voting behavior. Tr. 277:19-278:10. That is particularly true in southern states such as North Carolina, where many registered Democrats now consistently vote for Republicans. Id. As Dr. Chen again noted, Legislative Defendants did not have Dr. Hofeller use party registration to measure partisanship in drawing the 2017 Plans. Tr. 278:11-15. 45 100. The Court finds the use of statewide elections by Plaintiffs’ experts to measure the partisanship of simulated and enacted districts is a reliable methodology. 101. To measure the partisanship of his simulated districts and the enacted districts, Dr. Chen determined the set of Census Blocks that comprise each district. Tr. 278:24-283:10; PX1 at 20-22 (Chen Report). Dr. Chen then aggregated the elections results from the ten 2010-2016 statewide elections for that set of Census Blocks. Id. In other words, Dr. Chen calculated the total votes cast for Democratic candidates in those ten 20102016 statewide elections across the relevant set of Census Blocks and the total votes cast for Republican candidates in that set of Census Blocks. Id. If there were more votes in aggregate for the Democratic candidates, Dr. Chen classified the district as a Democratic district, and if there were more votes for the Republican candidates, Dr. Chen classified the district as a Republican district. Id. 102. Using this measure of partisanship, Dr. Chen compared the number of Democratic districts under the enacted 2017 House plan and under the 1,000 simulated plans in his House Simulation Set 1. While the enacted House plan has 42 Democratic districts using the 2010-2016 statewide elections, not a single one of the 1,000 simulated plans produce so few Democratic districts. Tr. 285:15-287:8; PX1 at 29-30 (Chen Report). The vast majority of simulated plans produce 46 to 51 Democratic districts using the 20102016 statewide elections, with the two most common outcomes in the simulations being 46 or 47 Democratic districts—i.e., four or five more Democratic districts than exist under the enacted House plan. Id. From these results, Dr. Chen concluded with over 99% statistical certainty that the current House plan is an extreme partisan outlier, and one that could not have occurred under a districting process that adhered to the traditional districting criteria. Tr. 287:2-8; PX1 at 29 (Chen Report). 46 103. Plaintiffs’ Exhibit 9 depicts the distribution of Democratic seats under the enacted House plan and under the 1,000 simulations in Dr. Chen’s House Simulation Set 1: 104. Dr. Chen explained that the number of Democratic districts estimated for his simulated plans is depressed by the fact that the 2010-2016 statewide elections he used were relatively favorable for Republicans. Tr. 284:1-285:12; PX1 at 29 (Chen Report). Three of the four elections cycles in this period—2010, 2014, and 2016—were favorable for Republicans nationally. Id. Consequently, the aggregate Democratic share of the twoparty vote across the ten statewide elections in the 2010-2016 composite used by Dr. Chen was just 47.92%. Id. 105. Dr. Chen also measured the number of Democratic districts that would exist under his simulated plans and the enacted House plan under electoral environments that 47 are more neutral or even favorable to Democrats. Tr. 287:15-22. First, Dr. Chen analyzed the number of Democratic districts using only the 2016 Attorney General election, which was a near tie. Tr. 287:19-289:14; PX1 at 29 (Chen Report). Using the 2016 Attorney General results, the enacted House plan produces 44 Democratic districts, while the 1,000 simulated House plans produce 48 to 55 Democratic districts, with the most common outcome being 52 Democratic districts. Tr. 287:24-289:14; PX119; PX1 at 29, 174, A1. The gap between the enacted House plan and the simulated plans therefore grows to eight Democratic seats in the most common outcome under the neutral electoral environment that was the 2016 Attorney General election. Id. 106. Dr. Chen also performed a “uniform swing” analysis to compare the enacted plan and the simulated plans under different electoral environments. Uniform swing analysis is a common technique used in academic literature and the redistricting community to measure how districts would perform under varying electoral conditions. Tr. 289:25-290:8. For his uniform swing analysis, Dr. Chen started with the Democratic vote share in every enacted and simulated district using the 2010-2016 statewide elections, and then increased or decreased the Democratic vote share uniformly in every district in 0.5% increments. Tr. 290:4-296:3. 107. Dr. Chen’s uniform swing analysis revealed a “striking trend.” Tr. 296:7. As the uniform swing increases in the direction of more favorable Democratic performance, the gap between the number of Democratic districts under the enacted plan and the simulated plans grows more and more. Tr. 296:7-20. In other words, “in electoral environments that are more favorable to Democrats, the gap between the enacted plan and all of the computersimulated plans is widened.” Tr. 296:18-20. 108. Plaintiffs’ Exhibit 10 below depicts Dr. Chen’s uniform swing analysis for House Simulation Set 1. The starting point is the row on the vertical axis for “47.92%,” 48 which represents the statewide Democratic vote share under the ten 2010-2016 statewide elections. Tr. 290:23-296:3; PX1 at 31-33 (Chen Report). Each row above this point represents the results when increasing the Democratic vote share in every enacted and simulated district by increments of 0.5%. Id. The red stars in each row represent the number of Democratic districts under the enacted 2017 House plan, and the numbers to the right of each red star represent the number of simulations (out of 1,000) that produce the number of Democratic districts found on the horizontal axis below. Id. For instance, for the starting row of a 47.92% statewide Democratic vote share, the enacted plan (the red star) produces 42 Democratic districts, six simulated plans produce 43 Democratic districts, 48 simulated plans produce 44 Democratic districts, 172 simulated plans produce 45 Democratic districts, and so on. Id. 49 109. Dr. Chen found that the gap between the enacted and simulated plans not only grew as the electoral environment became more favorable for Democrats, but the gap is “widest” at the point when Democrats would start winning a majority of House seats under the simulated plans. Tr. 296:20-297:21. Plaintiffs’ Exhibit 11 (Figure U2) below depicts Dr. Chen’s results for a uniform swing corresponding to a statewide Democratic vote share of 52.42%. In this scenario, the enacted House plan contains only 48 Democratic districts, but roughly one-third of the 1,000 simulations produce 60 or more Democratic districts, with a 60-60 tie being the second most common outcome. Tr. 298:2-299:7. Plaintiffs’ Exhibit 12 (Figure U3) below depicts Dr. Chen’s results for a uniform swing corresponding to a statewide Democratic vote share of 52.92%. In this scenario, there are 60 or more Democratic districts in nearly two-thirds of the simulations, and Democrats would win a majority (61 or more seats) in more than 40% of the simulations. Tr. 299:16301:12. But Democrats would hold just 51 districts under the enacted House plan. Id. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 50 [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 51 110. Dr. Chen analyzed the type of electoral environment that would produce 55 Democratic districts under the enacted House plan, which is the number of House districts that Democrats won in 2018. Tr. 301:16-302:14. Dr. Chen found that, in the type of electoral environment that would produce 55 Democratic districts under the enacted plan in his uniform swing analysis, Democrats would win 60 or more House districts in over 99% of his simulated plans, and would win a majority of districts in over 98% of the simulated plans. Id.; PX10. In other words, while Democrats improved their seat share in 2018, they may well have won a majority had a nonpartisan plan been in place. 111. The Court finds Dr. Chen’s uniform swing analysis to be substantial evidence of the intent and effects of Legislative Defendants’ partisan gerrymander. The analysis establishes that the effects of the gerrymander are most extreme in electoral environments that are better for Democrats, specifically in electoral environments where Democrats could 52 win a majority of House seats under a nonpartisan map. Dr. Chen’s uniform swing analysis is persuasive evidence the enacted House plan was designed specifically to ensure that Democrats would not win a majority of House seats under any reasonably foreseeable electoral environment. 112. The Court further gives weight to Dr. Chen’s overall conclusions from his House Simulation Set 1. Dr. Chen concluded with over 99% statistical certainty that partisanship predominated in the drawing of the enacted House plan and subordinated the traditional districting criteria of compactness, avoiding splitting municipalities, and avoiding splitting VTDs. Tr. 307:12-24. The Court adopts these conclusions and finds the current House districts, regardless of whether they were drawn in 2017 or 2011, subordinated these three traditional districting criteria in order to accomplish Legislative Defendants’ predominant partisan goals. 113. In his House Simulation Set 2, Dr. Chen programmed his algorithm to add avoiding pairing incumbents as an additional criterion. Dr. Chen performed this analysis to determine whether a hypothetical, nonpartisan effort to avoid pairing the incumbents in place at the time each of the relevant districts was drawn could account for the extreme partisan bias and subordination of traditional districting principles that Dr. Chen found in his Simulation Set 1. Tr. 308:15-21. Dr. Chen programmed his algorithm in Simulation Set 2 to avoid pairing the maximum number of incumbents possible who were in office at the time of the relevant redistrictings, and to ensure that the very same incumbents who were not paired with another incumbent under the enacted plans were not paired in the simulations. Tr. 308:3-14, 310:21-311:16; PX1 at 43 (Chen Report). 114. The method by which Dr. Chen avoided pairing incumbents in Simulation Set 2 is consistent with the Adopted Criteria’s incumbency protection provision. The Court gives no weight to Legislative Defendants’ contention that the Adopted Criteria required 53 incumbency protection beyond merely avoiding pairing incumbents; namely, that the Adopted Criteria required creating districts politically favorable to incumbents. As Representative Lewis stated, this criterion was interpreted as simply an intent to avoid pairing incumbents. See FOF ¶ 28. At the time of the 2017 redistricting, Republicans held supermajorities in both chambers of the General Assembly. Hence, seeking to enhance the reelection chances of every incumbent, Democrat and Republican alike, would have been a means of seeking to lock-in the Republican supermajorities. It would also have been particularly inappropriate to seek to preserve the “core” of the existing districts, as Legislative Defendants’ expert Dr. Brunell suggested, since many of the existing districts had been found to constitute illegal racial gerrymanders. 115. In addition, the Court finds that Legislative Defendants did not seek to protect Democratic and Republican incumbents alike in a neutral manner. For example, in Buncombe County, the enacted plan paired two Democratic incumbents who were in office at the time these House districts were drawn in 2011, but Dr. Chen’s algorithm was able to avoid pairing these two Democratic incumbents in all 1,000 of his simulations. Tr. 312:14313:9; PX1 at 45, 47 (Chen Report). Legislative Defendants thus unnecessarily paired these two Democratic incumbents in creating the Buncombe County House districts, ensuring that one of the two would not be reelected. Id. Dr. Hofeller’s Excel files further show that, in 2017, Dr. Hofeller focused solely on concerns for Republican incumbents and not Democratic incumbents. FOF § B.2.a. Dr. Hofeller analyzed “Pressure Points for GOP Incumbents” in both the House and the Senate, but performed no similar analysis for Democratic incumbents. Id. 116. Based on his House Simulation Set 2 analysis, Dr. Chen found that a nonpartisan effort to avoid pairing incumbents cannot explain the extreme partisan bias of the enacted House plan or its subordination of traditional districting criteria. Dr. Chen 54 found that the enacted House plan is an extreme outlier with respect to the number of Democratic districts it produces, the number of municipalities and VTDs it splits, and the compactness of its districts compared to the 1,000 simulated plans in House Simulation Set 2. Tr. 313:11-317:24; PX7; PX18; PX23; PX1 at 44-56 (Chen Report). The Court gives weight to Dr. Chen’s findings in House Simulation Set 2 and finds that a nonpartisan effort to protect incumbents cannot explain the extreme partisan bias and subordination of traditional districting principles in the enacted House plan. 117. For the Senate, Dr. Chen ran two sets of 1,000 simulations just as he did for the House. Tr. 318:11-319:9. Dr. Chen’s Senate Simulation Set 1 applied the same algorithm used for House Simulation Set 1, prioritizing and equally weighting the traditional districting principles within the Adopted Criteria of compactness and avoiding splitting municipalities and VTDs.3 Dr. Chen ran his algorithm 1,000 times for each Senate county grouping, producing 1,000 unique statewide plans in Senate Simulation Set 1. Tr. 319:10-320:10. 118. With respect to municipal splits, Dr. Chen found the enacted Senate plan splits 25 municipalities, while the 1,000 simulated plans in Senate Simulation Set 1 split between just 8 and 12 municipalities. Tr. 320:12-321:9; PX1 at 69, 71 (Chen Report). From this, Dr. Chen concluded with over 99.9% statistical certainty that the enacted Senate plan subordinates the traditional districting criterion of following municipal boundaries, and splits far more municipalities than is reasonably necessary. Tr. 321:12-17. 3 Dr. Chen used the same Senate county groupings that exist under the enacted Senate plan, minimized the number of county traversals, and applied the Adopted Criteria’s equal population and contiguity requirements. Tr. 318:11-319:9. 55 119. Plaintiffs’ Exhibit 34 depicts the number of municipalities split under the enacted Senate plan and the 1,000 simulations in Senate Simulation Set 1: 120. The Court finds the enacted Senate plan subordinates to partisanship the traditional districting principle of avoiding the unnecessary splitting of municipalities. The Court finds the current Senate districts split substantially more municipalities than would be split if the map-drawing process had not subordinated to partisanship this traditional districting principle. 121. With respect to VTDs, Dr. Chen found the enacted Senate plan splits 5 VTDs, while his simulations split between 0 and 3 VTDs. Tr. 321:19-322:9; PX1 at 69, 72 (Chen Report). From this, Dr. Chen concluded with over 99.9% statistical certainty that the enacted Senate plan subordinates the traditional districting criterion of following VTD boundaries, and splits more VTDs than is reasonably necessary. Tr. 322:12-15. 56 122. Plaintiffs’ Exhibit 35 depicts the number of VTDs split under the enacted Senate plan and the 1,000 simulations in Senate Simulation Set 1: 123. The Court finds the enacted Senate plan subordinates to partisanship the traditional districting principle of avoiding the unnecessary splitting of VTDs. The Court finds the current Senate districts split more VTDs than would be split if the map-drawing process had not subordinated to partisanship this traditional districting principle. 124. Dr. Chen found the enacted Senate plan is also less compact than all 1,000 of his Senate simulations. Using both the Reock and Polsby-Popper measures of compactness, all 1,000 simulated plans in Senate Simulation Set 1 are more compact than the enacted Senate plan. Tr. 322:17-324:3; PX1 at 67-69 (Chen Report). From this, Dr. Chen concluded with over 99% statistical certainty that the enacted Senate plan subordinates the traditional districting criterion of compactness, and that the current districts are less 57 compact than they would be under a map-drawing process that prioritizes and follows the traditional districting criteria. Tr. 324:6-15. 125. Plaintiffs’ Exhibit 33 depicts the compactness of the enacted Senate plan and the 1,000 simulations in Senate Simulation Set 1: 126. The Court finds the enacted Senate plan subordinates to partisanship the traditional districting principle of compactness. The Court finds the current Senate districts are less compact than they would be under a map-drawing process that had not subordinated to partisanship this traditional districting criteria. 127. As with the House, Dr. Chen compared the partisanship of his simulated Senate plans to the partisanship of the enacted Senate plan using the same ten statewide elections from 2010-2016 that Representative Lewis stated would be used. Tr. 324:16325:5. 128. Using the 2010-2016 statewide elections, Dr. Chen found that the enacted Senate plan produces 18 Democratic districts. Tr. 325:7-326:11; PX1 at 57, 60 (Chen 58 Report). In contrast, none of the 1,000 simulated plans produce such an outcome. Id. The simulated Senate plans produce 19 to 21 Democratic districts using the 2010-2016 statewide elections, with the most common outcome in the simulations being 20 Democratic districts—i.e., two more Democratic districts than exist under the enacted Senate plan. Id. From these results, Dr. Chen concluded with over 99% statistical certainty that the current Senate plan is an extreme partisan outlier, and one that could not have occurred under a districting process that adhered to the traditional districting criteria. Tr. 326:12-21; PX1 at 59 (Chen report). 129. Plaintiffs’ Exhibit 28 depicts the distribution of Democratic seats under the enacted Senate plan and under the 1,000 simulations in Senate Simulation Set 1: 59 130. Like he did for the House, Dr. Chen measured the number of Democratic districts that would exist under his simulated plans and the enacted plan under electoral environments that are more neutral or even favorable to Democrats. Dr. Chen again analyzed the number of Democratic districts when using just the 2016 Attorney General election, which was a near tie. Tr. 327:8-11; PX121; PX1 at 59, 61, A3 (Chen Report). Dr. Chen found that the enacted Senate plan produces 20 Democratic districts using the 2016 Attorney General results, while the 1,000 simulated Senate plans most commonly produce 23 Democratic districts under the 2016 Attorney General results. Tr. 328:1-13. The gap between the enacted Senate plan and the simulated plans therefore grows to three Democratic seats in the most common outcome under the neutral electoral environment of the 2016 Attorney General election. Id. 131. Dr. Chen also performed a uniform swing analysis to compare the enacted Senate plan to the simulated Senate plans under different electoral environments. Just as he did for the House, in his uniform swing analysis for the Senate, Dr. Chen started with the Democratic vote share in every enacted and simulated district using the 2010-2016 statewide elections and then increased or decreased the Democratic vote share uniformly in every district in 0.5% increments. Tr. 328:25-329:7. 132. Dr. Chen found the same trend in his uniform swing analysis of the Senate that he found for the House. Tr: 330:7-23. He found that as he increases the uniform swing in the more Democratic direction, the gap between the number of Democratic districts under the enacted Senate plan and the simulated plans grows. Id. And the gap again becomes widest around the points where Democrats would come close to gaining a majority or would actually gain a majority under the nonpartisan simulated plans. Id. 133. Plaintiffs’ Exhibit 29 below depicts Dr. Chen’s uniform swing analysis for the Senate. The red stars again reflect the number of Democratic districts under the enacted 60 Senate plan and the numbers to the right of the red stars reflect the number of simulations (out of 1,000) that produce the number of Democratic districts listed on the horizontal axis. 134. Plaintiffs’ Exhibit 30 (Figure U8) below depicts Dr. Chen’s Senate results for a uniform swing corresponding to a statewide Democratic vote share of 51.92%. The figure reveals that, in this scenario, the enacted Senate plan contains only 22 Democratic districts, but the vast majority of simulations would give Democrats a tie or an outright majority in the Senate. Tr. 331:2-332:23. Plaintiffs’ Exhibit 31 (Figure U9) below depicts Dr. Chen’s Senate results for a uniform swing corresponding to a statewide Democratic vote share of 52.42%. In this environment, Democrats would win half or more of the districts in over 95% of the simulations and would win an outright majority in over 62% of the simulations. Tr. 333:7-334:2. Yet, under the enacted Senate plan, Democrats would hold just 22 Senate districts in this scenario. Id. 61 [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 62 135. Dr. Chen also analyzed the type of electoral environment that would produce 21 Democratic districts under the enacted plan, which is the number of Senate districts that Democrats won in 2018. Tr. 334:3-335:7. Dr. Chen found that, in the type of environment that would produce 21 Democratic districts under the enacted plan in his uniform swing analysis, Democrats would win 25 or more Senate districts in the vast majority of simulations. Id.; PX29. In other words, while Democrats improved their seat share in 2018, they may well have won a majority had a nonpartisan plan been in place. 136. The Court again finds Dr. Chen’s uniform swing analysis to be substantial evidence of the intent and effects of the partisan gerrymander. Dr. Chen’s analysis establishes that the effects of the gerrymander are most extreme in electoral environments that are better for Democrats, and in particular in environments under which Democrats could win a majority of Senate seats under a nonpartisan map. Dr. Chen’s uniform swing 63 analysis is persuasive evidence that the enacted Senate plan was designed specifically to ensure that Democrats would not win a majority of Senate seats under any reasonably foreseeable electoral environment. 137. The Court further gives weight to Dr. Chen’s overall conclusions from his Senate Simulation Set 1. Dr. Chen concluded with over 99% statistical certainty that partisanship predominated in the drawing of the enacted Senate plan and subordinated the traditional districting criteria of compactness, avoiding splitting municipalities, and avoiding splitting VTDs. Tr. 336:22-337:7. The Court adopts these conclusions and finds the current Senate districts, regardless of whether they were drawn in 2017 or 2011, subordinated these three traditional districting criteria in order to accomplish Legislative Defendants’ predominant partisan goals. 138. Dr. Chen generated 1,000 more simulated plans in his Senate Simulation Set 2, adding the same incumbency criteria he used for the House. Dr. Chen found that a hypothetical, nonpartisan effort to avoid pairing the incumbents in place at the time each of the relevant districts was drawn could not explain the extreme partisan bias of the enacted Senate plan and its subordination of traditional districting principles. Tr. 341:18-342:8. Dr. Chen found the enacted Senate plan is an extreme outlier with respect to the number of Democratic districts it produces, the number of municipalities and VTDs it splits, and the compactness of its districts compared to the 1,000 simulated plans in Senate Simulation Set 2. Tr. 337:8-341:22, 26, 37, 42; PX1 at 73-85 (Chen Report). The Court gives weight to Dr. Chen’s findings in Senate Simulation Set 2 and finds a nonpartisan effort to protect incumbents cannot explain the extreme partisan bias and subordination of traditional districting principles in the enacted Senate plan. 139. The Court also gives weight to and adopts Dr. Chen’s conclusions that the partisan bias of the 2017 House and Senate Plans cannot be explained by North Carolina’s 64 political geography, meaning the geographic locations of Republican and Democratic voters. Tr. 307:3-11, 336:11-19. Political geography can create a natural advantage for Republicans in winning seats where, for example, Democratic voters are clustered in urban areas. Tr. 304:9-18; PX1 at 7-8 (Chen Report). But Dr. Chen designed his simulations with the specific purpose of accounting for North Carolina’s political geography and any other built-in advantages either party may have in redistricting. Tr. 304:19-305:19; see PX1 at 78 (Chen Report). The simulations build districts using the same Census geographies and population data that existed when the enacted plans were drawn; thus, the simulated plans capture any natural advantage one party may have had based on population patterns when the General Assembly passed the enacted plans. Id. 140. Dr. Chen found that Republicans may have a small degree of natural advantage in winning districts in both the House and Senate; Dr. Chen’s analysis suggests that even under his nonpartisan plans, Democrats may win less than 50% of the seats when they win 50% of the votes. Tr. 305:21-307:2, 335:17-336:10; PX1 at 36, 66 (Chen Report). But Dr. Chen concluded, and the Court finds, that the enacted House and Senate plans are extreme partisan outliers compared to Dr. Chen’s simulations that account for political geography and any other built-in advantages Republicans may have, and thus political geography and other built-in advantages cannot explain the enacted plans’ extreme partisan bias. Tr. 307:3-11, 336:11-19. 141. The Court also rejects Legislative Defendants’ critiques of the way in which Dr. Chen’s simulation algorithm applied the traditional districting principles of compactness and avoiding splitting municipalities and precincts. 142. Dr. Chen’s interpretation and application of the traditional districting principles is fully consistent with the guidance provided by Legislative Defendants at the time of the 2017 redistricting. At the first public hearing after the draft plans were 65 unveiled, Representative Lewis explained the Adopted Criteria meant “trying to keep towns, cities and precincts whole where possible.” PX607 at 10:5-6. Representative Lewis made similar statements at the committee hearing where the Adopted Criteria were proposed and debated; he asserted, for example, that the criterion regarding municipal splits “says that the map drawer may and rightfully should consider municipality boundaries when they can.” PX603 at 67:16-18. Representative Lewis added that “municipality, precinct lines are things that are all community-of-interest-type things that we’re going to seek to preserve.” Id. at 77:12-14. Representative Lewis did not qualify in these statements that the Redistricting Committees would seek only to promote these traditional principles up to a point, or would seek to intentionally split some minimum number of municipalities and VTDs. 143. The Court further gives weight to Dr. Chen’s testimony that his application of these criteria is consistent with generally accepted redistricting principles and practice. Dr. Chen testified that no jurisdiction in the country prefers to split a higher number of municipalities or VTDs or wants less compact districts. Tr. 603:2-605:21, 774:5-21. Nor does any jurisdiction seek to split some minimum number of municipalities or VTDs or impose a cap on how compact the districts should be. Id. 144. Legislative Defendants did not introduce persuasive evidence of nonpartisan reasons why the enacted plans split particular municipalities or VTDs or made particular districts less compact. 145. The Court also rejects any suggestion that Dr. Chen should not have applied these traditional districting criteria in simulating county groupings that were drawn in 2011 because these principles were not expressly stated as official criteria during the 2011 redistricting process. See Tr. 629:19-636:12. The principles of compactness and avoiding split municipalities and VTDs were traditional districting criteria since well before 2011. 66 Tr. 776:8-777:8; see, e.g., Stephenson v. Bartlett, 355 N.C. 354, 371, 562 S.E.2d 377, 389 (2002). That the General Assembly did not list these traditional districting principles as official criteria in 2011 does not change the fact that Legislative Defendants subordinated these principles to partisan considerations in drawing the 2011 districts at issue in this case. Id. And the fact that the General Assembly reenacted these districts without change in 2017 does not mean these districts no longer subordinate traditional districting principles to partisan considerations. Id. 146. Dr. Chen’s analysis demonstrates the current districts subordinate these nonpartisan traditional principles to partisan intent. b. 147. Dr. Mattingly Jonathan Mattingly, Ph.D., is a North Carolina native, the chairman of the Duke University Mathematics Department, and the James B. Duke Professor of Mathematics at Duke University. Tr. 1080:7-20. He also is a professor in the Duke Statistics Department. Id. Dr. Mattingly was accepted as an expert in applied mathematics, probability, and statistical science. Tr. 1083:1-10. 148. Dr. Mattingly developed his method of evaluating partisan gerrymandering in his academic research. Tr. 1086:20-24. He has since created a project at Duke called “Quantifying Gerrymandering.” Tr. 1084:9-1085:4. In the one previous case in which Dr. Mattingly testified, a federal partisan gerrymandering case relating to North Carolina’s congressional districts, the federal court credited Dr. Mattingly’s testimony and concluded his analysis “provide[d] strong evidence” of partisan gerrymandering. Rucho, 279 F. Supp. 3d at 644. The court found his simulations “not only evidence[d] the General Assembly’s discriminatory intent, but also provide[d] evidence of the 2016 Plan’s discriminatory effects.” Id. at 666. 67 149. For this case, Dr. Mattingly generated a collection, or “ensemble,” of nonpartisan, alternative redistricting maps using the Markov chain Monte Carlo computer algorithm, which is a well-established algorithm dating back at least to the Manhattan Project. Tr. 1089:11-24; Tr. 1090:19-22. Dr. Mattingly generated approximately 1.1 x 10108 statewide maps in the House (of which 6.6 x 1086 were unique), and approximately 3.7 x 1093 statewide maps in the Senate (of which 5.3 x 1030 were unique). Tr. 1090:1-14; PX359 at 4. The number of maps that Dr. Mattingly generated is greater than the number of atoms in the known universe. Tr. 1090:12-14. 150. To generate the maps, Dr. Mattingly used all of the nonpartisan redistricting criteria identified by the General Assembly in its Adopted Criteria. The Markov chain Monte Carlo algorithm that Dr. Mattingly employed ensured that the collection of maps was a random and representative sample from the distribution of nonpartisan maps that adhere to North Carolina’s political geography and nonpartisan redistricting criteria. Tr. 1094:5-1095:3. All of Dr. Mattingly’s simulated maps followed North Carolina’s Whole County Provision and split no counties that were kept whole under the enacted plans; he ensured population deviations were within the 5% threshold; he required contiguity; and he tuned his algorithm to ensure that the nonpartisan qualities of the simulated maps were similar to the nonpartisan qualities of the enacted map with respect to compactness and the number of counties, municipalities, and precincts split. Tr. 1091:3-1093:1; PX359 at 3-4. Dr. Mattingly did not try to optimize or maximize any particular criterion such as compactness; instead, he took a random, representative sample of the distribution of all maps that are comparable to the enacted maps in terms of compactness and municipal splits. Tr. 1091:3-23. 151. The Court finds that Dr. Mattingly’s simulated maps provide a reliable and statistically accurate baseline against which to compare the 2017 Plans. Tr. 1089:11-24. 68 Dr. Mattingly’s collection of nonpartisan maps tracked all the nonpartisan criteria adopted by the Committees. By comparing Dr. Mattingly’s simulated plans to the enacted plans, the Court can reliably assess whether the characteristics and partisan outcomes under the enacted plans could plausibly have resulted from a nonpartisan process or be explained by North Carolina’s political geography. The Court can also reliably assess whether the enacted plans reflect extreme partisan gerrymanders. The partisan bias Dr. Mattingly identified by comparing the enacted plans to his nonpartisan ensemble of plans could not be explained by political geography or natural packing. Tr. 1095:9-1096:8. Moreover, Dr. Mattingly’s analysis did not rest on any assumption about proportional representation. Tr. 1132:6-1133:5; Tr. 1103:24-1104:5. 152. After creating a representative sample of hundreds of trillions of nonpartisan maps, Dr. Mattingly used votes from 17 prior North Carolina statewide elections to compare the partisan performance and characteristics of the 2017 Plans to the simulated plans. Dr. Mattingly chose all major statewide elections from 2008-2016 that were available to him, and those 17 elections demonstrated a range of Democratic support and Republican support and a range of spatial structures and vote patterns. Tr. 1097:8-1098:8; PX487 at 5. 153. The elections Dr. Mattingly considered and their statewide Democratic vote share are listed in the table below (PX778 at 7; Tr. 1097:8-1098:8): 69 154. Dr. Mattingly concluded that the 2017 Plans displayed a “systematic, persistent bias toward the Republican Party, both on the statewide level and on the county cluster level.” Tr. 1087:22-25. He concluded that the enacted plans were “extreme partisan outlier[s]” when compared to maps that respect the political geography of North Carolina and are similar to the enacted plans in terms of the nonpartisan Adopted Criteria such as compactness and splitting municipalities. Tr. 1088:1-7. He concluded that the “extreme partisan bias” was durable and persisted across a broad range of possible voting patterns and election results. Tr. 1088:1-7. He concluded that the gerrymander was particularly effective at preventing Democrats from breaking the Republican supermajority in both chambers when they would expect to do so under a nonpartisan plan, and from breaking 70 the Republican majority in both chambers when they would expect to do so under a nonpartisan plan. Tr. 1088:8-11. And Dr. Mattingly concluded that the probability that the General Assembly would have enacted the 2017 Plans without intentionally searching for such a biased plan was “astronomically small.” Tr. 1088:12-14, Tr. 1158:3-8. The Court gives great weight to those conclusions. 155. With respect to the Senate, Dr. Mattingly concluded that the enacted Senate plan shows a systematic bias toward the Republican Party. Tr. 1110:22-1111:3. In 15 of the 17 elections he considered, the enacted Senate plan produces an atypical bias toward the Republican Party with respect to the number of expected Democrat and Republican seats using the results of these prior statewide elections. Tr. 1116:2-12. The probability of seeing such a consistent pro-Republican bias across so many elections was 0.005%, Tr. 1116:18-21; PX487 at 23, meaning that the chance the General Assembly would have picked such a partisan map if it were not looking for it is five in a million, Tr. 1116:22-1117:2. 156. Dr. Mattingly concluded that the enacted Senate plan is an extreme outlier not just with respect to how consistently it favors Republicans, but with respect to the amount by which it favors Republicans. PX363 (Mattingly Report Figure 3). The enacted map caused Democrats to lose between 2 to 3 seats in the Senate in 13 of the 17 elections that Dr. Mattingly analyzed. Id. The Court finds this seat deviation to be significant. Tr. 1106:12-15. 157. Dr. Mattingly concluded that the 2017 Senate Plan’s extreme partisan bias was responsible for creating firewalls protecting the Republican supermajority and majority in the Senate. He plotted the results of the statewide elections using the enacted Senate plan and his nonpartisan simulations (PX362). Tr. 1106:17-1110:4. He ordered the elections vertically from bottom (most Republican vote share) to top (most Democratic vote share), and then plotted the number of seats that Democrats would expect to receive under 71 the nonpartisan plans using blue histograms. Id. Using nonpartisan maps, the Democratic seat count would be expected to fall in the tallest part of the blue histogram. Tr. 1108:7-24. Dr. Mattingly used purple dots to report how many seats Democrats would win in the Senate using the results of each statewide election under the enacted Senate plan. Tr. 1109:3-10. Dr. Mattingly then used three vertical dotted lines to represent the point at which Democrats would break the Republican supermajority, the Republican majority, or win a supermajority themselves. Tr. 1111:5-24.4 If the enacted plan is a pro-Republican outlier, the purple dot is to the left of the blue histogram (meaning the enacted plan elects fewer Democratic seats). If a purple dot is to the left of the Republican supermajority or majority line, and the bulk of the blue histogram is to the right, that is an election in which the enacted plan protects the Republican supermajority or majority where Democrats would break the firewalls in a nonpartisan plan. Tr. 1111:5-1112:24. Dr. Mattingly plotted only 13 of the 17 elections he considered in PX362 for visual clarity reasons, Tr. 1115:1-12, but he provided all the data for all 17 elections in Figure 3 (PX363) and Table 3 of his report (PX417). 4 72 158. Plaintiffs’ Exhibit 362 is reproduced below: 159. Dr. Mattingly’s analysis demonstrates that the enacted Senate plan creates two “firewalls,” protecting Republican supermajorities and majorities which Democrats would break under a nonpartisan plan. Dr. Mattingly testified that, in elections where 73 Democrats win enough votes that they would typically be expected to break the Republican supermajority under nonpartisan plans, the Republicans win the supermajority in the enacted plan. Tr. 1112:8-24. This is visually demonstrated by Plaintiffs’ Exhibit 362, which shows that the Democratic seat count in the enacted plan consistently stays to the left of the supermajority line even as the Democratic vote share rises and the nonpartisan plans break through the Republican supermajority line. PX362. In many cases the enacted plan is completely outside the distribution of nonpartisan plans. Tr. 1112:8-24. 160. The results of the Attorney General 2016 election illustrate Dr. Mattingly’s conclusion that the enacted map is an extreme, pro-Republican partisan gerrymander. Tr. 1114:9-11. This was a relatively even election where Democrats won 50.20% of the statewide vote, and in 99.999% of the nonpartisan maps, the Democrats broke the Republican supermajority. But, using the results of this election, the enacted map preserves the Republican supermajority. Tr. 1112:25-1114:11. 161. Overall, in 5 of the 17 elections that Dr. Mattingly considered, the Democrats would have almost certainly broken the Republican supermajority in the nonpartisan plans but failed to do so under the enacted plan (the 2012 Lieutenant Governor; 2016 President, 2008 President, 2016 Governor, and 2016 Attorney General elections). PX363; PX487 at 25 (Mattingly Rebuttal Report). In two others (the 2014 U.S. Senate and 2012 President elections), the Democrats would have had a chance of breaking the Republican supermajority in the nonpartisan plans, but never do in the enacted plan. PX362; PX417. In all seven of those elections where the Democrats would be expected to break the supermajority under nonpartisan plans, the enacted plan is an “extreme outlier.” See PX363 (fifth column). 162. In elections where the Democrats won so many votes that the enacted Senate plan’s Republican supermajority firewall breaks, Dr. Mattingly showed that the enacted 74 Senate plan creates a second firewall preventing the Democrats from breaking the Republican majority. Tr. 1114:14-25. Using the results of the 2008 Commissioner of Insurance and 2008 Lieutenant Governor elections—both elections in which the Democrats won over 52.5% of the statewide vote—the enacted plan protects a Republican majority even where the overwhelming majority of nonpartisan plans would break its majority. Id.; PX362. 163. Dr. Mattingly found similar results for the House. Tr. 1087:22-25. Once again, in 15 of the 17 elections he considered, the enacted House Plan produced an atypical bias toward the Republican Party with respect to the number of Democrat and Republican seats. Tr. 1121:23-1122:5. The probability of seeing such a consistent pro-Republican bias across so many elections was 1.4%, Tr. 1122:6-13; PX359 at 11 (Mattingly Report), making it extremely unlikely that the General Assembly would have picked such a partisan map if it were not looking for it, Tr. 1122:14-17. 164. Dr. Mattingly concluded that the enacted House plan is an extreme outlier not just with respect to how consistently it favors the Republicans, but with respect to the amount by which it favors the Republicans. PX359 at 11 (“We never see any plans that favor the Republican Party to the same extent” in terms of seats); PX366 (Mattingly Report Figure 6). The House plan becomes a greater and greater pro-Republican outlier under elections that have more Democratic votes, and becomes an “incredibly extreme outlier” in such elections. Tr. 1120:4-11; Tr. 1119:14-20. The enacted map caused Democrats to lose between 2 and 11 seats in the House in 13 of the 17 elections that Dr. Mattingly analyzed. PX366. The Court finds this seat deviation to be significant. 165. Dr. Mattingly concluded that the enacted House plan’s extreme partisan bias is responsible for creating firewalls protecting the Republican supermajority and majority in the House. Tr. 1120:15-1121:18. As with the Senate, Dr. Mattingly plotted the results of 75 various statewide elections using the enacted House plan and his nonpartisan simulations in Figure 5 of his report (PX365). Tr. 1118:5-1120:14. 166. Plaintiffs’ Exhibit 365 is reproduced below: 76 167. As Dr. Mattingly testified, Plaintiff’s Exhibit 365 illustrates how the enacted House plan becomes a greater and greater pro-Republican outlier as Democrats win more votes statewide, and how the enacted House plan creates firewalls protecting the Republican supermajority and majority which Democrats would break under a nonpartisan plan. Tr. 1120:4-1121:18. In the elections in the lower left of the figure where the Republicans have more statewide votes and have a supermajority even in the nonpartisan plans, the enacted plan is generally within the distribution of nonpartisan plans. PX365 (see, e.g., the 2016 Lieutenant Governor and 2016 U.S. Senate elections). Dr. Mattingly explained that this makes sense from the mapmaker’s perspective, because the mapmaker would not design the map for environments where Republicans are assured a “commanding supermajority” no matter what. Tr. 1123:17-24. 168. Plaintiffs’ Exhibit 365 shows that in elections where the Democrats begin to break the Republican supermajority in the nonpartisan plans, the enacted plan becomes an outlier and consistently protects the Republican supermajority. Tr. 1120:15-1121:8. Dr. Mattingly testified that the enacted map “has a firewall that retards the advance of the Democratic Party particularly when they’re about to break through and break the Republican supermajority.” Tr. 1121:6-8. 169. Overall, in 4 of the 17 elections that Dr. Mattingly considered, the Democrats would have almost certainly broken the Republican supermajority in the nonpartisan plans but failed to do so under the enacted plan (2008 President, 2012 Lieutenant Governor, 2016 Attorney General, 2016 Governor). See PX366 (Mattingly Report Figure 6). By contrast, the enacted map never creates a Democratic supermajority in the House when one would not be expected under the nonpartisan ensemble. PX359 at 13-14. 170. In elections where the Democrats win so many votes that the enacted House plan’s Republican supermajority firewall breaks, Dr. Mattingly showed that the enacted 77 House plan creates a second firewall preventing the Democrats from breaking the Republican majority. Tr. 1119:14-20; Tr. 1121:9-18. Using the results of the 2008 U.S. Senate, 2008 Lieutenant Governor, or 2008 Commissioner of Insurance elections, where the Democrats virtually always have a majority in the collection of hundreds of trillions of nonpartisan plans and sometimes have a supermajority, the Democrats never win a majority under the enacted plan. Tr. 1121:11-18; PX365 (Mattingly Report Figure 5); PX359 at 13. 171. In a race like the 2008 U.S. Senate election—where the Democrats won 54.32% of the statewide vote—the enacted map is a particularly extreme pro-Republican outlier. Tr. 1121:11-18. Using that election, the Republicans win 11 more seats in the enacted House plan than they would expect to win under the nonpartisan collection of plans. PX366 (Mattingly Report Figure 6). In more than 40.1% of the plans in the nonpartisan collection, Democrats actually win a supermajority, but the Democrats do not even win a majority under the enacted plan. PX359 at 14; PX418 (Mattingly Report Table 4). By contrast, there were no historical elections under which the Republicans would have been expected to receive a majority under the nonpartisan House plans but would not receive a majority in the enacted House plan. PX359 at 13. 172. Dr. Mattingly also performed a uniform swing analysis that confirmed the enacted plan’s persistent, durable, and extreme bias toward the Republican party. Tr. 1123:25-1131:5. Using six different historical elections ranging from very pro-Republican (e.g., 2012 Governor, where the Democrats won 44.13% of the statewide vote) to very proDemocratic (e.g., 2008 U.S. Senate, where the Democrats won 54.32% of the statewide vote), Dr. Mattingly showed that the House plan’s gerrymandered protection of the Republican supermajority and majority was highly robust over many different electoral structures and statewide vote fractions. Tr. 1127:15-18; Tr. 1129:5-1131:5; PX488 78 (Mattingly Rebuttal Report Figure 1). Each of the elections end up looking “remarkably the same” as the Democratic vote share increases; in all of the elections, the enacted map creates a firewall protecting the Republican supermajority and majority. Tr. 1129:111130:2; Tr. 1130:23-1131:5. Dr. Mattingly concluded on the basis of his uniform swing analysis that the House plan was “designed” to “consistently protect” the Republican supermajority and majority across all of the “very different” elections he studied, which contain many different “spatial vote patterns” and “historical voting patterns from the state of North Carolina.” Tr. 1130:23-1131:5. 173. In particular, under the nonpartisan maps, the Republicans do not win a supermajority when the Democratic statewide vote share rises above 50 percent, but in the enacted plan, the Republicans do. Tr. 1130:7-19. And the uniform swing analysis shows that the enacted plan becomes an especially extreme outlier whenever the Democrats would win a majority of seats under the ensemble of nonpartisan plans. Tr. 1128:12-1129:4; Tr. 1130:3-6. Dr. Mattingly’s uniform swing analysis shows that the enacted map prevents Democrats from winning a majority of the seats in the House unless they have around 55% of the statewide vote. Tr. 1131:6-16. That is well more than the Democrats would need in a non-gerrymandered plan to win a majority of House seats. See PX488 (Mattingly Rebuttal Report Figure 1). 79 174. Plaintiffs’ Exhibit 488 (Mattingly Rebuttal Report Figure 1) shows Dr. Mattingly’s uniform swing analysis of the House plans: 175. Dr. Mattingly preferred to compare the enacted plan to nonpartisan plans election-by-election, because taking an average seat shift across a set of elections can obscure a gerrymander’s effect in close elections where control of the Senate or House is at issue. Tr. 1214:8-13, 1216:16-19, 1216:22-1217:3. Even considering the average, however, Dr. Mattingly found that the enacted plan is an extreme pro-Republican outlier. Tr. 1216:4-12. Comparing the enacted Senate plan to the median Senate plan in the ensemble for each of the 17 elections, the enacted plan causes Democrats to lose on average 1.94 seats 80 in the Senate across all 17 elections. PX363. Not a single one of Dr. Mattingly’s 3.7 x 1093 statewide maps in the Senate favors the Republican Party as much as the enacted plan under this metric. PX363 (bottom right image); PX487 at 23 (Mattingly Rebuttal Report). Similarly, comparing the enacted House plan to the median House plan in the ensemble for each of the 17 elections, the enacted plan causes Democrats to lose on average 3.35 seats in the House across all 17 elections. Not a single one of Dr. Mattingly’s 1.1 x 10108 statewide maps in the House favors the Republican Party as much as the enacted plan under this metric. PX366 (bottom right image); PX359 at 11 (Mattingly Report) (noting that the average seat difference in favor of the Republicans across all 17 elections is “greater than all plans in the ensemble”). 176. Dr. Mattingly’s separate analysis of the structure of the enacted House and Senate plans provided further confirmation that both plans are extreme partisan gerrymanders, even putting aside the effect on seat count in any particular election. He demonstrated that the General Assembly cracked and packed Democratic voters for partisan gain across the House and the Senate plans, with a particular focus on cracking Democratic voters out of the middle seats that determine supermajority and majority control of both Chambers. 177. Dr. Mattingly ordered the 120 districts in the House in his ensemble of nonpartisan plans from lowest to highest based on the Democratic vote fraction in each district. He did this for each of the 17 statewide elections he analyzed. Tr. 1159:4-15; PX483. 178. Below is an example of Dr. Mattingly’s structural analysis of the 120 districts in the House using the votes from the 2016 Attorney General’s Election. See PX483 at 13; PX778 at 33 (Mattingly PowerPoint presentation). 81 179. The purple dots in the ranked-ordered box plots from Plaintiffs’ Exhibit 483 represent the Democratic vote fraction in the enacted plan for each district ordered from least to most Democratic; the boxes represent the Democratic vote fraction across Dr. Mattingly’s ensemble of nonpartisan plans. Tr. 1159:4-1162:1. The key in the top left-hand corner shows the statewide election and the Democratic statewide vote fraction in that election. 180. Dr. Mattingly explained that in the 40 seats in the middle—between the 40th most Democratic seat and the 80th most Democratic seat—the Democratic vote fraction in the enacted plan is far below the boxes representing the nonpartisan plans. Tr. 1162:7-25. Those “are the seats that determine who has a supermajority and who has the majority,” and they are the “critical seats for the structure of the House.” Tr. 1162:19-25. But in the most Democratic districts, beginning around the 99th least Democratic seat, the Democratic vote fraction is much higher in the enacted plan. Tr. 1162:7-12. In other words, across the map, Democrats have been cracked out of the districts that determine control of the House and packed into districts they would win anyway. Tr. 1162:7-25. In the 2016 Attorney General election, this structural gap between the Democratic vote share in the enacted plan and the nonpartisan plans in the critical districts means that the Republicans kept the supermajority even though they would have lost it under the ensemble of nonpartisan plans. Tr. 1163:3-25. 83 181. An examination of the dis tricts between the 40th least Democratic district and the 80th least Democratic district in the House using the 2016 Attorney General election further demonstrates the cracking of Democratic voters in these critical seats. (PX485 at 13; PX778 at 34): 182. Dr. Mattingly testified that the large gap between the Democratic vote fraction in the enacted plan and in the ensemble at the 72-seat marker is the structural feature of the House map that is responsible for the firewall protecting the Republican supermajority. Tr. 1164:1-9. 183. Dr. Mattingly’s ranked-ordered box plot using the results of the 2012 Presidential election revealed that same structural anomaly (PX485 at 11; PX778 at 35): 84 184. Using the results of the 2012 Presidential election, Dr. Mattingly testified that again the enacted map shows a “huge depletion of Democratic voters” in these districts that matter for supermajority and majority control. Tr. 1164:17-1165:7; PX485 at 11. Dr. Mattingly explained that, although the Presidential 2012 election was a fairly Republican election where the Republicans would win a House majority even under the nonpartisan plans, the significant deviation in the Democratic vote fraction in the seats that matter most will have a “dramatic effect” in elections where the Democrats get more votes statewide. Tr. 1166:1-17. 185. Plaintiffs’ Exhibit 484 contains Dr. Mattingly’s ranked-ordered box plots for the Senate. Dr. Mattingly ordered all 50 Senate districts in his ensemble from lowest to highest based on the Democratic vote fraction in each district. He did this for each of the 17 statewide elections he analyzed. PX484. Below is an example of Dr. Mattingly’s structural 85 analysis of the 50 Senate districts using the 2016 Lieutenant Governor election. PX484 at 15; PX778 at 40. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 86 87 186. The ranked-ordered box plot using the 2016 Lieutenant Governor results demonstrates the same significant suppression of Democratic votes in the enacted plan in the districts that matter most—the 25th most Democratic district, which determines who wins the majority in the Senate, and the 29th least Democratic district, which the Democrats need to win to break the supermajority. Tr. 1175:12-24; PX484 at 15. Dr. Mattingly testified that the gap between the enacted plan and the ensemble around the 25th and 29th/30th district shows that the enacted plan is an “extreme outlier.” Tr. 1176:59. In turn, in the most Democratic districts, the enacted plan has significantly more Democrats than in the nonpartisan ensemble, PX484 at 15—representing packing of Democrats into these districts. Tr. 1175:4-9. 187. As noted, Dr. Mattingly performed this same structural analysis of the House and Senate enacted plans using all 17 of his statewide elections. PX483, PX484. He testified that all 34 of his ranked-ordered box plots overwhelmingly show the same gaps between the enacted plan and the ensemble in the Democratic vote fraction in the seats that matter most in the Senate and the House, and overwhelmingly show the firewalls protecting the Republican supermajorities and majorities. Tr. 1176:10-23. Dr. Mattingly testified that it would “almost be impossible to build this structure” in the absence of an intentional choice to do so. Tr. 1176:24-1177:2. The Court gives great weight to this conclusion. 