Raleigh Wake Citizens Assoc v. Wake County Board of Elections

PUBLISHED

UNITED STATES COURT OF APPEALS
FOR THE FOURTH CIRCUIT

No. 16-1270

RALEIGH WAKE CITIZENS ASSOCIATION; JANNET B. BARNES;
BEVERLEY S. CLARK; WILLIAM B. CLIFFORD; BRIAN FITZSIMMONS;
GREG FLYNN; DUSTIN MATTHEW INGALLS; AMY T. LEE; ERWIN
PORTMAN; SUSAN PORTMAN; JANE ROGERS; BARBARA VANDENBERGH;
JOHN G. VANDENBERGH; AMYGAYLE L. WOMBLE; PERRY WOODS,

Plaintiffs - Appellants,

v.

WAKE COUNTY BOARD OF ELECTIONS,

Defendant - Appellee,

and

CHAD BAREFOOT, in his official capacity as Senator and
primary sponsor of SB 181; PHILLIP E. BERGER, in his
official capacity as President Pro Tempore of the North
Carolina Senate; TIM MOORE, in his official capacity as
Speaker of the North Carolina House of Representatives,

Defendants.

No. 16-1271

CALLA WRIGHT; WILLIE J. BETHEL; AMY T. LEE; AMYGAYLE L.
WOMBLE; JOHN G. VANDENBERGH; BARBARA VANDENBERGH; AJAMU G.
DILLAHUNT; ELAINE E. DILLAHUNT; LUCINDA H. MACKETHAN;
WILLIAM B. CLIFFORD; ANN LONG CAMPBELL; GREG FLYNN; BEVERLEY
S. CLARK; CONCERNED CITIZENS FOR AFRICAN-AMERICAN CHILDREN,
d/b/a Coalition of Concerned Citizens for African-American
Children; RALEIGH WAKE CITIZENS ASSOCIATION,

Plaintiffs - Appellants,
 v.

WAKE COUNTY BOARD OF ELECTIONS,

Defendant - Appellee,

and

STATE OF NORTH CAROLINA,

Defendant.

Appeals from the United States District Court for the Eastern
District of North Carolina, at Raleigh. James C. Dever III,
Chief District Judge. (5:15-cv-00156-D; 5:13-cv-00607-D)

Argued: May 9, 2016 Decided: July 1, 2016

Before MOTZ, GREGORY, and WYNN, Circuit Judges.

Reversed and remanded in part and affirmed in part by published
opinion. Judge Wynn wrote the majority opinion, in which Judge
Gregory joined. Judge Motz wrote a dissenting opinion.

ARGUED: Anita Sue Earls, Allison Jean Riggs, SOUTHERN COALITION
FOR SOCIAL JUSTICE, Durham, North Carolina, for Appellants.
Charles Foster Marshall, III, BROOKS, PIERCE, MCLENDON, HUMPHREY
& LEONARD, L.L.P., Raleigh, North Carolina, for Appellee. ON
BRIEF: George E. Eppsteiner, SOUTHERN COALITION FOR SOCIAL
JUSTICE, Durham, North Carolina, for Appellants. Matthew B.
Tynan, Jessica Thaller-Moran, BROOKS, PIERCE, MCLENDON, HUMPHREY
& LEONARD, L.L.P., Raleigh, North Carolina, for Appellee.

2
WYNN, Circuit Judge:

The right to vote is “fundamental,” and once that right “is

granted to the electorate, lines may not be drawn which are

inconsistent with the Equal Protection Clause of the Fourteenth

Amendment.” Bush v. Gore, 531 U.S. 98, 104-05 (2000) (quotation

marks and citation omitted). “It must be remembered that” the

right to vote “can be denied by a debasement or dilution of the

weight of a citizen’s vote just as effectively as by wholly

prohibiting the free exercise.” Id. (quoting Reynolds v. Sims,

377 U.S. 533, 555 (1964)).

In these consolidated cases, Plaintiffs, registered voters

and civic organizations in Wake County, North Carolina, claim

that under the two (identically drawn) redistricting laws they

challenge, some Wake County School Board and Wake County Board

of County Commissioners districts have been over-populated,

while others have been under-populated. Plaintiffs further

assert that these discrepancies result in some votes counting

more while others count less, and that the discrepancies stem

from illegitimate redistricting factors. As explained below, we

agree, hold that Plaintiffs have proven their state and federal

one person, one vote claims, and therefore reverse.

Plaintiffs also claim that one discrete district was the

product of racial gerrymandering. We hold that the district

3
court did not clearly err in rejecting that claim and thus

affirm.

I.

In the years leading up to 2013, the Wake County School

Board (“School Board”) consisted of nine members elected from

single-member districts. Those districts were subject to change

every ten years following the decennial census.

In 2010, the census showed that Wake County’s population

had grown by 43.51% over the preceding decade, causing the then-

existing districting plan to have a maximum population deviation

of 47.89%. 1 The School Board, at that time dominated by

registered Republicans, 2 redrew its districts in light of the

2010 census.

