WYNN, Circuit Judge:

Benjamin W. Thorpe, Emmet J. Bondurant, Jason J. Carter, Bondurant Mixson & Elmore, LLP, Atlanta, GA, Caroline P. Mackie, Steven B. Epstein, Edwin M. Speas, Jr., Poyner Spruill, LLP, Raleigh, NC, Gregory L. Diskant, Peter A. Nelson, Patterson Belknap Webb & Tyler, New York, NY,Allison Jean Riggs, Emily E. Seawell, Anita S. Earls, Southern Coalition for Social Justice, Durham, NC, Anna E. Bodi, Danielle M. Lang, J. Gerald Hebert, Molly Danahy, Paul M. Smith, Campaign Legal Center, Washington, DC, Annabelle E. Harless, Ruth M. Greenwood, Campaign Legal Center, Nicholas O. Stephanopoulos, University of Chicago Law School, Chicago, IL, for Plaintiffs.

Alexander McClure Peters, James Bernier, Jr., N.C. Department of Justice, Michael Douglas McKnight, Phillip John Strach, Thomas A. Farr, Ogletree Deakins Nash Smoak & Stewart, P.C., Raleigh, NC, for Defendants.

Before WYNN, Circuit Judge, and OSTEEN, JR., District Judge, and BRITT, Senior District Judge.

MEMORANDUM OPINION

WYNN, Circuit Judge: In these consolidated cases, two groups of Plaintiffs allege that North Carolina's 2016 Congressional Redistricting Plan (the "2016 Plan") constitutes a partisan gerrymander in violation of the First Amendment, the Equal Protection Clause of the Fourteenth Amendment, and Article I, Sections 2 and 4 of the Constitution. Legislative Defendants do not dispute that the General Assembly intended for the 2016 Plan to favor supporters of Republican candidates and disfavor supporters of non-Republican candidates. Nor could they. The Republican-controlled North Carolina General Assembly expressly directed the legislators and consultant responsible for drawing the 2016 Plan to rely on "political data"—past election results specifying whether, and to what extent, particular voting districts had favored Republican or Democratic candidates, and therefore were likely to do so in the future—to draw a districting plan that would ensure Republican candidates would prevail in the vast majority of the state's congressional districts. Ex. 1007.

Legislative Defendants also do not argue—and have never argued—that the 2016 Plan's intentional disfavoring of supporters of non-Republican candidates advances any democratic, constitutional, or public interest. Nor could they. Neither the Supreme Court nor any lower court has recognized any such interest furthered by partisan gerrymandering—"the drawing of legislative district lines to subordinate adherents of one political party and entrench a rival party in power." Ariz. State Legislature v. Ariz. Indep. Redistricting Comm'n , ––– U.S. ––––, 135 S.Ct. 2652, 2658, 192 L.Ed.2d 704 (2015). And, as further detailed below, partisan gerrymandering runs contrary to numerous fundamental democratic principles and individual rights enshrined in the Constitution.

Rather than seeking to advance any democratic or constitutional interest, the state legislator responsible for drawing the 2016 Plan said he drew the map to advantage Republican candidates because he "think[s] electing Republicans is better than electing Democrats." Ex. 1016, at 34:21–23. But that is not a choice the Constitution allows legislative mapdrawers to make. Rather, "the core principle of [our] republican government [is] that the voters should choose their representatives, not the other way around." Ariz. State Leg. , 135 S.Ct. at 2677 (internal quotation marks omitted). Accordingly, and as further explained below, we conclude that Plaintiffs prevail on all of their constitutional claims.

I.

A.

Over the last 30 years, North Carolina voters repeatedly have asked state and federal courts to pass judgment on the constitutionality of the congressional districting plans drawn by their state legislators. The first such challenge involved a redistricting plan adopted by the North Carolina General Assembly after the 1990 census, which increased the size of North Carolina's congressional delegation from 11 to 12 members. See Shaw v. Reno (Shaw I ), 509 U.S. 630, 633–34, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993). When the General Assembly set out to redraw the state's congressional districts to incorporate the new seat, the Department of Justice, pursuant to its "max-black" policy, pushed for the creation of a second majority-black district to augment, it maintained, the representation of the state's African–American voters in Congress. Id. at 635, 113 S.Ct. 2816. In response, the General Assembly prepared a revised district map that included the majority-black First and Twelfth Districts (the "1992 Plan"). Id.

Several dozen North Carolina voters, most of whom were Republican, challenged the 1992 Plan as a partisan gerrymander, in violation of the Equal Protection Clause, the First Amendment, and Article I, Section 2 of the United States Constitution. Pope v. Blue , 809 F.Supp. 392, 394–95, 397–98 (W.D.N.C. 1992), aff'd 506 U.S. 801, 113 S.Ct. 30, 121 L.Ed.2d 3 (1992). A divided three-judge panel dismissed the action, holding that the plaintiffs failed to adequately allege that the redistricting plan had a legally cognizable "discriminatory effect" on any "identifiable [political] group," under the standard set forth in the Supreme Court's decision in Davis v. Bandemer , 478 U.S. 109, 127, 106 S.Ct. 2797, 92 L.Ed.2d 85 (1986) (plurality op.). Pope , 809 F.Supp. at 397.

Separately, a group of North Carolina voters challenged the 1992 Plan as a racial gerrymander, in violation of the Equal Protection Clause. Shaw I , 509 U.S. at 636–37, 113 S.Ct. 2816. After several years of litigation, the Supreme Court held that the General Assembly's use of race as the predominant factor in drawing the second majority-black district in the 1992 Plan violated the Equal Protection Clause, and enjoined the use of that district in future elections. Shaw v. Hunt (Shaw II ), 517 U.S. 899, 905–18, 116 S.Ct. 1894, 135 L.Ed.2d 207 (1996). In 1997, a politically divided General Assembly enacted a remedial plan expected to elect six Republican and six Democratic Representatives, rendering each party's share of the state's congressional delegation proportional to its share of the statewide vote in the most recent congressional election. Cromartie v. Hunt , 133 F.Supp.2d 407, 412–13 (E.D.N.C. 2000), rev'd sub nom. Easley v. Cromartie , 532 U.S. 234, 121 S.Ct. 1452, 149 L.Ed.2d 430 (2001) ; id. at 423–24 (Thornburg, J., dissenting). In 2001, after several more years of litigation, the Supreme Court approved that remedial plan. See Easley , 532 U.S. 234, 121 S.Ct. 1452, 149 L.Ed.2d 430 (holding that three-judge panel's finding that race constituted the predominant motivation in redrawing remedial districts was not supported by substantial evidence).

Just as litigation regarding the 1992 Plan came to an end, the results of the 2000 census entitled North Carolina to another seat in Congress, and the General Assembly again set out to redraw the state's congressional districts to include the additional seat. The resulting plan, which was adopted in 2001 (the "2001 Plan"), was used in each of the State's congressional elections between 2001 and 2010. In all but one of these elections, the party receiving more statewide votes for their candidates for the House of Representatives also won a majority of the seats in North Carolina's congressional delegation (the only exception being the 2010 election, in which Republicans won 54 percent of votes statewide but only 6 of the 13 seats). Exs. 1021–25. Although the 2001 Plan did not include any majority-black districts, black voters in the First and Twelfth Districts were consistently successful in electing their preferred candidates. Harris v. McCrory , 159 F.Supp.3d 600, 606–07 (M.D.N.C. 2016), aff'd sub nom. Cooper v. Harris , ––– U.S. ––––, 137 S.Ct. 1455, 197 L.Ed.2d 837 (2017). Unlike the 1992 Plan, the 2001 Plan did not generate significant federal litigation. Id. at 607.

B.

In 2010, for the first time in more than a century, North Carolina voters elected Republican majorities in both the North Carolina Senate and the North Carolina House of Representatives, giving Republicans exclusive control over the decennial congressional redistricting process. See id. at 607. The House of Representatives and Senate each established redistricting committees, which were jointly responsible for preparing a proposed congressional redistricting plan. Id. Representative David Lewis, in his capacity as the senior chair of the House Redistricting Committee, and Senator Robert Rucho, in his capacity as senior chair of the Senate Redistricting Committee, were responsible for developing the proposed redistricting plan. Id.

Through private counsel, the committees engaged Dr. Thomas Hofeller, who had previously worked as the redistricting coordinator for the Republican National Committee, to draw the new congressional districting plan. Id. Concurrent with his work on the 2011 North Carolina congressional redistricting plan, Dr. Hofeller also served on a "redistricting team" established as part of the Republican State Leadership Committee's ("RSLC") Redistricting Majority Project, commonly referred to as "REDMAP." Ex. 2015, at ¶ 13. According to RSLC, REDMAP sought to elect Republican candidates to state legislatures so that Republicans would control such legislatures' redistricting efforts and thereby "solidify conservative policymaking at the state level and maintain a Republican stronghold in the U.S. House of Representatives for the next decade." Id. at ¶ 10. With regard to North Carolina, in particular, REDMAP sought to "[s]trengthen Republican redistricting power by flipping [state legislative] chambers from Democrat to Republican control." Ex. 2020.

Representative Lewis and Senator Rucho, both of whom are Republican, orally instructed Dr. Hofeller regarding the criteria he should follow in drawing the new districting plan. Dep. of Thomas B. Hofeller ("Hofeller Dep.") 20:7–19, Jan. 24, 2017, ECF Nos. 101–34, 110–1. According to Dr. Hofeller, Representative Lewis and Senator Rucho's "primar[y] goal" in drawing the new districts was "to create as many districts as possible in which GOP candidates would be able to successfully compete for office." Id . at 123:1–7.

In accordance with Representative Lewis and Senator Rucho's instructions, Dr. Hofeller testified that he sought "to minimize the number of districts in which Democrats would have an opportunity to elect a Democratic candidate." Id. at 127:19–22. In order to minimize the electoral opportunities of Democratic candidates, Dr. Hofeller used the results of past statewide elections to predict whether a particular precinct or portion of a precinct was likely to vote for a Republican or Democratic congressional candidate in future elections. See id. at 132:22–134:13, 159:20–160:12. According to Dr. Hofeller, "past voting behavior," as reflected in "past election results," is "the best predictor of future election success." Ex. 2037. Past election data have become "the industry standard" for predicting the partisan performance of a districting plan, he explained, because "as more and more voters ... register non-partisan or independent," party registration data have decreased in predictive value. Id.

Using past election data to "draw maps that were more favorable to Republican candidates," Dr. Hofeller moved district lines "to weaken Democratic strength in Districts 7, 8, and 11 ... by concentrating Democratic voting strength in Districts 1, 4, and 12." Ex. 2043, at 33–34; see also Hofeller Dep. 116:19–117:25 ("The General Assembly's goal [in 2011] was to increase Republican voting strength in New Districts 2, 3, 6, 7, and 13. This could only be accomplished by placing all the strong Democratic [census voting districts ("VTDs") ] in either New Districts 1 or 4."). Dr. Hofeller conceded that, by doing so, the 2011 Plan "diminished ... [t]he[ ] opportunity to elect a Democratic candidate in the districts in which [he] increased Republican voting strength." Hofeller Dep. 128:17–21.

Believing (incorrectly) that Section 2 of the Voting Rights Act required the creation of majority-black districts "where possible," Representative Lewis and Senator Rucho also directed Dr. Hofeller to re-establish two majority-black districts in the state. Harris , 159 F.Supp.3d at 608. This goal worked hand-in-hand with the General Assembly's partisan objective because, as Legislative Defendants acknowledge, "race and politics are highly correlated." Ex. 2043, at ¶ 120. Thus, Dr. Hofeller drew the map to further concentrate black voters, who are more likely to vote for Democratic candidates, into the state's First and Twelfth Congressional districts, where Dr. Hofeller already was planning to concentrate Democratic voting strength. Harris , 159 F.Supp.3d at 607–09. As a result, the proportion of black voters in those districts increased from 47.76 percent to 52.65 percent and from 43.77 percent to 50.66 percent, respectively. Id. The General Assembly enacted the 2011 Plan on July 28, 2011. Id. at 608.

North Carolina conducted two congressional elections using the 2011 Plan. In 2012, Republican candidates received a minority of the statewide vote (49%), Ex. 3023, but won a supermajority of the congressional seats (9 of 13), Ex. 1020. In 2014, Republican candidates received 54 percent of the statewide vote, and won 10 of the 13 congressional seats. Ex. 1019.

Meanwhile, voters living in the two majority-black districts challenged the 2011 Plan in both state and federal court, alleging that lines for the two districts constituted unconstitutional racial gerrymanders. Harris , 159 F.Supp.3d at 609–10. The North Carolina Supreme Court twice ruled that the 2011 Plan did not violate the state or federal constitution. Dickson v. Rucho , 368 N.C. 481, 781 S.E.2d 404, 410–11 (N.C. 2015), vacated , ––– U.S. ––––, 137 S.Ct. 2186, 198 L.Ed.2d 252 (2017) (mem.); Dickson v. Rucho , 367 N.C. 542, 766 S.E.2d 238 (N.C. 2014), vacated , ––– U.S. ––––, 135 S.Ct. 1843, 191 L.Ed.2d 719 (2015) (mem.). However, on February 5, 2016, a three-judge panel presiding in the U.S. District Court for the Middle District of North Carolina struck down the districts as unconstitutional racial gerrymanders and enjoined their use in future elections. Harris , 159 F.Supp.3d at 627.

With both chambers of the North Carolina General Assembly still controlled by Republicans, Representative Lewis and Senator Rucho again took charge of drawing the remedial districting plan. On February 6, 2016, Representative Lewis decided to again engage Dr. Hofeller to draw the remedial plan. Dep. of Rep. David Lewis ("Lewis Dep.") 44:2–4, Jan. 26, 2017, ECF Nos. 101–33, 108–3, 110–3, 110–4; see also Ex. 4061. Soon thereafter, Representative Lewis spoke with Dr. Hofeller over the phone regarding the drawing of the new plan. Lewis Dep. 44:12–24; Ex. 4061. Even before he spoke with Representative Lewis, Dr. Hofeller had begun working on a remedial plan using redistricting software and data on his personal computer. Hofeller Dep. 130:2–9.

On February 9, 2016, Representative Lewis and Senator Rucho met with Dr. Hofeller at his home and provided him with oral instructions regarding the criteria he should follow in drawing the remedial plan. Ex. 4061; Lewis Dep. 48:19–49:7; Dep. of Sen. Robert Rucho ("Rucho Dep.") 170:13–170:17, Jan. 25, 2017, ECF Nos. 101–32, 110–5. Once again, Representative Lewis and Senator Rucho did not reduce their instructions to Dr. Hofeller to writing. Lewis. Dep. 60:1–13. In addition to directing Dr. Hofeller to remedy the racial gerrymander, Representative Lewis and Senator Rucho again directed Dr. Hofeller to use political data—precinct-level election results from all statewide elections, excluding presidential elections, dating back to January 1, 2008—in drawing the remedial plan. Ex. 2043, at ¶ 38; Lewis Dep. 162:24–163:7; Hofeller Dep. 100:3–102:5, 180:10–16. Representative Lewis and Senator Rucho further instructed Dr. Hofeller that he should use that political data to draw a map that would maintain the existing partisan makeup of the state's congressional delegation, which, as elected under the racially gerrymandered plan, included 10 Republicans and 3 Democrats. Ex. 2043, at ¶ 38; Lewis Dep. 162:24–163:7; Hofeller Dep. 175:19–23, 178:14–20, 188:19–190:2.

With these instructions, Dr. Hofeller continued to prepare draft redistricting plans on his personal computer. To achieve Representative Lewis and Senator Rucho's partisan objectives—and in accordance with his belief that "past voting data" serve as the best predictor of future election results—Dr. Hofeller drew the draft plans using an aggregate variable he created to predict partisan performance. For each census block, the variable compared the sum of the votes cast for Republican candidates in seven statewide races occurring between 2008 and 2014 with the sum of the average total number of votes cast for Democratic and Republican candidates in those same races. Exs. 1017, 2002, 2039, 2043 at ¶¶ 18, 47, 49, 50; Dep. of Thomas Hofeller, vol. II ("Hofeller Dep. II") 262:21–24, Feb. 10, 2017, ECF No. 110–2. Dr. Hofeller testified that he used the averaged results from the seven elections so as "to get a pretty good cross section of what the past vote had been," Hofeller Dep. 212:16–213:9, and "[t]o give [him] an indication of the two-party partisan characteristics of VTDs," Hofeller Dep. II 267:5–6. Dr. Hofeller explained that "he had drawn numerous plans in the state of North Carolina over decades," and in his "experience[,] ... the underlying political nature of the precincts in the state does not change no matter what race you use to analyze it." Ex. 2045, at 525:6–10; Hofeller Dep. at 149:5–18. "So once a precinct is found to be a strong Democratic precinct, it's probably going to act as a strong Democratic precinct in every subsequent election. The same would be true for Republican precincts." Ex. 2045, at 525:14–17; see also Hofeller Dep. II at 274:9–12 ("[I]ndividual VTDs tend to carry ... the same characteristics through a string of elections.").

When he drew district lines, Dr. Hofeller displayed his partisanship variable on his computer screen by color-coding counties, VTDs, or precincts to reflect their partisan performance. Ex. 5116, at ¶ 8, fig.1; Hofeller Dep. 103:5–105:24; Hofeller Dep. II 267:18–278:4. Dr. Hofeller would use the partisanship variable to assign a VTD "to one congressional district or another," Hofeller Dep. 106:23–107:1, 132:14–20, and "as a partial guide" in deciding whether and where to split VTDs or counties, id. at 203:4–5; Hofeller Dep. II at 267:10–17. In assigning a county, VTD, or precinct to a particular district, Dr. Hofeller also sought to preserve the "core" constituency of the districts in the 2011 Plan. Ex. 5001, at ¶ 31. Using his partisanship variable—and in accordance with his effort to preserve the "cores" of the districts in the 2011 Plan—Dr. Hofeller drew, for example, the Fourth and Twelfth Districts to be "predominantly Democratic," as those districts had been under the 2011 Plan. Hofeller Dep. 192:7–12. After drawing a draft plan, Dr. Hofeller also would use his seven-election variable to assess the partisan performance of the plan on a district-by-district basis and as a whole. Id. at 247:18–23; Hofeller Dep. II 283:15–19, 284:20–285:4. Dr. Hofeller then would convey his assessment of the partisan performance of each district to Representative Lewis. Hofeller Dep. II 290:17–25.

The following day, February 10, 2016, Dr. Hofeller met with Representative Lewis and Senator Rucho and showed them several draft redistricting plans. Rucho Dep. 31:16–31:18, 37:7–37:8. "Nearly every time" he reviewed Dr. Hofeller's draft maps, Representative Lewis assessed the plans' partisan performance using the results from North Carolina's 2014 Senate race between Senator Thom Tillis and former Senator Kay Hagan. Lewis Dep. 63:9–64:17. Representative Lewis visited Dr. Hofeller's house several more times over the next few days to review additional draft remedial plans. On either February 12 or February 13, Dr. Hofeller presented the near-final 2016 Plan to Representative Lewis, which Representative Lewis found acceptable. Id. at 77:7–20.

On February 12, 2016, the leadership of the North Carolina General Assembly appointed Representative Lewis and Senator Rucho as co-chairs of a newly formed Joint Select Committee on Congressional Redistricting (the "Committee"), comprised of 25 Republican and 12 Democratic legislators, to draw the remedial district plan. Ex. 2009. On February 15, 2016, the co-Chairs held a public hearing on the redistricting effort. Ex. 1004. Dr. Hofeller did not attend the public hearing. Rucho Dep. 55:4–6. The Committee also solicited written comments regarding the redistricting efforts on its website. Id. at 55:10–23. Dr. Hofeller was not apprised of any of the comments made at the public hearing or in the written submissions. Id. at 55:4–56:13. Because Dr. Hofeller finished drawing the 2016 Plan before the public hearing and the opening of the window for members of the public to submit written comments, Hofeller Dep. 177:9–21, the 2016 Plan did not reflect any public input.

On February 16, 2016—after Dr. Hofeller, at Representative Lewis and Senator Rucho's direction, had completed drawing the remedial maps, id. ; Ex. 5001, at ¶ 33—the Committee met for the first time. At that meeting, Representative Lewis and Senator Rucho proposed the following criteria to govern the drawing of the remedial districts:

Equal Population: The Committee will use the 2010 federal decennial census data as the sole basis of population for the establishment of districts in the 2016 Contingent Congressional Plan. The number of persons in each congressional district shall be as nearly as equal as practicable, as determined under the most recent federal decennial census.

Contiguity: Congressional districts shall be comprised of contiguous territory. Contiguity by water is sufficient.

Political Data: The only data other than population data to be used to construct congressional districts shall be election results in statewide contests since January 1, 2008, not including the last two presidential contests. Data identifying the race of individuals or voters shall not be used in the construction or consideration of districts in the 2016 Contingent Congressional Plan. Voting districts ("VTDs") should be split only when necessary to comply with the zero deviation population requirements set forth above in order to ensure the integrity of political data.

Partisan Advantage: The partisan makeup of the congressional delegation under the enacted plan is 10 Republicans and 3 Democrats. The Committee shall make reasonable efforts to construct districts in the 2016 Contingent Congressional Plan to maintain the current partisan makeup of North Carolina's congressional delegation.

Twelfth District: The current General Assembly inherited the configuration of the Twelfth District from past General Assemblies. This configuration was retained because the district had already been heavily litigated over the past two decades and ultimately approved by the courts. The Harris court has criticized the shape of the Twelfth District citing its "serpentine" nature. In light of this, the Committee shall construct districts in the 2016 Contingent Congressional Plan that eliminate the current configuration of the Twelfth District.

Compactness: In light of the Harris court's criticism of the compactness of the First and Twelfth Districts, the Committee shall make reasonable efforts to construct districts in the 2016 Contingent Congressional Plan that improve the compactness of the current districts and keep more counties and VTDs whole as compared to the current enacted plan. Division of counties shall only be made for reasons of equalizing population, consideration of incumbency and political impact. Reasonable efforts shall be made not to divide a county into more than two districts.

Incumbency: Candidates for Congress are not required by law to reside in a district they seek to represent. However, reasonable efforts shall be made to ensure that incumbent members of Congress are not paired with another incumbent in one of the new districts constructed in the 2016 Contingent Congressional Plan.

Ex. 1007. No other criteria were discussed by the Committee or in legislative debate on the 2016 Plan.

Representative Lewis explained the relationship between the "Political Data" and "Partisan Advantage" criteria as follows: the Partisan Advantage criterion "contemplate[s] looking at the political data ... and as you draw the lines, if you're trying to give a partisan advantage, you would want to draw lines so that more of the whole VTDs voted for the Republican on the ballot than they did the Democrat." Ex. 1005, at 57:10–16. And he further explained that "to the extent [we] are going to use political data in drawing this map, it is to gain partisan advantage."Id. at 54. Representative Lewis "acknowledge[d] freely that this would be a political gerrymander," which he maintained was "not against the law." Id. at 48:4–6.

Democratic state Senator Floyd McKissick, Jr., objected to the "Partisan Advantage" criterion, stating that "ingrain[ing]" the 10–3 advantage in favor of Republicans was not "fair, reasonable, [or] balanced" because, as recently as 2012, Democratic congressional candidates had received more votes on a statewide basis than Republican candidates. Id. at 49:16–50:5, 50:14–22. In response, Representative Lewis said that he "propose[d] that [the Committee] draw the maps to give a partisan advantage to 10 Republicans and 3 Democrats because [he] d[id] not believe it[ would be] possible to draw a map with 11 Republicans and 2 Democrats." Id. at 50:7–10. Democratic Committee members also expressed concern that the Partisan Advantage criterion would "bake in partisan advantage that was achieved through the use of unconstitutional maps." Id. at 62:1–3. In response, Representative Lewis again reiterated that "the goal" of the criterion "is to elect 10 Republicans and 3 Democrats." Id. at 62:18–19.

That same day, Committee members adopted, on a bipartisan basis, the Equal Population, Contiguity, Twelfth District, and Incumbency criteria. Id. at 14:16–18:3, 21:9–24:18, 91:17–94:17, 95:15–98:20. The remaining two criteria—Political Data and Partisan Advantage—were adopted on party-line votes. Id. at 43:21–47:5, 67:2–69:23. Additionally, the Committee authorized the chairmen to engage a consultant to assist the Committee's Republican leadership in drawing the remedial plan. Ex. 2003.

Also on February 16, 2016, after receiving authorization to hire a redistricting consultant, Representative Lewis and Senator Rucho sent Dr. Hofeller an engagement letter, which Dr. Hofeller signed that same day. Ex. 2003. Upon his engagement, Dr. Hofeller downloaded the 2016 Plan, which he had completed several days earlier, from his personal computer onto a legislative computer. Lewis Dep. 138:6–8; Ex. 1009, at 45:7–45:11; Ex. 1014, at 21:10–21:24; Ex. 4061. Democratic Committee members were not allowed to consult with Dr. Hofeller nor were they allowed access to the state computer systems to which he downloaded the 2016 Plan. Ex. 1011, at 36:9–20; Ex. 1014, at 44:23–45:15; Ex. 2008. According to Representative Lewis, Senator Rucho, and Dr. Hofeller, the 2016 Plan adhered to the Committee's Partisan Advantage and Political Data criteria. Ex. 1014, at 36:25–37:6; Ex. 1016, at 37:3–7; Hofeller Dep. 129:14–15.

The following day, Representative Lewis and Senator Rucho presented the 2016 Plan to the Committee. Ex. 1008. As part of the presentation, Representative Lewis provided Committee members with spreadsheets showing the partisan performance of the proposed districts in twenty previous statewide elections. Ex. 1017. Representative Lewis stated that he and Senator Rucho believed that the 2016 Plan "will produce an opportunity to elect ten Republican members of Congress," but it was "a weaker map than the [2011 Plan]" from the perspective of Partisan Advantage. Ex. 1008, at 12:3–7. The Committee approved the 2016 Plan by party-line vote. Id. at 67:10–72:8.

On February 19, 2016, the North Carolina House of Representatives debated the 2016 Plan. During that debate, Representative Lewis further explained the rationale behind the Partisan Advantage criterion, stating: "I think electing Republicans is better than electing Democrats. So I drew this map to help foster what I think is better for the country." Ex. 1016, at 34:21–23. Following that debate, the North Carolina Senate and North Carolina House of Representatives approved the 2016 Plan, with one slight modification, on February 18 and February 19, respectively, in both cases by party-line votes. Ex. 1011, at 110:13–22; Ex. 1016, at 81:6–16.

The 2016 Plan splits 13 counties and 12 precincts. Ex. 5023. Under several statistical measures of compactness, the districts created by the 2016 Plan are, on average, more compact than the districts created by the 2011 Plan. Ex. 5048. The 2016 Plan paired 2 of the 13 incumbents elected under the unconstitutional 2011 Plan. Ex. 2012, at 15–19. Ten of the thirteen districts in the 2016 Plan retained at least 50 percent of their constituency under the 2011 Plan. Ex. 5001, tbl.1.

The Harris plaintiffs filed objections to the Plan with the three-judge court presiding over the racial gerrymandering case. Harris v. McCrory , No. 1:13-cv-949, 2016 WL 3129213, at *1 (M.D.N.C. June 2, 2016). Among those objections, the Harris plaintiffs asked the court to reject the 2016 Plan as an unconstitutional partisan gerrymander. Id. at *2. Noting that the Supreme Court had not agreed to a standard for adjudicating partisan gerrymandering claims and that the "plaintiffs ha[d] not provided the Court with a ‘suitable standard’ " for evaluating such claims, the court rejected the partisan gerrymandering objection "as presented." Id. at *3 (quoting Ariz. State Leg. , 135 S.Ct. at 2658). The court twice made clear, however, that its "denial of plaintiffs' objections does not constitute or imply an endorsement of, or foreclose any additional challenges to, the [2016 Plan]." Id. at *1, *3 (emphasis added).

In November 2016, North Carolina conducted congressional elections using the 2016 Plan. In accordance with the objective of the Partisan Advantage criterion, Republican candidates prevailed in 10 of the 13 (76.92%) congressional districts established by the 2016 Plan. Ex. 1018. Republican candidates received 53.22 percent of the statewide vote. Ex. 3022.

C.

On August 5, 2016, Common Cause, the North Carolina Democratic Party, and fourteen North Carolina voters (collectively, "Common Cause Plaintiffs"), filed a complaint alleging that the 2016 Plan constituted a partisan gerrymander. Compl., Common Cause v. Rucho , No. 1:16–cv–1026, Aug. 5, 2016, ECF No. 1. The League of Women Voters of North Carolina (the "League") and twelve North Carolina voters (collectively, "League Plaintiffs," and together with Common Cause Plaintiffs, "Plaintiffs") filed their partisan gerrymandering action on September 22, 2016. Compl., League of Women Voters of N.C. v. Rucho , No. 1:16–cv–1164, Sept. 22, 2016, ECF No. 1. Both parties named as defendants Legislative Defendants; A. Grant Whitney, Jr., in his official capacity as Chairman of the North Carolina State Board of Elections (the "Board of Elections"); the Board of Elections; and the State of North Carolina (collectively, with Chairman Whitney and the Board of Elections, "State Defendants," and with Legislative Defendants, "Defendants").

In their operative complaints, both Common Cause Plaintiffs and League Plaintiffs allege that the 2016 Plan violates the Equal Protection Clause, by intentionally diluting the electoral strength of individuals who previously opposed, or were likely to oppose, Republican candidates, and the First Amendment, by intentionally burdening and retaliating against supporters of non-Republican candidates on the basis of their political beliefs and association. First Am. Compl. for Decl. J. and Inj. Relief ("Common Cause Compl.") ¶¶ 25–45, Common Cause v. Rucho , No. 16–cv–1026, Sept. 7, 2016, ECF No. 12; Am. Compl. ("League Compl.") ¶¶ 69–83, League of Women Voters of N.C. v. Rucho , No. 16–cv–1164, Feb. 10, 2017, ECF No. 41. Common Clause Plaintiffs further allege that the 2016 Plan violates Article I, Section 2 of the United States Constitution, which provides that members of the House of Representatives will be chosen "by the People of the several States," by usurping the right of "the People" to select their preferred candidates for Congress, and Article I, Section 4, by exceeding the States' delegated authority to determine "the Times, Places and Manner of holding Elections" for members of Congress. Common Cause Compl. ¶¶ 46–54.

On February 7, 2017, this Court consolidated the two actions for purposes of discovery and trial. Order, Feb. 7, 2017, ECF No. 41. Three days later, League Plaintiffs amended their complaint to reflect the results of the 2016 congressional election conducted under the 2016 Plan and empirical analyses of those results.

On February 21, 2017, Defendants moved to dismiss both complaints under Federal Rule of Civil Procedure 12(b)(6), principally asserting that (1) Pope v. Blue , 809 F.Supp. 392 (W.D.N.C. 1992), which the Supreme Court summarily affirmed, 506 U.S. 801, 113 S.Ct. 30, 121 L.Ed.2d 3 (1992), required dismissal of Plaintiffs' actions, and (2) the Supreme Court's splintered opinions regarding the justiciability of—and, to the extent such claims are justiciable, the legal framework for—partisan gerrymandering claims foreclosed Plaintiffs' claims. Mot. to Dismiss for Failure to State a Claim, Feb. 21, 2017, ECF No. 45. In a memorandum opinion and order entered March 3, 2017, this Court denied Defendants' motions to dismiss. Common Cause v. Rucho , 240 F.Supp.3d 376 (M.D.N.C. 2017) ; Order, March 3, 2017, ECF No. 51.

Beginning on October 16, 2017, this Court held a four-day trial, during which the Common Cause Plaintiffs, League Plaintiffs, and Legislative Defendants introduced evidence and presented testimony from their expert witnesses. Although counsel for the State Defendants attended trial, they did not participate and took no position as to how this Court should resolve the case.

In post-trial briefing, League Plaintiffs set forth a single, three-part test for determining whether a state congressional redistricting plan violates the First and Fourteenth Amendments. Under their proposed test, a plaintiff alleging that a state redistricting body engaged in unconstitutional partisan gerrymandering bears the burden of proving: (1) that the redistricting body enacted the challenged plan with the intent of discriminating against voters who support candidates of a disfavored party and (2) that the challenged plan had a "large and durable" discriminatory effect on such voters. League of Women Voters Pls.' Post–Trial Br. ("League Br.") 3, Nov. 6, 2017, ECF No. 113. If the plaintiff makes such a showing, then the burden shifts to the governmental defendant to provide (3) a legitimate, non-partisan justification for the plan's discriminatory effect. Id.

League Plaintiffs point to the Political Advantage and Partisan Advantage criteria and the chairmen's official explanations of those criteria as evidence of the General Assembly's intent to discriminate against voters who support Democratic candidates. Id. at 7–8. To establish the plan's discriminatory effect, League Plaintiffs introduced expert analyses of the 2016 Plan's alleged "partisan asymmetry" to establish that the plan makes it substantially more difficult for voters who favor Democratic candidates to translate their votes into representation, and that this substantial difficulty is likely to persist throughout the life of the 2016 Plan. Id. at 12–16. Finally, League Plaintiffs assert that Legislative Defendants have failed to provide any evidence of a legitimate justification for the 2016 Plan's alleged partisan asymmetry, such as the state's political geography or other legitimate redistricting goals. Id. at 21–24.

By contrast, Common Cause Plaintiffs advance distinct legal frameworks for their First Amendment, Equal Protection, and Article I claims. Regarding the First Amendment, Common Cause Plaintiffs assert that the 2016 Plan's disfavoring of voters who previously opposed Republican candidates or associated with non-Republican candidates or parties amounts to viewpoint discrimination and passes constitutional muster only if narrowly tailored to serve a compelling state interest. Common Cause Pls.' Post–Trial Br. ("Common Cause Br.") 5–8, Nov. 6, 2017, ECF No. 116. According to Common Cause Plaintiffs, the General Assembly's use of individuals' past voting history to assign such individuals to congressional districts with the purpose of advantaging Republican candidates on a statewide basis constitutes evidence of viewpoint discrimination. Id. at 7–15. Common Clause Plaintiffs further contend that Legislative Defendants have provided no compelling interest justifying such viewpoint discrimination. Id. at 9.

Turning to the Equal Protection Clause, Common Cause Plaintiffs suggest that the level of scrutiny to which a court must subject a redistricting plan turns on the degree to which the redistricting body intended to pursue partisan advantage. Id. at 15–17. According to Common Cause Plaintiffs, the General Assembly predominantly pursued partisan advantage in drawing the 2016 Plan, and therefore this Court should apply strict scrutiny, upholding the plan only if Legislative Defendants show that the plan was narrowly tailored to advance a compelling state interest. Id. As proof of the General Assembly's predominant intent to burden voters who support non-Republican candidates, Common Cause Plaintiffs point to the Political Data and Partisan Advantage criteria, the chairmen's explanations of the purpose behind those criteria, and expert analyses showing that the 2016 Plan is an "extreme statistical outlier " with regard to its pro-Republican tilt relative to thousands of other simulated districting plans conforming to non-partisan districting principles. Id. at 17. Common Cause Plaintiffs further argue that, even if this Court finds that the General Assembly did not draw the 2016 Plan with a predominantly partisan motive, the plan nonetheless fails to pass constitutional muster under intermediate or rational basis scrutiny. Id. at 18–19.

Finally, Common Cause Plaintiffs allege that the 2016 Plan exceeds the General Assembly's delegated authority under Article I, Section 4—commonly referred to as the "Elections Clause"—because it amounts to an unconstitutional effort " ‘to dictate electoral outcomes’ " and " ‘to favor ... a class of candidates.’ " Id. at 20–21 (quoting Cook v. Gralike , 531 U.S. 510, 523–24, 121 S.Ct. 1029, 149 L.Ed.2d 44 (2001) ). And Common Clause Plaintiffs further assert that the 2016 Plan violates Article I, Section 2 because it gives voters who favor Republican candidates "a greater voice in choosing a Congressman" than voters who favor candidates put forward by other parties. Id. at 22–23 (quoting Wesberry v. Sanders , 376 U.S. 1, 13–14, 84 S.Ct. 526, 11 L.Ed.2d 481 (1964) ).

In response, Legislative Defendants first argue that both sets of Plaintiffs lack Article III standing to assert any of their claims. Legislative Defs.' Post–Trial Br. ("Leg. Defs.' Br.") 12, Nov. 6, 2017, ECF No. 115. Legislative Defendants next contend that, even if Plaintiffs have standing, neither set of Plaintiffs has offered a judicially manageable standard under any constitutional provision for evaluating a partisan gerrymandering claim, and, therefore, that Plaintiffs' actions must be dismissed as raising nonjusticiable political questions. Id. at 9. To that end, Legislative Defendants criticize Plaintiffs' expert statistical analyses, in particular, on grounds that such analyses are "a smorgasbord of alleged ‘social science’ theories" that fail to answer what Legislative Defendants see as the fundamental question in partisan gerrymandering cases: "how much politics is too much politics in redistricting?" Id. at 2, 9–11. As to the merits, Legislative Defendants assert that the 2016 Plan was not a "partisan gerrymander"—as they define that term—because, among other reasons, (1) the General Assembly did not try to "maximize" the number of Republican seats, and (2) the districts created by the 2016 Plan conform to a number of traditional redistricting principles such as compactness, contiguity, and adherence to county lines. Id. at 3, 7–8.