188. In his report, Dr. Mattingly conducted a statistical analysis to quantify the statewide cracking and packing of Democratic voters in the House and Senate plans that the ranked-ordered box plots from Plaintiffs’ Exhibits 483 and 484 visually illustrate. That analysis confirms to a high degree of statistical significance that the structure of the enacted plans reflects extreme bias in favor of the Republicans that will persist in election after election. 88 189. Specifically, in the House, Dr. Mattingly analyzed the 48th to the 72nd least Democratic districts (again, the range that determines majority and supermajority control). PX359 at 13 (Mattingly Report). Dr. Mattingly found that in 15 of the 17 elections, there is less than a 0.0005% chance of finding a plan in the ensemble that had fewer Democratic votes across those districts than did the enacted plan. Id.; PX359 at 13. In the remaining 2 elections, there was less than a 0.02% and 0.3% chance of finding a plan in the ensemble with as much cracking of Democrats out of the middle districts as the enacted plan. Id. 190. Dr. Mattingly’s statewide quantification of the Senate showed the same extreme cracking of Democrats out of the districts that determine majority and supermajority control. For the Senate, Dr. Mattingly considered the 20th to 30th least Democratic districts. PX359 at 9. He found that in 14 of the 17 statewide elections, there is less than a 0.0005% chance of finding an ensemble plan with fewer Democratic votes across those districts than the enacted plan. Id. In two other elections, the enacted plan was still an extreme outlier, at the 0.1% level. Id. 191. Dr. Mattingly also created video animations of his uniform swing analysis using six different elections in both the House and Senate. PX772 (video animations). In the videos, the blue histograms represent the distribution of seats using Dr. Mattingly’s nonpartisan plans; the “enacted” marker represents the enacted plan, and the three vertical lines represent the Republican supermajority, Republican majority, and Democratic supermajority lines. Id. Dr. Mattingly played two of the videos for the Court, representing uniform swing analysis in the House using the results of the 2012 Presidential election and 2016 Lieutenant Governor election. Tr. 1168:4-8, 1169:17-1172:15; PX778 at 37, 38 (PowerPoint slides); PX772 (video animations). The 2012 Presidential election video showed that the enacted plan started out looking fairly typical of the ensemble of nonpartisan plans; that is the video starts with a 45% Democratic vote share where 89 Republicans retain the supermajority under the nonpartisan plans as well. Tr. 1169:17-25. As the Democratic vote fraction increases, the blue histograms representing the nonpartisan plans shifts to the right and the number of seats that Democrats win increase. Tr. 1169:25-1170:9. But the enacted plan begins to lag “dramatically” behind the nonpartisan plans. Tr. 1170:6-13. In particular, at the Republican supermajority and majority lines, the enacted plan “sticks” on the Republican side of the line even as the blue histogram representing the nonpartisan plans move completely past those lines. Tr. 1171:8-21. The gerrymander is sometimes so effective that it retains a Republican supermajority in the enacted plan even where the Democrats win a majority in the nonpartisan plans. Tr. 1172:6-10. 192. Dr. Mattingly’s video animation of a uniform swing analysis of the 2016 Lieutenant Governor election showed the same thing, Tr. 1172:17-1174:20, as do Dr. Mattingly’s four remaining videos, PX772. 193. The Court finds that these video animations provide significant evidence confirming Dr. Mattingly’s conclusions that the enacted House and Senate maps exhibit extreme partisan bias and create partisan firewalls protecting the Republican supermajority and majority. The Court finds that Dr. Mattingly’s uniform swing videos are also significant evidence that the gerrymanders cause the enacted House and Senate maps to be largely nonresponsive to the actual votes cast in North Carolina’s elections. Moreover, as Dr. Mattingly explained, the ranked-ordered box plots that he created using all 17 statewide elections showing the systematic suppression of Democratic vote fractions in the districts that matter most for the House and Senate demonstrate—without any need to conduct uniform swing analysis—that the enacted plan will be nonresponsive to the votes actually cast in North Carolina elections. Tr. 1174:25-1176:9. 90 194. Dr. Mattingly’s findings regarding the firewall to protect the Republican majorities in the General Assembly are significantly similar to Dr. Chen’s findings. Dr. Chen, like Dr. Mattingly, found that the gap between the number of Democratic districts under the enacted plans and under his simulated plans gets wider in electoral environments that are better for Democrats, and are at their widest around the point where Democrats would win a majority of seats in the House or Senate in his simulated plans. The independent findings of Drs. Chen and Mattingly strengthen and reinforce the conclusion that Legislative Defendants drew the enacted House and Senate plans with the specific goal of making it extremely difficult, if not impossible, for Democrats to take control of either chamber of the General Assembly. 195. Dr. Mattingly’s county-grouping analysis, discussed in greater detail below, also allowed him to draw statistically significant conclusions about the intent of the mapmaker in creating the statewide Senate and House plans. Tr. 1157:24-1158:8. In particular, he explained that the design of each county grouping in the House and Senate plans represented an independent choice by the mapmaker, because “how you redistrict one county cluster does not affect how you redistrict the next one since you can’t cross county cluster lines.” Tr. 1157:17-23. Dr. Mattingly found that numerous county groupings in the House and Senate were extreme pro-Republican partisan outliers at the 100% or 99% level. PX778 at 29-30. He testified that the probability that the extreme partisan bias in the enacted maps was unintentional was “astronomically small,” because the chance of making so many independent choices “with such extreme bias” in one map was “astronomically small if you are not looking for it.” Tr. 1158:3-8. 196. Dr. Mattingly conducted a secondary analysis in which he only considered plans that preserved incumbents “to the same extent, or better, than they are preserved” in the enacted plan in each grouping. PX359 at 81. Dr. Mattingly found that accounting for 91 the effects of incumbency did not change his conclusion that the enacted plans are extreme pro-Republican gerrymanders. Tr. 1093:21-1094:3. Defendants failed to offer evidence sufficient to rebut Dr. Mattingly’s conclusion that the enacted plan’s extreme bias could not be explained by a nonpartisan effort to avoid pairing incumbents. 197. Dr. Mattingly performed extensive robustness checks establishing that his results were insensitive to the choices he made and criteria he used to generate the distribution of nonpartisan plans. Among other things: Dr. Mattingly went through every district in every grouping he analyzed to confirm that the compactness and municipal splits in the ensemble tracked those qualities in the enacted plan. PX359 at 57-80 (Mattingly Report). He performed a secondary analysis considering only plans that were equal to or better than the enacted plan along the dimension of compactness and municipal splits and found that it did not affect his results. PX359 at 82; PX468, 472-473. He created different collections of nonpartisan maps using six different sets of weights for compactness and other nonpartisan criteria and confirmed that changing the weights did not change the results. PX487 at 11 (Mattingly Rebuttal Report). And when Defendants’ experts raised various speculative critiques in their reports—asking whether changing one criterion or another would make a difference—Dr. Mattingly performed a follow-up analysis in his rebuttal report confirming that it did not. Id. at 6-11. 198. The Court finds that none of Legislative Defendants’ objections to Dr. Mattingly’s analysis calls into question its persuasive value. The fact that, in a few individual elections, the enacted plan is not an extreme outlier relative to the ensemble of plans in terms of seat count alone does not undermine Dr. Mattingly’s conclusion that the enacted plans are extreme partisan gerrymanders designed to protect Republican supermajorities and majorities. Tr. 1117:9-11 (Senate); Tr. 1122:18-1123:24 (House). First, Dr. Mattingly explained that the underlying structure of the enacted plans reflected a 92 trade-off. To crack Democrats out of districts where it matters, the mapmaker had to pack Democrats into other districts. Tr. 1123:5-24. Under certain circumstances—i.e., in Republican wave elections—the packing of Democratic voters in the enacted plan causes Republicans to lose districts that they would have won in nonpartisan plans that did not pack Democratic voters into these districts. But such an electoral environment is one in which Republicans would already win a commanding supermajority. Id. As Dr. Mattingly explained, someone gerrymandering a map would happily hold the supermajority or the majority in elections where their control is at risk, even if the cost is a few less seats in elections where they will always have a commanding supermajority anyway. Id. 199. The 2012 Governor election—a highly Republican election where the Republicans win a supermajority in Dr. Mattingly’s nonpartisan plans—provides an example. When Dr. Mattingly conducted a uniform swing analysis using the 2012 Governor election, the enacted map became an “extreme outlier in favor of the Republican Party” as the statewide vote swings to the Democrats and the Democrats approached the point where they would break the Republican supermajority and majority under his nonpartisan plans. Tr. 1126:7-1127:9; PX488. Although the 2012 Governor election may not appear to be a partisan outlier for the Republicans, Dr. Mattingly testified that in fact “it is.” Tr. 1127:19-1128:11. 200. During Dr. Mattingly’s cross examination, Legislative Defendants suggested that he should have included other purportedly nonpartisan criteria in his simulated plans beyond the ones listed in the adopted criteria. The Court, however, gives no weight to Legislative Defendants’ suggestions that secret and undisclosed nonpartisan agreements between “representatives of different political parties” might explain the partisan bias that Dr. Mattingly identified. E.g., Tr. 1204:11-14. The Court also gives no weight to the suggestion that Dr. Mattingly should have accounted for “communities of interest” in a 93 manner other than by avoiding splitting counties, cities, and towns, see, e.g., Tr. 1192:191193:4, considering Legislative Defendants expressly declined to include “communities of interest” as a criterion for the 2017 Plans. Tr. 1223:8-1224:1; see PX603 at 67:14-25 (Rep. Lewis stating that “communities of interest” is not a “criteria that we have proposed” because the Committee “couldn’t find a concise definition”); id. at 73:16-20 (Rep. Lewis stating that he opposed listing “communities of interest” as a criteria because “municipalities are defined and understood” but the Committee couldn’t “agree[]” on what a community of interest was beyond that); id. at 77:3-25 (Rep. Lewis again rejecting the use of “communities of interest”); id. at 106:10-11 (Rep. Lewis stating that “I don’t believe [communities of interest] belongs in this criteria”). 201. When asked by interrogatory to “identify and describe all criteria that were considered or used in drawing or revising districting boundaries for the 2017 Plans,” Legislative Defendants made a binding concession that the only “criteria used to draw the 2017 plans is the criteria adopted by the Redistricting Committees.” PX579 at 13. As such, the Court gives little credence to Legislative Defendants’ critique that Plaintiffs’ experts failed to include criteria not in the Adopted Criteria, or a claim that other considerations purportedly explain the contours of the 2017 Plans. c. 202. Dr. Pegden Wesley Pegden, Ph.D., is an Associate Professor in the Department of Mathematical Sciences at Carnegie Mellon University, and testified as an expert in probability. Tr. 1294:19-21, 1302:6-12; PX509. Dr. Pegden has published numerous papers on discrete mathematics and probability in high-impact, peer-reviewed journals, and has been awarded multiple prestigious grants, fellowships, and awards. Tr. 1295:4-20; PX509. He has been appointed by the Governor of Pennsylvania to that state’s Redistricting Reform Commission. Tr. 1301:24-1302:5. 94 203. Dr. Pegden’s academic work on redistricting involves Markov chains. A Markov chain is “a random walk around some abstract space.” Tr. 1295:23-1296:1. For example, if a person walks around a city, and whenever she reaches an intersection, she chooses which way to turn at random, her position over time “would evolve as a Markov chain.” Tr. 1296:5-7. In the context of redistricting, one can imagine taking a random walk “over the space of maps.” Tr. 1296:8-14. 204. In 2017, before Dr. Pegden had ever served as an expert in redistricting litigation, he published a peer-reviewed article (PX510) entitled “Assessing Significance in a Markov Chain Without Mixing” in the Proceedings of the National Academy of Sciences—a top-ranked, science-wide journal. Tr. 1295:13-17, 1296:24-1297:1. This article provides a new way to demonstrate that a given object is an outlier compared to a set of possibilities. Tr. 1297:2-7. 205. Dr. Pegden explained that there are three ways to show that a given object is an outlier. The first, most basic way is simply to examine every single member of the entire set of possibilities, and then determine whether the object in question is different than all or most of those possibilities. The second form of outlier analysis is to take a random sample from the set of possibilities, and then compare the object in question to that sample. This type of analysis is the basis of most modern statistics, and is the form of outlier analysis used by Drs. Chen and Mattingly in generating nonpartisan simulated plans and comparing the enacted plans to those random nonpartisan plans. Tr. 1297:10-1298:11, 1309:10-18. 206. The third form of outlier analysis, developed by Dr. Pegden and his co- authors, is a kind of “sensitivity analysis” that begins with the object in question, uses a Markov chain to make a series of small, random changes to the object, and then compares the objects generated by making the small changes to the original object. Tr. 1298:16- 95 1299:4. Dr. Pegden’s article illustrates this methodology using a redistricting plan. Tr. 1299:8-18. The article demonstrates that, by using an existing plan as a starting point and then making small random changes to the district boundaries, one can prove the extent to which the existing plan is an outlier compared to all possible maps meeting certain criteria. Dr. Pegden’s article proves mathematical theorems showing that this approach can establish a redistricting plan’s outlier status in a way that is “completely statistically rigorously grounded in mathematics.” Tr. 1299:1-4. 207. In mid-2018, before this case was filed, Dr. Pegden began working on a new article entitled “Practical Tests for Significance in Markov Chains.” Tr. 1300:8-1301:4; PX511. This article further develops this new, third form of outlier analysis with new, more powerful statistical tools. Tr. 1301:5-12. Though unpublished, this second article has been vetted by the mathematical community, including through detailed presentations Dr. Pegden gave at the Duke Statistical and Applied Mathematical Sciences Institute and the Harvard Center for Mathematical Sciences and Applications. Tr. 1300:13-23. 208. In this case, Dr. Pegden used this new, third form of outlier analysis to evaluate whether and to what extent the 2017 Plans were drawn with the intentional and extreme use of partisan considerations. Tr. 1302:24-1303:1. To do so, using a computer program, Dr. Pegden began with the enacted plans, made a sequence of small random changes to the maps while respecting certain nonpartisan constraints, and then evaluated the partisan characteristics of the resulting comparison maps. Tr. 1304:1-1306:21. As explained in further detail below, Dr. Pegden found that the enacted House and Senate plans are more favorable to Republicans than 99.999% of the comparison maps his algorithm generated by making small random changes to the enacted plans. Tr. 1304:1418, 1342:10-18, 1344:18-1345:3; PX515; PX519. And based on these results, Dr. Pegden’s theorems prove that the enacted House and Senate maps are more carefully crafted to favor 96 Republicans than at least 99.999% of all possible maps of North Carolina satisfying the nonpartisan constraints imposed in his algorithm. Tr. 1342:13-25, 1344:18-1345:7; PX515; PX519. 209. Dr. Pegden’s analysis proceeded in several steps. He began with the enacted House or Senate map. His computer program then randomly selected a geographic unit on the boundary line between two districts and attempted to move or “swap” the unit from the district it is in into the neighboring district. Tr. 1309:19-24, 1311:1-5; PX508 at 9 (Pegden Report). 210. Dr. Pegden’s method uses two different geographic units, VTDs and geounits. Tr. 1309:25-1310:2; PX508 at 9 (Pegden Report). His method uses VTDs when analyzing enacted maps that split few or no VTDs. Such maps include the enacted Senate map and the Senate county groupings Dr. Pegden analyzed. Tr. 1310:3-6; PX508 at 9 (Pegden Report). When analyzing enacted maps that split many VTDs—including the enacted House map and certain House county groupings Dr. Pegden analyzed—Dr. Pegden’s method uses a sub-VTD geographic unit known as a “geounit.” Tr. 1310:3-11; PX508 at 9 (Pegden Report). Created by a computer program, geounits are compact collections of census blocks that lie entirely within one VTD and one district, containing roughly 5001000 people. There are roughly six or seven geounits per VTD. Tr. 1310:12-25; PX508 at 9 (Pegden Report). 211. When attempting to swap a randomly selected VTD or geounit from one district to another, Dr. Pegden allowed the swap to occur only if certain constraints were satisfied. Tr. 1311:1-8; PX508 at 7-8 (Pegden Report). These constraints were based on the 2017 Adopted Criteria, and were designed to ensure that the comparison maps generated by Dr. Pegden’s algorithm are “good, reasonable comparisons to the enacted map.” Tr. 1311:9-12, 1317:25-1318:25. The constraints that Dr. Pegden imposed included contiguity, 97 population deviation, compact districts, county preservation, municipality preservation, precinct preservation, and incumbency protection. Tr. 1311:13-1317:10; PX508 at 7-8 (Pegden Report). Dr. Pegden also froze boundary lines redrawn by the Special Master in 2017. Tr. 1319:1-22. 212. Dr. Pegden applied these constraints in a conservative way, so as to “accept choices the mapmaker made.” Tr. 1312:19-22. For example, with respect to population deviation, while the 2017 enacted criteria allows districts to vary between plus-or-minus 5% from the ideal district population, the actual enacted House map does not use all of that range, and instead varies between plus 5% to minus 4.97% from ideal. Dr. Pegden accepted that choice by the mapmaker and required all of his comparison maps to fall within that slightly narrower range. Tr. 1312:1-22; PX508 at 8 (Pegden Report). Similarly, with respect to county preservation, Dr. Pegden’s algorithm not only respected North Carolina’s county groupings, capped the number of county traversals, and preserved the same number of counties as in the enacted map—his algorithm also preserved whole the very same counties preserved whole in the enacted plan. Tr. 1314:9-1315:3. Likewise, with respect to municipality preservation, Dr. Pegden’s algorithm not only preserved the same number of municipalities preserved in the enacted map, but also preserved the very same municipalities, and preserved them within the very same districts as in the enacted plan. Tr. 1315:4-19. 213. Dr. Pegden’s conservative application of these constraints “ties [his] comparisons very strongly to the enacted map itself.” Tr. 1315:22-24. This makes it all the more remarkable that the enacted maps are such outliers in his analysis, even against this very similar comparison set. Tr. 1315:24-1316:2, 1331:6-10. 214. Dr. Pegden also constrained the compactness of his comparison maps. In his main analysis, Dr. Pegden required that the average compactness score for each 98 comparison map not exceed the corresponding average for the enacted plan, with an error of up to 5%. Tr. 1312:23-1313:5; PX508 at 8 (Pegden Report). Dr. Pegden also ran robustness checks using several other compactness constraints—a 10% error, a 0% error, and a completely different measure based on total district perimeter—and found that altering the compactness constraint did not affect his results. Tr. 1313:6-1314:8; PX508 at 32-34 (Pegden Report). 215. For some county groupings, because of Dr. Pegden’s conservative application of his constraints, it was impossible for his algorithm to find a swap that satisfied all of the constraints. Tr. 1319:25-1320:10. When this occurred, Dr. Pegden ran a modification of his algorithm allowing multiple swaps in one step. Tr. 1320:11-25; PX508 at 9-10 (Pegden Report). 216. For some county groupings, even with multi-move swaps, Dr. Pegden’s algorithm still was unable to generate any comparison maps—or only a very small number—meeting all of his constraints. Where this occurred, Dr. Pegden was unable to draw any conclusions about the county groupings in question. Tr. 1321:1-16. Dr. Pegden, however, credibly explained that this does not mean that the maps in those groupings were not drawn with the intentional use of partisanship. For example, partisan considerations could have predominated in choosing which municipalities to preserve whole in which districts, a choice Dr. Pegden’s comparison maps took as a given. Tr. 1321:17-25, 1349:111350:4; PX508 at 10-11 (Pegden Report). 217. Once Dr. Pegden’s algorithm made a swap satisfying his constraints, his algorithm evaluated the partisan characteristics of the comparison map that resulted from the swap. Tr. 1322:1-6. For his main analysis, Dr. Pegden used data from the 2016 Attorney General race to analyze the whole House and Senate maps, the subset of House and Senate districts redrawn in 2017, and any House or Senate county grouping last 99 changed in 2017. Dr. Pegden then used data from the 2008 Commissioner of Insurance race to analyze the subset of House and Senate districts last changed in 2011, as well as any House or Senate county grouping last changed in 2011. Dr. Pegden used these particular elections because they were reasonably close, statewide, down-ballot elections that were available to the General Assembly at the relevant times. Tr. 1322:7-24. Dr. Pegden explained that the “point of [his] analysis is really to get at the intent of the legislature,” to “understand the decisions they made with information available to them at the time.” Tr. 1322:25-1323:3. 218. Dr. Pegden also re-ran his analysis using four additional elections—the 2016 Governor election, the 2014 U.S. Senate election, the 2012 Presidential election, and the 2012 Lieutenant Governor election. Tr. 1323:4-12; PX508 at 35-36 (Pegden report). Using these different historical elections did not alter Dr. Pegden’s conclusions. Tr. 1323:13-15. 219. To evaluate the partisan characteristics of each comparison map, Dr. Pegden’s algorithm calculates the number of seats Republican candidates would win, on average, if a random uniform swing were repeatedly applied to the historical voting data being used. This metric captures how a given comparison map would perform over a range of electoral environments centered around the base election being used (i.e., the 2016 Attorney General’s election for Dr. Pegden’s primary analysis). Tr. 1324:8-1326:20. 220. Dr. Pegden also re-ran his analysis using a different partisan metric, which measures the Republican vote share in the 61st-most Republican House district, or the 26th-most Republican Senate district. This metric captures, for a given comparison map, how comfortably Republicans would win the seat that would give them the majority in the relevant chamber of the General Assembly. Put differently, this metric captures how large of a Democratic wave election the Republican House or Senate majority could withstand. Tr. 1326:21-1327:20. 100 221. In his rebuttal report, in response to certain criticisms by Legislative Defendants’ experts, Dr. Pegden also re-ran his analysis yet again, this time using a third partisanship metric. In this analysis, Dr. Pegden’s algorithm simply measured the number of seats Republicans would have won in an election precisely mirroring the 2016 Attorney General election, without any uniform swing or rank-ordering of districts by Republican vote share. Tr. 1327:21-1328:10. 222. Dr. Pegden’s analysis is statistically robust across three different partisanship metrics, none of which altered his conclusions. Tr. 1326:21-1327:15. 223. Dr. Pegden’s algorithm repeats the foregoing steps billions or trillions of times in sequence. The algorithm begins with the enacted map, makes a small random change complying with certain constraints, and uses historical voting data to evaluate the partisan characteristics of the resulting map. The algorithm then repeats those steps, each time using the comparison map generated by the previous change as the starting point. By repeating this process many times, Dr. Pegden’s algorithm generates a large number of comparison maps in sequence, each map differing from the previous map only by one small random change. Tr. 1328:22-1329:12. 224. Each sequence of billions or trillions of small changes in Dr. Pegden’s analysis is one “run.” His algorithm performs multiple runs for each map being analyzed, with each run beginning with the enacted plan as the starting point. Dr. Pegden ran his algorithm with a sufficient number of steps and runs in order to generate results that are statistically significant but capable of being replicated within a reasonable time. Tr. 1329:3-22. 225. The comparison maps generated by Dr. Pegden’s algorithm are not intended to provide a baseline for what neutral, nonpartisan maps of the North Carolina House or Senate should look like. Instead, Dr. Pegden’s comparison maps are intended to be similar 101 to the enacted map in question with respect to each map’s relevant nonpartisan characteristics, in order to assess how carefully created the enacted plan is to maximize partisan advantage. Tr. 1308:4-12, 1309:10-18, 1329:23-1330:6, 1362:23-1363:6, 1369:251370:4. 226. Dr. Pegden performed two levels of analysis on the comparison maps generated by his algorithm. Dr. Pegden’s first-level analysis simply “report[s] what happened” in each run when his algorithm made random swaps to the enacted plan’s district boundaries. Tr. 1332:8-16. For the enacted House and Senate maps, Dr. Pegden reports that—in every run—the enacted map was more favorable to Republicans than 99.999% of the comparison maps generated by his algorithm making small random changes to the district boundaries. PX515; PX519. 227. Dr. Pegden’s first-level analysis provides clear, intuitive evidence that the 2017 Plans were meticulously crafted for Republican partisan advantage. 228. Dr. Pegden provided a stark illustration from his first-level analysis of how precisely the enacted plans are drawn to maximize partisan advantage. Dr. Pegden explained that, in his runs for the Wake-Franklin county grouping in the Senate, after “the first fraction of a second,” his algorithm “never again” encountered a “single comparison map as advantageous to the Republican Party as the enacted plan itself.” Tr. 1308:151309:7. 229. Dr. Pegden’s second-level analysis provides mathematically precise calculations of how “carefully crafted” the 2017 Plans are—that is, how precisely the district boundaries align with partisan voting patterns so as to advantage Republicans— when compared not just to the comparison maps generated in each run of his algorithm, but to all possible maps of North Carolina that satisfy his constraints. Tr. 1332:24-1335:20. In other words, Dr. Pegden is able to determine—to a mathematical certainty—the extent to 102 which the enacted plan is an outlier relative to every single other possible House or Senate map of North Carolina that could exist meeting the contiguity, equal population, compactness, political subdivision, and Special Master constraints that his algorithm applies. For the enacted House and Senate maps, Dr. Pegden reports that under this second-level analysis the enacted map is more carefully crafted for Republican partisan advantage than at least 99.999% of all possible maps of North Carolina satisfying his constraints. PX515; PX519. 230. The results of Dr. Pegden’s second-level analyses follow from his theorems, which have been validated by other mathematicians. Tr.1337:9-18. And the results of Dr. Pegden’s second-level analyses are intuitive. In effect, Dr. Pegden’s analysis shows that the 2017 Plans not only are quite advantageous to Republicans, but also are surrounded in the space of maps by a plethora of other maps that are less advantageous to Republicans. It is simply not possible, even in principle, for a typical map of North Carolina (or any other state) to be favorable to Republicans and be surrounded by maps that are less favorable. The only explanation is that the map drawer intentionally crafted the district boundaries to maximize partisan advantage. Tr. 1337:9-1340:8; see PX508 at 7 (“In other words, it is mathematically impossible for any state, with any political geography of voting preferences and any choice of districting criteria, to have the property that a significant fraction of the possible districtings of the state satisfying the chosen districting criteria appear carefully crafted.”) 231. For both the House and the Senate, Dr. Pegden performed three different analyses. First, using voting data from the 2016 Attorney General election, Dr. Pegden analyzed the entire House and Senate maps. Second, again using voting data from the 2016 Attorney General election, Dr. Pegden analyzed only the districts that were redrawn in 2017, while freezing the districts that were last changed in 2011. Third, using voting 103 data from the 2008 Commissioner of Insurance election, Dr. Pegden analyzed only the districts that were last changed in 2011, while freezing the districts that were redrawn in 2017. Tr. 1340:14-1341:15. 232. Dr. Pegden’s statewide analyses conclusively show that the pertinent districts drawn in 2011, the districts drawn in 2017, and the maps as a whole were all drawn with the intentional and extreme use of partisan considerations. The following demonstrative chart summarizes Dr. Pegden’s statewide results: PX904; see also PX515-517, 519-521; Tr. 1341:18-1346:16. 233. These results cannot be explained by North Carolina’s political geography. Dr. Pegden’s algorithm compares the enacted map to other maps of North Carolina, with the very same political geography. And Dr. Pegden’s theorems do not depend on any aspect of North Carolina’s political geography—the theorems are mathematically valid for any state with any political geography. Indeed, Dr. Pegden’s theorems are mathematically valid not just for redistricting plans, but for any abstract space on which one could imagine taking a random walk using a Markov chain. Tr. 1333:14-24, 1401:9-1402:5. 104 234. The results of Dr. Pegden’s statewide analyses also conclusively show that it is possible for a North Carolina map drawer to make intentional and extreme use of partisan considerations even within the Whole County Provision and the other constraints set forth in the 2017 Adopted Criteria. All of Dr. Pegden’s comparison maps respect the Whole County Provision and the other constraints set forth in the 2017 Adopted Criteria. And in his algorithm, Dr. Pegden applied those constraints in a very conservative way that respects the choices made by the map drawer with respect to compactness and the divisions and preservation of particular counties and municipalities. Even within those tight constraints, there were many different maps for a map drawer to choose from, and the enacted maps demonstrate that the map drawer intentionally chose maps that were more carefully crafted for Republican partisan advantage than at least 99.999% of all possible alternatives. Tr. 1402:15-1403:8; PX515; PX519. 235. The Court gives great weight to Dr. Pegden’s testimony, analysis, and conclusions. d. 236. Dr. Cooper Christopher A. Cooper, Ph.D., has resided in North Carolina for 17 years and is the Robert Lee Madison Distinguished Professor and Department Head of Political Science and Public Affairs at Western Carolina University. Tr. 848:18-849:7. Dr. Cooper was accepted as an expert in political science with a specialty in the political geography and political history of North Carolina. Tr. 861:21-862:5. 237. As Dr. Cooper explained, North Carolina is a “purple state” that, on the whole, is politically moderate. Tr. 862:21-22. In statewide elections, which are not susceptible to gerrymandering, Democratic candidates perform as well as Republican candidates. Tr. 859:14-18, 864:1-8, 865:5-18. Dr. Cooper’s analysis demonstrated that North Carolina is a “two-party” state where Democrats can compete and succeed with 105 respect to U.S. Presidential elections, Tr. 863:2-864:8; PX255; PX253 at 5-6 (Cooper Report), and elections for North Carolina’s Council of State, Tr. 864:21-865:18; PX256; PX253 at 6-7 (Cooper Report). 238. Dr. Cooper also analyzed the aggregate vote share of Democratic and Republican candidates in General Assembly elections since 2012, finding that Democrats received close to or over 50% of the vote in each election. Tr. 865:23-866:16; PX257. But over the same period, Republicans controlled the North Carolina General Assembly, winning supermajorities in both chambers from 2012-2016 and majorities in 2018. Tr. 866:24-868:12; PX259. Despite winning close to or more than 50% of the statewide vote in General Assembly elections since 2012, Democrats have “never approached” a roughly corresponding percentage of seats, a sign of “gross disproportionality.” Tr. 868:4-12; PX257; PX259; PX264; PX253 at 8, 11 (Cooper Report). 106 239. Dr. Cooper also used the results of the 2018 elections to show how, under the enacted House and Senate plans, Democratic votes translate to seats far less efficiently than Republican votes. Consistent with the packing and cracking of Democratic voters, when Democrats win seats in the House and Senate, they win by large margins, meaning that many votes tend to be “wasted.” Republicans win by significantly narrower margins. Tr. 869:23-871:3; PX262; PX263; PX253 at 14-16 (Cooper Report). 107 240. The Court finds Dr. Cooper’s analysis of the 2018 elections to be persuasive and consistent with Plaintiffs’ experts’ findings regarding the packing and cracking of Democratic voters within individual county groupings, described below. 108 C. The 2017 Plans Were Designed Intentionally and Effectively to Maximize Republican Partisan Advantage Within Specific County Groupings 241. Each of Plaintiffs’ four experts analyzed seven county groupings in the Senate and 16 county groupings in the House. Plaintiffs’ experts concluded that partisan gerrymandering and bias in these groupings was responsible for the extreme partisan bias that they found in their statewide analysis of the House and Senate. Tr. 1134:1-5 (Dr. Mattingly). 1. Senate County Groupings a. 242. Mecklenburg The Mecklenburg Senate county grouping contains Senate Districts 37, 38, 39, 40, and 41. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 243. For each House and Senate county grouping that Plaintiffs’ experts analyzed, Dr. Cooper produced a map showing the district boundaries within the grouping and the partisanship of every VTD within the grouping using the results of the 2016 Attorney General election. In each map, darker red shading indicates a larger Republican vote share in the VTD, darker blue shading indicates a larger Democratic vote share in the VTD, and lighter colors indicate VTDs that were closer to evenly split in Democratic and Republican vote shares. 109 244. Plaintiffs’ Exhibit 285 is Dr. Cooper’s map for this county grouping: 245. As Dr. Cooper explained, the mapmaker packed Democratic voters into Senate Districts 37, 38, and 40 to make Senate Districts 39 and 41 as favorable for Republicans as possible. Tr. 901:16-20; PX253 at 47 (Cooper Report). 246. Senate District 41 stretches from the farthest northern boundaries of Mecklenburg County all the way to the farthest south, traversing two narrow passageways. One passageway is so narrow that the district’s contiguity is maintained only by the Martin 110 Marietta Arrowood Quarry, which is less than a mile wide. Tr. 902:22-903:4; PX287; PX253 at 48 (Cooper Report). The Court is persuaded that the clear intent of this elongated district is to connect the Republican areas north of Charlotte with the Republican-leaning areas in the southern tip of Charlotte. Tr. 902:5-8. 247. Senate District 39 contains the Republican-leaning VTDs in the southern portion of Charlotte, which resemble a “pizza slice” in Dr. Cooper’s maps. Tr. 901:11-15, 902:7-10; PX285; PX286. Those Republican VTDs in Charlotte are grouped with the Republican-leaning areas in the south of Mecklenburg County, outside of Charlotte, so that Senate District 39 is more favorable to Republicans. Tr. 901:18-20; PX253 at 47. 248. Dr. Cooper also illustrated the packing and cracking of Democratic voters in this grouping by focusing just on the division of Charlotte. As illustrated in Plaintiffs’ Exhibit 286 below, the enacted plan places Charlotte’s most Democratic VTDs in Senate Districts 37, 38, and 40, while placing all of Charlotte’s Republican-leaning VTDs in Senate Districts 39 and 41. Tr. 902:1-9; PX253 at 47 (Cooper Report). As Dr. Cooper explained, with large municipalities such as Charlotte, the mapmaker’s partisan intent is not apparent from the mere fact that a municipality is split, but rather from “where do those municipal splits take place and what are the partisan effects.” Tr. 900:12-21; see Tr. 877:24-25. In the Mecklenburg Senate county grouping, the Court is persuaded the mapmaker split Charlotte strictly along partisan lines for partisan gain. 111 249. Legislative Defendants’ expert Dr. Johnson offered alternative explanations for the configuration of this grouping. While Dr. Johnson admitted that he had no personal knowledge as to why Dr. Hofeller or the General Assembly drew the districts this way, Tr. 1972:18-1973:6, Dr. Johnson stated that Senate District 41 was “drawn to capture as much of” the Charlotte suburbs as possible into a single district, Tr. 1844:11-12, and that Senate 39 similarly reflected an effort to “unite[] the southern suburbs” of Charlotte, LDTX289 at 4; Tr. 1845:4-9. 250. The Court rejects Dr. Johnson’s explanations as it appears to be purely speculative, and in any event his speculation does not withstand minimal scrutiny. Rather than seeking to create a “suburban” district, Senate District 41 stretches to Mecklenburg County’s southern tip in order to pick up areas of the City of Charlotte itself, and specifically Republican-leaning VTDs in Charlotte. Tr. 1972:7-1974:15. In so doing, Senate 112 District 41 avoids suburban areas north of Charlotte, with those suburbs packed into Senate District 38 instead because they are Democratic-leaning. Id. Similarly, Senate District 39 cuts into the heart of Charlotte, taking all of Charlotte’s most Republicanleaning areas, while avoiding suburbs in southeast Mecklenburg County. Tr. 1975:51976:14. The Court finds Dr. Johnson’s speculative alternative explanations for the configuration of the Mecklenburg Senate county grouping not credible. 251. Dr. Johnson also opined at trial that the enacted plan version of this county grouping is not the most favorable possible configuration of this grouping for Republicans. Dr. Johnson created an alternative version of this grouping that he asserted would be even more favorable for Republicans. Tr. 1840:17-1841:19. However, Dr. Johnson’s alternative map suffered from a critical error: it paired the two Republican incumbents who were in office at the time of the 2017 redistricting. Tr. 1977:2-1978:7. Clearly, the most favorable possible configuration of this grouping for Republicans would not pair the only two Republican incumbents together, and Dr. Johnson conceded that he did not analyze whether the enacted plan represents the most favorable possible configuration of this grouping possible that would not have paired those two Republican incumbents. Id. 252. The simulations of Plaintiffs’ other experts confirm and independently establish that this county grouping is an extreme partisan gerrymander. 253. Dr. Chen analyzed individual county groupings by comparing the most Democratic district in the grouping under the enacted plan with the most Democratic district in the grouping under the simulated plans, comparing the second most Democratic district in the grouping under the enacted plan with the second most Democratic district in the grouping under the simulated plans, and so on. 254. Using this methodology, Dr. Chen found that the Mecklenburg Senate county grouping has four districts in the enacted plan that are extreme partisan outliers. PX098; 113 see Tr. 377:8-14. Dr. Chen found that Senate Districts 39 and 41 have a lower Democratic vote share than their corresponding districts in all 1,000 of his simulated plans of this grouping, and that Senate Districts 37 and 40 have a higher Democratic vote share than 99.99% and 100% than their corresponding districts in his simulations. Dr. Chen’s findings show the packing of Democratic voters into certain districts in this grouping and the cracking of Democratic voters in Senate Districts 39 and 41, in an effort to create two districts as favorable for Republicans as possible. The Court gives weight to Dr. Chen’s findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 98 below:5 255. Dr. Mattingly analyzed individual county groupings by plotting the Democratic vote fraction in each district in the grouping, ordered from least to most Democratic. He conducted this analysis for the enacted plan (represented by a black dot in his county-grouping-level figures) and for his ensemble of nonpartisan plans (represented 5 Unless otherwise noted, Dr. Chen’s results for individual House and Senate county groupings were materially the same for his Simulation Set 2 as for his Simulation Set 1. Tr. 349:12-18. 114 by the blue histograms), using six prior statewide elections. Tr. 1134:14-1138:6. If the black dot representing the enacted plan is above the dotted black line at 50%, the Democrats win that district under the enacted plan. Tr. 1135:23-1136:6. If all or the bulk of the blue histogram representing the ensemble is above the dotted black line at 50%, the Democrats would expect to win that district under the ensemble. Tr. 1137:8-1138:6. Dr. Mattingly labeled the historical election whose statewide vote counts he was using in the upper left corner of the plots. Black dots that are at the bottom of the corresponding blue histogram represent districts that Democrats have been cracked out of, because the enacted plan has many fewer Democrats than would be expected in the nonpartisan plans; black dots that are at the top of the corresponding blue histogram represent districts that Democrats have been packed into. Tr. 1138:14-1139:4. 256. Plaintiffs’ Exhibit 370 shows Dr. Mattingly’s analysis of the Mecklenburg Senate county grouping: 115 257. As the figure above shows, Democrats were cracked out of the two most Republican districts in this grouping, and packed into heavily Democratic districts. In the enacted plan, there is a significant jump in Democratic vote share between: (i) the two least Democratic districts (Senate Districts 39 and 41), and (ii) the three most Democratic districts (Senate Districts 40, 37, and 38). PX370; PX 359 at 16 (Mattingly Report). Dr. Mattingly testified that the jump signifies intentional gerrymandering—he called it “signature gerrymandering”—and means that elections in the grouping will be nonresponsive to the votes cast. Tr. 1139:19-21; see 1146:13-21; see PX 359 at 14-15 (Mattingly Report). As the figure above shows, the gerrymander cost Democrats one or two seats in certain electoral environments, because the black dots for Senate Districts 39 and 41 often fall below the 50% line while the blue histograms often rise above it. Tr. 1142:221143:1. 258. Dr. Mattingly mathematically quantified the “jump”—i.e., the cracking and packing in this grouping—using all 17 statewide elections he studied. Specifically, Dr. Mattingly calculated the average Democratic vote share in the two least Democratic districts and the average Democratic vote share in the three most Democratic districts, for both the enacted plans and his ensemble plans. PX 359 at 16 (Mattingly Report). He found that the two least Democratic districts in the enacted plan had fewer Democratic voters than 100% of the comparable districts in the nonpartisan ensemble, while the three most Democratic districts in the enacted plan had more average Democratic votes than 100% of the comparable Democratic districts in the nonpartisan ensemble, meaning that not a single plan in his nonpartisan ensemble showed as much of a jump—i.e., as much cracking and packing—as the enacted plan. Tr. 1143:2-20. Dr. Mattingly concluded that the Mecklenburg Senate grouping is an extreme pro-Republican partisan gerrymander, Tr. 1143:21-24, and the Court gives weight to his conclusion. 116 259. Dr. Pegden found that the Mecklenburg Senate county grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.9985% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.995% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1356:25; PX540. The Court gives weight to Dr. Pegden’s analysis and conclusions. 260. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme and intentional partisan gerrymander. b. 261. Franklin-Wake The Franklin and Wake Senate county grouping contains Senate Districts 14, 15, 16, 17, and 18. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 117 262. Plaintiffs’ Exhibit 276 is Dr. Cooper’s map for this county grouping: 263. As Dr. Cooper testified and is clear from a visual inspection, this grouping packs Democratic voters into Senate Districts 14, 15, and 16 in order to make Senate Districts 17 and 18 as favorable for Republicans as possible. Tr. 892:11-13; PX253 at 36 (Cooper Report). 118 264. Senate District 18 includes Franklin County and the only Republican-leaning VTDs within Raleigh, near the center of the city. Tr. 892:13-23; PX278; PX253 at 37-38 (Cooper Report). 265. As with Charlotte, the fact that Raleigh is split is not itself revealing, but how and “where Raleigh is split” illustrates the partisan intent behind the districts in this grouping. Tr. 893:16-894:21; PX253 at 37-38. Plaintiffs’ Exhibit 278, reproduced below, shows how the mapmaker put the most Democratic VTDs in Raleigh in Senate Districts 14, 15, and 16, and put all of Raleigh’s moderate and Republican-leaning VTDs in Senate District 18. Id. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 119 266. Senate District 17 includes all of the Republican VTDs in southern Wake County while carefully avoiding heavily Democratic areas. PX276; PX253 at 36 (Cooper Report). 267. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of Senate Districts 17 and 18. At trial, Legislative Defendants focused on an amendment that Democratic Senator Daniel Blue 120 introduced that altered this grouping, but that amendment did not affect the contours of Senate Districts 17 and 18. Senator Blue testified that he was told by Republican leadership that he could not change the boundaries of Senate Districts 17 and 18, but instead could only shift population between the heavily Democratic districts in this grouping. Tr. 155:20-156:15. Senator Blue’s amendment did just that, as it only shifted population between Senate Districts 14 and 15, both of which had been packed with Democratic voters. Tr. 150:5-8; PX619. Senator Blue’s amendment did not result in, and cannot explain, the composition of Senate Districts 17 and 18 and their extreme partisan outlier status. 268. The simulations of Plaintiffs’ other experts confirm and independently establish that this county grouping is an extreme partisan gerrymander. 269. Dr. Chen found that this county grouping contains three districts that are extreme partisan outliers. Tr. 381:2-18. Senate District 14 has a higher Democratic vote share than its corresponding district in all of the simulations, while Senate Districts 17 and 18 have lower Democratic vote shares than their corresponding districts in all of the simulations. Id.; PX97. Dr. Chen’s findings show the packing of Democratic voters into districts in this grouping in an effort to create two districts (Senate Districts 17 and 18) that are as favorable for Republicans as possible. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 97 below. 121 270. Plaintiffs’ Exhibit 372 shows Dr. Mattingly’s analysis of this grouping: 271. Dr. Mattingly’s analysis shows that Democrats were cracked out of the two least Democratic districts in this grouping (Districts 17 and 18), and packed into heavily Democratic districts. PX372; Tr. 1145:2-7. In the enacted plan, there is a significant jump 122 between the Democratic vote share in the least two Democrats districts and the three most Democratic districts. PX372. Dr. Mattingly found that not a single plan in his ensemble showed as much of a jump between these sets of districts as the enacted plan, Tr. 1145:1114, and concluded that this grouping showed more pro-Republican advantage than 100% of the maps in his ensemble. Tr. 1153:24-1154:4. As the figure above shows, the gerrymander causes Democrats to lose two seats in this grouping in many electoral environments, because the black dots for Senate Districts 17 and 18 fall below the 50% line while the blue histograms often rise above it. See Tr. 1142:22-1143:1. Dr. Mattingly concluded that the Wake-Franklin Senate grouping is an extreme pro-Republican partisan gerrymander, Tr. 1153:17-23, and the Court gives weight to his conclusion. 272. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.99999995% of the maps that his algorithm encountered by making small changes to the district boundaries. Tr. 1356:23-24; PX539. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.99999985% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Id. Dr. Pegden also testified that the changes made by Senator Blue to the boundaries between Senate Districts 14 and 15 cannot explain his results for this county grouping. See Tr. 1352:2-1354:22. The Court gives weight to Dr. Pegden’s analysis and conclusions. 273. The analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. 123 c. 274. Nash-Johnston-Harnett-Lee-Sampson-Duplin The Nash-Johnston-Harnett-Lee-Sampson-Duplin Senate county grouping contains Senate Districts 10, 11, and 12. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 275. Plaintiffs’ Exhibit 274 is Dr. Cooper’s map of this county grouping: 276. Dr. Cooper explained how the district boundaries connect the most Republican VTDs in Johnston County with the Democratic stronghold of Rocky Mount in Senate District 11, ensuring that those Rocky Mount Democratic voters are separated from 124 the moderate and Democratic-leaning VTDs in Johnston County, diluting the voting strength of these various Democratic voters. Tr. 890:4-891:17; PX253 at 33 (Cooper Report). Dr. Hofeller’s Maptitude files further illustrate this intentional cracking of Democratic voters. Dr. Hofeller’s file, below in Plaintiffs’ Exhibit 332, reveals how he drew these districts with “remarkable precision” by “building a fence” around the moderate and Democratic-leaning VTDs in central Johnston County—shaded yellow and red in the image below—making sure to keep these VTDs in Senate District 10 separate from Rocky Mount’s voters in Senate District 11. Tr. 968:12-969:8. 277. Dr. Hofeller’s Microsoft Excel files provide evidence that Dr. Hofeller placed special attention on this country grouping and its partisan composition. In a file titled “Johnston Senate Switch,” Dr. Hofeller compared two alternative drafts of this county 125 grouping and the expected Republican performance of the three districts in this grouping under each of the two alternatives. Tr. 469:5-470:3; PX166; PX123 at 68-69 (Chen Rebuttal Report). The file analyzed no information other than partisanship considerations, demonstrating Dr. Hofeller’s predominant partisan intent in constructing the districts in this grouping. Id. 278. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 279. The simulations of Plaintiffs’ other experts confirm and independently establish that this county grouping was gerrymandered to favor Republicans. 280. Dr. Chen found that all three districts in this county grouping are extreme partisan outliers. Tr. 375:14-25. Senate District 11 has a lower Democratic vote share than its corresponding district in all the simulations, while Senate Districts 10 and 12 have a higher Democratic vote share than their corresponding districts in all the simulations. PX96. Dr. Chen’s findings demonstrate the cracking of Democratic voters across all three districts in this grouping to ensure that all three districts are safe Republican seats. The most Democratic district in this grouping would be far more competitive or even Democratic-leaning under a nonpartisan plan, particular in electoral environments that are more neutral or favorable for Democrats than the 2010-2016 statewide elections. Tr. 376:18. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 96 below: 126 281. Plaintiffs’ Exhibit 382 shows Dr. Mattingly’s analysis of this grouping: 282. Dr. Mattingly concluded that this grouping reflects a pro-Republican partisan bias, Tr. 1154:20-1155:1, and the Court gives weight to Dr. Mattingly’s conclusion. Dr. Mattingly’s analysis shows that, in this grouping, the number of Democrats in the districts 127 was flattened or squeezed to advantage the Republicans. PX778 at 29; Tr. 1154:20-22. Squeezing represents pure cracking, Tr. 1150:22-1151:2. Here, Democrats were cracked out of the most Democratic district and placed in the two least Democratic districts where their presence would not affect the results. When Dr. Mattingly mathematically quantified the cracking in this grouping using all 17 statewide elections, he found that the least two Democratic districts in the enacted plan had more Democratic voters than 77.21% of the comparable districts in the nonpartisan ensemble. Although Dr. Mattingly did not label this grouping an “outlier” because he used a 90% threshold, he explained that the proRepublican bias evidence in this grouping still contributed to the extreme pro-Republican bias he found statewide. Tr. 1151:21-1153:2, 1154:23-1155:1. Because the lines in each county grouping are independent of each other, if the mapmaker time after time makes choices that systematically bias each grouping to one party, that effect accumulates across the map. Tr. 1151:21-1153:2. 