1“[C]ourts usually analyze[] apportionment plan[s] in terms
of the maximum population deviation among the districts.
Generally, to calculate maximum deviation, the court first
constructs a hypothetical ideal district by dividing the total
population of the political unit (e.g., state or county) by the
total number of representatives who serve that population.
Then, the court determines how much the actual population of
each district varies from the population of the ideal district.
This deviation is expressed as a percentage of the ideal
population. Maximum deviation is the sum of the absolute value
of the deviation of the district with the smallest population
and that of the district with the largest population.” Daly v.
Hunt, 93 F.3d 1212, 1215 n.2 (4th Cir. 1996).
2While the School Board is nominally non-partisan, its
members are routinely registered and affiliated with the
Democratic and Republican Parties, and uncontroverted trial
testimony showed a high level of partisanship in “what’s
supposed to be a nonpartisan election.” J.A. 234; see also,
(Continued)
4
 That effort led to a redistricting plan with geographically

compact districts having a maximum population deviation of 1.75%

and no district deviating from the ideal district population by

even 1%. The first election under the new districting, in Fall

2011, resulted in a School Board with a Democratic majority.

In 2013, the Republican-controlled North Carolina General

Assembly (“General Assembly”), over the objection of a majority

of the School Board and every Democratic and African-American

legislator in the General Assembly, passed a local bill, Session

Law 2013–110, making numerous changes to the School Board’s

method of selection. Among other things, Session Law 2013-110

changed the School Board’s make-up from nine single-member

districts to seven single-member districts and set less

geographically compact boundaries for this new set of districts.

The maximum population deviation among the new single-member

districts swelled to over 7%.

Additionally, Session Law 2013-110 created two “super

districts” that overlaid the single-member districts. J.A. 160.

One super district formed a donut of outer, more rural areas of

the county, while the other formed a donut hole in the inner,

e.g., J.A. 254 (noting that such local races have “become more
partisan-based” due to “block candidates,” the “political party
machine,” and “money”).

5
urban area. The maximum population deviation between the super

districts exceeded even that of the single-member districts—just

shy of 10%. Session Law 2013-110 moved elections to even-

numbered years, and limited the School Board’s ability to make

changes to its method of election until 2021.

In August 2013, thirteen individuals and two civic

organizations filed suit in the United States District Court for

the Eastern District of North Carolina, challenging the

constitutionality of the districts that Session Law 2013-110

established. The complaint alleged that the plan unevenly

weighted the votes of citizens in the county for impermissible

reasons, thereby violating the one-person, one-vote guarantees

of the federal and state constitutions. In March 2014, the

district court dismissed Plaintiffs’ suit for failure to state a

claim. Wright v. North Carolina, 975 F. Supp. 2d 539 (E.D.N.C.

2014). Plaintiffs appealed.

In April 2015, while Plaintiffs’ appeal was pending before

this Court, the General Assembly enacted Session Law 2015-4,

making the electoral system for the Wake County Board of County

Commissioners (“Board of County Commissioners”) identical to the

system it had created for the School Board with Session Law

6
2013-110. 3 With Session Law 2015-4, too, the General Assembly

forced a local bill on Wake County despite opposition from the

majority of the Board of County Commissioners, polled Wake

County voters, nearly every Democratic state legislator, and

every African-American legislator in the General Assembly.

Fourteen individuals and a civic organization filed suit shortly

thereafter, challenging the Board of County Commissioners’

redistricting plan as violating the one person, one vote

guarantees of the state and federal constitutions.

In Plaintiffs’ appeal from the district court’s March 2014

dismissal, this Court, in May 2015, held that “Plaintiffs’

allegations in support of their claim that [Session Law 2013-

110] violates the one person, one vote principle suffice to

survive a motion to dismiss for failure to state a claim.”

Wright v. North Carolina, 787 F.3d 256, 269 (4th Cir. 2015). We

therefore reinstated Plaintiffs’ complaint against the Wake

County Board of Elections.

On remand, the district court consolidated the suits

challenging Session Law 2013-110 and Session Law 2015-4 and

expedited discovery. Discovery was further limited by the state

3
Previously, members of the Board of County Commissioners
were elected at-large, subject to the requirement that one
member had to be elected from each of the county’s seven
residency districts.

7
legislators’ refusing Plaintiffs’ discovery requests, claiming

legislative privilege. 4 In December 2015, the district court

held a bench trial, in which Plaintiffs presented numerous

witnesses, including legislators, citizens, and experts, as well

as copious documentary evidence, with 481 exhibits including:

expert reports and supporting data; school assignment maps;

campaign finance reports; results data from various elections;

excerpts of legislative transcripts; and public polling results.

By contrast, Defendant, the Board of Elections that administers

elections with no stake in the “political interests of the

General Assembly,” Trial Tr. vol. I, 13:24-25, presented none of

its own. Defendant simply cross-examined Plaintiffs’ witnesses

and made legal argument.

Nevertheless, the district court ruled for Defendant.

Raleigh Wake Citizens Ass’n v. Wake Cty. Bd. of Elections, No.