For the reasons that follow, we reject Legislative Defendants' standing and justiciability arguments. We further conclude that the 2016 Plan violates the Equal Protection Clause because the General Assembly enacted the plan with the intent of discriminating against voters who favored non-Republican candidates, the plan has had and likely will continue to have that effect, and no legitimate state interest justifies the 2016 Plan's discriminatory partisan effect. We also conclude that the 2016 Plan violates the First Amendment by unjustifiably discriminating against voters based on their previous political expression and affiliation. Finally, we hold that the 2016 Plan violates Article I by exceeding the scope of the General Assembly's delegated authority to enact congressional election regulations and interfering with the right of "the People" to choose their Representatives.

II.

Before addressing the merits of Plaintiffs' claims, we first address Legislative Defendants' threshold standing and justiciability arguments. As detailed below, we conclude that Plaintiffs have standing to raise statewide and district-by-district partisan gerrymandering challenges to the 2016 Plan. We further conclude that Plaintiffs' partisan gerrymandering claims are not barred by the political question doctrine, either in theory or as proven.

A.

Article III's "case" or "controversy" requirement demands that a plaintiff demonstrate standing—that the plaintiff has "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." Baker v. Carr , 369 U.S. 186, 204, 82 S.Ct. 691, 7 L.Ed.2d 663 (1962). To establish standing, a plaintiff first must demonstrate "an ‘injury in fact’—an invasion of a legally protected interest which is (a) concrete and particularized, ... and (b) ‘actual or imminent, not conjectural or hypothetical.’ " Lujan v. Defs. of Wildlife , 504 U.S. 555, 560, 112 S.Ct. 2130, 119 L.Ed.2d 351 (1992) (citations and some internal quotation marks omitted). "Second, there must be a causal connection between the injury and the conduct complained of—the injury has to be ‘fairly ... trace[able] to the challenged action of the defendant, and not ... th[e] result [of] the independent action of some third party not before the court.’ " Id. (alterations in original) (quoting Simon v. E. Ky. Welfare Rights Org. , 426 U.S. 26, 41–42, 96 S.Ct. 1917, 48 L.Ed.2d 450 (1976) ). "Third, it must be ‘likely,’ as opposed to merely ‘speculative,’ that the injury will be ‘redressed by a favorable decision.’ " Id. at 561, 112 S.Ct. 2130 (quoting Simon , 426 U.S. at 41–42, 96 S.Ct. 1917 ). Plaintiffs bear the burden of establishing standing. DaimlerChrysler Corp. v. Cuno , 547 U.S. 332, 342, 126 S.Ct. 1854, 164 L.Ed.2d 589 (2006).

Plaintiffs comprise individual North Carolina voters; two non-profit organizations concerned with promoting open, honest, and accountable government and fostering education and engagement in elections; and the North Carolina Democratic Party. These individuals and entities assert a variety of injury types: vote dilution; elected representatives who, with victory all-but assured, are less willing to engage in democratic dialogue and meaningfully consider contrary viewpoints; statewide chilling of association and discourse through decreased democratic participation, fundraising, and candidate recruitment; increased statewide costs for voter education and candidate recruitment; and a statewide congressional delegation that fails to adequately reflect the interests of all North Carolina voters. League Plaintiffs—who reside in most, but not all, of the state's thirteen congressional districts—assert that these alleged injuries allow them to lodge a statewide challenge under the Equal Protection Clause and First Amendment. Common Cause Plaintiffs—who reside in all thirteen congressional districts—claim that they have standing to assert both statewide and district-by-district challenges to the 2016 Plan under the Equal Protection Clause, the First Amendment, and Article I.

Legislative Defendants do not dispute that, to the extent Plaintiffs suffered an injury-in-fact, the injury was caused by the 2016 Plan. Nor do they dispute that Plaintiffs' claimed injuries are redressable by a favorable decision of this Court. Instead, Legislative Defendants argue that all Plaintiffs lack standing for three reasons: (1) a plaintiff may not rely on statewide standing to challenge an entire congressional redistricting plan as a partisan gerrymander; (2) individual Plaintiffs lack standing to lodge both statewide and district-by-district challenges because they have not suffered constitutionally cognizable injuries-in-fact; and (3) organizational Plaintiffs lack standing because no individual member has standing and no organizational Plaintiff suffered a concrete harm attributable to the 2016 Plan. We reject each argument.

1.

Two strands of Supreme Court precedent dealing with standing in gerrymandering cases under the Equal Protection Clause potentially bear on whether a partisan gerrymandering plaintiff has standing to raise a statewide challenge to a congressional redistricting plan. In racial gerrymandering cases, a plaintiff lacks standing to challenge a districting plan on a statewide basis. Ala. Leg. Black Caucus , 135 S.Ct. at 1265. The Supreme Court explained that only those voters who "live[ ] in the district attacked"—as opposed to voters "who live[ ] elsewhere in the State"—"normally [have] standing to pursue a racial gerrymandering claim" because "the harms that underlie a racial gerrymandering claim ... are personal." Id. "They include being personally subjected to a racial classification, as well as being represented by a legislator who believes his primary obligation is to represent only the members of a particular racial group." Id. (internal citation, quotation marks, and alterations omitted). A racial gerrymander, therefore, "reinforces the perception that members of the same racial group—regardless of their age, education, economic status, or the community in which they live—think alike, share the same political interests, and will prefer the same candidates at the polls." Shaw I , 509 U.S. at 647, 113 S.Ct. 2816. Such harms "threaten[ ] to stigmatize individuals by reason of their membership in a racial group and to incite racial hostility." United States v. Hays , 515 U.S. 737, 744, 115 S.Ct. 2431, 132 L.Ed.2d 635 (1995). Put differently, the harm associated with a racial gerrymander is that the state redistricting body drew district lines that "embody stereotypes that treat individuals as the product of their race, evaluating their thoughts and efforts—their very worth as citizens—according to a criterion barred to the Government by history and the Constitution." Miller v. Johnson , 515 U.S. 900, 912, 115 S.Ct. 2475, 132 L.Ed.2d 762 (1995) (quoting Metro Broad., Inc. v. F.C.C. , 497 U.S. 547, 604, 110 S.Ct. 2997, 111 L.Ed.2d 445 (1990) (O'Connor, J., dissenting)).

By contrast, in one-person, one-vote cases—in which a plaintiff in an overpopulated district alleges that she is injured because the districting plan dilutes her vote relative to voters in underpopulated districts—the plaintiff may challenge the districting plan on a statewide basis. See, e.g. , Wesberry , 376 U.S. at 7, 84 S.Ct. 526 (permitting voters in a single overpopulated district to raise one-person, one-vote challenge to districting plan as a whole); Gray v. Sanders , 372 U.S. 368, 370, 375, 83 S.Ct. 801, 9 L.Ed.2d 821 (1963) (holding that plaintiff, "who [wa]s qualified to vote in primary and general elections in Fulton County, Georgia," had standing to lodge statewide challenge to Georgia's "county unit system as a basis for counting votes in a Democratic primary for the nomination of a United States Senator and statewide officers"); Baker , 369 U.S. at 187, 205–07, 82 S.Ct. 691 (holding that plaintiffs, who lived in five Tennessee counties, had standing to challenge districting plan's "apportioning [of] the members of the General Assembly among the State's 95 counties" because "voters who allege facts showing disadvantage to themselves as individuals have standing to sue"). Like racial gerrymandering cases, the Supreme Court's approach to standing in one-person, one-vote cases reflects the type of harms associated with malapportionment. The injury in a malapportionment case is "a gross disproportion of representation to voting population." Baker , 369 U.S. at 207, 82 S.Ct. 691. "[T]his classification disfavors the voters in [overpopulated districts], placing them in a position of constitutionally unjustified inequality vis-à-vis voters in irrationally favored [districts]." Id. at 207–08, 82 S.Ct. 691. Put differently, in a one-person, one-vote case, a plaintiff who resides in an overpopulated district suffers an injury because her vote is diluted relative to other voters in the jurisdiction. Reynolds v. Sims , 377 U.S. 533, 568, 84 S.Ct. 1362, 12 L.Ed.2d 506 (1964) ("[A]n individual's right to vote ... is unconstitutionally impaired when its weight is in a substantial fashion diluted when compared with votes of citizens living [i]n other parts ...."). Importantly, in the context of one-person, one-vote challenges to a congressional districting plan, like the 2016 Plan, the Supreme Court has found that malapportionment causes structural harms, as well as individual harms, by contravening the legislative structure and republican principles put in place by the Framers. Wesberry , 376 U.S. at 15–18, 84 S.Ct. 526.

Legislative Defendants assert that this Court should follow the Supreme Court's racial gerrymandering cases and deny Plaintiffs statewide standing for two reasons: (1) partisan gerrymandering cases involve the "same representational harms" as racial gerrymandering cases, and (2) "race-based claims allege a more serious violation of the Constitution than do partisan-based claims." Leg. Defs.' Proposed Findings of Fact and Conclusion of Law ("Leg. Defs.' FOF") 112–13, Nov. 6, 2017, ECF No. 114. As to the first argument, we agree that some of the injuries flowing from partisan gerrymandering are analogous to the injuries attributable to a racial gerrymander. For example, a plaintiff subject to an invidious partisan gerrymander is harmed by "being represented by a legislator who believes his primary obligation is to represent only the members of a particular ... group." Ala. Leg.Black Caucus, ––– U.S. ––––, 135 S.Ct. 1257, 1265, 191 L.Ed.2d 314 (2015) (internal quotation marks omitted). But the injuries attributable to partisan gerrymanders also meaningfully differ from those associated with racial gerrymanders. For instance, partisan gerrymandering plaintiffs do not suffer the same stigmatic and dignitary harms as those suffered by racial gerrymandering plaintiffs. And partisan gerrymandering plaintiffs endure the same dilutionary harms that permit voters residing in overpopulated districts to lodge statewide challenges in one-person, one-vote cases. See Davis v. Bandemer , 478 U.S. 109, 114, 132–33, 143, 106 S.Ct. 2797, 92 L.Ed.2d 85 (1986) (plurality op.) (treating partisan gerrymandering as a form of "unconstitutional vote dilution"); id. at 173, 106 S.Ct. 2797 (Powell, J., concurring in part and dissenting in part) (same). Additionally, like one-person, one-vote challenges to congressional districting plans, partisan gerrymanders of congressional districts produce structural harms as well as personal harms. See infra Parts II.B.1, V.

As to the relative severity of racial and partisan gerrymandering claims, the Fourteenth Amendment no doubt prohibits unjustified reliance on race in districting. Shaw I , 509 U.S. at 657, 113 S.Ct. 2816. But both the Constitution and statutes enacted by Congress permit state redistricting bodies to consider race in certain circumstances. For example, Section 2(b) of the Voting Rights Act, enacted pursuant to Congress's authority to enforce the Fifteenth Amendment, requires states to ensure that members of a protected class do not have "less opportunity than other members of the electorate to ... elect representatives of their choice." 52 U.S.C. § 10301(b) ; see also Voinovich v. Quilter , 507 U.S. 146, 154, 113 S.Ct. 1149, 122 L.Ed.2d 500 (1993). To that end, a state may rely on race in drawing district lines when it has "good reasons to think that it would transgress the [Voting Rights] Act if it did not draw race-based district lines." Cooper , 137 S.Ct. at 1464 (internal quotation marks omitted). Even when the Voting Rights Act does not compel states to take into account race in drawing district lines, the Supreme Court has recognized that states have an important "interest in eradicating the effects of past discrimination," including through their redistricting plans. Shaw I , 509 U.S. at 656, 113 S.Ct. 2816. Accordingly, state legislatures involved in the "delicate task" of redistricting, see Miller , 515 U.S. at 905, 115 S.Ct. 2475, can—and, in certain circumstances, should—consider the impact of a redistricting plan on minority groups, including groups of voters previously subject to race-based discrimination. And in appropriate circumstances, states may rely on race-conscious redistricting to advance the interests of members of minority groups subject to past discrimination.

Whereas both Congress and the Supreme Court have recognized that the consideration of race in redistricting can advance constitutionally cognizable interests, Legislative Defendants offer no argument or authority, nor have we found any, identifying any legitimate state interest, let alone a constitutionally cognizable state interest, served by partisan gerrymandering—"the drawing of legislative district lines to subordinate adherents of one political party and entrench a rival party in power." Ariz. State Leg. , 135 S.Ct. at 2658. Because race-conscious redistricting, in appropriate circumstances, can advance legitimate governmental objectives, and because partisan gerrymandering does not serve any such objective, we reject Legislative Defendants' assertion "that race-based claims allege a more serious violation of the Constitution than do partisan-based claims." Leg. Defs.' FOF 113–14.

Given the differences between partisan gerrymandering and racial gerrymandering claims—and the similarities between the harms associated with partisan gerrymandering and malapportionment, particularly in the case of congressional districts—we conclude that the Supreme Court's approach to standing in one-person, one-vote cases should guide the standing inquiry in partisan gerrymandering cases. Under that approach, we find that both groups of Plaintiffs, some of whom reside in districts in which their votes have been diluted, have standing to challenge the 2016 Plan as a whole. Accord Whitford v. Gill , 218 F.Supp.3d 837, 927–28 (W.D. Wis. 2016) (three-judge panel) (concluding that partisan gerrymandering plaintiffs, who resided in a small minority of the districts established by a redistricting plan, had standing to challenge the redistricting plan as a whole), appeal docketed , 137 S.Ct. 2289 (2017).

The injuries associated with Plaintiffs' First Amendment and Article I claims also support statewide standing. Partisan gerrymandering implicates the "the First Amendment interest of not burdening or penalizing citizens because of their participation in the electoral process, their voting history, their association with a political party, or their expression of political views." Vieth v. Jubelirer , 541 U.S. 267, 314, 124 S.Ct. 1769, 158 L.Ed.2d 546 (2004) (Kennedy, J., concurring in the judgment). Among other types of "burden[s]" on First Amendment rights, partisan gerrymandering "purposely dilut[es] the weight of certain citizens' votes to make it more difficult for them to achieve electoral success because of the political views they have expressed through their voting histories and party affiliations." Shapiro v. McManus , 203 F.Supp.3d 579, 595 (D. Md. 2016) (three-judge panel). To that end, the First Amendment injury associated with partisan gerrymandering echoes the harms attributable to malapportionment. See id. (explaining that "while a State can dilute the value of a citizen's vote by placing him in an overpopulated district, a State can also dilute the value of his vote by placing him in a particular district because he will be outnumbered by those who have affiliated with a rival political party. In each case, the weight of the viewpoint communicated by his vote is ‘debased’ " (quoting Bd. of Estimate of City of N.Y. v. Morris , 489 U.S. 688, 693–94, 109 S.Ct. 1433, 103 L.Ed.2d 717 (1989) ). Partisan gerrymandering also implicates additional, non-district-specific First Amendment harms, such as infringing on the right to associate with likeminded voters to fund, attract, and elect candidates of choice. See Williams v. Rhodes , 393 U.S. 23, 30, 89 S.Ct. 5, 21 L.Ed.2d 24 (1968) (explaining that "[w]e have repeatedly held that freedom of association is protected by the First Amendment," including "the right of individuals to associate for the advancement of political beliefs"). Because the First Amendment harms attributable to partisan gerrymandering are analogous to one-person, one-vote claims and are not district-specific, we conclude that partisan gerrymandering claims under the First Amendment need not be asserted on a district-by-district basis.

The injuries underlying Common Cause Plaintiffs' Article I claims—which allege that the 2016 Plan exceeds the General Assembly's authority under the Elections Clause and usurps the power of "the People" to elect their representatives—also do not stop at a single district's lines. Rather, like the malapportionment of congressional districts, these injuries reflect structural violations amenable to statewide standing. Cf. U.S. Term Limits, Inc. v. Thornton , 514 U.S. 779, 808–09, 115 S.Ct. 1842, 131 L.Ed.2d 881 (1995) ("The Convention debates make clear that the Framers' overriding concern was the potential for States' abuse of the power to set the ‘Times, Places and Manner’ of elections."); id. at 809, 115 S.Ct. 1842 ("As Hamilton later noted: 'Nothing can be more evident than that an exclusive power of regulating elections for the national government, in the hands of the State legislatures, would leave the existence of the Union entirely at their mercy."). Indeed, malapportionment challenges to congressional districting plans, which permit statewide standing, are governed by Article I, Section 2, one of two Article I provisions under which Common Cause Plaintiffs seek relief. See Wesberry , 376 U.S. at 7–8, 84 S.Ct. 526.

Our conclusion that Plaintiffs may rely on statewide standing in pursuing their partisan gerrymandering claims also finds support in the facts and circumstances surrounding the General Assembly's drawing and enactment of the 2016 Plan. As reflected in the later-adopted Partisan Advantage criterion, Representative Lewis and Senator Rucho instructed Dr. Hofeller to draw a plan that would elect ten Republicans and three Democrats. Ex. 2043, at ¶ 38; Lewis Dep. 162:24–163:7; Hofeller Dep. 175:19–23, 178:14–20, 188:19–190:2. Representative Lewis further testified that he sought to draw a plan that elected as many Republican candidates as feasible. Ex. 1005, at 50:7–10. To achieve that statewide goal, the 2016 Plan sacrificed a number of district-specific objectives, such as preventing the pairing of all incumbents elected under the 2011 Plan, respecting the lines of political subdivisions, and further improving on the compactness of the districts in the 2011 Plan. See Ex. 2012, at 15–19; infra Part III.A.2.b. Accordingly, in drawing the 2016 Plan, the General Assembly sought to achieve a statewide partisan effect. In such circumstances, we find it appropriate to view the 2016 Plan as inflicting a statewide partisan injury.

2.

Legislative Defendants next argue that Plaintiffs, at least one of whom resides in each of the thirteen districts created by the 2016 Plan, have not suffered the injuries-in-fact necessary to assert either statewide or district-by-district challenges to the plan. In particular, Legislative Defendants maintain that none of the Plaintiffs have suffered an injury-in-fact because: (1) certain Plaintiffs conceded they were able to elect the representative of their choice and (2) certain other Plaintiffs reside in districts that since 2002 have elected only a single political party's candidates. We disagree.

To begin, the 2016 Plan diluted the votes of those Plaintiffs who supported non-Republican candidates and reside in the ten districts that the General Assembly drew to elect Republican candidates. That dilution constitutes a legally cognizable injury-in-fact. See Whitford , 218 F.Supp.3d at 927 (finding evidence that "the electoral influence of plaintiffs and other Democratic voters statewide has been unfairly and disproportionately reduced" by partisan gerrymander proved the plaintiffs' injury-in-fact).

Other Plaintiffs in the groups identified by Legislative Defendants testified to legally cognizable non-dilutionary injuries. For example, Plaintiffs in both groups testified to decreased ability to mobilize their party's base, to attract volunteers, and to recruit strong candidates. See, e.g. , Dep. of Elizabeth Evans ("Evans Dep.") 16:1–12, April 7, 2017, ECF No. 101–7; Dep. of John West Gresham ("Gresham Dep.") 38:5–18, March 24, 2017, ECF No. 101–25; Dep. of Melzer Aaron Morgan, Jr. ("Morgan Dep.") 22:16–19, 23:20–25, April 7, 2017, ECF No. 101–16; Palmer Dep. 27:19–23, 50:10–23; Dep. of Gunther Peck ("Peck Dep.") 27:8–24, 34:6–20, March 22, 2017, ECF No. 101–3; Dep. of Cheryl Taft ("C. Taft Dep.") 17:6–11, March 30, 2017, ECF No. 101–11; Dep. of Aaron J. Sarver ("Sarver Dep.") 26:9–27:23, 34:8–15, 37:24–39:4, April 10, 2017, ECF No. 101–23; Dep. of Russell Grady Walker, Jr. ("Walker Dep.") 29:17–30:8, April 7, 2017, ECF No. 101–27. Plaintiffs who live in districts that have consistently elected candidates from the same party also testified to voters feeling frozen out of the democratic process because "their vote never counts," which in turn affects voter mobilization and educational opportunities and the ability to attract strong candidates. See, e.g. , Dep. of Elliott J. Feldman ("Feldman Dep.") 27:8–22, March 24, 2017, ECF No. 101–20; Dep. of William Halsey Freeman ("Freeman Dep.") 17:17–18:10, April 7, 2017, ECF No. 101–14; Fox Dep. 29:21–30:7, 51:18–52:9; Morgan Dep. 23:2–8; Dep. of John J. Quinn, III ("Quinn Dep.") 38:1–39:5, April 10, 2017, ECF No. 101–22; C. Taft Dep. 17:6–11. The Supreme Court has recognized that these types of harms constitute cognizable injuries. See, e.g. , Anderson v. Celebrezze , 460 U.S. 780, 792, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983) (finding that plaintiff was injured by election law that made "[v]olunteers ... more difficult to recruit and retain, media publicity and campaign contributions ... more difficult to secure, and voters ... less interested in the campaign").

In sum, Plaintiffs' dilutionary and non-dilutionary injuries are sufficient to ensure the sharply adversarial presentation of issues the standing doctrine contemplates. Indeed, if partisan gerrymandering "does produce a legally cognizable injury, the[se] [Plaintiffs] are among those who have sustained it. They are asserting ‘a plain, direct and adequate interest in maintaining the effectiveness of their votes.’ " Baker , 369 U.S. at 208, 82 S.Ct. 691 (quoting Coleman v. Miller , 307 U.S. 433, 438, 59 S.Ct. 972, 83 L.Ed. 1385 (1939) ).

3.

Finally, Legislative Defendants argue that all of the organizational Plaintiffs lack standing. Specifically, Legislative Defendants assert that no organizational Plaintiff can rely on its members for standing nor has any organizational Plaintiff suffered injury in its own right sufficient to confer standing. However, our analysis above forecloses Legislative Defendants' arguments that individual members of the Plaintiff organizations lack standing. See supra Part II.A.2. And even if Plaintiff organizations could not rely on their members' injuries to establish standing, the Plaintiff organizations each have suffered additional costs and burdens due to the 2016 Plan sufficient to establish Article III standing.

The League, for example, seeks to educate voters regarding a fair and evenhanded democracy, which includes redistricting. Klenz Dep. 30:22–32:9. The 2016 Plan has required the League to increase those educational efforts and therefore forced the League to incur additional costs. Id. at 33:7–20, 59:7–60:25, 80:1–81:7. Common Cause engages in similar efforts, which in turn have required increased expenditures due to the 2016 Plan. 30(B)(6) Dep. of Common Cause by Bob Phillips ("Common Cause Dep.") 64:13–25, 66:10–22, 74:6–75:15, 149:17–150:19, April 14, 2017, ECF Nos. 101–29, 110–6. Finally, the North Carolina Democratic Party testified that the 2016 Plan has made it more difficult for the party to raise resources and to recruit candidates. See Goodwin Dep. 97:18–98:9. Taken together, these specific and direct harms to each organizational Plaintiff—stemming from the 2016 Plan and which would abate if this Court invalidated the 2016 Plan—are independently sufficient to confer standing on the Plaintiff organizations. See, e.g. , Havens Realty Corp. v. Coleman , 455 U.S. 363, 379, 102 S.Ct. 1114, 71 L.Ed.2d 214 (1982) ("[T]here can be no question that the organization has suffered injury in fact.... [C]oncrete and demonstrable injury to the organization's activities—with the consequent drain on the organization's resources—constitutes far more than simply a setback to the organization's abstract social interests."); Warth v. Seldin , 422 U.S. 490, 511, 95 S.Ct. 2197, 45 L.Ed.2d 343 (1975) ("There is no question that 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.").

* * * * *

In conclusion, we find that both the individual and organizational Plaintiffs have suffered injuries-in-fact attributable to the 2016 Plan, and, based on those injuries, Plaintiffs have standing to challenge the 2016 Plan as a whole. Even absent statewide standing, because Plaintiffs reside in each of the state's thirteen districts and have all suffered injuries-in-fact, Plaintiffs, as a group, have standing to lodge district-by-district challenges to the entire 2016 Plan.

B.

Next, Legislative Defendants argue that although partisan gerrymandering claims are justiciable "in theory," Plaintiffs' specific partisan gerrymandering claims should be dismissed because, as alleged and proven, they raise nonjusticiable political questions. Leg. Defs.' FOF 93. The political question doctrine dates to Justice Marshall's opinion in Marbury v. Madison , 5 U.S. (1 Cranch) 137, 2 L.Ed. 60 (1803), and rests on the principle that certain disputes are not appropriate for or amenable to resolution by the courts because they raise questions constitutionally reserved to the political branches, id. at 170 ("Questions, in their nature political, or which are, by the constitution and laws, submitted to the executive, can never be made in this court.").

The political question doctrine has played a central role in apportionment cases. The Supreme Court set forth its current test for determining whether a claim raises a political question in a case dealing with the justiciability of one-person, one-vote claims. See Baker v. Carr , 369 U.S. 186, 82 S.Ct. 691, 7 L.Ed.2d 663 (1962). Prior to Baker , in Colegrove v. Green , 328 U.S. 549, 66 S.Ct. 1198, 90 L.Ed. 1432 (1946), several Justices took the position that certain apportionment challenges raised political questions because the Constitution expressly delegated authority over apportionment to the States, subject to the supervision of Congress, thereby leaving no place for judicial review. Id. at 553–55, 66 S.Ct. 1198.

Baker confronted a one-person, one-vote challenge under the Equal Protection Clause to a state legislative districting plan. The Court concluded such claims were justiciable, and distinguished Colegrove on grounds that Colegrove involved a challenge under the Guaranty Clause, Article IV, Section 4, which the Court had previously held was not "the source of a constitutional standard for invalidating state action." 369 U.S. at 209–10, 223, 82 S.Ct. 691 (citing Taylor v. Beckham , 178 U.S. 548 (1900)). In concluding that one-person, one-vote apportionment claims are justiciable, Baker held that an issue poses a political question if there is:

A textually demonstrable constitutional commitment of the issue to a coordinate political department; or a lack of judicially

discoverable and manageable standards for resolving it; or the impossibility of deciding without an initial policy determination of a kind clearly for nonjudicial discretion; or the impossibility of a court's undertaking independent resolution without expressing lack of the respect due coordinate branches of government; or an unusual need for unquestioning adherence to a political decision already made; or the potentiality of embarrassment from multifarious pronouncements by various departments on one question.

Id. at 217, 82 S.Ct. 691. Applying this test, the Court concluded one-person, one-vote claims were justiciable under the Fourteenth Amendment because they involved a determination of "the consistency of state action with the Federal Constitution"—a question constitutionally assigned to the Judiciary. Id. at 226, 82 S.Ct. 691. The Court further emphasized that the resolution of the question was "judicially manageable" because "[j]udicial standards under the Equal Protection Clause are well developed and familiar, and it has been open to courts since the enactment of the Fourteenth Amendment to determine, if on the particular facts they must, that a discrimination reflects no policy, but simply arbitrary and capricious action." Id. The Court subsequently extended Baker 's justiciability holding to one-person, one-vote challenges to congressional districts under Article I, Section 2. See Wesberry , 376 U.S. at 5–6, 84 S.Ct. 526.

1.

In Davis v. Bandemer , 478 U.S. 109, 106 S.Ct. 2797, 92 L.Ed.2d 85 (1986), the Supreme Court applied the Baker framework to partisan gerrymandering claims, holding that such claims do not raise nonjusticiable political questions, see id. at 123, 106 S.Ct. 2797 (plurality op.); id. at 161–65, 106 S.Ct. 2797 (Powell, J., concurring in part and dissenting in part). Writing for the Court, Justice White emphasized that the Court had previously concluded that one-person, one-vote and racial gerrymandering claims were justiciable, thereby establishing that apportionment claims implicating "issue[s] of representation" are justiciable. Id. at 124, 106 S.Ct. 2797 (plurality op.). Justice White further stated that there was no reason to believe that the "standards ... for adjudicating this political gerrymandering claim are less manageable than the standards that have been developed for racial gerrymandering claims." Id. at 125, 106 S.Ct. 2797. Although the Court recognized the justiciability of partisan gerrymandering claims under the Equal Protection Clause, a majority could not agree as to the substantive standard for proving such claims. Compare id. at 127–37, 106 S.Ct. 2797, with id. at 161–62, 106 S.Ct. 2797 (Powell, J., concurring in part and dissenting in part).

The Court revisited the justiciability of partisan gerrymandering claims in Vieth v. Jubelirer , 541 U.S. 267, 124 S.Ct. 1769, 158 L.Ed.2d 546 (2004). Conceding "the incompatibility of severe partisan gerrymanders with democratic principles," id. at 292, 124 S.Ct. 1769 (plurality op.), a four-justice plurality nonetheless took the position that no judicially manageable standard exists to adjudicate partisan gerrymandering claims and therefore would have reversed Bandemer 's holding of justiciability, id. at 281, 124 S.Ct. 1769. Justice Kennedy agreed with the plurality that the Vieth plaintiffs had failed to put forward a legally cognizable standard for evaluating partisan gerrymandering claims, therefore warranting dismissal of the action for failure to allege "a valid claim on which relief may be granted." Id. at 306, 313, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment). But Justice Kennedy rejected the plurality's conclusion that partisan gerrymandering claims are categorically nonjusticiable. See id. at 309–10, 124 S.Ct. 1769. And the remaining four Justices agreed with Justice Kennedy's refusal to reverse Bandemer 's justiciability holding. Id. at 317, 124 S.Ct. 1769 (Stevens, J., dissenting) ("[F]ive Members of the Court ... share the view that, even if these appellants are not entitled to prevail, it would be contrary to precedent and profoundly unwise to foreclose all judicial review of similar claims that might be advanced in the future."). Two years later, the Supreme Court again refused to revisit Bandemer 's holding that partisan gerrymandering claims are justiciable. League of United Latin Am. Citizens v. Perry (LULAC ), 548 U.S. 399, 414, 126 S.Ct. 2594, 165 L.Ed.2d 609 (2006).

Accordingly, under controlling Supreme Court precedent, a challenge to an alleged partisan gerrymander presents a justiciable case or controversy. See Common Cause , 240 F.Supp.3d at 387. For good reason.

As the Supreme Court recently held, " ‘[p]artisan gerrymanders ... [are incompatible] with democratic principles.’ " Ariz. State Leg. , 135 S.Ct. at 2658 (quoting Vieth , 541 U.S. at 292, 124 S.Ct. 1769 (plurality op.)). That statement accords with the unanimous conclusion of the Justices in Vieth . See 541 U.S. at 292, 124 S.Ct. 1769 (plurality op.) (recognizing "the incompatibility of severe partisan gerrymanders with democratic principles"); id. at 312, 316–17, 124 S.Ct. 1769 (Kennedy, J., concurring) ("If a State passed an enactment that declared ‘All future apportionment shall be drawn so as most to burden Party X's rights to fair and effective representation, though still in accord with one-person, one-vote principles,’ we would surely conclude the Constitution had been violated."); id. at 326, 124 S.Ct. 1769 (Stevens, J., dissenting) ("State action that discriminates against a political minority for the sole and unadorned purpose of maximizing the power of the majority plainly violates the decisionmaker's duty to remain impartial"); id. at 345, 124 S.Ct. 1769 (Souter, J., dissenting) ("[T]he increasing efficiency of partisan redistricting has damaged the democratic process to a degree that our predecessors only began to imagine."); id. at 360, 124 S.Ct. 1769 (Breyer, J., dissenting) (holding that redistricting plan violates Constitution if it amounts to an "unjustified use of political factors to entrench a minority in power").

On its most fundamental level, partisan gerrymandering violates "the core principle of republican government ... that the voters should choose their representatives, not the other way around." Ariz. State Leg. , 135 S.Ct. at 2677 (internal quotation marks omitted); see also Powell v. McCormack , 395 U.S. 486, 540–41, 89 S.Ct. 1944, 23 L.Ed.2d 491 (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))). Put differently, partisan gerrymandering represents " ‘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 , 548 U.S. at 456, 126 S.Ct. 2594 (Stevens, J., concurring in part and dissenting in part) (quoting Balderas v. Texas , Civ. Action No. 6:01CV158, App. to Juris. Statement 209a–10a, 2001 WL 36403750 (E.D. Tex. 2006) ).

Partisan gerrymandering runs contrary to both the structure of the republican form of government embodied in the Constitution and fundamental individual rights preserved by the Bill of Rights. As detailed more fully below, partisan gerrymandering of congressional districts constitutes a structural violation because it insulates Representatives from having to respond to the popular will, and instead renders them responsive to state legislatures or political factions thereof. See infra Part V. Unlike the Senate, which, at the time of the founding, represented the interests of the States, the Framers intended for the House of Representatives to be the governmental body directly responsive to "the People." U.S. Const. art. I, § 2; see also Wesberry , 376 U.S. at 13, 84 S.Ct. 526 (explaining that "William Samuel Johnson of Connecticut had summed [the Great Compromise] up well: ‘in one branch the people, ought to be represented; in the other, the States’ "). As James Madison explained, "it is essential to liberty that the government in general should have a common interest with the people, so it is particularly essential that the [House of Representatives] should have an immediate dependence on, and an intimate sympathy with, the people. " See The Federalist No. 52 (James Madison), at 295 (Clinton Rossiter ed., 1999) (emphasis added). On this point, both the Federalists and Anti–Federalists agreed. See, e.g. , James Madison, Notes of Debates in the Federal Convention of 1787 39 (W. W. Norton & Co. 1987) (1787) (hereinafter "Debates ") (reporting that George Mason "argued strongly for an election of the larger branch by the people. It was to be the grand depository of the democratic principle of the government."); id. at 167 (reporting that James Wilson stated that he "considered the election of the first branch by the people not only as the corner Stone, but as the foundation of the fabric: and that the difference between a mediate and immediate election was immense").

Emphasizing that the House of Representatives was the repository of the People's power, the Framers repeatedly expressed concern about state legislatures, or political factions thereof, interposing themselves between Representatives and the People. For example, James Madison explained that "[i]t is essential " that a Republican government "derive[ its powers] from the great body of society, not from an inconsiderable proportion or a favored class of it ; otherwise a handful of tyrannical nobles, exercising their oppressions by a delegation of their powers, might aspire to the rank of republicans and claim for their government the honorable title of republic." The Federalist No. 39 (James Madison), at 209 (second emphasis added); Debates at 40 (reporting that James Wilson stated that "[a]ll interference between the general and local government should be obviated as much as possible"). The Framers expressed particular concern that State legislatures would seek to influence Congress by enacting electoral regulations that favored candidates aligned with, and responsive to, the interests of the legislatures, rather than the public at large. See Debates at 167 (reporting that Rufus King expressed concern that "the Legislatures would constantly choose men subservient to their own views as contrasted to the general interest; and that they might even devise modes of election that would be subversive of the end in view"). Surveying these and other founding-era authorities, the Supreme Court recognized that "[i]t would defeat the principle solemnly embodied in the Great Compromise ... to hold that, within the states, legislatures may draw the lines of congressional districts in such a way as to give some voters a greater voice in choosing a Congressman than others." Wesberry , 376 U.S. at 14, 84 S.Ct. 526. Partisan gerrymandering—drawing district lines to enhance the electoral power of voters who support a favored party and diminish the electoral power of voters who support disfavored parties—amounts to a legislative effort "to give some voters a greater voice in choosing a Congressman than others," id. , contrary to the republican system put in place by the Framers.

Partisan gerrymandering also runs afoul of rights that "are individual and personal in nature," Reynolds , 377 U.S. at 561, 84 S.Ct. 1362, because it subverts the foundational constitutional principle that the State govern "impartially"—that "the State should treat its voters as standing in the same position, regardless of their political beliefs or party affiliation." Davis , 478 U.S. at 166, 106 S.Ct. 2797 (Powell, J., concurring in part and dissenting in part); see also infra Part III. And partisan gerrymandering infringes on core political speech and associational rights by "burdening or penalizing citizens because of their participation in the electoral process, their voting history, their association with a political party, or their expression of political views." Vieth , 541 U.S. at 314, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment); see also infra Part IV.

That partisan gerrymandering encroaches on these individual rights by undermining the right to vote—the principle vehicle through which the public secures other rights and prevents government overreach—magnifies the constitutional harm. As the Supreme Court explained in Wesberry , "[o]ur Constitution leaves no room for classification of people in a way that unnecessarily abridges [the right to vote]" because "[o]ther rights, even the most basic, are illusory if the right to vote is undermined." 376 U.S. at 17–18, 84 S.Ct. 526. To that end, the Supreme Court long has held that "legislation which restricts those political processes which can ordinarily be expected to bring about repeal of undesirable legislation, is to be subjected to more exacting judicial scrutiny under the general prohibitions of the Fourteenth Amendment than are most other types of legislation." United States v. Carolene Prods. Co. , 304 U.S. 144, 152 n.4, 58 S.Ct. 778, 82 L.Ed. 1234 (1938).

A partisan gerrymander that is intended to and likely has the effect of entrenching a political party in power undermines the ability of voters to effect change when they see legislative action as infringing on their rights. And as James Madison warned, a legislature that is itself insulated by virtue of an invidious gerrymander can enact additional legislation to restrict voting rights and thereby further cement its unjustified control of the organs of both state and federal government. See Debates at 424 ("[T]he inequality of the Representation in the Legislatures of particular States, would produce like inequality in their representation in the Natl. Legislature, as it was presumable that the Counties having the power in the former case would secure it to themselves in the latter."). That is precisely what occurred in the late Eighteenth Century when Democratic legislatures used aggressive partisan gerrymanders to secure Democratic control of the House of Representatives and then, by virtue of that control, restrict earlier federal efforts to enforce the Fifteenth Amendment in the South, thereby facilitating the return of de jure and de facto segregation. See Erik J. Engstrom, Partisan Gerrymandering and the Construction of American Democracy 94–121 (2013).