283. Moreover, while Dr. Mattingly’s “jump” analysis evaluated the districts in this grouping using all 17 statewide elections, analyzing the most Democratic district in this grouping based on the more recent elections depicted in the figure above reveals the intent and effects of the gerrymander. Dr. Mattingly’s figure shows that the most Democratic district in this grouping under the enacted plan, which is Senate District 11 in most of the elections shown, has less Democrats than the most Democratic district in almost all of his simulations under these more recent six statewide elections. PX382. 284. Dr. Pegden found evidence that this county grouping is an extreme partisan gerrymander. Due to Dr. Pegden’s conservative methodology, his algorithm was only able to generate 18 comparison maps for this Senate county grouping. Tr. 1355:5-23; PX542. Of those 18 maps, Dr. Pegden found that the enacted map for this county grouping is more 128 favorable to Republicans than every single one. Tr. 1356:3-8. The Court gives weight to Dr. Pegden’s analysis and conclusions. 285. The analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. d. 286. Guilford-Alamance-Randolph The Guilford-Alamance-Randolph Senate county grouping contains Senate Districts 24, 26, 27, and 28. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 129 287. Plaintiffs’ Exhibit 281 is Dr. Cooper’s map for this county grouping: 288. For this county grouping, the Covington court tasked the Special Master with redrawing Senate District 28 because the General Assembly’s enacted version of Senate District 28 did not cure the racial gerrymander. 2017 WL 11049096, at *1-2 (M.D.N.C. Nov. 1, 2017). In redrawing Senate District 28, the Special Master also made changes to Senate District 24. See LDTX159 at 19; Covington, ECF No. 220 at 34. Plaintiffs do not challenge Senate Districts 24 and 28 in this case and do not seek relief with respect to them. 130 289. Unlike Senate Districts 24 and 28, the Special Master did not make any changes to the General Assembly’s enacted version of Senate District 26. See Covington, ECF No. 220 at 34 (“2017 Enacted Senate District 26 remains untouched”); Tr. 378:9-16. The Special Master made certain changes to Senate District 27 in carrying out his assignment to redraw Senate District 28, but in so doing, the Special Master did not alter any part of the border between Senate Districts 27 and 26. See Chen Demonstrative D6 at 3; LDTX159 at 19. According to estimates presented at trial by Legislative Defendants’ expert Dr. Johnson, of the current population of Senate District 27, 77% of the population was put into the district by the General Assembly under the enacted 2017 Senate plan. 290. In drawing Senate District 26, the mapmaker cracked Democratic voters in Guilford County, placing the Democratic stronghold of High Point in Senate District 26 and separating these voters from Democratic voters in the Greensboro suburbs. Tr. 895:15896:25; PX254 at 42-43 (Cooper Report). This has the effect of “washing out” the influence of High Point’s Democratic voters, who are joined with the heavily Republican Randolph County in a safe Republican district (Senate District 26), preventing them from influencing the competitive Senate District 27 and thereby making Senate District 27 more favorable for Republicans. Id. 291. Dr. Hofeller’s Maptitude files confirm that he was using VTD-level partisanship data in constructing the districts in this and other county groupings. Tr. 971:16-18; 975:2-5. For example, Dr. Hofeller drew the boundaries of Senate District 26 to grab only the most Democratic VTDs on the border of Randolph County. Tr. 975:10-13, 974:19-975:5. The partisan implications of which are illustrated by Dr. Hofeller’s draft map, which is Plaintiffs’ Exhibit 334: 131 292. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the decision to place High Point’s most-Democratic VTDs in Senate District 26. 293. The simulations of Plaintiffs’ other experts confirm and independently establish that Senate Districts 26 and 27 are extreme partisan gerrymanders. 294. Drs. Chen, Mattingly, and Pegden all froze Senate Districts 24 and 28 in this grouping. Tr. 378:17-379:19; PX359 at 23 (Mattingly Report); PX508 at 30 (Pegden Report). 295. Dr. Chen explained in unrebutted testimony that his simulations of the Alamance-Guilford-Randolph House county grouping did not make any changes to the portion of Senate District 27 added by the Covington Special Master, and instead altered only the southwest portion of Senate District 27 that borders Senate District 26. Tr. 773:822; Chen Demonstrative D6 at 4, 5; PX1 at 18-19 (Chen Report). The Court finds that 132 because Dr. Chen’s simulations altered only portions of Senate District 27 drawn by the mapmaker, and did not touch the portions of the district added by the Special Master, the mapmaker necessarily is responsible for the extreme partisan bias that Dr. Chen finds for Senate District 27. 296. Dr. Chen found that both districts in this county grouping that he did not freeze are extreme partisan outliers. Senate District 26 has a higher Democratic vote shares than its corresponding district in all of the simulations, while Senate District 27 has a lower Democratic vote share that its corresponding district in all of the simulations. Tr. 380:1-18; PX94. Dr. Chen’s findings show the mapmaker’s intentional placing of High Point’s Democratic voters into Senate District 26 to make Senate District 27 as favorable for Republicans as possible. The Court gives weight to Dr. Chen’s findings and analysis for this grouping, which are reflected in Plaintiffs’ Exhibit 94 below: 133 297. Plaintiffs’ Exhibit 380 shows Dr. Mattingly’s analysis of the Guilford- Alamance-Randolph Senate county grouping: 298. Setting aside the frozen districts, Dr. Mattingly’s analysis shows that Democrats were cracked between the grouping’s two remaining districts—an example of what Dr. Mattingly called flattening or squeezing. PX380; PX778 at 29; PX359 at 23. Not a single plan in Dr. Mattingly’s nonpartisan ensemble showed as much cracking of Democratic voters in the grouping as was present in the enacted plan, PX359 at 23, and thus the grouping has more pro-Republican advantage than 100% of the maps in his nonpartisan ensemble. Tr. 1153:24-1154:4. Dr. Mattingly concluded that this grouping is an extreme pro-Republican partisan gerrymander, Tr. 1153:17-23; PX778 at 29; PX359 at 23, and the Court gives weight to this conclusion. 299. Dr. Pegden found that this Senate county grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.95% of the maps that his 134 algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.85% of all possible districtings of this grouping that satisfy the criteria Dr. Pegden used. Tr. 1357:1; PX543. The Court gives weight to Dr. Pegden’s analysis and conclusions. 300. The analyses of Plaintiffs’ experts independently and together demonstrate that Senate Districts 26 and 27 are extreme partisan gerrymanders. e. Davie-Forsyth 301. The Davie-Forsyth Senate county grouping contains Senate Districts 31 and 32. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 135 302. Plaintiffs’ Exhibit 282 is Dr. Cooper’s map for this county grouping: 303. Dr. Cooper explained what is apparent from the above map: the mapmaker packed Democratic voters into Senate District 32, thereby ensuring that Senate District 31 would be a safe Republican district. Tr. 897:9-24; PX253 at 44 (Cooper Report). 304. This packing occurred not only at the grouping-level, but within Winston- Salem. The map packs all of Winston-Salem’s most Democratic VTDs into Senate District 32, and puts almost all of the city’s Republican-leaning VTDs in Senate District 31. Tr. 136 898:1-16; PX283; PX253 at 44 (Cooper Report). As shown in Plaintiffs’ Exhibit 283 below, Senate District 31 wraps around Winston-Salem to avoid the Democratic-leaning VTDs in the city, while taking in the Republican-leaning VTDs on the western, northern, and eastern sides of the city: 305. Dr. Hofeller’s Maptitude files confirm his predominant partisan intent in drawing this grouping. The district boundaries are drawn “almost perfectly” so that the green areas on the map, which reflect Republican VTDs, are all placed in Senate District 31. Tr. 976:24-977:4; PX335; PX329 at 11 (Cooper Rebuttal Report). The “bite mark” on the west side of Winston-Salem, where Republican-leaning VTDs were carved out of Senate District 32, is evident on Dr. Hofeller’s draft map of these districts, which is Plaintiffs’ Exhibit 335: 137 306. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 307. The simulations of Plaintiffs’ other experts confirm and independently establish that the Davie-Forsyth county grouping is an extreme partisan gerrymander. 308. Dr. Chen found that both districts in this grouping are extreme partisan outliers. Tr. 373:18-374:12. Senate District 32 has a far higher Democratic vote share than its corresponding district in all of the simulations, while Senate District 31 has a far lower Democratic vote share than its corresponding district in all of the simulations. PX95. Dr. Chen’s findings demonstrate the packing of Democratic voters into Senate District 32 in order to make Senate District 31 a safe Republican seat. As Dr. Chen explained, the less Democratic district in this grouping would be far more competitive for Democrats under a nonpartisan plan, particularly in electoral environment that are more neutral or favorable 138 for Democrats than the 2010-2016 statewide elections. Tr. 374:13-23. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 95 below: [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 139 309. Plaintiffs’ Exhibit 374 shows Dr. Mattingly’s analysis of this county grouping: 310. Dr. Mattingly’s analysis shows that Democrats were cracked out of the most Republican district in this county grouping, and packed into the most Democratic district. PX374; PX778 at 29. Dr. Mattingly found that not a single plan in his nonpartisan ensemble showed as much packing of Democratic voters in the Davie-Forsyth Senate grouping as was present in the enacted plan, PX359 at 18, and thus the grouping has a more pro-Republican advantage than 100% of the maps in his nonpartisan ensemble, Tr. 1153:24-1154:4. Dr. Mattingly concluded that this grouping is an extreme pro-Republican partisan gerrymander, Tr. 1153:17-23; PX778 at 29; PX359 at 18, and the Court gives weight to his conclusion. 311. Dr. Pegden found that this county grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of the grouping is more favorable to Republicans than 99.993% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second 140 level analysis, Dr. Pegden found that the grouping is more carefully crafted to favor Republicans than at least 99.98% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1356:25; PX538. The Court gives weight to Dr. Pegden’s analysis and conclusions. 312. The analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. f. 313. Bladen-Pender-New Hanover-Brunswick The Bladen-Pender-New Hanover-Brunswick Senate county grouping, drawn in 2011 and left unchanged in 2017, contains Senate Districts 8 and 9. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 141 314. Plaintiffs’ Exhibit 272 is Dr. Cooper’s map of this county grouping: 315. In this grouping, the population of New Hanover County is slightly too large to fit into one Senate district, and thus the mapmaker had to place a small portion of New Hanover in Senate District 8. Tr. 887:8-9. The mapmaker chose to take heavily Democratic VTDs in Wilmington, separating them from the rest of Wilmington (which is in Senate District 9) and grouping them instead with heavily Republican areas in Bladen, Pender, and Brunswick counties. Tr. 887:5-888:8; PX253 at 29-31 (Cooper Report). As Dr. Cooper explained, the clear intent and effect of this decision was to waste the votes of the 142 Democratic voters in these Wilmington VTDs, placing them in a heavily Republican district (Senate District 8) and removing them from a highly competitive district (Senate District 9) where their votes could make a difference. Id. Plaintiffs’ Exhibit 273 provides a zoomed-in view of the cracking of the Democratic voters in these two VTDs, which has come to be known as the “Wilmington Notch”: 316. Dr. Cooper credibly testified that the enacted plan is the most maximally favorable construction of the grouping possible for Republicans. Tr. 887:24-25. This grouping illustrates Dr. Cooper’s conclusion about all of the groupings he analyzed: “whenever there’s discretion to be exercised, that discretion tended to go in favor of one party, in this case the Republican Party, and against the other party, in this case the Democrat party.” Tr. 889:22-25. 143 317. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. While they noted that some portion of New Hanover County must be placed in Senate District 9 for equal population purposes, Legislative Defendants failed to rebut the fact that alternative ways to draw the grouping would not split municipalities in the manner that the enacted plan does. Over 97% of Dr. Mattingly’s simulations of this county grouping do not split Wilmington. PX429. 318. The simulations of Plaintiffs’ other experts confirm that the Bladen- Brunswick-New Hanover-Pender Senate county grouping is an outlier. 319. Because this county grouping was drawn in 2011 and remained unchanged in 2017, in analyzing this individual county grouping, Dr. Chen used the statewide elections from 2004 to 2010 that the General Assembly used during the 2011 redistricting process, rather than the 2010-2016 statewide elections. Tr. 366:8-367:1, 382:23-383:11; PX720. Dr. Chen used these 2004-2010 statewide elections because, to assess the question of partisan intent, he wanted to use the same elections data that the mapmaker had available and was considering when it drew this grouping in 2011. Tr. 367:2-23; PX1 at 21-24 (Chen Report). 320. Dr. Chen found that both districts in this county grouping are extreme partisan outliers. Tr. 384:2-386:19. Senate District 9 has a lower Democratic vote share than all of its corresponding districts in all of the simulations, while Senate District 8 has a higher Democratic vote share than all of its corresponding districts in all of the simulations. Id.; PX100. Dr. Chen’s analysis demonstrates that the moving of Democratic voters in the Wilmington Notch into Senate District 8 made Senate District 9 as favorable for Republicans as possible. The Court gives weight to Dr. Chen’s findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 100 below: 144 321. Dr. Mattingly similarly concluded that the Bladen-Pender-New Hanover- Brunswick Senate grouping was “certainly an outlier” but when on to state that “there were some features of [the Bladen] district that meant that the type of analysis that [he] had initially chosen was not as illuminating in that district. So [he] couldn’t say something is conclusive.” Tr. 1154:11-16. When he mathematically quantified cracking in the Bladen grouping across all 17 statewide elections, he found that the most Democratic district in the Bladen grouping had fewer Democrats than in 92.46% of plans in the nonpartisan ensemble. PX359 at 19-20 (Mattingly Report); PX778 at 29.6 322. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme and intentional partisan gerrymander. 6 Dr. Pegden was unable to generate any comparison districtings of this county grouping due to his conservative methodology. Tr. 1357:12-23; PX544. As Dr. Pegden testified, the fact that his algorithm does not generate any comparison districtings for a given county grouping does not mean that the mapmaker did not make extreme and intentional use of partisan considerations in that county grouping. See Tr. 1321:17-25, 1349:11-1350:4. 145 g. 323. Buncombe-Henderson-Transylvania The Buncombe-Henderson-Transylvania Senate county grouping, drawn in 2011 and left unchanged in 2017, contains Senate Districts 48 and 49. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 324. Plaintiffs’ Exhibit 288 is Dr. Cooper’s map of this county grouping: 146 325. Dr. Cooper explained how these district boundaries combine the heavily Democratic VTDs in Asheville with Democratic VTDs in Black Mountain, packing those Democratic voters to create a safe Democratic district in Senate District 49, allowing Senate District 48 to comfortably favor Republicans. Tr. 903:23-904:13; PX253 at 50 (Cooper Report). 326. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 327. The simulations of Plaintiffs’ other experts confirm and independently establish that this county grouping is an extreme partisan gerrymander. 328. Dr. Chen found that both districts in this county grouping are extreme partisan outliers. Tr. 383:12-19.7 Senate District 49 has a higher Democratic vote share than its corresponding district in nearly all of the simulations, while Senate District 48 has a lower Democratic vote share than its corresponding district in nearly all of the simulations. PX99. Dr. Chen’s findings demonstrate the packing of Democratic voters into Senate District 49 to make Senate District 48 a safe Republican seat. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 99 below: 7 Because this county grouping was drawn in 2011, Dr. Chen used the 2004 to 2010 statewide elections to analyze this county grouping. Tr. 383:16-22; PX99. 147 329. Plaintiffs’ Exhibit 378 shows Dr. Mattingly’s analysis of the Buncombe- Transylvania-Henderson Senate county grouping: 330. Dr. Mattingly’s analysis shows that Democrats were cracked out of Senate District 48 and packed into Senate District 49. PX378; PX778 at 29; Tr. 1153:7-1154:9. Dr. 148 Mattingly found that the least Democratic district in the enacted plan has fewer Democratic votes than in 95.44% of the plans in his ensemble, meaning that the grouping showed more pro-Republican partisan advantage than 95.44% of the nonpartisan plans. PX778 at 29; PX359 at 21-22. Dr. Mattingly concluded that this grouping reflects a proRepublican partisan gerrymander, Tr. 1154:6-10; PX778 at 29; PX359 at 21-22, and the Court gives weight to his conclusion. 331. Dr. Pegden found that this county grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of the grouping is more favorable to Republicans than 99.8% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that the grouping is more carefully crafted to favor Republicans than at least 99.4% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1357:2; PX541. The Court gives weight to Dr. Pegden’s analysis and conclusions. 332. The analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. 2. House County Groupings a. 333. Robeson-Columbus-Pender The Robeson-Columbus-Pender House county grouping contains House Districts 16, 46, and 47. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 149 334. Plaintiffs’ Exhibit 301 is Dr. Cooper’s map of this county grouping: 335. Dr. Cooper explained that House District 47 packs as “many . . . Democratic voters as possible” into that district, including in Lumberton and the area around UNC Pembroke. The packing of Democrats in House District 47 makes House Districts 16 and 46 more favorable to Republicans. Tr. 912:19-913:3; PX253 at 70 (Cooper Report). 336. Dr. Hofeller’s Maptitude files confirm he “had full knowledge of the partisan effects of drawing those lines exactly where they were drawn, essentially drawing a fence between districts 47 and 46 . . . between Democratic and Republican voters.” Tr. 985:15-19; PX342; PX329 at 18 (Cooper Rebuttal Report). In the files for his draft House plan, Dr. Hofeller shaded more Democratic VTDs darker blue, more Republican VTDs red and orange, and moderate VTDs green and yellow. Tr. 979:20-980:19. As shown in Plaintiffs’ 150 Exhibit 342, Dr. Hofeller placed all of the Republican-leaning VTD near Lumberton (shaded orange and red) on the right side of the red line, in House District 46, rather than in House District 47: 337. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of the districts in this county groupings. 338. The simulations of Plaintiffs’ other experts independently establish that the Columbus-Pender-Robeson county grouping is an extreme partisan gerrymander. 339. Dr. Chen found that all three House districts in this county are extreme partisan outliers. Dr. Chen found that House District 47 has a higher Democratic vote share than the corresponding districts in all of Dr. Chen’s simulated plans. Tr. 346:4347:14. Dr. Chen found that House District 46 has a lower Democratic vote share than the corresponding districts across all of Dr. Chen’s simulations, while House District 16 has a 151 higher Democratic vote share than the corresponding districts in all of Dr. Chen’s simulations. Tr. 347:16-348:7. Dr. Chen’s findings demonstrate the packing of Democratic voters into House District 47 and the cracking of Democratic voters across House Districts 16 and 46. Dr. Chen finds that, as a result of this packing and cracking, almost all of his simulations would produce two Democratic-leaning districts in this county grouping, while the enacted House plan produces just one such district in this grouping. Tr. 348:8-23. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 47 below: 152 340. Plaintiffs’ Exhibit 388 shows Dr. Mattingly’s analysis of the Columbus- Pender-Robeson House county grouping: 341. Dr. Mattingly’s analysis shows that Democrats were cracked in the two least Democratic districts in this grouping (Districts 16 and 46) and packed into the most Democratic district (District 47). PX388; PX359 at 28; PX778 at 30. There is a significant jump between the number of Democratic votes in the two least and the most Democratic districts in the enacted plan. Id. Dr. Mattingly found that the two least Democratic districts in the enacted plan have fewer Democratic voters than 97.98% of the comparable districts in the nonpartisan ensemble. Id. As the figure above shows, the gerrymander causes Democrats to lose a seat in this grouping in certain electoral environments. Dr. Mattingly concluded that this grouping reflects a clear pro-Republican partisan gerrymander, PX778 at 30; Tr. 1155:17-21; PX359 at 28, and the Court gives weight to Dr. Mattingly’s conclusion. 342. Dr. Pegden found that this county grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version 153 of this grouping is more favorable to Republicans than 98.7% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 96% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:8; PX526. The Court gives weight to Dr. Pegden’s analysis and conclusions. 343. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. b. 344. Cumberland The Cumberland House county grouping contains House Districts 42, 43, 44, and 45. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 154 345. Plaintiffs’ Exhibit 305 is Dr. Cooper’s map of this county grouping: 346. Dr. Cooper described how House District 45 has a “backwards C-shape” that is “a very clear attempt to connect these Republican leaning [VTDs] all together and avoid . . . the Democratic leaning VTDs.” Tr. 917:7-14. In such a way, the district boundaries make House District 45 more favorable for Republicans, while packing the Democraticleaning VTDs in the Fayetteville area into House Districts 42 and 43. Tr. 917:14-16; PX253 at 76 (Cooper Report). 347. The district boundaries in this grouping, shown below in Plaintiffs’ Exhibit 306, divide Fayetteville between all four districts in a way that does not correspond to 155 Fayetteville’s boundaries of or any other municipality. Tr. 917:23-918:5; PX253 at 76 (Cooper Report). 348. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 349. The simulations of Plaintiffs’ other experts independently establish that the Cumberland county grouping is an extreme partisan gerrymander. 350. Dr. Chen found that this county grouping contains three districts that are extreme partisan outliers. Dr. Chen found that House Districts 42 and 43 have a higher Democratic vote shares than their corresponding districts in all or almost all of Dr. Chen’s simulated plans, while House District 45 has a much lower Democratic vote share that the corresponding district in all of the simulations. Tr. 350:2-12. Dr. Chen’s findings demonstrate the packing of Democratic voters into House Districts 42 and 43 in order to make House District 45 as favorable for Republicans as possible. Indeed, the least Democratic district in this grouping would be very competitive or even Democratic-leaning 156 in Dr. Chen’s simulations. The Court gives weight to Dr. Chen’s findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 48 below: [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 157 351. Plaintiffs’ Exhibit 390 shows Dr. Mattingly’s analysis of the Cumberland House county grouping: 352. Dr. Mattingly’s analysis shows that the least Democratic district (District 45) show cracking of Democrats, while the two most Democratic districts (District 43 and 42) show extreme packing of Democrats, in comparison to the nonpartisan plans. PX390; PX778 at 30; PX359 at 29. He found that the two most Democratic districts in the enacted plan have more Democratic votes than 99.79% of the comparable Democratic districts in the nonpartisan ensemble. Id. As the figure above shows, the gerrymander causes Democrats to lose a seat in this grouping in certain electoral environments, because the black dot in House District 45 always falls below the 50% line while the blue histogram often rises above it. Dr. Mattingly concluded that the Cumberland House grouping is an extreme proRepublican partisan gerrymander, Tr. 1155:5-16; PX778 at 30; PX359 at 29; PX390, and the Court gives weight to Dr. Mattingly’s conclusion. 353. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version 158 of this grouping is more favorable to Republicans than 98.3% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 95% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:9; PX529. The Court gives weight to Dr. Pegden’s analysis and conclusions. 354. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. c. 355. Person-Granville-Vance-Warren The Person-Granville-Vance-Warren House county grouping contains House Districts 2 and 32. 356. Plaintiffs’ Exhibit 289 is Dr. Cooper’s map of this county grouping: 159 357. Several of Plaintiffs’ experts testified that there are only a limited number of possible ways to draw this county grouping. Tr. 359:4-360:2 (Dr. Chen), 905:17-19 (Dr. Cooper); 1156:25-1157:16 (Dr. Mattingly). Because of the Whole County Provision, the only differences between the alternative ways to draw this grouping involve which of Granville County’s few VTDs are placed in each of the two districts. See id. 358. This county grouping is one of two drawn by Campbell Law students and ultimately adopted by Dr. Hofeller. Tr. 474:7-475:23; PX123 at 71. The evidence from Dr. Hofeller’s files suggests that Dr. Hofeller intentionally chose to include this configuration because it most favored Republicans, to the detriment of Democratic voters. See Tr. 905:21906:8. 359. However, because of the limited possible configurations for this county grouping, and the limited statistical evidence that could be generated by Plaintiffs’ experts, the Court does not find that this grouping, or the districts contained therein, constitute an extreme partisan gerrymander. See PX051 (Dr. Chen Figure 31 showing Democratic vote share of each district well below his extreme partisan outlier threshold); Tr. 1156:251157:16 (Dr. Mattingly found very few possible unique maps for this grouping that satisfied his criteria); Tr. 1349:11-1350:4; PX536 (Dr. Pegden was unable to generate any comparison districtings of this House county grouping due to his conservative methodology). 360. The Court, though, does find that this county grouping does reflect a clear pro-Republican partisan tilt that can contribute to the extreme pro-Republican bias statewide. 160 d. 361. Franklin-Nash The Franklin-Nash House county grouping contains House Districts 7 and 25. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 362. Plaintiffs’ Exhibit 293 is Dr. Cooper’s map of this county grouping: 363. These district boundaries avoid grouping the more Democratic-leaning and competitive VTDs on Nash County’s western border in House District 7, instead stretching 161 House District 7 into the southeast corner of Nash County to grab the heavily Republican VTDs there. The placement of this district boundary made House District 7 more favorable to Republicans. As Dr. Cooper explained, if the mapmaker had included “any other VTD” in House District 7 from Nash County, House District 7 would have been less favorable to Republican candidates. Tr. 907:4-13; PX253 at 59 (Cooper Report). 364. The Court gives little weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. They noted that the enacted version of this county grouping matches the draft drawn by the Campbell Law students, but the mapmaker adopted these districts because they were maximally favorable for Republicans, FOF § B.2.a., and as the simulations of Plaintiffs’ experts Dr. Chen and Dr. Mattingly confirm and independently establish, the Nash-Franklin House county grouping is indeed an extreme partisan gerrymander. 365. Dr. Chen found that both districts in county grouping are extreme partisan outliers. Dr. Chen found that House District 25 has a higher Democratic vote share than its corresponding district in all of Dr. Chen’s simulated plans, while House District 7 has a lower Democratic vote share that the corresponding district in all of the simulations. Tr. 356:8-17. Dr. Chen’s findings demonstrate the packing of Democratic voters into House Districts 25 in order to make House District 7 a safe Republican seat. In Dr. Chen’s simulations, the less Democratic district in this grouping would be more competitive for Democrats, particularly in a more favorable electoral environment for them than the 20102016 statewide elections. Tr. 356:18-357:1. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 50 below: 162 366. Plaintiffs’ Exhibit 402 shows Dr. Mattingly’s analysis of the Nash-Franklin House county grouping: 163 367. Dr. Mattingly concluded that the most Democratic district shows extreme packing of Democrats, while the most Republican district shows extreme cracking of Democrats, in comparison to the nonpartisan plans. Tr. 1149:2-9. He found that the least Democratic district in the enacted plan has fewer Democratic voters than 95.58% of the comparable districts in the nonpartisan ensemble, demonstrating packing. PX778 at 30; PX359 at 36-37. As the figure above shows, the gerrymander could cause the Democrats to lose a seat in this grouping in certain electoral environments, because the black dot for House District 7 falls below the 50% line while the blue histogram sometimes rises above it or gets very close. Dr. Mattingly concluded that the Nash-Franklin House grouping is a pro-Republican partisan gerrymander, PX778 at 30; Tr. 1155:17-21; PX359 at 36-37, and the Court gives weight to Dr. Mattingly’s conclusion.8 368. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. e. 369. Pitt-Lenoir The Pitt-Lenoir House county grouping contains House Districts 8, 9, and 12. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 8 Dr. Pegden was unable to generate any comparison districtings of this House county grouping due to his conservative methodology. Tr. 1351:22-1352:10; PX537. 164 370. Plaintiffs’ Exhibit 294 is Dr. Cooper’s map of this county grouping: 371. The districts in this county grouping split Greenville between all three House districts and even bisect East Carolina University’s campus. The district lines pack the most Democratic-leaning VTDs in Greenville into House District 8, while placing all but one of the Republican-leaning VTDs into House District 9. Tr. 908:3-8, 909:23-910:8; PX253 at 61 (Cooper Report). Plaintiffs’ Exhibit 295 below shows the municipalities within this county grouping and how the districts split Greenville. Tr. 908:16-23. 165 372. The Maptitude files from Dr. Hofeller’s hard drive confirm he used VTD-level partisanship data with “surgical precision” to construct the districts in this grouping. Tr. 983:5-984:7; PX340; PX329 at 16 (Cooper Rebuttal Report). Dr. Hofeller’s Maptitude file, reproduced below in Plaintiffs’ Exhibit 340, demonstrates how Dr. Hofeller meticulously packed all of Greenville’s bluest VTDs into House District 8 (on the left side of the red line), in order to make House Districts 9 and 12 favorable for Republicans. 166 373. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of the districts in this county grouping. 374. The simulations of Plaintiffs’ other experts independently establish that the Lenoir-Pitt county grouping is an extreme partisan gerrymander. 375. Dr. Chen found that House District 8 has a higher Democratic vote shares than its corresponding districts in all Dr. Chen’s simulated plans, while House District 9 has a lower Democratic vote share than the corresponding district in all of the simulations. PX52; Tr. 360:16-22. Dr. Chen further found that the remaining district in this grouping, House District 12, is less Democratic than over 81% of the corresponding districts across Dr. Chen’s simulations. Id. Dr. Chen’s findings demonstrate the packing of Democratic voters into House District 8 and the cracking of Democratic voters in House Districts 9 and, 167 to some extent, 12. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 52 below: 376. Plaintiffs’ Exhibit 408 shows Dr. Mattingly’s analysis of this grouping: 168 377. Dr. Mattingly concluded that the two most Republican districts show extreme cracking of Democrats, while the most Democratic shows extreme packing of Democrats, as evidence by the “jump” between these sets of districts. PX408; PX778 at 30; PX359 at 41. Dr. Mattingly found that the two least Democratic districts in the enacted plan have fewer Democratic voters than 99.98% of the comparable districts in the nonpartisan ensemble, while the most Democratic district in the enacted plan has more Democratic votes than 99.95% of the comparable Democratic districts in the ensemble. PX778 at 30; PX359 at 43. As the figure above shows, the gerrymander causes the Democrats to lose one or possibly two seats in this grouping in certain electoral environment, because the black dot in House Districts 9 and 12 often falls below the 50% line while the blue histograms rise above it. Dr. Mattingly concluded that the Pitt-Lenoir House grouping is an extreme pro-Republican partisan gerrymander, Tr. 1155:5-16; PX778 at 30; PX359 at 41; PX408, and the Court gives weight to Dr. Mattingly’s conclusion. 378. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.97% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.91% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:6; PX532. The Court gives weight to Dr. Pegden’s analysis and conclusions. 379. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. 169 f. 380. Guilford The Guilford House county groupings contains House Districts 57, 58, 59, 60, 61, and 62. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 381. This grouping contains several districts that were altered by the Covington Special Master. The Covington court tasked the Special Master with redrawing House District 57 after the court found that the enacted House plan did not cure the racial gerrymander of the district. Covington, 2017 WL 11049096, at *1-2. In directing the Special Master to redraw House District 57, the court further directed that “the redrawn lines shall also ensure that the unconstitutional racial gerrymanders in 2011 Enacted House Districts 58 and 60 are cured.” Id. at *2. The Covington court did not direct the Special Master to redraw House District 59, and did not even mention House District 59 in its order. 382. Consistent with the court’s guidance, the Special Master redrew House District 57, and in so doing, also made substantial changes to House District 61 and 62. Tr. 351:14-25; see LDTX 159 at 27-29 (Special Master’s Recommend Plan). In redrawing these three districts, the Special Master also made what he described as “minor changes” to House District 59 to equalize population. Covington, ECF No. 220 at 46. The Special Master explained that he altered House District 59 “only a little.” LDTX 159 at 28. Specifically, the Special Master moved one precinct from the enacted District 59 into the Special Master’s District 57, and added “two additional precincts” to the northwest corner of House District 59 to equalize population. Covington, ECF No. 220 at 46; see Chen Demonstrative D5 at 3; Tr. 352:1-21. According to estimates presented at trial by Legislative Defendants’ expert Dr. Johnson, of the current population of House District 59, 92% of the population was put into the district by the General Assembly under the enacted 170 House plan. LDTX314; Tr. 1978:19-22. The Special Master did not make any changes at all to House Districts 58 and 60. Plaintiffs do not bring allegations, and do not seek relief, with respect to the three House districts that the Special Master substantially redrew, House Districts 57, 61, and 62. 383. Plaintiffs’ Exhibit 310 is Dr. Cooper’s map for this grouping: 384. The mapmaker packed Democratic voters into House Districts 58 and 60 to make House District 59 favorable to Republicans. Tr. 923:3-23; PX253 at 82 (Cooper 171 Report). House District 58 has “boot-like appendages” to grab Democratic VTDs and ensure these voters could not make House District 59 competitive or Democratic-leaning. Id. 385. The Maptitude files from Dr. Hofeller’s hard drive confirm Dr. Hofeller drew this grouping with extreme partisan intent. Tr. 986:13-987:9. Specifically, Dr. Hofeller drew the boundaries of House Districts 58, 59, and 60 “almost like a fence” “separating [Republican voters] from the Democratic voters” in the southern portion of Guilford County. Tr. 987:20-988:5; PX344; PX329 at 20 (Cooper Rebuttal Report). Plaintiffs’ Exhibit 344 depicts the Dr. Hofeller’s Maptitude file showing the Guilford grouping. 386. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries the mapmaker drew for House Districts 58, 59, and 60. 172 387. The simulations of Plaintiffs’ other experts independently establish that the Guilford county grouping is an extreme partisan gerrymander. 388. Drs. Chen, Mattingly, and Pegden all froze three districts in this grouping that were substantially redrawn by the Covington Special Master: House Districts 57, 61, and 62. Tr. 352:24-353:3; PX359 at 33 (Mattingly Report); PX508 at 19 (Pegden Report). 389. Dr. Chen explained in unrebutted testimony that his simulations of the Guilford House grouping did not make any changes to the portion of House District 59 added by the Special Master. Tr. 770:12-771:12; Chen Demonstrative D5 at 4. The Court finds that because Dr. Chen’s simulations altered only portions of House District 59 drawn by the mapmaker, and did not touch the very small portions of the district added by the Special Master, the mapmaker necessarily is responsible for the extreme partisan bias that Dr. Chen finds for House District 59. 390. Dr. Chen found that all three districts in the Guilford grouping that he did not freeze are extreme partisan outliers. He found that House Districts 58 and 60 have higher Democratic vote shares than their corresponding districts in all of Dr. Chen’s simulations, while House District 59 has a much lower Democratic vote share that the corresponding district in all of the simulations. Tr. 353:17-21; PX45. Dr. Chen’s findings demonstrate the packing of Democratic voters into House Districts 58 and 60 to make House District 59 favorable for Republicans. Indeed, the least Democratic district in this grouping would be competitive or Democratic-leaning in Dr. Chen’s simulations, whereas House District 59 under the enacted plan is much less favorable for Democrats using the 2010-2016 statewide elections. The Court gives weight to Dr. Chen’s findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 45 below. 173 391. Plaintiffs’ Exhibit 398 shows Dr. Mattingly’s analysis of the Guilford grouping: 392. Setting aside the frozen districts, Dr. Mattingly concluded that the least Democratic district (House District 59) shows extreme cracking of Democrats, while the remaining two districts (House Districts 58 and 60) shows extreme packing of Democrats, 174 in comparison to the nonpartisan plans. PX398; PX778 at 30; PX359 at 33-34. Dr. Mattingly found that House 59 has fewer Democratic voters than 99.89% of the comparable districts in the nonpartisan ensemble, while House Districts 58 and 60 have more average Democratic votes than 99.86% of the comparable Democratic districts in the nonpartisan ensemble. PX778 at 30; PX359 at 33-34; PX398. As the figure above shows, the gerrymander could cause the Democrats to lose a seat in this grouping in certain electoral environments, because the black dot for House District 59 falls below the 50% line while the blue histogram sometimes rises above it or gets very close. Dr. Mattingly concluded that the Guilford House grouping is an extreme pro-Republican partisan gerrymander, Tr. 1155:5-16; PX778 at 30; PX359 at 33-34; PX398, and the Court gives weigh to Dr. Mattingly’s conclusion. 393. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 93.9% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 82% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:10; PX527. The Court gives weight to Dr. Pegden’s analysis and conclusions. 394. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. 175 g. 395. Davie-Rowan-Cabarrus-Stanly-Montgomery-Richmond The Davie-Rowan-Cabarrus-Stanly-Montgomery-Richmond House county grouping contains House Districts 66, 67, 76, 77, 82, and 83. The Court gives weight to the analysis of Plaintiffs’ experts and finds that significant portions of this county grouping are an extreme partisan gerrymander. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 176 396. Plaintiffs’ Exhibit 314 is Dr. Cooper’s map for this county grouping: 397. This county grouping cracks Democratic voters across its districts. In particular, Dr. Cooper explained how the mapmaker “maximize[d] partisan advantage” by splitting municipalities in “critical ways” that crack Democratic voters. Tr. 926:18-24. The cities of Kannapolis and Concord are both split across House Districts 82 and 83, cracking the Democratic voters across these districts to dilute their voting power. Tr. 926:23-927:24; 177 PX253 at 87-88 (Cooper Report). The Democratic voters from both of these cities are kept separate from the Democratic voters in Salisbury, which is placed in House District 76. Id. Plaintiffs Exhibit 315 depicts the splitting and treatment of these municipalities (Concord is shaded green, Kannapolis is pink, and Salisbury is yellow). 398. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 178 399. Dr. Chen found that, in his House Simulation Set 1, one of the districts in this grouping, House District 83, is an extreme partisan outlier, as it has a lower Democratic vote than its corresponding district in nearly all of the simulations. Tr. 363:612; PX46. Dr. Chen further found, however, that this grouping has three districts (House Districts 76, 82, and 83) that are partisan outliers in his House Simulation Set 2 that avoided pairing the incumbents in office in 2017. Tr. 363:14-364:10; PX70. Dr. Chen’s findings demonstrate the cracking of Democratic voters across the districts in this grouping, particularly given Legislative Defendants’ representations that the General Assembly sought to avoid pairing incumbents in 2017. See Tr. 364:11-22. The Court gives weight to Dr. Chen’s findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 70 below. 179 400. Plaintiffs’ Exhibit 392 shows Dr. Mattingly’s analysis of this grouping: 401. When Dr. Mattingly mathematically quantified cracking in this grouping across all 17 statewide elections, he found that the four most Democratic districts in the Davie grouping had more Democrats than in 97.38% of plans in the nonpartisan ensemble. PX359 at 30; PX778 at 30; PX392.9 Dr. Mattingly concluded that this grouping reflects an “anomalous structure,” Tr. 1156:1-16, and the Court gives weight to that conclusion. 402. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that significant portions of this county grouping are an extreme partisan gerrymander that was drawn to dilute the votes of Democratic voters and maximize the number of Republican districts in this grouping. 9 Dr. Pegden’s conservative methodology resulted in comparison maps that are very similar to the enacted plan for this grouping. Tr. 1351:17-1352:10. In particular, Dr. Pegden’s conservative choice to allow his algorithm to split the same municipalities that are split under the enacted plan results in his comparison maps frequently splitting the Democratic strongholds of Kannapolis and Concord. PX535; PX508 at 24 (Pegden Report). 180 h. 403. Yadkin-Forsyth The Yadkin-Forsyth House County grouping contains House Districts 71, 72, 73, 74, and 75. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 404. Plaintiffs’ Exhibit 316 is Dr. Cooper’s map for this county grouping: 405. Legislative Defendants packed Democratic voters into House Districts 71 and 72. Tr. 928:20-21; PX253 at 90 (Cooper Report). Legislative Defendants then cracked the remaining Democratic voters in this grouping across the remaining districts, where those Democratic voters’ influence is washed out by heavily Republican VTDs. House District 73 includes all of Republican-leaning Yadkin County and just two Democratic-leaning VTDs on the west side of Winston-Salem, ensuring that it will be a safe Republican district. House Districts 74 and 75 include Democratic-leaning VTDs on the northern and southern 181 sides of Winston-Salem, respectively, but both of those districts wrap around the city to include Republican-dominated VTDs on either side of Forsyth County. Indeed, in order to join Republican VTDs, House District 75 traverses an extremely narrow passageway on the border of Forsyth County. Tr. 928:5-21; PX253 at 90-91 (Cooper Report). 406. The Maptitude files from Dr. Hofeller’s hard drive illustrate the “anatomy of this gerrymander.” Tr. 988:17-989:4; PX345; PX329 at 21 (Cooper Rebuttal Report). They show Dr. Hofeller’s intentional packing of all of the most Democratic VTDs in Forsyth County into House Districts 71 and 72, while putting all of the moderate and Republicanleaning VTDs (shaded tan, yellow, light green, and red) into House Districts 73, 74, and 75. Id. Plaintiffs’ Exhibit 345 shows Dr. Hofeller’s Maptitude file containing this county grouping: 182 407. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 408. The simulations of Plaintiffs’ other experts independently establish that the Forsyth-Yadkin county grouping is an extreme partisan gerrymander. 409. Dr. Chen found that, in his House Simulation Set 1, two of the districts in this grouping (House Districts 71 and 75) are extreme partisan outliers above the 95% level, and another two districts in the grouping (House Districts 72 and 74) have higher or lower Democratic vote shares than over 80% of their corresponding districts. Tr. 354:1-20; PX49. Dr. Chen further found, however, that all four of these districts are extreme partisan outliers in his House Simulation Set 2 that avoided pairing the incumbents in office in 2017. Tr. 355:1-18. In Simulation Set 2, House Districts 71 and 72 have higher Democratic vote shares than nearly all of their corresponding districts in the simulations, while House Districts 74 and 75 have lower Democratic vote shares than nearly all of their corresponding districts in the simulations. Id. Dr. Chen’s findings demonstrate the packing of Democratic voters into House Districts 71 and 72 and the cracking of Democratic voters in the remaining districts in this grouping, particularly given Legislative Defendants’ representations that the General Assembly sought to avoid pairing incumbents in 2017. See Tr. 355:19-356:4. The Court gives weight to Dr. Chen’s findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 67 below. 183 410. Plaintiffs’ Exhibit 414 shows Dr. Mattingly’s analysis of this grouping: 184 411. Dr. Mattingly concluded that the three least Democratic districts show extreme cracking of Democrats while the two most Democratic districts shows extreme packing of Democrats, as evidenced by the significant jump between these sets of districts. Tr. 1144:3-9. Dr. Mattingly’s analysis showed that the three least Democratic districts in the enacted plan had fewer average Democratic votes than 99.46% of the comparable districts in the nonpartisan ensemble, while the two most Democratic districts in the enacted plan had more average Democratic votes than 99.84% of the comparable Democratic districts in the nonpartisan ensemble. PX778 at 30; PX359 at 44. As the figure above shows, the gerrymander causes the Democrats to lose one, possibly two, seats in this grouping in certain electoral environments, because the black dots for House District 74 and 75 always below the 50% line while the blue histograms sometimes rise above it. Tr. 1144:6-9. Dr. Mattingly concluded that the Yadkin-Forsyth grouping is an extreme proRepublican partisan gerrymander, Tr. 1144:13-16, and the Court gives weight to his conclusion. 412. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.7% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.1% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:7; PX530. The Court gives weight to Dr. Pegden’s analysis and conclusions. 413. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. 185 i. 414. Mecklenburg The Mecklenburg House County grouping contains House Districts 88, 92, 98, 99, 100, 101, 102, 103, 104, 105, 106, and 107. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 415. Plaintiffs’ Exhibit 319 is Dr. Cooper’s map for this county grouping: 416. Dr. Cooper detailed how House Districts 88, 92, and 101 pack Democratic voters on the western side of Mecklenburg County while House Districts 99, 100, 102, and 106 pack Democratic voters on the eastern and central portions of the county. There is not 186 a single Republican-leaning VTD included in any of these packed House Districts. Tr. 930:13-24; PX253 at 93 (Cooper Report). 417. House Districts 103, 104, and 105, meanwhile, include all of the Republican- leaning VTDs on the southern side of Mecklenburg County, allowing those districts to be “as competitive as possible for Republicans.” Tr. 930:25-931:7; PX253 at 93 (Cooper Report). 418. House District 98, on the northern boundary of Mecklenburg County, includes almost all Republican-leaning VTDs, avoiding the Democrat-heavy VTDs that are packed into House Districts 106 and 107. Tr. 931:7; PX253 at 93 (Cooper Report). 419. As depicted in Plaintiffs’ Exhibit 320, these district boundaries split Charlotte between 11 House Districts but manage to place every Republican-leaning VTD within the city—the “red pizza” slice—into House Districts 103, 104, and 105. Tr. 932:1-17; PX320; PX253 at 93 (Cooper Report). 420. Dr. Hofeller’s Maptitude files confirm he drew the districts in this grouping to maximize partisan gain. The “pizza slice” that contains the Republican-leaning VTDs within Charlotte is evident in Dr. Hofeller’s color-coded draft map, which groups those 187 Republican-leaning VTDs into three House Districts and packs almost all of the Democratic VTDs into other districts. Tr. 990:4-21; PX329 at 22 (Cooper Rebuttal Report). Plaintiffs’ Exhibit 346 shows Dr. Hofeller’s Maptitude files containing this county grouping: 421. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 422. The simulations of Plaintiffs’ other experts independently establish that the Mecklenburg county grouping is an extreme partisan gerrymander. 423. Dr. Chen found that this county grouping contains six districts that are extreme partisan outliers above the 95% outlier level, and another three districts that are 188 outliers above the 90% level. Tr. 361:20-22; PX53. The enacted plan packs Democratic voters into a number of districts in order to create four districts—House Districts 98, 103, 104, and 105—that are less Democratic than all of nearly of their corresponding districts in Dr. Chen’s simulations. PX53. Dr. Chen’s findings demonstrate the packing and cracking of Democratic voters in this grouping. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which is reflected in Plaintiffs’ Exhibit 53 below. 424. As Dr. Chen explained at trial, the fact that Democrats won House Districts 98, 103, 104, and 105 by small or extremely small margins in 2018 does not contradict his findings. Tr. 362:2-363:2; see JSF ¶¶ 125, 132-35. Rather, Dr. Chen’s simulations suggest that Democrats very likely would have won each of these districts by larger margins if not for the gerrymander. Id. Moreover, Dr. Hofeller’s own assessment of these districts demonstrates that he believed these districts to be Republican-leaning, and that it took the 189 Democratic wave of 2018 to squeak out wins in them. Dr. Hofeller estimated that House District 98 would have a 62.76% Republican vote share and he characterized it as a “strong Rep. district in Mecklenburg.” PX246 at 3. Dr. Hofeller similarly estimated that House Districts 103, 104, and 105 would have 62% to 64% Republican vote shares. Id. Dr. Hofeller’s spreadsheets evidence the partisan intent behind the creation of these districts and the strong possibility that Democratic could lose them in the next election under the current district lines intended to produce that result. 425. Plaintiffs’ Exhibit 400 shows Dr. Mattingly’s analysis of this grouping: 426. Dr. Mattingly concluded that the four most Republican districts showed extreme cracking of Democrats while the next four districts showed extreme packing of Democrats, as evidenced by the significant jump between these sets of districts. Tr. 1138:71139:4. Dr. Mattingly found that the least four Democratic districts in the enacted plan had fewer average Democratic votes than 99.9% of the comparable districts in the nonpartisan ensemble, while the eight most Democratic districts in the enacted plan had 190 more average Democratic votes than 99.5% of the comparable Democratic districts in the nonpartisan ensemble. Tr. 1141:8-25; PX778 at 30; PX359 at 34-35. As the figure above shows, the gerrymander causes the Democrats to lose up to three, possibly four, seats in this grouping in certain electoral environments, because the black dots for House Districts 98, 103, 104, and 105 often fall below the 50% line while the blue histograms rise above it. Tr. 1140:12-1140:25. Dr. Mattingly concluded that this grouping is an extreme proRepublican partisan gerrymander, Tr. 1142:1-4, and the Court gives weight to his conclusion. 427. Like Dr. Chen, Dr. Mattingly explained that the fact that Democrats won all the seats in the Mecklenburg grouping in the 2018 election does not undermine his conclusion that the grouping is an extreme pro-Republican partisan gerrymander. Tr. 1142:5-14. That the Democrats did well in one election and were able to prevail over the gerrymander does not change the fact that the grouping provides an extreme and atypical structural advantage to the Republicans that could cause the Democrats to lose seats in the next election. Tr. 1142:10-17. 428. Dr. Pegden found that this county grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.994% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.98% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:5-6; PX531. The Court gives weight to Dr. Pegden’s analysis and conclusions. 429. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. 191 j. 430. Wake The Wake House county grouping contains House Districts 11, 33, 34, 35, 36, 37, 38, 39, 40, 41, and 49.10 431. Plaintiffs’ Exhibit 297 is Dr. Cooper’s map for this county grouping: 432. The 2017 versions of House Districts 11, 33, 38, and 49 packed Democratic voters to allow House Districts 35, 36, 37, and 40, on the north and south sides of Wake 10 Plaintiffs presented evidence at trial that the enacted 2017 version of the Wake House county grouping was a partisan gerrymander, but Plaintiffs presented no evidence regarding this grouping as revised pursuant to this Court’s ruling in North Carolina State Conference of NAACP Branches, et al. v. David Lewis, et al. Plaintiffs do not seek a remedy for the current, revised version of this grouping. However, the analysis and findings of Plaintiffs’ experts with respect to the 2017 version of this county grouping is evidence of Legislative Defendants’ intentional and systematic gerrymandering across the State during the 2017 redistricting. 192 County to be more favorable to Republicans. Tr. 911:15-912:16; PX253 at 65 (Cooper Report). 433. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these 2017 districts. 434. The simulations of Plaintiffs’ other experts independently establish that the 2017 enacted House plan version of the Wake grouping was an extreme partisan gerrymander. 435. Dr. Chen found that the 2017 version of this county grouping contained three districts that were extreme partisan outliers above the 95% outlier level. Tr. 365:15-366:1; PX54. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping. 436. Dr. Mattingly’s analysis showed that the four most Republican districts in the 2017 version of this grouping show extreme cracking of Democrats, while the next four districts show extreme packing of Democrats, in comparison to the nonpartisan plans. PX412; PX778 at 30; PX359 at 43. His analysis showed that the least Democratic districts in the enacted plan had fewer Democratic voters than 99.98% of the comparable districts in the nonpartisan ensemble, while the most Democratic districts in the enacted plan had more average Democratic votes than 99.99% of the comparable Democratic districts in the ensemble. PX778 at 30; PX359 at 43; PX412. The Court gives weight to Dr. Mattingly’s analysis and conclusions for this grouping. 437. Dr. Pegden found that the 2017 version of this grouping constituted an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.9997% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.9991% of all possible districtings of 193 this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:4; PX533. The Court gives weight to Dr. Pegden’s analysis and conclusions. 438. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that the 2017 version of this county grouping was an extreme partisan gerrymander. While Plaintiffs do not challenge any individual House districts in Wake County as currently drawn, the Court gives weight to the findings and conclusions of Plaintiffs’ experts in regard to the consistency of the partisan intent throughout the statewide map. k. 439. New Hanover-Brunswick The New Hanover-Brunswick House county grouping, drawn in 2011 and left unchanged in 2017, contains House Districts 17, 18, 19, and 20. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 194 440. Plaintiffs’ Exhibit 302 is Dr. Cooper’s map of this county grouping: 441. As Dr. Cooper testified, House District 18 packs the most Democratic-leaning VTDs in this grouping into that district, thereby making House Districts 17, 19, and 20 more favorable to Republicans. Tr. 913:17-914:7; PX253 at 72 (Cooper Report). 442. Wilmington is split between House Districts 18, 19, and 20, with the most Democratic-leaning VTDs in that city packed into House District 18 and the Republicanleaning VTDs placed in the two adjacent districts. In order to accomplish the packing of voters in House District 18, the district boundaries split Wilmington and the UNC Wilmington campus. Tr. 914:13-20; PX253 at 73 (Cooper Report); PX303. By dividing the campus in this manner, the district boundaries enable House District 20 to connect to Republican-leaning VTDs in the Wilmington area, creating a boot-like appendage in the southwest portion of House District 20. PX253 at 75 (Cooper Report); Tr. 916:12-21. 195 Plaintiffs’ Exhibit 303 show which portions of Wilmington are placed into each of the three districts: 443. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 444. The simulations of Plaintiffs’ other experts independently establish that the Brunswick-New Hanover county grouping is an extreme partisan gerrymander. 445. Dr. Chen found that this county grouping contains three districts that are extreme partisan outliers. Tr. 369:3-7.11 House District 18 has a higher Democratic vote share than its corresponding district in all the simulations, while House Districts 17 and 19 have lower Democratic vote shares than their corresponding districts in all or nearly all of the simulations. Dr. Chen’s findings demonstrate the packing of Democratic voters in 11 For all House county groupings drawn in 2011 and unchanged in 2017, Dr. Chen used the 2004 to 2010 statewide elections to analyze these county groupings. 196 House District 18 and the cracking of Democratic voters across the other districts. The vast majority of Dr. Chen’s simulations would produce up to two additional districts in this grouping that are competitive or even Democratic-leaning, compared to the enacted plan. PX57. The Court gives weight to Dr. Chen’s analysis and findings for this grouping, which are reflected in Plaintiffs’ Exhibit 57 below: 197 446. Plaintiffs’ Exhibit 404 shows Dr. Mattingly’s analysis of this grouping: 447. Dr. Mattingly concluded that the most Democratic district shows extreme packing of Democrats, while the three least Democratic districts show extreme cracking of Democrats, as evidenced by the significant jump between these sets of districts. Tr. 1145:17-1146:12. Dr. Mattingly found that the most Democratic district in the enacted plan had more Democratic voters than 92.01% of the comparable districts in the nonpartisan ensemble. PX778 at 30; PX359 at 38. As the figure above shows, the enacted map causes the Democrats to lose one seat in this grouping in certain electoral environments, because the black dot in the second most Democratic district always falls below the 50% line while the blue histograms often rise above it. Tr. 1146:5-9. Dr. Mattingly concluded that the New Hanover-Brunswick House grouping reflected a pro-Republican partisan gerrymander, Tr. 1146:22-1147:2, and the Court gives weight to his conclusion. 448. Dr. Pegden found that this county grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version 198 of this grouping is more favorable to Republicans than 99.97% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.91% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:6-7; PX524. The Court gives weight to Dr. Pegden’s analysis and conclusions. 449. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. l. 450. Duplin-Onslow The Duplin-Onslow House county grouping, drawn in 2011 and left unchanged in 2017, contains House Districts 4, 14, and 15. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 199 451. Plaintiffs’ Exhibit 291 is Dr. Cooper’s map for this county grouping: 452. Legislative Defendants split Jacksonville across House Districts 14 and 15, pairing the Democratic-leaning “shark’s tooth” in Jacksonville with heavily Republicanleaning VTDs in House District 15. Tr. 906:10-23; PX253 at 53-57 (Cooper Report). The map also ensures that none of Jacksonville’s voters are joined with the Democratic-leaning and moderate VTDs in Duplin County, in House District 4. Id. The map cracks Democratic 200 voters across all three districts in this grouping, ensuring that House District 14 “becomes Republican and [House District 4] also stays safely Republican.” Id. 453. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 454. The simulations of Plaintiffs’ other experts independently establish that the Duplin-Onslow county grouping is an extreme partisan gerrymander. 455. Dr. Chen found that all three districts in this grouping are extreme partisan outliers. Tr. 370:16-371:1. House Districts 4 and 14 have lower Democratic vote shares than their corresponding districts in nearly all the simulations, while House District 15 has a higher Democratic vote share than its corresponding district in nearly all the simulations. PX60. Dr. Chen’s findings demonstrate the cracking of Democratic voters across the three districts. The vast majority of Dr. Chen’s simulations would produce two districts that are more competitive using the 2004-2010 statewide elections compared to the enacted plan. PX60. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, reflected in Plaintiffs’ Exhibit 60: 201 456. Plaintiffs’ Exhibit 394 shows Dr. Mattingly’s analysis of this grouping: 457. This grouping is another example of what Dr. Mattingly called “squeezing” or “flattening,” where Democrats are cracked across all of the districts in the grouping. See 202 Tr. 1149:19-1150:2; Tr. 1150:22-1151:2. Dr. Mattingly’s analysis showed that the two most Democratic districts in the enacted plan had fewer Democratic voters than 92.4% of the comparable districts in the nonpartisan ensemble, meaning that the Duplin-Onslow House grouping showed clear cracking of Democratic voters. PX778 at 30; PX359 at 31. As the figure above shows, the gerrymander could cause the Democrats to lose at least one seat in certain electoral environments. Dr. Mattingly concluded that this grouping reflects a clear pro-Republican partisan gerrymander, Tr. 1155:17-21, PX778 at 30, and the Court gives weight to Dr. Mattingly’s conclusion. 458. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 98% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 94% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:9; PX528. The Court gives weight to Dr. Pegden’s analysis and conclusions. 459. The Court finds that the analyses of all Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. m. 460. Anson-Union The Anson-Union county grouping, drawn in 2011 and left unchanged in 2017, contains House Districts 55, 68, and 69. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 203 461. Plaintiffs’ Exhibit 307 is Dr. Cooper’s map for this county grouping: 462. Dr. Cooper detailed how this county grouping cracks the Democratic voters in Monroe between two districts (House Districts 68 and 69), and then ensures that none of these voters are joined with the Democratic voters in Anson County (in House District 55). The map thus dilutes the voting power of the Democratic voters in this grouping, ensuring that House Districts 68 and 69 are reliable Republican districts. Tr. 919:3-16; PX253 at 7980 (Cooper Report). Plaintiffs’ Exhibit 308 illustrates the cracking of Monroe (which is colored pink). 204 463. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 464. Dr. Hofeller’s Maptitude files confirm his intentional use of partisanship data to crack Democratic voters. The relevant Maptitude file, which was last modified in June 2011 and is depicted in Plaintiffs’ Exhibit 353 below, shows Dr. Hofeller’s use of the 2008 Presidential election results to separate Democratic VTDs across the three districts in this grouping. Tr. 995:20-998:7; PX329 at 31 (Cooper Rebuttal Report). 205 465. The simulations of Plaintiffs’ other experts independently establish that this county grouping is an extreme partisan gerrymander. 466. Dr. Chen found that all three districts in this county grouping are extreme partisan outliers. Tr. 368:7-15. House District 55 has a lower Democratic vote share than its corresponding district in nearly all of the simulations, while House Districts 68 and 69 have higher Democratic vote shares than their corresponding districts in nearly all of the simulations. Dr. Chen’s findings demonstrate the cracking of Democratic voters across the three districts in this grouping. In the vast majority of Dr. Chen’s simulations, this county grouping would produce a district that is Democratic-leaning using the 2004-2010 statewide elections. PX56. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 56 below: 206 467. Plaintiffs’ Exhibit 410 shows Dr. Mattingly’s analysis of this grouping: 207 468. This grouping is another example of what Dr. Mattingly called “squeezing” or “flattening,” where the Democrats are cracked across all of the districts in the grouping. See Tr. 1149:19-1150:2; Tr. 1150:22-1151:2. Dr. Mattingly’s analysis showed that the two most Democratic districts in the enacted plan had fewer Democratic voters than 100% of the comparable districts in the nonpartisan ensemble, meaning that not a single plan in his nonpartisan ensemble showed as much cracking of Democratic voters in this grouping as the enacted plan. PX778 at 30; PX359 at 42. As the figure above shows, the gerrymander causes the Democrats to lose one seat in certain electoral environment, as the black dot for House District 55 is always below the dotted line but the blue histogram often rises above it. Dr. Mattingly concluded that the Anson-Union House grouping reflected an extreme pro-Republican partisan gerrymander, Tr. 1155:8-16, PX778 at 30, and the Court gives weight to his conclusion. 469. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 98.5% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 95.5% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:8-9; PX523. The Court gives weight to Dr. Pegden’s analysis and conclusions. 470. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. 208 n. 471. Alamance The Alamance House county grouping, drawn in 2011 and left unchanged in 2017, contains House Districts 63 and 64. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 472. Plaintiffs’ Exhibit 311 is Dr. Cooper’s map for this county grouping: 473. Dr. Cooper described how House District 63 takes the shape of a “duck’s head” in the Burlington area, cracking the Democratic voters in and around Burlington between House Districts 63 and 64 to reduce those voters’ influence. Tr. 924:3-25; PX253 at 84 (Cooper Report). And the map carefully places Burlington’s Republican-leaning-VTDs 209 (in the “duck’s head”) in House Districts 63, helping to ensure that House District 63 will consistently elect a Republicans. Plaintiffs’ Exhibit 312 depicts the division of Burlington (shaded green): 474. Dr. Hofeller’s Maptitude files confirm the partisan intent and “partisan consequences” of cracking Democratic voters in this grouping. Tr. 998:18-19. In particular, Dr. Hofeller’s draft map for House Districts 63 and 64 (which was last modified in June 2011 while this district was being drawn) demonstrates how the “duck’s head” portion put Burlington’s most moderate and Republican-leaning VTDs (shaded tan and light green) in House District 63, while Burlington’s bluest VTDs were grouped with heavily Republican 210 areas in northern and southern Alamance County. Tr. 998:9-25; PX354; PX329 at 32 (Cooper Rebuttal Report). Plaintiffs’ Exhibit 354 shows Dr. Hofeller’s Maptitude file containing the Alamance grouping. 475. Election results demonstrate that the gerrymandering of this grouping has been highly effective. Although Intervenor Defendants presented testimony claiming that “candidate quality” resulted in the Democratic loss in one of the districts in 2018 (Tr. 2245:9-2246:25), in fact, Republicans have won both districts in this grouping in all four elections since the districts were drawn in 2011, across a range of candidates. JSF at Ex. 2; Tr. 2253:15-2256:10. 476. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of the districts in this county groupings. 211 477. The simulations of Plaintiffs’ other experts independently establish that the Alamance county grouping is an extreme partisan gerrymander. 478. In his House Simulation Set 1, Dr. Chen found that House District 63 has a lower Democratic vote than its corresponding district in over 77% of the simulations while House District 64 has a higher Democratic vote share than its corresponding district in over 74.5% of the simulations. Tr. 371:10-372:6; PX55. More importantly, Dr. Chen found that both districts in this county grouping are extreme partisan outliers in House Simulation Set 2 that avoids pairing the incumbents in office at the time this grouping was drawn. Tr. 372:8-373:5; PX76. Dr. Chen thus concluded with over 99% statistical certainty that the districts in this grouping are extreme partisan outliers if the mapmaker was trying to protect incumbents in drawing the districts in the grouping. Tr. 372:23-373:5. Indeed, across the vast majority of 2,000 simulations in House Simulation Sets 1 and 2, this county grouping would produce a Democratic-leaning district in the simulations, whereas it does not in the enacted plan. PX55; PX76. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 76 below: 212 [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 213 479. Plaintiffs’ Exhibit 384 shows Dr. Mattingly’s analysis of this grouping: 480. This grouping reflects what Dr. Mattingly called “squeezing” or “flattening,” where Democratic districts are cracked across all of the districts. Tr. 1149:19-1151:2. Dr. Mattingly found that this grouping reflected more cracking of Democratic voters than 77% of the comparable districts in the nonpartisan ensemble. Tr. 1151:10-17; PX778 at 30; PX359 at 26. Although Dr. Mattingly did not label this grouping an “outlier” because he used a 90% threshold, he testified that the pro-Republican bias in the grouping still contributed to the extreme pro-Republican bias he found statewide. Tr. 1151:21-1153:2, Tr. 1154:23-1155:1. What’s more, the pro-Republican tilt has a significant effect; as the figure above shows, the gerrymander causes the Democrats to lose one seat in this grouping in many electoral environments. Tr. 1151:3-9. Dr. Mattingly concluded that the Alamance House grouping reflected a clear pro-Republican partisan tilt, Tr. 1151:24-1153:2; PX778 at 30, and the Court gives weight to his conclusion. 214 481. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version of this grouping is more favorable to Republicans than 99.9998% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.996% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:5; PX522. The Court gives weight to Dr. Pegden’s analysis and conclusions. 482. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. o. 483. Cleveland-Gaston The Cleveland-Gaston House county grouping, drawn in 2011 and left unchanged in 2017, contains House Districts 108, 109, 110, and 111. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 215 484. Plaintiffs’ Exhibit 323 is Dr. Cooper’s map for this county grouping: 485. As Dr. Cooper testified, this grouping is a textbook example of cracking. The Democratic voters in Gastonia are cracked across House Districts 108, 109, and 110, and the Democratic voters in Shelby across House Districts 110 and 111. Tr. 932:23-934:10; PX253 at 97-98 (Cooper Report). Plaintiffs’ Exhibit 325 illustrates the splitting of these municipalities: 216 486. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 487. The simulations of Plaintiffs’ other experts independently establish that the Cleveland-Gaston county grouping is an extreme partisan gerrymander. 488. Dr. Chen found that this county grouping contains three districts that are extreme partisan outliers. Tr. 370:5-13. House Districts 109 and 111 have lower Democratic vote shares than their corresponding district in all or nearly all of the simulations, while House District 108 has a higher Democratic vote shares than its corresponding district in all of the simulations. PX59. Dr. Chen’s findings demonstrate the cracking of Democratic voters across the districts in this county grouping. The Court gives weight to Dr. Chen’s analysis and findings for this county grouping, which are reflected in Plaintiffs’ Exhibit 59 below. 217 [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 218 489. Plaintiffs’ Exhibit 396 shows Dr. Mattingly’s analysis of this grouping: 490. This grouping reflects what Dr. Mattingly called “squeezing” or “flattening,” where Democratic voters are cracked across all of the districts. See Tr. 1149:19-1150:2; Tr. 1150:22-1151:2. Dr. Mattingly found that this grouping reflected more cracking of Democratic voters than 82.86% of the comparable districts in the nonpartisan ensemble. PX778 at 30; PX359 at 32. Although he did not label this grouping an “outlier” because he used a 90% threshold, he testified that the pro-Republican bias in the Gaston-Cleveland still contributed to the extreme pro-Republican bias he found statewide. See Tr. 1151:211156:21. Moreover, as the figure above shows, the gerrymander could cause Democrats to lose at least one seat in certain electoral environments. Dr. Mattingly concluded that the Gaston-Cleveland grouping reflects a clear pro-Republican partisan tilt that can contribute to the extreme pro-Republican bias statewide, Tr. 1156:17-24, PX778 at 30, and the Court gives weight to his conclusion. 219 491. Dr. Pegden’s conservative methodology resulted in comparison maps that are very similar to the enacted plan for this grouping. Tr. 1351:17-1352:10. 492. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. p. 493. Buncombe The Buncombe House county grouping, drawn in 2011 and left unchanged in 2017, contains House Districts 114, 115, and 116. The Court gives weight to the analysis of Plaintiffs’ experts and finds that this county grouping is an extreme partisan gerrymander. 494. Plaintiffs’ Exhibit 326 is Dr. Cooper’s map for this county grouping: 495. The mapmaker packed the most Democratic VTDs in and around Asheville into House District 114, in an effort to make House Districts 115 and 116 as competitive for Republicans as possible. Tr. 934:17-935:1; PX253 at 100 (Cooper Report). 220 496. The Court does not give weight to any nonpartisan explanation Legislative Defendants offered with respect to the boundaries of these districts. 497. The simulations of Plaintiffs’ other experts independently establish that the Buncombe county grouping is an extreme partisan gerrymander. 498. Dr. Chen found that all three districts in this county grouping are extreme partisan outliers. Tr. 369:22-370:1. House District 114 has a higher Democratic vote share than its corresponding district in all the simulations, while House Districts 115 and 116 have lower Democratic vote shares than their corresponding districts in all the simulations. Dr. Chen’s findings demonstrate the packing of Democratic voters into House District 114 to make House Districts 115 and 116 as competitive for Republicans as possible. PX58. The Court gives weight to Dr. Chen’s analysis and findings for this grouping, which are reflected in Plaintiffs’ Exhibit 58: 221 499. Plaintiffs’ Exhibit 386 shows Dr. Mattingly’s analysis of this grouping: 500. Dr. Mattingly’s analysis shows that Democrats were cracked out of the two least Democratic districts in this grouping and packed into the most Democratic district. PX778 at 30; PX359 at 27; PX386. The two least Democratic districts in the enacted plan had fewer Democratic voters than 85.45% of the comparable districts in the nonpartisan ensemble. PX778 at 30; PX359 at 27; PX386. Although Dr. Mattingly did not label this grouping an “outlier” because he used a 90% threshold, he explained that the proRepublican bias still contributed to the extreme pro-Republican bias he found statewide. See Tr. 1151:21-1156:24. As the figure above shows, the gerrymandering could cause Democrats to lose one or two districts in certain electoral environments. Dr. Mattingly concluded that the Buncombe House grouping reflected a pro-Republican partisan bias, Tr. 1156:17-21, and the Court gives weight to his conclusion. 501. Dr. Pegden found that this grouping constitutes an extreme partisan gerrymander. In his first level analysis, Dr. Pegden found that the enacted plan’s version 222 of this grouping is more favorable to Republicans than 99.9997% of the maps that his algorithm encountered by making small changes to the district boundaries. In his second level analysis, Dr. Pegden found that this grouping is more carefully crafted to favor Republicans than at least 99.999% of all possible districtings of this county grouping that satisfy the criteria Dr. Pegden used. Tr. 1351:4-5; PX525. The Court gives weight to Dr. Pegden’s analysis and conclusions. 502. The Court finds that the analyses of Plaintiffs’ experts independently and together demonstrate that this county grouping is an extreme partisan gerrymander. D. The 2017 Plans Protected the Republican Majorities in the 2018 Elections 503. In the 2018 House elections, Republican candidates won a minority—48.8%— of the two-party statewide vote, but still won 65 of 120 seats (54%). JSF ¶¶ 68-69. Democrats thus broke the Republican supermajority, but not the majority. Id.; Tr. 163:21164:19 (Rep. Meyer). 504. In the 2018 Senate elections, Republican candidates won a minority— 49.5%—of the two-party statewide vote, but still won 29 of 50 seats (58%). JSF ¶¶ 142-43; Tr. 117:5-19 (Sen. Blue). Democrats broke the Republican supermajority by a single seat, after narrowly prevailing in Senate Districts 9 and 27 by margins of 0.1% and 0.5%. Id. 505. Democrats were unable to win majorities in either chamber despite strong efforts to fuel voter enthusiasm, recruit candidates, and fundraise, and despite favorable political conditions nationally and in North Carolina. Tr. 76:5-11 (Phillips); Tr. 118:19-21, 124:9-13 (Sen. Blue); Tr. 163:21-164:5 (Rep. Meyer); Tr. 1269:4-14, 1283:15-1284:1 (Goodwin). Democrats raised and spent more money than Republicans in the 2018 cycle, running the most well-funded campaign operation in the history of North Carolina. Tr. 223 117:20-117:25, 124:20-24 (Sen. Blue); Tr. 163:21-164:5, 171:3-6 (Rep. Meyer); Tr. 1284:11-17 (Goodwin). 506. Consistent with the findings of Drs. Chen and Mattingly, Senator Blue testified that, under the current Senate plan, Democrats would have needed to win over 55% of the statewide vote to win a majority of seats in the Senate. Tr. 119:19-120:4. E. The 2017 Plans Harm the Organizational and Individual Plaintiffs 1. 507. The 2017 Plans Harm the North Carolina Democratic Party Elections, voting, and redistricting are central to the mission and purposes of Plaintiff the North Carolina Democratic Party (the “NCDP”). The NCDP is “an association of like-minded individuals”—“predominantly registered Democrats”—“who support and also help develop policies that they agree on.” Tr. 1264:1-6 (Goodwin). As the NCDP’s chair, Mr. Goodwin testified, the “basic purpose” of the NCDP is to “encourage like-minded folks to come together, to help recruit candidates and to support candidates who favor those policies and favor the development of policies that Democrats support.” Tr. 1265:2-5. The NCDP “persuade[s] voters to support the nominees of the Democratic Party during the general election.” Tr. 1265:7-9. The Court gives weight to Mr. Goodwin’s testimony regarding the NCDP’s mission and purposes. 508. The Court gives further weight to Mr. Goodwin’s testimony that district lines significantly affect the NCDP’s ability to fulfill its mission and purposes. To achieve its purposes, the NCDP must “have good candidates that we recruit and that we support”; it needs “enthusiasm for the party and its candidates and its message and mission”; and it needs “the appropriate financial resources to get a message [out]” and to fund all “the things that are involved with elections.” Tr. 1264:15-21. All of those things are affected by district boundaries. Tr. 1265:22-24. For that reason, to “accomplish [NCDP’s] mission,” it is 224 “vital” that the NCDP have “fair, nondiscriminatory district lines for the candidates that run in districts across the State.” Id. 509. The current district lines have harmed the NCDP and will continue to do so. The lines drawn in 2011 “had a tremendously negative impact on the ability of the North Carolina Democratic Party to achieve the purposes for which it exists.” Tr. 1266:9-16. Under the 2011 districts, “it was more difficult to recruit candidates, it was more difficult to raise the funds necessar[]y, [and] enthusiasm was down tremendously because of . . . unfair []districts.” Id. 510. Upon enactment of the 2017 Plans, the NCDP “knew it was still going to be a difficult, difficult race because of . . . [the] district lines.” Tr. 1267:11-13. Because of the 2017 Plans, the NCDP “had to expend extraordinary amounts of time and resources and the like in a way that, in a set of fair maps across the State, [it] wouldn’t have had to do.” Tr. 1270:10-14; see Tr. 1284:18-22. The NCDP had to spend more money than it would have under nonpartisan maps, both statewide and in individual districts. For example, in House District 103 in Mecklenburg County, “to make that election competitive,” Democrats had to recruit the daughter of former Governor Jim Hunt and “her election had to be financed at a level that no previous House election had ever been financed in the history of state elections,” with Democrats spending over a million dollars in support of Ms. Hunt. Tr. 189:17-190:23 (Rep. Meyer). Even then, Ms. Hunt won the election by fewer than 100 votes. Id. The simulations of Drs. Chen and Mattingly each establish that, under nonpartisan maps, House District 103 in Mecklenburg County would be more favorable for Democrats than it is under the current House plan, FOF § C.2.i., meaning that Democrats would not need to devote as many resources to this district and would be able to spend those resources in other districts across the State instead. The Court finds that the NCDP has established that the current districts have injured the NCDP as an organization by 225 requiring it to spend and divert more financial resources than it would have under nonpartisan maps, both statewide and in individual districts 511. The Court finds that the current districts have injured the NCDP in other ways. As Mr. Goodwin testified, “notwithstanding the tremendous[,] palpable level of enthusiasm” for Democratic candidates nationwide and in North Carolina in 2018, “notwithstanding raising the most funds ever raised for a mid-term election for the [D]emocratic [P]arty,” and “notwithstanding the fact that . . . there was a [D]emocratic [G]overnor and [a] unique partnership” with the Governor, the NCDP’s “efforts and enthusiasm and . . . money did not translate into seats.” Tr. 1268:16-1269:3. “[D]espite everyone going [the NCDP’s] way, the lines were drawn in such a way that [the NCDP] could not breach that seawall that protected the [R]epublican majority.” Tr. 1268:13-15. 512. The Court finds that the current districts will also continue to injure the NCDP in the 2020 elections absent judicial relief. The NCDP will continue to need to spend and divert financial resources as a result of the gerrymanders, and it will continue to be extremely unlikely that Democratic candidates will be able to win majorities in either chamber of the General Assembly under the current districts. Moreover, although the NCDP was able to recruit a candidate in every district the favorable national environment that existed for Democrats in 2018 recruitment is more difficult under partisan plans. As Mr. Goodwin explained, unfair districts make it “more difficult to recruit candidates.” Tr. 1266:12-13. 513. In addition to harming the NCDP itself, the enacted plans also have harmed the NCDP’s members, and continue to do so. The NCDP’s members include every registered Democratic voter in North Carolina. Tr. 1269:8-17. There are “well over two million registered Democrats in North Carolina.” Tr. 1269:10-11. “There are registered Democrats in every precinct in the State, every House District, [and] every Senate District.” 226 Tr. 1269:15-20. The NCDP thus has members in every House and Senate district at issue in this case, and those members are harmed by the enacted plans. The gerrymanders dilute the voting power of the NCDP’s members by intentionally making it more difficult for some Democratic voters to elect candidates of their choice and making it extremely difficult for Democratic voters statewide to obtain Democratic majorities in the General Assembly. See FOF § E.3. 514. The NCDP’s “support scores” do not undermine the harms that the 2017 Plans cause the NCDP and its members. As Democratic Representative Graig Meyer testified, “support scores” are purchased scores that are assigned to all registered voters based on “a combination of consumer data as well as geographic and other factors that give you a sense of the likelihood someone is going to support a Democratic candidate.” Tr. 164:22-165:12. These scores are made available by the NCDP to Democratic candidates’ campaigns, Tr. 1270:24-1271:19 (Goodwin), which then, in their discretion, may use them “to determine which voters [they] should target for paid communications, such as digital or mail, or for individual communications, such as canvassing and knocking on voters’ doors,” Tr. 164:23-165:2 (Rep. Meyer). Even then, Democratic campaigns “almost always use [support scores] in conjunction with other measures, such as a turnout score, which tells you how likely someone is to actually vote.” Tr. 165:13-15. 515. Several of Legislative Defendants’ Exhibits purportedly show—based on support scores that are aggregated on a district-by-district basis—that Democratic candidates should be competitive, and in fact could win, in a comfortable majority of House and Senate districts under the 2017 Plans. See LDTX 145-147, 278; see Tr. 2072:21-2074:22 (Dr. Hood). 516. The Court gives little weight to Defendants’ arguments related to aggregated district-level support scores. Neither the NCDP nor any Democratic campaign or candidate 227 “ever use[s] . . . aggregated support scores for any purpose,” Tr. 1271:20-24 (Goodwin), and they do not use them “to determine the electability of a district,” Tr. 194:1-2 (Rep. Meyer). Support scores are “not reliable in the aggregate,” Tr. 167:5-6 (Rep. Meyer), and “[a]ggregated support scores wouldn’t be all that helpful because individual support scores can be misleading,” Tr. 165:24-166:1 (Rep. Meyer). “They’re imprecise measures, and then if you aggregate imprecise measures like that they tend to get less and less precise in the aggregate.” Tr. 166:7-9 (Rep. Meyer). Moreover, the aggregated support scores include all registered voters in a district, not likely or actual voters, which tends to overstate Democratic support. Tr. 2091:6-2092:14 (Dr. Hood). Rather than use aggregated support scores, the NCDP uses other metrics to assess a district’s competitiveness, such as the “Democratic Performance Index” (DPI) or the results of specific recent statewide elections. Tr. 1272:3-11 (Goodwin); Tr. 177:3-11 (Rep. Meyer). 517. Additionally, Legislative Defendants’ expert Dr. Hood, who presented an analysis based on the aggregated support scores, conceded that he is not aware of anyone who has ever “used those scores to make predictions” of how a district will perform in an election. Tr. 2092:3-24. Nor did Dr. Hood present any analysis to substantiate any claim that aggregated support scores are accurate predictors of a district’s competitiveness, and Representative Meyer credibly explained that they are not. Representative Meyer gave several examples where the district-level aggregated support scores differ considerably from actual election results, demonstrating why the NCDP and Democratic campaigns “don’t use support scores to determine electability of a district.” Tr. 194:1-2; see Tr. 193:17196:12. 518. Defendants presented no persuasive evidence that Democrats have a realistic possibility of winning majorities in the General Assembly under the metrics that are used 228 to assess a district’s likely performance, such as the DPI and prior statewide elections results. 519. The total number of registered Democrats in particular districts likewise does not undermine the harm the enacted plans cause the NCDP and its members. Legislative Defendants’ Exhibit 280 purportedly indicates that Democrats and unaffiliated voters, when combined together, hold a registration advantage over Republicans in all Senate districts and all House districts but one. See Tr. 1279:25-1281:15 (Goodwin). The Court gives little weight to Legislative Defendants’ arguments based on statewide party registration numbers. 520. As Mr. Goodwin explained, Legislative Defendants’ Exhibit 280 presents “an extreme hypothetical assuming that everyone who’s registered for his or her respective party actually vote and vote only based on their party registration, and assuming that unaffiliateds all vote for the Democratic candidate. That is not going to happen.” Tr. 1281:21:25. The notion that Democrats could win 169 of 170 total seats in the General Assembly is not credible. 521. As Dr. Chen further explained, party registration has been “studied in the academic literature[,] and it’s well known that in a lot of different Southern states, including in some parts of North Carolina, party registration is not necessarily a reliable indicator of one’s actual partisan voting habits.” Tr. 277:22-278:1. For example, “there are conservative Democrats, or what we call blue dog democrats sometimes, who in the past used to vote Democratic and have, for the last couple of decades, switched over to voting Republican, but their party registration may still remain as Democrats.” Tr. 278:3-10. 522. The Court finds that party registration is not a reliable indicator of the competitiveness of any individual district or of the enacted plans as a whole. 229 2. 523. The 2017 Plans Harm Common Cause Redistricting is central to the mission and purposes of Plaintiff Common Cause. Bob Phillips—Executive Director of Common Cause’s local chapter, Common Cause North Carolina—testified that Common Cause advocates for “[s]trengthening democracy” and “for more open, honest and accountable government.” Tr. 40:23-41:1, 41:10-16, 42:1317. And “there is nothing . . . that’s really more significant, consequential in a legislative session than redistricting.” Tr. 42:23-25. Redistricting “really locks in . . . everything” “for the next decade,” including “who gets elected and what the power share will be” and “[u]ltimately what kind of laws and policies are going to be emphasized and then [] will not be, what will be ignored.” Tr. 42:25-43:4. The Court gives weight to Mr. Phillips’s testimony. 524. Common Cause has long advocated to end partisan gerrymandering in North Carolina. Tr. 43:10-52:20. The 2017 Plans harm Common Cause as an organization by substantially impeding this longtime goal because, as Mr. Phillips testified, majorities in the General Assembly, as the beneficiaries of gerrymandered plans, are unlikely to adopt meaningful redistricting reform. Tr. 52:1-20. 525. The enacted plans also harm Common Cause by impeding its mission and objectives in other ways. As Mr. Phillips explained, “[o]ne of the central missions to Common Cause is to help citizens understand that they do have an obligation and that they can hold their elects accountable. How do you do that when so many—90 percent of our legislative seats are preordained . . . ?” Tr. 48:8-12. When “we already know [on] the filing date, basically, who is going to win,” it is “hard to get citizens, voters[,] to participate, to think that their vote really matters.” Tr. 48:25-49:3. 526. In addition to Common Cause itself, the enacted plans also harm Common Cause’s members. Common Cause has 25,000 members across North Carolina, including in the districts at issue here. See Tr. 41:17-42:12; PX644 (listing Common Cause members by 230 district). The enacted plans harm Common Cause’s members in the same ways they harm the NCDP’s members and the individual voter-plaintiffs in this case. 3. 527. The 2017 Plans Harm the Individual Plaintiffs The Individual Plaintiffs are thirty-seven individual North Carolina voters who prefer Democratic candidates and have consistently voted for Democratic candidates running for the North Carolina General Assembly. See PX678-714. 528. The evidence demonstrates that the 2017 Plans disadvantage the Individual Plaintiffs and other Democratic voters across North Carolina. Two of the Individual Plaintiffs testified live at trial, and the remaining 35 testified through affidavits. PX678714.12 529. Plaintiff Derrick Miller testified live at trial. Dr. Miller, a professor of German at the University of North Carolina Wilmington, resides in Senate District 8 in the “Wilmington Notch.” Tr. 202:11-14. Dr. Miller testified that by splitting off this small portion of Wilmington where he lives, the General Assembly has “made it impossible for [him] and [his] Democratic neighbors to elect a Democrat, a candidate of our choice, in Senate District 8.” Tr. 205:9-19. In 2018, the Republican candidate won Senate District 8 with around 60% of the vote. Tr. 204:3-4. As a fifth-generation North Carolinian, Dr. Miller cares deeply about issues such as public education and preserving North Carolina’s natural resources, and he believes that “Democrats much more reliably and [a] Democratic majority much more reliably would protect those resources, the educational resources and the natural resources of our state.” Tr. 206:8-12. 530. Dr. Miller also lives in House District 18, Tr. 204:5-7, where the General Assembly packed Democrats in Wilmington and Leland into a single, reliably Democratic 12 See, however, COL § I.C., wherein the Court concludes that nine Individual Plaintiffs lack sufficient standing. 231 district, PX302. Dr. Miller testified that while such packing does assure him a Democratic representative in House District 18, “it does so at the expense of multiple safe districts for Republicans along the . . . neighboring districts,” Tr. 205:9-19, making it more likely that the Republicans would gain control of the General Assembly. 531. The other Individual Plaintiff who testified at trial, Joshua Brown, is a locksmith apprentice from High Point who resides in Senate District 26. Tr. 830:7-12. As shown in Plaintiffs’ Exhibit 281, the General Assembly split off the most heavily Democratic area of Guilford County where Mr. Brown lives and appended it to conservative Randolph County: 532. Mr. Brown testified that by drawing his Senate District in this manner, the General Assembly “clearly dilute[d] the ability of Democrats to even attempt to run a fair race.” Tr. 833:19-21. Like Dr. Miller, Mr. Brown cares about a number of issues before the General Assembly, including a living wage, the environment, and Medicaid expansion. Tr. 834:5-6. Mr. Brown’s mother was recently hospitalized, and he believes that she would not be facing certain health issues if North Carolina had approved the Medicaid expansion. Tr. 232 834:15-835:3. He believes that the Republican Party in the General Assembly today has “opposing” stances on these issues that he cares about. Tr. 835:4-7. 533. Mr. Brown also lives in House District 60, where Democrats such as Mr. Brown are packed to create an overwhelmingly Democratic district. See Tr. 833:25-834:2; PX310. Mr. Brown testified that by packing Democrats in this manner, the General Assembly “reduced the odds of surrounding districts electing a Democrat,” Tr. 833:25-834:2, making it more difficult for Democrats to gain control of the General Assembly. 534. The affidavits submitted by the remaining thirty-five Individual Plaintiffs establish that each of these Individual Plaintiffs (i) has voted for the Democratic candidate running for the North Carolina General Assembly in each year that such an election was held since at least 2011, (ii) has a preference for electing Democratic legislators and a majority-Democratic General Assembly, and (iii) believes that if the Democratic Party made up a majority of the members in the General Assembly, the policies proposed and enacted would more closely represent the Plaintiff’s personal and political views. PX678-713. 535. Plaintiffs’ expert Dr. Chen quantified the effects of the gerrymander on the partisan composition of the districts in which each Individual Plaintiffs resides. For each of his 4,000 simulations (2,000 in the House and 2,000 in the Senate), Dr. Chen determined the House or Senate district in which each Individual Plaintiff would live based on that Plaintiff’s residential address. Tr. 387:14-388:6; PX1 at 167-68 (Chen Report). Dr. Chen then compared the Democratic vote share of the districts in which a particular Plaintiff would live under his simulations to the Democratic vote share of the Plaintiff’s districts under the enacted plans. Id. 536. Plaintiffs’ Exhibit 238 (reproduced below) shows Dr. Chen’s results for his House Simulation Set 1. In each row, the red star represents the Democratic vote share in the Individual Plaintiff’s House district under the enacted plan using the ten 2010-2016 233 statewide elections, while the gray circles represent the Democratic vote share of that Plaintiff’s district under each of the 1,000 simulated plans in House Simulation Set 1. Tr. 388:14-389:12. For instance, the figure shows that Rebecca Johnson’s House district in the enacted plan has a roughly 40% Democratic vote share using the 2010-2016 statewide elections, but Ms. Johnson would live in a House district with a higher Democratic vote share in 99% of the simulations, with most of the simulations putting her in a district with an over 50% Democratic vote share. Tr. 390:6-391:20. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 234 537. Dr. Chen found that the following Plaintiffs live in House districts that are extreme partisan outliers compared to their districts in House Simulation Set 1: Vinod 235 Thomas, Paula Ann Chapman, Kristin Parker, Julie Ann Frey, Jackson Thomas Dunn Jr., Rebecca Johnson, Lily Nicole Quick, Joshua Perry Brown, Dwight Jordan, David Dwight Brown, Electa E. Person, Donald Allan Rumph, Amy Claire Oseroff, Lesley Brook Wischmann, Derrick Miller, Carlton E. Campbell Sr., Rosalyn Sloan, Mark S. Peters, Joseph Thomas Gates, Stephen Douglas McGrigor, and Rebecca Harper. Tr. 393:9-17. Dr. Chen further found that Plaintiff Leon Schaller lives in a district that is a 68.1% outlier in House Simulation Set 1, but a 100% outlier in House Simulation Set 2. Tr. 394:2-10; see PX239. 538. Plaintiffs’ Exhibit 117 shows the same analysis for the Senate, comparing the Democratic vote share in certain Individual Plaintiffs’ districts under the enacted Senate plan to their districts under Dr. Chen’s Senate Simulation Set 1. [THE REMAINDER OF THIS PAGE INTENTIONALLY LEFT BLANK] 236 539. Dr. Chen found that the following Plaintiffs live in Senate districts that are outliers or extreme partisan outliers compared to their districts in his Senate simulations: Vinod Thomas, Paula Anna Chapman, Pamela Morton, Kristin Parker, Jackson Tomas 237 Dunn, Jr., Rebecca Johnson, Dwight Jordan, David Dwight Brown, Karen Sue Holbrook, James Mackin Nesbit, George David Gauck, Derrick Miller, Mark S. Peters, Joseph Thomas Gates, William Service, Stephen Douglas McGrigor, Rebecca Harper, Nancy Bradley, Aaron Wolff, and Kathleen Barnes. Tr. 395:7-22. Dr. Chen found that the same Plaintiffs lived in districts that are outliers under his Senate Simulation Set 2. Tr. 396:1-7; PX118. 540. Plaintiffs’ expert Dr. Cooper further demonstrated how the 2017 Plans, as a whole, disadvantage the Individual Plaintiffs. As Dr. Cooper explained, under the 2017 Plans, Democrats cannot translate their votes into seats as efficiently as Republicans. Tr. 870:11-14. 541. One of Legislative Defendants’ experts, Dr. Brunell, also testified about the ways in which partisan gerrymandering harms individual voters. Dr. Brunell testified that “the responsiveness of a legislator to the voters in the voter’s district is critical to democratic representation.” Tr. 23531:3-6. He testified that a change in the party representing a given district generates “a huge difference” in the policies for which the representative will vote. Tr. 2354:20-23. He also testified that partisan gerrymandering is a problem in modern redistricting because it “can distort how voter preferences get translated into public policy.” Tr. 2355:7-9. F. Defendants Offered No Meaningful Defense of the 2017 Plans 1. 542. No Witness Denied That the Plans Are Intentional and Effective Partisan Gerrymanders Defendants did not persuasively rebut Plaintiffs’ extensive direct evidence that the 2017 Plans were drawn with the predominant purpose of maximizing Republican advantage. 238 543. Defendants presented unpersuasive evidence to rebut evidence that the Hofeller files show that Dr. Hofeller primarily focused on maximizing partisan advantage. Defendants did not identify any file showing that Dr. Hofeller was motivated by anything other than partisanship in drawing the enacted House and Senate plans. Defendants identified no file, for example, showing that Dr. Hofeller at any point during the 2011 and 2017 redistricting processes considered “communities of interest,” cf. Tr. 1059:3-1060:5, or sought to preserve the “cores” of existing districts, cf. Tr. 1212:20-24, or drew or altered any district to avoid splitting a municipality or VTD or to make the district more compact, or constructed any district as a “product of the nuance of legislative negotiation,” cf. Tr. 1204:2-1206:4. 544. Defendants’ experts did not persuasively contest that the plans sought to ensure Republican control of the legislature. Defendants’ experts offered no methodology to attempt to evaluate whether the enacted plans were (or were not) extreme partisan gerrymanders. None offered an opinion on that question. Rather, as explained below, Defendants’ experts offered theories of why the analyses by Plaintiffs’ experts was somehow incomplete or unreliable. The Court gives little weight to these criticisms. 2. Defendants’ Criticisms of Plaintiffs’ Experts Were Not Persuasive a. 545. Dr. Thornton Legislative Defendants offered expert testimony from Dr. Janet Thornton to criticize the analyses and conclusions of Plaintiffs’ simulation experts, Drs. Chen, Mattingly, and Pegden. Tr. 1618:10-13; LDTX 286 at 4 (Thornton report). Dr. Thornton offered three main critiques of Plaintiffs’ experts: (a) Dr. Pegden’s and Dr. Mattingly’s conclusions supposedly were skewed by the particular statewide elections they used to measure the partisan lean of their simulated plans versus the enacted plans, LDTX 286 at 239 6-10; (b) their simulations purportedly deviated in various ways from the 2017 Adopted Criteria, id. at 10-19; and (c) their simulations supposedly are not statistically significantly different from the enacted plans in terms of the number of Democratic-leaning districts, id. at 20-29. See Tr. 1622:5-1623:11. But Dr. Thornton’s testimony was not persuasive, her analysis is unreliable, and her opinions are given little weight. 546. Dr. Thornton has a masters and a doctorate in economics from Florida State University. Tr. 1571:6-11. She has a bachelor’s degree in economic and political science from the University of Central Florida. Id. 547. Dr. Thornton is currently a managing director at Berkeley Research Group and has worked as an economist and applied statistician for 35 years. Tr. 1571:15-1572:3. Dr. Thornton has prepared statistical analysis in voting cases, limited, however, to analysis of statistical differences in voter participation rates by race and minority status. Tr. 1574:321. 548. Dr. Thornton has taught statistics and quantitative methods for the business school at Florida State University. Tr. 1573:12-15; LDTX 286 at 39. 549. Dr. Thornton is a member of the American Economic Association and the National Association of Forensic Economists. She has published in peer-reviewed publications including the Journal of Forensic Economics and the Journal of Legal Economics. Tr. 1573:16-1574:2. 550. Dr. Thornton was accepted by the Court as an expert in the fields of economic and applied statistical analysis. Tr. 1578:7-17. She has been qualified as an expert in other cases regarding these subjects. Tr. 1576:12-1577:13. Dr. Thornton has never been excluded from testifying. Id. 551. Dr. Thornton has no academic experience involving gerrymandering and instead specializes in expert witness testimony and other consulting-type work in various 240 areas, including employment, insurance, and credit decisions. Tr. 1619:19-1620:20, 1621:217; LDTX 286 at App’x A (Thornton CV). Dr. Thornton has no degree in mathematics, no degree in statistics, and only an undergraduate degree in political science. Tr. 1620:211621:1. She purported to critique the work of Plaintiffs’ simulations experts, each of whom is a full-time academic with years of academic experience in using computer simulations to evaluate partisan gerrymandering. Tr. 1618:14-1619:18. 552. In her report and testimony in this case, Dr. Thornton offered no methodology for determining whether a particular redistricting plan is or is not a partisan gerrymander, or whether a particular plan is or is not the product of extreme partisan considerations. Tr. 1621:18-25. Nor did Dr. Thornton offer any opinion as to whether the enacted plans were drawn as partisan gerrymanders to benefit Republicans. When asked whether she was offering such an opinion, Dr. Thornton responded, “I have no way of knowing.” Tr. 1622:1-4. (i) 553. Criticisms Concerning Choice of Statewide Elections Dr. Thornton’s criticisms of the specific statewide elections used by Drs. Pegden and Mattingly suffered from critical flaws. 554. Dr. Thornton stated in her report that Dr. Pegden “considered” only “two elections” in his analysis. LDTX 286 at 10; see id. 8-11; Tr. 1626:9-16. However, Dr. Pegden used six prior election results—two discussed in the body of his report, and four more summarized in an appendix. PX508 at 11, 34-37 (Pegden Report). Dr. Thornton corrected this mistake only after Dr. Pegden’s rebuttal report pointed it out and she was confronted with it at deposition. Tr. 1627:22-1628:4. At trial, Dr. Thornton presented a revised version of a table from her report, in which she (without acknowledging the change during her direct testimony) had added asterisks showing that Dr. Pegden in fact used six prior elections. Tr. 1626:17-1627:3; compare LDTX 286 at 7 (tbl. 1) with LDTX 302 (Thornton 241 Demonstrative 1). Dr. Thornton’s apparent oversight of the number of elections used in Dr. Pegden’s analysis led to her to conclude that “Dr. Pegden’s choice of elections influence[d] his conclusions.” Tr. 1604:21-1605:7; see Tr. 1591:20-1592:10 (presenting LDTX 91, a chart purported to show the average Democratic vote share of the elections “included by each expert,” but using just the 2016 Attorney General and 2008 Commissioner of Insurance for Dr. Pegden). 555. On cross examination, Dr. Thornton did not dispute that, when Dr. Pegden tested his results using the four additional elections summarized in his appendix, he found that it did not change his results. Tr. 1628:17-1629:4. Dr. Thornton did not test Dr. Pegden’s results using other prior elections. Tr. 1629:7-25. 