The Constitution sharply curtails restrictions on electoral speech and the right to vote because, in our republican form of democracy, elected representatives in power have a strong incentive to enact legislation or policies that preserve their position, at the expense of public interest. As Justice Scalia explained, "[t]he first instinct of power is the retention of power, and, under a Constitution that requires periodic elections, that is best achieved by the suppression of election-time speech." McConnell v. Fed. Election Comm'n , 540 U.S. 93, 263, 124 S.Ct. 619, 157 L.Ed.2d 491 (2003) (Scalia, J., concurring in part and dissenting in part). Casting a vote and associating with a political party are among the most fundamental forms of "election-time speech." See Williams , 393 U.S. at 30, 89 S.Ct. 5 (recognizing "the right of individuals to associate for the advancement of political beliefs, and the right of qualified voters, regardless of their political persuasion, to cast their votes effectively"); Reynolds , 377 U.S. at 555, 84 S.Ct. 1362 ("The right to vote freely for the candidate of one's choice is of the essence of a democratic society, and any restrictions on that right strike at the heart of representative government."); Alexander Meiklejohn, The First Amendment is an Absolute , 1961 Sup. Ct. Rev. 245, 254 (1961) ("The revolutionary intent of the First Amendment is ... to deny to [the government] authority to abridge the freedom of the electoral power of the people."). Partisan gerrymandering is no different than legislative efforts to curtail other forms of election-time speech because in both cases "[p]oliticians have deep-seated incentives to bias translation of votes into seats." Engstrom, supra at 192. Accordingly, because partisan gerrymandering encroaches on individuals' right to engage in "election-time speech"—including the right to vote—allegations of partisan gerrymandering "must be carefully and meticulously scrutinized" by the judiciary. Reynolds , 377 U.S. at 562, 84 S.Ct. 1362.

Because partisan gerrymandering targets voting rights, the deference to the policy judgments of the political branches animating the political question doctrine is inapplicable. In Wesberry , the defendant state asserted that claims premised on malapportionment of congressional districts raise political questions because the Elections Clause—which empowers state "Legislatures," subject to congressional regulation, to "prescribe[ ] ... The Times, Places and Manner of holding Elections for ... Representatives"—textually commits apportionment questions to Congress and the States. 376 U.S. at 6–7, 84 S.Ct. 526. In rejecting that argument, the Supreme Court refused to "support ... a construction [of the Elections Clause] that would immunize state congressional apportionment laws which debase a citizen's right to vote from the power of courts to protect the constitutional rights of individuals from legislative destruction, a power recognized at least since our decision in Marbury v. Madison ." Id. "The right to vote is too important in our free society to be stripped of judicial protection by such an interpretation of Article I," the Court held. Id.

Further, "a textually demonstrable constitutional commitment" of authority to a coordinate branch provides the strongest basis for treating a claim as a political question. Vieth , 541 U.S. at 278, 124 S.Ct. 1769 (plurality op.) (characterizing the "textually demonstrable constitutional commitment" test as the most "importan[t] and certain[ ]" test for the existence of a political question). Given that the Supreme Court has recognized that the importance of the right to vote warrants not treating malapportionment claims as political questions, notwithstanding the alleged textual commitment of such claims in the Elections Clause, a purported lack of judicially manageable standards provides an even weaker basis for "stripp[ing] of judicial protection" the right to vote when a legislature seeks to destroy that right through partisan gerrymandering. Wesberry , 376 U.S. at 6–7, 84 S.Ct. 526.

Importantly, and contrary to Legislative Defendants' claims, the judiciary's refusal to treat alleged infringements on the right to vote—like claims of partisan gerrymandering—as political questions reflects an effort to advance the interests served by the political question doctrine, rather than usurp the role of the political branches. As the Supreme Court has explained, "[t]he voting rights cases, indeed, have represented the Court's efforts to strengthen the political system by assuring a higher level of fairness and responsiveness to the political processes, not the assumption of a continuing judicial review of substantive political judgments entrusted expressly to the coordinate branches of government." Gilligan v. Morgan , 413 U.S. 1, 11, 93 S.Ct. 2440, 37 L.Ed.2d 407 (1973). Put differently, because the judiciary jealously protects the right to vote—and thereby ensures that the People retain the means to counteract any encroachment by the political branches on substantive individual rights—the judiciary can give the political branches greater latitude to make substantive policy decisions. See John Hart Ely, Democracy and Distrust: A Theory of Judicial Review 102 (1980) (explaining that by "devoting itself instead to policing the mechanisms by which [our constitutional] system seeks to ensure that our elected representatives will actually represent," the judiciary "recognizes the unacceptability of the claim that appointed and life-tenured judges are better reflectors of conventional values than elected representatives").

In sum, partisan gerrymandering infringes on a variety of individual rights and does so by targeting the right to vote—the constitutional mechanism through which the People repel legislative encroachment on their rights. The Supreme Court has long recognized that when the Constitution preserves individual rights, courts have an obligation to enforce those rights. Marbury , 5 U.S. at 166 ("[W]here a specific duty is assigned by law, and individual rights depend upon the performance of that duty, it seems equally clear that the individual who considers himself injured, has a right to resort to the laws of his country for a remedy."). We find no basis to disregard that obligation here.

Notably, the State defendant in Reynolds made arguments against judicial oversight of state redistricting similar to those advanced by Legislative Defendants here—namely, that it is improper for courts to embroil themselves in inherently political issues and that courts lack the capability of identifying a judicially manageable standard to determine whether, and to what degree, malapportionment violates the Constitution. Rejecting each of these arguments, the Supreme Court reaffirmed the principle first recognized by Chief Justice Marshall in Marbury : "We are cautioned about the dangers of entering into political thickets and mathematical quagmires. Our answer is this: a denial of constitutionally protected rights demands judicial protection; our oath and our office require no less of us." Reynolds , 377 U.S. at 566, 84 S.Ct. 1362. Our oath and our office impose that same obligation here.

2.

Legislative Defendants nonetheless argue that, regardless of whether partisan gerrymandering claims are justiciable "in theory," this Court should dismiss Plaintiffs' claims as nonjusticiable because Plaintiffs have failed to put forth a "judicially manageable standard" for resolving their claims. Leg. Defs.' Br. 2, 11, 17; Leg. Defs.' FOF 93. Legislative Defendants argue that the analytical frameworks and empirical analyses advanced by Plaintiffs fail to provide a judicially manageable standard for three reasons. First, Legislative Defendants assert that Plaintiffs' legal frameworks and expert analyses fail to address, much less resolve, what Legislative Defendants see as the fundamental question bearing on the constitutionality of partisan gerrymandering: "how much politics is too much politics in redistricting"? Leg. Defs.' Br. 2, 9–11. Second, Legislative Defendants argue that the empirical analyses on which Plaintiffs rely—which Legislative Defendants characterize as "a smorgasbord of alleged ‘social science’ theories"—lack any constitutional basis, and instead amount to "academically inspired proposed judicial amendments to the Constitution." Id. at 2, 17. Finally, Legislative Defendants maintain that allowing the judiciary to strike down a redistricting plan as a partisan gerrymander would interfere with the political branches' decision, rendered pursuant to Congress's authority under the Election Clause, to require election of representatives from single-member districts. Id. at 13. We reject all three arguments.

a.

Legislative Defendants' assertion that any judicially manageable partisan gerrymandering framework must distinguish "reasonable" partisan gerrymandering from "too much" partisan gerrymandering rests on the premise that some degree of partisan gerrymandering—again, defined by the Supreme Court as "the drawing of legislative district lines to subordinate adherents of one political party and entrench a rival party in power," Ariz. State Leg. , 135 S.Ct. at 2658—is permissible. To justify that premise, Legislative Defendants assert that (1) historical practice indicates that the Framers viewed some amount of partisan gerrymandering as constitutionally permissible and (2) the Supreme Court repeatedly has sanctioned at least some degree of partisan gerrymandering. Neither claim is correct.

As to the historical pedigree of partisan gerrymanders, Legislative Defendants, like the plurality in Vieth , correctly note that partisan gerrymanders date to the colonial era. See Leg. Defs.' Br. 17; 541 U.S. at 274, 124 S.Ct. 1769 (plurality op.). And without question, several notorious partisan gerrymanders were drawn soon after the Founding, including the "salamander"-shaped state legislative district attributed to Massachusetts Governor Elbridge Gerry in 1812 that gave rise to the term "gerrymander." Vieth , 541 U.S. at 274, 124 S.Ct. 1769 ; Engstrom, supra at 21 ("Partisan collisions over districting pervaded the early republic, and even had antecedents in the colonial legislatures"). State legislatures gerrymandered state legislative and congressional districts to favor one party or candidate at the expense of another in a variety of ways: through the manipulation of district lines; by using regional or state-wide, multi-member districts, as opposed to single-member districts; and, most commonly, by creating districts with unequal population. Engstrom, supra at 22–23.

Neither founding-era records nor historical practice, however, supports Legislative Defendants' contention that the Framers viewed some level of partisan gerrymandering as constitutionally acceptable. Rather, "the Constitution did not contemplate the rise of political parties—indeed, it was designed to discourage their emergence—let alone the modern era's highly integrated national and state parties." Richard H. Pildes, Foreword, The Constitutionalization of Democratic Politics , 118 Harv. L. Rev. 28, 81 (2004). Given that the Framers sought to discourage the rise of political parties, there is no basis to find, as Legislative Defendants suggest, that the Framers intended to allow elected members of a political party to draw district lines so as to undermine the electoral prospects of their opposition.

On the contrary, founding-era records reflect a concerted effort by the Framers to forestall the enactment of election regulations that would favor one party or faction at the expense of others. This concern is most evident in the Framers' debates regarding whether, and to what extent, the federal government should be empowered to displace States' authority to administer and regulate congressional elections. On the one hand, James Madison argued that "the Legislatures of the States ought not to have the uncontrouled right of regulating the times places and manner of holding elections [as i]t was impossible to foresee all the abuses that might be made of the discretionary power." Debates at 423. "Whenever the State Legislatures had a favorite measure to carry, they would take care so to mould their regulations as to favor the candidates they wished to succeed ," Madison explained. Id. at 424 (emphasis added). Likewise, Alexander Hamilton argued that the federal government should have some supervisory authority over the States' regulation of elections because there was no reason to believe that "it is less probable that a predominant faction in a single State should, in order to maintain its superiority, incline to a preference of a particular class of electors , than that a similar spirit should take possession of the representatives of thirteen States, spread over a vast region, and in several respects distinguishable from each other by a diversity of local circumstances, prejudices, and interests." The Federalist No. 61, at 342 (Alexander Hamilton) (Clinton Rossiter ed., 1999) (emphasis added).

On the other hand, delegates who opposed federal intrusion on state regulation of elections saw such intrusion "as an avenue through which Congress might perpetuate itself in power or ... institute unfair at-large voting methods in the states so as to favor particular interests ." Jamal Greene, Note, Judging Partisan Gerrymanders Under the Elections Clause , 114 Yale L.J. 1021, 1036 (2005) (emphasis added). Thus, although the delegates disagreed as to whether, and to what extent, to lodge authority over the regulation of congressional elections in the federal government, they were united in their view that the Constitution should be drafted to minimize the possibility that political bodies would adopt electoral regulations that favored particular parties or factions. See Note, A New Map: Partisan Gerrymandering as a Federalism Injury , 117 Harv. L. Rev. 1196, 1201 (2004). Significantly, delegates at the Constitutional Convention sought to design the Constitution so as to prevent Congress from being plagued by "what Madison called the ‘vicious representation’ in Great Britain whereby ‘rotten boroughs’ with few inhabitants were represented in Parliament on or almost on a par with cities of greater population." Wesberry , 376 U.S. at 14–15, 84 S.Ct. 526.

Notwithstanding the Framers' efforts to prevent the formation of political parties and partisan gerrymandering, the early Nineteenth Century saw the rise of political parties, and with that rise, several notable partisan gerrymanders. Engstrom supra 21–42. But the founding generation did not view such gerrymanders as constitutionally permissible. On the contrary, such gerrymanders were widely criticized as antidemocratic. For example, the newspaper cartoon that coined the term "Gerry–Mander" described partisan redistricting as "a grievous wound on the Constitution,—it in fact subverts and changes our form of Government , which ceases to be Republican as long as an Aristocratic House of Lords under the form of a Senate tyrannizes over the People, and silences and stifles the voice of the Majority ." The Gerry–Mander, or Essex South District Formed into a Monster! , Salem Gazette, Apr. 2, 1813. Numerous other Nineteenth–Century partisan gerrymanders faced similar condemnation from politicians, the press, the judiciary, and the public. See Br. of Amici Curiae Historians in Supp. of Appellees at 23–34, Gill v. Whitford , No. 16–1161 (S.Ct. Sept. 5, 2017).

Even if founding-era practice did support Legislative Defendants' assertion that some degree of partisan gerrymandering was viewed as permissible—which it does not—long-standing, and even widespread, historical practice does not immunize governmental action from constitutional scrutiny. See Reynolds , 377 U.S. at 582, 84 S.Ct. 1362 (holding that malapportionment of state legislative districts violates Equal Protection Clause, notwithstanding that malapportionment was widespread in Nineteenth and Twentieth Centuries). That is particularly true when, as here, the legal bases for challenging the conduct were unavailable at the time of the Founding. See id. The Equal Protection Clause, which fundamentally altered the relationship between the States and the federal government, post-dates the founding era by decades. See Fitzpatrick v. Bitzer , 427 U.S. 445, 455, 96 S.Ct. 2666, 49 L.Ed.2d 614 (1976) ("There can be no doubt that this line of cases has sanctioned intrusions by Congress, acting under the Civil War Amendments, into the judicial, executive, and legislative spheres of autonomy previously reserved to the States."); Libertarian Party of Va. v. Alcorn , 826 F.3d 708, 715 (4th Cir. 2016) (Wilkinson, J.) ("Of course, the Reconstruction Amendments ... materially altered the division of labor [between the federal government and the States] established by the Framers for the regulation of elections."). Likewise, the Supreme Court did not recognize the incorporation of the First Amendment against the States through the Fourteenth Amendment until 1943. See Murdock v. Pennsylvania , 319 U.S. 105, 108, 63 S.Ct. 870, 87 L.Ed. 1292 (1943). And until the Reconstruction Congress adopted Section 1983, there was no basis for a plaintiff to challenge a congressional redistricting plan as a partisan gerrymander under Article I or any other federal constitutional provision. See The Enforcement Act of 1871, 17 Stat. 13 (1871), codified as amended at 42 U.S.C. § 1983.

Accordingly, even if some degree of partisan gerrymandering had been acceptable during the founding era, that does not mean that the ratification of the Fourteenth Amendment and the incorporation of the First Amendment against the States did not subsequently render unconstitutional the drawing of district lines to frustrate the electoral power of supporters of a disfavored party. That is precisely what the Supreme Court concluded in holding that racial gerrymandering and malapportionment violated the Constitution, notwithstanding that both practices were widespread during the Nineteenth and early Twentieth Centuries. See Reynolds , 377 U.S. at 556 n.30, 567 n.43, 84 S.Ct. 1362 ; Gomillion v. Lightfoot , 364 U.S. 339, 345–46, 81 S.Ct. 125, 5 L.Ed.2d 110 (1960).

Legislative Defendants' contention that the Supreme Court has sanctioned some degree of partisan gerrymandering—the drawing of district lines to undermine the electoral prospects of supporters of candidates of a disfavored party—fares no better. To be sure, the Supreme Court has recognized certain purposes for which a state redistricting body may take into account political data or partisan considerations in drawing district lines. For example, in appropriate circumstances, a legislature may draw district lines to avoid the pairing of incumbents. See Karcher v. Daggett , 462 U.S. 725, 740, 103 S.Ct. 2653, 77 L.Ed.2d 133 (1983). Likewise, the Supreme Court has held that a state redistricting body does not violate the 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 , 412 U.S. at 752, 93 S.Ct. 2321. And the Supreme Court has recognized that a redistricting body may draw district lines to respect political subdivisions or maintain "communities of interest." Abrams v. Johnson , 521 U.S. 74, 100, 117 S.Ct. 1925, 138 L.Ed.2d 285 (1997).

But the Supreme Court's acceptance of state legislatures' reliance on partisan considerations and political data for certain purposes does not establish that a state legislature may pursue any partisan objective, as Legislative Defendants contend. In particular, the Supreme Court has never recognized that a legislature may draw district lines for the purpose of diminishing or minimizing the voting strength of supporters of a particular party or citizens who previously voted for representatives of a particular party—the legislative action challenged here. On the contrary, the Supreme Court recently held that such efforts are "[incompatible] with democratic principles." Ariz. State Leg. , 135 S.Ct. at 2658 (alteration original); see also Reynolds , 377 U.S. at 578–79, 84 S.Ct. 1362 (condemning "[i]ndiscriminate districting, without any regard for political subdivision or natural or historical boundary lines, [as] little more than an open invitation to partisan gerrymandering " (emphasis added)). And in approving the "proportionality" gerrymander in Gaffney , the Court expressly distinguished gerrymanders that seek "to minimize or eliminate the political strength of any group or party." 412 U.S. at 754, 93 S.Ct. 2321 ; see also id. at 751, 93 S.Ct. 2321 ("A districting plan may create multimember districts perfectly acceptable under equal population standards, but invidiously discriminatory because they are employed to minimize or cancel out the voting strength of racial or political elements of the voting population." (emphasis added) (internal quotation marks omitted)). Likewise, the Supreme Court did not include burdening or punishing citizens for voting for candidates from an opposing party among its list of "legitimate" redistricting factors that justify deviating from population equality in congressional districts. See Harris v. Ariz. Indep. Redistricting Comm'n , ––– U.S. ––––, 136 S.Ct. 1301, 1306–07, 194 L.Ed.2d 497 (2016).

In sum, neither historical practice nor Supreme Court precedent supports Legislative Defendants' assertion that it is sometimes permissible for a state redistricting body to draw district lines for the purpose of burdening voters who supported or are likely to support a disfavored party or candidate. Because the Constitution does not authorize state redistricting bodies to engage in such partisan gerrymandering, a judicially manageable framework for evaluating partisan gerrymandering claims need not distinguish an "acceptable" level of partisan gerrymandering from "excessive" partisan gerrymandering. Vieth , 541 U.S. at 316, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment) (recommending against "a standard that turns on whether partisan interests in the redistricting process were excessive" because a government body is "culpable" regardless of whether it seeks to maximize its partisan advantage or "proceeds by a more subtle effort, capturing less than all the seats in each State"). Rather, the framework must distinguish partisan gerrymandering from the results of legitimate districting objectives, including those objectives that take into account political data or permissible partisan considerations. Put differently, "[a] determination that a gerrymander violates the law must rest ... on a conclusion that [political] classifications, though generally permissible, were applied in an invidious manner or in a way unrelated to any legitimate legislative objective." Id. at 307, 124 S.Ct. 1769. As explained below, we conclude that Plaintiffs' proposed legal frameworks and supporting evidence do just that.

b.

Legislative Defendants next argue that the empirical analyses introduced by Plaintiffs do not offer a judicially manageable standard for adjudicating partisan gerrymandering claims, but instead are "a smorgasbord of alleged ‘social science’ theories" that lack any constitutional basis. Leg. Defs.' Br. 2. As detailed more fully below, Plaintiffs offer two groups of empirical analyses to support their Equal Protection and First Amendment claims. The first group of analyses relies on thousands of computer-generated districting plans that conform to most traditional redistricting criteria, including those relied on by the General Assembly in drawing the 2016 Plan. According to Plaintiffs, when these plans are evaluated using the precinct-by-precinct results of recent North Carolina elections, the 2016 Plan is an "extreme statistical outlier" with regard to the degree to which it disfavors voters who oppose Republican candidates. See infra Parts III.A.2.b, III.B.2.c. Plaintiffs assert that these analyses prove that the General Assembly intended to burden voters who supported non-Republican candidates and that the 2016 Plan had the effect of burdening such voters. The second group of analyses assess the 2016 Plan's "partisan symmetry"—whether the plan allows supporters of the two principal parties to translate their votes into representation with equal effectiveness. See infra Part III.B.2.b. According to Plaintiffs, a variety of measures of the 2016 Plan's partisan symmetry reveal that, throughout the life of the plan, supporters of non-Republican candidates will likely have a significantly more difficult time translating their votes into representation.

Legislative Defendants are correct that none of these empirical analyses appear in the Constitution. But Plaintiffs need not show that a particular empirical analysis or statistical measure appears in the Constitution to establish that a judicially manageable standard exists to resolve their constitutional claims. See, e.g. , Brown v. Thomson , 462 U.S. 835, 842–43, 103 S.Ct. 2690, 77 L.Ed.2d 214 (1983) (holding that "an apportionment plan with a maximum population deviation under 10% falls within th[e] category" of "minor deviations ... from mathematical equality among state legislative districts [that] are insufficient to make out a prima facie case of invidious discrimination under the Fourteenth Amendment," notwithstanding that the plain language of the Constitution references no such statistical threshold). Rather, Plaintiffs must identify cognizable constitutional standards to govern their claims, and provide credible evidence that Defendants have violated those standards. And contrary to Legislative Defendants' assertions, Plaintiffs do not seek to constitutionalize any of the empirical analyses they have put forward to support their claims, nor does this Court do so. Instead, Plaintiffs argue that these analyses provide evidence that the 2016 Plan violates a number of well-established constitutional standards—that the government act impartially, not infringe the right to vote, and not burden individuals based on the exercise of their rights to political speech and association.

The Supreme Court long has relied on statistical and social science analyses as evidence that a defendant violated a standard set forth in the Constitution or federal law. In the context of the Equal Protection Clause, in particular, the Supreme Court has relied on statistical and social science evidence as proof that a government action was motivated by discriminatory intent or had a discriminatory effect—the same purposes for which Plaintiffs seek to use such evidence here. For example, in Yick Wo v. Hopkins , 118 U.S. 356, 6 S.Ct. 1064, 30 L.Ed. 220 (1886), the Court held that an ordinance providing a municipal board of supervisors with the discretion to grant or withhold its consent to use wooden buildings as laundries, although neutral on its face, was administered in a manner that discriminated on the basis of national origin, id. at 366, 374, 6 S.Ct. 1064. As proof, the Court noted that the board withheld consent from 200 individuals, "all of whom happen to be Chinese subjects," whereas "eighty others, not Chinese subjects, [we]re permitted to carry on the same business under similar conditions." Id. at 374, 6 S.Ct. 1064.

Likewise, in Brown v. Board of Education of Topeka , 347 U.S. 483, 74 S.Ct. 686, 98 L.Ed. 873 (1954), the Supreme Court cited numerous academic studies of the psychological impact of segregation on children and youth as evidence that "[s]eparate educational facilities are inherently unequal," and therefore violate the Equal Protection Clause, id. at 494–95, 74 S.Ct. 686 & n.11. And the Supreme Court has recognized that "[s]tatistical analyses have served and will continue to serve an important role as one indirect indicator of racial discrimination in access to service on governmental bodies." Mayor of Phila. v. Educ. Equal. League , 415 U.S. 605, 620, 94 S.Ct. 1323, 39 L.Ed.2d 630 (1974). The Court also embraced the use of statistical evidence to determine whether a governmental body was justified, under the Fourteenth Amendment, in using "race-based measures to ameliorate the effects of past discrimination." City of Richmond v. J.A. Croson Co. , 488 U.S. 469, 476–77, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989) (plurality op.); see also id. at 509, 109 S.Ct. 706 ("[E]vidence of a pattern of individual discriminatory acts can, if supported by appropriate statistical proof, lend support to a local government's determination that broader remedial relief is justified.").

The Supreme Court has relied on statistical and social science evidence in cases involving voting rights and redistricting, in particular. For example, to support their racial gerrymandering claim, the plaintiffs in Gomillion alleged that the City of Tuskegee, Alabama, redrew its municipal boundaries "to remove from the city all save only four or five of its 400 Negro voters while not removing a single white voter or resident." 364 U.S. at 341, 81 S.Ct. 125. The Court concluded that the plaintiffs alleged adequate facts to support a claim under the Equal Protection Clause, explaining that "[i]f these allegations upon a trial remain uncontradicted or unqualified, the conclusion would be irresistible, tantamount for all practical purposes to a mathematical demonstration , that the legislation is solely concerned with segregating white and colored voters." Id. (emphasis added). More recently, the Court relied on statistical analyses to strike down as unconstitutional the coverage formula in Section 4(b) of the Voting Rights Act, citing evidence that the gap between white and black voter registration percentages had fallen substantially since Congress first adopted the coverage formula in 1965, as had the percentage of proposed voting changes facing objections from the Attorney General. Shelby Cty. v. Holder , 570 U.S. 529, 133 S.Ct. 2612, 2626, 186 L.Ed.2d 651 (2013). And of particular note, in its decision holding that the 2011 Plan constituted a racial gerrymander, the Supreme Court in part relied on an expert statistical analysis—which found that the General Assembly disproportionately moved blacks into the racially gerrymandered districts, even when controlling for party registration—as proof that the General Assembly predominantly relied on race, rather than partisan considerations, in drawing district lines. Cooper , 137 S.Ct. at 1477–78.

Contrary to Legislative Defendants' assertion that Plaintiffs must identify a specific empirical test derived from the language of the Constitution to prove the existence of a judicially manageable standard to adjudicate their constitutional claims, in none of these cases did the Supreme Court hold that the particular statistical or social science analyses upon which it relied had—or had to have—constitutional pedigree, or that the plaintiff had to identify a specific empirical threshold, across which the relevant constitutional provision would be violated. For example, the Gomillion Court did not state that a statistical analysis revealing that the municipal boundary plan had fenced out, say, only 80 percent of blacks, as opposed to 99 percent, would be inadequate to establish a constitutional violation. Nor did the Court require that the plaintiffs identify the particular percentage of fenced-out blacks at which a boundary plan would violate the Equal Protection Clause. Likewise, the Brown Court did not point to any specific constitutional basis for its reliance on psychological research demonstrating the impact of segregation on children and youth, nor did it require the plaintiffs to identify a specific degree of adverse psychological impact necessary to support an Equal Protection claim. And the Shelby County Court did not require the states seeking invalidation of the coverage formula to identify a specific gap between white and black voter registration percentages or a specific percentage of proposed voting changes facing objections from the Attorney General at which Congress would be constitutionally barred from displacing the states' rights to administer elections. Rather, in all of the cases, the Supreme Court treated the empirical analyses as evidence of a violation of an established constitutional standard—that governmental entities must act impartially, that governmental entities must not invidiously discriminate based on race or national origin, that the federal government may not interfere in traditional areas of state authority absent a compelling justification, and that the federal government must have a legitimate reason for subjecting certain states to more intrusive scrutiny than others.

Contrary to Legislative Defendants' assertion, therefore, courts are not foreclosed from considering statistical analyses and " ‘social science’ theories" as evidence of a violation of a constitutional or statutory standard. Leg. Defs.' Br. 2. But that does not mean courts must blindly accept such analyses either. On the contrary, in all cases courts play an essential gatekeeping role in ensuring that an expert analysis—including each analysis introduced by Plaintiffs and Legislative Defendants—is sufficiently reliable, in that it "is based on sufficient facts or data," "is the product of reliable principles and methods," and the principles and methods have "been reliably applied ... to the facts of the case." Fed. R. Evid. 702 ; see also Daubert v. Merrell Dow Pharm., Inc. , 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993). And when, as here, the court also serves as the finder-of-fact, the court must carefully weigh empirical evidence, and discount such evidence's probative value if it fails to address the relevant question, lacks rigor, is contradicted by more reliable and compelling evidence, or is otherwise unworthy of substantial weight.

Here, in arguing that Plaintiffs' empirical evidence fails to provide a judicially manageable standard for adjudicating their claims, Legislative Defendants identify what they see as a number of specific flaws, limitations, and weaknesses of that evidence—that the partisan asymmetry measures cannot be applied in all states, that the simulated maps fail to take into account certain criteria on which the General Assembly relied, that several of the analyses rely on hypothetical election results, to name a few. Although we ultimately find these objections either unfounded or insufficiently compelling to overcome the significant probative value of the analyses, see infra Part III, these are fair criticisms. But—as evidenced by their consistent placement of "social science" in quotation marks and their characterization of Plaintiffs' evidence as "academically inspired"—Legislative Defendants' judicial manageability argument appears to rest on a more cynical objection: that we should dismiss Plaintiffs' actions as nonjusticiable simply because much of the evidence upon which Plaintiffs' rely has its genesis in academic research and is the product of an effort by scholars to apply novel, and sometimes complex, methodological approaches to address a previously intractable problem. To the extent Legislative Defendants are in fact making such an argument, it fails as a matter of both fact and law.

As a matter of fact, Legislative Defendants are correct that the application of Plaintiffs' empirical methods to redistricting, to date, has largely occurred in academic research. But see Raleigh Wake Citizens Ass'n v. Wake Cty. Bd. of Elections , 827 F.3d 333, 344 (4th Cir. 2016) (relying on analysis of hundreds of computer-simulated districting plans as evidence that population deviations in municipal districting plan were attributable to illegitimate partisan purpose rather than legitimate redistricting objectives); Whitford , 218 F.Supp.3d at 890–906 (relying on predictions of vote percentages based on historical election data, a "uniform swing analysis," and a measure of partisan asymmetry to conclude Wisconsin legislative redistricting plan adversely affected representational rights of non-Republican voters). But the empirical methods themselves have been developed and broadly applied inside and outside of academia to address a wide variety of problems. For example, Dr. Jowei Chen, a political science professor at the University of Michigan, testified that the computational algorithms and statistical theories he used in generating simulated redistricting plans to assess the partisan performance of the 2016 Plan are used by logistics companies to optimize their distribution chains. Trial Tr. II, at 25:2–24. And other empirical methods on which Plaintiffs' expert witnesses relied are broadly used by governments, the business community, and academia in a variety of other fields ranging from national defense, to public safety, to finance, and to health care. Br. Amicus Curiae Eric S. Lander in Supp. of Appellees 23–25, Gill v. Whitford , No. 16–1161 (S. Ct. Aug. 31, 2017).

To hold that such widely used, and relied upon, methods cannot provide a judicially manageable standard for adjudicating Plaintiffs' partisan gerrymandering claims would be to admit that the judiciary lacks the competence—or willingness—to keep pace with the technical advances that simultaneously facilitate such invidious partisanship and provide an opportunity to remedy it. See Vieth , 541 U.S. at 312–13, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment) (explaining that advances in technology in redistricting pose both a "threat"—because technology increases "the temptation to use partisan favoritism in districting"—and a "promise"—because "these new technologies may produce new methods of analysis that make more evident the precise nature of the burdens gerrymanders impose on the representational rights of voters and parties"). But "the Constitution forbids ‘sophisticated as well as simpleminded modes of discrimination.’ " Reynolds , 377 U.S. at 563, 84 S.Ct. 1362 (quoting Lane v. Wilson , 307 U.S. 268, 275, 59 S.Ct. 872, 83 L.Ed. 1281 (1939) ). Accordingly, the judiciary likewise has an obligation to keep pace with technological and methodological advances so it can effectively fulfill its constitutional role to police ever-more sophisticated modes of discrimination.

As a legal matter, the empirical analyses' sophistication and genesis in academic research also do not preclude this Court from concluding that Plaintiffs' claims are judicially manageable. To be sure, the statistical analyses and social science theories used by Plaintiffs' experts are more advanced than the bare descriptive statistics upon which the Supreme Court relied in Yick Wo , Gomillion , and Shelby County . But the Court has not hesitated to accept sophisticated or novel empirical methods as evidence. For example, in Thornburg v. Gingles , 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986), the Supreme Court endorsed the use of "extreme case analysis and bivariate ecological regression analysis," id. 52–53, 106 S.Ct. 2752, in determining whether an electoral district exhibits "racially polarized" voting, within the meaning of Section 2 of the Voting Rights Act, id. at 61, 106 S.Ct. 2752 (plurality op.). Notably, both forms of analysis derived from social science literature, as did the definition of "racially polarized" voting adopted by the Court. Id. at 53, 106 S.Ct. 2752 nn.20–21. Outside of the voting context, the Supreme Court has embraced new social science theories and empirical analyses to resolve a variety of constitutional and statutory disputes. See, e.g. , Leegin Creative Leather Prods., Inc. v. PSKS, Inc. , 551 U.S. 877, 881–82, 889–92, 127 S.Ct. 2705, 168 L.Ed.2d 623 (2007) (appealing to "the theoretical literature" and a variety of economic analyses to support its decision to reverse century-old precedent treating vertical price restraints as a per se violation of the Sherman Act); Utah v. Evans , 536 U.S. 452, 465, 122 S.Ct. 2191, 153 L.Ed.2d 453 (2002) (holding that Census Bureau's use of "hot-deck imputation" to conduct decennial census did not violate census statute or the Constitution, relying on the "technical literature" to determine whether hot-deck imputation constitutes "sampling"); Maryland v. Craig , 497 U.S. 836, 855, 857, 110 S.Ct. 3157, 111 L.Ed.2d 666 (1990) (appealing to "the growing body of academic literature documenting the psychological trauma suffered by child abuse victims who must testify in court" in holding that the Confrontation Clause did not categorically prohibit state laws permitting victims of child abuse to testify outside the presence of their alleged abusers).

As the judiciary's understanding and application of statistical and empirical methods have increased, it has come to appreciate that the attractive simplicity of less sophisticated methods—like the descriptive statistics relied on in Yick Wo , Gomillion , and Shelby County —comes with costs. For example, although descriptive statistics may reveal that an allegedly disfavored group of employees has a lower average salary than another group, that does not mean that the average salary difference is attributable to invidious discrimination, as the allegedly disfavored group's lower average salary may reflect a variety of nondiscriminatory reasons that can be accounted for adequately only by using more advanced statistical methods. See Tagatz v. Marquette Univ. , 861 F.2d 1040, 1044 (7th Cir. 1988) (Posner, J.) ("Correlation is not causation."); Ste. Marie v. E. R.R. Ass'n , 650 F.2d 395, 400 (2d Cir. 1981) (Friendly, J.); see also Jeffrey M. Wooldridge, Econometric Analysis of Cross Section and Panel Data § 1.1 (2002) ("Simply finding that two variables are correlated is rarely enough to conclude that a change in one variable causes a change in another.").

Advances in statistical and empirical theory and application, therefore, have the potential to allow parties, experts, and amici to provide courts with more rigorous and probative evidence, thereby decreasing the risk that courts will render a decision that later proves to have rested on an errant empirical analysis. Consequently, it makes no practical or legal sense for courts to close their eyes to new scientific or statistical methods—as Legislative Defendants implicitly suggest—to prove or disprove claims premised on established legal standards. As Justice Kennedy recognized in Vieth , "new technologies may produce new methods of analysis that make more evident the precise nature of the burdens gerrymanders impose on the representational rights of voters and parties." 541 U.S. at 312–13, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment). That is precisely what we find Plaintiffs' empirical methods have done. See infra Part III.

More fundamentally, there is no constitutional basis for dismissing Plaintiffs' claims as judicially unmanageable—not because they are irrelevant, unreliable, or incorrectly applied, but simply because they rely on new, sophisticated empirical methods that derive from academic research. The Constitution does not require the federal courts to act like Galileo's Inquisition and enjoin consideration of new academic theories, and the knowledge gained therefrom, simply because such theories provide a new understanding of how to give effect to our long-established governing principles. See Timothy Ferris, Coming of Age in the Milky Way 97–101 (1989). That is not what the founding generation did when it adopted a Constitution grounded in the then-untested political theories of Locke, Montesquieu, and Rousseau. That is not what the Supreme Court did when it recognized that advances in our understanding of psychology had proven that separate could not be equal. And that is not what we do here.

Legislative Defendants' characterization of the empirical evidence introduced by Plaintiffs' as a "smorgasbord" also suggests that Legislative Defendants view the sheer number of analyses upon which Plaintiffs' rely as rendering their claims judicially unmanageable. Leg. Defs.' Br. 2. But when a variety of different pieces of evidence, empirical or otherwise, point to the same conclusion—as is the case here—courts have greater confidence in the correctness of the conclusion because even if one piece of evidence is subsequently found infirm other probative evidence remains. See, e.g. , Strickler v. Greene , 527 U.S. 263, 293, 296, 119 S.Ct. 1936, 144 L.Ed.2d 286 (1999) (holding that exculpatory evidence withheld by government was not "material" for purposes of Brady v. Maryland , 373 U.S. 83, 83 (S.Ct. 1194, 10 L.Ed.2d 215 1963), when "there was considerable forensic and other physical evidence linking [the defendant] to the crime"). Even if none of the analyses introduced by Plaintiffs could, by itself, provide definitive evidence that the 2016 Plan constitutes an unconstitutional partisan gerrymander—which we do not necessarily believe is the case—"[a] case of discrimination can ... be made by assembling a number of pieces of evidence, none meaningful in itself, consistent with the proposition of statistical theory that a number of observations, each of which supports a proposition only weakly can, when taken as a whole, provide strong support if all point in the same direction: a number of weak proofs can add up to a strong proof." Sylvester v. SOS Children's Vills. Ill., Inc. , 453 F.3d 900, 903 (7th Cir. 2006) (Posner, J.) (internal quotation marks omitted).

In sum, Plaintiffs' reliance on academically derived, social science evidence to support their partisan gerrymandering claims does not render their claims judicially unmanageable.

c.

Finally, Legislative Defendants contend that rejecting their nonjusticiability argument would be tantamount to nullifying the political branches' decision to require representatives to be elected from single-member districts. See Leg. Defs.' Br. 13 ("[W]hat plaintiffs are asking the Court to do is sub silentio eliminate district-based congressional redistricting in North Carolina."). Again, we disagree.