556. Dr. Thornton criticized Dr. Mattingly for using a different and broader set of statewide elections than the 10 elections identified by Representative Lewis, and she specifically criticized Dr. Mattingly’s use of several 2008 elections. Tr. 1686:10-22; LDTX 286 at 8. However, Dr. Hofeller likewise used 2008 elections—including many of the same ones as Dr. Mattingly—in the partisanship formula Dr. Hofeller used to draw the 2017 Plans. Compare PX153 (Hofeller partisanship formula) with PX359 at 4 (Mattingly Report). When asked whether she knew this fact, Dr. Thornton responded that she “do[es]n’t know one way or the other,” is “not aware of anything regarding Dr. Hofeller,” and did not investigate what elections the mapmaker himself used in drawing the 2017 Plans. Tr. 1686:23-1689:5. 557. In any event, Dr. Thornton’s critique of Dr. Mattingly’s use of election results, and her analysis of various “averages” across the different elections he used, misses the point of his analysis. Dr. Mattingly analyzed, on an election-by-election basis, how the partisan bias of the enacted plan relative to the ensemble varies in different electoral environments. 242 (ii) 558. Criticisms Concerning Use of the Adopted Criteria Dr. Thornton’s assertion that Plaintiffs’ simulation experts deviated from the Adopted Criteria also suffers from critical flaws. Additionally, Dr. Thornton failed to show that any of her criticisms would have made any difference to Plaintiffs’ experts’ conclusions. 559. Dr. Thornton stated in her report that “[a] review of Dr. Pegden’s simulation code suggests that in reality, he did not actually apply a compactness criterion.” LDTX 286 at 33. However, Dr. Pegden did apply a compactness criterion. PX508 at 8, 34 (Pegden Report); Tr. 1358:11-24 (Dr. Pegden). As Dr. Pegden explained in his rebuttal report, if he had not applied a compactness criterion, his simulated plans would have looked completely different—dramatically less compact. PX551 at 17-19 (Pegden Rebuttal Report); Tr. 1358:25-1360:1 (Dr. Pegden). When asked about this mistake on cross examination, Dr. Thornton testified that “in retrospect” she “should have written it in a different way.” Tr. 1623:12-25. 560. While Dr. Thornton criticized Dr. Pegden for not specifically applying a Reock compactness threshold, she did no work to assess whether adding such a threshold would change Dr. Pegden’s simulations or results. Tr. 1624:23-1626:3. Nor did she do any work to test whether adding a Reock threshold would change Dr. Pegden’s conclusion that the enacted plans are extreme outliers carefully crafted to favor Republicans. Tr. 1626:4-8. The Adopted Criteria state that the 2017 Plans should “improve the compactness” over the 2011 Plans, and when asked whether Dr. Pegden’s simulated plans “are, in fact, an improvement in terms of compactness over the districting in the 2011 map,” Dr. Thornton responded, “I don’t know.” Tr. 1625:13-18. Dr. Thornton did no work to figure it out. Tr. 1625:19-1626:3. 561. Dr. Thornton testified that Dr. Pegden did not “make any adjustment for incumbency.” Tr. 1604:8-9. This is incorrect. Dr. Pegden included as a criterion in all of his 243 simulations avoiding pairing the incumbents who were in office at the time the districts were drawn. PX508 at 8 (listing “Incumbency protection” as criterion). 562. Dr. Thornton also suggested that Dr. Pegden could not draw valid conclusions about the 2017 Plans without reaching “equilibrium” in his Markov Chain—without comparing the 2017 Plans to the entire universe of potential House and Senate districtings. Tr. 1631:2-11. In this regard, Dr. Thornton analogized Dr. Pegden’s analysis to looking for a lost key in a bedroom without considering that the key might be somewhere else in the house. But as Dr. Pegden explained, the purpose of his approach and the accompanying mathematical theorems he has proved is that they allow for drawing statistically significant conclusions about how the enacted plans compare to the universe of all possible plans meeting the relevant criteria without achieving “equilibrium,” i.e., without needing to generate a representative sample of the universe of possible maps. PX551 at 2 (Pegden Rebuttal Report); Tr. 1360:2-1361:21. Dr. Thornton acknowledged that she has no expertise in proving mathematical theorems, nor did she offer any opinion that Dr. Pegden’s theorems are wrong. Tr. 1631:12-1632:9. 563. Dr. Thornton stated in her report that Dr. Mattingly “did not consider incumbency protection as defined in the 2017 enacted map criteria.” LDTX 286 at 19. Dr. Thornton repeated this assertion in her direct testimony, stating that Dr. Mattingly did not “control, in any respect, for incumbency protection.” Tr. 1610:20-22. This is false. Dr. Mattingly added incumbency protection as a criterion in checking the robustness of his results, and he concluded that it did not change his results. PX359 at 81-85; Tr. 1093:151094:4. 564. On cross examination, Dr. Thornton said that Dr. Mattingly may not have considered incumbency protection “simultaneously” “[w]ith respect to all the other factors, as I recall.” Tr. 1633:14-24. This too is incorrect. Dr. Mattingly added incumbency 244 protection as a criterion in conjunction with the criteria used to generate his primary ensemble, and he ran a separate analysis that “consider[ed] the joint effect of both ensuring incumbents are preserved and requiring more stringent redistricting criteria” with respect to the traditional districting criteria. PX359 at 81-82. 565. Dr. Thornton criticized Dr. Mattingly for using only Polsby-Popper compactness scores, and not Reock scores. Tr. 1633:25-1634:3. But she did no work to determine whether the Reock scores for his simulated plans were too low, or whether applying a Reock threshold would change his results. Tr. 1634:4-21. In his rebuttal report, Dr. Mattingly calculated Reock scores for all of his simulated districts, and he reported that there was not a single district in any of his simulated Senate plans with a Reock score less than or equal to 0.15—the threshold referenced in the Adopted Criteria. PX487 at 8-9. There were very few such districts in his simulated House plans—only 1 out of 550,000 simulated Wake districts, and 7 out of 486,588 Mecklenburg districts. PX487 at 8; Tr. 1634:22-161635:14. Dr. Mattingly concluded that removing those districts would not change his results, id., and Dr. Thornton did no work of her own to determine whether he was wrong, Tr. 1635:15-25. 566. Dr. Thornton criticized Dr. Pegden’s and Dr. Mattingly’s weighting of the various criteria they applied to create their simulated plans. LDTX 286 at 17-18; Tr. 1636:13-24. But Dr. Thornton acknowledged that she did not know whether the legislature “did weighting” at all, or how it may have done so. Tr. 1636:25-1637:13. She did not suggest any better way than Dr. Mattingly’s approach to weighting the various criteria. Tr. 1637:14-25. She did not rerun Dr. Mattingly’s computer code using any different weighting system to determine if using a different weighting system could have affected Dr. Mattingly’s conclusions. Tr. 1638:1-6. In his rebuttal report, Dr. Mattingly tried six different ways of weighting the various criteria, and he concluded that none changed his 245 results. PX487 at 10-11. When asked about this analysis on cross examination, Dr. Thornton merely said, “I don’t recall.” Tr. 1638:7-14. 567. Dr. Thornton testified that Dr. Chen’s use of a “T score” meant that his simulations did not follow the Adopted Criteria regarding compactness, avoiding splitting municipalities, and avoiding splitting VTDs. Tr. 1599:18-1600:3. Dr. Thornton suggested that Dr. Chen restricted his algorithm to only accept plans below a particular T Score, Tr. 1597:25-1598:19, and she asserted in her report that “[a] t-score evaluation was not among the actual criteria” in the Adopted Criteria, LDTX286 at 15. Dr. Thornton testified that, if Dr. Chen “changed the value of the T scores,” used a “value other than 1.75” in the T score, or “added a random element,” his results would have been entirely different. Tr. 1597:251598:19. 568. Dr. Thornton’s testimony misapprehends Dr. Chen’s algorithm. Dr. Chen’s “T score” does not impose a numerical threshold that restricts the maps the algorithm generates. Rather, the T score is just a way of equally weighting and jointly tracking the three traditional districting criteria of compactness, avoiding municipal splits, and avoiding VTD splits. For any given county grouping, the algorithm randomly draws an initial set of districts, and then proposes a random change to the border between a random pair of adjoining districts. Tr. 261:23-262:16. If the border change would, on net, improve the districting of the grouping across the three criteria of compactness, avoiding municipal splits, and avoiding VTD splits, the algorithm accepts the change. Id. But if the change would make the districting worse off, on net, with respect to these criteria, the algorithm rejects the change. Id. The T score is merely a way of giving the three criteria equal weight and then tracking whether a proposed random change improves the districting across these criteria. Tr. 263:4-8 The algorithm considers thousands of these random changes, one at a time in an iterative fashion, in drawing districts within a given grouping. Tr. 261:18-262:23. 246 569. Dr. Thornton is thus incorrect that Dr. Chen’s algorithm lacks a “random element.” Tr. 1598:7-8. She misapprehends the T score’s function in suggesting that raising or lowering the “T score value” would be less “restrictive.” Tr. 1598:5-10. The T score’s sole purpose is to equally weight the three criteria of compactness, avoiding split municipalities, and avoiding split VTDs. Dr. Thornton does not dispute that Dr. Chen’s T score accurately gives equal weight to these three criteria. 570. Moreover, while Dr. Thornton asserted that Dr. Chen may not have found the enacted plans to be statistical outliers if he had used “different T scores,” Tr. 1598:201599:13, Dr. Thornton offered no proof or analysis to substantiate this claim, Tr. 1645:141647:15. 571. Dr. Thornton also criticized Dr. Chen’s approach to incumbency protection in his Simulation Set 2. Tr. 1638:15-1639:8. She acknowledged that Dr. Chen’s Simulation Set 2 successfully avoided pairing incumbents, but she asserted that Dr. Chen failed to comply with the second sentence of the Adopted Criteria’s incumbency protection criterion, which provided that “the committees may make reasonable efforts to ensure voters have a reasonable opportunity to elect non-paired incumbents.” Tr. 1610:23-1611:3. Dr. Thornton claimed that this sentence meant the Committees should make efforts “to allow for incumbents to win” by placing them in politically favorable districts, LDTX286 at 16, and that “it would have been interesting” if Dr. Chen had applied “some sort of weighting” to carry this out, Tr. 1639:12-1640:3. Dr. Thornton’s interpretation is contrary to the contemporaneous explanation of this sentence by Representative Lewis, who stated at an August 10, 2017 hearing that the sentence “is simply saying that mapmakers may take reasonable efforts to not pair incumbents unduly.” PX603 at 122:4-18; Tr. 1640:16-1641:12. That direction matches Dr. Chen’s approach to incumbency protection. 247 572. Dr. Thornton did not analyze whether any of the supposed deviations made any difference to the experts’ conclusions. On cross examination, Dr. Thornton was asked whether, “for every single criticism you’ve leveled, there’s no instance in which you took any of plaintiffs’ experts’ code, substituted whatever you thought was an improved criteria, ran the code with the improved criteria and showed us that it made a difference to their work; isn’t it true in your report there’s no place that you did that?” Tr. 1647:3-13. Dr. Thornton responded that, “given the time, [she] did not have sufficient time to do so.” Tr. 1647:14-15. (iii) 573. Criticisms Concerning Statistical Significance Dr. Thornton opined that the enacted plans are “not statistically significantly different from the simulated maps with respect to the number of Democratic districts.” LDTX286 at 21 (capitalization omitted). Dr. Thornton wrote in her report that she compared “the enacted plan’s number of Democratic districts and the number predicted by the simulated maps,” and “determined the number of standard deviations associated with the difference between the enacted plan and simulated number of Democratic districts.” LDTX286 at 24. However, Dr. Thornton did not use the actual results of Plaintiffs’ experts’ “simulated plans,” or the actual “standard deviation” of the simulated plans. 574. Instead, Dr. Thornton created her own distribution of the predicted number of Democratic seats won under a nonpartisan plan, using a “binomial distribution.” She then calculated the “standard deviation” of her own distribution, and used that standard deviation to assess statistical significance. See PX551 at 10 (Pegden Rebuttal Report). Dr. Thornton used this binomial distribution methodology as the foundation for her criticisms of all three of Plaintiffs’ simulation experts. LDTX286 at 22; Tr. 1685:9-22. 575. Contrary to Dr. Thornton’s approach, the distribution of districting maps is not a binomial distribution, and thus it is inappropriate to use a binomial distribution in the redistricting context. When confronted with the flaws in using a binomial distribution 248 in the redistricting context, Dr. Thornton’s responses were not persuasive. The Court gives her testimony concerning statistical significance little weight. 576. It is undisputed that a binomial distribution applies only when two conditions are met: (1) each trial (in this case, each House or Senate district) is independent of one another; (2) each trial has the exact same percentage chance of producing a particular outcome (in this case, that a Democrat wins the district). Tr. 1669:4-8, 1676:1-5 (Dr. Thornton); Tr. 1378:24-1382:2 (Dr. Pegden); PX551 at 10 (Pegden Rebuttal Report); PX487 at 11-12 (Mattingly Rebuttal Report); PX123 at 171-72 (Chen Rebuttal Report). Thus, the classic example of the binomial distribution is a coin flip, because the likelihood of landing on heads on any flip of a coin is independent of the result of every other flip, and the percent chance of landing on heads is the same in each flip (50%). Tr. 1669:11-1670:5. 577. By applying a binomial-distribution methodology, Dr. Thornton assumed that district elections, like coin flips, are independent of each other, and also that Democrats have the same chance—specifically, a roughly 40% chance—of winning each and every district House or Senate district, no matter where in North Carolina the district is located. Tr. 1670:6-1671:2 (Dr. Thornton); see Tr. 1381:15-1382:2 (Dr. Pegden); PX551 at 10 (Pegden Rebuttal Report); PX487 at 11-12 (Mattingly Rebuttal Report); PX123 at 171-72 (Chen Rebuttal Report). 578. Both assumptions are incorrect in the redistricting context. First, unlike a coin flip, each House (or Senate) district is not independent of one another. Tr. 1379:221381:10 (Dr. Pegden); PX551 at 10 (Pegden Rebuttal Report). In a given county grouping, if a particular set of Democratic voters is placed in one district, then those voters cannot be put in any other district in the grouping. Id. The partisan makeup of the districts are thus intertwined and not independent of one another; increasing the number of Democratic 249 voters in a particular district necessarily decreases the number of Democratic voters in neighboring districts. Id. 579. The second assumption underlying Dr. Thornton’s binomial distribution— that Democrats have the exact same percentage chance of winning each House (or Senate) seat—is contrary to reality. Dr. Thornton assumes, for example, that Democrats have the same percentage chance of winning a House district in Wake County as in Caldwell County. Tr. 1381:15-1382:2 (Dr. Pegden); see PX487 at 11-12 (Mattingly Rebuttal Report); see PX123 at 171-72 (Chen Rebuttal Report). This is not the case. 580. The following example illustrates these flaws in Dr. Thornton’s analysis. In the Alamance County House grouping, there are two districts of roughly equal population. Assuming, as a hypothetical, that Republicans will win 60% of the total vote across the County in a particular election, it is mathematically impossible for Democrats to win both districts in the election. Tr. 1673:14-19. But under Dr. Thornton’s binomial-distribution methodology, Democrats will win both districts 16% of the time—because she assumes that Democrats have an equal and independent 40% of winning each of the two districts. Tr. 1671:10-17; see also Tr. 1379:1-1381:10 (Dr. Pegden). When asked about this on cross examination, Dr. Thornton repeatedly asserted that she did not “understand” the illustration. Tr. 1671:3-1673:13. 581. Dr. Thornton’s binomial-distribution methodology was recently rejected by a federal court in a partisan gerrymandering case in Ohio. There, as here, Dr. Thornton used a binomial distribution in her expert analysis on behalf of the Republican legislative defendants, and the three-judge federal district court rejected her analysis. The court stated: “Dr. Thornton also performed her own analysis using a binomial distribution, but we do not give any weight to that analysis.” Ohio A. Philip Randolph Inst. v. Householder, 373 F. Supp. 3d 978, 1056 (S.D. Ohio 2019); see Tr. 1673:20-1674:20. The court explained 250 that Dr. Thornton’s binomial-distribution analysis “incorporates yet another faulty assumption that each district has a 51% chance of being won by a Republican because Republicans won 51% of the congressional vote across the State; this assumption does not comport with basic understandings of congressional elections, i.e., that although some districts may be competitive (a 51% Republican to 49% Democrat district), other districts lean heavily in favor of one party or the other.” Ohio A. Philip Randolph Inst., 373 F. Supp. 3d at 1056; see Tr. 1677:23-1678:15. 582. While Dr. Thornton claimed that her use of a binomial distribution here is different from the Ohio case, Tr. 1677:19-22, the Court disagrees and finds that Dr. Thornton’s methodology here suffers from the same flaws identified by the federal court in the Ohio case. Assuming that districts are independent, and that Democrats have a roughly 40% chance of winning every House and Senate district, does not comport with basic understandings and reality of North Carolina House and Senate elections. Dr. Thornton could not identify literature or precedent supporting the use of a binomial distribution in a redistricting context. Tr. 1680:6-14. 583. Dr. Thornton’s use of a binomial distribution skewed her statistical significance analysis. Due to the independence and equal probability assumptions, the binomial produces a much wider distribution of the number of possible districts Democrats could win in the House or the Senate than the actual distribution produced by each expert’s simulations. That wider distribution in turn results in Dr. Thornton estimating much larger standard deviations than the actual standard deviations of each expert’s simulated plans, allowing Dr. Thornton to claim that the enacted plan is less than two standard deviations from each expert’s average simulation and therefore purportedly not a statistically significant outlier. LDTX286 at 9-13. For instance, in Dr. Chen’s House Simulation Set 1, his simulated maps produce a range of results from 43 Democratic 251 districts to 51 Democratic districts, with 90 percent of those results between 45 and 48 Democratic districts, whereas the enacted 2017 House plan produces only 42 Democratic districts—an extreme outlier, completely off the distribution. PX234; Tr. 1647:16-1648:16. The actual standard deviation of Dr. Chen’s House Simulation Set 1 is 1.36 seats, and the enacted plan is more than three standard deviations from the average simulated plan. Id. But Dr. Thornton’s unsubstantiated binomial distribution suggests that Democrats could win as few as 30 districts and as many as 63, and has a standard deviation of 5.34 seats. PX123 at 170-76. 584. Similarly, Dr. Thornton’s binomial distribution is completely different from the actual distribution of simulated plans she created using a modification of Dr. Pegden’s computer code. For the House, while the simulations generated between 46 and 50 Democratic seats, Dr. Thornton’s binomial distribution generated between 35 and 60 Democratic seats and a much larger standard deviation. Plaintiffs’ Exhibit 554, a figure from Dr. Pegden’s rebuttal report, depicts these dramatic differences: 252 585. Dr. Thornton’s binomial distribution likewise is completely different from the actual distribution of simulated plans created by Dr. Mattingly. PX495. When Dr. Mattingly used the “actual distribution” of his results to calculate statistical significance as opposed to Dr. Thornton’s “grossly inaccurate seat distribution,” he found that the enacted maps are “well outside two or three standard deviations” and are “extreme outliers.” PX487 at 11-12. 586. Dr. Thornton made other significant methodological errors in her analysis of statistical significance. For instance, in modifying Dr. Pegden’s computer code to generate simulated plans of her own, Dr. Thornton used the wrong command and froze every single district drawn in 2011 and left unchanged in 2017. Tr. 1363:7-1364:8 (Dr. Pegden); PX551 at 6 (Pegden Rebuttal Report). Dr. Thornton’s suggestion that she intended to freeze the 2011 districts, Tr. 1666:16-21, is not credible, given that her report nowhere mentions this decision and in fact claims that it is analyzing the entire enacted map—all 120 House districts and all 50 Senate districts. LDTX286 at 75 (tbl. 3). 587. Dr. Thornton’s freezing errors ran in both directions. In her report, Dr. Thornton presented a graph purporting to show differences in Democratic vote share between the enacted plans’ districts and the districts she drew using her modified version of Dr. Pegden’s code. The evident goal of these charts—titled “Comparison of the Enacted Plan and the Average Across Dr. Pegden’s Simulations for Each Non-Frozen House [and Senate] District”—was to suggest that the vote shares in the enacted districts were not markedly different from those in the nonpartisan simulations. LDTX286 at 28-29 (emphasis added). But Dr. Thornton’s charts included many districts that were frozen on account of the Whole County Provision, which misleadingly suggested a high degree of similarity between the enacted plan and the simulations. Tr. 1680:24-1684:9. Dr. Pegden pointed out a number of other problems with this chart—e.g., using thick lines, stretching the data out 253 over an unnecessarily long vertical axis, and needlessly connecting the data points using lines, all which served to obscure the significant gaps in vote share between the enacted and simulated districts. Tr. 1391:6-1395:19. 588. Setting aside the flaws in her analysis, Dr. Thornton’s results show a statistically significant difference between the enacted 2017 Plans and the simulated plans she created using a modification of Dr. Pegden’s code. As shown in Dr. Pegden’s rebuttal report, only 0.001% of Dr. Thornton’s simulated plans are as Republican-favorable as the enacted House plan, and only 0.182% of Dr. Thornton’s simulated plans are as Republicanfavorable as the enacted Senate plan. PX551 at 8-9 (Pegden Rebuttal Report); Tr. 1369:41371:18. 589. Thus, even including errors, Dr. Thornton’s results were still consistent with the conclusions of Plaintiffs’ experts. Tr. 1400:10-21 (Dr. Pegden). b. 590. Dr. Brunell Legislative Defendants offered expert testimony from Dr. Thomas Brunell, who was asked to read and respond to the reports of Drs. Pegden, Cooper, Mattingly and Chen. Tr. 2276:19-20. Dr. Brunell is a tenured political science professor at the University of Texas, Dallas. For over 20 years, Dr. Brunell has taught, lectured and published on representational and redistricting issues. LDTX292. Dr. Brunell was accepted by the Court as an expert on redistricting and political science. Tr. 2275:4-12. Dr. Brunell offered no opinion on whether the 2017 Plans are partisan gerrymanders. Tr. 2316:10-12. 591. The Court finds Dr. Brunell’s opinions were unpersuasive, sometimes inconsistent with prior testimony he has given, and gives them little weight. 592. Dr. Brunell testified that Plaintiffs’ experts have not shown “what is too much politics in this political process.” Tr. 2306:24-2307:2. However, this critique contradicts Dr. Brunell’s own expert analysis and conclusions in a prior case. In 2011, Dr. 254 Brunell opined as an expert witness for the Nevada Republican Party that state legislative maps were excessive partisan gerrymanders—based on an analysis less robust than the analyses of Plaintiffs’ experts here. Tr. 2337:5-2338:23. Using two statewide elections, Dr. Brunell conducted a uniform swing analysis and concluded that the maps at issue gave Democrats 60% of the seats when Democrats won only 50% of the votes statewide. Tr. 2340:16-2345:5. Dr. Brunell concluded exclusively on the basis of that analysis that the maps were “unfair” and showed “heavy pro-Democratic bias”—“clearly a pattern of partisan bias, i.e., gerrymandering.” Tr. 2342:4-2345:11. Dr. Brunell further opined, based solely on his uniform swing analysis and the disconnect between Democrats winning 60% of the seats with only 50% of the statewide vote, that he could be “absolutely conclusive” that the maps were not just partisan gerrymanders, but a “leading candidate for gerrymander of the decade.” Tr. 2345:12-2346:15. 593. In this case, Dr. Brunell conceded that Plaintiffs’ experts’ analyses—using both uniform swing analysis and actual results of prior statewide elections—demonstrated that when Republicans get 50% of the votes in either chamber of the General Assembly, they win at least 60% of the seats. Tr. 2346:16-2350:2. Thus, under Dr. Brunell’s own approach, the Court could find, in his own words, a “heavy pro-[Republican] bias” and “clearly a pattern of partisan bias i.e., gerrymandering.” Tr. 2350:3-8. 594. The Court also rejects Dr. Brunell’s testimony that simulation methods for evaluating partisan gerrymandering have not been sufficiently vetted by academics and courts. Tr. 2292:15-2293:23. Dr. Brunell testified on direct examination that he was unaware of any peer-reviewed political science papers that provide a “basis” for “using [simulations] as an evaluation for partisanship.” Tr. 2293:11-17. He testified that a 2013 paper by Dr. Chen and Dr. Jonathan Rodden “uses simulations, I think,” “[b]ut in terms of using it as an evaluation for partisanship, I don’t think there have been any such 255 publications yet.” Tr. 2293:11-17. Dr. Brunell later acknowledged that the 2013 Chen and Rodden paper was in fact a peer-reviewed political science paper that “uses simulation techniques to measure partisanship.” Tr. 2307:19-2308:5; see PX1 at 179. He also acknowledged that he was unfamiliar with three other peer-reviewed political science papers by Dr. Chen published between 2015 and 2017 that use computer simulations to evaluate partisan gerrymandering. Tr. 2308:10-2309:9; PX1 at 180. Dr. Brunell was also unaware that Dr. Pegden’s paper on using simulations to measure gerrymandering, published in the Proceedings of the National Academy of Sciences, was peer reviewed by a political scientist. Tr. 2309:12-22; see Tr. 1413:7-16. 595. Dr. Brunell was also unfamiliar with court decisions approving the use of simulations to measure partisanship. He testified on direct that “we’ve only just started to see [simulations] used in law suits,” Tr. 2292:24-2293:1, that simulations therefore “may not be ready for prime time yet,” Tr. 2292:22-24, and that he himself did not learn about the simulation method until 2017 or 2018, Tr. 2293:7-10. However, as he acknowledged, multiple courts have credited simulations by Drs. Chen, Mattingly, and Pegden as a method of establishing whether a particular map is a partisan gerrymander. Tr. 2310:82312:1. Dr. Brunell was “unaware” that the Fourth Circuit credited Dr. Chen’s simulations in a 2016 decision, in a gerrymandering case filed in 2013. Tr. 2311:4-2312:1; see Raleigh Wake Citizens Ass’n v. Wake Cty. Bd. of Elections, 827 F.3d 333 (4th Cir. 2016). The court rejected the criticism Dr. Brunell makes here, namely that Dr. Chen’s simulations “ignor[ed] partisanship.” Tr. 2311:17-20; see Raleigh Wake, 827 F.3d at 344. 596. The Court rejects Dr. Brunell’s testimony that simulated maps are only useful if the algorithm draws “partisan districts” as opposed to “nonpartisan districts.” Tr. 2277:13-20; 2280:4-16. Dr. Brunell acknowledged that the 2017 Plans were drawn for partisan gain, but argued that simulations can tell if an enacted map is an “extreme 256 partisan outlier” only if the simulations include some level of partisanship. LDTX291 at 3; Tr. 2277:13-20; 2280:4-16. Dr. Brunell’s criticisms miss the point. Dr. Mattingly’s and Dr. Chen’s simulations quantify the effects of the gerrymandering and how extreme it is. Both find that the enacted plans are outside the entire distribution of their simulated plans— sometimes by many seats. For instance, Dr. Chen found in his uniform swing analysis that, in electoral environments corresponding to a 52.42% statewide Democratic vote share, Democrats win 11 to 12 fewer seats in the House and 3 to 4 fewer seats in the Senate than they would typically win under the simulated plans. See PX1 at 34, 65 (Chen Report). Dr. Mattingly found similar results. See PX359 at 12 (Mattingly Report); PX487 at 25 (Mattingly Rebuttal Report). 597. Additionally, Dr. Pegden’s analysis demonstrates that the 2017 Plans are extreme partisan outliers even in comparison to other partisan maps. Although Dr. Brunell criticized “all three of” Plaintiffs’ simulation experts for using “nonpartisan districts” as the point of comparison, Tr. 2277:13-20, this misunderstands Dr. Pegden’s methodology. Dr. Pegden started with the enacted plan and made a sequence of small random changes, observing how those changes affected the partisan characteristics of the plan. Tr. 1304:31305:7; PX515; PX519. Dr. Pegden’s comparison maps thus “are not supposed to be neutral comparison maps drawn from scratch of North Carolina,” and “even against a set of extremely similar maps which were generated from the enacted map and which share all sorts of qualities with the enacted map, the enacted map is still an extreme outlier.” Tr. 1304:14-1305:7. Dr. Pegden’s comparison maps are “tied strongly to the enacted map” and “baked in” intentional partisan choices by the mapmaker. Tr. 1405:1-13, 1406:2-19. This makes it all the more remarkable that the enacted plans are such outliers in his analysis, even against this very similar comparison set. Tr. 1315:22-1316:2. 257 598. The Court gives no weight to Dr. Brunell’s criticisms of uniform swing analysis. Dr. Brunell stated in his report that uniform swing analysis is “not reliable,” LDTX291 at 4, and he testified that the assumption of uniform swing analysis was “clearly wrong,” Tr. 2289:14-22. But again, when Dr. Brunell was evaluating partisan bias in the Nevada case in 2011, he testified that uniform swing analysis allowed him to be “absolutely conclusive” in finding legislative maps to be heavily biased and gerrymandered. Tr. 2351:19-2352:7. 599. Dr. Brunell’s report and testimony contained numerous statements that were erroneous and reflect a failure to understand the work of Plaintiffs’ experts. Dr. Brunell’s report asserts that Dr. Pegden “use[d] the results of just two elections for his simulations” and that “both of them have Democratic winners.” LDTX291 at 15. In fact, Dr. Pegden used six elections, two of which—2012 Lieutenant Governor and 2014 U.S. Senate—had Republican winners. PX508 at 34-37 (Pegden Report). On the stand, Dr. Brunell explained his assertion by stating that Dr. Pegden “does some quick checks with other elections in his appendix, but he only uses [] two elections for his full simulation,” that he “uses one particular metric . . . but not all of it,” and that he did not use “the four additional elections in his appendix to perform his entire statewide analysis.” Tr. 2323:1-15. In fact, Dr. Pegden re-ran his entire statewide analysis using all six elections. PX508 at 34-37 (Pegden Report). 600. Dr. Brunell wrote in his report that he was “confused” by aspects of Dr. Pegden’s analysis, Tr. 2318:19-22, that were clearly explained in Dr. Pegden’s initial report. Tr. 2318:23-2319:24. Dr. Brunell criticized Dr. Pegden for failing to explain how many changes he made to the enacted map before comparing the simulated maps to the enacted map, LDTX291 at 13, but Dr. Pegden’s report made clear that he evaluated the partisanship of the new map after every step, meaning every swap, PX508 at 5. Dr. Brunell also criticized Dr. Pegden for purportedly failing to explain terms like “fragility” and 258 “carefully crafted,” Tr. 2320:8-18, but Dr. Pegden’s report specifically defined those terms. Tr. 2321:15-2322:2. 601. In criticizing Dr. Chen’s application of the Adopted Criteria, Dr. Brunell testified that Dr. Chen’s “programmatic algorithm . . . maximizes geographic compactness,” Tr. 2295:10-16, but Dr. Brunell had not reviewed Dr. Chen’s code, Tr. 2333:23-25, and he got it wrong, Tr. 262:24-263:12. When confronted with his error at trial, Dr. Brunell testified that whether Dr. Chen maximized compactness did not matter because Dr. Chen’s “algorithm” was “different from the legislative criteria” in unspecified other ways relating to splitting VTDs. Tr. 2334:6-13. However, Dr. Brunell “didn’t know” how Dr. Chen’s algorithm “worked” with respect to other issues, Tr. 2297:9-14, and he did no work to determine whether a different weighting would have affected Dr. Chen’s conclusions, Tr. 2334:18-21. 602. Dr. Brunell’s report inaccurately criticized Dr. Mattingly and Dr. Pegden for failing to preserve incumbents, when both ran simulations that avoided pairing incumbents. LDTX291 at 3; Tr. 2326:13-25; Tr. 2329:2-5. 603. The Court rejects Dr. Brunell’s testimony that the simulated maps are not proper comparisons to the enacted map to the extent they do not preserve the “core” of an incumbent’s district. Tr. 2283:21-2284:19. Dr. Brunell acknowledged that he had “no idea if and to what extent core preservation was used” in the enacted map, Tr. 2329:21-2330:1, and no other witness testified that the 2017 Plans preserved district cores. Neither Dr. Brunell nor any other witness for Legislative Defendants analyzed whether a hypothetical effort to preserve district cores could explain the extreme partisan bias in the 2017 Plans. As Representative Lewis explained, the Adopted Criteria’s incumbency protection provision referred only to “not pair[ing] incumbents unduly”—not core preservation. PX603 at 122. As Dr. Brunell acknowledged, core preservation also can be a partisan criterion, Tr. 259 2332:12-25, and that, when, as here, the prior plan was an unlawful racial gerrymander, preserving cores might also preserve racial gerrymanders, Tr. 2333:1-12. 604. Additionally, Plaintiffs proved that a hypothetical effort to preserve the “cores” of an incumbent’s district could not explain the enacted plans’ extreme partisan bias. Dr. Pegden’s simulations preserved the “cores” of each incumbent’s prior district. Tr. 1316:24-1317:10 (Dr. Pegden); see Tr. 2330:16-19. 605. The Court gives little weight to Dr. Brunell’s testimony that Figure 8 and Figure 20 of Dr. Chen’s report do not show that the enacted plan is an “outlier.” Tr. 2302:12-2303:15. Figure 8 of Dr. Chen’s report shows at least a five-seat difference between the bulk of his House simulations and the enacted plan, and shows that the enacted plan is off the distribution entirely—it elects fewer Democrats than 100% of his simulated plans. PX1 at 48 (Chen Report). The Court rejects Dr. Brunell’s testimony that a five-seat difference is only a “slight[]” difference. Tr. 2302:24-2303:2. Likewise, Figure 20 of Dr. Chen’s report shows a two-seat difference between the typical result of his Senate simulations and the enacted plan, and again shows that the enacted plan is off the distribution entirely—it elects fewer Democrats than 100% of his simulated plans. PX1 at 48 (Chen Report). Dr. Brunell also speculated that changing Dr. Chen’s criteria “could shift this over and then this wouldn’t be an outlier at all,” Tr. 2303:4-9, but the Court gives no weight to Dr. Brunell’s untested conjecture. The Court likewise rejects Dr. Brunell’s testimony about Plaintiffs’ Exhibit 48, which is Figure 28 of Dr. Chen’s report and shows cracking and packing in the Cumberland House grouping. PX1 at 93. Dr. Brunell testified that this figure did not show the enacted plan to be an “outlier” because “the enacted districts are in the gray clouds,” Tr. 2303:16-21, but in fact the figure demonstrates that two districts (HD-45 and HD-43) are entirely outside the “gray clouds” and show more 260 cracking (HD-45) and packing (HD-43) of Democrats that 100% of the districts in Dr. Chen’s simulations. PX1 at 93. c. 606. Dr. Hood Legislative Defendants offered the testimony of Dr. M.V. (Trey) Hood III to respond to Plaintiffs’ experts Dr. Cooper and Dr. Chen. LDTX 284; Tr. 2037:21-2038:3. 607. Dr. Hood is a tenured professor of political science at the University of Georgia, a position he has held for 20 years. Tr. 2032:19-2033:5. He holds three degrees in political science: a Ph.D. from Texas Tech University; a Master of Arts degree from Baylor University, and a Bachelor of Science degree from Texas A&M University. Tr. 2032:14-18. 608. Dr. Hood is also the director of the School of Public and International Affairs’ Survey Research Center which performs public opinion research and polling for entities including the Atlanta Journal-Constitution. Tr. 2033:6-19. 609. Dr. Hood teaches courses in American politics and policy, Southern politics, research methods and election administration, including redistricting. Tr. 2033:20-2034:9. 610. Dr. Hood also conducts research on redistricting and has published articles in peer-reviewed journals on topics that include redistricting. Tr. 2034:10-18. Dr. Hood’s work has appeared in peer-reviewed journals approximately 50 times. Tr. 2034:13-21. He currently serves on the editorial boards of Social Science Quarterly and Election Law Journal, with the latter journal dealing with issues of election administration, including redistricting. Tr. 2034:22-2035:2. 611. Dr. Hood was accepted by the Court as an expert in American politics and policy, Southern politics, quantitative political analysis, and election administration, including redistricting. Tr. 2037:13-20. 261 612. Dr. Hood testified about the role of the Whole County Provision and 2017 Adopted Criteria in limiting the mapmaker’s discretion in drawing the 2017 Plans, the results of the 2018 elections, and North Carolina’s political geography. 613. Dr. Hood’s testimony was not persuasive, and the Court gives it little weight. 614. Dr. Hood’s expert testimony has been rejected by courts in numerous prior redistricting and other voting rights cases. See, e.g., Tr. 2095:6-2096:9 (in recent Ohio partisan gerrymandering case, stating that Dr. Hood drew “some inapt comparisons”); Tr. 2096:14-24 (in Texas voter ID case, stating that Dr. Hood’s testimony and analysis was “unconvincing” and given “little weight”); Tr. 2096:25-2097:19 (in Arizona voting rights case, “afford[ing] little weight to Dr. Hood’s opinions” “[f]or a number of reasons”); Tr. 2097:22-2098:6 (in Georgia voter ID case, finding that “Dr. Hood’s absentee voting analysis is unreliable or not relevant to the questions the court must resolve”); Tr. 2098:9-16 (in Ohio case involving absentee ballots, affording Dr. Hood’s opinions “little weight”); Tr. 2098:22-2099:6 (in recent Virginia racial gerrymandering case, stating: “We do not credit Dr. Hood’s testimony for several reasons.”); Tr. 2099:9-2100:1 (in Ohio voting rights case, finding Dr. Hood’s views “of little value,” and explaining that “Dr. Hood’s testimony and report are in large part irrelevant to the issues before the court and also reflected methodological errors that undermine his conclusions”). 615. Dr. Hood did not offer—and does not have—any methodology for determining whether or not a map was drawn to create a partisan lean or bias. Tr. 2078:1-2079:3. 616. Dr. Hood’s testimony supports the view that the enacted plans were drawn intentionally to favor Republicans. Dr. Hood generally agreed that “the party that controls the legislative process is going to make the maps in their favor,” and that the enacted plans “were drawn to favor Republicans” using prior election results. Tr. 2079:4-2081:2. 262 (i) 617. Dr. Hood’s testimony about the redistricting process in North Carolina was unpersuasive Dr. Hood testified that the 2017 redistricting was a “fairly formulaic process” because the Whole County Provision and 2017 Adopted Criteria “really limits the discretion, to a large extent, of the map drawers.” Tr. 2038:4-2039:12; LDTX284 at 9-10 (“[T]he process is quite constrained, which greatly limits the ability of map drawers to create districts where partisan motives predominate.”). However, Dr. Hood did no work to determine whether any of those criteria actually prevented the mapmaker from gerrymandering the enacted plans to advantage Republicans. Tr. 2077:10-15. 618. Dr. Hood’s assertion that the Adopted Criteria “constrained” the “map drawer” is incorrect. The Adopted Criteria were not passed by the House and Senate Redistricting Committees until August 10, 2017. As discussed below, Dr. Hofeller had completed much of the General Assembly’s eventually enacted House and Senate districts by June 2017, a month and a half before the Adopted Criteria were passed. FOF § F.7. Logically, Dr. Hofeller could not have been following the Adopted Criteria when he was drafting these districts by June 2017. 619. Dr. Hofeller’s files further refute Dr. Hood’s assertions that the 2017 redistricting process was “quite constrained” and that it is difficult to prove the partisan intent behind the 2017 Plans. PX123 at 48-49 (Chen Response Report). Those files show Dr. Hofeller’s continuous efforts and exercise of his discretion to draw the district lines to maximize Republican advantage within the confines of the Whole County Provision, including various drafts that considered alternative possible districtings. FOF § B.2.b. (ii) 620. Dr. Hood’s testimony about the 2018 elections was unpersuasive For his analysis of the 2018 election results, Dr. Hood compared the number of seats Democrats actually won in 2018 to the number districts in Dr. Chen’s simulated 263 plans that lean Democratic using the 2010-2016 composite statewide election results. Tr. 2083:14-25. But that is an apples-to-oranges comparison, because the 2018 elections were different than the 2010-2016 composite statewide election results. Tr. 2084:1-5. In the 2010-2016 composite statewide election results, the Democratic vote share is 47.9%, whereas 2018 was a far more favorable environment for Democrats. Tr. 2084:12-24. 621. Dr. Hood made no attempt to perform an apples-to-apples comparison by comparing the actual 2018 election results under the enacted 2017 Plans to the performance of alternative nonpartisan plans under the 2018 election results. Tr. 2084:252087:19. (iii) 622. Dr. Hood’s testimony about North Carolina’s political geography was unpersuasive Dr. Hood’s analysis of North Carolina’s political geography is unpersuasive because Dr. Hood did not attempt to determine whether the Republican lean in the enacted 2017 Plans can be explained by political geography. Tr. 2094:18-21. By contrast, Dr. Hood agreed that the work of Drs. Chen, Mattingly, and Pegden does address whether political geography could explain the extreme partisan lean of the 2017 Plans. Tr. 2094:22-2095:2. 623. For his analysis of political geography, Dr. Hood analyzed how the partisan makeup of the State of North Carolina would change if its six largest counties were removed. Tr. 2089:14-17; LDTX140. But it is not possible to remove any counties from North Carolina, much less the six largest counties. Of course, hypothetically removing North Carolina’s six largest counties would make the state “[m]uch more rural,” Tr. 2089:18-22, and much more Republican-leaning, just as would removing New York City from the State of New York. d. Dr. Barber 264 624. Intervenor Defendants’ expert, Dr. Michael Barber, received his Bachelor of Arts degree in International Relations with an emphasis in Political Economy from Brigham Young University in 2008, his Masters in Political Science from Princeton University in 2011, and his Ph.D. in 2014. Tr. 2106:7–22, 2107:4–13, ID Ex. 98 p. 1. 625. Dr. Barber is currently an Assistant Professor at Brigham Young University and an affiliated faculty member with the Center for the Study of Elections and Democracy. Tr. 2109:9–18. 626. Dr. Barber teaches classes on Congress and the legislative process (which includes state-level legislative research), statistical analysis, and a seminar course on contemporary research in American politics. Tr. 2110:14–2111:13. 627. Dr. Barber recently testified as an expert witness in an election law case involving a dispute over ballot order in Federal Court in Florida. Tr. 2113:10–2114:6. 628. Dr. Barber has published 11 peer-reviewed articles involving American Politics, and an additional 5 articles that have been accepted for upcoming publication. Tr. 2111:22–2112:4, 2113:6–9; ID Ex. 98 pp.1–2. Many of these articles involve political ideology, issues of campaign finance, electoral politics, survey research methodologies, [and] political polarization. Tr. 2111:24–2112:4. 629. Dr. Barber was admitted by the Court as an expert in American politics, specifically on the topics of ideology and partisanship, geography of voters, and the analysis of elections results. Tr. 2118:2–13. 630. Dr. Barber offered no opinion as to whether North Carolina’s state legislative district maps were gerrymandered. 631. The Court finds that Dr. Barber’s criticisms of Dr. Cooper’s analysis unpersuasive and gives them little weight. 265 632. At the outset, the Court notes that none of Dr. Barber’s academic research or published articles concern redistricting or North Carolina, nor was redistricting in North Carolina “something [he] had given a lot of thought to” before being retained by Intervenor Defendants in this case. Tr. 2169:19-2170:19. Dr. Barber admitted that he was not an expert on North Carolina’s political geography, nor had he spent time in North Carolina other than two vacations in the Outer Banks and one visit to Duke’s campus. Tr. 2168:122169:13, 2216:4-8. Most importantly, Dr. Barber did not analyze the specific district boundaries or county groupings the Court is reviewing and he could not comment on any of Dr. Cooper’s extended analysis of the packing and cracking of Democratic voters in those districts and county groupings. Tr. 2117:24-2118:12, 2213:25-2214:15 633. The majority of Dr. Barber’s testimony concerned the opinions Dr. Cooper offered regarding the aggregate political ideology of the North Carolina electorate and that of the General Assembly, including Dr. Cooper’s comparison between the two. The Court finds it unnecessary to determine whether the General Assembly is “out of step” with the electorate and therefore, makes no findings regarding Dr. Cooper’s testimony, or Dr. Barber’s criticism of that testimony, relating thereto. 634. Dr. Barber also sought to rebut opinions Dr. Cooper offered regarding the disproportionality between Democratic seat share and the Democrats’ statewide vote share in the General Assembly after the 2011 redistricting. Dr. Barber observed that “it’s actually not as rare as you might think” that a party wins a majority of votes for the North Carolina House or Senate statewide, but only a minority of seats. Tr. 2149:21-2150:2. But since Dr. Barber did not analyze the extent to which any of these instances of disproportionality between votes and seats were attributable to gerrymandered district boundaries, his analysis is less useful to the Court. Dr. Barber admitted that it was “very possible” that those instances from 2002-2006 where the Democrats won a minority of the 266 statewide vote and yet a majority of seats in a chamber of the General Assembly “could have been because the Democrats did a good job of gerrymandering the maps that were in place during those elections.” Tr. 2203:12-16. 635. In support of his opinion regarding the translation of seats from votes, Dr. Barber created a chart providing the “absolute difference” in percentage between the vote share and seat share for each party in House and Senate elections since 1994. IDTX23. But as Dr. Barber acknowledged, the greatest difference between the percentage of Republican vote share and seat share in the House occurred in the 2012 election, just after the 2011 redistricting. Tr. 2207:3-12. The difference in the Senate between the percentage of Republican votes received and seats won was also relatively large in 2012, and represented a significant increase from the 2010 election, just before redistricting. Tr. 2207:13-22. If anything, Dr. Barber’s analysis suggests that the 2011 redistricting led to more disproportionality between votes cast and seats won, as Dr. Cooper observed. See Tr. 2207:23-2212:16. 636. Finally, Dr. Barber noted that there is “academic research that points to political party geography as an important factor in representation and legislatures,” suggesting that the geographic distribution of voters “is something that should be investigated” in this case. Tr. 2152:10-14. Specifically, Dr. Barber referenced a 2013 article co-authored by Plaintiffs’ expert, Dr. Chen, focused on the political geography of Florida and Florida’s congressional districts, an article in which Dr. Chen used simulations to measure whether political geography created a natural advantage for Republicans in redistricting in Florida. Tr. 2153:2-24. Despite acknowledging that Dr. Chen’s co-authored 2013 article did not include any analysis of North Carolina, Tr. 2153:25-2154:2, Dr. Barber testified that the article “invites the question as to what it would look like if we looked to see if this relationship also existed in North Carolina,” Tr. 2154:5-7. 267 637. Dr. Chen performed that analysis in this case and concluded that North Carolina’s political geography does not account for the extreme partisan bias of the enacted plans. Tr. 2216:11-2220:21. Similarly, at the time he conducted his analysis and arrived at the opinions he offered regarding the potential partisan bias of North Carolina’s political geography, Dr. Barber was unaware that Dr. Chen’s co-author in the same 2013 paper, Dr. Jonathan Rodden, had come to the conclusion that North Carolina’s Democratic voters were relatively efficiently distributed throughout the State. Tr. 2222:9-2223:4, 2224:6-2225:8. 638. Dr. Barber did not engage in the type of analysis that Dr. Chen performed to account for and measure the extent to which “natural” partisan bias in North Carolina’s political geography could account for electoral outcomes favoring Republicans, but the analysis that Dr. Barber did conduct of the distribution of North Carolina’s Democratic voters actually supports Plaintiffs’ claims. Dr. Barber observed a positive correlation between the population density of North Carolina’s VTDs and their support for Democratic candidates, but he acknowledged that there were “a lot of other Democratic-leaning VTDs” spread across the state, even outside the urban centers of Raleigh and Charlotte. Tr. 2216:11-16. Dr. Barber’s analysis fails to offer the Court any information about how the many Democratic-leaning VTDs across North Carolina fit into specific county groupings and specific districts and therefore, his analysis is not directly relevant to the questions the Court faces. Unlike Dr. Cooper, who performed an extensive analysis of North Carolina’s House and Senate Districts at the county grouping level, Dr. Barber admitted that he could not offer any opinion to rebut Plaintiffs’ evidence regarding gerrymandering within those county groupings. Tr. 2217:8-2218:12. 639. In light of the above shortcomings in Dr. Barber’s analysis, the Court gives little weight to his testimony. 268 e. 640. Dr. Johnson Legislative Defendants’ expert Dr. Douglas Johnson has a Bachelor of Arts in Government from Claremont McKenna College, a Master of Business Administration from the Anderson School at UCLA, and a Ph.D. in Political Science from Claremont Graduate University. Tr. 1812:15-21; LDTX288. The focus of Dr. Johnson’s graduate studies in Political Science was American politics, and he wrote his dissertation on redistricting. Tr. 1812:22-25. 641. Dr. Johnson is a fellow at the Rose Institute of State and Local Government at Claremont McKenna College. Tr. 1813:1-6. In that role, he leads the Institute’s research into census and redistricting issues. Tr. 1813:1-6. 642. Dr. Johnson is also the President of National Demographics Corporation (“NDC”), where he has been employed full-time since 2001. Tr. 1814:7-19. NDC is engaged in redistricting work, including liability analyses, polarized voting studies, and other related redistricting issues. Tr. 1814:20-25. 643. Dr. Johnson has used Maptitude for Redistricting software (“Maptitude”) for his work for 20 to 30 hours a week since 2001. Tr. 1816:16-23. 644. Dr. Johnson has served as an expert witness in redistricting litigation numerous times; specifically, he has been involved in hundreds of challenges to at-large elections for city councils, school boards, counties, etc. Tr. 1817:5-7; 1817:14-21. Dr. Johnson has also served as an expert witness in challenges to state redistricting plans. Tr. 1817:2224. Dr. Johnson has never been excluded as an expert witness by any court. Tr. 1817:8-10. 645. Dr. Johnson was accepted by the Court as an expert in the fields of political science, political geography, redistricting, and Maptitude for Redistricting software. Tr. 1818:11-20. 646. Dr. Johnson offered primarily two sets of opinions in this case. First, Dr. Johnson purported to show that one could draw a Senate map even more favorable to 269 Republicans if one ignored the North Carolina Constitution’s Whole County Provision. Second, Dr. Johnson attempted to critique Dr. Chen’s analysis of Dr. Hofeller’s files. 647. The Court finds Dr. Johnson’s analysis unpersuasive and gives his opinions little weight. 648. Dr. Johnson has testified as a live expert witness in four cases previously, and the courts in all four cases have rejected his analysis. Tr. 1886:21-1891:14; see Covington, 283 F. Supp. 3d at 450 (finding “Dr. Johnson’s analysis and opinion . . . unreliable and not persuasive”); Luna v. Cnty. of Kern, 291 F. Supp. 3d 1088, 1137 (E.D. Cal. 2018) (holding that defendants’ argument based on Dr. Johnson’s analysis “lacks merits”); Garrett v City of Highland, 2016 WL 3693498, at *2 (Cal. Super. Apr. 06, 2016) (finding Dr. Johnson’s methodology “inappropriate”); Jauregui v City of Palmdale, No. BC483039, 2013 WL 7018375, at *2 (Cal. Super. Dec. 23, 2013) (describing Dr. Johnson’s work in the case was “unsuitable” and “troubling”). This Court joins these other courts in rejecting Dr. Johnson’s methodologies, analyses, and conclusions. 649. Dr. Johnson created a “test map” for the North Carolina Senate that ignored the Whole County Provision entirely. Tr. 1892:21-1893:4. Based on this test map, Dr. Johnson purported to find that one could draw a Senate map even more favorable for Republicans than the enacted Senate plan if one were to ignore the county groupings and traversal rules. Tr. 1893:17-22. The Court finds Dr. Johnson’s analysis using his test map to be of little probative value to the legal and factual issues in this case. 650. Dr. Johnson performed no statewide analysis of the House or the Senate to determine the extent to which, within the confines of the Whole County Provision, the enacted House and Senate plans constitute the most favorable maps for Republicans possible. Tr. 1894:13-1896:7. The only individual county groupings for which Dr. Johnson performed partisanship analysis within the confines of the Whole County Provision were 270 Mecklenburg County in the Senate, id., and Wake County in the House, and Dr. Johnson’s partisanship analysis of the Mecklenburg Senate districts was erroneous and not credible for the reasons already explained. See supra, para 251. Dr. Johnson did not analyze any other individual House or Senate county grouping to determine whether the enacted plans’ version of that grouping is the most favorable configuration of the grouping possible for Republicans. Id. Dr. Johnson thus offered no rebuttal to the testimony of Plaintiffs’ experts demonstrating that the enacted plans constitute extreme partisan gerrymanders of specific county groupings. 651. Dr. Johnson instead ignored the Whole County Provision in creating his Senate test map, but as he acknowledged, the Whole County Provision is a state constitutional requirement. Tr. 1896:8-10. The General Assembly lacks authority to ignore the state constitutional county groupings and traversals requirements in creating redistricting plans. Dr. Johnson’s test map analysis is thus no more relevant or helpful than would be a test map that ignores other constitutional requirements, such as the equal population requirement for districts. One could draw a map ignoring the equal population requirement that is even more favorable for Republicans than Dr. Johnson’s test map, and certainly more favorable for Republicans than the enacted plan. Tr. 1896:11-1900:21. But the fact that one could draw such a hypothetical map in no way sheds light on whether the enacted plan is an extreme partisan gerrymander. See id. It provides no information as to whether the General Assembly acted within extreme partisan intent in drawing districts within the confines of the accepted constitutional requirements, and it provides no information as to the effects of the gerrymander on the number of Republican and Democratic-leaning districts relative to a nonpartisan plan. See id. Dr. Johnson’s test map analysis is of little probative value to the legal or factual issues in this case. 271 652. With respect to Dr. Johnson’s testimony regarding Dr. Hofeller’s files, as described above, the Court struck all of Dr. Johnson’s affirmative analysis of Dr. Hofeller’s 2017 draft House and Senate plans and the extent to which they overlap with other plans including the final enacted plans. Tr. 1988:11-1990:4. The Court struck this testimony and all related portions of Dr. Johnson’s rebuttal report under Rule 702 and Rule 403 after it was uncovered on cross-examination that Dr. Johnson had made a series of significant errors. Id. 3. Dr. Karen Owen’s Testimony on “Representation” and “Competitive Elections” and Representative John Bell’s Testimony on Competitive Districts Was Unpersuasive a. Dr. Karen Owen 653. Legislative Defendants offered expert testimony of Dr. Karen Owen on the issues of “representation” and “competitive elections” in North Carolina. Tr. 1488:6-22; LDTX 293 (Owen report). 654. Dr. Owen is an assistant professor of political science at West Georgia University, and focuses on southern politics, political representation, legislative politics, campaigns and elections and research methodology, and developed her expertise through both academic and professional work. Tr. 1481:18-22, 1483:16-24, 1484:2-1485:24, 1486:411; LDTX293 at 1-2, 28-34. 655. Dr. Owen has particular expertise in the area of southern politics; she has presented papers and been a lead discussant at the Citadel’s Symposium on Southern Politics for over 10 years, she has taught and studied courses in southern politics. Tr. 1480:15-1481:4. 656. Dr. Owen’s work in southern politics has included writing and presenting a paper in 2016 titled “Growth and Geography in the South: Representation in the North Carolina and Texas State Legislatures.” Tr. 1481:5-11; LDTX293 at 31. 272 657. The Court admitted Dr. Owen as an expert. Tr. 1487:24-1488:1. 658. Dr. Owen has very little experience or expertise with politics, elections, or representation in North Carolina specifically. Dr. Owen has never lived or worked in North Carolina. LDTX 293 at 28-29. With the exception of the aforementioned paper, she has never written or published about North Carolina politics, elections, or representation. Tr. 1555:19-1557:25. She has never participated in or spoken at any conference about North Carolina politics, elections, or representation. Tr. 1558:1-1559:16. She has never been interviewed by any media outlet about North Carolina politics, elections, or representation. Tr. 1559:17-25. She has never taught a class focused on North Carolina politics, elections, or representation—the closest she came was teaching a single course in “Southern Politics” three years ago. LDTX 293 at 32; Tr. 1560:11-24. 659. The methodologies Dr. Owen employed to evaluate “representation” and “competitive elections” in North Carolina were unpersuasive. In conducting her research and analysis for this case, Dr. Owen did not speak to any current or former North Carolina legislator, or any winning or losing North Carolina candidate, or any North Carolina voter. Tr. 1561:7-1564:14. Nor did she consult any North Carolina polling data or survey data. Tr. 1564:15-19. Instead, Dr. Owen’s analysis of representation in North Carolina was based on her conversations with several staff members in the General Assembly’s Legislative Services Commission. Tr. 1561:7-1562:1. Her analysis of competitive elections in North Carolina was based on her reading of newspaper articles and a website called “Real Facts North Carolina.” Tr. 1566:5-13. 660. Based on her lack of relevant expertise and the inadequate methodologies she employed in this case, the Court gives little weight to Dr. Owen’s opinions about “representation” and “competitive elections” in North Carolina. 273 661. In addition, as described below, Dr. Owen’s analysis and opinions are unhelpful in resolving the issues in this case. i. 662. Dr. Owen’s analysis of “representation” was unpersuasive In support of her opinion that Republican members of the General Assembly meaningfully “represent” their Democratic constituents, Dr. Owen emphasized that the members “are noticeably involved in more than producing and passing laws,” LDTX 293 at 22, and that they provide “constituent services” to Republican and Democratic voters alike, regardless of their political beliefs, party affiliation, or past votes. Tr. 1567:15-1568:18; see also Tr. 1801:17-1803:2 (similar testimony by Rep. Bell); Tr. 2000:21-2001:6 (Sen. Brown). 663. The Court finds, however, that the mere provision of constituent services does not mean that voters of one particularly party are meaningfully “represented” by a member of the other party political and does not mean the voter receives the same “representation” that the voter would if he or she could elect the candidate of that voter’s choice. Constituent services are only one part of a legislator’s responsibilities. In addition to providing constituent services, members of the North Carolina House and Senate participate in enacting the State’s laws and policies. Tr. 1803:3-9 (Rep. Bell). Legislative Defendants’ own expert, Dr. Brunell, testified that, among the ways in which a legislator “represents” his or her constituents, providing constituent services may be “an important part, but if you are sort of, you know, worried about the hierarchy of the things that they do, I think that how they vote on the major issues of the day is more important.” Tr. 2353:11-2354:4. Dr. Brunell agreed that “policy responsiveness” is a “higher form of representation” and “more critical to the notion of representing someone.” Tr. 2354:5-10; see Tr. 2353:3-6 (agreeing that “the responsiveness of a legislator to the voters on questions on policy in particular is critical to Democratic representation”). As “just one example of the many issues from which policy responsiveness is the more central form of representing 274 the people in the legislature,” Dr. Brunell agreed that if a legislator casts a vote for gun control, the legislator is “not giving good representation to the voters in [his or her] district who don’t want gun control.” Tr. 2354:11-19. Thus, as Dr. Brunell agreed, “a change in the party that represents a given district generates a huge difference in the policies for which the representative of that district will vote.” Tr. 2354:20-23. Another witness for Legislative Defendants, Senator Harry Brown, also testified that “in order to push legislation that we thought was important to this state,” a political party must “be in the majority.” Tr. 2023:20-22. 664. Other purported indicia of “representation” discussed by Dr. Owen likewise were unhelpful. For example. Dr. Owen pointed to a form “welcome letter” that members of the General Assembly can send to new voters in their districts. LDTX 293 at 22; Tr. 1514:4-1516:23. But sending a form letter does not signify meaningful representation. ii. 665. Dr. Owen’s analysis of “competitive elections” was unpersuasive In her analysis of “competitive elections,” Dr. Owen suggested that Democrats’ failure to win certain House and Senate races in 2018 was the result of poor “candidate quality,” rather than the district boundaries. Tr. 1540:13-1542:9; LDTX 293 at 6-7. Dr. Owen’s methodology was unreliable, and her conclusions were unpersuasive. 666. The sole criterion that Dr. Owen applied for assessing candidate quality turns on whether the candidate “had held prior elected office.” Tr. 1533:5-21. Under this “dichotomous measure,” any person who has previously held elective office is a “quality” candidate, and any person without prior experience holding elective office is not “quality.” LDTX 293 at 10. This approach ignores other important factors and is an unreliable measure of whether a person is a quality candidate. 667. For instance, Dr. Owen classified a Democratic candidate who is a U.S. Army Colonel as a “nonquality” candidate. Tr. 1566:18-25; LDTX 293 at 12. She classified 275 another Democratic candidate who is a “small business owner” and “community leader” as a “nonquality” candidate. Tr. 1567:1-7; LDTX 293 at 12. And she classified a “young Air Force veteran and attorney” as a non-quality candidate. LDTX 293 at 16. These examples illustrate the shortcomings in Dr. Owen’s methodologies. b. Representative John Bell 668. Legislative Defendants also offered the testimony of Representative John Bell, IV, who testified about the competitiveness of various House districts. 669. Representative Bell is the majority leader for the North Carolina House of Representatives and represents House District 10. Tr. 1739:16-22. 670. As Majority Leader, Representative Bell assists the Conference chair to achieve two goals: 1) recruit candidates and 2) win elections. Tr. 1740:5-6. 671. Representative Bell also pointed to candidate quality as a purported factor in House districts he claimed might be “competitive” in 2020. Tr. 1752:13-1754:18. But Representative Bell’s claim that certain House districts could be “competitive” in 2020, and only were not close in 2018 due to purported candidate quality issues is not persuasive. Representative Bell included on his list of purportedly competitive districts numerous districts that were not only extremely lopsided in the 2018 state House elections, but that feature similarly lopsided vote shares under the results of prior statewide elections, including the 2012 Presidential election, the 2016 Presidential election, and the 2016 Governor election. Tr. 1788:5-1801:16. Representative Bell included on his list of purportedly competitive districts a handful of districts in which the Republican candidate won over 60% of the vote share in the district across all of these various elections. Id. Moreover, for many of the districts he identified, Representative Bell testified that the race could be competitive only if it was an “open seat”—that is, if the incumbent Republican member either retires or does not run again in 2020. Tr. 1767:3-23, 1772:16-20, 1773:24- 276 1774:2. However, there is no evidence that any of those Republicans members will not run in 2020. Tr. 1786:4-10. The Court finds that Representative Bell’s testimony does not provide a reliable basis for assessing the competitiveness of current House districts. 4. 672. The Whole County Provision Did Not Prevent Systematic Gerrymandering of the Plans for Partisan Gain Throughout trial, Legislative Defendants and their experts emphasized the existence of the North Carolina Constitution’s Whole County Provision, which the North Carolina Supreme Court has held requires dividing the State into discrete county groupings and restricting the traversal of county lines for districts within a county grouping. Tr. 252:17-257:10. The Court finds that Legislative Defendants overstate the constraints imposed by the Whole County Provision, and that Legislative Defendants intentionally and effectively gerrymandered the enacted plans for partisan gain within the confines of the Whole County Provision. 673. Legislative Defendants overstate the impact of the Whole County Provision. Dr. Chen explained in unrebutted testimony that the Whole County Provision dictates the contours of only 13 of 120 House districts and 17 of 50 Senate districts. Tr. 782:2-783:1. Legislative Defendants thus had discretion in drawing 107 of 120 House districts and 33 of 50 Senate districts—constituting over 82% of all districts across both enacted plans. Id. 674. As detailed above, the evidence establishes that Legislative Defendants engaged in systematic gerrymandering for partisan gain in the districts in which they did have discretion. All four of Plaintiffs’ experts concluded that Legislative Defendants acted with extreme partisan intent within the confines of the Whole County Provision. Plaintiffs’ simulations experts—Drs. Chen, Mattingly, and Pegden—simulated plans that adhered to the existing House and Senate county groupings, and all three experts found that the enacted plans are extreme outliers compared to nonpartisan plans that follow the same 277 county groupings. And all three experts found that specific county groupings are extreme outliers compared to other, simulated versions of the same county grouping that contain the same number of traversals as the enacted plan in that grouping. Dr. Cooper independently established—in unrebutted testimony—that the enacted plans pack and crack Democratic voters within specific county groupings. 5. 675. Plaintiffs Do Not Seek Proportional Representation Contrary to Legislative Defendants’ claim, Plaintiffs do not seek proportional representation. As described in more detail below, Plaintiffs assert that the General Assembly may not intentionally discriminate against voters and may not attempt to predetermine election outcomes and control of the General Assembly. Dr. Chen and Dr. Mattingly established through their simulations that nonpartisan plans that do not intentionally discriminate against Democratic voters may well not provide for proportional representation. Under Dr. Chen’s and Dr. Mattingly’s simulations, there are scenarios where Democrats would win 50% of the statewide vote but less than 50% of the seats in either chamber. Tr. 306:16-307:2 (Dr. Chen); Tr. 1103:24-1104:5, 1132:6-1133:13 (Dr. Mattingly). Dr. Pegden’s simulations also did not rely on any notion of proportional representation. Tr. 1306:22-24. 676. Legislative Defendants’ presentation regarding the proportionality of seats to votes in specific county groupings like Wake and Mecklenburg Counties, Tr. 2068:102069:13, was not persuasive. As Dr. Pegden explained, analyzing proportionality at the local level of a county grouping is “completely useless” and can be misleading in the context of a gerrymandered map. Tr. 1452:17-1454:18. In a county grouping that contains a small number of districts and in which one party wins an overwhelming share of the vote across the grouping, one would expect that party to win a disproportionate share of the seats under a nonpartisan map, and likely all of the seats. Tr. 1452:23-1453:12. Under a 278 Republican gerrymander, however, Republican mapmakers will allow that natural outcome to occur in county groupings that strongly favor Republicans but will gerrymander the more Democratic county groupings in a way that may result in proportional outcomes just in those Democratic county groupings—e.g., by gerrymandering the grouping to elect one or two Republican seats. Tr. 1452:17:22-1454:18. Thus, the fact that the enacted plans may have resulted in proportional seats-to-votes outcomes in individual county groupings that are heavily Democratic is not evidence of a lack of gerrymandering. 6. 677. Legislative Defendants Did Not Seek to Comply with the VRA and Did Not Show Nonpartisan Plans Would Violate the VRA Defendants did not present persuasive evidence at trial to substantiate any federal defense under the Voting Rights Act or Fourteenth or Fifteenth Amendments. Defendants did not introduce persuasive evidence at trial to establish any of the prerequisites to application of the Voting Rights Act under Thornburg v. Gingles, 478 U.S. 30 (1986). For example, Defendants presented no expert testimony or any other evidence to establish the existence of legally sufficient racially polarized voting in any area of North Carolina, or any particular state House or state Senate district. Nor did Defendants introduce any evidence to establish the minimum African-American percentage of the voting age population (“BVAP”) needed in any particular area of the State for the African American community to be able to elect the candidate of its choice. 678. Notably, Legislative Defendants retained Dr. Jeffrey Lewis, a political scientist from UCLA, who analyzed and provided estimates of the minimum BVAP needed in certain county groupings for African-American-preferred candidates to win. See PX773 (Amended Table 4 from Lewis Report). But Legislative Defendants chose not to have Dr. Lewis testify at trial. At the conclusion of trial, Legislative Defendants attempted to introduce expert reports that a different political scientist (Dr. Alan Lichtman) had 279 prepared on behalf of different parties in previous lawsuits in North Carolina years ago, but the Court sustained Plaintiffs’ objections to the admission of these reports. Tr. 2376:2-3. The Court excluded these reports as inadmissible hearsay and undisclosed expert work, particularly given that Plaintiffs dispute Legislative Defendants’ characterization of those reports. Tr. 2363:16-2364:25. 679. Defendants did not demonstrate that the relief Plaintiffs seek would violate the VRA or federal equal protection requirements. Plaintiffs established that it would not. Using Dr. Lewis’s estimates of the minimum BVAP needed in certain county groupings for an African-American-preferred candidate to win a state House or Senate election, Dr. Chen determined how many of his simulations of those county groupings contained districts exceeding Dr. Lewis’s BVAP-threshold estimates. Tr. 512:15-517:6. Dr. Chen determined that for every county grouping that Dr. Lewis analyzed except one in the House and one in the Senate, all of Dr. Chen’s simulations produce at least as many districts above Dr. Lewis’s BVAP-threshold estimate as does the enacted House or Senate plan. Id.; see PX775; PX776. For the two remaining county groupings, which are Forsyth-Yadkin in the House and Davie-Forsyth in the Senate, a majority of Dr. Chen’s simulations of each grouping produce at least as many districts above Dr. Lewis’s BVAP-threshold estimate as the enacted plan. Id.; see PX775; PX776. The evidence at trial thus demonstrated that, based on the BVAP-threshold estimates of Legislative Defendants’ own expert, adopting nonpartisan House and Senate plans would not diminish the ability of African Americans to elect the candidate of their choice. 680. While Defendants’ failure to introduce any evidence at trial necessary to the legal elements of a racial vote dilution defense is dispositive of any such defense, the Court further finds that—as a factual matter—Legislative Defendants did not draw or adopt any district under the 2017 Plans in an effort to comply with the VRA. 280 681. One of the Adopted Criteria, titled “No Consideration of Racial Data,” stated that “[d]ata identifying the race of individuals or voters shall not be used in the drawing of legislative districts in the 2017 House and Senate plans.” LDTX155. When submitting the plans to the Covington court for approval, Legislative Defendants stated that “[d]ata regarding race was not used in the drawing of districts for the 2017 House and Senate redistricting plans.” PX629 at 10. 682. Legislative Defendants have claimed in this case that, even though they did not use racial data in drawing the districts, they purportedly checked the racial demographics of the districts on the “back end” to ensure that “the VRA was satisfied.” See, e.g., Leg. Defs.’ Pre-Trial Brief at 44. Legislative Defendants presented no testimony at trial to substantiate this assertion, and the Court finds the assertion not credible for multiple reasons. 683. Throughout the 2017 redistricting process, Legislative Defendants asserted that the reason they were ignoring racial considerations entirely in drawing the new districts was because they had concluded that the “third Gingles factor” was not “present” anywhere in the State of North Carolina. PX593 at 52 (statement of Sen. Berger); see also id. (“we cannot prove the third Gingles factor”) (statement of Sen. Berger). Legislative Defendants repeatedly told the Covington court that they could not “justify the use of race in drawing districts” in the 2017 Plans—and thus could not seek to hit a “racial numerical quota” for any district—because they had insufficient evidence of “legally sufficient racially polarized voting.” Covington, No. 15-cv-399, ECF No. 184 at 10; ECF No. 192 at 12; see also ECF No. 184-17 at 12. 684. The existence of legally sufficient racially polarized voting is a “prerequisite[]” to VRA liability; if any Gingles factor is not met, “§ 2 simply does not 281 apply.” Cooper v. Harris, 137 S. Ct. 1455, 1472 (2017). Hence, when Legislative Defendants concluded that the third Gingles factor was not met, they necessarily concluded that the VRA did not impose requirements for the racial composition of any state House or state Senate district. Any assertion by Legislative Defendants now that they sought to “satisfy” the VRA in adopting the 2017 Plans does not make sense as a legal or factual matter given their assertions at the time. 685. Moreover, the mere timing of when Legislative Defendants learned of the racial composition of the new districts belies their claim that they reviewed the data to ensure VRA compliance. The Stat Packs that Legislative Defendants produced when they released the initial drafts of the House and Senate plans did not include racial data on any of the draft districts.13 At the August 24, 2017 hearing at which the Senate Redistricting Committee passed the Senate plan out of committee, Senator Hise insisted, “I have not seen any racial data for these districts.” PX606 at 46:2-3. Representative Lewis said the same the next day at the hearing at which the House plan was passed out of the House Redistricting Committee. PX605 at 20:11-21:18. Only after this point did legislative staff produce racial data on the districts—at the request of Democratic legislators over Legislative Defendants’ objections. PX600 at 11. Even then, Legislative Defendants claimed to have remained unaware of the racial composition of the districts. Representative Lewis asserted that he did not “see” any data on the racial composition of the House districts until after the House plan was passed by the full House chamber. Id. at 12. Legislative Defendants clearly did not have assure themselves that the plans satisfied See https://bit.ly/2YJnaRP (Stat Pack for Senate draft plan released on August 21, 2017); https://bit.ly/2YPch0L (Stat Pack for House draft plan released on August 20, 2017). 13 282 the VRA by meeting particular racial thresholds when they purportedly had no knowledge of the racial composition of the districts. 686. Legislative Defendants have pointed to a single floor statement by Senator Berger near the end of the legislative process that mentioned the VRA, but that statement does not establish that Senator Berger, let alone any other Legislative Defendant, actually undertook efforts to comply with the VRA. Senator Berger made that statement immediately after declaring that the third Gingles factor was not met, which if true would preclude VRA application as a matter of law. PX593 at 52-54. And neither Senator Berger nor anyone else has pointed to any change that was made to any House or Senate district to ensure VRA compliance. 687. The Court finds that the General Assembly did not enact any House or Senate district under the 2017 Plans with the specific intent of complying with the VRA, and that Defendants have not established that the VRA requires maintaining any of the districts that Plaintiffs challenge in its current form. 688. Indeed, the Court finds that Legislative Defendants’ stated concern that “unpacking” heavily-Democratic districts could dilute the voting power of AfricanAmericans to be a pretext for partisan gerrymandering. Unrebutted evidence presented at trial established that Legislative Defendants themselves created districts with artificially low BVAPs when it was politically advantageous. In particular, while Legislative Defendants now accuse Plaintiffs of seeking to “crack” African American voters, the unrebutted evidence established that Legislative Defendants cracked African American voters in rural and semi-rural parts of the state where cracking Democratic voters would maximize Republican victories. 689. Dr. Chen demonstrated that, for several rural and semi-rural House county groupings, all or nearly all of his simulated plans (which ignored racial data in drawing the 283 districts) produced a district in the grouping with a higher or much higher BVAP than any districts in that grouping under the enacted plan. Tr. 519:6-523:9. These county groupings include the Anson-Union, Cleveland-Gaston, Columbus-Pender-Robeson, and DuplinOnslow county groupings, all of which are county groupings in which Legislative Defendants cracked Democratic voters to dilute their political power. Id.; see PX225; PX226; PX227; PX228. Dr. Chen’s findings significantly undermine Legislative Defendants’ claims that they seek to create higher-BVAP districts to promote the political power of African-American communities. Id. 7. 690. Legislative Defendants, through Dr. Hofeller, substantially completed drafting the Enacted Maps in June 2017 Based on an analysis of draft maps from June 2017 found on Dr. Hofeller’s storage devices, see FOF § B.2., Plaintiffs’ expert Dr. Jowei Chen demonstrated that Dr. Hofeller had begun drawing the 2017 Plans prior to July 2017, and that he had already substantially completed them by that point. Dr. Chen’s analysis compared the draft maps found on Dr. Hofeller’s hard drive, each of which is dated by the metadata, with the Enacted 2017 House and Senate maps to determine the degree of similarity between the drafts and the Enacted Plans. 691. For the Senate, Dr. Chen analyzed a draft map that Dr. Hofeller last modified on June 24, 2017. Tr. 400:7-10, 402:5-403:8; see also PX572 (showing “last modified” date); PX123 at 25 (Chen Rebuttal Report). Dr. Chen found that Dr. Hofeller had already finished assigning 97.6% of the State’s census blocks and 95.6% of the State’s population to their final Senate districts in this June 24, 2017, draft map. Tr. 400:6-25. 692. To show the extent to which Dr. Hofeller had already completed drawing the new Senate plan, Dr. Chen compared individual Senate county groupings in the June 24, 2017, draft map to the final version of the same grouping in the enacted Senate plan. The 284 figure below, PX142 [Chen rebuttal report, Figure 19], shows one such comparison for a Senate county grouping containing multiple districts that was redrawn in 2017. Tr. 416:1520; PX123 at 27-38 (Chen Rebuttal Report). Dr. Chen repeated this analysis for every Senate county grouping containing multiple districts that was redrawn in 2017, and the Court adopts, by reference to Dr. Chen’s trial testimony and as illustrated in his Rebuttal Report, each of those illustrations as if fully set forth herein. Tr. 404:19-417:13; PX140; PX141; PX142; PX143; PX144; PX145; PX146; PX147 [Chen rebuttal report, Figures 17-24]. 693. In Dr. Chen’s illustrations, as shown by the example below, the map on the bottom left is Dr. Hofeller’s June 24, 2017, draft, the map on the bottom right is the final enacted plan, and the top half of the figure reports the percentage of the population in each district in Dr. Hofeller’s draft (on the vertical axis) that were assigned to the corresponding district in the final enacted plan (on the horizontal axis). Tr. 405:5-407:18. For instance, the figure included below shows that 99.42% of the population assigned to Senate District 19 in Dr. Hofeller’s June 24, 2017 draft was also assigned to Senate District 19 in the enacted Senate plan, while 100% of the population in Dr. Hofeller’s draft Senate District 21 was assigned to Senate District 21 in the enacted plan. Id. 285 694. Based on Dr. Chen’s analysis of each Senate county grouping containing multiple districts that was redrawn in 2017, the Court finds that by June 24, 2017—nearly seven weeks before the Adopted Criteria were passed on August 10, 2017—Dr. Hofeller had 286 fully or at least substantially completed drawing every Senate county grouping redrawn in 2017. Tr. 404:23-417:13. The only Senate districts that were not an over-90% match to their final corresponding districts were a few heavily Democratic districts in Wake and Mecklenburg Counties. Tr. 412:5-414:12; see PX146; PX147. 695. Contrary to Legislative Defendants’ contention, the North Carolina Constitution’s Whole County Provision is not responsible for the high degree of overlap between Dr. Hofeller’s draft Senate plan and the final enacted plan. As Dr. Chen testified, the Whole County Provision did not dictate the contours of Senate districts in counties such as Cumberland, Forsyth, Johnston, Durham, Wake, Mecklenburg, and Guilford Counties, and Dr. Hofeller’s June 24, 2017 draft districts in these counties distinctly match the final versions. Tr. 408:13-416:1. 696. As with the Senate, Dr. Chen found that Dr. Hofeller had substantially completed drawing the new House plan by June 2017. Analyzing a draft House plan that Dr. Hofeller last modified on June 28, 2017, see PX569, Dr. Chen found that Dr. Hofeller had already finished assigning 90.9% of North Carolina’s census blocks and 88.2% of the State’s population into their final House districts in the June 28, 2017 draft plan. Tr. 401:15-23, 417:14-418:2, PX123 at 2-3 (Chen Rebuttal Report). 697. The figure below, PX124 [Chen rebuttal report, Figure 1], shows Dr. Chen’s analysis comparing Dr. Hofeller’s June 28, 2017, draft House map to the final enacted House map for a single House county grouping, in this instance, Mecklenburg County. Dr. Chen repeated this analysis for every House county grouping containing multiple districts that was redrawn in 2017, and the Court adopts, by reference to Dr. Chen’s trial testimony and as illustrated in his Rebuttal Report, each of those illustrations as if fully set forth herein. Tr. 417:14-427:15; PX124; PX125; PX126; PX127; PX128; PX129; PX131; PX132; PX133 [Chen rebuttal report, Figures 1 – 6, 8-10] 287 698. Based on Dr. Chen’s analysis, the Court finds that by June 28, 2017—over six weeks before the Adopted Criteria were passed—Dr. Hofeller had fully or at least 288 substantially completed drawing numerous House county groupings redrawn in 2017. Tr. 419:12-427:1. 699. Contrary to Legislative Defendants’ contention, the Whole County Provision is not responsible for the high degree of overlap between Dr. Hofeller’s June 28, 2017 draft House plan and the final enacted House plan. Tr. 419:12-427:1. The Whole County Provision does not dictate the contours of House districts in counties such as Mecklenburg, Harnett, Wayne, Sampson, Orange, Durham, Pitt, Robeson, Granville, Forsyth, and Rockingham Counties, and Dr. Hofeller’s June 28, 2017, draft House districts in these counties were near-exact matches to the final districts. Id. 700. The Court finds Dr. Chen’s comparisons of Dr. Hofeller’s June 2017 draft plans to the enacted plans to be highly credible and persuasive. Notably, Dr. Chen’s analysis stands unrebutted. Legislative Defendants presented testimony from Dr. Douglas Johnson in an attempt to rebut Dr. Chen’s analysis. However, the Court struck all of Dr. Johnson’s analysis comparing Dr. Hofeller’s draft districts and the final enacted districts after Plaintiffs’ cross-examination exposed a series of significant errors and unreliable methodology. Tr. 1988:11-1990:4. 701. As for Dr. Johnson’s remaining criticisms of Dr. Chen’s methodology for calculating the overlap between Dr. Hofeller’s June 2017 draft plans and the final enacted plans, the Court assigns them no weight. The Court finds that Dr. Chen employed a reasonable methodology to estimate the degree of similarity between the draft and final plans, by simply calculating the percentage of census blocks and population in each draft district that was also assigned to the most closely corresponding district in the final enacted House or Senate plan. See Tr. 398:3-399:15. Dr. Chen’s methodology and findings also accord with a visual comparison of the draft House and Senate districts to the corresponding final versions. No party has disputed that the maps presented in Plaintiffs’ 289 Exhibits 124-129, 131-133, and 140-147 accurately reflect the district boundaries in Dr. Hofeller’s June 2017 draft plans and the final enacted plans. 702. The Court concludes from this showing, and therefore finds, that Dr. Hofeller, and consequently the Legislative Defendants who retained him, by having largely completed the drafting of House and Senate maps by June, 2017, did so with little regard for the Adopted Criteria, or the neutral, non-partisan criteria contained therein, which were not adopted by the Senate Redistricting Committee and House Select Committee on Redistricting until August 10, 2017, and provided to Dr. Hofeller on August 11, 2017. PX 603 at 4:23-5:5; PX629. The Court finds that this is further compelling evidence of the intent of Legislative Defendants to create legislative districts by subordinating Democratic voters for partisan gain and to entrench the power of the Republican majority. 703. Since Dr. Hofeller’s files came to light, Legislative Defendants have asserted that they did not know at the time that Dr. Hofeller was developing draft maps prior to August 2017 or that Plaintiffs cannot “connect” Dr. Hofeller’s draft maps to the General Assembly. See, e.g., Leg. Defs’. Pre-trial Brief, p. 36. The Court finds this argument unpersuasive. Dr. Hofeller was retained by the General Assembly on June 27, 2017, for the purposes of drawing the 2017 House and Senate maps. PX641. The Court finds it highly improbable that in the days leading up to his engagement, or in the nearly six weeks following, Dr. Hofeller never mentioned his draft maps to anyone connected with Legislative Defendants until after he received the Adopted Criteria on August 11, 2017— especially since, merely eight or nine days later, Legislative Defendants were able to reveal final drafts of his House and Senate maps. PX605 at 16:2-17:16; PX629 at 7. 704. The Court is troubled by representations made by Legislative Defendants, or attorneys working on their behalf, in briefs and arguments to the Covington Court and to General Assembly colleagues at committee meetings that affirmatively stated that no draft 290 maps had been prepared even as late as August 4, 2017. See, e.g., Covington, ECF No. 161 at 2, 4, 13, and 28-29; PX601 at 11-12; PX602 at 72-73; and PX629 at 3, 4, 6 and 10 (Covington, ECF No. 184). For the purposes of determining liability for the claims asserted in this litigation,14 the Court finds it unnecessary to delve further into these concerns, other than to note that the Court, as previously stated, is persuaded, and specifically finds, that Dr. Hofeller’s intent and actions, as evidenced throughout his map-drawing process from at least early June 2017, are attributable in full to Legislative Defendants. 14 In considering the appropriate remedy, the Court does take this finding into account, among others, when mandating that the remedial process be more transparent to the Court, the public, and the entire General Assembly. 291 CONCLUSIONS OF LAW I. THE STANDING OF PLAINTIFFS 1. The North Carolina Constitution provides: “All courts shall be open; every person for an injury done him in his lands, goods, person, or reputation shall have remedy by due course of law; and right and justice shall be administered without favor, denial, or delay.” N.C. Const. art. I, § 18. 2. “[B]ecause North Carolina courts are not constrained by the ‘case or controversy’ requirement of Article III of the United States Constitution, our State’s standing jurisprudence is broader than federal law.” Davis v. New Zion Baptist Church, 811 S.E.2d 725, 727 (N.C. Ct. App. 2018) (quotation marks omitted); accord Goldston v. State, 361 N.C. 26, 35, 637 S.E.2d 876, 882 (2006) (“While federal standing doctrine can be instructive as to general principles . . . and for comparative analysis, the nuts and bolts of North Carolina standing doctrine are not coincident with federal standing doctrine.”). At a minimum, a plaintiff in a North Carolina court has standing to sue when it would have standing to sue in federal court. 3. The North Carolina Supreme Court has broadly interpreted Article I, § 18 to mean that “[a]s a general matter, the North Carolina Constitution confers standing on those who suffer harm.” Mangum v. Raleigh Bd. of Adjustment, 362 N.C. 640, 642, 669 S.E.2d 279, 281 (2008). The “gist of the question of standing” under North Carolina law is whether the party seeking relief has “alleged such a personal stake in the outcome of the controversy as to assure that concrete adverseness which sharpens the presentation of issues upon which the court so largely depends for illumination of difficult constitutional questions.” Goldston, 361 N.C. at 30, 637 S.E.2d at 879 (quoting Stanley v. Dep’t of Conservation & Dev., 284 N.C. 15, 28, 199 S.E.2d 641, 650 (1973)). Although the North Carolina Supreme Court “has declined to set out specific criteria necessary to show 292 standing in every case, [it] has emphasized two factors in its cases examining standing: (1) the presence of a legally cognizable injury; and (2) a means by which the courts can remedy that injury.” Davis, 811 S.E.2d at 727-28. A. The North Carolina Democratic Party Has Standing 4. The Court determines that the North Carolina Democratic Party (NCDP) has standing, both to sue on its own behalf as an organization and to sue on behalf of its members. 5. “An association may have standing in its own right to seek judicial relief from injury to itself and to vindicate whatever rights and immunities the association itself may enjoy.” River Birch Assoc. v. Raleigh, 326 N.C. 100, 129, 388 S.E.2d 538, 555 (1990) (quoting Warth v. Seldin, 422 U.S. 490, 511, 95 S. Ct. 2197, 2211 (1975)). The Court finds instructive the United States Supreme Court holdings under federal standing principles that state political parties and organizations similar to the NCDP have standing to bring voting-rights challenges on their own behalf. See, e.g., Crawford v. Marion County Election Bd., 553 U.S. 181, 189 n.7 (2008); id. at 204-09 (Scalia, J., concurring); id. at 209 n.2 (Souter, J., dissenting); Gill v. Whitford, 138 S. Ct. 1916, 1938 (2018) (Kagan, J., concurring) (explaining how these standards can apply to political parties and similar organizations in a partisan gerrymandering case); Ohio A. Philip Randolph Inst. v. Householder, 373 F. Supp. 3d 978, 1076 (S.D. Ohio 2019); League of Women Voters of Mich. v. Johnson, 352 F. Supp. 3d 777, 801 (E.D. Mich. 2018). Indeed, the federal court in Common Cause v. Rucho held that the NCDP had standing to bring a partisan gerrymandering challenge on its own behalf—based in part on the testimony of Mr. Goodwin. See, Common Cause v. Rucho, 318 F. Supp. 3d 777, 830 (M.D.N.C. 2018), vacated on other grounds, 139 S. Ct. 2484 (2019). 293 6. The NCDP has standing in its own right to seek judicial relief in this case because the NCDP has sufficiently demonstrated the presence of a legally cognizable injury to NCDP and a means by which the courts of our State can remedy that injury.15 7. An association also “has standing to bring suit on behalf of its members when: (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.” River Birch Assoc., 326 N.C. at 130, 388 S.E.2d at 555 (quoting Hunt v. Wash. State Apple Adver. Comm'n, 432 U.S. 333, 343, 97 S. Ct. 2434, 2441 (1977)). An associational plaintiff need not show that all of its members would have standing to sue in their own right when seeking declaratory or injunctive relief; rather, it is sufficient if any “one” member would have individual standing. Id.; see also State Employees Ass’n of N.C., Inc. v. State, 357 N.C. 239, 580 S.E.2d 693 (2003) (reversing lower court decision that had required every member of association or organization to have standing). The Court finds instructive federal court holdings that organizations similar to the NCDP have standing to bring partisan gerrymandering challenges on behalf of their members. See, e.g., League of Women Voters of Mich., 373 F. Supp. 3d at 933, 937-38; Ohio A. Philip Randolph Inst., 373 F. Supp. 3d at 1072-73; Rucho, 318 F. Supp. 3d at 827, 835-36 (holding that the NCDP had standing to bring a partisan gerrymandering claim on behalf of its members). 8. The NCDP has standing to sue on behalf of its members in this case because its members—registered Democratic voters located in every state House and state Senate District across our State—otherwise have standing to sue in their own right, the interests 15 Furthermore, even under the federal standing requirements of (1) injury, (2) causation, and (3) redressability, see Gill v. Whitford, 138 S. Ct. 1916, 1929 (2018), the NCDP has such a personal stake in the outcome of the controversy that it has standing under this more stringent standard. 294 that the NCDP seeks to protect are germane to the NCDP’s purpose, and neither the claims asserted nor the declaratory and injunctive relief requested requires the participation of individual NCDP members in this lawsuit. B. Common Cause Has Standing 9. The Court further holds that Common Cause has standing, both to sue on its own behalf as an organization and to sue on behalf of its members. 10. The Court finds instructive federal court holdings that organizations similar to Common Cause have standing to bring partisan gerrymandering challenges on their own behalves and on behalf of their members. See, e.g., League of Women Voters of Mich., 373 F. Supp. 3d at 933, 937-38; Ohio A. Philip Randolph Inst., 373 F. Supp. 3d at 1072-75; Rucho 318 F. Supp. 3d at 830-31 (holding that Common Cause had standing to bring a partisan gerrymandering challenge). 11. Like the NCDP, Common Cause has standing in its own right to seek judicial relief in this case because Common Cause has sufficiently demonstrated the presence of a legally cognizable injury to Common Cause and a means by which the courts of our State can remedy that injury.16 12. Common Cause also has standing to sue on behalf of its members in this case because at least one of its individual members has standing to sue in his or her own right, the interests Common Cause seeks to protect in this case are germane to Common Cause’s purposes, and neither the claims asserted nor the declaratory and injunctive relief requested requires the participation of individual Common Cause members in this lawsuit. 16 Furthermore, even under the federal standing requirements of (1) injury, (2) causation, and (3) redressability, see Gill, 138 S. Ct. at 1929, Common Cause has such a personal stake in the outcome of the controversy that it has standing under this more stringent standard. 295 C. The Standing of Individual Plaintiffs 13. Individual Plaintiffs also have standing to challenge each of their individual districts as well as their county groupings. All of the Individual Plaintiffs detailed below have shown “a personal stake in the outcome of the controversy,” Goldston, 361 N.C. at 30, 637 S.E.2d at 879, and that the 2017 Plans cause them to “suffer harm,” Mangum, 362 N.C. at 642, 669 S.E.2d at 281. 14. Certain Individual Plaintiffs have standing to challenge their own districts. Plaintiffs introduced extensive district-specific evidence demonstrating how, through cracking and packing, the 2017 Plans dilute the voting power of Individual Plaintiffs and other Democratic voters. Plaintiffs also introduced unrebutted, district-specific evidence demonstrating that twenty-two Individual Plaintiffs live in House districts that are outliers in partisan composition relative to the districts in which they live under Dr. Chen’s nonpartisan simulated plans and that twenty Individual Plaintiffs live in Senate districts that are outliers in the same manner. FOF § E.3. Each of these Individual Plaintiffs thus established a personal stake in the outcome of the controversy and a specific harm directly attributable to the partisan gerrymandering of the district in which they reside. Goldston, 361 N.C. at 30, 637 S.E.2d at 879; Mangum, 362 N.C. at 642, 669 S.E.2d at 281; see, e.g., Rucho, 318 F. Supp. 3d at 817; Ohio A. Philip Randolph Inst., 373 F. Supp. 3d at 1063; League of Women Voters of Mich., 373 F. Supp. 3d at 916; Benisek, 348 F. Supp. 3d 493, 517 (D. Md. 2018), vacated on other grounds, 139 S. Ct. 2484 (2019). Moreover, these Individual Plaintiffs have demonstrated, through extensive district-specific evidence, the presence of a legally cognizable injury and, as discussed in great detail below, a means by which the courts of our State can remedy that injury. 15. These Individual Plaintiffs challenge not only the individual districts in which they reside, but also the county groupings as a whole in which they reside. The 296 United States Supreme Court has held that individual voters have standing under the federal Constitution to challenge only their own districts on partisan gerrymandering grounds, Gill, 138 S. Ct. at 1930-31; however, in light of the less stringent standing requirements in our State, and because the manner in which one district is drawn in a county grouping necessarily is tied to the drawing of some, and possibly all, of the other districts within that same grouping, a challenge to the entire county grouping by these Individual Plaintiffs constitutes the necessary “personal stake in the outcome of the controversy” for a plaintiff to have standing in this case. Goldston, 361 N.C. at 30, 637 S.E.2d at 879; see Erfer v. Commonwealth, 794 A.2d 325, 330 (Pa. 2002) (recognizing that a “reapportionment plan acts as an interlocking jigsaw puzzle, each piece reliant upon its neighbors to establish a picture of the whole” and that an “allegation that a litigant’s district was improperly gerrymandered necessarily involves a critique of the plan beyond the borders of his district”), abrogated on other grounds by League of Women Voters of Pa. v. Commonwealth, 178 A.3d 737 (Pa. 2018). 16. On the other hand, several named Individual Plaintiffs do not have standing to challenge either the individual House or Senate District in which they reside because, under Dr. Chen’s analysis, the district in which they would reside is not an outlier—based upon the location of that Individual Plaintiff’s residence—when compared to all of Dr. Chen’s nonpartisan simulated House or Senate maps.17 Therefore, these Individual Plaintiffs have not demonstrated a cognizable injury and a means by which the Court could remedy that injury; however, with respect to the challenged districts in which these 17 These Individual Plaintiffs without standing to challenge either their individual House or Senate district are: Virginia Walters Brien, Leon Charles Schaller, Howard Du Bose, Jr., Deborah Anderson Smith, Alyce Machak, John Balla, John Mark Turner, Ann McCracken, and Mary Ann Peden-Coviello. FOF § E.3.; PX238; PX117. The Court notes that although some Individual Plaintiffs may not have standing to challenge both of their House and Senate districts, they do have standing to challenge at least a district in which they reside. 297 Individual Plaintiffs reside, because the NCDP has standing to bring partisan gerrymandering claims on behalf of its members, the Court concludes that Plaintiffs’ challenges to these districts do not fail for lack of standing. II. THE 2017 PLANS VIOLATE THE NORTH CAROLINA CONSTITUTION’S FREE ELECTIONS CLAUSE 17. Two months ago, in Rucho v. Common Cause, 139 S. Ct. 2484 (2019), the United States Supreme Court considered constitutional challenges to political gerrymandering of Congressional districts in North Carolina and Maryland. 18. The North Carolina Congressional map under consideration by the Supreme Court, adopted by the General Assembly on February 19, 2016, arose in remarkably similar circumstances as the maps under consideration by this trial court, which were adopted August 31, 2017: both the 2016 Congressional map and the 2017 legislative maps were required after a federal court declared existing maps unconstitutional; both were drawn under the direction of many of the same actors working on behalf of the Republicancontrolled General Assembly; both were drawn by Dr. Thomas Hofeller; both were drawn in large part before the General Assembly’s redistricting committee met and approved redistricting criteria; and both, as has been found above with respect to the 2017 legislative maps, were drawn with the intent to maximize partisan advantage and, in fact, achieved their intended partisan effects. 19. In the majority opinion of the Rucho Court, the Justices found the Congressional maps before them to be “highly partisan, by any measure,” id. at 2491, and “blatant examples of partisanship driving districting decisions,” id. at 2505. The majority further reaffirmed that “partisan gerrymanders are incompatible with democratic principles.” Id. at 2506 (citing Ariz. State Legislature v. Ariz. Indep. Redistricting Comm’n, 135 S. Ct. 2652, 2658 (U.S. 2016)). 298 20. Nonetheless, the Supreme Court concluded, in the majority opinion, that “partisan gerrymandering claims present political questions beyond the reach of the federal courts.” Rucho, 139 S. Ct. at 2506-07 (emphasis added). The Court held that the federal courts “have no commission to allocate political power and influence in the absence of a constitutional directive or legal standards to guide us in the exercise of such authority,” id. at 2508, and that the United States Constitution “confines the federal courts to a properly judicial role,” because there is no “no plausible grant of authority in the [United States] Constitution, and no legal standards to limit and direct their decisions,” id. at 2507 (emphasis added). 21. The Supreme Court hastened to add, however, that “our conclusion does not condone excessive partisan gerrymandering” and nor does its conclusion “condemn complaints about districting to echo into a void.” Id. 22. Rather, the Supreme Court held, “[t]he States . . . are actively addressing the issue on a number of fronts,” and “[p]rovisions in state statutes and state constitutions can provide standards and guidance for state courts to apply.” Id. (emphasis added). 23. The North Carolina Constitution, in the Declaration of Rights, Article I, § 10, declares that “[a]ll elections shall be free.” 24. The Free Elections Clause, Article I, § 10, is one of the clauses that makes the North Carolina Constitution more detailed and specific than the federal Constitution in the protection of the rights of its citizens. Corum v. Univ. of N.C. ex rel. Bd. of Gov’rs, 330 N.C. 761, 783, 413 S.E.2d 276, 290 (1992). The federal Constitution contains no similar counterpart to this declaration, although several other states’ constitutions do. 25. The broad language of the Free Elections Clause has not heretofore been extensively interpreted by our appellate courts. However, “it is emphatically the province 299 and duty of the judicial department to say what the law is.” Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803). 26. The North Carolina Supreme Court has long recognized the fundamental role of the will of the people in our democratic government. “Our government is founded on the will of the people. Their will is expressed by the ballot.” People ex rel. Van Bokkelen v. Canaday, 73 N.C. 198, 220 (1875). 27. In particular, the North Carolina Supreme Court has directed that in construing provisions of the Constitution, “we should keep in mind that this is a government of the people, in which the will of the people--the majority--legally expressed, must govern.” State ex rel. Quinn v. Lattimore, 120 N.C. 426, 428, 26 S.E. 638, 638 (1897) (citing N.C. Const. art. I, § 2). 28. Therefore, our Supreme Court continued, because elections should express the will of the people, it follows that “all acts providing for elections, should be liberally construed, that tend to promote a fair election or expression of this popular will.” Id. “[F]air and honest elections are to prevail in this state.” McDonald v. Morrow, 119 N.C. 666, 673, 26 S.E. 132, 134 (1896). 29. Our Supreme Court has elevated this principle to the highest legal standard, noting that it is a “compelling interest” of the State “in having fair, honest elections.” State v. Petersilie, 334 N.C. 169, 184, 432 S.E.2d 832, 840 (1993). As to this there is little room for debate; the Court has recognized that “there is also agreement as to the compelling government interest in ensuring honest and fair elections.” Id. (citing Burson v. Freeman, 504 U.S. 191, 198-99, 112 S. Ct. 1846, 1851-52 (1992)). 30. In giving meaning to the Free Elections Clause, this Court’s construction of the words contained therein must therefore be broad to comport with the following 300 Supreme Court mandate: “We think the object of all elections is to ascertain, fairly and truthfully, the will of the people--the qualified voters.” Hill v. Skinner, 169 N.C. 405, 415, 86 S.E. 351, 356 (1915) (quoting R. R. v. Comrs., 116 N.C. 563, 568, 21 S.E. 205, 207 (1895)). 31. As such, the Court concludes that the meaning of the Free Elections Clause is that elections must be conducted freely and honestly to ascertain, fairly and truthfully, the will of the people. This, the Court concludes, is a fundamental right of the citizens enshrined in our Constitution’s Declaration of Rights, a compelling governmental interest, and a cornerstone of our democratic form of government. 