Legislative Defendants are correct that, by statute, each State must "establish[ ] by law a number of districts equal to the number of Representatives to which such State is so entitled, and Representatives shall be elected only from districts so established, no district to elect more than one Representative." 2 U.S.C. § 2c. But our invalidation of the 2016 Plan as an unconstitutional partisan gerrymander in no way impacts North Carolina's authority—indeed, statutory obligation—to draw a congressional redistricting plan using single-member districts. Rather, it simply requires that the General Assembly, in drawing congressional district lines, not seek to diminish the electoral power of voters who supported or are likely to support candidates of a particular party.

Of equal significance, judicial restriction of partisan gerrymandering advances the purpose behind single-member districts, rather than undermines it. The Supreme Court long has recognized that the "basic aim" of requiring districting is to "achiev[e] ... fair and effective representations for all citizens." Reynolds , 377 U.S. at 565–66, 84 S.Ct. 1362. To that end, "[t]he very essence of districting is to produce a different—a more ‘politically fair’—result than would be reached with elections at large, in which the winning party would take 100% of the legislative seats." Gaffney , 412 U.S. at 753, 93 S.Ct. 2321. The use of districting, as opposed to elections at large, serves a number of specific beneficial purposes. For example, unlike at-large electoral systems, which in politically divided states can lead to a wholesale change in the state's congressional delegation with only a small shift in votes between parties, see Engstrom, supra at 22–28, 84 S.Ct. 526, single-member districting systems "maintain[ ] relatively stable legislatures in which a minority party retains significant representation," Vieth , 541 U.S. at 360, 124 S.Ct. 1769 (Breyer, J., dissenting). Additionally, single-member districts "diminish the need for coalition governments" and thereby "make[ ] it easier for voters to identify which party is responsible for government decision-making (and which rascals to throw out)." Id. at 357, 124 S.Ct. 1769. And single-member districts make it easier for a representative to understand the interests of her constituency and act on behalf of those interests because she serves a limited group of constituents, rather than the entire state. S. Rep. 90–291, at 28 (1967) (Individual Views of Sen. Bayh). The use of single-member districts comes with democratic costs, as well. Most notably, the stability achieved by single-member districts necessarily entails that a legislative body will be less responsive to shifts in popular will.

Recall that the Supreme Court defines "partisan gerrymandering" as "the drawing of legislative district lines to subordinate adherents of one political party and entrench a rival party in power," Ariz. State Leg. , 135 S.Ct. at 2658. Therefore, by definition, partisan gerrymandering—not judicial oversight of such gerrymandering—contravenes the purpose of district-based congressional districting because it is intended not to "achiev[e] ... fair and effective representations for all citizens," Reynolds , 377 U.S. at 565–66, 84 S.Ct. 1362 (emphasis added), and not to produce a "more ‘politically fair’ " result, Gaffney , 412 U.S. at 753, 93 S.Ct. 2321. And partisan gerrymandering undermines several of the specific benefits of single-member districts. It poses a risk that "a representative may feel more beholden to the cartographers who drew her district than to the constituents who live there." LULAC , 548 U.S. at 470, 126 S.Ct. 2594 (Stevens, J., concurring in part and dissenting in part). And by "entrenching" a party in power, Ariz. State Leg. , 135 S.Ct. at 2658, even in the face of shifting voter preferences, LULAC , 548 U.S. at 470–71, 126 S.Ct. 2594 (Stevens, J., concurring in part and dissenting in part), partisan gerrymandering makes it harder for voters "to throw the rascals out," Vieth , 541 U.S. at 357, 124 S.Ct. 1769 (Breyer, J., dissenting) (internal quotation marks omitted), magnifying the downsides to the use of single-member districts.

Not only does partisan gerrymandering contradict the purpose behind single-member districting—and enhance its drawbacks—the legislative history of Section 2c reveals that Congress did not intend for the statute to empower state legislatures to engage in partisan gerrymandering. Congress adopted the current version of the single-member district statute in 1967, in the wake of the Supreme Court's invalidation of widespread malapportionment of congressional districts in Wesberry . See S. Rep. 90–291, at 2. The draft of the statute reported out of the House required that congressional districts be "in as reasonably a compact form as the State finds practicable." Id. at 4. The House intended for the compactness requirement to reflect a "congressional policy against gerrymandering" and to "prevent gerrymandering," including gerrymandering to "attempt ‘to minimize or cancel out the voting strength of racial or political elements of the voting population.’ " Id. (emphasis added) (quoting Burns v. Richardson , 384 U.S. 73, 89, 86 S.Ct. 1286, 16 L.Ed.2d 376 (1965) ). Congress removed the compactness provision from the final version of the statute after a group of senators expressed concern that the ambiguity of the reasonableness standard would be "an invitation to gerrymander, especially to gerrymander at the expense of urban minority groups." Id. at 19 (Minority Views of Sens. Kennedy, Dodd, Hart, and Tydings). Accordingly, although legislators were divided as to whether the compactness provision would be an effective tool to combat gerrymandering, they agreed that the statute should not serve as an "invitation" to state legislatures to engage in partisan gerrymandering, as we find Legislative Defendants did here.

* * * * *

In sum, we conclude that Plaintiffs have standing to challenge each of the districts created by the 2016 Plan and the 2016 Plan as a whole. We further hold that each of Plaintiffs' claims is justiciable, and, in reaching that conclusion, we reject Legislative Defendants' argument that Plaintiffs have failed to provide this Court with a judicially manageable standard for resolving their claims.

III.

Having disposed of Legislative Defendants' standing and justiciability arguments, we now turn to Plaintiffs' claims under the Equal Protection Clause of the Fourteenth Amendment. The Equal Protection Clause prohibits a State from "deny[ing] to any person within its jurisdiction the equal protection of the laws." U.S. Const. amend XIV. Partisan gerrymandering potentially runs afoul of the Equal Protection Clause 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, 77 L.Ed.2d 614 (1983) ("The concept of equal justice under law requires the State to govern impartially."). Put differently, a redistricting plan violates the Equal Protection Clause if it "serve[s] no purpose other than to favor one segment—whether racial, ethnic, religious, economic or political —that may occupy a position of strength ... or to disadvantage a politically weak segment." Karcher , 462 U.S. at 748, 103 S.Ct. 2653 (Stevens, J. concurring).

As this Court explained in denying Defendants' motions to dismiss, the Supreme Court's splintered partisan gerrymandering decisions establish that in order to prove a prima facie partisan gerrymandering claim under the Equal Protection Clause, "a plaintiff must show both [1] discriminatory intent and [2] discriminatory effects." Common Cause , 240 F.Supp.3d at 387 (citing Bandemer , 478 U.S. at 127, 106 S.Ct. 2797 (plurality op.); id. at 161, 106 S.Ct. 2797 (Powell, J., concurring and dissenting)). Plaintiffs further propose—and we agree—that if Plaintiffs establish that the 2016 Plan was enacted with discriminatory intent and resulted in discriminatory effects, the plan will nonetheless survive constitutional scrutiny if its discriminatory effects are attributable to the state's political geography or another legitimate redistricting objective. League Br. 21; Common Cause Br. 17–19; see also Bandemer , 478 U.S. at 141–42, 106 S.Ct. 2797 (plurality op.) (recognizing justification step); cf. Whitford , 218 F.Supp.3d at 884 ("[T]he Equal Protection clause prohibit[s] a redistricting scheme which (1) is intended to place a severe impediment on the effectiveness of the votes of individual citizens on the basis of their political affiliation, (2) has that effect, and (3) cannot be justified on other, legitimate legislative grounds.").

Although the three-step framework governing partisan gerrymandering claims under the Equal Protection Clause is not in dispute, neither the Supreme Court nor the parties agree as to the standard of proof for each of those elements—or whether Plaintiffs satisfied those standards—the questions to which we now turn.

A.

The Supreme Court long has required that a plaintiff seeking relief under the Equal Protection Clause to establish that a challenged official action can "be traced to a ... discriminatory purpose." Washington v. Davis , 426 U.S. 229, 240, 96 S.Ct. 2040, 48 L.Ed.2d 597 (1976). The discriminatory purpose or intent requirement extends to Equal Protection challenges to redistricting plans, in particular, including partisan gerrymandering challenges. See, e.g. , Bandemer , 478 U.S. at 127, 106 S.Ct. 2797 (plurality op.); id. at 161, 106 S.Ct. 2797 (Powell, J., concurring in part and dissenting in part); Rogers v. Lodge , 458 U.S. 613, 617, 102 S.Ct. 3272, 73 L.Ed.2d 1012 (1982) ; see also Cooper , 137 S.Ct. at 1463 (holding that to establish a racial gerrymandering claim under the Equal Protection Clause, a plaintiff must show "that race was the predominant factor motivating the legislature's decision to place a significant number of voters within or without a particular district"). 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." Washington , 426 U.S. at 241, 96 S.Ct. 2040. Rather, "an invidious discriminatory purpose may often be inferred from the totality of the relevant facts." Id. at 242, 96 S.Ct. 2040. In determining whether an "invidious discriminatory purpose was a motivating factor" behind the challenged action, evidence that the impact of the challenged action falls "more heavily" on one group than another "may provide an important starting point." Vill. of Arlington Heights v. Metro. Hous. Dev. Corp. , 429 U.S. 252, 266, 97 S.Ct. 555, 50 L.Ed.2d 450 (1977). "Sometimes a clear pattern, unexplainable on grounds other than [invidious discrimination], emerges from the effect of the state action even when the governing legislation appears neutral on its face." Id. Likewise, "[t]he historical background of the decision" may be probative of discriminatory intent, "particularly if it reveals a series of official actions taken for invidious purposes." Id. at 267, 97 S.Ct. 555. "The specific sequence of events leading up to the challenged decision also may shed some light on the decisionmaker's purposes," including whether the legislative process involved "[d]epartures from the normal procedural sequence." Id. Additionally, "[t]he legislative or administrative history may be highly relevant, especially where there are contemporary statements by members of the decision-making body, minutes of its meetings, or reports." Id. at 268, 97 S.Ct. 555.

1.

Although the discriminatory intent requirement and the types of evidence probative of such intent are well-established, it remains unclear what level of intent a plaintiff must prove to establish a partisan gerrymandering claim. Common Cause Plaintiffs assert that the degree of partisan intent motivating the drawing of the districting plan's lines determines the level of scrutiny under which a court must review the plan. Common Cause Br. 16–18. For example, if a partisan purpose "predominated" over other legitimate redistricting criteria, then the 2016 Plan warrants strict scrutiny, Common Cause Plaintiffs maintain. Id. at 17. If partisan advantage was only "a purpose" motivating the 2016 Plan, then, according to Common Cause Plaintiffs, the plan should be reviewed under the "sliding scale" standard of review set forth in Anderson v. Celebrezze , 460 U.S. 780, 788, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983), and Burdick v. Takushi , 504 U.S. 428, 434, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992). Common Cause Br. 18. By contrast, League Plaintiffs argue that Supreme Court precedent forecloses a "predominant" or "sole" intent standard in partisan gerrymandering cases. League Br. 6. Rather, League Plaintiffs assert that a partisan gerrymandering plaintiff will meet its burden under the intent prong if it proves that the redistricting body acted with the intent to "disadvantage[e] one party's (and favor[ ] the other party's) voters and candidates." Id. at 5.

We agree with League Plaintiffs that Supreme Court precedent weighs against a "predominant intent" standard. In Bandemer , the plurality opinion did not require that a plaintiff establish the mapmakers were solely or primarily motivated by invidious partisanship, but instead required proof of "intentional discrimination against an identifiable political group." 478 U.S. at 127, 106 S.Ct. 2797. And in Vieth , the plurality expressly rejected a "predominant" intent standard as judicially unmanageable. See 541 U.S. at 284–85, 124 S.Ct. 1769 (plurality op.) (stating a "predominant motivation" requirement would not be judicially manageable because it is "indeterminate" and "vague," particularly when a plaintiff lodges a challenge to an entire plan, as opposed to a single district).

The Bandemer and Vieth pluralities' rejection of a "primary" or "predominant" intent standard accords with Equal Protection principles. In describing the intent requirement for Equal Protection claims in Arlington Heights , the Supreme Court held that a plaintiff generally need not prove that a legislature took a challenged action with the "sole," "dominant," or "primary" purpose of discriminating against an identifiable group. 429 U.S. at 265–66, 97 S.Ct. 555. The Court rejected such a heightened intent requirement because "[t]he search for legislative purpose is often elusive enough ... without a requirement that primacy be ascertained." Id. at 265, 97 S.Ct. 555 n.11 (internal citation omitted). "Legislation is frequently multipurposed: the removal of even a ‘subordinate’ purpose may shift altogether the consensus of legislative judgment supporting the statute." Id. ; see also LULAC , 548 U.S. at 417–18, 126 S.Ct. 2594 (opinion of Kennedy, J.) (rejecting partisan gerrymandering framework premised on "sole" intent requirement because legislative actions are "a composite of manifold choices," making it difficult to identify the sole or predominant motivation behind the decision).

Another question bearing on the discriminatory intent requirement is what type of intent a partisan gerrymandering plaintiff must prove. As explained above, there are a number of purposes for which a state redistricting body permissibly may rely on political data or take into account partisan considerations. See supra Part II.B.2.a. 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 partisan considerations. Rather, the plaintiff must show that the redistricting body intended to apply partisan classifications "in an invidious manner or in a way unrelated to any legitimate legislative objective." Vieth , 541 U.S. at 307, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment); id. at 339, 124 S.Ct. 1769 (Stevens, J., dissenting) (holding redistricting plan would violate Equal Protection Clause if it reflected "a naked desire to increase partisan strength"); see also Romer v. Evans , 517 U.S. 620, 632, 116 S.Ct. 1620, 134 L.Ed.2d 855 (1996) (defining an "invidious" classification as "a classification of persons undertaken for its own sake ... inexplicable by anything but animus towards the class it affects"). To that end, a plaintiff satisfies the discriminatory purpose or intent requirement by introducing evidence establishing that the state redistricting body acted with an intent to "subordinate adherents of one political party and entrench a rival party in power." Ariz. State Leg. , 135 S.Ct. at 2658.

2.

We agree with Plaintiffs that a wealth of evidence proves the General Assembly's intent to "subordinate" the interests of non-Republican voters and "entrench" Republican domination of the state's congressional delegation. In particular, we find that the following evidence proves the General Assembly's discriminatory intent: (a) the facts and circumstances surrounding the drawing and enactment of the 2016 Plan; (b) empirical analyses of the 2016 Plan; and (c) the discriminatory partisan intent motivating the 2011 Plan, which the General Assembly expressly sought to carry forward when it drew the 2016 Plan.

a.

Several aspects of the 2016 redistricting process establish that the General Assembly sought to advance the interests of the Republican Party at the expense of the interests of non-Republican voters. First , Republicans had exclusive control over the drawing and enactment of the 2016 Plan. The Committee's Republican leadership and majority denied Democratic legislators access to the principal mapdrawer, Dr. Hofeller. Ex. 1011, at 36:9–20; Ex. 1014, at 44:23–45:15; Ex. 2008. And with the exception of one small change to prevent the pairing of Democratic incumbents, Dr. Hofeller finished drawing the 2016 Plan before Democrats had an opportunity to participate in the legislative process. Additionally, all of the key votes—including the Committee votes adopting the Political Data and Partisan Advantage criteria and approving the 2016 Plan, and the House and Senate votes adopting the 2016 Plan—were decided on a party-line basis. Ex. 1008, at 12:3–7, 67:10–72:8; Ex. 1011, at 110:13–22; Ex. 1016, at 81:6–16. As the Bandemer plurality recognized, when a single party exclusively controls the redistricting process, "it should not be very difficult to prove that the likely political consequences of the reapportionment were intended." Bandemer , 478 U.S. at 129, 106 S.Ct. 2797 (plurality op.); Pope , 809 F.Supp. at 396.

Second , the legislative process "[d]epart[ed] from the normal procedural sequence." Arlington Heights , 429 U.S. at 267, 97 S.Ct. 555. Representative Lewis and Senator Rucho instructed Dr. Hofeller regarding the criteria he should follow in drawing the 2016 Plan before they had been appointed co-chairs of the Committee and before the Committee debated and adopted those criteria. Lewis Dep. 77:7–20. Indeed, Dr. Hofeller completed drawing the 2016 Plan before the Committee met and adopted the governing criteria. Id. And notwithstanding that the Committee held public hearings and received public input, Dr. Hofeller never received, much less considered, any of that input in drawing the 2016 Plan. Rucho Dep. 55:4–56:13; Hofeller Dep. 177:9–21.

Third , the plain language of the "Partisan Advantage " criterion reflects an express legislative intent to discriminate—to favor voters who support Republican candidates and subordinate the interests of voters who support non-Republican candidates. Ex. 1007 (emphasis added). Moreover, the Partisan Advantage criterion reflects an express intent to entrench the Republican supermajority in North Carolina's congressional delegation by seeking to "maintain" the partisan make-up of the delegation achieved under the unconstitutional 2011 Plan. Id. The official explanation of the purpose behind that criterion by Representative Lewis—who co-chaired the Committee and, in that capacity, developed the Adopted Criteria and oversaw the drawing of the 2016 Plan—demonstrates as much. Representative Lewis explained that "to the extent [we] are going to use political data in drawing this map, it is to gain partisan advantage." Ex. 1005 at 54; see also Ex. 1016, at 29:12–13 ("We did seek a partisan advantage in drawing the map." (Statement of Rep. Lewis)). To that end, the Partisan Advantage criterion required "draw[ing] lines so that more of the whole VTDs voted for the Republican on the ballot than they did the Democrat," he explained. Ex. 1005, 57:10–16. And Representative Lewis "acknowledge[d] freely that this would be a political gerrymander," Id. at 48:4–5—a sentiment with which Senator Rucho "s[aw] nothing wrong," Rucho Dep. 118:20–119:10.

Fourth , the process Dr. Hofeller followed in drawing the 2016 Plan, in accordance with Representative Lewis and Senator Rucho's instructions, reflected the General Assembly's intent to discriminate against voters who were likely to support non-Republican candidates. In particular, in accordance with the Political Data criterion, Dr. Hofeller used past election results—which Dr. Hofeller, Representative Lewis, and Senator Rucho agree serve as the best predictor of whether a geographic area is likely to vote for a Republican or Democratic candidate, Ex. 1016, at 30:23–31:3; Hofeller Dep. 25:1–17; Rucho Dep. 95:15–16—to create a composite partisanship variable indicating whether, and to what extent, a particular precinct was likely to support a Republican or Democratic candidate, Hofeller Dep. II 262:21–24, 267:5–6. Of particular relevance to the mapdrawers' intent to draw a plan that would favor Republicans for the remainder of the decade, Dr. Hofeller testified that he believed that because "the underlying political nature of the precincts in the state does not change," his composite partisanship variable indicated whether a particular precinct would be a "strong Democratic precinct [or Republican precinct] in every subsequent election ." Ex. 2045, at 525:14–17 (emphasis added); see also Hofeller Dep. II 274:9–12 (explaining partisan characteristics of particular VTD, as reflected in Dr. Hofeller's composite partisanship variable, are likely to "carry ... through a string of elections").

Dr. Hofeller then used the partisanship variable to assign a county, VTD, or precinct "to one congressional district or another," Hofeller Dep. 106:23–107:1, 132:14–20, and "as a partial guide" in deciding whether and where to split VTDs, municipalities, or counties, id. 203:4–5; Hofeller Dep. II 267:10–17. For example, Dr. Hofeller split—or, in redistricting parlance, "cracked"—the Democratic city of Asheville between Republican Districts 10 and 11 and the Democratic city of Greensboro between Republican Districts 6 and 13. Ex. 4066, 4068. And Dr. Hofeller drew the Districts 4 and 12 to be "predominantly Democratic," Hofeller Dep. 192:7–12, by concentrating—or "packing"—Democratic voters in Durham, Mecklenburg, and Wake Counties in those two districts, Ex. 4070, Ex. 4072.

After drawing a draft plan, Dr. Hofeller then would use his partisanship variable to assess the partisan performance of the plan on a district-by-district basis and as a whole. Id. at 247:19–23; Hofeller Dep. II 283:15–22, 284:20–285:4. Based on that review, Dr. Hofeller would convey his assessment of the partisan performance of the plan to Representative Lewis. Hofeller Dep. II 290:17–25. The evidence establishes that Representative Lewis's appraisal of the various draft plans provided by Dr. Hofeller focused on such plans' likely partisan performance. Representative Lewis admitted as much during debate on the proposed map, stating that he believed "electing Republicans is better than electing Democrats," and therefore that he "drew this map in a way to help foster" the election of Republican candidates. Ex. 1016, at 34:21–23. And Representative Lewis testified that when he assessed the draft plans, "[n]early every time" he used the results from North Carolina's 2014 Senate race between Senator Thom Tillis and former Senator Kay Hagan to evaluate the plans' partisan performance in "future elections." Lewis Dep. 63:9–64:17.

b.

We also find that empirical evidence reveals that the 2016 Plan "bears more heavily on [supporters of candidates of one party] than another." Washington , 426 U.S. at 242, 96 S.Ct. 2040. In particular, two empirical analyses introduced by Plaintiffs demonstrate that the pro-Republican partisan advantage achieved by the 2016 Plan cannot be explained by the General Assembly's legitimate redistricting objectives, including legitimate redistricting objectives that take into account partisan considerations.

Dr. Jonathan Mattingly, a mathematics and statistics professor at Duke University and an expert in applied computational mathematics, drew an ensemble of 24,518 simulated districting plans from a probability distribution of all possible North Carolina congressional redistricting plans. Ex. 3002, at 9–10. To create the ensemble, Dr. Mattingly programmed a computer first to draw a random sample of more than 150,000 simulated plans using a Markov chain Monte Carlo algorithm—a widely employed statistical method used in a variety of settings —that randomly perturbed the lines of an initial districting plan to generate successive new plans. Id. at 13–15. The computer algorithm then eliminated from the 150,000 plan sample all "unreasonable" districting plans—plans with noncontiguous districts, plans with population deviations exceeding 0.1 percent, plans that were not reasonably compact under common statistical measures of compactness, plans that did not minimize the number of county and VTD splits, and plans that did not comply with the Voting Rights Act —yielding the 24,518–plan ensemble. Id. at 15–17. The criteria Dr. Mattingly used to eliminate "unreasonable" plans from his sample reflect traditional redistricting criteria, see Harris , 136 S.Ct. at 1306 (recognizing compactness, contiguity, maintaining integrity of political subdivisions, and, potentially, compliance with the Voting Rights Act, as "legitimate" considerations for deviations from population equality in state redistricting plans), and nearly all non-partisan criteria adopted by the Committee, see Ex. 1007.

After constructing the 24,518–plan ensemble, Dr. Mattingly analyzed the partisan performance of the 2016 Plan relative to the plans in his ensemble using precinct-level actual votes from North Carolina's 2012 and 2016 congressional elections. Dr. Mattingly's analysis, therefore, "assumed that the candidate does not matter, that a vote for the Democrat or Republican will not change, even after the districts are rearranged." Ex. 3002, at 23. Dr. Mattingly found that 0.36 percent (89/24,518) of the plans yielded a congressional delegation of 9 Republicans and 4 Democrats—the outcome that would have occurred under the 2016 Plan—when he evaluated the ensemble using actual 2012 votes. Id. at 3; Ex. 3040, at 7. The ensemble most frequently yielded plans that would have elected 7 (39.52%) or 6 (38.56%) Republicans. Ex. 3002, at 4; Ex. 3040, at 7. Using actual 2016 congressional votes, a congressional delegation of 10 Republicans and 3 Democrats—the outcome that occurred under the 2016 Plan—occurred in less than 0.7 percent of the simulated plans (162/24,518), with a delegation of 8 Republicans and 5 Democrats occurring in approximately 55 percent of the plans. Ex. 3040, at 19. Put differently, using both actual 2012 or 2016 votes, more than 99 percent of the 24,518 simulated maps produced fewer Republican seats than the 2016 Plan. Trial Tr. I, at 35:9–10.

Dr. Mattingly's analysis of the simulated plans also demonstrated that the General Assembly "cracked" and "packed" Democratic voters. Dr. Mattingly ordered the 13 congressional districts in each of the 24,518 simulated plans from lowest to highest based on the percentage of Democratic votes that would have been cast in the districts in the 2012 and 2016 elections. Ex. 3002, at 5–7. When analyzed using the results of both the 2012 and 2016 election, the medians of the Democratic vote share in each of the 13 districts "form a relatively straight, gradually increasing line from the most Republican district ... to the most Democratic." Id. at 7; Ex. 3040, at 12. An identical plot of the Democratic vote percentages under a plan drawn by a bipartisan commission of former judges took on the same linear form. Id. at 13.

By contrast, when Dr. Mattingly conducted the same analysis using the 2016 Plan, he found that the line connecting the medians of the Democratic vote share in each of the 13 districts took on an "S-shaped" form, which Dr. Mattingly characterized as "the signature of gerrymandering," because the 2016 Plan places "significantly more Democrats in the three most Democratic districts and fairly safe Republican majorities in the first eight most Republican districts." Ex. 3002, at 8; Ex. 3040, at 18, 30, 39; Trial Tr. I, 35:19–22 ("[T]here were clearly many, many more Democrats packed into those Democratic districts [in the 2016 Plan]; and on the other hand, that allowed there to be many more Republicans in the next group of districts."). Using 2012 votes, for example, the percentage of votes cast for Democratic candidates in the three most Democratic districts in the 2016 Plan was significantly higher than the percentage of votes casts for Democratic candidates in the three most Democratic districts in the 24,518 plan sample, and the percentage of votes cast for Democratic candidates in the sixth through tenth most Republican districts was significantly lower than in the equivalent districts in the sample. Ex. 3040, at 15–19; Trial Tr. I, at 60:6–23 (describing the sixth through thirteenth most Republican districts in 2016 Plan as "extreme outliers" relative to the simulated plans). Dr. Mattingly found the same pattern of packing Democratic voters in the three most Democratic districts when he used the votes from the 2016 election. Ex. 3040, at 27–30.

To determine whether the 2016 Plan's pro-Republican bias could have resulted from chance, Dr. Mattingly analyzed how "slight[ ]" changes in the boundaries of the districts in the 2016 Plan impacted the plan's partisan performance. Trial Tr. I, at 36:3–12. That analysis found that "when [he] shifted just as little as 10 percent of the boundary," the new map produced a "very, very different" partisan result that was "[m]uch, much less advantageous to Republicans." Id. Dr. Mattingly performed a number of additional analyses to validate his results by assessing their sensitivity to changes in his model—including seeking to reduce the number of county splits in his sample, reducing the population deviation threshold, and altering the compactness threshold—all of which confirmed the robustness of his results. Ex. 3040, at 35–38; Trial Tr. I, at 83:23–84:1, 85:9–20, 85:21–86:24.

Based on his principal analyses and sensitivity and robustness tests, Dr. Mattingly concluded that the 2016 Plan is "heavily gerrymandered" and "dilute[s] the votes" of supporters of Democratic candidates. Ex. 3002, at 9. He further concluded that the General Assembly could not "have created a redistricting plan that yielded [the pro-Republican] results [of the 2016 Plan] unintentionally." Trial Tr. I, at 62:9–12; see also id. at 73:8–9 (stating the pro-Republican partisan results of the 2016 Plan, when analyzed using 2016 votes, "would be essentially impossible to generate randomly"); id. at 92:24–93:8 (opining that 2016 Plan was "specifically tuned" to achieve a pro-Republican "partisan advantage"). And Dr. Mattingly further opined "that it's extremely unlikely that one would have produced maps that had that level of packing here and that level of depletion [of Democratic votes] here unintentionally or using nonpartisan criteria." Id. at 71:24–72:2.

We find that Dr. Mattingly's analyses, which he confirmed through extensive sensitivity testing, provide strong evidence that the General Assembly intended to subordinate the interests of non-Republican voters and entrench the Republican Party in power. In particular, given that 99 percent of Dr. Mattingly's 24,518 simulated plans—which conformed to traditional redistricting criteria and the non-partisan criteria adopted by the Committee—would have led to the election of at least one additional Democratic candidate, we agree with Dr. Mattingly's conclusion that the 2016 Plan's pro-Republican bias is not attributable to a legitimate redistricting objective, but instead reflects an intentional effort to subordinate the interests of non-Republican voters. Dr. Mattingly's analysis that the packing and cracking of non-Republican voters had to have been the product of an intentional legislative effort reinforces that conclusion. And Dr. Mattingly's finding that the 2016 Plan produced "safe Republican majorities in the first eight most Republican districts," Ex. 3002, at 8, shows that the General Assembly intended for the partisan advantage to persist. That the 2016 Plan's intentional pro-Republican bias exists when Dr. Mattingly used the actual votes from both 2012 (a relatively good year for Democrats) and 2016 (a relatively good year for Republicans) also speaks to the imperviousness of the 2016 Plan's partisan advantage to changes in candidates and the political environment.

Dr. Jowei Chen, a political science professor at the University of Michigan and expert in political geography and redistricting, also evaluated the 2016 Plan's partisan performance relative to simulated districting plans. Trial Tr. I, at 157:2–4. But rather than creating a representative ensemble of districting plans by randomly perturbing an initial plan, as Dr. Mattingly did, Dr. Chen created a computer algorithm to draw three random sets of 1,000 simulated districting plans that comply with specific criteria. Ex. 2010, at 2. To determine "whether the distribution of partisan outcomes created by the [2016 Plan] could have plausibly emerged from a non-partisan districting process," id. at 4, Dr. Chen, like Dr. Hofeller, then analyzed the partisan performance of the 2016 Plan relative to the plans in his three 1,000–plan samples using precinct-level election results, id. at 9. Unlike Dr. Hofeller, who used results from North Carolina's 2012 and 2016 congressional elections, Dr. Chen used two equally-weighted averages of precinct-level votes cast in previous statewide elections: (1) the seven statewide elections Dr. Hofeller included in his composite partisanship variable and (2) the twenty elections included in the Committee's Political Data criterion. Id. at 9–10. As the Fourth Circuit explained, "Dr. Chen's computer simulations are based on the logic that if a computer randomly draws [1,000] redistricting plans following traditional redistricting criteria, and the actual enacted plan[ ] fall[s] completely outside the range of what the computer has drawn [in terms of partisanship], one can conclude that the traditional criteria do not explain that enacted plan." Raleigh Wake Citizens Ass'n , 827 F.3d at 344.

Dr. Chen programmed the computer to draw the first set of districting plans to follow what he deemed to be the non-partisan criteria included in the Committee's Adopted Criteria: population equality, contiguity, minimizing county and VTD splits, and maximizing compactness. Ex. 2010, at 6. The 1,000 simulated plans generated by the computer split the same or fewer counties and VTDs as the 2016 Plan and significantly improved the compactness of the 2016 Plan under the Reock and Popper–Polsby measures of compactness adopted by the Committee. Id. at 6–7. Dr. Chen found that none of the 1,000 plans yielded a congressional delegation of 10 Republicans and 3 Democrats—the outcome that would have occurred under the 2016 Plan—when he evaluated the sample using Dr. Hofeller's seven-election average. Id. at 13–14. The sample most frequently yielded plans that would have elected 6 (32.4%) or 7 (45.6%) Republicans. Id . at 13. Using the results of the twenty elections referenced in the Adopted Criteria, a congressional delegation of 10 Republicans and 3 Democrats—the outcome that would have occurred under the 2016 Plan—again occurred in none of the simulated plans, with a delegation of 6 (52.5%) Republicans occurring most frequently. Id. Based on these results, Dr. Chen concluded that "the [2016 Plan] is an extreme partisan outlier when compared to valid, computer-simulated districting plans" and that the Committee's "partisan goal—the creation of 10 Republican districts—predominated over adherence to traditional districting criteria." Id. at 10–11.

To test whether the Committee's goal of protecting incumbents called into question the validity of his results, Dr. Chen next programmed his computer to draw maps that adhered to the requirements it used to draw the first set of simulated maps, and also to not pair in a single district any of the 13 incumbents elected under the 2011 Plan. Id. at 15. By comparison, the 2016 Plan paired 2 of the 13 incumbents elected under the 2011 Plan. Id. Like the first set of simulations, the second set of simulated plans split the same or fewer counties and VTDs as the 2016 Plan and improved the compactness of the 2016 Plan under the Reock and Popper–Polsby measures. Id. at 18. Dr. Chen again found that none of the 1,000 plans yielded a congressional delegation of 10 Republicans and 3 Democrats—the outcome that would have occurred under the 2016 Plan—when he evaluated the sample using Dr. Hofeller's seven-election average. Id. at 16–17. A majority of the plans included in the sample (52.9%) would have elected 7 Republicans. Id. at 16. Using twenty elections in the Adopted Criteria, a congressional delegation of 10 Republicans and 3 Democrats again occurred in none of the simulated plans, with a delegation of 6 (50.3%) or 7 (30.6%) Republicans occurring most frequently. Id. Based on these results, Dr. Chen concluded that the General Assembly's desire to avoid pairing incumbents did not explain the 2016 Plan's pro-Republican partisan advantage. Id. at 18–19.

To further test the validity of his results, Dr. Chen's third set of simulations sought to match the number of split counties (13) and paired incumbents (2) in the 2016 Plan, rather than minimize such criteria. Id. at 19–20. Adhering to these characteristics of the 2016 Plan did not meaningfully alter Dr. Chen's results. In particular, he again found that none of the 1,000 plans yielded a congressional delegation of 10 Republicans and 3 Democrats—the outcome that would have occurred under the 2016 Plan—when he evaluated the sample using Dr. Hofeller's seven-election average. Id. at 21–22. A majority of the plans included in the sample (53%) would have elected 7 Republicans. Id. at 21. Using the twenty elections in the Adopted Criteria, a congressional delegation of 10 Republicans and 3 Democrats again occurred in none of the simulated plans, with a delegation of 6 Republicans and 7 Democrats occurring most frequently (52.3%). Id. Based on these results, Dr. Chen concluded that the General Assembly's decision not to minimize the number of county splits or paired incumbents could not "have justified the plan's creation of a 10–3 Republican advantage." Id. at 20.

Analyzing the results of his three simulation sets as a whole, Dr. Chen concluded that the 2016 Plan "is an extreme statistical outlier in terms of its partisanship." Trial Tr. I, at 213:22–23. He further concluded "that the pursuit of that partisan goal ... of creating a ten Republican map, not only predominated [in] the drawing of the map, but it subordinated the nonpartisan portions of the Adopted Criteria," including the goals of increasing compactness and avoiding county splits. Trial Tr. I, at 158:20–159:2.

Like Dr. Mattingly's analyses, we find that Dr. Chen's analyses provide compelling evidence that the General Assembly intended to subordinate the interests of non-Republican voters in drawing the 2016 Plan. In particular, we find it significant that none of the 3,000 simulated districts plans generated by Dr. Chen's computer algorithm, which conformed to all of the traditional nonpartisan districting criteria adopted by the Committee, produced a congressional delegation containing 10 Republican and 3 Democrats—the result the General Assembly intended the 2016 Plan to create, and the result the 2016 Plan in fact created. That the 2016 Plan continued to be an "extreme statistical outlier" in terms of its pro-Republican tilt under three separate specifications of criteria for drawing the simulated plans reinforces our confidence that Dr. Chen's conclusions reflect stable and valid results.

Legislative Defendants raise two objections to Dr. Mattingly's and Dr. Chen's analyses, neither of which we find undermines the persuasive force of their conclusions. To begin, Legislative Defendants assert that Dr. Mattingly's and Dr. Chen's analyses rest on the "baseless assumption" that "voters vote for the party, and not for individual candidates." Leg. Defs.' Br. 10–11. Although we agree that the quality of individual candidates may impact, to a certain extent, the partisan vote share in a particular election, we do not find that this assumption undermines the probative force of the two simulation analyses, and for several reasons.

To begin, we find it significant that Dr. Mattingly and Dr. Chen used four different sets of actual votes—2012 and 2016 congressional votes in Dr. Mattingly's case and the seven– and twenty-statewide race averages in Dr. Chen's case—and reached essentially the same conclusion. As Legislative Defendants' expert in congressional elections, electoral history, and redistricting Sean Trende acknowledged, Trial Tr. III, at 30:14–15, the sets of votes used by Dr. Mattingly and Dr. Chen included elections in which Republican candidates performed well and elections in which Democratic candidates performed well, Ex. 5101, at 25, 36 (describing 2008 election as a "Democratic wave" and 2010 election as a "Republican wave"). The twenty-race average used by Dr. Chen, in particular, encompassed forty race/candidate combinations occurring over four election cycles, meaning that it reflected a broad variety of candidates and electoral conditions. Given that Dr. Mattingly and Dr. Chen reached consistent results using data reflecting numerous candidates and races—and confirmed those results in numerous sensitivity analyses—we believe that the strength or weakness of individual candidates does not call into question their key findings. That Dr. Chen found that the 2016 Plan produced a 10–Republican, 3–Democrat delegation using Dr. Hofeller's seven-race average and the twenty-race average derived from the Adopted Criteria—the same partisan make-up as the congressional delegation elected by North Carolina voters in the 2016 race—further reinforces our confidence that Dr. Mattingly and Dr. Chen's assumption regarding the partisan behavior of voters did not materially impact their results.