32. The Court now turns to the issue of whether extreme partisan gerrymandering of legislative districts run afoul of the mandate of the Free Elections Clause by depriving citizens of elections that are conducted freely and honestly to ascertain, fairly and truthfully, the will of the people. 33. At its most basic level, partisan gerrymandering is defined as: “the drawing of legislative district lines to subordinate adherents of one political party and entrench a rival party in power.” Ariz. State Legislature, 135 S. Ct. at 2658. 34. The danger of partisan gerrymandering is that it has the potential to violate “the core principle of republican government . . . that the voters should choose their representatives, not the other way around.” Id. at 2677; see also Powell v. McCormack, 395 U.S. 486, 540-41, 89 S. Ct. 1944, 1974 (1969) (“[T]he true principle of a republic is, that the people should choose whom they please to govern them.” (quoting Alexander Hamilton in 2 Debates of the Federal Constitution 257 (J. Elliott ed. 1876))). Moreover, it can represent “an abuse of power that, at its core, evinces a fundamental distrust of voters, serving the self-interest of the political parties at the expense of the public good.” LULAC v. Perry, 548 301 U.S. 399, 456, 126 S. Ct. 2594, 2631 (2006) (Steven, J., concurring in part and dissenting in part) (quotation and citation omitted). 35. Partisan gerrymandering operates through vote dilution—the devaluation of one citizen’s vote as compared to others. A mapmaker draws district lines to “pack” and “crack” voters likely to support the disfavored party. See generally Gill, 138 S. Ct. 1916. The mapmaker packs supermajorities of those voters into a relatively few districts, in numbers far greater than needed for their preferred candidates to prevail. Then the mapmaker cracks the rest across many more districts, spreading them so thin that their candidates will not be able to win. Whether the person is packed or cracked, his vote carries less weight—has less consequence—than it would under a neutrally drawn (nonpartisan) map. See id., 138 S. Ct. at 1935-36 (Kagan, J., concurring). In short, the mapmaker has made some votes count for less, because they are likely to go for the other party. Rucho, 2513-14 (Kagan, J., dissenting). 36. Seen in this light, it is clear to the Court that extreme partisan gerrymandering—namely redistricting plans that entrench politicians in power, that evince a fundamental distrust of voters by serving the self-interest of political parties over the public good, and that dilute and devalue votes of some citizens compared to others—is contrary to the fundamental right of North Carolina citizens to have elections conducted freely and honestly to ascertain, fairly and truthfully, the will of the people. 37. Extreme partisan gerrymandering does not fairly and truthfully ascertain the will of the people. Voters are not freely choosing their representatives. Rather, representatives are choosing their voters. It is not the will of the people that is fairly ascertained through extreme partisan gerrymandering. Rather, it is the will of the map drawers that prevails. 302 38. The Court is further persuaded that the history of the Free Elections Clause comports with the interpretation applied in this case. 39. The Free Elections Clause dates back to the North Carolina Declaration of Rights of 1776. The framers of the North Carolina Declaration of Rights based the Free Elections Clause on a provision of the 1689 English Bill of Rights providing that “election of members of parliament ought to be free.” Bill of Rights 1689, 1 W. & M. c. 2 (Eng.); see John V. Orth, North Carolina Constitutional History, 70 N.C. L. Rev. 1759, 1797-98 (1992). 40. This provision of the 1689 English Bill of Rights grew out of the king’s efforts to manipulate parliamentary elections, including by changing the electorate in different areas to achieve “electoral advantage.” J.R. Jones, The Revolution of 1688 in England 148 (1972). The king’s attempt to maintain control of parliament by manipulating elections led to a revolution, and after dethroning the king, the revolutionaries called for a “free and lawful parliament” as a critical reform. Grey S. De Krey, Restoration and Revolution in Britain: A Political History of the Era of Charles II and the Glorious Revolution 241, 247-48, 250 (2007). 41. A number of states included versions of a free election clause in their early Declarations of Rights, all drawing inspiration from the 1689 English Bill of Rights. The Framers of North Carolina’s Declaration of Rights in turn drew inspiration for North Carolina’s Free Elections Clause from these other states, which included Pennsylvania, Maryland, and Virginia. See Orth, 70 N.C. L. Rev. at 1797-98. 42. Like the 1689 English Bill of Rights, North Carolina’s Free Elections Clause, in conjunction with the companion provision of the State Constitution now found in Article I, § 9 concerning redress of grievances, mandates that elections in North Carolina must be “free from interference or intimidation” by the government, so that all North Carolinians are freely able, through the electoral process, to pursue a “redress of grievances and for 303 amending and strengthening the laws.” John V. Orth & Paul M. Newby, The North Carolina State Constitution 55-57 (2d ed. 2013) (hereinafter “Orth & Newby”). “[T]his pair of sections concerns the application of the principle of popular sovereignty.” Id. at 55. As the North Carolina Supreme Court explained nearly a century ago, the Free Elections Clause reflects that “[o]ur government is founded on the consent of the governed,” and the right to free elections “must be held inviolable to preserve our democracy.” Swaringen v. Poplin, 211 N.C. 700, 191 S.E. 746, 747 (1937). 43. North Carolina has broadened and strengthened the Free Elections Clause since its adoption in 1776 to make these purposes clear. The original clause stated that “elections of members, to serve as Representatives in the General Assembly, ought to be free.” N.C. Declaration of Rights, VI (1776). The next version of the State’s Constitution, adopted in 1868, declared that “[a]ll elections ought to be free,” expanding the principle to include all elections in North Carolina. N.C. Const. art. I, § 10 (1868). In the current State Constitution, adopted in 1971, the Free Elections Clause now mandates that “[a]ll elections shall be free.” N.C. Const. art. I, § 10 (emphasis added). This change was intended to “make [it] clear” that the Free Elections Clause and the other rights secured to the people by the Declaration of Rights “are commands and not mere admonitions” to proper conduct on the part of the government. N.C. State Bar v. DuMont, 304 N.C. 627, 635, 639, 286 S.E.2d 89, 94, 97 (1982) (quoting Report of the N.C. State Constitution Study Comm’n to the N.C. State Bar and the N.C. Bar Ass’n, 75 (1968)). 44. The North Carolina Supreme Court has enforced the Free Elections Clause to invalidate laws that interfere with voters’ ability to freely choose their representatives. In Clark v. Meyland, the North Carolina Supreme Court struck down a law that required voters seeking to change their party affiliation to take an oath supporting the party’s 304 nominees “in the next election and . . . thereafter.” 261 N.C. 140, 141, 134 S.E.2d 168, 169 (1964). The Court held that this attempt to manipulate the outcome of future elections “violate[d] the constitutional provision that elections shall be free.” Id. at 143, 134 S.E.2d at 170. 45. The partisan gerrymandering of the 2017 Plans strikes at the heart of the Free Elections Clause. Using their control of the General Assembly, Legislative Defendants manipulated district boundaries, to the greatest extent possible, to control the outcomes of individual races so as to best ensure their continued control of the legislature. 46. Plaintiffs’ experts demonstrated that the 2017 Plans were designed, specifically and systematically, to maintain Republican majorities in the state House and Senate. Drs. Chen and Mattingly each independently established that the 2017 Plans were gerrymandered to be most resilient in electoral environments where Democrats could win majorities in either chamber under nonpartisan plans. FOF § B.3.a, b. Their analyses establish that it is nearly impossible for Democrats to win majorities in either chamber in any reasonably foreseeable electoral environment. Id. Elections are not free when partisan actors have tainted future elections by specifically and systematically designing the contours of the election districts for partisan purposes and a desire to preserve power. In doing so, partisan actors ensure from the outset that it is nearly impossible for the will of the people—should that will be contrary to the will of the partisan actors drawing the maps—to be expressed through their votes for State legislators. 47. The 2017 Plans also unlawfully seek to predetermine election outcomes in specific districts and county groupings. Drs. Chen and Mattingly each found numerous districts and county groupings that result in safe or relatively safe Republican seats under the enacted plans but would be far more competitive or even Democratic-leaning under nonpartisan plans. In the remaining county groupings, Drs. Chen and Mattingly similarly 305 found that Legislative Defendants placed their thumbs heavily on the scale to favor Republicans. See FOF § C. 48. The harm caused by this manipulation of election outcomes subverts another key purpose of the Free Elections Clause, which, in conjunction with Article I, § 9, is to facilitate the ability of North Carolina citizens to seek a “redress of grievances and for amending and strengthening the law.” Orth & Newby, at 56. Democratic voters in North Carolina cannot meaningfully seek to redress their grievances or amend the laws consistent with their policy preferences when they cannot obtain a majority of the General Assembly. 49. For the foregoing reasons, the Court concludes that Plaintiffs have met their burden of showing, plainly and clearly without any reasonable doubt, that the enacted plans violate the North Carolina Constitution’s guarantee of free elections in Article I, Section 10 of the North Carolina Constitution by demonstrating that Legislative Defendants, with the predominant intent to control and predetermine the outcome of legislative elections for the purpose of retaining partisan power in the General Assembly, manipulated the current district boundaries. And Plaintiffs have met their burden to establish that the manipulation of district boundaries by Legislative Defendants resulted in extreme partisan gerrymandering, subordinating traditional redistricting criteria, so that the resulting maps cracked and packed voters to achieve these partisan objectives. The 2017 Plans, individually and collectively, deprive North Carolina citizens of the right to vote for General Assembly members in elections that are conducted freely and honestly to ascertain, fairly and truthfully, the will of the people. 306 III. THE 2017 PLANS VIOLATE THE NORTH CAROLINA CONSTITUTION’S EQUAL PROTECTION CLAUSE 50. The Equal Protection Clause of the North Carolina Constitution guarantees to all North Carolinians that “[n]o person shall be denied the equal protection of the laws.” N.C. Const., art. I, § 19. 51. Generally, partisan gerrymandering runs afoul of the State’s obligation to provide all persons with equal protection of law because, by seeking to diminish the electoral power of supporters of a disfavored party, a partisan gerrymander treats individuals who support candidates of one political party less favorably than individuals who support candidates of another party. Cf. Lehr v. Robertson, 463 U.S. 248, 265, 103 S. Ct. 2985 (1983) (“The concept of equal justice under law requires the State to govern impartially.”) A. North Carolina’s Equal Protection Clause Provides Greater Protection for Voting Rights Than its Federal Counterpart 52. North Carolina’s Equal Protection Clause provides greater protection for voting rights than federal equal protection provisions. Stephenson v. Bartlett, 355 N.C. 354, 377-81 & n.6, 562 S.E.2d 377, 393-96 & n.6 (2002); Blankenship v. Bartlett, 363 N.C. 518, 522-28, 681 S.E.2d 759, 763-66 (2009). “It is beyond dispute that [North Carolina courts] ha[ve] the authority to construe [the North Carolina Constitution] differently from the construction by the United States Supreme Court of the Federal Constitution, as long as our citizens are thereby accorded no lesser rights than they are guaranteed by the parallel federal provision.” Stephenson, 355 N.C. at 381 n.6, 562 S.E.2d at 395 n.6. North Carolina courts can and do interpret even “identical term[s]” in the State’s Constitution more broadly than their federal counterparts. Northampton Cnty. Drainage Dist. No. One v. Bailey, 326 N.C. 742, 749, 392 S.E.2d 352, 357 (1990). 307 53. The North Carolina Supreme Court has held that North Carolina’s Equal Protection Clause protects “the fundamental right of each North Carolinian to substantially equal voting power.” Stephenson, 355 N.C. at 379, 562 S.E.2d at 394 (emphasis added). “It is well settled in this State that ‘the right to vote on equal terms is a fundamental right.’” Id. at 378, 562 S.E.2d at 393 (quoting Northampton Cnty., 326 N.C. at 747, 392 S.E.2d at 356) (emphasis added). These principles apply with full force in the redistricting context, and because a fundamental right is implicated, strict scrutiny applies. See id. at 377-78, 562 S.E.2d at 393-94. 54. The North Carolina Supreme Court has applied this broader state constitutional protection to invalidate redistricting schemes and other elections laws under Article I, § 19, irrespective of whether they violated federal equal protection guarantees. In Stephenson, the Court held that use of single-member and multi-member districts in a redistricting plan violated Article I, § 19. Id. at 377-81 & n.6, 562 S.E.2d at 393-95 & n.6. The Court explained that, although such a redistricting scheme did not violate the United States Constitution, it restricted the “fundamental right under the State Constitution” to “substantially equal voting power and substantially equal legislative representation.” Id. at 382, 562 S.E.2d at 396. Because the “classification of voters” between single-member and multi-member districts created an “impermissible distinction among similarly situated citizens,” it “necessarily implicate[d] the fundamental right to vote on equal terms,” triggering “strict scrutiny.” Id. at 377-78, 562 S.E.2d at 393-94. 55. In Blankenship, the Court held that Article I, § 19 mandates one-person, one- vote in judicial elections, even though the United States Constitution does not. 363 N.C. at 522-24, 681 S.E.2d at 762-64. The Court stressed that “[t]he right to vote on equal terms in 308 representative elections . . . is a fundamental right” and therefore “triggers heightened scrutiny.” Id. 56. And in Northampton County, the Court applied strict scrutiny to invalidate certain rules related to voting for drainage districts, holding that the rules at issue deprived one county’s residents of the “fundamental right” to “vote on equal terms” with residents of a neighboring county. 326 N.C. at 747, 392 S.E.2d at 356. 57. Although the North Carolina Constitution provides greater protection for voting rights than the federal Equal Protection Clause, our courts use the same test as federal courts in evaluating the constitutionality of challenged classifications under an equal protection analysis. Duggins v. N.C. State Bd. of Certified Pub. Accountant Exam’rs, 294 N.C. 120, 131, 240 S.E.2d 406, 413 (1978); Richardson v. N.C. Dep’t of Corr., 345 N.C. 128, 134, 478 S.E.2d 501, 505 (1996). 58. Generally, this test has three parts: (1) intent, (2) effects, and (3) causation. First, the plaintiffs challenging a districting plan must prove that state officials’ “predominant purpose” in drawing district lines was to “entrench [their party] in power” by diluting the votes of citizens favoring their rival. Ariz. State Legis., 135 S. Ct. at 2658. Second, the plaintiffs must establish that the lines drawn in fact have the intended effect by “substantially” diluting their votes. Rucho, 318 F. Supp. 3d at 861. Finally, if the plaintiffs make those showings, the State must provide a legitimate, non-partisan justification (i.e., that the impermissible intent did not cause the effect) to preserve its map. Rucho, 139 S. Ct. at 2516 (Kagan, J., dissenting). B. The 2017 Plans Were Created with the Intent to Discriminate Against Plaintiffs and Other Democratic Voters 59. To establish a discriminatory purpose or intent, a plaintiff need not show that the discriminatory purpose is “express or appear[s] on the face of the statute.” 309 Washington v. Davis, 426 U.S. 229, 241, 96 S. Ct. 2040, 2048 (1976). Rather, “an invidious discriminatory purpose may often be inferred from the totality of the relevant facts.” Id. at 242, 96 S. Ct. at 2048. 60. The United States Supreme Court has recognized that there are certain purposes for which a state redistricting body may take into account political data or partisan considerations in drawing district lines. For example, a legislature may, under appropriate circumstances, draw district lines to avoid the pairing of incumbents. Karcher v. Daggett, 462 U.S. 725, 740, 103 S. Ct. 2653, 2663 (1983). Likewise, a state redistricting body does not violate the United States Constitution by seeking “to create a districting plan that would achieve a rough approximation of the statewide political strengths of the Democratic and Republican Parties.” Gaffney v. Cummings, 412 U.S. 735, 752, 93 S. Ct. 2321, 2331 (1973). And a redistricting body may draw district lines to respect municipal boundaries or maintain communities of interest. Abrams v. Johnson, 521 U.S. 74, 100, 117 S. Ct. 1925, 1940 (1997). Accordingly, a plaintiff in a partisan gerrymandering case cannot satisfy the discriminatory intent requirement simply by proving that the redistricting body intended to rely on political data or to take into account political or partisan considerations. Rather, the plaintiff must show that the redistricting body intended to apply partisan classifications or deprive citizens of the right to vote on equal terms “in an invidious manner or in a way unrelated to any legitimate legislative objective.” Vieth, 541 U.S. at 307, 124 S. Ct. at 1793 (Kennedy, J., concurring in the judgment). 61. “Blatant examples of partisanship driving districting decisions,” Rucho, 139 S. Ct. at 2505, are unrelated to any legitimate legislative objective. Indeed, partisan gerrymanders are incompatible with democratic principles. Vieth, 541 U.S. at 292, 124 S. 310 Ct. at 1785 (plurality opinion); id., at 316, 124 S. Ct. at 1798 (Kennedy, J., concurring in judgment); Ariz. State Legislature, 135 S. Ct. at 2658. 62. Partisan gerrymanders are also contrary to the compelling governmental interests established by the North Carolina Constitution “in having fair, honest elections,” see Petersilie, 334 N.C. at 182, 432 S.E.2d at 840, where the “will of the people” is ascertained “fairly and truthfully,” Skinner, 169 N.C. at 415, 86 S.E. at 356. Partisan gerrymandering contravenes the legitimate purposes of redistricting because it is intended to hamper, rather than to “achiev[e,] . . . fair and effective representation for all citizens.” Reynolds v. Sims, 377 U.S. 533, 565-66, 84 S. Ct. 1362, 1383 (1964). 63. Moreover, the intentional “classification of voters” based on partisanship in order to pack and crack them into districts is an “impermissible distinction among similarly situated citizens” aimed at denying equal voting power. See Stephenson, 355 N.C. at 377-78, 562 S.E.2d at 393-94 (“The classification of voters into both single-member and multimember districts within plaintiffs’ proposed remedial plans necessarily implicates the fundamental right to vote on equal terms . . . These classifications, as used within plaintiffs’ proposed remedial plans, create an impermissible distinction among similarly situated citizens based upon the population density of the area in which they reside.”). “A state may not dilute the strength of a person’s vote to give weight to other interests.” Texfi Indus., Inc. v. Fayetteville, 301 N.C. 1, 13, 269 S.E.2d 142, 150 (1980) (citing Evans v. Cornman, 398 U.S. 419, 90 S. Ct. 1752 (1970)). 64. Legislative Defendants openly admitted that they used prior election results to draw districts to benefit Republicans in both 2011 and 2017. FOF § B.1. Dr. Hofeller’s own files provide even more direct evidence that the predominant goal of the 2017 Plans was to maximize Republicans’ political advantage by drawing Democratic voters into 311 districts where their votes would be diluted, and in many cases where their votes would not matter. FOF § B.2. 65. The analysis and conclusions of Plaintiffs’ experts confirm the point. Dr. Chen’s analysis confirms that the General Assembly intentionally subordinated traditional districting principles to maximize Republican advantage. FOF § B.3.a. Dr. Mattingly’s analysis confirms that the enacted plans’ extreme partisan bias could only have been intentional. FOF § B.3.b. Dr. Pegden’s sensitivity analysis shows that the enacted plans are more carefully crafted to favor Republicans than 99.999% of all possible plans of North Carolina meeting the same nonpartisan criteria laid out in the Adopted Criteria. FOF § B.3.c. And Dr. Cooper demonstrated, by analyzing the district boundaries within each relevant county grouping, that the enacted plans intentionally and systematically pack and crack Democratic voters. FOF § C. 66. As such, the Court concludes that, in drawing the 2017 House and Senate Maps, Legislative Defendants acted with the intent, unrelated to any legitimate legislative objective, to classify voters and deprive citizens of the right to vote on equal terms. Legislative Defendants did so by subordinating Democratic voters to Legislative Defendants’ partisan goals—in other words, by devaluing their vote as compared to the votes of Republican voters with the aim of entrenching the Republican Party in power—and the Court concludes that this intent was the predominant purpose of drawing the district lines in individual districts and statewide. C. The 2017 Plans Deprive Plaintiffs and Other Democratic Voters of Substantially Equal Voting Power and the Right to Vote on Equal Terms 67. The United States Supreme Court has recognized that the injury associated with partisan gerrymandering “arises from the particular composition of the voter’s own district, which causes his vote – having been packed or cracked – to carry less weight than 312 it would carry in another hypothetical district.” Gill, 138 S. Ct. at 1931. It is the “voter’s placement in a ‘cracked’ or ‘packed’ district” that causes injury. Id. 68. Therefore, to prevail, Plaintiffs must also establish that the enacted legislative districts actually had the effect of discriminating against—or subordinating— voters who support candidates of the Democratic Party by virtue of district lines that crack or pack those voters, thereby depriving them of substantially equal voting power in an effort to entrench the Republican Party in power, in violation of Article I, § 19. 69. The manipulation of district boundaries in the enacted plans prevents Democratic voters from obtaining a majority in the House or the Senate even in election environments where Democrats would obtain a majority under virtually any nonpartisan map. Dr. Chen and Dr. Mattingly each independently found that the effects of the gerrymanders are most extreme in circumstances where Democrats could win majorities in one or both chambers under nonpartisan plans. FOF § B.3.a, b. There is nothing “equal” about the “voting power” of Democratic voters when they have a vastly less realistic chance of winning a majority in either chamber under the enacted plans. “The right to vote is the right to participate in the decision-making process of government.” Texfi Indus., 301 N.C. at 13, 269 S.E.2d at 150. Democratic voters are significantly hindered from meaningfully participating in the decision-making process of government when the maps are drawn to systematically prevent Democrats from obtaining a majority in either chamber of the General Assembly. 70. Beyond the issue of majority control, Dr. Chen and Dr. Mattingly also concluded that the gerrymanders deprive Democratic voters of multiple seats in the House and the Senate across a variety of electoral environments. FOF § B.3.a, b. The 2017 Plans achieve these effects by cracking and packing Democratic voters in districts contained within county grouping after county grouping. FOF § C. This packing and cracking 313 diminishes the “voting power” of Democratic voters in these districts and groupings; packing dilutes the votes of Democratic voters such that their votes, when compared to the votes of Republican voters, are substantially less likely to ultimately matter in deciding the election results, and the entire purpose of cracking likeminded voters across multiple districts is so they do not have sufficient “voting power” to join together and elect a candidate of their choice. 71. Moreover, although not necessary to establish Plaintiffs’ equal protection claim, the Court similarly concludes that the 2017 Plans not only deprive Democratic voters of equal voting power in terms of electoral outcomes, but also deprive them of substantially equal legislative representation. See Stephenson, 355 N.C. at 379, 562 S.E.2d at 394. Partisan gerrymandering insulates legislators from popular will and renders them unresponsive to portions of their constituencies. See Reynolds, 377 U.S. at 565 (“Since legislatures are responsible for enacting laws by which all citizens are to be governed, they should be bodies which are collectively responsible to the popular will.”). When a district is created solely to effectuate the interests of one group, the elected official from that district is “more likely to believe that their primary obligation is to represent only the members of that group, rather than their constituency as a whole.” See Shaw I, 509 U.S. at 648 (in the context of racial gerrymandering). 72. Just as the “political reality” is that “legislators are much more inclined to listen to and support a constituent than an outsider,” Stephenson, 355 N.C. at 380, 562 S.E.2d at 395, the reality is that legislators are far more likely to represent the interests and policy preferences of voters of the same party. Legislative Defendants’ own expert, Dr. Brunell, agreed that “a voter whose candidate of choice loses will on average be less wellrepresented than a voter who voted for the winning candidate.” Tr. 2370:22-2371:2. 314 D. The 2017 Plans Cannot be Justified by any Legitimate Governmental Interest 73. Once a plaintiff establishes a prima facia case that boundaries of legislative districts violate the Equal Protection Clause of the North Carolina Constitution, which Plaintiffs have done in this case by establishing a discriminatory intent and a discriminatory effect, the burden shifts to Legislative Defendants to prove that a legitimate state interest or other neutral factor justified such discrimination. 74. Legislative Defendants offer limited neutral justifications for the enacted maps. They contend that the plans “satisfy the equal-population rule and the strict countygrouping and transversal rules of Article II of the State Constitution” and that “[t]he districts were far more compact than in 2011 or prior years; they split fewer VTDs than in 2011 or prior years; they . . . minimized incumbency pairings; and they preserved core constituency-incumbent relations.” Leg. Defs.’ Post-Trial Brief at p. 28. 75. While all of this may be true, these neutral justifications do not provide a sufficient justification for the substantial evidence, proffered by Plaintiffs and given substantial weight by this Court, showing that Legislative Defendants’ predominant intent was to classify voters and deprive citizens of the right to vote on equal terms and substantially equally voting power. Legislative Defendants did so by subordinating Democratic voters to Legislative Defendants’ partisan goals—in other words, by devaluing their vote as compared to the votes of Republican voters with the aim of entrenching the Republican Party in power—and the Court concludes that this intent was the predominant purpose of drawing the district lines in individual districts and statewide. 76. Nor do these justifications address the substantial evidence that the neutral criteria offered by Legislative Defendants, and indeed all other neutral objectives of the Adopted Criteria, were subordinated by Legislative Defendants in the map drawing process 315 in order to attain the discriminatory effects of the resulting extreme partisan gerrymandering. 77. Because the 2017 Plans impermissibly interfere with the exercise of the fundamental right to vote, strict scrutiny applies. See Stephenson, 355 N.C. at 377-78, 562 S.E.2d at 393. Legislative Defendants have not established that the 2017 Plans are narrowly tailored to advance a compelling governmental interest. See Id. Advantaging a particular political party or discriminating against voters based on how they vote for the purposes of entrenching a political party’s power is not a compelling government interest. 78. For the foregoing reasons, the Court concludes that Plaintiffs have met their burden of showing, plainly and clearly without any reasonable doubt, that the enacted plans violate the North Carolina Constitution’s guarantee of equal protection in Article I, Section 19 of the North Carolina Constitution by demonstrating that (1) Legislative Defendants acted with the intent, unrelated to any legitimate legislative objective, to classify voters and deprive citizens of the right to vote on equal terms by subordinating Democratic voters to Legislative Defendants’ partisan goals—in other words, by devaluing their vote as compared to the votes of Republican voters with the aim of entrenching the Republican Party in power—and this intent was the predominant purpose of drawing the district lines in individual districts and statewide; (2) that the legislative maps drawn by Legislative Defendants with this intent had the effect of depriving disfavored voters in North Carolina of substantially equal voting power and the right to vote on equal terms, as well as substantially equal legislative representation; and (3) Legislative Defendants have not provided a neutral justification or a compelling governmental rationale for their actions. 79. Specifically, voters in specific districts in the following county groupings are unlawfully deprived of equal protection under the law in violation of the North Carolina Constitution. In these districts, Plaintiffs have demonstrated through Dr. Chen, Dr. 316 Mattingly, and Dr. Cooper, whose expert testimony has been given substantial weight by the Court, that Democratic voters were packed or cracked into extreme gerrymandered districts so that the effect upon these voters was to deprive them of substantially equal voting power and the right to vote on equal terms, as well as substantially equal legislative representation. County groupings including these districts are as follows: Senate Districts: FOF § C.1.a (Mecklenburg); C.1.b (Franklin-Wake); C.1.c (NashJohnston-Harnett-Lee-Sampson-Duplin); C.1.d. (Guilford-AlamanceRandolph); C.1.e (Davie-Forsyth); C.1.g (Buncombe-HendersonTransylvania); House Districts: FOF § C.2.a (Robeson-Columbus-Pender); C.2.b (Cumberland); C.2.d (Franklin-Nash); C.2.e (Pitt-Lenoir); C.2.f (Guilford); C.2.g (DavieRowan-Cabarrus-Stanly-Montgomery-Richmond); C.2.h (Yadkin-Forsyth); C.2.i (Mecklenburg); C.2.k (New Hanover-Brunswick); C.2.l (Duplin-Onslow); C.2.m (Anson-Union); C.2.n. (Alamance); C.2.o (Cleveland-Gaston); C.2.p (Buncombe). In the remaining county groupings challenged by Plaintiffs, Drs. Chen and Mattingly similarly found that Legislative Defendants placed their thumbs heavily on the scale to favor Republicans. See FOF § C. IV. THE 2017 PLANS VIOLATE THE NORTH CAROLINA CONSTITUTION’S FREEDOM OF SPEECH AND FREEDOM OF ASSEMBLY CLAUSES 80. The Freedom of Speech Clause in Article I, § 14 of the North Carolina Constitution provides that “[f]reedom of speech and of the press are two of the great bulwarks of liberty and therefore shall never be restrained.” The Freedom of Assembly Clause in Article I, § 12 provides, in relevant part, that “[t]he people have a right to 317 assemble together to consult for their common good, to instruct their representatives, and to apply to the General Assembly for redress of grievances.” 81. The 2017 Plans violate the North Carolina Constitution’s guarantees of free speech and assembly, irrespective of whether the plans violate the U.S. Constitution. See Michigan v. Long, 463 U.S. 1032, 103 S. Ct. 3469 (1983). A. North Carolina’s Constitution Protects the Rights of Free Speech and Assembly Independently from the Federal Constitution 82. “[I]n construing provisions of the Constitution of North Carolina,” the North Carolina Supreme Court “is not bound by opinions of the Supreme Court of the United States construing even identical provisions in the Constitution of the United States.” State v. Hicks, 333 N.C. 467, 483, 428 S.E.2d 167, 176 (1993). While the North Carolina Supreme Court gives “great weight” to decisions of the United States Supreme Court that interpret corresponding provisions in the federal constitution, Hicks, 333 N.C. at 484, 428 S.E.2d at 176, only North Carolina courts can “answer[] with finality” questions of North Carolina constitutional law, State v. Arrington, 311 N.C. 633, 643, 319 S.E.2d 254, 260 (1984). North Carolina courts thus “have the authority to construe [the State’s] own constitution differently from the construction by the United States Supreme Court of the Federal Constitution, as long as [its] citizens are thereby accorded no lesser rights than they are guaranteed by the parallel federal provision.” State v. Carter, 322 N.C. 709, 713, 370 S.E.2d 553, 555 (1988). 83. The North Carolina Supreme Court has held that the North Carolina Constitution’s Free Speech Clause provides broader rights than does federal law. In particular, the Court has held that the North Carolina Constitution affords a direct cause of action for damages against government officers in their official capacity for speech violations, even though federal law does not. Corum, 330 N.C. at 783, 413 S.E.2d at 290. 318 Noting that “[o]ur Constitution is more detailed and specific than the federal Constitution in the protection of the rights of its citizens,” the Court explained that the North Carolina courts “give our Constitution a liberal interpretation in favor of its citizens with respect to those provisions which were designed to safeguard the liberty and security of the citizens in regard to both person and property.” Id. Indeed, in recognizing a direct cause of action under the State Constitution, the Court expressly relied on the lack of a federal remedy, which left plaintiffs with “no other remedy . . . for alleged violations of his constitutional freedom of speech rights.” Id. 84. Similarly, in Evans v. Cowan, the Court of Appeals reversed a trial court that had dismissed a claim under Article I, § 14, on the erroneous ground that it was res judicata based on a prior dismissal of the plaintiff’s claim under the federal First Amendment. 122 N.C. App. 181, 183-84, 468 S.E.2d 575, 577-78, aff’d, 477 S.E.2d 926 (N.C. 1996). While “both the North Carolina Constitution and the United States Constitution contain similar provisions proclaiming certain principles of liberty,” North Carolina courts “are not bound by the opinions of the federal courts.” Id. at 183-84, 468 S.E.2d at 577. “[A]n independent determination of plaintiff’s constitutional rights under the state constitution [was] required, and the state courts reserve the right to grant relief under the state constitution in circumstances under which no relief might be granted under the federal constitution.” Id. at 184, 468 S.E.2d at 577 (citation and internal quotations marks omitted); see also McLaughlin v. Bailey, 240 N.C. App. 159, 172, 771 S.E.2d 570, 579-80 (2015), aff’d, 781 S.E.2d 23 (N.C. 2016); see also Lenzer v. Flaherty, 106 N.C. App. 496, 418 S.E.2d 276 (1992). 85. In the context of partisan gerrymandering, it is especially important that North Carolina courts give independent force to North Carolina’s constitutional protections. 319 The United States Supreme Court recently held that federal courts applying the federal constitution have no power to adjudicate claims of partisan gerrymandering. Rucho, 139 S. Ct. 2484. That ruling does not mean that partisan gerrymandering complies with the constitution; it means that federal courts have no power to decide whether the practice complies with the constitution. “Having no other remedy,” the North Carolina Constitution “guarantees [P]laintiff[s] a direct action under the State Constitution for alleged violations of [their] constitutional freedom of speech rights.” Corum, 330 N.C. at 783, 413 S.E.2d at 290. B. Voting, Banding Together in a Political Party, and Spending on Elections Are Protected Expression and Association 86. Voting for the candidate of one’s choice and associating with the political party of one’s choice are core means of political expression protected by the North Carolina Constitution’s Freedom of Speech and Freedom of Assembly Clauses. The 2017 Plans burden that protected expression and thus are subject to scrutiny under those clauses. 87. Voting provides citizens a direct means of expressing support for a candidate and his views. See Buckley v. Valeo, 424 U.S. 1, 21, 96 S. Ct. 612, 635 (1976). Indeed, if donating money to a candidate constitutes a form of protected speech, then voting for that same candidate necessarily does as well. “There is no right more basic in our democracy than the right to participate in electing our political leaders”—including, of course, the right to “vote.” McCutcheon v. FEC, 572 U.S. 185, 191, 134 S. Ct. 1434, 1440 (2014) (plurality op.). “[P]olitical belief and association constitute the core of those activities protected by the First Amendment.” Elrod v. Burns, 427 U.S. 347, 356, 96 S. Ct. 2673, 2681 (1976). 88. Plaintiffs’ expression is no less protected “merely because it involves the ‘act’” of casting a ballot. State v. Bishop, 368 N.C. 869, 874, 787 S.E.2d 814, 818 (2016). “[M]uch 320 speech requires an ‘act’ of some variety—whether putting ink to paper or paint to canvas, or hoisting a picket sign, or donning a message-bearing jacket.” Id. Voting, like donating money to a candidate or signing a petition for a referendum, constitutes “expressive activity” that “express[es] [a] view” about the State’s laws and policies. Winborne v. Easley, 136 N.C. App. 191, 198, 523 S.E.2d 149, 153 (1999); Doe v. Reed, 561 U.S. 186, 195, 130 S. Ct. 2811, 2817 (2010). Voting’s expressive force is not diminished by the fact that it “is a legally operative legislative act.” Id. at 195; see also Nev. Comm’n on Ethics v. Carrigan, 564 U.S. 117, 134, 131 S. Ct. 2343, 2355 (2011) (Alito, J., concurring) (“[T]he act of voting is not drained of its expressive content when the vote has a legal effect.”). Having “cho[sen] to tap the energy and the legitimizing power of the democratic process,” the government “must accord the participants in that process the First Amendment rights that attach to their roles.” Republican Party of Minn. v. White, 536 U.S. 765, 788, 122 S. Ct. 2528, 2541 (2002) (quotation omitted). The ballots cast by Plaintiffs and other Democratic voters to elect candidates to the North Carolina General Assembly are protected by North Carolina’s Freedom of Speech Clause. 89. Expression aside, the Freedom of Assembly Clause independently protects Plaintiffs’ voting and their association with the Democratic Party. The Freedom of Assembly Clause—part of North Carolina’s original 1776 Declaration of Rights—protects the right of the people “to assemble together to consult for their common good, to instruct their representatives, and to apply to the General Assembly for redress of grievances.” N.C. Const. art. I, § 12; see N.C. Const. art. I, § 18 (1776). In North Carolina, the right to assembly encompasses the right of association. Feltman v. City of Wilson, 238 N.C. App. 246, 253, 767 S.E.2d 615, 620 (2014). 90. Just as voting is a form of protected expression, banding together with likeminded citizens in a political party is a form of protected association. “[C]itizens form 321 parties to express their political beliefs and to assist others in casting votes in alignment with those beliefs.” Libertarian Party of N.C. v. State, 365 N.C. 41, 49, 707 S.E.2d 199, 20405 (2011). “[F]or elections to express the popular will, the right to assemble and consult for the common good must be guaranteed.” John V. Orth, The North Carolina State Constitution 48 (1995). 91. A final form of relevant protected expression involves the expenditure of funds in support of candidates. It is now well-settled that “political contributions and expenditures” constitute “expressive activity” that are constitutionally protected. Winborne, 136 N.C. App. at 198, 523 S.E.2d at 153-54. C. The 2017 Plans Burden Protected Expression and Association 92. The 2017 Plans are subject to strict scrutiny because they burden Plaintiffs’ and Democratic voters’ political expression and association. 1. 93. The 2017 Plans Burden Protected Expression Based on Viewpoint by Making Democratic Votes Less Effective It is “axiomatic” that the government may not infringe on protected activity based on the individual’s viewpoint. Rosenberger v. Rector & Visitors of Univ. of Va., 515 U.S. 819, 828, 115 S. Ct. 2510, 2516 (1995). “The government must abstain from regulating speech when the specific motivating ideology or the opinion or perspective of the speaker is the rationale for the restriction.” Id. at 829, 115 S. Ct. at 2516. The guarantee of free expression “stands against attempts to disfavor certain subjects or viewpoints.” Citizens United v. FEC, 558 U.S. 310, 340, 130 S. Ct. 876, 898 (2010). 94. Viewpoint discrimination is most insidious where the targeted speech is political. “[I]n the context of political speech, . . . [b]oth history and logic” demonstrate the perils of permitting the government to “identif[y] certain preferred speakers” while burdening the speech of “disfavored speakers.” Id. at 340-41, 130 S. Ct. at 899. The 322 government may not burden the “speech of some elements of our society in order to enhance the relative voice of others” in electing officials. McCutcheon, 572 U.S. at 207, 134 S. Ct. at 1450; see also Winborne, 136 N.C. App. at 198, 523 S.E.2d at 154 (“political speech” has “such a high status” that free speech protections have their “fullest and most urgent application” in this context (quotations marks omitted)). 95. Here, Legislative Defendants “identified[] certain preferred speakers” (Republican voters), while targeting certain “disfavored speakers” (Plaintiffs and other Democratic voters) for “disfavored treatment” because of disagreement with the views they express when they vote. Citizens United, 558 U.S. at 340-41, 130 S. Ct. at 899; see Sorrell v. IMS Health Inc., 564 U.S. 552, 565, 131 S. Ct. 2653, 2663 (2011). Legislative Defendants analyzed the voting histories of every VTD in North Carolina, identified VTDs that favor Democratic candidates, and then singled out the voters in those VTDs for disfavored treatment by packing and cracking them into districts with the aim of diluting their votes and, in the case of cracked districts, ensuring that these voters are significantly less likely, in comparison to Republican voters, to be able to elect a candidate who shares their views. 96. The fact that Democratic voters can still cast ballots under gerrymandered maps changes nothing. The government unconstitutionally burdens speech where it renders disfavored speech less effective, even if it does not ban such speech outright. The government may not restrict a citizen’s “ability to effectively exercise” their free speech rights. Heritage Vill. Church & Missionary Fellowship, Inc. v. State, 40 N.C. App. 429, 451, 253 S.E.2d 473, 486 (1979), aff’d, 299 N.C. 399, 263 S.E.2d 726 (1980). “It is thus no answer to say that petitioners can still be ‘seen and heard’” if the burdens placed on their speech “have effectively stifled petitioners’ message.” McCullen v. Coakley, 573 U.S. 464, 489-90, 134 S. Ct. 2518, 2537 (2014). 323 97. In McCullen, for instance, the United States Supreme Court invalidated a law that imposed a buffer zone around abortion clinics because the law “compromise[d] [the] ability” of the plaintiffs to “initiate the close, personal conversations that they view as essential” to effectively communicate their message. 573 U.S. at 487, 134 S. Ct. at 2535. And in Sorrell, the United States Supreme Court invalidated on viewpoint discrimination grounds a state law that burdened drug manufacturers by denying them information that made their marketing more effective. 564 U.S. at 580, 131 S. Ct. at 2672. The Court stressed that “the distinction between laws burdening speech is but a matter of degree and the Government’s content-based burdens must satisfy the same rigorous scrutiny as its content-based bans.” Id. at 555-56, 131 S. Ct. at 2664 (quotation marks omitted). 98. These principles apply equally to burdens on political expression. In Davis v. FEC, the United States Supreme Court struck down a law that disfavored candidates who self-financed their campaigns. 554 U.S. 724, 128 S. Ct. 2759 (2008). The law in question did not limit how much money self-financing candidates could spend, but it still unconstitutionally “diminishe[d] the effectiveness of [their] speech.” Id. at 736, 128 S. Ct. at 2770. The Court held the same in Ariz. Free Enterprise Club’s Freedom Club PAC v. Bennett, where it invalidated a public-matching scheme because it rendered the money spent by privately financed candidates “less effective.” 564 U.S. 721, 747, 131 S. Ct. 2806, 2824 (2011); see also Randall v. Sorrell, 548 U.S. 230, 248-49, 126 S. Ct. 2479, 2492 (2006) (invalidating limit on campaign donations that made such donations less “effective”). 99. North Carolina courts have recognized “several paths” leading to the conclusion that laws burdening protected expression are impermissibly discriminatory and thus “subject to strict scrutiny.” State v. Bishop, 368 N.C. 869, 875, 787 S.E.2d 814, 819 (2016). A finding of discrimination “can find support in the plain text of a statute, or the 324 animating impulse behind it, or the lack of any plausible explanation besides distaste for the subject matter or message.” Id. The 2017 Plans thus need not explicitly mention any particular viewpoint to be impermissibly discriminatory. See, e.g., Reed v. Town of Gilbert, 135 S. Ct. 2218, 2227 (2015). 100. Here, all paths lead to the same conclusion: the 2017 Plans reflect viewpoint discrimination against Plaintiffs and other Democratic voters that render their protected political expression less effective. 101. Overwhelming, unrebutted evidence establishes that the 2017 Plans were laced with viewpoint-driven intent. Legislative Defendants directed Dr. Hofeller to assign voters to districts using “election data” reflecting the contents of their prior votes for Democratic or Republican candidates, and Dr. Hofeller abided, using a color-coded shading system to track voters based on their partisan preferences and voting histories. FOF § C. Within county groups, Dr. Hofeller placed Democratic voters in this district or that one based solely on their political views. If this direct evidence left any doubt, the expert testimony showed that the mapmaker crafted the plans with partisanship as the predominant (if not sole) focus. Dr. Cooper in particular illustrated the intentional packing and cracking of specific Democratic voters and communities. FOF § C. 102. This sorting of Plaintiffs and other Democratic voters based on disfavor for their political views has burdened their speech by making their votes less effective. Many Plaintiffs and other Democratic voters live in districts where their votes are guaranteed to be less effective—either because the districts are packed such that Democratic candidates will win by astronomical margins or because the Democratic voters are cracked into seats that are safely Republican. Plaintiff Derrick Miller testified that he is one such voter: with the Wilmington Notch having been placed in Senate District 8, it is “impossible for [he] and Democratic neighbors to elect a Democrat, a candidate of our choice.” Tr. 205:13-15. 325 Plaintiff Joshua Brown similarly testified that the mapmaker’s placing High Point’s Democrats into Senate District 26 “clearly dilutes the ability of Democrats to even attempt to run a fair race.” Tr. 833:20-21. 103. By packing and cracking Democratic voters to make it harder for them to translate votes into legislative seats, the 2017 Plans “single[] out a subset of messages for disfavor based on the views expressed.” Matal v. Tam, 137 S. Ct. 1744, 1766 (2017) (Kennedy, J., concurring). “This is the essence of viewpoint discrimination.” Id. 104. Even were Legislative Defendants permitted to consider voters’ political beliefs when drawing district maps, the 2017 Plans would still be unlawful. In arenas where the government is allowed (or even required) to consider the content or viewpoint of expression that it regulates, it is still forbidden from intentionally elevating one viewpoint over the other. In Board of Education v. Pico, for example, the Supreme Court recognized that, while local school boards “possess significant discretion to determine the content of their school libraries,” their discretion may “not be exercised in a narrowly partisan or political manner.” 457 U.S. 853, 870, 102 S. Ct. 2799, 2810 (1982). As the Court observed, “[i]f a Democratic school board, motivated by party affiliation, ordered the removal of all books written by or in favor of Republicans, few would doubt that the order violated the constitutional rights of the students denied access to those books.” Id. at 870-71, 102 S. Ct. at 2810. So too here. Legislative Defendants did not simply look at partisan data to satisfy their curiosity. They drew the 2017 Plans in a way that deliberately minimized the effectiveness of the votes of citizens with whom they disagree. 2. 105. The 2017 Plans Burden Plaintiffs’ Ability to Associate The 2017 Plans independently violate Article I, § 12 by burdening the ability of the NCDP, Common Cause, and Democratic voters to associate effectively. 326 106. The 2017 Plans severely burden—if not outright preclude—the ability of the NCDP, Common Cause, and Democratic voters “to instruct their representatives, and to apply to the General Assembly for redress of grievances.” N.C. Const. art. I, § 12. Democratic voters who live in cracked districts have little to no ability to instruct their representatives or obtain redress from their representatives on issues important to those voters. FOF § E.3. And as a result of the gerrymanders, Democratic voters across the state, as well as the NCDP, will be unlikely to obtain redress from “the General Assembly” on important policy issues, because they will unlikely be able to obtain Democratic majorities in the General Assembly. Id. Common Cause likewise cannot instruct representatives or obtain redress on the issues central to its mission due to the gerrymanders. FOF § E.2. The 2017 Plans “burden[] the ability of like-minded people across the State to affiliate in a political party and carry out [their] activities and objects.” Gill, 138 S. Ct. at 1939 (Kagan J., concurring). 107. The 2017 Plans separately violate NCDP’s associational rights by “debilitat[ing] [the] party” and “weaken[ing] its ability to carry out its core functions and purposes.” Id. Due to the unfair playing field created by the 2017 Plans, the NCDP “face[s] difficulties fundraising, registering voters, attracting volunteers, generating support from independents, and recruiting candidates to run for office.” Id. at 1938; see FOF § E.1. And, even when overcoming these difficulties through extraordinary efforts, fundraising and enthusiasm, as was evidenced in the 2018 election cycle, the 2017 Plans nonetheless debilitate the NCDP and weaken its ability to translate its effort, funds and enthusiasm into a meaningful opportunity to gain majority control of the General Assembly. FOF § E.1. 327 3. 108. The 2017 Plans Burden the NCDP’s Expression Through Financial Support for Candidates The 2017 Plans independently violate the NCDP’s free expression and assembly rights under Article I, §§ 12 and 14 by burdening their campaign donations and expenditures. The NCDP must spend more money than it would need to under nonpartisan plans, both statewide and in individual races, and the money that the NCDP spends is less effective than it would be under nondiscriminatory maps. FOF § E.1. The NCDP’s political opponent, the North Carolina Republican Party, faces no such burdens. 109. The operation of the 2017 Plans is analogous to the laws struck down in Davis and Bennett in this regard. Those laws did not preclude or limit any campaign expenditures, but were still held unconstitutional because they “diminishe[d] the effectiveness” of the expenditures of some candidates. See Bennett, 564 U.S. at 736, 131 S. Ct. at 2818 (quoting Davis, 554 U.S. at 736, 128 S. Ct. at 2770). The same is true here. The 2017 Plans create “a political hydra” that forces the NCDP to drain and divert resources across the State merely to avoid being relegated to a super-minority. Id. at 738. D. The 2017 Plans Fail Strict Scrutiny—and Indeed Any Scrutiny 110. Because the 2017 Plans discriminate against Plaintiffs and other Democratic voters based on their protected expression and association, the burden shifts to the Legislative Defendants to establish that the 2017 Plans were narrowly tailored to achieve a compelling government interest. See Petersilie, 334 N.C. at 206, 432 S.E.2d at 853-54 (Mitchell, J., dissenting). 111. As noted above, COL § III.D., Legislative Defendants have offered no credible justification for their partisan discrimination. Nor could they have. Discriminating against citizens based on their political beliefs does not serve any legitimate government interest. 328 E. The 2017 Plans Impermissibly Retaliate Against Voters Based on Their Exercise of Protected Speech 112. The 2017 Plans violate the Freedom of Speech and Assembly Clauses for an independent reason. In addition to forbidding discrimination, those clauses also bar retaliation based on protected speech and expression. See McLaughlin, 240 N.C. App. at 172, 771 S.E.2d at 579-80. Courts carefully guard against retaliation by the party in power. See Elrod, 427 U.S. at 356, 96 S. Ct. at 2681; Branti v. Finkel, 445 U.S. 507, 100 S. Ct. 1287 (1980); Rutan v. Republican Party of Ill., 497 U.S. 62, 110 S. Ct. 2729 (1990). When patronage or retaliation restrains citizens’ freedoms of belief and association, it is “at war with the deeper traditions of democracy embodied in the First Amendment.” Elrod, 427 U.S. at 357, 96 S. Ct. at 2682 (quotation marks omitted). 