Second, Dr. Chen investigated the reasonableness of the assumption Legislative Defendants challenge by analyzing his set of simulated districting plans using VTD-specific predicted Republican and Democratic vote shares generated by a regression model. Ex. 2010, at 26–31. The regression model controlled for incumbency and turnout, factors correlated with candidate quality and electoral conditions. Id. at 27. Dr. Chen found that even when controlling for incumbency and turnout on a VTD-by-VTD basis, over 67 percent of his simulated maps yielded a congressional delegation of 7 Republicans and 6 Democrats, and none of his maps produced a delegation of 10 Republicans and 3 Democrats—the outcome the 2016 Plan would have produced. Id. at 36. Based on that finding, Dr. Chen reaffirmed his conclusion that the 2016 Plan "could have been created only through a process in which the explicit pursuit of partisan advantage was the predominant factor." Id. at 30.

Third, and most significantly, Dr. Mattingly's and Dr. Chen's assumption that Legislative Defendants characterize as "baseless"—that the partisan characteristics of a particular precinct do not materially vary with different candidates or in different races—is the same assumption on which the Committee, Representative Lewis, Senator Rucho, and Dr. Hofeller relied in drawing the 2016 Plan. As Dr. Hofeller—who has been involved in North Carolina redistricting for more than 30 years, Ex. 2045, at 525:6–10—testified: "[T]he underlying political nature of the precincts in the state does not change no matter what race you use to analyze it ." Ex. 2045, at 525:9–10 (emphasis added); Hofeller Dep. 149:5–18. "So once a precinct is found to be a strong Democratic precinct, it's probably going to act as a strong Democratic precinct in every subsequent election. The same would be true for Republican precincts." Ex. 2045, at 525:14–17; see also Hofeller Dep. II 274:9–12 ("[I]ndividual VTDs tend to carry ... the same characteristics through a string of elections ." (emphasis added)). Representative Lewis, Senator Rucho, and the Committee agreed with Dr. Hofeller that, at least in North Carolina, past election results serve as the best predictor of whether, and to what extent, a particular precinct will favor a Democratic or Republican candidate, Ex. 1016, at 30:23–31:2; Rucho Dep. 95:15–16, and therefore directed Dr. Hofeller to use past election results to draw a plan that would elect 10 Republicans and 3 Democrats, see Ex. 1007. And Dr. Hofeller, Representative Lewis, and the rest of the Committee relied on past election results—the same election results upon which Dr. Chen relied—in evaluating whether the 2016 Plan achieved its partisan objective. Ex. 1017 (spreadsheet Representative Lewis presented to the Committee, immediately before it voted to approve the 2016 Plan, showing the partisan performance of the plan using votes cast in twenty previous statewide elections).

Importantly, the past election results upon which both Dr. Hofeller and Representative Lewis relied to assess the 2016 Plan involved different candidates—a composite of seven statewide races in Dr. Hofeller's case and the results of the 2014 Tillis–Hagan Senate race in Representative Lewis' case—than those who ran in the 2016 congressional elections. Legislative Defendants and the expert mapdrawer they employed, therefore, believed that Dr. Mattingly's and Dr. Chen's allegedly "baseless" assumption was sufficiently reasonable, at least in the case of North Carolina, to rely on it to draw the 2016 Plan. Likewise Legislative Defendants' expert in American politics and policy, southern politics, quantitative political analysis, and election administration, Dr. M.V. Hood, III, conceded that he relied on the same assumption in assessing the likely partisan performance of the districts created by the 2016 Plan. Trial Tr. IV, at 11:8–12, 71:1–15 (acknowledging that by averaging partisan results of past elections with different candidates, as Dr. Hofeller and Dr. Chen did, "candidate effects are going to average out so we'll get a pretty good fix on what the partisan composition of an area is"). In such circumstances, we cannot say that that assumption calls into question the significant probative force of Dr. Mattingly's and Dr. Chen's analyses, particularly given how extreme a partisan outlier the 2016 Plan was in each of the two analyses.

Legislative Defendants next contend that both sets of simulated maps fail to account for a number of criteria implicitly relied upon by the General Assembly, including: that more populous, rather than less populous counties should be divided; that the "core" of the 2011 Plan districts should be retained; that a district line should not traverse a county line more than once; and that, to ensure compliance with the Voting Rights Act, one district should have a BVAP of at least 42 percent and another should have a BVAP of at least 35 percent. Leg. Defs.' FOF 78–86.

None of these alleged criteria were among the seven criteria adopted by the Committee, Ex. 1007, nor are any of these criteria mentioned in the legislative record. Additionally, both the Adopted Criteria and the legislative record expressly contradict the purported BVAP threshold criterion, as the Adopted Criteria state that "[d]ata identifying the race of individuals or voters shall not be used in the construction or consideration of districts," Ex. 1007 (emphasis added), and Representative Lewis and Dr. Hofeller repeatedly disclaimed any reliance on race or effort to preserve BVAP percentages in the 2016 Plan, see, e.g. , Ex. 1016 at 62:9–20; Hofeller Dep. 145:9–12, 146:4–146:8, 183:22–184:8. And even if the General Assembly had implicitly adopted a BVAP threshold criterion—which the record proves it did not—Dr. Mattingly's analysis accounted for that criterion by requiring that any simulated plan included in his final ensemble include one district with a BVAP of at least 40 percent and a second district with a BVAP of at least 33.5 percent. Trial Tr. I, at 41:23–25

The only two of the alleged implicit criteria that find any support in the record of this case—the alleged criteria requiring preservation of the "cores" of the districts in the 2011 Plan and the division of populous counties—are criteria that would serve to advance the General Assembly's invidious partisan objective. By preserving the "cores" of the districts in the 2011 Plan, the General Assembly perpetuated the partisan effects of a districting plan expressly drawn "to minimize the number of districts in which Democrats would have an opportunity to elect a Democratic candidate." Hofeller Dep. 127:19–22. And the alleged criterion requiring division of populous counties—which is referenced in a single line of an affidavit provided by Dr. Hofeller after the trial, see Ex. 5116, at 5—effectively required "cracking" areas of Democratic strength because more populous counties tend to be Democratic whereas less populous counties tend to be Republican. This is precisely what the 2016 Plan did by dividing populous Democratic counties like Buncombe and Guilford. Exs. 4066, 4068. Given that most of these alleged implicit criteria have no support in the record and the remaining purported criteria work hand-in-hand with the General Assembly's partisan objective, the omission of these purported criteria from Dr. Mattingly's and Dr. Chen's analyses does not in any way call into question the persuasive force of their results.

c.

Finally, although we find the facts and analyses specifically relating to the 2016 Plan sufficient, by themselves, to establish the General Assembly's discriminatory intent, we further note that evidence regarding the drawing and adoption of the 2011 Plan also speaks to the General Assembly's discriminatory intent in drawing and enacting the 2016 Plan . Typically, it would be improper for a court to rely on evidence regarding a different districting plan in finding that a redistricting body enacted a challenged plan with discriminatory intent. The "Partisan Advantage" criterion proposed by the Chairs and adopted by the Committee, however, expressly sought to carry forward the partisan advantage obtained by Republicans under the unconstitutional 2011 Plan. Ex. 1007 ("The Committee shall make reasonable efforts to construct districts in the 2016 ... Plan to maintain the current partisan makeup of North Carolina's congressional delegation."). Accordingly, to the extent invidious partisanship was a motivating purpose behind the 2011 Plan, the Committee expressly sought to carry forward—and thereby entrench—the effects of that partisanship.

As with the 2016 Plan, Republicans exclusively controlled the drawing and adoption of the 2011 Plan. The 2011 redistricting effort coincided with the RSLC's REDMAP, in which Dr. Hofeller participated and which sought to "solidify conservative policymaking at the state level and maintain a Republican stronghold in the U.S. House of Representatives for the next decade ." Ex. 2015, at ¶ 10; Ex. 2026, at 1 (emphasis added). As chairs of the committees responsible for drawing the 2011 Plan, Representative Lewis and Senator Rucho's "primary goal" was "to create as many districts as possible in which GOP candidates would be able to successfully compete for office." Hofeller Dep. 123:1–7. Defendants conceded as much in the Harris litigation, in which Dr. Hofeller stated in an expert report that "[p]olitics was the primary policy determinant in the drafting of the ... [2011] Plan." Ex. 2035, at ¶ 23.

To effectuate the General Assembly's partisan intent, Dr. Hofeller drew the 2011 Plan "to minimize the number of districts in which Democrats would have an opportunity to elect a Democratic candidate." Hofeller Dep. 127:19–22 (emphasis added). In particular, Dr. Hofeller "concentrat[ed]" Democratic voters in three districts, Ex. 2043, at 33–34, and thereby "increase[d] Republican voting strength" in five new districts, Hofeller Dep. 116:19–117:25. Notably, the three districts in the 2011 Plan that elected Democratic candidates were the same three districts in the 2016 Plan that elected Democratic candidates, and the ten districts in the 2011 Plan that elected Republican candidates were the same ten districts in the 2016 Plan that elected Republican candidates. Exs. 1018–19. Accordingly, the 2016 Plan carried forward the invidious partisan intent motivating the 2011 Plan.

3.

Legislative Defendants nonetheless argue that the General Assembly failed to act with the requisite discriminatory intent for two reasons: (1) the General Assembly did not seek to "maximize partisan advantage" and (2) the General Assembly adhered to a number of "traditional redistricting criteria," such as compactness, contiguity, and equal population. Neither argument, however, calls into question our finding that Plaintiffs satisfied their burden as to the discriminatory intent requirement.

Legislative Defendants' reliance on the General Assembly's purported lack of intent to "maximize partisan advantage" fails as a matter of both law and fact. As a matter of law, Legislative Defendants cite no authority, controlling or otherwise, stating that a governmental body must seek to "maximize" partisan advantage in order to violate the Equal Protection Clause. To be sure, the Supreme Court has indicated that evidence that a legislative body sought to maximize partisan advantage would prove that the legislature acted with discriminatory intent. See Gaffney , 412 U.S. at 751, 93 S.Ct. 2321 ("A districting plan may create multimember districts perfectly acceptable under equal population standards, but invidiously discriminatory because they are employed to ‘minimize or cancel out the voting strength of racial or political elements of the voting population.’ " (quoting Fortson v. Dorsey , 379 U.S. 433, 439, 85 S.Ct. 498, 13 L.Ed.2d 401 (1965) ); Vieth , 541 U.S. at 316, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment) ("If a State passed an enactment that declared ‘All future apportionment shall be drawn so as most to burden Party X's rights to fair and effective representation, though still in accord with one-person, one-vote principles,’ we would surely conclude the Constitution had been violated.").

That does not mean, however, that to establish a constitutional violation a plaintiff must prove that a districting body sought to maximize partisan advantage. The Supreme Court does not require that a redistricting plan maximally malapportion districts for it to violate the one-person, one-vote requirement. Nor does the Supreme Court require that a redistricting plan maximally disadvantage voters of a particular race to constitute an unconstitutional racial gerrymander. And in the context of partisan gerrymandering, in particular, Justice Kennedy has rejected a "maximization" requirement, explaining that a legislature is "culpable" regardless of whether it engages in an "egregious" and "blatant" effort to "capture[ ] every congressional seat" or "proceeds by a more subtle effort, capturing less than all seats." Vieth , 541 U.S. at 316, 124 S.Ct. 1769.

Another basis for not imposing a maximization requirement is that, in the context of a partisan gerrymander, what constitutes "maximum partisan advantage" is elusive, and turns on political strategy decisions that courts are ill suited to render. A party may not seek to maximize the number of seats a redistricting plan could allow it to win in a particular election because, by spreading out its supporters across a number of districts to achieve such a goal, its candidates would face a greater risk of losing either initially or in subsequent elections. See Bernard Grofman & Thomas Brunnell, The Art of the Dummymander , in Redistricting in the New Millennium 192–93 (Peter F. Galderisi ed., 2005) (finding, for example, that North Carolina's 1991 decennial redistricting plan, which was drawn by a Democrat-controlled General Assembly, created districts with sufficiently narrow margins in favor of expected Democratic voters that Republicans were able capture seats later in the decade). Accordingly, different partisan redistricting bodies may have different perspectives on what constitutes maximum partisan advantage.

As a matter of fact, Plaintiffs presented compelling evidence that the General Assembly did seek to maximally burden voters who were likely to support non-Republican candidates. Most significantly, in explaining the proposed Partisan Advantage criterion to the Committee, Representative Lewis said that he "propose[d] that [the Committee] draw the maps to give a partisan advantage to 10 Republicans and 3 Democrats because [he] d[id] not believe it[ would be] possible to draw a map with 11 Republicans and 2 Democrats ." Ex. 1005, at 50:7–10 (emphasis added). Legislative Defendants assert that this statement establishes that Representative Lewis did not draw the map to maximize partisan advantage because he did not believe that it would be possible to draw a plan that could elect 11 Republicans without violating other criteria, "such as keeping ... counties whole and splitting fewer precincts." Leg. Defs.' Br. 5. Put differently, Legislative Defendants maintain that the 2016 Plan's adherence to other traditional redistricting criteria establishes that the General Assembly did not pursue maximum partisan advantage. Id.

But Representative Lewis acknowledged during his deposition that had the 2016 Plan split a large number of precincts and counties, as the 2011 Plan did, there was a significant risk that the Harris court would "throw it out" on grounds that it failed to remedy the racial gerrymander. Lewis Dep. 166:13–168:8. Accordingly, Representative Lewis's testimony indicates that he believed the 2016 Plan offered the maximum lawful partisan advantage—the maximum partisan advantage that could be obtained without risking that the Harris court would "throw" the plan out as perpetuating the constitutional violation.

Dr. Mattingly's and Dr. Chen's analyses further evidence that the 2016 Plan reflected an effort to maximize partisan advantage. In particular, when Dr. Mattingly evaluated his 24,518–plan ensemble using the votes cast in North Carolina's 2012 congressional election, none of the plans produced an 11–2 pro-Republican partisan advantage. Ex. 3040, at 7. And Dr. Mattingly found the same result when he used votes from the 2016 election—none of the simulated plans produced an 11–2 partisan advantage. Id. at 19. Likewise, regardless of whether Dr. Chen applied the seven-race formula used by Dr. Hofeller or the twenty-race formula adopted by the Committee, none of his 3,000 simulated plans produced a 10–3 pro-Republican partisan advantage, let alone an 11–2 partisan advantage. Ex. 2010, at 12, 16, 21, 36–37.

Finally, the facts and circumstances surrounding the drawing and enactment of the 2011 Plan—the partisan effects of which the Committee expressly sought to carry forward in the 2016 Plan, Ex. 1007—further establish that the General Assembly drew the 2016 Plan to maximize partisan advantage. In particular, Representative Lewis and Senator Rucho's "primar[y] goal" in drawing the 2011 Plan was "to create as many districts as possible in which GOP candidates would be able to successfully compete for office." Hofeller Dep. 123:1–7 (emphasis added). And, in accordance with that goal, Dr. Hofeller testified that he drew the plan "to minimize the number of districts in which Democrats would have an opportunity to elect a Democratic candidate." Id. at 127:19–22 (emphasis added).

Nor does the General Assembly's reliance on a number of traditional redistricting criteria undermine our finding that invidious partisan intent motivated the 2016 Plan. As a matter of law, the Supreme Court long has held that a state redistricting body can engage in unconstitutional partisan gerrymandering even if it complies with the traditional redistricting criterion of population equality. Gaffney , 412 U.S. at 751, 93 S.Ct. 2321. More recently, the Supreme Court rejected an identical argument in a racial gerrymandering case, holding that "inconsistency between the [challenged] plan and traditional redistricting criteria is not a threshold requirement " to establish such a claim. Bethune–Hill v. Va. State Bd. of Elections , ––– U.S. ––––, 137 S.Ct. 788, 799, 197 L.Ed.2d 85 (2017) (emphasis added). The rationale supporting the Bethune–Hill Court's refusal to allow compliance with traditional redistricting criteria to immunize a plan from scrutiny under the Equal Protection Clause is equally compelling in the partisan gerrymandering context. As the Whitford Court explained in holding that compliance with traditional redistricting criteria is not a "safe harbor" from a partisan gerrymandering claim, "[h]ighly sophisticated mapping software now allows lawmakers to pursue partisan advantage without sacrificing compliance with traditional districting criteria." 218 F.Supp.3d at 889. "A map that appears congruent and compact to the naked eye may in fact be an intentional and highly effective partisan gerrymander." Id.

As a matter of fact, the 2016 Plan does not conform to all traditional redistricting principles. Although the plan is equipopulous, contiguous, improves on the compactness of the 2011 Plan, and reduces the number of county and precinct splits relative to the 2011 Plan, the 2016 Plan fails to adhere to the traditional redistricting principle of "maintaining the integrity of political subdivisions." Harris , 136 S.Ct. at 1306. In particular, Legislative Defendants' expert Dr. Hood conceded that the 2016 Plan divided numerous political subdivisions, see, e.g. , Trial Tr. IV, at 41:2–18, 42:6–43:4, including the City of Asheville, Buncombe County, Cumberland County, the City of Greensboro, Guilford County, Johnston County, the City of Charlotte, Mecklenburg County, and Wake County, Exs. 4066–70, 4072. Notably, the Committee voted, on a party-line basis, against adopting a proposed criterion that would have directed the mapdrawers to make reasonable efforts to respect the lines of political subdivisions and preserve communities of interest. See Ex. 1006, at 27–28. The division of political subdivisions allowed the General Assembly to achieve its partisan objectives, by packing non-Republican voters in certain districts and submerging non-Republican voters in majority-Republican districts. Trial Tr. IV, at 41:2–18, 42:6–43:4.

* * * * *

In sum, we find that Plaintiffs presented more-than-adequate evidence to satisfy their burden to demonstrate that the General Assembly was motivated by invidious partisan intent in drawing the 2016 Plan. Although we do not believe the law requires a finding of predominance, we nonetheless find that Plaintiffs' evidence—particularly the facts and circumstances surrounding the drawing and enactment of the 2016 Plan and Dr. Mattingly's and Dr. Chen's analyses—establish that the pursuit of partisan advantage predominated over the General Assembly's non-partisan redistricting objectives. And given that Dr. Chen found that the General Assembly's desire to protect incumbents and express refusal to try to avoid dividing political subdivisions failed to explain the 2016 Plan's partisan bias, we find that Plaintiffs' evidence distinguishes between permissible redistricting objectives that rely on political data or consider partisanship, and what instead here occurred: invidious partisan discrimination.

B.

Having concluded that the General Assembly intended to discriminate against voters who supported or were likely to support non-Republican candidates, we now must determine whether the 2016 Plan achieved its discriminatory objective.

1.

The discriminatory effects prong is the principal reason the Supreme Court has failed to agree on a standard for proving a partisan gerrymandering claim. For nearly two decades, the plurality opinion in Bandemer provided what was widely treated as the controlling test for determining whether a redistricting plan had the effect of discriminating against voters based on their partisan affiliation. See, e.g. , Pope , 809 F.Supp. at 395 ("[The Bandemer ] plurality opinion must be considered controlling as the position which concurs in the judgment on the narrowest grounds."). In Bandemer , a group of Indiana Democrats sued Indiana state officials alleging that the State's decennial state legislative redistricting—which was enacted by a Republican-controlled legislature and approved by a Republican governor—violated the Equal Protection Clause by intentionally discriminating against Democrats, notwithstanding that the plan satisfied the one-person, one-vote requirement. 478 U.S. at 113–14, 106 S.Ct. 2797 (plurality op.). As evidence of the districting plan's discriminatory effects, the plaintiffs alleged that the legislature drew district lines that packed Democratic voters into certain districts and fragmented Democratic votes in other districts in order to debase Democratic voting strength. Id. at 115, 106 S.Ct. 2797. Additionally, the legislature allegedly used multi-member districts to further diminish Democrats' voting strength. Id. In the first election following the redistricting, Democratic candidates received 51.9 percent of the vote but won 43 percent (43 of 100) of the seats in the state House. Id. In the Senate, Democratic candidates received 53.1 percent of the vote, and won 52 percent (13 of 25) of the seats up for election. Id.

Writing for a four-Justice plurality, Justice White stated that a partisan gerrymandering plaintiff must prove that it "has been unconstitutionally denied its chance to effectively influence the political process" or that the "electoral system [has been] arranged in a manner that will consistently degrade a voter's or a group of voters' influence on the political process as a whole." Id. at 132–33, 142–43, 106 S.Ct. 2797. Because legislators are presumed to represent all of their constituents, "even in a safe district where the losing group loses election after election," a "mere lack of proportional representation will not be sufficient to prove unconstitutional representation." Id. at 132, 106 S.Ct. 2797. Rather, a plaintiff must provide evidence "of continued frustration of the will of a majority of the voters or effective denial to a minority of voters of a fair chance to influence the political process." Id. at 133, 106 S.Ct. 2797.

Applying this test, the plurality concluded the plaintiffs failed to meet their burden. Id. at 134, 106 S.Ct. 2797. In particular, the plurality stated that the results of a single election were insufficient to demonstrate that Indiana Democrats would be relegated to minority status throughout the decade, particularly because Indiana was a "swing [s]tate" and voters would "sometimes prefer Democratic candidates, and sometimes Republican." Id. at 135, 106 S.Ct. 2797. The plurality further emphasized that the district court did not find that the redistricting plan would preclude Democrats from taking control of the assembly in a subsequent election, nor did the district court ask "by what percentage the statewide Democratic vote would have had to increase to control either the House or the Senate." Id. And the plaintiffs provided no proof that the redistricting plan would "consign the Democrats to a minority status in the Assembly throughout the [decade]." Id.

The Bandemer plurality's discriminatory effects test proved virtually impossible for future plaintiffs to satisfy. See, e.g. , Pope , 809 F.Supp. at 397 (dismissing partisan gerrymandering action because the plaintiffs did "not allege, nor c[ould] they, that the state's redistricting plan ... caused them to be ‘shut out of the political process’ " or that they had "been or w[ould] be consistently degraded in their participation in the entire political process"); Badham v. Eu , 694 F.Supp. 664, 670 (N.D. Cal. 1988) (dismissing partisan gerrymandering claim because the plaintiffs failed to allege any "interfer[ence] with [the allegedly disfavored party's] registration, organizing, voting, fund-raising, or campaigning" or that the interests of supporters of the disfavored party were "being ‘entirely ignore[d]’ by their congressional representatives" (third alteration in original) (quoting Bandemer , 478 U.S. at 132, 106 S.Ct. 2797 )). As one commentator explained, "by its impossibly high proof requirements the Court in Bandemer essentially eliminated political gerrymandering as a meaningful cause of action, but only after it had essentially declared the practice unconstitutional." John Hart Ely, Gerrymanders: The Good, the Bad, and the Ugly , 50 Stan. L. Rev. 607, 621 (1998) ; see also Samuel Issacharoff, Pamela S. Karlan & Richard H. Pildes, The Law of Democracy 563 (1998) (" Bandemer has served almost exclusively as an invitation to litigation without much prospect of redress.").

In Vieth , all of the Justices rejected Bandemer 's discriminatory effects test. 541 U.S. at 283, 124 S.Ct. 1769 (plurality op.) ("Because this standard was misguided when proposed [and] has not been improved in subsequent application, ... we decline to affirm it as a constitutional requirement."); id. at 308, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment); id. at 318, 339, 124 S.Ct. 1769 (Stevens, J., dissenting); id. at 344–45, 124 S.Ct. 1769 (Souter, J., dissenting); see id. at 360, 124 S.Ct. 1769 (Breyer, J., dissenting). And the Justices appeared to agree that one of the principal problems with the Bandemer plurality's discriminatory effects test is that it created an evidentiary standard so high that no plaintiff could satisfy it, even in the face of strong evidence of partisan discrimination. See id. at 280–81, 124 S.Ct. 1769 (plurality op.) (noting that under Bandemer 's test, "several districting plans ... were upheld despite allegations of extreme partisan discrimination, bizarrely shaped districts, and disproportionate results"); id. at 312, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment) (rejecting Bandemer 's effects test as establishing "a single, apparently insuperable standard"); id. at 344–45, 124 S.Ct. 1769 (Souter, J., dissenting) (rejecting Bandemer effects test on grounds that it "required a demonstration of such pervasive devaluation over such a period of time as to raise real doubt that a case could ever be made out").

In light of Vieth 's rejection of Bandemer 's discriminatory effects test—and the Supreme Court's failure to agree on a replacement—there is an absence of authority regarding the evidentiary burden a plaintiff must meet to prove that a districting plan discriminates against voters who are likely to support a disfavored candidate or party. League Plaintiffs propose that to prove that a districting plan has a discriminatory effect, a plaintiff must demonstrate that the plan "exhibits a large and durable partisan asymmetry." League Br. 10. League Plaintiffs assert that their proposed magnitude requirement would ensure that courts do not unduly intrude on a state districting efforts. Id. at 11. And according to League Plaintiffs, the durability requirement speaks to one of the Court's principal concerns with partisan gerrymandering: entrenchment. Id. at 11–12. By contrast, although Common Cause Plaintiffs concede that a plaintiff must prove that a districting plan "burden[ed]" the rights of supporters of a disfavored candidate, they assert that neither "the Constitution [n]or controlling precedent require either a large or a durable effect before the Court can intervene." Common Cause Br. 4.

Drawing on the Supreme Court's definition of "partisan gerrymandering," we conclude that to meet the discriminatory effects requirement, the Equal Protection Clause demands that a partisan gerrymandering plaintiff show that a challenged districting plan "subordinate[s the interests] of one political party and entrench[es] a rival party in power." Ariz. State Leg. , 135 S.Ct. at 2658. A plaintiff proves a districting plan "subordinates" the interests of supporters of a disfavored candidate party by demonstrating that the redistricting plan is biased against such individuals. The bias requirement reflects the Equal Protection Clause's animating dictate that states must govern "impartially"—that "the State should treat its voters as standing in the same position, regardless of their political beliefs or party affiliation." Davis , 478 U.S. at 166, 106 S.Ct. 2797 (Powell, J., concurring in part and dissenting in part).

The entrenchment requirement addresses another principal constitutional concern with partisan gerrymandering—that it insulates legislators from popular will and renders them unresponsive to portions of their constituencies. See Reynolds , 377 U.S. at 565, 84 S.Ct. 1362 ("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."). As the Supreme Court explained with regard to racial gerrymanders, "[w]hen a district obviously is created solely to effectuate the perceived common interests of one ... group, elected officials are more likely to believe that their primary obligation is to represent only the members of that group, rather than their constituency as a whole." Shaw I , 509 U.S. at 648, 113 S.Ct. 2816. To prove entrenchment, a plaintiff need not meet Bandemer 's "apparently insuperable standard," id. at 312, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment), which required a showing that supporters of a disfavored party had been entirely ignored by their representatives and for years had been frozen out of key aspects of the political process. Instead, a plaintiff must show that a districting plan's bias towards a favored party is likely to persist in subsequent elections such that an elected representative from the favored party will not feel a need to be responsive to constituents who support the disfavored party.

2.

We find that Plaintiffs satisfied their burden under the discriminatory effects prong by proving the 2016 Plan dilutes the votes of non-Republican voters and entrenches Republican control of the state's congressional delegation. In reaching this conclusion we rely on the following categories of evidence: (a) the results of North Carolina's 2016 congressional election conducted using the 2016 Plan; (b) expert analyses of those results revealing that the 2016 Plan exhibits "extreme" partisan asymmetry; (c) Dr. Mattingly's and Dr. Chen's simulation analyses; and (d) the results of North Carolina's 2012 and 2014 elections using the 2011 Plan—the partisan effects of which the General Assembly expressly sought to carry forward when it drew the 2016 Plan—and empirical analyses of those results. a.

We begin with the results of North Carolina's 2016 congressional election conducted under the 2016 Plan. The General Assembly achieved its goal: North Carolina voters elected a congressional delegation of 10 Republicans and 3 Democrats. Exs. 1018, 3022. That the 2016 Plan resulted in the outcome Representative Lewis, Senator Rucho, Dr. Hofeller, and the General Assembly intended proves both that the precinct-level election data used by the mapdrawers served as a reliable predictor of the 2016 Plan's partisan performance and that the mapdrawers effectively used that data to draw a districting plan that perfectly achieved the General Assembly's partisan objectives.

Following the 2016 election, Republicans hold 76.9 percent of the seats in the state's thirteen-seat congressional delegation, whereas North Carolina voters cast 53.22 percent of their votes for Republican congressional candidates. Ex. 3022. Notably, the Whitford court found that less significant disparities between the favored party's seat-share and vote-share (60.7% v. 48.6% and 63.6% v. 52%) provided evidence of a challenged districting plan's discriminatory effects. 218 F.Supp.3d at 901. As the court explained, "[i]f it is true that a redistricting ‘plan that more closely reflects the distribution of state party power seems a less likely vehicle for partisan discrimination,’ ... then a plan that deviates this strongly from the distribution of statewide power suggests the opposite." Id. at 902 (quoting LULAC , 548 U.S. at 419, 126 S.Ct. 2594 (opinion of Kennedy, J.)).

The results of the 2016 election also reveal that the 2016 Plan "packed" and "cracked" voters who supported Republican candidates. In particular, in the three districts in which Democratic candidates prevailed, the Democratic candidates received an average of 67.95 percent of the vote, whereas Republican candidates received an average of 31.24 percent of the vote. See Ex. 3022. By contrast, in the ten districts in which Republican candidates prevailed, the Republican candidates received an average of 60.27 percent of the vote, and Democratic candidates received an average of 39.73 percent of the vote. See id. Democratic candidates, therefore, consistently won by larger margins than Republican candidates. Additionally, the Democratic candidate's margin in the least Democratic district in which a Democratic candidate prevailed (34.04%) was nearly triple that of the Republican candidate's margin in the least Republican district in which a Republican candidate prevailed (12.20%), see id. , reflecting the "S-shaped curve" that Dr. Mattingly described as "the signature of [partisan] gerrymandering," Trial Tr. I, at 76:18–77:5.

And the results of the 2016 congressional election establish that the 2016 Plan's discriminatory effects likely will persist through multiple election cycles. To begin, the Republican candidate's vote share (56.10%) and margin of victory (12.20%) in the least Republican district electing a Republican candidate, District 13, exceed the thresholds at which political science experts, including Legislative Defendants' expert Dr. Hood, consider a seat to be "safe"—i.e. , highly unlikely to change parties in subsequent elections. See Ex. 5058, at 25, Trial Tr. IV, at 29:16–22, 86:21–88:5; LULAC , 548 U.S. at 470–71, 126 S.Ct. 2594 (Stevens, J., dissenting in part) (characterizing 10 percent advantage as a threshold for a "safe" seat and explaining that "[m]embers of Congress elected from such safe districts need not worry much about the possibility of shifting majorities, so they have little reason to be responsive to political minorities in their district"). Indeed, all of the districts—including all ten Republican districts—in the 2016 Plan are "safe" under that standard. Ex. 3022.

Additionally, Dr. Simon Jackman—a professor of political science at the University of Sydney and expert in statistical methods in political science, elections and election forecasting, and American political institutions, Trial Tr. II, at 32:5–9—performed a "uniform swing analysis," which is used by both researchers and courts to assesses the sensitivity of a districting plan to changing electoral conditions, Ex. 4002, at 15–16, 54–59; Whitford , 218 F.Supp.3d at 899–903. To conduct his uniform swing analysis, Dr. Jackman took the two-parties' statewide vote share in the 2016 election, and then shifted those shares by one-percent increments ranging from 10 percent more Republican to 10 percent more Democratic. Ex. 4002, at 54. The analysis assumed that votes shift in all districts by the same amount. Id. Dr. Jackman found that "[i]f Democrats obtained a statewide, uniform swing of even six points—taking Democratic share of the two-party vote to 52.7%—no seats would change hands relative to the actual 2016 results ." Id. at 59 (emphasis added). Accordingly, even if Democratic candidates obtained a 52.7 percent of the statewide vote, they would comprise only 23.1 percent of the state's congressional delegation. And if Democratic candidates captured the same percentage of the vote (53.22%) that elected Republican candidates in ten districts in 2016, Democratic candidates would prevail in only four districts. Ex. 3022.

b.

We also find that other analyses performed by Dr. Jackman assessing the 2016 Plan's "partisan asymmetry"—whether supporters of each of the two parties are able to translate their votes into representation with equal ease—provide additional evidence of the 2016 Plan's discriminatory effects. Trial Tr. II, at 34:20–22 (explaining that a redistricting plan exhibits partisan asymmetry if there is "a gap between the parties with respect to the way their votes are translated into seats"). The concept of partisan symmetry, at least in its modern form, dates to the 1970s, but scholars did not begin to widely view it as a measure of partisan gerrymandering until the last 20 years. Id. at 33:24–34:11. Dr. Jackman analyzed three standard measures of partisan symmetry: (i) the "efficiency gap," (ii) "partisan bias," and (iii) "the mean-median difference." Id. at 34:13–17.

i.

The efficiency gap, which was the focus of Dr. Jackman's report and is the newest measure of partisan asymmetry, evaluates whether a districting plan leads supporters of one party to "waste" more votes than supporters of the other. Ex. 4002, at 5. The concept of "wasted" votes derives from two of the principal mechanisms mapdrawers use to diminish the electoral power of a disfavored party or group: (1) packing—concentrating members or supporters of the party or group in a limited number of districts—and (2) cracking—dispersing members or supporters of the party or group across a number districts so that they are relegated to minority status in each of those districts. Trial Tr. II, at 45:19–46:11. "Wasted" votes are votes cast for a candidate in excess of what the candidate needed to win a given district, which increase as more voters supporting the candidate are "packed" into the district, or votes cast for a losing candidate in a given district, which increase, on an aggregate basis, when a party's supporters are "cracked." Id. at 35:9–23, 45:19–46:11. Dr. Jackman calculated the efficiency gap by subtracting the sum of one party's wasted votes in each district in a particular election from the sum of the other party's wasted votes in each district in that election and then dividing that figure by the total number of votes cast for all parties in all districts in the election. Ex. 4002, at 18; Ex. 4078. Efficiency gaps close to zero, which occur when the two parties waste approximately the same number of votes, reflect a districting plan that does not favor, invidiously or otherwise, one party or the other.

Using the results of the 2016 congressional elections conducted under the 2016 Plan, Dr. Jackman calculated an efficiency gap favoring Republican candidates of 19.4 percent. Ex. 4002, at 7–8. That constituted the third largest efficiency gap (pro-Republican or pro-Democratic) in North Carolina since 1972, surpassed only by the efficiency gaps exhibited in the 2012 and 2014 elections using the 2011 Plan. Trial Tr. II, at 54:21–24.

To put the 19.4 percent figure further in perspective, Dr. Jackman estimated the efficiency gaps for 512 congressional elections occurring in 25 states between 1972 and 2016. He determined that the distribution of those efficiency gaps was normal with its mean and median centered on zero, meaning that, on average, the districting plans in his sample did not tend to favor either party. Ex. 4002, at 26–28. Dr. Jackman found that North Carolina's 2016 congressional election under the 2016 Plan yielded the 13th most pro-Republican efficiency gap of the 512 elections in the database, and that 95 percent of the plans in the database had efficiency gaps that were smaller in magnitude (in favor of either Republicans or Democrats). Id. at 7, 65. Dr. Jackman also calculated the average efficiency gap for the 136 unique districting plans included in his 512–election database, and found that the 2016 Plan produced the fourth-largest average efficiency gap of the 136 plans. Id. at 10; Trial Tr. II, at 60:15–17. And Dr. Jackman compared North Carolina's efficiency gap in 2016 with that of 24 other states for which his database contained 2016 data, finding that the 2016 Plan produced the largest efficiency gap of any of those plans. Ex. 4002, at 9.

To further put the 19.4 percent figure in context, Dr. Jackman used his database of elections to analyze what magnitude of efficiency gap would likely lead to at least one congressional seat changing hands—a "politically meaningful" burden on a disfavored party's supporters. Ex. 4002, at 37; Trial Tr. II, at 64:6–12. Dr. Jackman found that in states with congressional delegations with 7 to 15 representatives, like North Carolina, an 8 percent efficiency gap is associated with at least one seat likely changing hands. Ex. 4002, at 39–41. Under that threshold, North Carolina's 2016 efficiency gap of 19.4 percent indicates that the 2016 Plan allowed Republicans to prevail in at least one more district than they would have in an unbiased plan. Based on these results, Dr. Jackman concluded that the 2016 Plan creates "a systematic advantage for Republican candidates,"id. at 62, and that that advantage "is generating tangible consequences in terms of seats being won," Trial Tr. IIII, at 82:13–16.

Dr. Jackman also sought to test whether, given the magnitude of North Carolina's 2016 efficiency gap, the pro-Republican bias of the 2016 Plan is likely to persist in future elections. To do so, he performed regressions using his multi-state dataset to analyze the relationship between the first efficiency gap observed in the first election conducted under a particular districting plan and the average efficiency gap over the remaining elections in which that plan was used. Ex. 4002, at 47–54. Using data from the 108 plans in his dataset that were used in at least three elections, Dr. Jackman estimated that a plan with an initial efficiency gap of 19.4 percent in favor of a particular party, like the 2016 Plan, likely would have an 8 percent average efficiency gap in favor of the same party in the remaining elections conducted under the plan, with the plan resulting in an average efficiency gap in that same party's favor over 90 percent of the time. Id. at 47. When Dr. Jackman restricted his data set to the 44 plans that have been used at least three times since 2000, he found that an efficiency gap of 19.4 percent in favor of one party would likely have a 12 percent efficiency gap in that party's favor over the remainder of the plan's use. Id. Based on these analyses, Dr. Jackman concluded that the evidence "strongly suggests" that the 2016 Plan "will continue to produce large, [pro-Republican] efficiency gaps (if left undisturbed), generating seat tallies for Democrats well below those that would be generated from a neutral districting plan." Id. at 66.