113. To establish a violation of the North Carolina Constitution under a retaliation theory, Plaintiffs must show, in addition to their engagement in protected expression or association, that (1) the 2017 Plans take adverse action against them, (2) the 2017 Plans were created with an intent to retaliate against their protected speech or conduct, and (3) the 2017 Plans would not have taken the adverse action but for that retaliatory intent. See McLaughlin, 240 N.C. App. at 172, 771 S.E.2d at 579-80. Plaintiffs proved all of these elements. 114. First, the 2017 Plans take adverse action against Plaintiffs. For the Individual Plaintiffs and the Organizational Plaintiffs’ members, the plans dilute the weight of their votes. The enacted plans adversely affect the individual Plaintiffs’ associational rights. In relative terms, Democratic voters under the 2017 Plans are far less able to succeed in electing candidates of their choice than they would be under plans that were not so carefully crafted to dilute their votes. And in absolute terms, Plaintiffs are 329 significantly foreclosed from succeeding in electing preferred candidates or a Democratic majority. 115. Second, the Plans were clearly crafted with an intent to retaliate against Plaintiffs and other Democratic voters on the basis of their voting history. Again, Dr. Hofeller’s files showed that when drafting the House and Senate maps he intentionally targeted Democratic voters based on their voting histories. Legislative Defendants cannot escape a finding of retaliatory intent by re-characterizing their actions as helping Republicans rather than hurting Democrats. In two-party elections, an intent to help one party necessarily implies an intent to hurt the other party. Nor does it matter that Legislative Defendants did not target specific individual voters. Plaintiffs were targeted for disfavored treatment because of a shared marker of political belief—their status as Democratic voters. That suffices. See Miller v. Johnson, 515 U.S. 900, 920, 115 S. Ct. 2475, 2490 (1995) (condemning State’s targeting of areas with “dense majority-black populations”). 116. Third, Legislative Defendants’ impermissible partisan intent caused the burden on Plaintiffs’ expression and association. The adverse effects described above would not have occurred if Legislative Defendants had not cracked and packed Democratic voters and thereby diluted their votes. In particular, Dr. Chen compared the districts in which the Individual Plaintiffs currently reside under the enacted plans with districts in which they would have resided under each of his simulated plans. Many of the Individual Plaintiffs’ actual districts are extreme partisan outliers when compared with their districts under the simulated plans. 117. For the foregoing reasons, the Court concludes that Plaintiffs have met their burden of showing, plainly and clearly without any reasonable doubt, that the enacted 330 plans violate the North Carolina Constitution’s guarantees of free speech and assembly under Article I, Sections 12 and 14 of the North Carolina Constitution. V. PARTISAN GERRYMANDERING CLAIMS ARE JUSTICIABLE UNDER THE NORTH CAROLINA CONSTITUTION 118. In all but the most exceptional circumstances, North Carolina courts are duty-bound to say what the law of this State is and to adjudicate cases on the merits. 119. In cases brought under the North Carolina Constitution, “[i]t has long been understood that it is the duty of the courts to determine the meaning of the requirements of our Constitution.” Leandro v. State, 346 N.C. 336, 345, 488 S.E.2d 249, 253 (1997). “When a government action is challenged as unconstitutional, the courts have a duty to determine whether that action exceeds constitutional limits.” Id. “It is the duty of this Court to ascertain and declare the intent of the framers of the Constitution and to reject any act in conflict therewith.” Maready v. City of Winston-Salem, 342 N.C. 708, 716, 467 S.E.2d 615, 620 (1996). 120. State courts’ duty to decide constitutional cases applies with full force in the redistricting context. Although the North Carolina Constitution directs the General Assembly to revise and reapportion districts after each census, “[t]he people of North Carolina chose to place several explicit limitations upon the General Assembly’s execution of the legislative reapportionment process,” which state courts have not hesitated to enforce. Stephenson, 355 N.C. at 370, 562 S.E.2d at 389. North Carolina courts have adjudicated claims that redistricting plans violated the Whole County Provision, the middecade redistricting bar, the Equal Protection Clause, and other provisions of the North Carolina Constitution. See Stephenson, 355 N.C. at 376, 380-81, 562 S.E.2d at 392, 395; State ex rel. Martin v. Preston, 325 N.C. 438, 385 S.E.2d 473 (1989); NAACP v. Lewis, 18 CVS 2322 (N.C. Super. Ct. Nov. 2, 2018). “[W]ithin the context of . . . redistricting and 331 reapportionment disputes, it is well within the power of the judiciary of [this] State to require valid reapportionment or to formulate a valid redistricting plan.” Stephenson, 355 N.C. at 362, 562 S.E.2d at 384 (quotation marks omitted). 121. Courts of other states have decided constitutional challenges to redistricting plans, including partisan gerrymandering claims, on the merits. In adjudicating a recent partisan gerrymandering suit, the Pennsylvania Supreme Court held that “it is the duty of the Court, as a co-equal branch of government, to declare, when appropriate, certain acts unconstitutional.” League of Women Voters of Pa., 178 A.3d at 822. The Florida Supreme Court similarly held that “there can hardly be a more compelling interest than the public interest in ensuring that the Legislature does not engage in unconstitutional partisan political gerrymandering.” League of Women Voters of Fla. v. Detzner, 172 So. 3d 363, 416 (Fla. 2015). And in another constitutional redistricting challenge, the Texas Supreme Court held that “[t]he judiciary . . . is both empowered and, when properly called upon, obliged to declare whether an apportionment statute enacted by the Legislature is valid.” Terrazas v. Ramirez, 829 S.W.2d 712, 717 (Tex. 1991). “A judicial determination that an apportionment statute violates a constitutional provision is no more an encroachment on the prerogative of the Legislature than the same determination with respect to some other statute.” Id.; see also, e.g., Johnson v. State, 366 S.W.3d 11, 23 (Mo. 2012) (similar). 122. Indeed, state courts are particularly well-positioned to adjudicate redistricting disputes, as the public may “more readily accept state court intervention . . . than . . . federal intervention in matters of state government.” Brooks v. Hobbie, 631 So. 2d 883, 890 (Ala. 1993). “The power of the judiciary of a State to require valid reapportionment or to formulate a valid redistricting plan has not only been recognized by th[e United States Supreme] Court but . . . has been specifically encouraged.” Scott v. 332 Germano, 381 U.S. 407, 409 (1965). In Rucho, the United States Supreme Court recently made clear that partisan gerrymandering claims are not “condemn[ed] . . . to echo in the void,” because although the federal courthouse doors may be closed, “state constitutions can provide standards and guidance for state courts to apply.” 139 S. Ct. at 2507. 123. If unconstitutional partisan gerrymandering is not checked and balanced by judicial oversight, legislators elected under one partisan gerrymander will enact new gerrymanders after each decennial census, entrenching themselves in power anew decade after decade. When the North Carolina Supreme Court first recognized the power to declare state statutes unconstitutional, it presciently noted that absent judicial review, members of the General Assembly could “render themselves the Legislators of the State for life, without any further election of the people.” Bayard v. Singleton, 1 N.C. 5, 7 (1787). Those legislators could even “from thence transmit the dignity and authority of legislation down to their heirs male forever.” Id. Extreme partisan gerrymandering reflects just such an effort by a legislative majority to permanently entrench themselves in power in perpetuity. 124. The fact that the process employed by the Legislative Defendant in crafting the 2017 Maps is a process that has been used in North Carolina for decades—albeit in less precise and granular detail—by Democrats and Republicans alike does render political gerrymandering nonjusticiable. Long standing, and even widespread historical practices do not immunize governmental action from constitutional scrutiny. See e.g., Citizens United v. FEC, 558 U.S. 310, 365 (2010); Reynolds v. Sims, 377 U.S. 533, 582 (1964) (holding that malapportionment of state legislative districts violates Equal Protection Clause, notwithstanding that malapportionment was widespread in the Nineteenth and early Twentieth Centuries.) 333 125. In rare instances, North Carolina courts have held that certain exceptional cases are non-justiciable because they present a “political question.” “The political question doctrine controls, essentially, when a question becomes not justiciable because of the separation of powers provided by the Constitution.” Cooper v. Berger, 370 N.C. 392, 407, 809 S.E.2d 98, 107 (2018) (quotation marks omitted; cleaned up). “The doctrine excludes from judicial review those controversies which revolve around policy choices and value determinations constitutionally committed for resolution to the legislative or executive branches of government.” Id. at 408, 809 S.E.2d at 107 (quotation marks omitted; cleaned up). The “dominant considerations” in determining whether the political question doctrine applies are “the appropriateness under our system of government of attributing finality to the action of the political departments and also the lack of satisfactory criteria for a judicial determination.” Id. (quotation marks omitted). 126. The Court concludes that partisan gerrymandering claims are justiciable under the North Carolina Constitution. Such claims fall within the broad, default category of constitutional cases the North Carolina courts are empowered and obliged to decide on the merits, and not within the narrow category of exceptional cases covered by the political question doctrine. 127. The Court concludes that partisan gerrymandering does not “involve a textually demonstrable constitutional commitment of the issue to a coordinate political department.” Bacon v. Lee, 353 N.C. 696, 717, 549 S.E.2d 840, 854 (2001) (quotation marks omitted). 128. Although Article II, §§ 3 and 5, of the North Carolina Constitution direct the General Assembly to revise and reapportion state House and Senate districts after each decennial census, North Carolina courts often decide constitutional challenges to state redistricting plans. COL ¶ 125 (citing cases). These cases conclusively refute any notion 334 that redistricting is “committed to the sole discretion of the General Assembly” without judicial review by the courts. Cooper, 370 N.C. at 409, 809 S.E.2d at 108 (emphasis added). 129. “[T]he General Assembly’s authority pursuant to [Article II, §§ 3 and 5] is necessarily constrained by the limits placed upon that authority by other provisions.” Cooper, 370 N.C. at 410, 809 S.E.2d at 109. The North Carolina Supreme Court has held that the State Constitution’s Equal Protection Clause constrains the General Assembly’s exercise of its redistricting authority pursuant to Article II, §§ 3 and 5. Stephenson, 355 N.C. at 376-82, 562 S.E.2d at 392-96. The people of North Carolina amended the Free Elections Clause to mandate that “all elections” not only “ought to be” but “shall be free.” N.C. Const. art. I, § 10 (emphasis added). This change “ma[d]e [it] clear” that the Free Elections Clause is a “command[] and not mere[ly] [an] admonition” to proper conduct on the part of the government. DuMont, 304 N.C. at 639, 286 S.E.2d at 97 (quotation marks omitted). And the North Carolina Supreme Court has held that North Carolinians must have a judicial “remedy for the violation of plaintiff's constitutionally protected right of free speech.” Corum, 330 N.C. at 784, 413 S.E.2d at 290. 130. In North Carolina, cases presenting “a conflict between . . . competing constitutional provisions” involve proper “constitutional interpretation, . . . rather than a nonjusticiable political question arising from nothing more than a policy dispute.” Cooper, 370 N.C. at 412, 809 S.E.2d at 110. The Court held in Cooper that a challenge to a statute creating a new State Board of Elections and Ethics Enforcement did not present a political question, because the General Assembly’s authority over the functions and powers of administrative agencies was limited by the Governor’s constitutional duty to “take care that the laws be faithfully executed.” Id. at 417-18, 809 S.E.2d at 113-14. Similarly, in News & Observer Publ’g Co. v. Easley, the Court held that a suit seeking public records related to 335 clemency applications was not a political question, because the Governor’s power over clemency was limited by the General Assembly’s power to enact laws “relative to the manner of applying for pardons.” 182 N.C. App. 14, 16, 641 S.E.2d 698, 700 (2007). So too, partisan gerrymandering claims do not present a political question because the General Assembly’s redistricting authority under Article II, §§ 3 and 5 is limited by the Equal Protection Clause, the Free Elections Clause, and the Freedom of Speech and Assembly Clauses. This Court’s task is “to identify where the line should be drawn” between these provisions. Id. at 15-16, 641 S.E.2d at 700. “There can be no doubt that [the Court has] the power and the responsibility to do so.” Id. 131. This case bears no resemblance to cases in which North Carolina courts have applied the political question doctrine. In Bacon v. Lee, for example, the North Carolina Supreme Court rejected a claim seeking a disinterested arbiter for a clemency application because the North Carolina Constitution “expressly commits the substance of the clemency power to the sole discretion of the Governor.” 353 N.C. at 698, 717, 549 S.E.2d at 843, 854 (emphasis added). Similarly, in Hoke Cnty. Bd. of Educ. v. State, the Supreme Court rejected a challenge to a statute setting the proper age for children to attend public school because the Constitution placed “the determination of the proper age for school children . . . squarely . . . in the hands of the General Assembly.” 358 N.C. 605, 639, 599 S.E.2d 365, 391 (2004). These cases centered on the appropriate exercise of authority under a single constitutional provision that was committed to the sole discretion of one of the political branches. Other cases cited by Legislative Defendants are similarly inapposite. See Leg. Defs.’ Pre-Trial Brief at 17 (citing cases). 132. The Court also concludes that “satisfactory and manageable criteria [and] standards . . . exist” for adjudicating partisan gerrymandering claims under the North 336 Carolina Constitution. Hoke, 358 N.C. at 639, 599 S.E.2d at 391. Plaintiffs have articulated satisfactory, manageable standards for each of their claims for relief. 133. The standard for Plaintiffs’ claim under the Free Elections Clause is based on the venerable history of that clause, as well as the commonsense insight that elections are not “free” where the partisan will of the mapmaker predominates over the ascertainment of the fair and truthful will of the voters. COL § II. The Court concludes this standard is satisfactory and manageable. 134. The standard for Plaintiffs’ claim under the Equal Protection Clause is based on the fundamental right to “substantially equal voting power” and to “vote on equal terms.” Stephenson, 355 N.C. at 378-79, 562 S.E.2d at 393-94. The North Carolina Supreme Court has previously applied this long-recognized standard, including in redistricting cases. See id.; Blankenship, 363 N.C. at 522-24, 681 S.E.2d at 762-64; Northampton Cnty., 326 N.C. at 747, 392 S.E.2d at 356. This standard is not only “manageable”—the North Carolina Supreme Court has already managed to apply it to resolve actual cases. The Court concludes that this standard is satisfactory and manageable. 135. The standards for Plaintiffs’ claims under the Free Speech and Free Assembly Clauses are based on longstanding doctrine, which recognizes that (1) voting is an expressive and associative act, and (2) government actions that burden or discriminate against protected expression or association, are subject to strict scrutiny. COL § IV.B-D. Plaintiffs also rely on longstanding retaliation doctrine, which prohibits the government from taking adverse actions based on protected expression or association. COL § IV.E. North Carolina courts routinely apply these standards to numerous government actions and programs in various contexts. The Court concludes that these standards are satisfactory and manageable. 337 136. Plaintiffs’ claims are justiciable notwithstanding that they arise under broad constitutional provisions that require interpretation. Courts routinely interpret broad constitutional text, adopt legal standards to operationalize such text, and then apply those legal standards to adjudicate the constitutionality of statutes. That is exactly what the North Carolina Supreme Court did in Stephenson. There, the Court interpreted a broad constitutional requirement that “[n]o county shall be divided in the formation of a [district],” N.C. Const. art. II, §§ 3 and 5, to require a detailed, multi-step procedure for redistricting, 355 N.C. at 383-84, 562 S.E.2d at 396-97. In adopting this standard, the Court explained that it was “not permitted to construe the [Whole County Provision] mandate as now being in some fashion unmanageable.” Id. at 382, 562 S.E.2d at 396. “Any attempt to do so,” the Court explained, “would be an abrogation of the Court’s duty to follow a reasonable, workable, and effective interpretation that maintains the people’s express wishes.” Id. So too here, it is the Court’s responsibility to distill the Free Elections Clause, the Equal Protection Clause, and Free Speech and Free Assembly Clauses into a “reasonable, workable, and effective interpretation.” 137. In Stephenson, the North Carolina Supreme Court also noted that “[p]rogress demands that government should be further refined in order to best respond to changing conditions.” Id. (quotation marks omitted). Like the Whole County Provision, the constitutional provisions invoked by Plaintiffs in this case “provide the elasticity which ensures the responsive operation of government.” Id. (quotation marks omitted). As the North Carolina Supreme Court asked rhetorically more than a century ago: “Is it true that we are living in a popular government, depending upon free and fair elections, and have a constitution that prohibits the legislature from authorizing a judge or a justice of the supreme court to investigate alleged irregularities of the election officers? If this were so, elections would become a farce, and free government a failure. But, fortunately for the 338 people and the government, in our opinion, this is not true, and fair and honest elections are to prevail in this state.” McDonald, 119 N.C. at 666, 26 S.E. at 134. 138. Legislative Defendants, joined by the Intervening Defendants, assert that this matter is not justiciable because when a claim, like they contend Plaintiffs’ to be, is that a districting plan is “somehow harmful to democracy,” there is “no way for the Court to address these concerns under a neutral, manageable standard.” Leg. Defs.’ and Int. Defs.’ Proposed Findings of Fact and Conclusions of Law at para. 800. They further suggest that judicial review of political redistricting claims will amount to “freewheeling policymaking,” id. at 803, and that “this court is not capable of controlling the exercise of power on the part of the General Assembly,” id. at 806 (citing Howell v. Howell, 66 S.E. 571, 573 (N.C. 1911)). 139. However, this is not a case where this Court is called upon to answer whether partisan gerrymandering is harmful to democracy (although the United States Supreme Court has certainly suggested that partisan gerrymandering is indeed harmful to democracy. See, Veith v. Jubelirer, 541 U.S. 267, 292, 124 S. Ct. 1769, 1785 (plurality opinion); id. at 316, 124 S. Ct. at 1798 (Kennedy, J., concurring); Ariz. State Legislature, 135 S. Ct. at 2658.). Nor is it a case where this Court is called upon to engage in policymaking by comparing the enacted maps with others that might be “ideally fair” under some judicially-envisioned criteria. It is not a case that threatens the General Assembly’s broad discretionary powers to create legislative districts, or threatens the General Assembly’s consideration of political data for legitimate purposes when crafting such districts. Rather this is a case where the Court is called upon to take the Adopted Criteria that the General Assembly itself, in its sole discretion, established, and compare the resulting maps with those criteria to see “how far the State had gone off that track because of its politicians’ effort to entrench themselves in office.” Rucho, 139 S. Ct. at 2521 (Kagan, J., dissenting). 339 140. Allowing the General Assembly discretion to establish its own redistricting criteria and craft maps accordingly is what the North Carolina Constitution requires; systematically packing and cracking voters to the extent that their votes are subordinated and devalued for no legitimate governmental purpose, but rather the purposes of entrenching a political party in power, is what the North Carolina Constitution forbids. When the Court is presented with evidence of the scope and quality proffered by Plaintiffs that shows widespread and extreme partisan gerrymandering—multiple districts showing a greater partisan skew than any of 3,000 randomly generated maps (all with the State’s political geography and districting criteria built in)—the standard is indeed clear and manageable. Such extreme partisan gerrymanders violate the fundamental constitutional rights of free elections, equal protection, speech, assembly and association. It is the Court’s duty to say so. 141. The separation of powers—which is expressly guaranteed by the North Carolina Constitution, art. I, § 6, and which underlies the political question doctrine— underscores the Court’s obligation to craft manageable judicial standards to adjudicate partisan gerrymandering claims. Each of the constitutional provisions invoked by Plaintiffs in this case appears in the Declaration of Rights in Article I of the North Carolina Constitution. And “[t]he civil rights guaranteed by the Declaration of Rights in Article I of our Constitution are individual and personal rights entitled to protection against state action.” Corum, 330 N.C. at 782, 413 S.E.2d at 289. “The very purpose of the Declaration of Rights is to ensure that the violation of these rights is never permitted by anyone who might be invested under the Constitution with the powers of the State.” Id. at 783, 413 S.E.2d at 290. And “[i]t is the state judiciary that has the responsibility to protect the state constitutional rights of the citizens.” Id. Indeed, “this obligation to protect the fundamental rights of individuals is as old as the State.” Id. 340 142. This Court is not bound by dicta from Stephenson that “[t]he General Assembly may consider partisan advantage and incumbency protection in the application of its discretionary redistricting decisions.” 355 N.C. at 371, 562 S.E.2d at 390. To begin with, the Supreme Court in Stephenson stated that any such considerations “must” be “in conformity with the State Constitution.” Id. In this case, Plaintiffs allege that partisan gerrymandering of the 2017 Plans violates provisions of the State Constitution, and there is an extensive trial record concerning those allegations. By contrast, Stephenson did not involve any partisan gerrymandering claim—let alone partisan gerrymandering claims under the constitutional provisions Plaintiffs invoke here—nor was there any record concerning partisan gerrymandering. The statements in Stephenson were “mere obiter dictum and [are] not binding on this Court or any other.” Taylor v. J.P. Stevens & Co., 300 N.C. 94, 100-01, 265 S.E.2d 144, 148 (1980). In a case with such important consequences, the Court will decide Plaintiffs’ claims on the basis of the record and arguments presented by the parties here, rather than follow dicta from prior cases involving different claims and evidence. 143. In order to reject Defendants’ invocation of the political question doctrine, this Court need not decide that the legal standards governing Plaintiffs’ claims would apply in all future cases, including a hypothetical close case. This case is not close. The extreme, intentional, and systematic gerrymandering of the 2017 Plans runs far afoul of the legal standards set forth above, or any other conceivable legal standard that could govern Plaintiffs’ constitutional claims. As Dr. Pegden testified, “[t]hese maps are so gerrymandered that no matter how you do the analysis, no matter who does the analysis, no matter which side is doing the analysis, you reach the same answer.” Tr. 1400:18-21. 144. The Court concludes that partisan gerrymandering claims are justiciable under the North Carolina Constitution. 341 VI. ANY LACHES DEFENSE LACKS MERIT 145. To the extent Defendants contend that Plaintiffs’ claims are barred by laches, that defense lacks merit. North Carolina courts have recognized that laches is inapplicable to continuing obligations. See Malinak v. Malinak, 242 N.C. App. 609, 612-13, 775 S.E.2d 915, 917 (2015). State and federal courts alike routinely refuse to apply laches in votingrights and other constitutional cases seeking solely prospective relief. E.g., Sprague v. Casey, 550 A.2d 184, 188-89 (Pa. 1988); Garza v. Cnty. of Los Angeles, 918 F.2d 763, 772 (9th Cir. 1990); Am. Trucking Ass’ns, Inc. v. N.Y. State Thruway Auth., 199 F. Supp. 3d 855, 872 (S.D.N.Y. 2016), vacated on other grounds, 238 F. Supp. 3d 527 (S.D.N.Y. 2017); Miller v. Bd. of Comm’rs of Miller Cnty., 45 F. Supp. 2d 1369, 1373 (M.D. Ga. 1998). Multiple federal courts have held that laches does not apply to partisan gerrymandering claims as a matter of law. See League of Women Voters of Mich., 373 F. Supp. 3d at 909; Ohio A. Philip Randolph Inst. v. Smith, 335 F. Supp. 3d 988, 1001-02 (S.D. Ohio 2018). 146. Moreover, “laches is an affirmative defense which the pleading party bears the burden of proving.” Malinak, 242 N.C. App. at 611, 775 S.E.2d at 916. Defendants presented no evidence at trial supporting laches. 147. Defendants offered no evidence of any “unreasonable” delay in filing this case. Id. at 612, 775 S.E.2d at 916. Plaintiffs commenced this case just fourteen months after the 2017 Plans were enacted. 148. Even if there had been any delay, Defendants presented no evidence that it “worked to the[ir] disadvantage, injury or prejudice.” Id. While Defendants have suggested that the time pressures of this case prevented their experts from conducting additional or more thorough analyses, any limitation on the time for Defendants’ expert reports was not the result of any delay by Plaintiffs. Rather, any such limitation resulted from Defendants’ own discovery misconduct in this case, which led the Court to extend the time for Plaintiffs’ 342 expert reports at the expense of the time for Defendants. See Order of Mar. 25, 2019. And the Court later granted Defendants a one-week extension to file their expert reports. Order of May 1, 2019. VII. DEFENDANTS’ FEDERAL DEFENSES LACK MERIT 149. Legislative Defendants and Intervenor Defendants raise a series of defenses under federal law, but none of these defenses has merit. A. The Covington Remedial Order Does Not Bar Changes to the 2017 Plans 150. Legislative Defendants contend that the Covington court’s remedial order in January 2018 precludes any changes being made to the current House and Senate plans. Legislative Defendants argue that the Covington remedial order contained an “express command that the 2017 plans be used in future elections,” so as to purportedly immunize the 2017 Plans from any state-law challenge. Leg. Defs.’ Pre-Trial Br. at 39. 151. Legislative Defendants made this same argument when they removed this case to federal court in December 2017, and the federal district court rejected it. The federal court held that the Covington remedial order “does not mandate the specific existing apportionment to the exclusion of no others.” Common Cause v. Lewis, 358 F. Supp. 3d 505, 512 (E.D.N.C. 2019). That holding constitutes law-of-the-case, or at minimum is entitled to controlling deference. 152. In any event, the federal court’s holding was clearly correct. In the very same remedial order that Legislative Defendants now cite, the Covington district court made clear that the 2017 Plans could be challenged on state-law grounds in state court. At Legislative Defendants’ urging, the Covington court declined to address state-law objections that the Covington plaintiffs had raised to the 2017 Plans, because those objections involved “unsettled questions of state law.” Covington v. North Carolina, 283 F. Supp. 3d 343 410, 428 (M.D.N.C. 2018). In declining to address such “unsettled question of state law,” the Covington court expressly stated that its order was “without prejudice to Plaintiffs or other litigants asserting such arguments in separate proceedings, including in “state court.” Id. at 447 n.9. The Covington court even noted that any “partisan gerrymandering objection” to the 2017 Plans “would demand development of significant new evidence and therefore [would] be more appropriately addressed in a separate proceeding.” Id. at 427. These statements squarely refute Legislative Defendants’ contention that the Covington remedial order precludes any changes to the 2017 Plans based on state-law violations that a state court may find. 153. The United States Supreme Court’s holding on appeal from the Covington remedial order eliminates any doubt on this score. The Court held that “[t]he District Court’s remedial authority was . . . limited to ensuring that the plaintiffs were relieved of the burden of voting in racially gerrymandered legislative districts.” 138 S. Ct. 2548, 2554 (2018). The Court explained: “Once the District Court had ensured that the racial gerrymanders at issue in this case were remedied, its proper role in North Carolina’s legislative districting process was at an end.” Id. at 2555. The Covington district court thus had no authority to do anything other than ensure the curing of the prior racial gerrymandering. It did not and could not immunize the plans from future challenge. 154. The Covington remedial order does not preclude North Carolina courts from invalidating the 2017 Plans for violations of state law and ordering the creation of new plans. B. There Is No Conflict with Federal Civil Rights Laws 155. The Court also rejects Legislative Defendants’ arguments that affording Plaintiffs relief on their claims would necessarily violate federal civil rights laws. 344 156. As described, Legislative Defendants introduced no evidence at trial to establish that any of the three Gingles factors, including the existence of legally sufficient racially polarized voting, is present in any area of the State or any particular districts. Legislative Defendants’ failure to present any evidence to establish that the Gingles factors are met is “is fatal to [any] Section 2 defense” under the VRA. Covington v. North Carolina, 316 F.R.D. 117, 169 (M.D.N.C. 2016), aff’d, 137 S. Ct. 2211 (2017). 157. Indeed, Legislative Defendants affirmatively represented throughout the 2017 redistricting process that the third Gingles factor was not met. FOF § F.6. Legislative Defendants have presented no evidentiary basis for any change in that position. The Court concludes that Legislative Defendants have not established that the VRA justifies the current House or Senate districts or precludes granting Plaintiffs relief on their claims. 158. Legislative Defendants also have not established any defense under the Fourteenth or Fifteenth Amendment. Legislative Defendants argue that affording Plaintiffs relief would require intentionally lowering the BVAP in purported “crossover” districts below the level necessary to elect candidates of choice of African Americans, but Legislative Defendants again have advanced no evidence to substantiate this claim. They provided no evidence to establish any district qualifies as a “crossover district,” or that remedying the partisan gerrymander in any district or grouping would require lowering the BVAP of any crossover district below the level necessary for African Americans to elect candidates of their choice. 159. Indeed, Legislative Defendants’ own expert Dr. Lewis generated estimates of the minimum BVAP needed in certain county groupings for African-American-preferred candidate to win, and Dr. Chen demonstrated that his nonpartisan simulations produce districts within each such county grouping with BVAPs above Dr. Lewis’s estimates. FOF § F.6. 345 160. Legislative Defendants’ federal equal protection defense suffers from another fatal defect—it requires a showing of an intent to discriminate against African Americans. To establish a Fourteenth or Fifteenth Amendment violation, there must be “racially discriminatory intent,” Lee v. Va. State Bd. of Elections, 843 F.3d 592, 603 (4th Cir. 2016), which in the redistricting context means “intentional vote dilution,” i.e., “invidiously minimizing or canceling out the voting potential of racial or ethnic minorities,” Abbott v. Perez, 138 S. Ct. 2305, 2314 (2018) (quotation marks and alterations omitted). 161. The Court finds without difficulty that Plaintiffs have no intent to discriminate against racial minorities in seeking remedial plans to replace the current plans that violate state constitutional provisions. Further, Plaintiffs alone cannot adopt or approve remedial plans in this case. The remedial plans approved or adopted in this case, as ordered below, will not intentionally dilute the voting power of any North Carolina citizens. C. Granting Relief Will Not Violate the Fundamental Right to Vote 162. Finally, Legislative Defendants contend that affording Plaintiffs relief in this case will violate the “fundamental right to vote” under the Fourteenth Amendment. Legislative Defendants cite no federal precedent for this purported defense, but in any event it lacks merit. 163. Granting Plaintiffs relief will promote, not violate, the fundamental right to vote of North Carolina citizens. Legislative Defendants’ defense operates from the misapprehension that voting rights must be a zero-sum game, such that curing discrimination against one set of citizens necessarily requires discriminating against another set of citizens. The right that Plaintiffs seek to vindicate is the right to be free from intentional discrimination, and vindicating that right in no way requires or will result in discriminating against others. 346 VIII. THE COURT WILL ENJOIN USE OF THE 2017 PLANS IN FUTURE ELECTIONS AND THE GENERAL ASSEMBLY IS TO IMMEDIATELY BEGIN THE PROCESS OF REDRAWING THE RELEVANT DISTRICTS A. The Court Will Require the Redrawing of Specific County Groupings 164. For the reasons stated above, and as set forth in the decree below, the Court declares that there is no reasonable doubt the 2017 House and Senate Plans are unconstitutional under the North Carolina Constitution, and the Court enjoins their use in the 2020 primary and general elections. In particular, the Court enjoins use of the districts in the specific House and Senate county groupings as specified in the decree below. 165. The Court does not enjoin or order any relief with respect to the current House districts in Wake County. Shortly before the trial in this matter, those districts were redrawn pursuant to a separate litigation. See NAACP v. Lewis, No. 18 CVS 2322 (N.C. Super. Ct. Nov. 2, 2018); N.C. Sess. Laws 2019-46. Plaintiffs did not present evidence in this case regarding the new Wake County House districts and do not seek relief with respect to those districts. 166. The Court does not enjoin or order the redrawing of House Districts 57, 61, and 62 or Senate Districts 24 or 28, all of which were redrawn by the Covington Special Master. With respect to House District 59 and Senate District 27, for which small portions of the current districts were added by the Special Master in Covington, the Court will order that the remedial versions of these districts not alter any portions of these districts that were added by the Special Master, but any other portions of these districts may be redrawn. Neither House District 59 nor Senate District 27 were found by the Covington court to have been racially gerrymandered (under either the 2011 Plans or the 2017 Plans enacted by the General Assembly), and the Covington court did not direct the Special Master to redraw either of these districts. The Special Master nonetheless made small changes to these districts, principally to equalize population, in the course of constructing other districts he 347 was tasked with redrawing. While this Court concludes that there is no legal impediment to redrawing any portion of House District 59 and Senate District 27, including the portions that the Special Master added, the Court nonetheless imposes the limitation set forth in this paragraph out of an abundance of caution. B. The Court Will Require the Use of the Adopted Criteria, with certain exceptions, and Prohibit the Use of Other Criteria in Redrawing the Districts 167. As set forth in the Court’s decree below, the Court will require that Remedial Maps for the House and Senate legislative district maps for the 2020 election (hereinafter “Remedial Maps”) be drawn, and that the Remedial Maps comply with the criteria adopted by the General Assembly’s House and Senate Redistricting Committees on August 10, 2017, with several exceptions. 168. First, with respect to “Incumbency Protection,” the drafters of the Remedial Maps may take reasonable efforts to not pair incumbents unduly in the same election district. Because Representative David Lewis, Chair of the House Redistricting Committee, explained at the time of the adoption of the Adopted Criteria that the “Incumbency Protection” criteria was “simply saying that mapmakers may take reasonable efforts to not pair incumbents unduly,” PX603 at 122:4-18; Tr. 1640:16-1641:12, and the criteria was understood as such, see PX606 at 9:24-10:1 (Sen. Hise: “The Committee adopted criteria pledging to make reasonable efforts not to double-bunk incumbents”), the Remedial Maps shall comply with this explanation and understanding. 169. Second, the “Election Data” criteria shall not be permitted in the drafting of the Remedial Maps. In other words, partisan considerations and election results data shall not be used in the drawing of legislative districts in the Remedial Maps. The Court likewise will prohibit any intentional attempt to favor voters or candidates of one political party. 348 170. In redrawing the relevant districts in the Remedial Maps, the invalidated 2017 districts may not be used as a starting point for drawing new districts, and no effort may be made to preserve the cores of invalidated 2017 districts. See Covington, 283 F. Supp. 3d at 431-32 (holding that remedial plan could not seek to “preserve the ‘cores’ of unconstitutional districts”). 171. Any Remedial Maps must comply with the VRA and other federal requirements concerning the racial composition of districts. The Court will afford all parties an opportunity to submit briefing, which may attach expert analysis, on whether the Gingles factors are met in particular counties and county groupings and/or the minimum BVAP needed in particular counties and county groupings for African-Americans to be able to elect candidates of their choice to the General Assembly. Any such submission by Legislative Defendants, however, is subject to two limitations set forth below. a) First, if Legislative Defendants assert that the Gingles factors are met in any particular district or county grouping, they must not only provide evidentiary support for that assertion, but also must also show good cause why they did not compile such evidence during the 2017 redistricting process and must show good cause why they should not be held judicially estopped from arguing that the Gingles factors are met given their repeated representations to the Covington court in 2017 that the third Gingles factor was not met anywhere in the State. b) Second, for districts in counties and county groupings for which Legislative Defendants’ expert Dr. Lewis estimated the minimum BVAP needed for an African-American preferred candidate to prevail in a state legislative election, Legislative Defendants may not assert that the VRA or the United States Constitution requires or justifies making the BVAP of any such district higher than the minimum BVAP threshold estimated by Dr. Lewis in his Amended 349 Table 4 (which was admitted into evidence at trial) for the relevant county or county grouping. PX773. For districts in counties and county groupings that Dr. Lewis did not analyze, Legislative Defendants may not assert that the VRA or the United States Constitution requires or justifies any minimum BVAP for the districts in that county or county grouping. The Court holds that Legislative Defendants are bound by the BVAP threshold-estimates generated by the expert they retained in this case and are estopped from departing from those estimates, which were relied upon by Plaintiffs’ experts, at this late stage of the litigation. 172. The Court will afford the General Assembly two weeks from the date of this Order, namely through September 18, 2019, to enact Remedial Maps in conformity with this Order. See N.C.G.S. § 120-2.4. 173. The Court concludes that this two week period is consistent with N.C.G.S. § 120-2.4, which states that “in no event may a court impose its own substitute plan unless the court first gives the General Assembly a period of time to remedy any defects identified by the court in its findings of fact and conclusions of law. That period of time shall not be less than two weeks.” Although § 120-2.4 goes on to state that a longer period of time might be required in some instances, that longer period, the Court concludes, is applicable only if the General Assembly is not currently in session. See N.C. Sess. Laws 2018-146, § 4.7. The Court notes that the General Assembly, as of the date of this Order, is in session. 174. The Court will require Legislative Defendants and their agents to conduct the entire remedial process in full public view. At a minimum, that would require all map drawing to occur at public hearings, with any relevant computer screen visible to legislators and public observers. Given what transpired in 2017, the Court will prohibit Legislative Defendants and their agents from undertaking any steps to draw or revise the new districts outside of public view. 350 175. If Legislative Defendants wish to retain one or more individuals who are not current legislative employees to assist in the map-drawing process, the Court will require Legislative Defendants to obtain approval from the Court to engage any such individuals. 176. Notwithstanding the General Assembly having the opportunity to draw Remedial Maps in the first instance, the Court will still immediately appoint a Referee to (1) assist the Court in reviewing any Remedial Maps enacted by the General Assembly; and (2) to develop remedial maps for the Court should the General Assembly fail to enact lawful Remedial Maps within the time allowed. C. The Court Will Not Stay the Remedial Process Pending Appeal 177. The Court orders that the remedial process commence immediately upon entry of this Order, and the Court will not grant a stay of the remedial process pending appeal. 178. The central inquiry in deciding whether to grant a stay of relief pending appeal is a balancing of the prejudice and risk of irreparable harm to the parties. See 130 of Chatham, LLC v. Rutherford Elec. Mbrshp. Corp., 2014 WL 3809066, at *9 (N.C. Super. Ct. July 31, 2014). 179. Here, the balance of the equities weighs definitively against any stay. “[C]ourts evaluating redistricting challenges have generally denied motions for a stay pending appeal.” Harris v. McCrory, 2016 WL 6920368, at *1 n.1 (M.D.N.C. Feb. 9, 2016) (citing cases and denying stay pending appeal). In such cases, a stay pending appeal could “risk that the State would not be able to implement” the remedial plans “in time for the [next] elections in the event that the [appellate courts] affirm[] this Court’s judgment.” Covington, 2018 WL 604732, at *6 (denying stay pending appeal). “The risk of harm is particularly acute where Plaintiffs and other North Carolina voters have already cast their ballots under unconstitutional district plans” in every election this decade. Id. The 351 prejudice to Plaintiffs here would be magnified because the state legislators elected in 2020 will redraw the state House and Senate districts in 2021 following the Decennial Census, substantially compounding the effects of allowing the current unconstitutional plans to be used in the 2020 elections. 180. In contrast, Legislative Defendants will suffer little if any prejudice from refusing any stay pending appeal. If Legislative Defendants ultimately prevail in an appeal, then the current districts will remain in place for the 2020 elections, and there will be no tangible harm from having allowed the remedial process to move forward while the appeal was pending. On balance, the equities and the public interest counsel strongly against a stay. D. The Court Retains Discretion to Move the Primary Dates 181. Finally, the Court holds that the remedial schedule and process that the Court has set forth in this Order should ensure that remedial plans will be in place sufficiently in advance of the current primary date of March 3, 2020. However, the Court retains authority and discretion to move the primary date for the General Assembly elections, or all of the State’s 2020 primaries, including for offices other than the General Assembly, should doing so become necessary to provide effective relief in this case. 182. While the Court concludes that moving the 2020 primaries is not needed at this date, the Court may consider doing so if necessary to grant effective relief in this case. DECREE Having considered all of the evidence, the memoranda and arguments of counsel, and the record proper, the Court ORDERS the following: 1. The Court declares that the 2017 House and Senate Plans are unconstitutional and invalid because there is no reasonable doubt each plan violates the rights of Plaintiffs and other Democratic voters under the North Carolina Constitution’s 352 Equal Protection Clause, art. I, § 19; the Free Elections Clause, art. I, § 10; and the Freedom of Speech and Freedom of Assembly Clauses, art. I, §§ 12 & 14. 2. Legislative Defendants and State Defendants, and their respective agents, officers, and employees, are permanently enjoined from preparing for or administering the 2020 primary and general elections for House districts in the following House county groupings: a. Alamance b. Anson-Union c. Brunswick-New Hanover d. Buncombe e. Cabarrus-Davie-Montgomery-Richmond-Rowan-Stanly (except that House District 66 shall not be redrawn) f. Cleveland-Gaston g. Columbus-Pender-Robeson h. Cumberland i. Duplin-Onslow j. Franklin-Nash k. Forsyth-Yadkin l. Guilford (except that House Districts 57, 61, and 62 shall not be redrawn, and any portions of House District 59 added by the Covington Special Master shall not be altered) m. Lenoir-Pitt n. Mecklenburg 3. Legislative Defendants and State Defendants, and their respective agents, officers, and employees, are permanently enjoined from preparing for or administering the 2020 primary and general elections for Senate districts in the following Senate county groupings: 353 a) Alamance-Guilford-Randolph (except that Senate Districts 24 and 28 shall not be redrawn, and any portions of Senate District 27 added by the Covington Special Master shall not be altered) b) Bladen-Brunswick-New Hanover-Pender c) Buncombe-Henderson-Transylvania d) Davie-Forsyth e) Duplin-Harnett-Johnston-Lee-Nash-Sampson f) Franklin-Wake g) Mecklenburg 4. The Court will afford the General Assembly two weeks from the date of this Order, namely through September 18, 2019, to enact Remedial Maps for the House and Senate legislative districts for the 2020 election (hereinafter “Remedial Maps”) in conformity with this Order. 5. Except as otherwise noted in this Order, the following criteria shall exclusively govern the redrawing of districts in the House and Senate county groupings set forth above: a. Equal Population. The mapmakers shall use the 2010 federal decennial census data as the sole basis of population for drawing legislative districts in the Remedial Maps. The number of persons in each legislative district shall comply with the +/- 5 percent population deviation standard established by Stephenson v. Bartlett, 355 N.C. 354, 562 S.E. 2d 377 (2002). b. Contiguity. Legislative districts shall be comprised of contiguous territory. Contiguity by water is sufficient. c. County Groupings and Traversals. The mapmakers shall draw legislative districts in the Remedial Maps within county groupings as required by Stephenson v. Bartlett, 355 N.C. 354, 562 S.E. 2d 377 (2002) (Stephenson I), Stephenson v. Bartlett, 357 N.C. 301, 582 S.E.2d 247 (2003) (Stephenson II), Dickson v. Rucho, 367 N.C. 542, 766 S.E.2d 238 (2014) (Dickson I) and Dickson v. Rucho, 368 N.C. 481, 781 S.E.2d 460 (2015) (Dickson II). Within county groupings, county lines shall not be traversed except as authorized by Stephenson I, Stephenson II, Dickson I, and Dickson II. The county groupings utilized in the 2017 House and Senate Maps shall be utilized in the Remedial Maps. 354 d. Compactness. The mapmakers shall make reasonable efforts to draw legislative districts in the Remedial Maps that improve the compactness of the districts when compared to districts in place prior to the 2017 Enacted Legislative Maps. In doing so, the mapmaker may use as a guide the minimum Reock (“dispersion”) and Polsby-Popper (“perimeter”) scores identified by Richard H. Pildes and Richard G. Neimi in Expressive Harms, “Bizarre Districts,” and Voting Rights: Evaluating Election-District Appearances After Shaw v. Reno, 92 Mich. L. Rev. 483 (1993). e. Fewer Split Precincts. The mapmakers shall make reasonable efforts to draw legislative districts in the Remedial Maps that split fewer precincts when compared to districts in place prior to the 2017 Enacted Legislative Maps. f. Municipal Boundaries. The mapmakers may consider municipal boundaries when drawing legislative districts in the Remedial Maps. g. Incumbency Protection. The mapmakers may take reasonable efforts to not pair incumbents unduly in the same election district. h. Election Data. Partisan considerations and election results data shall not be used in the drawing of legislative districts in the Remedial Maps. 6. In redrawing the relevant districts in the Remedial Maps, the invalidated 2017 districts may not be used as a starting point for drawing new districts, and no effort may be made to preserve the cores of invalidated 2017 districts. 7. Any Remedial Maps must comply with the VRA and other federal requirements concerning the racial composition of districts. Within 14 days of this Order, all parties may submit briefing, which may attach expert analysis, on whether the Gingles factors are met in particular counties and county groupings and/or the minimum BVAP needed in particular counties and county groupings for African Americans to be able to elect candidates of their choice to the General Assembly. Any such submission by Legislative Defendants is subject to the limitations set forth in subparagraphs (a) and (b) immediately below. a) If Legislative Defendants assert that the Gingles factors are met in any counties or county groupings, they shall not only provide evidentiary support for that assertion, but shall also show good cause why they did not compile such evidence during the 2017 redistricting process and shall show good cause why they should not be held judicially estopped from 355 arguing that the Gingles factors are met given their repeated representations to the Covington court in 2017 that the third Gingles factor was not met anywhere in the State. b) For districts in counties and county groupings for which Legislative Defendants’ expert Dr. Lewis estimated the minimum BVAP needed for an African-American preferred candidate to prevail in a state legislative election, Legislative Defendants shall not assert that the VRA or the United States Constitution requires or justifies making the BVAP of any such district higher than the minimum BVAP threshold estimated by Dr. Lewis in his Amended Table 4 (PX773) for the relevant county or county grouping. For districts in counties and county groupings that Dr. Lewis did not analyze, Legislative Defendants shall not assert that the VRA or the United States Constitution requires or justifies any minimum BVAP for the districts in that county or county grouping. 8. Legislative Defendants and their agents shall conduct the entire remedial process in full public view. At a minimum, this requires all map drawing to occur at public hearings, with any relevant computer screen visible to legislators and public observers. Legislative Defendants and their agents shall not undertake any steps to draw or revise the new districts outside of public view. 9. To the extent that Legislative Defendants wish to retain one or more individuals who are not current legislative employees to assist in the map-drawing process, Legislative Defendants must seek and obtain prior approval from the Court to engage any such individuals. 10. Notwithstanding the General Assembly having the opportunity to draw Remedial Plans in the first instance, the Court, by subsequent Court Order, shall promptly appoint a Referee to (1) assist the Court in reviewing any Remedial Maps enacted by the General Assembly; and (2) to develop remedial maps for the Court should the General Assembly fail to enact lawful Remedial Maps within the time allowed. 14. No later than September 6, 2019, the parties may submit to the Court names and qualifications of suggested referees. The Court will thereafter appoint a referee by subsequent Court Order. 356 15. The Court orders that the remedial process will commence immediately upon entry of this Order. 17. The Court, on its own motion, denies a stay of the remedial process pending appeal. 18. The Court retains jurisdiction to move the primary date for the General Assembly elections, or all of the State’s 2020 primaries, including for offices other than the General Assembly, should doing so become necessary to provide effective relief in this case. SO ORDERED, this the 3rd day of September, 2019. /s/ Paul C. Ridgeway ____________________________________ Paul C. Ridgeway, Superior Court Judge /s/ Joseph N. Crosswhite ____________________________________ Joseph N. Crosswhite, Superior Court Judge /s/ Alma L. Hinton ____________________________________ Alma L. Hinton, Superior Court Judge 357