Additionally, Dr. Jackman evaluated the likely persistence of the 2016 Plan's pro-Republican bias by conducting a uniform swing analysis and determining the size of pro-Democratic swing necessary to eliminate the 2016 Plan's pro-Republican efficiency gap. Id. at 54–60. Dr. Jackman found that it would require a uniform swing of approximately 9 percentage points in Democrats' favor—on the order of the 1974 post-Watergate swing in favor of Democrats, the largest pro-Democratic swing that has occurred in North Carolina since 1972—for the efficiency gap to return to zero, and therefore for the 2016 Plan to lose its pro-Republican bias. Id. at 55–59. Based on these analyses, Dr. Jackman concluded that the 2016 Plan's pro-Republican efficiency gap "is durable," and that it would require a swing of votes in Democratic candidates' favor of "historic magnitude" to strip the 2016 Plan of its pro-Republican bias. Trial Tr. II, at 54:24–55:9; see also Ex. 4002, at 66 (concluding that the 2016 Plan's large, pro-Republican efficiency gap is "likely to endure over the course of the plan").

Legislative Defendants raise several objections to Dr. Jackman's efficiency gap analysis: (1) the efficiency gap cannot be applied in all states; (2) the efficiency gap is a measure of "proportional representation," and therefore is foreclosed by controlling Supreme Court precedent; (3) there are several problems with Dr. Jackman's efficiency gap thresholds for identifying when a particular plan is biased towards one party and when that bias is likely to persist; (4) the efficiency gap does not account for a variety of idiosyncratic factors that play a significant role in determining election outcomes; (5) the efficiency gap fails to flag as unconstitutional certain districting plans that bear certain hallmarks of a partisan gerrymander; (6) the efficiency gap cannot be administered prospectively, making it impossible for a legislature to predict whether a districting plan will violate the Constitution; and (7) the efficiency gap does not encourage mapmakers to draw more competitive districts. Leg. Defs.' FOF 62–66. Although we do not entirely discount all of these objections, we find that they do not individually, or as a group, materially undermine the persuasive force of Dr. Jackman's efficiency gap analysis regarding the 2016 Plan.

Dr. Jackman concedes that the sensitivity of the efficiency gap in jurisdictions with only a few districts—in the case of congressional districts, states with six or fewer districts—renders it difficult, if not impossible, to apply. See Ex. 4002, at 19. According to Legislative Defendants, this limitation requires this Court to categorically reject the efficiency gap as a measure of partisan gerrymandering because "[i]t would be untenable for a court to impose a constitutional standard on one state that literally cannot be imposed or applied in all other states." Leg. Defs.' Br. 10. But League Plaintiffs do not propose that this Court constitutionalize the efficiency gap—nor does this Court do so. Rather, League Plaintiffs argue—and this Court finds—that Dr. Jackman's efficiency gap analysis provides evidence that Defendants violated the governing constitutional standard: that a redistricting body must not adopt a districting plan that intentionally subordinates the interests of supporters of a disfavored party and entrenches a favored party in power. See supra Parts II.B.2.b. That constitutional standard does not vary with the size of a state's congressional delegation. In states entitled to a small number of representatives, a partisan gerrymandering plaintiff simply will have to rely on different types of evidence to prove that the redistricting body violated that constitutional standard. Importantly, in addition to the efficiency gap, this Court relies on a variety of other types of evidence probative of the 2016 Plan's discriminatory effects, much of which could be relied on in states with a smaller number of congressional districts.

Legislative Defendants also are correct that the Constitution does not entitle supporters of a particular party to representation in a state's congressional delegation in proportion to their statewide vote share. See LULAC , 548 U.S. at 419, 126 S.Ct. 2594 (opinion of Kennedy, J.) ("To be sure, there is no constitutional requirement of proportional representation ...."). But the efficiency gap, like other measures of partisan asymmetry, does not dictate strict proportional representation. Trial Tr. II, at 48:21–50:7; Trial Tr. III, at 70:5–7. In particular, the efficiency gap permits a redistricting body to choose to draw a districting plan that awards the party that obtains a bare majority of the statewide vote a larger proportion of the seats in the state's congressional delegation (referred to as a "winner's bonus"). The efficiency gap, therefore, is not premised on strict proportional representation, but rather on the notion that the magnitude of the winner's bonus should be the same for both parties. Trial Tr. II, at 49:8–17 (Dr. Jackman explaining that partisan symmetry is a "weaker property" than proportional representation because "[a]ll it insists on is that the mapping from votes into seats is the same for both sides of politics"). Even if the efficiency gap did amount to a measure of proportional representation, "[t]o say that the Constitution does not require proportional representation is not to say that highly dis proportionate representation may not be evidence of a discriminatory effect." Whitford , 218 F.Supp.3d at 906–07. On the contrary, a number of Justices have concluded that disproportionate representation constitutes evidence, although not conclusive evidence, of a redistricting plan's discriminatory effects—the same way in which we treat Dr. Jackman's efficiency gap evidence. LULAC , 548 U.S. at 419, 126 S.Ct. 2594 (opinion of Kennedy, J.) ("[A] congressional plan that more closely reflects the distribution of state party power seems a less likely vehicle for partisan discrimination than one that entrenches an electoral minority."); Bandemer , 478 U.S. at 132, 106 S.Ct. 2797 (plurality op.) ("[A] failure of proportional representation alone does not constitute impermissible discrimination under the Equal Protection Clause." (emphasis added)).

As to Dr. Jackman's proposed thresholds, Legislative Defendants are correct that in Whitford Dr. Jackman used a different method for calculating an efficiency gap and found "that an efficiency gap above 7% in any districting plan's first election year will continue to favor that party for the life of the plan." 218 F.Supp.3d at 905. By contrast, here Dr. Jackman concluded that, in states like North Carolina with 7 to 14 representatives, a 12 percent first-year efficiency gap indicates that the districting plan's partisan bias will persist in subsequent elections. Ex. 4002, at 51–54. Even under the more conservative threshold Dr. Jackman proposes in this case, approximately one-third of the post–2000 districting plans in such states that would trip Dr. Jackman's threshold did not have an average remainder-of-the-plan efficiency gap of sufficient magnitude to establish that the districting plan deprived the disfavored party of at least one seat. Id. at 53. We agree with Legislative Defendants that this error rate weighs against constitutionalizing Dr. Jackman's proposed thresholds. But we do not constitutionalize Dr. Jackman's efficiency gap thresholds. And given (1) that the magnitude of the 2016 Plan's efficiency gap in the 2016 congressional election (19.4 percent) significantly exceeded either threshold, (2) that most plans in Dr. Jackman's database that exceeded his proposed threshold continued to exhibit a meaningful bias throughout their life, and (3) that numerous other pieces of evidence provide proof of the 2016 Plan's discriminatory effects, we do not believe this concern strips Dr. Jackman's analyses of their persuasive force in this case. See Whitford , 218 F.Supp.3d at 907–08 (acknowledging different methods of calculating the efficiency could prove problematic in other cases but nonetheless relying on efficiency gap evidence because challenged legislative districting plan was not "at the statistical margins" and "both methods yield[ed] an historically large, pro-Republican [efficiency gap]").

Legislative Defendants next assert that the efficiency gap, as a "mathematical formula," does not take into account a number of idiosyncratic considerations that effect the outcome of particular elections, such as "the quality of ... candidates, the amount of money raised, the impact of traditional districting principles on election results, whether Democratic voters are more concentrated than Republican voters, and the impact of wave elections." Leg. Defs.' FOF 65. We agree that each of these considerations may impact the outcome of a particular election. But we reject Legislative Defendants' assertion that Dr. Jackman's conclusion that the 2016 Plan is an extreme partisan outlier does not account for these contest-specific factors. On the contrary, Dr. Jackman reached his conclusion by comparing the 2016 Plan's efficiency gap with efficiency gaps observed in the other 512 elections in his database. That database comprises results from 512 elections occurring in 25 states over a 44–year period. As Dr. Jackman explained, "all of those [election-specific] factors appeared in those 512 elections," including the Watergate and 1994 wave elections, candidates facing political scandals, candidates who were well-funded or poorly funded, states with political geography favoring one party or the other, and unique candidates at the top of the ballot like President Obama and President Trump. Trial Tr. IIII, at 69:5–18. Accordingly, comparing the 2016 Plan's efficiency gap to those observed in hundreds of other elections allowed Dr. Jackman to conclude that the election-specific factors that Legislative Defendants highlight do not explain the large magnitude of the 2016 Plan's pro-Republican efficiency gap.

Relatedly, Legislative Defendants contend that Dr. Jackman's proposed efficiency thresholds flag several bipartisan districting plans or districting plans drawn by courts or nonpartisan commissions and fail to flag as partisan gerrymanders a number of districting plans that bear other hallmarks of gerrymandering such as irregular shapes and widespread division of political subdivisions and voting precincts. See Ex. 5101, at 29–62. But if a districting plan is drawn on a bipartisan basis or by a nonpartisan body, a plaintiff will be unable to establish that it was drawn with discriminatory intent, and therefore the plan will pass constitutional muster. See Whitford , 218 F.Supp.3d at 908. Likewise, just as compliance with traditional redistricting criteria does not immunize a districting plan from constitutional scrutiny, see supra Part III.A.3, failure to comply with redistricting criteria does not necessarily prove the inverse—that a districting plan amounts to an actionable partisan gerrymander. And to the extent Dr. Jackman's threshold fails to flag certain unconstitutional plans, a plaintiff can rely on other types of evidence to prove a plan's discriminatory effects. Additionally, each of these concerns are not present in this case—the Republican-controlled General Assembly intended to dilute the votes of non-Republican voters and the 2016 Plan exhibited an extremely large efficiency gap in the 2016 election—meaning that those concerns, although potentially legitimate in other cases, do not significantly undermine the probative force of Dr. Jackman's efficiency gap conclusions as to the 2016 Plan. Accord Whitford , 218 F.Supp.3d at 908.

We also reject Legislative Defendants' assertion that a state redistricting body cannot apply the efficiency gap prospectively. In particular, Dr. Chen used the results from the seven races on which Dr. Hofeller relied and the twenty races included in the Committee's Political Data criterion to predict the efficiency gap for both the 2016 Plan and the 3,000 simulated plans he generated. Ex. 2010, at 32–34. Like Dr. Jackman's post hoc analysis, Dr. Chen's analysis revealed that the 2016 Plan's predicted efficiency gap was an extreme outlier relative to the simulated plans in his sample and significantly higher than the thresholds suggested by Dr. Jackman. Id. at 25. Accordingly, just as the General Assembly used the data relied on by Dr. Hofeller and prescribed by the Committee to predict (correctly) that the 2016 Plan would elect ten Republicans and three Democrats, so too could it have used that same data to predict the 2016 Plan's efficiency gap—and that the magnitude of that gap would provide strong evidence of the 2016 Plan's pro-Republican bias.

Finally, we agree with Legislative Defendants that the efficiency gap does not provide redistricting bodies with an incentive to draw districting plans with more competitive districts. But the 2016 Plan, which Legislative Defendants seek to keep in place, also creates uniformly "safe" districts. See Ex. 3022. And the Supreme Court has never held that the Constitution entitles voters to competitive districts. Accordingly, regardless of whether the efficiency gap's failure to encourage redistricting bodies to draw districting plans with competitive districts is desirable from a policy perspective, that failure does not render the efficiency gap constitutionally or legally infirm.

ii.

The second measure of partisan asymmetry calculated by Dr. Jackman, partisan bias, measures a districting plan's asymmetry by taking the two parties' statewide vote share in a particular election, and then imposing a uniform swing of the magnitude necessary to make the parties split the statewide vote equally. Trial Tr. II, at 47:7–21; LULAC , 548 U.S. at 420, 126 S.Ct. 2594 (explaining that partisan bias is measured by "comparing how both parties would fare hypothetically if they each (in turn) had received a given percentage of the vote" (internal quotation marks and alteration omitted)). After performing the uniform swing, the analyst then calculates the number of seats each party would win. Trial Tr. II, at 47:7–21. A districting plan "is biased in favor of the party that would win more than 50 percent of the seats, if it won 50 percent of the vote and is biased against the ... party that would win less than 50 percent of the seats if it were able to win 50 percent of the vote," Dr. Jackman explained. Id. at 46:15–47:4. When partisan bias is close to zero, a districting plan does not favor, invidiously or otherwise, one party or the other. Ex. 4002, at 13–17; Trial Tr. II, at 48:21–50:7. In LULAC , a majority of the Court agreed that partisan bias, at a minimum, has "utility in redistricting planning and litigation," even if, by itself, it is "not a reliable measure of unconstitutional partisanship." 548 U.S. at 420, 126 S.Ct. 2594 (opinion of Kennedy, J.); id. at 483–84, 126 S.Ct. 2594 (Souter, J. dissenting in part) (joined by Ginsburg, J., noting that "[i]nterest in exploring [partisan bias and other measures of partisan symmetry] is evident" and citing separate opinions of Kennedy, J., Stevens, J., and Breyer, J.).

Dr. Jackman found that the 2016 Plan exhibited a pro-Republican partisan bias of 27 percent. Ex. 4003, at 3–4. He again sought to put that figure in perspective by comparing it to previous North Carolina congressional elections and congressional elections across the country. Dr. Jackman found that the 2016 Plan's partisan bias in the 2016 election was the largest observed in North Carolina since 1972, the first year for which he had data. Id. And the 2016 Plan's partisan bias was the second largest observed among the 283 state congressional elections in his database, and "roughly three standard deviations from the historical mean." Id. at 4. Based on these findings, Dr. Jackman characterized the partisan bias exhibited by the 2016 Plan as "extreme"—"of quite literally historic magnitude, not just relative to North Carolina's history, but in the United States of America." Trial Tr. II, at 80:15, 80:24–81:1.

iii.

Finally, Dr. Jackman estimated the 2016 Plan's mean-median difference in North Carolina's 2016 congressional election. As its name suggests, the mean-median difference is the difference between a party's mean vote share in a particular election and median vote share in that election across all of the districts included in the subject districting plan. Ex. 4003, at 7. In his report, Dr. Jackman explained that the intuition behind the mean-median difference measure "is that when the mean and the median diverge significantly, the distribution of district-level vote shares is skewed in favor of one party and against its opponent—consistent with the classic gerrymandering techniques of ‘packing’ partisans into a relatively small number of districts and/or ‘cracking’ partisans among a larger number of districts." Id. As with the efficiency gap and partisan bias, the closer the mean-median difference is to zero, the less a plan is biased (invidiously or otherwise) towards one party or another.

Dr. Jackman found that the 2016 Plan exhibited a pro-Republican mean-median difference of 5.1 percent in North Carolina's 2016 congressional election. He explained that the mean-median difference arose from the packing of Democratic voters in the three districts in which Democratic candidates prevailed, and the dispersal of Democratic voters across the remaining districts. Trial Tr. II, at 81:17–21 ("[T]he skew here arises from the fact that there are three districts where Democratic vote share is in the 60s, and then there are ten where it's below 50 percent, where the Democrat lost."). Again seeking to put the 2016 Plan's 5.1 percent figure in historical perspective, Dr. Jackman found that "North Carolina's average mean-median difference from 1972 to 2016 was just 1.0%," Ex. 4003, at 8, and for the other state elections included in his database the average mean-median difference was "roughly ... zero." Trial Tr. II, at 81:22.

* * * * *

We find Dr. Jackman's partisan asymmetry analyses provide strong evidence that the 2016 Plan subordinates the interests of supporters of non-Republican candidates and serves to entrench the Republican Party's control of the state's congressional delegation. In particular, we find it significant that three different measures of partisan asymmetry all point to the same result—that the 2016 Plan poses a significant impediment to supporters of non-Republican candidates translating their votes into seats, and that the magnitude of that impediment is an extreme outlier relative to other congressional districting plans. We also find it significant that Dr. Jackman's analyses demonstrate the durability of the 2016 Plan's pro-Republican bias, both by comparing the 2016 Plan to other plans that were used in multiple elections and by demonstrating that 2016 Plan is likely to retain its pro-Republican bias "under any likely electoral scenario." Whitford , 218 F.Supp.3d at 899, 903. Given that durability, we find that the 2016 Plan has the effect of entrenching Republican candidates in power, even in the face of significant shifts in voter support in favor of non-Republican candidates, and thereby likely making Republican elected representatives less responsive to the interests of non-Republican members of their constituency.

c.

Next, we find that 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. As explained above, Dr. Mattingly created an ensemble of 24,518 simulated districting plans that conform to traditional redistricting criteria, and then assessed the electoral outcomes of those plans relative to the 2016 Plan using actual votes cast in North Carolina's 2012 and 2016 congressional elections. See supra Part III.A.2.b. When he evaluated the ensemble using actual 2012 votes, Dr. Mattingly found that nearly 80 percent of the simulated plans would have yielded two-to-three fewer seats for Republicans than the 2016 Plan, and more than 99 percent of the plans resulted in at least one less seat for Republicans. Ex. 3040, at 7–10. And using actual 2016 congressional votes, Dr. Mattingly found that more than 70 percent of the simulated plans produced two-to-three fewer seats for Republicans than the 2016 Plan, and more than 99 percent of the plans resulted in at least one less seat for Republicans. Id. at 19–22. Accordingly, Dr. Mattingly's analyses indicate that the 2016 Plan had a measurable tangible adverse impact on supporters of non-Republican candidates.

Dr. Chen's simulation analyses likewise indicate that the 2016 Plan had a measurable tangible adverse effect on supporters of non-Republican candidates. Analyzing his first set of 1,000 simulated plans—which sought to conform to the Committee's non-partisan criteria—using elections results reflected in Dr. Hofeller's seven-race formula, Dr. Chen found that 78 percent of the simulated plans would have elected three-to-four fewer Republican candidates, with all of the plans electing at least one less Republican candidate. See Ex. 2010, at 12–13. And using the Committee's twenty-race criterion, Dr. Chen found that 94.5 percent of the simulated plans would have elected two-to-four fewer Republican candidates, with all of the plans electing at least one less Republican candidate. Id. at 13. Dr. Chen found similar results when he used the 2,000 simulated plans in his simulated sets that sought to avoid pairing incumbents and match the county splits and incumbent protection of the 2016 Plan. Id. at 16, 21. Based on these results, Dr. Chen concluded that the 2016 Plan "creates 3 to 4 more Republican seats than what is generally achievable under a map-drawing process respecting non-partisan, traditional districting criteria." Id. at 2–3.

To assess the 2016 Plan's partisan effects, Dr. Chen also compared the 2016 Plan's efficiency gap with those of his simulated plans. For each of his three sets of 1,000 simulated districting plans, Dr. Chen found that the 2016 Plan yielded a significantly higher pro-Republican efficiency gap than all of the simulated plans, regardless of whether he used the results from the seven elections relied on by Dr. Hofeller or the twenty elections prescribed by the Committee. Id. at 32–34. Because the 2016 Plan yielded "improbabl[y]" high pro-Republican efficiency gaps, Dr. Chen concluded "with overwhelmingly high statistical certainty that neutral, non-partisan districting criteria, combined with North Carolina's natural political geography, could not have produced a districting plan as electorally skewed as the [2016 Plan]." Id. at 25.

Taken together, Dr. Mattingly's and Dr. Chen's analyses—which use multiple methods for generating districting plans and multiple sets of votes—provide additional strong evidence that the 2016 Plan had the effect of discriminating against non-Republican voters. As detailed above, none of Legislative Defendants' objections to Dr. Mattingly's and Dr. Chen's analyses call into question their persuasive force. See supra Part III.A.2.b.

d.

Finally, although not essential to our finding that the 2016 Plan had the effect of discriminating against supporters of non-Republican candidates, the results of the two congressional elections conducted under the 2011 Plan —and empirical analyses of those results—provide further evidence of the 2016 Plan 's discriminatory effects. As explained previously, see supra Part III.A.2.c, because the Adopted Criteria expressly sought to carry forward the 2011 Plan's partisan effects, Ex. 1007, any discriminatory partisan effects attributable to the 2011 Plan are probative of the 2016 Plan's discriminatory effects. That is particularly true given that, according to an analysis by Legislative Defendants' expert Dr. Hood, most of the districts created by the 2016 Plan retained the "core" of their constituency under the 2011 Plan, Ex. 5058, at 23, including the First, Fourth, and Twelfth Districts in which Dr. Hofeller expressly sought to "concentrat[e]" likely Democratic voters, Ex. 2043, at 33–34.

In North Carolina's 2012 election conducted under the 2011 Plan, North Carolina voters statewide cast 50.9 percent of the votes for Democratic congressional candidates, yet Democratic candidates won only 30.8 percent of the state's congressional seats (4 of 13). Ex. 4002, at 62. The 2011 Plan exhibited a 21.4 percent pro-Republican efficiency gap in the 2012 election. Id. In 2014, Democratic candidates won 46.2 percent of the statewide vote, and won 23.1 percent of the seats in the state's congressional delegation, producing a pro-Republican efficiency gap of 21.1 percent. Id. North Carolina's 2012 and 2014 efficiency gaps produced under the 2011 Plan were the twelfth– and fourteenth-largest by magnitude in Dr. Jackman's 512–election sample. Id. at 65. Therefore, as the durability analyses conducted by Dr. Jackman described above would indicate, the magnitude of the 2012 efficiency gap pointed to the large efficiency gap realized in 2014. See supra Part II.B.2.b.i.

Noting that the magnitude of North Carolina's efficiency gaps under the 2011 Plan were significantly higher than those exhibited by the 2001 Plan, Dr. Jackman concluded that the 2011 Plan "is the driver of the change, systematically degrading the efficiency with which Democratic votes translate into Democratic seats in North Carolina." Ex. 4002, at 66. Accordingly, because (1) the General Assembly drew the 2016 Plan to perpetuate the partisan effects of the 2011 Plan and (2) evidence reveals that the 2011 Plan was systematically biased to durably burden supporters of non-Republican candidates, we find that the pro-Republican bias of the 2011 Plan provides further evidence of the 2016 Plan's discriminatory effects.

* * * * *

When viewed in totality, we find Plaintiffs' evidence more than sufficient to prove that the 2016 Plan has discriminated, and will continue to discriminate, against voters who support non-Republican candidates. In reaching this conclusion, we find it significant that Plaintiffs' evidence proves the 2016 Plan's discriminatory effects in a variety of different ways. Plaintiffs' direct evidence based on the actual results of an election conducted under the 2016 Plan confirmed that the discriminatory effects intended by the 2016 Plan's architects and predicted by Dr. Mattingly's analyses—the election of 10 Republicans by margins that suggest they will retain their seats throughout the life of the plan—in fact occurred. That five different types of statistical analyses performed by three different experts all reached the same conclusion gives us further confidence that 2016 Plan produces discernible discriminatory effects. And although some of those analyses considered "unfair results that would occur in a hypothetical state of affairs," LULAC , 548 U.S. at 420, 126 S.Ct. 2594 (opinion of Kennedy, J.), others like the efficiency gap and the mean-median difference did not. Given that all of this evidence "point[s] in the same direction"—and Legislative Defendants failed to provide any evidence to the contrary—Plaintiffs have provided "strong proof" of the 2016 Plan's discriminatory effects. Sylvester , 453 F.3d at 903.

C.

We now must determine whether the 2016 Plan's discriminatory effects are justified by a legitimate state districting interest or neutral explanation. See Vieth , 541 U.S. at 307, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment) (noting that "[a] determination that a gerrymander violates the law" must "rest ... on a conclusion that [political] classifications ... were applied in ... a way unrelated to any legitimate legislative objective"); Bandemer , 478 U.S. at 141, 106 S.Ct. 2797 ("If there were a discriminatory effect and a discriminatory intent, then the legislation would be examined for valid underpinnings."). As a general matter, once a plaintiff establishes a prima facie case that a redistricting plan violates the Equal Protection Clause, the burden shifts to the governmental defendant to prove that a legitimate state interest or other neutral factor justified such discrimination. See, e.g. , Cooper , 137 S.Ct. at 1464 (racial gerrymandering); Brown , 462 U.S. at 842–43, 103 S.Ct. 2690 (one-person, one-vote). Plaintiffs contend—and Legislative Defendants do not dispute—that the same burden-shifting approach applies in partisan gerrymandering cases. Accordingly, we must determine whether Legislative Defendants have proven that the 2016 Plan's discriminatory effects are attributable to a legitimate state interest or other neutral explanation.

1.

Legislative Defendants first argue that Democratic voters tend to congregate in North Carolina's urban centers—i.e. , that North Carolina's political geography exhibits "natural packing"—and therefore the 2016 Plan's pro-Republican partisan bias is attributable to such natural packing, rather than invidious partisan discrimination. See Ex. 5058, at 10–13; Vieth , 541 U.S. at 289–90, 124 S.Ct. 1769 (plurality op.) (describing " ‘natural’ packing"). To support their natural packing argument, Legislative Defendants rely on a shaded map prepared by Dr. Hood reflecting the partisan makeup of North Carolina's VTDs. Ex. 5058, at 9–10. According to Dr. Hood, that map "visual[ly]" demonstrates that "Democrats appear to be located in urban areas (e.g. Charlotte, Asheville, Winston–Salem, Greensboro, Durham, and Raleigh) and within the blackbelt area of the state that runs through the coastal plain subregion," whereas "Republican partisans are much more geographically dispersed, producing a larger footprint within the state." Id. at 9–10 (footnote text altered). We agree with Legislative Defendants that supporters of Democratic candidates often cluster in North Carolina's urban areas, but we find that this clustering does not explain the 2016 Plan's pro-Republican discriminatory effects, and for several reasons.

First, Dr. Hood conceded on cross-examination that, in drawing the 2016 Plan, the General Assembly repeatedly divided Democratic clusters. For example, Dr. Hood conceded that the 2016 Plan "cracked" the naturally occurring Democratic cluster in the City of Asheville and Buncombe County into two districts that he classified as "safe" Republican districts. Trial Tr. IV, at 40:1–43:4. Dr. Hood further conceded that had the General Assembly kept that naturally occurring Democratic cluster whole, it would have been more likely that voters in the cluster would have elected a Democratic candidate. Id. at 42:23–43:4. Dr. Hood similarly conceded that the 2016 Plan "cracked" several other naturally occurring Democratic clusters and, by "submerg[ing]" likely Democratic voters in pro-Republican districts, made it easier for Republican candidates to prevail in more districts. Id. at 43:5–50:25. Accordingly, testimony by Legislative Defendants' expert belies any argument that natural packing explains the 2016 Plan's discriminatory partisan effect.

Second, Dr. Mattingly and Dr. Chen's simulation analyses, both of which account for the state's political geography, found that "natural packing" of Democratic voters did not explain the 2016 Plan's partisan effects. In particular, based on his ensemble of 24,518 simulated congressional districting plans—all of which conformed to traditional redistricting criteria such as population equality, contiguity, keeping political subdivisions and precincts whole, compactness, and complying with the Voting Rights Act—Dr. Mattingly concluded that "the background structure in the geopolitical makeup of North Carolina, ... its geography, where its people live, where its voters in each party are distributed, and whether the African–American population is, and what that necessitates relative to the Voting Rights Act" did not explain the 2016 Plan's partisan bias. Trial Tr. I, at 91:20–92:19. Dr. Chen's analysis of his simulated districting plans—which conformed to the nonpartisan criteria adopted by the Committee—reached the same conclusion: the "political geography of North Carolina voters" does not explain the 2016 Plan's pro-Republican bias. Id. at 212:14–214:2.

Legislative Defendants have not provided any persuasive basis for calling into question Dr. Mattingly's and Dr. Chen's methods, findings, and conclusions. See supra Part II.A.2.b. And other than Dr. Hood's "visual" analysis, Legislative Defendants have not provided any contrary empirical analysis showing that the state's political geography does, in fact, explain the 2016 Plan's discriminatory effects. See Whitford , 218 F.Supp.3d at 914–15 (concluding that Wisconsin's political geography did not explain legislative districting plan's partisan bias when the defendant's natural packing argument was "based largely on ... shaded maps rather than quantitative analysis"). Accordingly, we find that North Carolina's political geography does not explain the 2016 Plan's discriminatory effects on supporters of non-Republican candidates.

2.

Next, Legislative Defendants suggest that the 2016 Plan's discriminatory effects are attributable to the General Assembly's legitimate interest in protecting incumbents elected under the 2011 Plan and the electoral benefits attributable to incumbency. Legislative Defendants are correct that state redistricting bodies have a legitimate interest, at least outside the remedial context, in drawing districts so as to avoid pairing incumbents in a single district. See Karcher , 462 U.S. at 740, 103 S.Ct. 2653. But we find that the General Assembly's efforts to protect incumbents do not explain the 2016 Plan's discriminatory partisan effects.

In particular, Dr. Chen's simulation analyses demonstrate that the General Assembly could achieve its interest in avoiding the pairing of incumbents without drawing a plan exhibiting the discriminatory effects of the 2016 Plan. Ex. 2010, at 15–19. Indeed, Dr. Chen's simulated plans advanced the Committee's goal of avoiding pairing incumbents more effectively than the 2016 Plan: unlike the 2016 Plan, which paired two of the state's thirteen incumbents, Dr. Chen drew 1,000 plans that did not pair any incumbents. Id. at 3, 15–19 ("These simulation results clearly reject any notion that an effort to protect incumbents might have warranted the extreme partisan bias observed in the [2016 Plan].").

Additionally, to ensure that the election data upon which he relied—the same data relied upon by Dr. Hofeller and prescribed by the Committee's Political Data criterion—adequately accounted for the benefits of incumbency, Dr. Chen performed a sensitivity analysis that accounted for the electoral advantages associated with incumbency. Id. at 26–31. Although that sensitivity analysis revealed, as expected, that incumbents enjoy electoral advantages, id. at 27 (finding that North Carolina congressional incumbents receive, on average, approximately 3 percent greater electoral support than nonincumbents), Dr. Chen found that the revealed electoral advantage associated with incumbency did not explain the 2016 Plan's pro-Republican bias, id. at 28–30, 32–37.

Dr. Chen's finding that incumbency does not explain the 2016 Plan's partisan bias is unsurprising given that the 2016 Plan sought to protect the incumbents elected under the 2011 Plan. As explained above, the General Assembly expressly drew the 2011 Plan "to minimize the number of districts in which Democrats would have an opportunity to elect a Democratic candidate." Hofeller Dep. 127:19–22; see also supra Part III.A.2–3. And the 2011 Plan had the effect of discriminating against supporters of non-Republican candidates and entrenching Republican control of the state's congressional delegation. Accordingly, the General Assembly's effort to protect incumbents elected under the 2011 Plan when it drew the 2016 Plan served to perpetuate the discriminatory partisan effects of the 2011 Plan.

Legislative Defendants nevertheless argue that Republican candidates' success in the 2016 election under the 2016 Plan was attributable to advantages associated with incumbency, including that the Republican incumbents attracted less experienced opponents and raised significantly more money than their opponents. Ex. 5058, at 6–7; Trial Tr. IV, at 51:1–53:12. But Dr. Hood conceded on cross-examination that the likelihood an incumbent will prevail in a redrawn district impacts the incumbent's ability to raise money and whether he draws a strong opponent. Trial Tr. IV, at 54:23–55:12. To that end, Dr. Hood further conceded that the Republican incumbents may have attracted weak opponents and raised substantially more money because the General Assembly drew the Republican incumbents districts in which they were likely to prevail—a possibility that Dr. Hood did not consider, much less evaluate. Id. at 54:9–59:18.

Given that Legislative Defendants' expert acknowledged that the 2016 Plan's discriminatory lines may have caused Republican incumbents' observed advantages, and that Legislative Defendants failed to offer any analyses rebutting Dr. Chen's rigorous quantitative analysis showing that the General Assembly's goal of protecting incumbents did not explain the 2016 Plan's pro-Republican bias, we find the General Assembly's interest in protecting incumbents and the electoral advantages associated with incumbency do not explain the 2016 Plan's discriminatory partisan effect.

* * * * *

In sum, we find that the General Assembly drew and enacted the 2016 Plan with intent to subordinate the interests of non-Republican voters and entrench Republican control of North Carolina's congressional delegation. We further find that a variety of evidence demonstrates that the 2016 Plan achieved the General Assembly's discriminatory partisan objective. And we find that neither North Carolina's political geography nor the General Assembly's interest in protecting incumbents explains the 2016 Plan's discriminatory effects. Accordingly, we conclude that the 2016 Plan constitutes an unconstitutional partisan gerrymander in violation of the Equal Protection Clause of the Fourteenth Amendment.

IV.

Next, we consider Plaintiffs' claims under the First Amendment. The First Amendment, through the Due Process Clause of the Fourteenth Amendment, prohibits states from making any law "abridging the freedom of speech." U.S. Const. amend. I. Partisan gerrymandering—again, "the drawing of legislative district lines to subordinate adherents of one political party and entrench a rival party in power," Ariz. State Leg. , 135 S.Ct. at 2658—implicates First Amendment rights because "political 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, 49 L.Ed.2d 547 (1976). The First Amendment "has its fullest and most urgent application to speech uttered during a campaign for political office." Citizens United v. Fed. Election Comm'n , 558 U.S. 310, 339–40, 130 S.Ct. 876, 175 L.Ed.2d 753 (2010) (internal quotation marks omitted). To that end, the First Amendment protects "the right of individuals to associate for the advancement of political beliefs, and the right of qualified voters, regardless of their political persuasion , to cast their votes effectively." Williams , 393 U.S. at 30–31, 89 S.Ct. 5 (emphasis added).

A.

Several lines of precedent bear on the application of the First Amendment to partisan gerrymanders. To begin, by favoring one set of political beliefs over another, partisan gerrymanders implicate the First Amendment prohibition on "viewpoint discrimination." See Rosenberger v. Rector & Visitors of Univ. of Va. , 515 U.S. 819, 829, 115 S.Ct. 2510, 132 L.Ed.2d 700 (1995) ; Vieth , 541 U.S. at 314, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment) ("First Amendment concerns arise where a State enacts a law that has the purpose and effect of subjecting a group of voters or their party to disfavored treatment by reason of their views ." (emphasis added)). The First Amendment prohibits the government from favoring or disfavoring particular viewpoints, and, therefore, "[t]he 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." Rosenberger , 515 U.S. at 829, 115 S.Ct. 2510. "At its most basic, the test for viewpoint discrimination is whether—within the relevant subject category—the government has singled out a subset of messages for disfavor based on the views expressed." Matal v. Tam , ––– U.S. ––––, 137 S.Ct. 1744, 1766, 198 L.Ed.2d 366 (2017) (Kennedy, J., concurring in part and concurring in the judgment). Viewpoint discrimination is "presumptively unconstitutional," Rosenberger , 515 U.S. at 830, 115 S.Ct. 2510 (internal quotation marks omitted), and therefore subject to "strict scrutiny," McCullen v. Coakley , ––– U.S. ––––, 134 S.Ct. 2518, 2530, 189 L.Ed.2d 502 (2014) (explaining that a governmental action amounting to viewpoint discrimination survives strict scrutiny only if the action is "the least restrictive means of achieving a compelling state interest").

Relatedly, by seeking to dilute the electoral speech of supporters of disfavored parties or candidates, partisan gerrymandering runs afoul of the First Amendment's prohibition on laws that disfavor a particular group or class of speakers. Citizens United , 558 U.S. at 340, 130 S.Ct. 876 (explaining that "[s]peech restrictions based on the identity of the speaker are all too often simply a means to control content"). The First Amendment prohibits such laws because "[b]y taking the right to speak from some and giving it to others, the Government deprives the disadvantaged person or class of the right to use speech to strive to establish worth, standing, and respect for the speaker's voice." Id. at 340–41, 130 S.Ct. 876. In the context of political speech, in particular, the Supreme Court repeatedly has applied the First Amendment's prohibition on "restrictions on certain disfavored speakers" to strike down electoral laws that disfavor a particular group of speakers. Id. at 341, 130 S.Ct. 876 ; First Nat'l Bank of Bos. v. Bellotti , 435 U.S. 765, 784, 98 S.Ct. 1407, 55 L.Ed.2d 707 (1978). And when, as is the case with a partisan gerrymander, a restriction on one group of speakers "suggests an attempt to give one side of a debatable public question an advantage in expressing its views to the people, the First Amendment is plainly offended." Belotti , 435 U.S. at 785–86, 98 S.Ct. 1407 (footnote omitted). Like viewpoint discrimination, governmental actions that discriminate against a particular group or class of speakers are subject to "strict scrutiny." See Citizens United , 558 U.S. at 340, 130 S.Ct. 876.

Third, by disfavoring a group of voters based on their prior votes and political association, partisan gerrymandering implicates the First Amendment's prohibition on burdening or penalizing individuals for engaging in protected speech. Vieth , 541 U.S. at 314, 124 S.Ct. 1769 (2004) (Kennedy, J., concurring in the judgment) (explaining partisan gerrymandering violates "the First Amendment interest of not burdening or penalizing citizens because of their participation in the electoral process, their voting history, their association with a political party, or their expression of political views"). The Supreme Court has explained that the government cannot "penalize[ ]" a person for engaging in "constitutionally protected speech or associations" because such indirect regulation of speech would "allow the government to produce a result which it could not command directly." Perry v. Sindermann , 408 U.S. 593, 597, 92 S.Ct. 2694, 33 L.Ed.2d 570 (1972) (internal quotation marks and alterations omitted). The Supreme Court's First Amendment retaliation jurisprudence represents a specific application of the general principle that even when the law affords the government the authority to make discretionary decisions—like firing or promoting an employee or allowing public use of a governmental facility—the government may not exercise such discretion "in a narrowly partisan or political manner." Bd. of Educ., Island Trees Union Free Sch. Dist. No. 26 v. Pico , 457 U.S. 853, 870–71, 102 S.Ct. 2799, 73 L.Ed.2d 435 (1982) (plurality opinion). For example, although the government retains discretion to curate public school libraries, "[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. ; see also id. at 907, 102 S.Ct. 2799 (Rehnquist, J., dissenting) ("I can cheerfully concede all of this.").

Courts have distilled a three-prong test from the Supreme Court's First Amendment retaliation jurisprudence, examining whether (1) the plaintiff's "speech was protected;" (2) "the defendant's ... retaliatory action adversely affected the plaintiff's constitutionally protected speech;" and (3) "a causal relationship exists between [the plaintiff's] speech and the defendant's retaliatory action." See, e.g. , Suarez Corp. Indus. v. McGraw , 202 F.3d 676, 686 (4th Cir. 2000). Examining these considerations, the Supreme Court repeatedly has struck down as violative of the First Amendment government actions that burden or penalize an individual or group for engaging in political speech. See, e.g. , Rutan v. Republican Party of Ill. , 497 U.S. 62, 65, 110 S.Ct. 2729, 111 L.Ed.2d 52 (1990) (concluding that First Amendment prohibits government employers from making "promotion, transfer, recall, and hiring decisions involving low-level public employees ... based on party affiliation and support"); Elrod , 427 U.S. at 373, 96 S.Ct. 2673 (holding that First Amendment prohibits government officials from discharging or threatening to discharge lower-level public employees based on their political affiliation).

Finally, partisan gerrymandering implicates First Amendment precedent dealing with electoral regulations that have the potential to burden political speech or association. See, e.g. , Burdick v. Takushi , 504 U.S. 428, 112 S.Ct. 2059, 119 L.Ed.2d 245 (1992) ; Anderson v. Celebrezze , 460 U.S. 780, 103 S.Ct. 1564, 75 L.Ed.2d 547 (1983). The First Amendment demands judicial scrutiny of state election regulations because regulations that "govern[ ] the registration and qualifications of voters, the selection and eligibility of candidates, or the voting process itself, inevitably affect[ ]—at least to some degree—the individual's right to vote and his right to associate with others for political ends." Anderson , 460 U.S. at 788, 103 S.Ct. 1564. Because states' "important regulatory interests are generally sufficient to justify reasonable, nondiscriminatory restrictions," id. , the Supreme Court applies "sliding-scale" scrutiny to state election regulations, see Burdick , 504 U.S. at 433–34, 112 S.Ct. 2059. In particular, "[a] court considering a challenge to a state election law must weigh ‘the character and magnitude of the asserted injury to the rights protected by the First and Fourteenth Amendments that the plaintiff seeks to vindicate’ against ‘the precise interests put forward by the State as justifications for the burden imposed by its rule,’ taking into consideration ‘the extent to which those interests make it necessary to burden the plaintiff's rights.’ " Id. at 434, 112 S.Ct. 2059 (quoting Anderson , 460 U.S. at 789, 103 S.Ct. 1564, Tashjian v. Republican Party of Conn. , 479 U.S. 208, 213–14, 107 S.Ct. 544, 93 L.Ed.2d 514 (1986) ). Under this test, "[e]lection regulations that impose a severe burden on associational rights are subject to strict scrutiny." Wash. State Grange v. Wash. State Republican Party , 552 U.S. 442, 451, 128 S.Ct. 1184, 170 L.Ed.2d 151 (2008). By contrast, "[i]f a statute imposes only modest burdens ... then ‘the State's important regulatory interests are generally sufficient to justify reasonable, nondiscriminatory restrictions.’ " Id. at 452, 128 S.Ct. 1184 (quoting Anderson , 460 U.S. at 788, 103 S.Ct. 1564 ).

Applying that test, the Court has "repeatedly upheld reasonable, politically neutral regulations that have the effect of channeling expressive activity at the polls." Id. at 438, 112 S.Ct. 2059 (emphasis added). By contrast, the Supreme Court has repeatedly struck down as violative of the First Amendment facially neutral electoral regulations that had the effect of burdening particular parties, candidates, or groups of voters. See, e.g. , Tashjian , 479 U.S. at 225, 107 S.Ct. 544 (concluding that state's enforcement of statute requiring closed primaries, against the will of the Republican party, violated First Amendment); Anderson , 460 U.S. at 806, 103 S.Ct. 1564 (striking down state candidate filing deadline because it posed unjustified burden on third-party candidates and voters who supported such candidates, where the "interests of the voters who chose to associate together" for political ends constituted the Court's "primary concern"). These cases reflect the governing principle that "in exercising their powers over elections and in setting qualifications for voters, the States may not infringe upon basic constitutional protections," including enacting "election laws [that] so impinge upon freedom of association as to run afoul of the First and Fourteenth Amendments." Kusper v. Pontikes , 414 U.S. 51, 56–57, 94 S.Ct. 303, 38 L.Ed.2d 260 (1973).

Against these many, multifaceted lines of precedent, the First Amendment's applicability to partisan gerrymandering is manifest. How can the First Amendment prohibit the government from disfavoring certain viewpoints, yet allow a legislature to enact a districting plan that disfavors supporters of a particular set of political beliefs? How can the First Amendment bar the government from disfavoring a class of speakers, but allow a districting plan to disfavor a class of voters? How can the First Amendment protect government employees' political speech rights, but stand idle when the government infringes on voters' political speech rights? And how can the First Amendment ensure that candidates ascribing to all manner of political beliefs have a reasonable opportunity to appear on the ballot, and yet allow a state electoral system to favor one set of political beliefs over others? We conclude that the First Amendment does not draw such fine lines.

The 2016 Plan, in particular, implicates all four of these lines of precedent. The 2016 Plan discriminates against a particular viewpoint: voters who oppose the Republican platform and Republican candidates. The 2016 Plan also discriminates against a particular group of speakers: non-Republican candidates and voters who support non-Republican candidates. The General Assembly's use of Political Data—individuals' votes in previous elections—to draw district lines to dilute the votes of individuals likely to support non-Republican candidates imposes burdens on such individuals based on their past political speech and association. And the 2016 Plan's partisan favoritism excludes it from the class of "reasonable, politically neutral" electoral regulations that pass First Amendment muster. Burdick , 504 U.S. at 438, 112 S.Ct. 2059.

B.

Notwithstanding the evident applicability of the First Amendment to partisan gerrymandering, and the 2016 Plan in particular, neither the Supreme Court nor lower courts have settled on a framework for determining whether a partisan gerrymander violates the First Amendment. League Plaintiffs, in accordance with the approach taken in Whitford , assert that the three-prong framework governing partisan gerrymandering claims under the Equal Protection Clause also applies to partisan gerrymandering claims under the First Amendment. This requires a plaintiff to demonstrate (1) discriminatory intent, (2) discriminatory effects, and (3) a lack of justification for the discriminatory effects. League Br. 3; Whitford , 218 F.Supp.3d at 884. That inquiry mirrors the considerations the Supreme Court evaluates in First Amendment retaliation cases and First Amendment challenges to election regulations, see supra Part IV.A; infra Part IV.C, albeit using somewhat different nomenclature. Legislative Defendants agree that to the extent partisan gerrymandering is actionable under the First Amendment—and we conclude that it is, see supra Parts II.B, IV.A —the governing legal framework is no "different from any test which might apply under the Fourteenth Amendment." Leg. Defs.' FOF 105–06 (" ‘[T]he [F]irst amendment, like the [T]hirteenth, offers no protection of voting rights beyond that afforded by the [F]ourteenth and [F]ifteenth Amendments.’ " (quoting Washington v. Finlay , 664 F.2d 913, 927–28 (4th Cir. 1981) )).

Common Cause Plaintiffs, by contrast, assert that once a plaintiff proves that a redistricting body intended for a districting plan to discriminate against voters likely to support a disfavored candidate or party—and thereby intended to engage in discrimination against a particular viewpoint and group of speakers—a court must subject the plan to strict scrutiny, upholding the plan " 'only if [Defendants] prove[ ] that [it is] narrowly tailored to serve compelling state interests." Common Cause Br. 7–8 (quoting Reed v. Town of Gilbert , ––– U.S. ––––, 135 S.Ct. 2218, 2226, 192 L.Ed.2d 236 (2015) ). Accordingly, unlike League Plaintiffs, Common Cause Plaintiffs take the position that once a plaintiff demonstrates that a districting plan is motivated by invidious partisan intent, the First Amendment does not require a plaintiff to demonstrate that a plan has concrete discriminatory effects.

We agree with Common Cause Plaintiffs that the Supreme Court's demonstrated dim view of viewpoint discrimination, laws that discriminate against a class of speakers, and laws that impose severe burdens on associational rights provides strong theoretical support for their position that invidious partisan discrimination, even absent a showing of concrete discriminatory effects, "is itself an injury to the First Amendment rights of the intended targets or victims." Common Cause Br. 9. To that end, the Supreme Court repeatedly has struck down election laws and regulations that discriminate against a particular viewpoint or group of speakers, even in the absence of evidence that the law or regulation had, or would have, a concrete effect on the outcome of an election. See, e.g. , Citizens United , 558 U.S. at 365–66, 130 S.Ct. 876 (striking down statute placing certain restrictions on political advocacy by corporations); Fed. Election Comm'n v. Wisc. Right to Life, Inc. , 551 U.S. 449, 481, 127 S.Ct. 2652, 168 L.Ed.2d 329 (2007) (opinion of Roberts, C.J.) (same); id. at 504, 127 S.Ct. 2652 (Scalia, J., concurring in the judgment) (same). Likewise, courts reviewing election regulations under the Anderson / Burdick framework apply strict scrutiny to election regulations that are not "even-handed" or "politically neutral." Dudum v. Arntz , 640 F.3d 1098, 1106 (9th Cir. 2011) ; see also Clingman v. Beaver , 544 U.S. 581, 603–04, 125 S.Ct. 2029, 161 L.Ed.2d 920 (2005) (O'Connor, J. concurring in part) (concluding that burden imposed by electoral regulation was not "severe," and thus not subject to strict scrutiny, because it imposed "only a modest and politically neutral burden on associational rights").

Nevertheless, Supreme Court precedent appears to bar a plaintiff from successfully challenging a partisan gerrymander solely based on evidence that a redistricting body enacted a districting plan with discriminatory partisan intent. See LULAC , 548 U.S. at 418, 126 S.Ct. 2594 (opinion of Kennedy, J.) ("[A] successful claim attempting to identify unconstitutional acts of partisan gerrymandering must do what appellants' sole-motivation theory explicitly disavows: show a burden, as measured by a reliable standard, on the complainants' representational rights. For this reason, a majority of the Court rejected a test proposed in Vieth that is markedly similar to the one appellants present today."); id. at 511–12, 126 S.Ct. 2594 (Scalia, J., concurring in part and dissenting in part). To that end, the one lower court to put forward a unique framework for adjudicating partisan gerrymandering claims under the First Amendment since the Supreme Court decided LULAC required that a partisan gerrymandering plaintiff prove that he experienced a "demonstrable and concrete adverse effect" on his First Amendment rights. Shapiro , 203 F.Supp.3d at 598.

In light of this precedent, we assume that the Supreme Court would review First Amendment partisan gerrymandering claims in accordance with the intermediate scrutiny applied in retaliation cases and challenges to election regulations that do not impose a "severe" burden on voting rights. Drawing on that precedent, we derive a three-prong test requiring Plaintiffs to prove: (1) that the challenged districting plan was intended to favor or disfavor individuals or entities that support a particular candidate or political party, (2) that the districting plan burdened the political speech or associational rights of such individuals or entities, and (3) that a causal relationship existed between the governmental actor's discriminatory motivation and the First Amendment burdens imposed by the districting plan.

1.

The intent prong principally derives from the causation component in First Amendment retaliation cases. In such cases, a "plaintiff must show a causal connection between a defendant's retaliatory animus and subsequent injury in any sort of retaliation action." Hartman v. Moore , 547 U.S. 250, 259, 126 S.Ct. 1695, 164 L.Ed.2d 441 (2006) (emphasis added). Put differently, a plaintiff must show that her protected First Amendment activities were a "motivating factor" behind the challenged retaliatory action. Mt. Healthy City Sch. Dist. Bd. of Educ. v. Doyle , 429 U.S. 274, 287, 97 S.Ct. 568, 50 L.Ed.2d 471 (1977). The motivating-factor requirement in First Amendment retaliation claims parallels the intent requirement in Equal Protection Claims. Id. at 287, 97 S.Ct. 568 n.2 (citing Arlington Heights , 429 U.S. at 270–71, 97 S.Ct. 555 ). Relying on this precedent, lower courts have concluded that the motivating-factor requirement renders proof of a governmental actor's intent to burden speech or associational rights an essential element of First Amendment retaliation claims. See, e.g. , Greenwich Citizens Comm., Inc. v. Ctys. Of Warren & Washington Indus. Dev. Agency , 77 F.3d 26, 32 (2d Cir. 1996) ("[R]etaliatory intent is required for a retaliatory First Amendment claim."); Mendocino Envtl. Ctr. v. Mendocino Cty. , 14 F.3d 457, 464 (9th Cir. 1994) ("The defendant's intent is an element of the [retaliation] claim." (emphasis removed)); Shapiro , 203 F.Supp.3d at 597.

Applying the guidelines for assessing discriminatory intent in Arlington Heights , we previously found that Plaintiffs adduced more-than-sufficient evidence to prove that, in enacting the 2016 Plan, the General Assembly intended to "subordinate" the interests of entities and voters who supported, or were likely to support, non-Republican candidates. See supra Part III.A. Given that the Arlington Heights intent inquiry parallels the intent inquiry in First Amendment retaliation claims, see Mt. Healthy , 429 U.S. at 287 n.2, 97 S.Ct. 568, we likewise find that Plaintiffs satisfied their burden to demonstrate that the General Assembly intended to burden the speech and associational rights of such entities and voters.

2.

Next, we must determine whether the 2016 Plan in fact burdened First Amendment rights. The requirement that a plaintiff demonstrate that a partisan gerrymander burdens political speech or associational rights derives from both retaliation and election regulation cases. In the context of retaliation claims, even when, as here, a challenged governmental action does not flatly prohibit protected speech or association, the action nonetheless burdens First Amendment rights if it "has a chilling effect or an adverse impact" on speech or associational rights. The Balt. Sun Co. v. Ehrlich , 437 F.3d 410, 416 (4th Cir. 2005). To constitute an actionable First Amendment burden, the chilling effect or adverse impact must be more than de minimis . See, e.g. , McKee v. Hart , 436 F.3d 165, 170 (3d Cir. 2006) ; ACLU of Md., Inc. v. Wicomico Cty. , 999 F.2d 780, 786 n.6 (4th Cir. 1993). Likewise, the Anderson / Burdick framework applied in election regulation cases requires a plaintiff to establish that a challenged regulation imposed a "burden" on political speech or associational rights. Crawford v. Marion Cty. Election Bd. , 553 U.S. 181, 189–90, 128 S.Ct. 1610, 170 L.Ed.2d 574 (2008) (opinion of Stevens, J.). The Court has refused to impose "any litmus test for measuring the severity of a burden that a state law imposes on a political party, an individual voter, or a discrete class of voters," instead requiring that "[h]owever slight [a] burden may appear ... it must be justified by relevant and legitimate state interests sufficiently weighty to justify the limitation." Id. at 191, 128 S.Ct. 1610 (internal quotation marks omitted).

Legislative Defendants argue that partisan gerrymandering does not "burden" First Amendment rights because it does not "prohibit" supporters of a disfavored party or candidate from speaking nor does it "chill" speech or "deter" such supporters "from engaging in political speech or association." Leg. Defs.' FOF 139. Put differently, the 2016 Plan does not "chill" First Amendment activities because "Plaintiffs are every bit as free under [the 2016 Plan] to run for office, express their political views, endorse and campaign for their favorite candidates, vote, or otherwise influence the political process through their expression." Kidd v. Cox , No. 1:06-cv-0997, 2006 WL 1341302, at *12 (N.D. Ga. 2006).

A governmental action "chills" speech if it is "likely [to] deter a person of ordinary firmness from the exercise of First Amendment rights." Benham v. City of Charlotte , 635 F.3d 129, 135 (4th Cir. 2011) (alteration in original) (internal quotation marks omitted). "Any chilling effect must be objectively reasonable. Nevertheless, a claimant need not show [he] ceased those activities altogether to demonstrate an injury in fact." Id. (alterations and internal citation omitted).

Under that standard, the record reveals that the 2016 Plan has had a chilling effect on reasonable North Carolinians' First Amendment activities. Multiple Plaintiffs testified that in "the most recent election, a lot of people did not come out to vote"—despite concerted get-out-the-vote efforts—"[b]ecause they felt their vote didn't count." Evans Dep. 16:4–9; accord, e.g. , Peck Dep. 27:20–24 ("I can't tell you how many people told me this election, Republicans as well as Democrats, ‘This system is rigged. My vote doesn't count.’ It was really hard to try to galvanize people to participate."). Likewise, in the 2016 election under the 2016 Plan, many organizations' "biggest struggle was to get people to vote." Peck Dep. 40:5–6. Voters and advocacy organizations elected not to participate in congressional races because they believed they could not "have a democratic—small "D"—democratic impact. It doesn't really matter for those races because of the gerrymandering because they're not competitive." Peck Dep. 30:20–24.

Additionally the League had difficulty "inform[ing] ... [and] engag[ing] voters in the process of voting and civic participation in their government." Klenz Dep. 59:16–17; see id. 44:15–25 (explaining that the League of Women Voters engages in "voter registration" and "Get Out The Vote" efforts). For example, the League testified that it had difficulty finding ways for their members to interact with "candidate[s] that [were] expected to win and projected to win," because those candidates were often not "motivated" to participate "in voter forums, debates, [or] voter guides, because the outcome is so skewed in favor or in disfavor of one or the other." Id. at 59:16–17, 60:6–10. Individual Plaintiffs also testified to the adverse impact of the districting plan on their ability to interact with and influence their representatives. See, e.g. , Brewer Dep. 24:8–25:6 (explaining that in "non-competitive districts" representatives from "both parties are not required to reach out to voters in the other party or even truly independent voters," and therefore such voters tend "to be poorly represented because their views and their potential votes are not fairly considered"). The 2016 Plan also chilled the speech and associational rights of voters affiliated with the North Carolina Democratic Party. Because Democratic candidates were unlikely to prevail in districts drawn by the General Assembly to elect Republicans, it "ma[d]e[ ] it extremely difficult" for the North Carolina Democratic Party "to raise funds and have resources and get the attention of the national congressional campaign committees and other lawful potential funders for congressional races in those districts." Goodwin Dep. 98:1–5. For the same reasons, the party had difficult recruiting strong candidates. Id. at 41:20–42:20; 60:23–61:16. Individual Plaintiffs testified to similar difficulty raising money, attracting candidates, and mobilizing voters to support the political causes and issues such Plaintiffs sought to advance. E.g. , Quinn Dep. 39:1–3 ("[Extreme gerrymandering] makes it harder for me [as a local organizer] to raise money; it makes it harder for me to recruit candidates; makes it harder to just mobilize a campaign."); Palmer Dep. 27:19–23 (recounting that citizens in one district asked for "help [to] recruit a candidate for [the citizens'] county [because] ... no Democrats [we]re going to run [t]here" given the significant obstacle to success posed by the partisan gerrymander); Morgan Dep. 23:21–25 ("[P]eople ... say no sense in us giving money to that candidate because [he or she] is unlikely to prevail, notwithstanding the merit of their positions.").

Expert testimony confirmed the reasonableness of North Carolinians' feelings that their votes "did not count" and the corresponding chilling effects on speech and associational activities. For example, the Republican candidate's vote share (56.10%) and margin of victory (12.20%) in the least Republican district which elected a Republican candidate under the 2016 Plan exceeded the thresholds at which political science experts, including Legislative Defendants' expert Dr. Hood, consider a district to be "safe"—i.e. , highly unlikely to change parties in subsequent elections. Ex. 5058, at 25, Trial Tr. IV, at 29:16–22, 86:21–88:5. Likewise, Dr. Jackman testified that it would require a swing of votes in Democratic candidates' favor of "historic magnitude" to strip the 2016 Plan of its pro-Republican bias. Trial Tr. II, at 54:24–55:9. And Dr. Hood testified that when a district's lines are drawn so that a particular party's candidate is likely to prevail, the opposing party will have difficulty attracting a strong candidate and raising money to support that candidate. Trial Tr. IV, at 54:9–59:18.

All of these chilling effects on speech and association—difficulty convincing voters to participate in the political process and vote, attracting strong candidates, raising money to support such candidates, and influencing elected officials—represent cognizable, and recognized, burdens on First Amendment rights. See, e.g. , Anderson , 460 U.S. at 792, 103 S.Ct. 1564 (finding that plaintiff was injured by election law that made "[v]olunteers ... more difficult to recruit and retain, media publicity and campaign contributions ... more difficult to secure, and voters ... less interested in the campaign"); Libertarian Party of Ohio v. Blackwell , 462 F.3d 579, 587 (6th Cir. 2006) (recognizing that electoral restrictions that "affect a political party's ability to perform its primary functions—organizing and developing, recruiting supporters, choosing a candidate, and voting for that candidate in a general election"—can constitute "severe" First Amendment burdens); Benisek v. Lamone , 266 F.Supp.3d 799, 834 (D. Md. 2017) (Niemeyer, J., dissenting) ("[T]he purposeful reduction of one party's effectiveness may well chill the protected expression of that party's voters, even if no individual plaintiff establishes, as a factual matter, that he was so chilled."), appeal docketed ––– U.S. ––––, 138 S.Ct. 543, ––– L.Ed.2d –––– ( 2017). Importantly, that partisan gerrymanders do not bar citizens from voting or expressing their political views does not render these First Amendment burdens any less significant. Cal. Democratic Party v. Jones , 530 U.S. 567, 581, 120 S.Ct. 2402, 147 L.Ed.2d 502 (2000) ("We have consistently refused to overlook an unconstitutional restriction upon some First Amendment activity simply because it leaves other First Amendment activity unimpaired.").

Additionally, Legislative Defendants' myopic focus on whether a partisan gerrymander, and the 2016 Plan in particular, "chilled" or "deterred" protected speech or association ignores that a retaliatory governmental action also poses a constitutionally cognizable "burden" when it "adversely affects[s]" the speaker and the candidate or political groups with whom he seeks to associate. Rutan , 497 U.S. at 73, 110 S.Ct. 2729 ; Suarez , 202 F.3d at 686. As detailed above, myriad evidence establishes that the 2016 Plan makes it easier for supporters of Republican candidates to translate their votes into seats in the state's congressional delegation and diminishes the need for Republican representatives to respond to the interests of voters who support non-Republican candidates. See supra Part III.B. Accordingly, even if the speech of voters who support non-Republican candidates was not in fact chilled—if, for example, they had all continued to vote for, speak on behalf of, donate money to, and campaign for such candidates—the 2016 Plan nonetheless "adversely affected" such voters' First Amendment rights by diluting the electoral power of their votes. Shapiro , 203 F.Supp.3d at 597–98 (recognizing that "dilution" of disfavored party's electoral power constitutes adverse effect cognizable under the First Amendment).

The principle that vote dilution—the intentional diminishment of the electoral power of supporters of a disfavored party and enhancement of the electoral power of supporters of a favored party—constitutes an actionable adverse effect on political speech and associational rights derives from bedrock First Amendment principles. "[T]he concept that government may restrict the speech of some elements of our society in order to enhance the relative voice of others is wholly foreign to the First Amendment ." Buckley v. Valeo , 424 U.S. 1, 48–49, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976) (emphasis added), superseded by statute on other grounds as stated in McConnell v. Fed. Election Comm'n , 540 U.S. 93, 124 S.Ct. 619, 157 L.Ed.2d 491 (2003). To that end, the government may not cap the amount of independent expenditures individuals, entities, and political parties may make on behalf of a "clearly identified candidate." Id. at 45, 96 S.Ct. 612.

Likewise, it is beyond cavil that the First Amendment would forbid the government from making large public spaces available for speakers advocating for a favored political party, while allowing supporters of disfavored speakers only to speak in smaller public venues, simply because government officials preferred the message of the favored party's speakers. Nor is there any question that the government would violate the First Amendment if it allowed supporters or candidates of one party to speak with a bullhorn but barred candidates from other parties from doing the same. Although the supporters of the disfavored candidate or party remain free to speak as much as they wish—i.e. their speech is not chilled—the government nonetheless violates the First Amendment by "enhanc[ing] the relative voice" of the favored party. Buckley , 424 U.S. at 48–49, 96 S.Ct. 612. Just as the government may not altruistically "equaliz[e] the relative ability of individuals and groups to influence the outcome of elections," Citizens United , 558 U.S. at 350, 130 S.Ct. 876 (internal quotation mark omitted), neither may the government invidiously amplify one group of citizens' speech and reduce that of all other citizens in order to influence the outcome of elections, see Shapiro , 203 F.Supp.3d at 598 ("While citizens have no right to be assigned to a district that is likely to elect a representative that shares their views, the State also may not intentionally drown out the voices of certain voters by reason of their views." (emphasis added)). That is particularly true in the republican form of government adopted by the Framers, in which elected officials represent the interests of "the People" in making governing decisions. U.S. Const. art. I, § 2; see infra Part V. When a legislature draws a congressional districting plan designed to enhance the electoral power of voters likely to support candidates of a favored party and the districting plan achieves that intended goal by electing more Representatives from the favored party than would have prevailed under an unbiased plan—as was the case with the 2016 Plan in the 2016 election—then the legislature unconstitutionally has "enhanced the relative voice" of the favored party in Congress, at the expense of the viewpoint of the supporters of disfavored parties.

Contrary to Legislative Defendants' assertions, the 2016 Plan's chilling effects and adverse impacts are more than de minimis . Even a "slight" burden on "a political party, an individual voter, or a discrete class of voters" can violate the First Amendment if not supported by a justification of commensurate magnitude—as is the case here. See Crawford , 553 U.S. 181, 189–90, 128 S.Ct. 1610, 170 L.Ed.2d 574 (2008) (opinion of Stevens, J.). And the myriad burdens on political speech and associational rights attributable to the 2016 Plan—including decreased voter engagement, difficulty raising money and attracting candidates, and vote dilution—are of a different magnitude than numerous retaliatory actions that courts have found to constitute more than de minimis burdens on First Amendment rights. See, e.g. , Hines v. Gomez , 108 F.3d 265, 269 (9th Cir. 1997) (filing of single "false [disciplinary] charge infringed ... First Amendment right[s]"); Crawford–El v. Britton , 93 F.3d 813, 826 (D.C. Cir. 1996) ("[P]ecuniary losses ... sustained in the form of the costs of shipping ... boxes and replacing clothing, though small, might well deter a person of ordinary firmness ... from speaking again."), vacated on other grounds , 523 U.S. 574, 118 S.Ct. 1584, 140 L.Ed.2d 759 (1998) ; Sloman v. Tadlock , 21 F.3d 1462, 1470 (9th Cir. 1994) (holding that factfinder could reasonably conclude that a police officer's "decisions to issue a citation and warnings to" a citizen expressing his political beliefs "chilled the political expression of [the citizen] and his group"); see also Anderson , 460 U.S. at 792, 103 S.Ct. 1564 (1983) (finding that plaintiff candidate was burdened by election law that made "[v]olunteers ... more difficult to recruit and retain, media publicity and campaign contributions ... more difficult to secure, and voters ... less interested in the campaign," even in the absence of evidence the candidate would have prevailed in election).

Taken together, we find that Plaintiffs' evidence establishes that the 2016 Plan's pro-Republican bias had the effect of chilling the political speech and associational rights of individuals and entities that support non-Republican candidates. And we further find that the 2016 Plan adversely affected such individuals' and entities' First Amendment rights by diluting the electoral speech and power of voters who support non-Republican candidates. Therefore, we find that Plaintiffs' evidence is more-than-adequate to establish that the 2016 Plan burdened their political speech and associational rights.

3.

Like the burden requirement, the causation requirement derives from both First Amendment retaliation and election regulation cases. In retaliation cases, the causation element not only requires a plaintiff to demonstrate retaliatory intent, it also allows a governmental actor to escape liability if the actor demonstrates it would have taken the challenged action "even in the absence of the protected conduct." Mt. Healthy , 429 U.S. at 287, 97 S.Ct. 568 ; Hartman , 547 U.S. at 260, 126 S.Ct. 1695 (explaining that a governmental "action colored by some degree of bad motive does not amount to a constitutional tort if that action would have been taken anyway"). Similarly, the Anderson / Burdick framework applied in election regulation cases requires that courts assess " ‘the precise interests put forward by the State as justifications for the burden imposed by its rule,’ taking into consideration ‘the extent to which those interests make it necessary to burden the plaintiff's rights.’ " Burdick , 504 U.S. at 434, 112 S.Ct. 2059 (quoting Anderson , 460 U.S. at 789, 103 S.Ct. 1564, and Tashjian , 479 U.S. at 213–14, 107 S.Ct. 544 ). Accordingly, under the causation prong, a challenged districting plan that burdens political speech and associational rights nonetheless passes First Amendment muster if legitimate state interests, unrelated to the redistricting body's intent to burden the rights of supporters of a disfavored party, justify the First Amendment burdens imposed by the plan.

As explained above, the 2016 Plan burdens First Amendment rights both by chilling voters, candidates, and parties' participation in the political process and by diluting the electoral power of supporters of non-Republican candidates. In evaluating Plaintiffs' claims under the Equal Protection Clause, we found that neither North Carolina's political geography nor any other legitimate redistricting objective justified the 2016 Plan's subordination of the interests of non-Republican voters. See supra Part III.C. And it is axiomatic that the government has no legitimate interest in "restrict[ing] the speech of some elements of our society in order to enhance the relative voice of others." Buckley , 424 U.S. at 48–49, 96 S.Ct. 612. Accordingly, we find that the General Assembly's discriminatory animus against non-Republican voters, candidates, and parties caused the 2016 Plan's burdens on such voters, candidates, and parties' political speech and associational rights.

* * * * *

In sum, we find (1) that the 2016 Plan was intended to disfavor supporters of non-Republican candidates based on those supporters' past expressions of political beliefs, (2) that the 2016 Plan burdened such supporters' political speech and associational rights, and (3) that a causal relationship existed between the General Assembly's discriminatory motivation and the First Amendment burdens imposed by the 2016 Plan. Accordingly, we conclude that the 2016 Plan violates the First Amendment.

V.

Finally, we turn to Common Clause Plaintiffs' claims under Article I of the Constitution. Common Cause Plaintiffs assert the 2016 Plan runs afoul of two provisions in Article I: Article I, section 2, which provides that the "House of Representatives shall be composed of Members chosen ... by the People," and the Elections Clause, which provides that "the Times, Places and Manner of holding Elections for ... Representatives, shall be prescribed in each State by the Legislature thereof; but the Congress may at any time by Law make or alter such Regulations," U.S. Const. art. I, § 4, cl. 1. Although Common Cause Plaintiffs assert distinct claims under Article 1, section 2 and the Elections Clause, framing era records and Supreme Court doctrine reveal that the two provisions are closely intertwined.

A.

Because the right to elect Representatives to Congress "ar[ose] from the Constitution itself," the States have no "reserved" or "sovereign" authority to adopt laws or regulations governing congressional elections. Thornton , 514 U.S. at 802–05, 115 S.Ct. 1842 ; id. at 802, 115 S.Ct. 1842 ("As Justice Story recognized, ‘the states can exercise no powers whatsoever, which exclusively spring out of the existence of the national government, which the constitution does not delegate to them.... No state can say, that it has reserved, what it never possessed.’ " (quoting Story, 1 Commentaries on the Constitution of the United States § 627 (3d ed. 1858)). Rather, the Constitution—and the Elections Clause in particular—delegates to the States the power to impose certain types of laws and regulations governing congressional elections, including laws or regulations establishing congressional districts. Id. at 802–05, 115 S.Ct. 1842 ; see also Brown v. Sec'y of State of Fla. , 668 F.3d 1271, 1283 (11th Cir. 2012) ("[S]tates have the delegated power under the Elections Clause to create districts for congressional elections."). But unless the Elections Clause or another constitutional provision delegates to the States the authority to impose a particular type of election law or regulation, "such a power does not exist." Thornton , 514 U.S. at 805, 115 S.Ct. 1842.

The plain language of the Elections Clause confers on the States the authority to regulate the "Times, Places, and Manner" of holding congressional elections. U.S. Const. art. I, sec. 4. During the Constitutional Convention, James Madison provided a list of examples of the types of regulations that would fall within States' authority to regulate the "Times, Places, and Manner" of holding elections: "whether the electors should vote by ballot or viva voce , should assemble at this place or that place; should be divided into districts or all meet at one place, sh[oul]d all vote for all the representatives; or all in a district vote for a number allotted to the district." Debates at 423–24. The Framers, therefore, "understood the Elections Clause as a grant of authority to issue procedural regulations ." Thornton , 514 U.S. at 833, 115 S.Ct. 1842 (emphasis added).

In accordance with the intent of the Framers, the Supreme Court has held that "[t]he Elections Clause gives States authority ‘to enact numerous requirements as to procedure and safeguards which experience shows are necessary in order to enforce the fundamental right involved.’ " Id. (emphasis added) (quoting Smiley v. Holm , 285 U.S. 355, 366, 52 S.Ct. 397, 76 L.Ed. 795 (1932) ). Put differently, the Elections Clause empowers the States to promulgate "regulations designed to ensure that elections are fair and honest and that some sort of order rather than chaos accompanies the democratic processes." Id. at 834–35, 115 S.Ct. 1842 (emphasis added) (internal quotation marks and alterations omitted).

The States' broad, delegated power under the Election Clause, however, is not without limit. See, e.g. , Cook v. Gralike , 531 U.S. 510, 527, 121 S.Ct. 1029, 149 L.Ed.2d 44 (2001) (Kennedy, J. concurring) ("The Elections Clause thus delegates but limited power over federal elections to the States."); Montano v. Lefkowitz , 575 F.2d 378, 385 (2d Cir. 1978) (Friendly, J.) (" Wesberry makes clear that the apparent breadth of the power granted to state legislatures by [the Elections Clause], is not a carte blanche."). In particular, "in exercising their powers of supervision over elections and in setting qualifications for voters, the States may not infringe upon basic constitutional protections." Kusper , 414 U.S. at 56–57, 94 S.Ct. 303 ; see also Tashjian , 479 U.S. at 217, 107 S.Ct. 544 ("The power to regulate the time, place, and manner of elections does not justify, without more, the abridgement of fundamental rights."). For example, in Wesberry , the Court held that the Elections Clause does not "immunize state congressional apportionment laws which debase a citizen's right to vote." 376 U.S. at 7, 84 S.Ct. 526. Likewise, the Elections Clause does not serve "as a source of power [for States] to dictate electoral outcomes, to favor or disfavor a class of candidates, or to evade important constitutional restraints." Thornton , 514 U.S. at 833–34, 115 S.Ct. 1842. Put differently, the States' authority under the elections clause extends only to "neutral provisions as to the time, place, and manner of elections." Gralike , 531 U.S. at 527, 121 S.Ct. 1029 (emphasis added).

B.

Under this precedent, we conclude that the 2016 Plan exceeds the General Assembly's delegated authority under the Elections Clause for three reasons: (1) the Elections Clause did not empower State legislatures to disfavor the interests of supporters of a particular candidate or party in drawing congressional districts; (2) the 2016 Plan's pro-Republican bias violates other constitutional provisions, including the First Amendment, the Equal Protection Clause, and Article I, section 2 ; and (3) the 2016 Plan represents an impermissible effort to "dictate electoral outcomes" and "disfavor a class of candidates." Thornton , 514 U.S. at 833–34, 115 S.Ct. 1842.

1.

The Elections Clause was the product of a vigorous debate at the Constitutional Convention among the delegates regarding whether, and to what extent, to lodge authority over the regulation of congressional elections in Congress. On the one hand, those who feared the power of the new federal government did not want to give Congress the ability to override state election regulations. For example, the Anti–Federalist propagandist Federal Farmer argued that placing authority to promulgate election regulations in the national government would allow Congress to draft election laws that favored particular representatives or viewpoints. See Greene , supra at 1033, 119 S.Ct. 1936. " ‘[T]he general legislature may ... evidently so regulate elections as to secure the choice of any particular description of men.’ " Id. (quoting Letter from the Federal Farmer (Oct. 10, 1787), reprinted in Origins of the House of Representatives: A Documentary Record 52, 53 (Bruce A. Ragsdale ed., 1990)). Other Anti–Federalists, including Patrick Henry, expressed similar concerns about Congress manipulating election regulations to favor a particular group of candidates or their supporters. Id. at 1036.

On the other hand, supporters of congressional control over state election regulations—the position that ultimately prevailed—emphasized the risk that States would refuse to hold elections, and thereby strip the federal government of power, or, more relevant to the case at hand, enact election regulations—including districting plans—that would favor particular factions. For example, James Madison argued that "[w]henever the State Legislatures had a favorite measure to carry, they would take care so to mould their regulations as to favor the candidates they wished to succeed." Debates at 424. Likewise, a delegate at the Massachusetts ratifying convention "warned that ‘when faction and party spirit run high,’ a legislature might take actions like ‘making an unequal and partial division of the states into districts for the election of representatives.’ " Ariz. State Leg. , 135 S.Ct. at 2672 (quoting Theophilus Parsons in Debate in Massachusetts Ratifying Convention (16–17, 21 Jan. 1788), in 2 The Founders' Constitution 256 (P. Kurland & R. Lerner eds. 1987)).

Accordingly, although the Framers disagreed as to whether, and to what extent, the Elections Clause should empower Congress to displace state election regulations, the Framers agreed that, regardless of whether Congress retained such authority, the Elections Clause should not empower legislative bodies—be they state or federal—to impose election regulations that would favor or disfavor a particular group of candidates or voters. See Thornton , 514 U.S. at 833 n.47, 115 S.Ct. 1842 (" ‘The constitution expressly provides that the choice shall be by the people, which cuts off both from the general and state Legislatures the power of so regulating the mode of election, as to deprive the people of a fair choice.’ " (quoting "The Republican," Connecticut Courant (Hartford, Jan. 7, 1788), 1 Bailyn 710, 713)). To that end, the Supreme Court has expressly recognized that the Elections Clause was "intended to act as a safeguard against manipulation of electoral rules by politicians and factions in the States to entrench themselves or place their interests over those of the electorate." Ariz. State Leg. , 135 S.Ct. at 2672.

As explained above in drawing the 2016 Plan, the General Assembly "manipulat[ed]," id. , district lines in order to subordinate the interests of non-Republican candidates and their supporters and entrench Republican candidates in power. The 2016 Plan, therefore, does not amount to a "neutral," Gralike , 531 U.S. at 527, 121 S.Ct. 1029, or "fair" procedural regulation, Thornton , 514 U.S. at 853, 115 S.Ct. 1842, but rather an effort to achieve an impermissible substantive goal—providing the Republican party with a "Partisan Advantage," Ex. 1007. Accordingly, the 2016 Plan exceeds the General Assembly's delegated authority under the Elections Clause.

2.

We further conclude that the 2016 Plan's favoring of Republican candidates and their supporters and disfavoring of non-Republican candidates and their supporters violates the Elections Clause by "infring[ing] upon basic constitutional protections." Kusper , 414 U.S. at 56–57, 94 S.Ct. 303. As explained above, the 2016 Plan violates the Equal Protection Clause because it reflects a successful, and unjustified, effort by the General Assembly to subordinate the interests of non-Republican voters and entrench Republican Representatives in power. See supra Part III. Additionally, as an intentional, and successful, effort to burden the speech and associational rights of supporters of non-Republican candidates, the 2016 Plan violates the First Amendment. See supra Part IV.

The 2016 Plan also violates Article I, section 2's grant of authority to "the People" to elect their Representatives. The Framers decision to vest the power to elect Representatives in "the People" was—and is—significant. This feature differentiated the House of Representatives from every other federal government body at the time of the Framing. It is "the only textual reference to ‘the People’ in the body of the original Constitution and the only express, original textual right of the People to direct, unmediated political participation in choosing officials in the national government." Richard H. Pildes, The Constitution and Political Competition , 30 Nova L. REV. 253, 267 (2006). For example, at the time, Senators were elected by the state legislatures. U.S. Const. art. I, § 3 repealed by U.S. Const. amend. XVII. The President was and still is elected through an intermediate body—the Electoral College. U.S. Const. art. II, § 1. Only the House of Representatives was directly accountable to the People.

Article I, section 2 was a product of the so-called Great Compromise, which resolved a bitter dispute between delegates regarding whether representation in the national legislature would be determined by population, with representatives directly elected by the people, or would be awarded equally among the States, with representatives elected by state legislatures. See Wesberry , 376 U.S. at 12–13, 84 S.Ct. 526. Under the Great Compromise, the Senate represented the interests of the States, each State was awarded equal representation in that body, and Senators were elected by state legislatures. Id. at 13, 84 S.Ct. 526. By contrast, "[t]he House of Represen[t]atives, the Convention agreed, was to represent the people as individuals, and on the basis of complete equality for each voter." Id. at 14, 84 S.Ct. 526. The House of Representatives, therefore, provided "a direct link between the National Government and the people of the United States." Thornton, 514 U.S. at 803, 115 S.Ct. 1842.

The delegates at the Constitutional Convention decided to have the House of Representatives elected directly by the People for two major reasons. First, the Framers viewed popular election of at least one branch of government as an essential feature of a government founded on democratic principles. James Madison explained, for example, that "[a]s it is essential to liberty that the government in general should have a common interest with the people, so it is particularly essential that the [House of Representatives] should have an immediate dependence on, and an intimate sympathy with, the people." The Federalist No. 52, at 295 (James Madison). Other delegates at the constitutional convention also emphasized the critical importance of direct popular election of representatives in any republican form of government. Debates at 39 (reporting that George Mason "argued strongly for an election of the larger branch by the people, stating that "[i]t was to be the grand depository of the democratic principle of the government"); id. at 167, 124 S.Ct. 619 (reporting that James Wilson stated he "considered the election of the first branch by the people not only as the corner Stone, but as the foundation of the fabric: and that the difference between a mediate and immediate election was immense"). Put simply, Article I, Section 2 gives effect to the Framers' belief that " ‘[t]he true principle of a republic is, that the people should choose whom they please to govern them.’ " Powell , 395 U.S. at 540–41, 89 S.Ct. 1944 (quoting Alexander Hamilton in 2 Debates on the Federal Constitution 257 (J. Elliot ed. 1876)).

The Framers also saw popular election of Representatives as an important check on the States' power. See, e.g. , Debates at 40 (reporting that James Wilson stated that: "no government could long subsist without the confidence of the people. In a republican Government, this confidence was peculiarly essential.... All interference between the general and local government should be obviated as much as possible."); id. at 167 (reporting that Alexander Hamilton did not want state legislatures to elect both chambers of Congress, because "State influence ... could not be too watchfully guarded against"); id. (reporting that Rufus King worried that "the Legislatures would constantly choose men subservient to their own views as contrasted to the general interest; and that they might even devise modes of election that would be subversive of the end in view"). In sum, "the Framers, in perhaps their most important contribution, conceived of a Federal Government directly responsible to the people, possessed of direct power over the people, and chosen directly, not by States, but by the people ." Thornton , 514 U.S. at 821, 115 S.Ct. 1842 (emphasis added).

The 2016 Plan's invidious partisanship runs contrary to the Constitution's vesting of the power to elect Representatives in "the People." U.S. Const. art. I, § 2. To begin, partisan gerrymanders, like the 2016 Plan, violate "the core principle of republican government" preserved in Article I, Section 2 —"namely, that the voters should choose their representatives, not the other way around." Ariz. State Leg. , 135 S.Ct. at 2677 (internal quotation marks omitted). And by favoring supporters of Republican candidates over supporters of non-Republican candidates, the 2016 Plan "defeat[s] the principle solemnly embodied in the Great Compromise" because it reflects a successful effort by the General Assembly to "draw the lines of congressional districts in such a way as to give some voters a greater voice in choosing a Congressman than others." Wesberry , 376 U.S. at 14, 84 S.Ct. 526.

Additionally, rather than having " ‘an habitual recollection of their dependence on the people,’ " as the Framers intended, Ariz. State Leg. , 135 S.Ct. at 2677 (quoting The Federalist No. 57, at 320 (James Madison)), partisan gerrymanders render Representatives responsive to the controlling faction of the State legislature that drew their districts, Vieth , 541 U.S. at 331–32, 124 S.Ct. 1769 (Stevens, J. dissenting) ("The problem [with partisan gerrymandering], simply put, is that the will of the cartographers rather than the will of the people will govern."). By rendering Representatives responsive to the state legislatures who drew their districts rather than the People, the 2016 Plan also upsets the careful balance struck by the Framers in the Great Compromise by "interpos[ing]" the General Assembly between North Carolinians and their Representatives in Congress. See Gralike , 531 U.S. at 527, 121 S.Ct. 1029 (Kennedy, J., concurring) ("A State is not permitted to interpose itself between the people and their National Government as it seeks to do here."). "Neither the design of the Constitution nor sound principles of representative government are consistent with the right or power of a State to interfere with the direct line of accountability between the National Legislature and the people who elect it." Id. at 528, 121 S.Ct. 1029.

3.

Finally, the 2016 Plan amounts to a successful effort by the General Assembly to "disfavor a class of candidates" and "dictate electoral outcomes." Thornton , 514 U.S. at 833–34, 115 S.Ct. 1842. In Cook v. Gralike , 531 U.S. 510, 121 S.Ct. 1029, 149 L.Ed.2d 44 (2001), the Court considered an amendment to a state constitution that "instruct[ed]" each member of the state's congressional delegation "to use all of his or her delegated powers to pass the Congressional Term Limits Amendment," id. at 514, 121 S.Ct. 1029 (majority op.). To advance that goal, the amendment further provided that "the statement ‘DISREGARDED VOTERS' INSTRUCTION ON TERM LIMITS’ be printed on all primary and general [election] ballots adjacent to the name of a[n incumbent] Senator or Representative who fails to take any of one of eight [enumerated] legislative acts in support of the proposed amendment." Id. And the amendment further required that primary and general election ballots expressly indicate if a nonincumbent candidate " ‘DECLINED TO PLEDGE TO SUPPORT TERM LIMITS.’ " Id. at 514–15, 121 S.Ct. 1029.

The Court concluded that the amendment exceeded the state's authority under the Elections Clause. Id. at 524–27, 121 S.Ct. 1029. In reaching this conclusion, the Court reaffirmed that because the Elections Clause constitutes the States' sole source of "authority over congressional elections," "the States may regulate the incidents of such elections ... only within the exclusive delegation of power under the Elections Clause." Id. at 522–23, 121 S.Ct. 1029 (emphasis added). The Court concluded the amendment exceeded that delegated authority for two principal reasons. First, the amendment was "plainly designed to favor candidates who are willing to support the particular form of term limits amendment set forth in its text and to disfavor those who either oppose term limits entirely or would prefer a different proposal." Id. at 523–25, 121 S.Ct. 1029. Second, the placement of the "pejorative" or "negative" labels next to candidates who opposed the term limits amendment on the ballot "handicap[ped] [such] candidates ‘at the most crucial stage in the election process—the instant before the vote is cast.’ " Id. at 524–25, 121 S.Ct. 1029 (quoting Anderson v. Martin , 375 U.S. 399, 402, 84 S.Ct. 454, 11 L.Ed.2d 430 (1964) ). By "handicap[ping]" candidates who opposed the term limits amendment, the state constitutional amendment represented an "attempt[t] to ‘dictate election outcomes,’ " which "simply is not authorized by the Elections Clause." Id. at 524, 526, 121 S.Ct. 1029 (quoting Thornton , 514 U.S. at 833–34, 115 S.Ct. 1842 ); see also Chamness v. Bowen , 722 F.3d 1110, 1121 (9th Cir. 2013) (explaining that, under Gralike , the Elections Clause prohibits state election regulations that "dictate political outcomes or invidiously discriminate against a class of candidates"); Brown , 668 F.3d at 1284 (explaining that the Elections Clause, as interpreted in Thornton and Gralike , does not authorize a state legislature to enact an election regulation "meant to prevent or severely cripple the election of particular candidates").

Like the state constitutional amendment at issue in Gralike , the Partisan Advantage criterion—and the record evidence regarding Representative Lewis, Senator Rucho, and Dr. Hofeller's implementation of that criterion in drawing the 2016 Plan, see supra Parts I.B.2, III.A.2—establishes that the 2016 Plan was intended to disfavor non-Republican candidates and supporters of such candidates and favor Republican candidates and their supporters. And like the constitutional amendment in Gralike , the General Assembly's express intent to draw a redistricting plan that would elect a congressional delegation composed of 10 Republicans and 3 Democrats—coupled with the fact that the 2016 election under the 2016 Plan yielded a congressional delegation with the intended composition—demonstrates that the 2016 Plan amounted to a successful "attempt[ ] to ‘dictate election outcomes.’ " Gralike , 531 U.S. at 526, 121 S.Ct. 1029 (quoting Thornton , 514 U.S. at 833–34, 115 S.Ct. 1842 ). Accordingly, the 2016 Plan's demonstrated partisan favoritism "simply is not authorized by the Elections Clause." Id.

VI.

Having concluded that the 2016 Plan violates the Equal Protection Clause, the First Amendment, and Article I of the Constitution, we now must determine the appropriate remedy. Absent unusual circumstances, "such as where an impending election is imminent and a State's election machinery is already in progress," courts should take "appropriate action to insure that no further elections are conducted under the invalid plan." Reynolds , 377 U.S. at 585, 84 S.Ct. 1362. As the 2018 general election remains many months away and the 2018 election cycle has not yet formally begun, we find no such circumstances exist. Accordingly, we enjoin Defendants from conducting any further elections using the 2016 Plan.

As to the drawing of a remedial plan, as a general rule, once a federal court concludes that a state districting plan violates the Constitution or federal law, it should "afford a reasonable opportunity for the legislature to meet constitutional requirements by adopting a substitute measure rather than for the federal court to devise ... its own plan." Wise v. Lipscomb , 437 U.S. 535, 540, 98 S.Ct. 2493, 57 L.Ed.2d 411 (1978). This case presents an exceptional circumstance, however: the General Assembly enacted the 2016 Plan after another panel of this Court invalidated the 2011 Plan as a racial gerrymander. Harris , 159 F.Supp.3d at 627. When a court finds a remedial districting plan also violates the Constitution, courts generally do not afford a legislature a second "bite-at-the-apple" to enact a constitutionally compliant plan. See Chapman v. Meier , 420 U.S. 1, 27, 95 S.Ct. 751, 42 L.Ed.2d 766 (1975) (holding that if a state fails to enact "a constitutionally acceptable" remedial districting plan, "the responsibility falls on the District Court"); Reynolds , 377 U.S. at 586, 84 S.Ct. 1362 (holding that a district court "acted in a most proper and commendable manner" by imposing its own remedial districting plan, after the district court concluded that the remedial plan adopted by state legislature failed to remedy constitutional violation).

We nevertheless conclude that the General Assembly is entitled to a second opportunity to draw a constitutional congressional districting plan. Although the Supreme Court had recognized that partisan gerrymanders "are incompatible with democratic principles," Ariz. State Leg. , 135 S.Ct. at 2658 (internal quotation marks and alterations omitted), and that partisan gerrymandering claims were justiciable under the Equal Protection Clause, Bandemer , 478 U.S. at 123, 106 S.Ct. 2797 (plurality op.), at the time the General Assembly drew the 2016 Plan, the Court had not established a legal standard for adjudicating partisan gerrymandering claims. In such circumstances, we decline to pre-empt the legislature's primary role in redistricting and reapportionment.

In providing the General Assembly with such an opportunity, we also recognize that North Carolina voters have been deprived of a constitutional congressional districting plan for the better part of the decade. The Constitution entitles those voters a remedy that "so far as possible eliminate[s] the discriminatory effects of the past as well as bar[s] like discrimination in the future." Louisiana v. United States , 380 U.S. 145, 154, 85 S.Ct. 817, 13 L.Ed.2d 709 (1965). Enacting new congressional districts as quickly as possible will, at least partially, remedy the discriminatory effects of the 2016 Plan by giving elected legislators an incentive to "focus on representing the interests of the constituents in their new districts—rather than the districts we held constituted unconstitutional [partisan] gerrymanders." Covington v. North Carolina , ––– F. Supp. 3d. ––––, 2017 WL 4162335, No. 15-cv-399 (M.D.N.C. Sept. 19, 2017). That consideration—coupled with the fast approaching deadline for candidates to file to compete in the 2018 election and our obligation to review any remedial plan to ensure that it remedies the constitutional violation and is not otherwise "legally unacceptable," McGhee v. Granville Cty., N.C. , 860 F.2d 110, 115 (4th Cir. 1988) —counsels in favor of allowing the General Assembly a shorter window to remedy the constitutional violation. Accordingly, the General Assembly will have until 5 p.m. on January 24, 2018, to enact a remedial districting plan. That deadline will allow the General Assembly two weeks to draw a remedial plan, the amount of time state law affords the General Assembly to draw remedial districting plans. N.C. Gen. Stat. § 120–2.4(a).

No later than 5 p.m. on January 29, 2018, the State shall file with the Court any enacted proposed remedial plan, along with:

1. transcripts of all committee hearings and floor debates related to the proposed remedial plan;

2. the "stat pack" for the proposed remedial plan;

3. a description of the process the General Assembly, and any constituent committees or members thereof, followed in drawing and enacting the proposed remedial plan, including, without limitation, the identity of all participants involved in the process;

4. any alternative plans considered by the General Assembly, any constituent committee responsible for drawing the remedial plan, or the leadership of the General Assembly or any such committee; and

5. the criteria the General Assembly, any constituent committee responsible for drawing the remedial plan, and the leadership of the General Assembly or any such committee applied in drawing the proposed remedial plan, including, without limitation, any criteria related to partisanship, the use of political data, or the protection of incumbents.

No later than 5 p.m. on February 5, 2018, Plaintiffs and other interested parties may file objections to any enacted proposed remedial plan and submit an alternative remedial plan. No later than 5 p.m. on February 12, 2018, Defendants may file responses to any such objections.

Given the fast-approaching candidate-filing deadline, we further find it appropriate to take steps to ensure the timely availability of an alternative remedial plan for use in the event the General Assembly does not enact a remedial plan or enacts a plan that fails to remedy the constitutional violation or is otherwise legally unacceptable. To that end, we intend to appoint in short order a Special Master pursuant to Federal Rule of Civil Procedure 53 to assist the Court in drawing an alternative remedial plan. Rodriguez v. Pataki , 207 F.Supp.2d 123, 125 (S.D.N.Y. 2002) ("[T]he ‘eleventh hour’ is upon us, if indeed it has not already passed. It is therefore necessary for this Court to prepare for the possibility that this Court will be required to adopt an appropriate redistricting plan."). Accordingly, we direct the parties to confer and file no later than January 16, 2018, a list of three qualified and mutually acceptable candidates to serve as Special Master. In the event the parties fail to agree as to a list of candidates, the Court may identify a special master without input from the parties.

SO ORDERED

Circuit Judge Wynn wrote the majority opinion in which Senior District Judge Britt concurred. District Judge Osteen, Jr., wrote a separate opinion concurring in part and dissenting in part.

OSTEEN, JR., District Judge, concurring in part and dissenting in part:

I concur with the well-reasoned opinion of the majority that Plaintiffs have met their burden of proving a prima facie partisan gerrymandering claim under the Equal Protection Clause of the Fourteenth Amendment. Plaintiffs have shown both an intent to subordinate the interests of non-Republican voters and entrench Republican candidates in power, all with the effect of controlling electoral outcomes to continue a 10–3 Republican control of Congressional seats. However, in keeping with the standard established by the Supreme Court for racial gerrymandering claims, I would require Plaintiffs to prove that partisanship was the predominant factor motivating the General Assembly's decision to draw the 2016 Plan as it did. Because I agree that Plaintiffs met their burden, and also agree that Defendants have not justified the effects of the 2016 Plan, I concur with the majority's conclusion that the Plan violates the Equal Protection Clause.

I also join the majority's conclusion that Plaintiffs have shown that the 2016 Plan violates Article I, Sections 2 and 4 of the United States Constitution by proving that the drawers of the Plan intended to dictate and preordain election outcomes. However, assuming that partisan gerrymandering claims are justiciable under the First Amendment, I am unconvinced that Plaintiffs have proven an injury to their First Amendment rights, and dissent from the majority's conclusion that the 2016 Plan violates the First Amendment.

Before turning to my analysis of the claims in this case, I write to express my concerns with these claims generally. If writing on a blank slate, I would rely solely upon Article I to grant relief to Plaintiffs. In my opinion, Article I, Sections 2 and 4 set a clear limit on unconstitutional political gerrymandering: when the legislature, through its redistricting plan, controls the outcome of the election, whether as a result of partisan consideration or another factor, the plan is unconstitutional. Beyond a prohibition on dictating the outcome of an election, which protects the right of the people granted in Article I, Section 2, I would not find the Constitution provides additional protection to the voting strength of members of a political party or group so as to prohibit partisan considerations in redistricting.

Subject to regulation by Congress, see 2 U.S.C. § 2, the Constitution delegates redistricting power for federal elections to the States and their legislatures. Legislative action is a political process, and issues addressed by those legislative bodies affecting constitutional questions—redistricting, Second Amendment, First Amendment, abortion, and the like—are inherently political in nature. As the plurality in Davis v. Bandemer observed, "[i]t would be idle ... to contend that any political consideration taken into account in fashioning a reapportionment plan is sufficient to invalidate it.... Politics and political considerations are inseparable from districting and apportionment." Davis v. Bandemer , 478 U.S. 109, 128, 106 S.Ct. 2797, 92 L.Ed.2d 85 (1986) (plurality op.) (citations omitted) (quoting Gaffney v. Cummings , 412 U.S. 735, 752, 93 S.Ct. 2321, 37 L.Ed.2d 298 (1973) ). Although Bandemer has been abrogated to some degree, see Common Cause v. Rucho , 240 F.Supp.3d 376, 387 (M.D.N.C. 2017) (per curiam), this observation remains true today.

Previously in this case, we held that the partisan gerrymandering claims presented here were justiciable under the Equal Protection Clause, see id. at 389, and I agree with that conclusion for the reasons described in the memorandum opinion. While the majority opinion presents additional, logical, and compelling analysis of applicable cases and precedent, I continue to have fundamental concerns over the application of Equal Protection and First Amendment principles to partisan gerrymandering.

The Elections Clause limits partisan considerations in redistricting by prohibiting action that dictates election results. Analysis of partisan gerrymandering claims under the Equal Protection Clause and the First Amendment attempt to set a limit on partisan advantage somewhere between a politically neutral redistricting and the Elections Clause prohibition of dictating election results, a limit I am not convinced is required by those constitutional provisions. If there should be additional limits on partisan consideration beyond those of Article I, the Constitution provides the people of this State with the additional power to "seek relief from Congress, which can make or alter the regulations prescribed by the legislature. And the Constitution gives them another means of change. They can follow the lead of the reformers who won passage of the Seventeenth Amendment." Ariz. State Legislature v. Ariz. Indep. Redistricting Comm'n , ––– U.S. ––––, 135 S.Ct. 2652, 2692, 192 L.Ed.2d 704 (2015) (Roberts, C.J., dissenting). Partisan advantage is a part of all legislative action. Remedies exist for legislative overreach, even in reapportionment, so long as the voters, and not the legislature, are controlling the outcomes of elections.

Nevertheless, I agree that, absent a contrary ruling from the Supreme Court, partisan gerrymandering claims are justiciable under the Equal Protection Clause, and so the court is obliged to articulate a standard for adjudication. Having found that Plaintiffs have met that standard in this case, I join the majority opinion in finding an Equal Protection violation.

I. Equal Protection

Both the majority opinion and the Supreme Court have spoken of evaluating Equal Protection claims in political gerrymandering cases in terms of a "discriminatory intent." As Justice Kennedy noted in Vieth , "[a] determination that a gerrymander violates the law must rest ... on a conclusion that [political] classifications, though generally permissible, were applied in an invidious manner or in a way unrelated to any legitimate legislative objective." Vieth v. Jubelirer , 541 U.S. 267, 307, 124 S.Ct. 1769, 158 L.Ed.2d 546 (Kennedy, J., concurring in the judgment); see also Gaffney , 412 U.S. at 751, 93 S.Ct. 2321 ("A districting plan may create ... districts [that are] invidiously discriminatory because they are employed ‘to minimize or cancel out the voting strength of racial or political elements of the voting population.’ " (quoting Fortson v. Dorsey , 379 U.S. 433, 439, 85 S.Ct. 498, 13 L.Ed.2d 401 (1965) )). Determining, then, whether a legislative redistricting body's partisan considerations amount to an invidiously discriminatory intent is critical to determining whether the plan it produces violates the Equal Protection Clause.

Under the intent prong, League Plaintiffs claim that the Republican-led state legislature enacted the 2016 Plan "with the aim of disadvantaging one party's (and favoring the other party's) voters and candidates." (League of Women Voters Pls.' Post–Trial Br. 9, Nov. 6, 2017, ECF No. 113.) The aim of the Plan, as alleged by Common Cause Plaintiffs, was to "achieve a partisan goal." (Common Cause Pls.' Post–Trial Br. ("Common Cause Br.") 7, Nov. 6, 2017, ECF No. 116.) Stating the obvious, the alleged discriminatory intent was an effort to gain partisan advantage; that is, the Republican majority sought to draw districts to elect more Republican representatives, which in turn would disadvantage Democratic voters. In my opinion, discriminatory intent and partisan advantage are two sides of the same coin, that is, the political process. As a general proposition, the political process is one in which one side seeks to gain political advantage over the opposing party or issue. It is difficult to conceive of any political issue, including redistricting, where opposing sides would not possess some intent to gain partisan advantage and thereby hold some form of discriminatory intent as that term is used in this case.

The Court has recognized many times in redistricting and apportionment cases that some degree of partisanship and political consideration is constitutionally permissible in a redistricting process undertaken by partisan actors. See, e.g. , Hunt v. Cromartie , 526 U.S. 541, 551, 119 S.Ct. 1545, 143 L.Ed.2d 731 (1999) ("Our prior decisions have made clear that a jurisdiction may engage in constitutional political gerrymandering, even if it so happens that the most loyal Democrats happen to be black Democrats and even if the State were conscious of that fact."); Miller v. Johnson , 515 U.S. 900, 914, 115 S.Ct. 2475, 132 L.Ed.2d 762 (1995) ("[R]edistricting in most cases will implicate a political calculus in which various interests compete for recognition ...."); Gaffney , 412 U.S. at 753, 93 S.Ct. 2321 ("Politics and political considerations are inseparable from districting and apportionment."); see also Cooper v. Harris , ––– U.S. ––––, 137 S.Ct. 1455, 1488, 197 L.Ed.2d 837 (2017) (Alito, J., concurring in the judgment in part and dissenting in part) (recognizing the constitutionality of at least some amount of political gerrymandering); Whitford v. Gill , 218 F.Supp.3d 837, 934–35 (W.D. Wis. 2016) (Griesbach, J., dissenting) (collecting cases), appeal docketed , 137 S.Ct. 2289 (2017). And Congress, though it could presumably act to limit partisan gerrymandering under its Article I, Section 4 authority, has chosen only to require single-member districts. See 2 U.S.C. § 2c.

I do not find, therefore, that the Constitution forbids a political body from taking into account partisan considerations, and indeed partisan advantage, when producing a redistricting plan. A plaintiff satisfies the majority's intent requirement "by introducing evidence establishing that the state redistricting body acted with an intent to ‘subordinate adherents of one political party and entrench a rival party in power.’ " (Maj. Op. at 86 (quoting Ariz. State Legislature , 135 S.Ct. at 2658).) Because I find that this standard of intent sweeps more broadly than required by the Equal Protection Clause, I am unable to agree with the intent prong of the majority's three-prong test.

Rather, I would require Plaintiffs to prove that this intent predominated over other considerations in the redistricting process. Although "[l]egislation is frequently multipurposed," Vill. of Arlington Heights v. Metro. Hous. Dev. Corp. , 429 U.S. 252, 265 n.11, 97 S.Ct. 555, 50 L.Ed.2d 450 (1977), the Supreme Court has expressly held that courts are equipped, in the particular context of redistricting legislation, to discern whether one consideration predominated over others, see Miller , 515 U.S. at 915–16, 115 S.Ct. 2475 (holding, in the context of racial gerrymandering cases, that plaintiffs must prove that "race was the predominant factor motivating the legislature's decision to place a significant number of voters within or without a particular district"). I see no reason to believe that courts are not just as well equipped to determine whether partisan considerations predominated. In my view, this level of intent equals the "invidious" application of political classifications required for Plaintiffs to prove the first prong of their prima facie case.

Under this standard, Plaintiffs must show that the redistricting body "subordinated traditional [neutral] districting principles, including but not limited to compactness, contiguity, and respect for political subdivisions or communities defined by actual shared interests," to political considerations. See id. at 916, 115 S.Ct. 2475. The majority's opinion details at length the facts and circumstances surrounding the enactment of 2016 Plan, which do not need repeating here. (See, e.g. , Maj. Op. at Part I.C, III.A.2–3, III.C.) Suffice it to say that there is ample evidence in the record to find that Plaintiffs have met this burden. In particular, Dr. Hofeller's and legislative Defendants' statements and the lack of transparency and public participation in the map drawing process invite this conclusion.

For example, Dr. Hofeller admitted that he sought "to minimize the number of districts in which Democrats would have an opportunity to elect a Democratic candidate" under the 2011 Plan. (See Dep. of Thomas B. Hofeller ("Hofeller Dep.") 127:19–22, Jan. 24, 2017, ECF Nos. 101–34, 110–1.) Past voting behavior was used to draw the maps. (See id. at 132:22–134:13, 159:20–160:12.) After the 2011 Plan was enjoined due to two unconstitutionally racially gerrymandered districts, Dr. Hofeller was instructed to draw new maps that would maintain the existing partisan makeup of the congressional delegation achieved under the racially gerrymandered plan: ten Republicans and three Democrats. (See id. at 175:19–23, 188:5–190:2.) Dr. Hofeller began to work on the 2016 Plan on his personal computer after receiving verbal instructions from Representative Lewis, without comment or participation from the public and without written instructions. (See id. at 71:6–73:15, 129:8–130:9; Dep. of Rep. David Lewis ("Lewis Dep.") 44:12–24, 46:1–4, 73:19–22, 105:11–106:1, Jan. 26, 2017, ECF Nos. 101–33, 108–3, 110–3, 110–4.) He continued work on the Plan at his home, with Representative Lewis and Senator Rucho, operating under oral directions. (Lewis Dep. 48:19–49:7, 60:1–13; Dep. of Sen. Robert Rucho ("Rucho Dep.") 169:21–170:17, Jan. 25, 2017, ECF Nos. 101–32, 110–5.) Dr. Hofeller then presented the maps to Representative Lewis in "near-final" versions that Representative Lewis intended to submit to the legislature for adoption. (Lewis. Dep. 77:7–20.) In the subsequent committee meeting discussing the 2016 Plan, Representative Lewis noted that "the goal is to elect 10 Republicans and 3 Democrats." (Ex. 1005, at 62:18–19.) Comments from the one public hearing held and written comments solicited and received via the committee's website were not shared with Dr. Hofeller. (Ex. 1004; Rucho Dep. 55:4–56:13.) The official criteria for the 2016 Plan, which included neutral principles as well as partisan criteria, were not adopted until after the maps were mostly completed. (Ex. 1007; Hofeller Dep. 177:9–21.)

In determining whether partisan consideration predominated, intent may be proven by both direct and circumstantial evidence. Miller , 515 U.S. at 916, 115 S.Ct. 2475. In this case, the evidence that partisan consideration predominated is substantial, including the limited access to mapping information provided to all legislators and a stated intent of maintaining the current partisan advantage of 10–3. In short, while Dr. Hofeller, under the direction of Senator Rucho and Representative Lewis, considered neutral principles to some extent, (see, e.g. , Hofeller Dep. 174:10–25), the evidence shows that these considerations were secondary to Defendants' primary goal of entrenching Republican candidates in power by dictating the outcome of elections held under the 2016 Plan.

I concur with the sections of the majority opinion addressing the effects and justification prongs of its three-part test, and join the majority in holding that the 2016 Plan violates Plaintiffs' rights under the Equal Protection Clause.

II. First Amendment

Assuming that partisan gerrymandering claims are justiciable under the First Amendment, I find that the majority's adopted test would in effect foreclose all partisan considerations in the redistricting process—a result I am unable to conclude that the First Amendment requires—and would allow redress for an injury that Plaintiffs have not proven rises to a constitutional level. Therefore, I respectfully dissent.

No one disputes that the First Amendment protects political expression and association. See, e.g. , Citizens United v. Fed. Election Comm'n , 558 U.S. 310, 339–40, 130 S.Ct. 876, 175 L.Ed.2d 753 (2010) ; Buckley v. Valeo , 424 U.S. 1, 15, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976) (per curiam). But as another court aptly noted in rejecting plaintiffs' claim that the inability to elect a preferred candidate burdened their political expression, "[p]laintiffs are every bit as free under the new [redistricting] plan to run for office, express their political views, endorse and campaign for their favorite candidates, vote, or otherwise influence the political process through their expression." Radogno v. Ill. State Bd. of Elections , No. 1:11-CV-04884, 2011 WL 5025251, at *7 (N.D. Ill. Oct. 21, 2011) (second alteration in original) (quoting Kidd v. Cox , No. 1:06-CV-0997-BBM, 2006 WL 1341302, at *17 (N.D. Ga. May 16, 2006) ). As the Radogno court explained, "[i]t may very well be that Plaintiffs' ability to successfully elect their preferred candidate is burdened by the redistricting plan, but that has nothing to do with their First Amendment rights." Id. (citing Washington v. Finlay , 664 F.2d 913, 927–28 (4th Cir. 1981) ).

Plaintiffs are likewise free under the 2016 Plan to "field candidates for office, participate in campaigns, vote for their preferred candidate, or otherwise associate with others for the advancement of common political beliefs." Id. (quoting Kidd , 2006 WL 1341302, at *17 ). The fact that some Plaintiffs testified about difficulties involving voter outreach, fundraising, and candidate recruitment, (see, e.g. , Dep. of Elizabeth Evans 16:4–9, Apr. 7, 2017, ECF No. 101–7; Dep. of John J. Quinn, III 39:1–3, Apr. 10, 2017, ECF No. 101–22), fails to persuade me that the 2016 Plan objectively chilled the speech and associational rights of the citizens of North Carolina so as to prove a First Amendment violation.

Justice Kennedy, suggesting in Vieth that the First Amendment may be an applicable vehicle for addressing partisan gerrymandering claims, proposed that such an analysis should ask "whether political classifications were used to burden a group's representational rights." Vieth , 541 U.S. at 314–15, 124 S.Ct. 1769 (Kennedy, J., concurring in the judgment). The Vieth plurality rejected this proposal because "a First Amendment claim, if it were sustained, would render unlawful all consideration of political affiliation in districting, just as it renders unlawful all consideration of political affiliation in hiring for non-policy-level government jobs." Id. at 294, 124 S.Ct. 1769 (plurality op.). Common Cause Plaintiffs essentially agree, arguing that strict scrutiny is triggered once a plaintiff shows that a redistricting body intended for a plan to discriminate against a certain set of voters. (Common Cause Br. 5–8.) The majority adopts an intermediate scrutiny standard requiring the showing of a concrete burden to political speech or associational rights. (Maj. Op. at 162–63.) However, in practice, I find the result to be indistinguishable, for partisan consideration in a political process is an attempt to create some sort of political advantage for the supporters of a candidate or party. This advantage necessarily comes at the expense of or burden to the other.

As explained above, Congress has declined to expressly limit partisan gerrymandering by statute, see 2 U.S.C. § 2c, and the Court's cases accepting or tolerating some amount of partisan consideration are many, see, e.g. , Cromartie , 526 U.S. at 551, 119 S.Ct. 1545 ; Miller , 515 U.S. at 914, 115 S.Ct. 2475 ; Gaffney , 412 U.S. at 753, 93 S.Ct. 2321 ; see also Harris , ––– U.S. ––––, 137 S.Ct. at 1488 (Alito, J., concurring in the judgment in part and dissenting in part); Whitford , 218 F.Supp.3d at 934–35 (Griesbach, J., dissenting) (collecting cases). It might be desirable for a host of policy reasons to remove partisan considerations from the redistricting process. But I am unable to conclude that the First Amendment requires it, or that Plaintiffs here have proven violations of their speech or associational rights under the First Amendment.

III. Article I, Sections 2 and 4

I agree with the majority's conclusion that the 2016 Plan amounts to a successful attempt to dictate election outcomes. I join in the majority's opinion as to Article I, Sections 2 and 4 to the extent consistent with the discussion above. I specifically join in the analysis and holding in Part V.B.3. I differ slightly from the majority in that I do not find that the Elections Clause completely prohibits State legislatures from disfavoring a particular party. See Brown v. Sec'y of State of Fla. , 668 F.3d 1271, 1284 & n.10 (11th Cir. 2012) (rejecting the prohibition of all regulations influencing election outcomes and instead reading the cases as prohibiting States from attempting "to prevent or severely cripple the election of particular candidates").

"[T]he people should choose whom they please to govern them." U.S. Term Limits, Inc. v. Thornton , 514 U.S. 779, 783, 115 S.Ct. 1842, 131 L.Ed.2d 881 (1995) (quoting Powell v. McCormack , 395 U.S. 486, 547, 89 S.Ct. 1944, 23 L.Ed.2d 491 (1969) ). In this case, the legislature, not the people, dictated the outcome when the districts were drawn, and Defendants have presented no specific facts to support a finding that the election results were due to anything other than the maps being drawn to reach a specific result. General suggestions of other factors possibly contributing to the election results such as fundraising disparities, voter turnout, the quality of the candidates, and unforeseen candidate circumstances, (see, e.g. , Legislative Defs.' Post–Trial Br. 10–11, Nov. 6, 2017, ECF No. 115; Leg. Defs.' Proposed Findings of Fact and Conclusions of Law 67, Nov. 6, 2017, ECF No. 114), are insufficient to establish that something other than partisan consideration dictated the election results across the State.

IV. Remedy

I agree that the General Assembly is entitled to a second opportunity to draw a constitutional congressional districting plan. As noted in both the majority opinion and this opinion, the adjudication of partisan gerrymandering claims against a redistricting plan is a developing area of law, and the General Assembly should have the opportunity to remedy its plan under the standards set forth in the majority opinion. While there is merit to the majority's procedure in identifying a Special Master at this juncture, I would not appoint a Special Master prior to the General Assembly's unsatisfactory enactment of a remedial plan. I am not convinced any duties exist at this time for which an appointment is appropriate, nor do I believe there is an exceptional condition or any post-trial matter yet presented which cannot be effectively and timely addressed by the court. See Fed. R. Civ. P. 53.