5:13-CV-607-D, 2016 WL 1060378 (E.D.N.C. Feb. 26, 2016). The

district court discredited every single one of Plaintiffs’

witnesses, for example as “anecdotal,” id. at *28-29, and

“unhelpful,” id. at *32. It went on to hold, among other

things, that “in order to prove a prima facie case in a one

4
Pursuant to an agreement between Plaintiffs and particular
legislators, certain external communications between the
legislators and third parties—but no internal communications
amongst the legislators—were produced.

8
person one vote challenge, plaintiffs must at least negate the

most common legitimate reasons that could explain the

legislature’s action.” Id. at *22 (quotation marks and

citations omitted). The district court held that Plaintiffs

failed to meet this and the other requisite burdens. Plaintiffs

appealed.

II.

On appeal, “‘[w]e review judgments resulting from a bench

trial under a mixed standard of review: factual findings may be

reversed only if clearly erroneous, while conclusions of law are

examined de novo.’” Nat’l Fed’n of the Blind v. Lamone, 813

F.3d 494, 502 (4th Cir. 2016) (quoting Plasterers’ Local Union

No. 96 Pension Plan v. Pepper, 663 F.3d 210, 215 (4th Cir.

2011)). Findings will be deemed clearly erroneous if, for

example, “even though there is some evidence to support the

finding, the reviewing court, on review of the record, is left

with a definite and firm conviction that a mistake has been

made,” or if findings were made using “incorrect legal

standards.” Consol. Coal Co. v. Local 1643, United Mine Workers

of Am., 48 F.3d 125, 128 (4th Cir. 1995) (quotation marks and

citation omitted). “Of course, if the trial court bases its

findings upon a mistaken impression of applicable legal

principles, the reviewing court is not bound by the clearly

9
erroneous standard.” Inwood Labs., Inc. v. Ives Labs., Inc.,

456 U.S. 844, 855 n.15 (1982).

III.

With their primary argument on appeal, Plaintiffs contend

that the district court applied the wrong legal standard for

adjudicating their one person, one vote claim. For the reasons

explained below, we agree.

A.

The right to vote is “fundamental,” and once that right “is

granted to the electorate, lines may not be drawn which are

inconsistent with the Equal Protection Clause of the Fourteenth

Amendment.” Bush, 531 U.S. at 104-05 (quotation marks and

citation omitted). Indeed, allowing, through unequal

apportionment amongst districts, a vote to be “worth more in one

district than in another would . . . run counter to our

fundamental ideas of democratic government.” Reynolds, 377 U.S.

at 563 (quotation marks and citation omitted). This requirement

that all citizens’ votes be weighted equally, known as the one

person, one vote principle, applies not just to the federal

government but also to state and local governments—including

school boards and county governing bodies. Avery v. Midland

Cty., 390 U.S. 474, 480 (1968).

Courts have recognized that “[m]athematical exactness or

precision is hardly a workable constitutional requirement” and

10
thus do not require “identical numbers” in state and local

government districts. Reynolds, 377 U.S. at 577. Nevertheless,

governments must “make an honest and good faith effort” to

construct districts as close to equal population “as is

practicable.” Id. To assess what is “practicable,” the Supreme

Court has allowed some population deviation for “legitimate

considerations” such as compactness and contiguity, the

integrity of political subdivisions, and balance among political

parties. Harris v. Ariz. Indep. Redistricting Comm’n, 136 S.

Ct. 1301, 1306 (2016).

Generally, a districting plan “with a maximum population

deviation under 10% will not, by itself, support an equal

protection claim.” Wright, 787 F.3d at 264 (quotation marks

omitted and emphasis added). Rather, plaintiffs in such cases

“must show that it is more probable than not that a deviation of

less than 10% reflects the predominance of illegitimate

reapportionment factors rather than” legitimate considerations

such as compactness or the integrity of political subdivisions.

Harris, 136 S. Ct. at 1307.

In Harris, the Supreme Court’s most recent, and arguably

most lucid, pronouncement as to plaintiffs’ burdens in one

person, one vote cases below the 10% deviation threshold, the

Court unanimously noted that the plaintiffs there had claimed

that the plan’s deviations from “absolute equality of population

11
reflect . . . political efforts to help the Democratic party.”

Id. Crucially, however, the plaintiffs “failed to prove this

claim.” Id. Instead, “the record b[ore] out” that the

deviations “predominantly reflected . . . efforts to achieve

compliance with the federal Voting Rights Act, not to secure

political advantage for one party.” Id. In other words, the

plaintiffs in Harris foundered not because their one person, one

vote challenge failed as a matter of law, but because they did

not muster the evidence needed to show it to be

“more probable than not that [the] deviation of less than 10%

reflect[ed] the predominance of illegitimate reapportionment

factors.” Id.

By contrast, in Larios v. Cox, the plaintiffs succeeded in

proving their one person, one vote claims. 300 F. Supp. 2d 1320

(N.D. Ga.) (three-judge panel), aff’d, 542 U.S. 947 (2004)

(mem.). In Larios, a federal court struck down a Georgia

redistricting plan that disproportionately favored Democrats by

under-populating districts in the urban Atlanta region and the

rural south—both Democratic strongholds—while over-populating

suburban districts with Republican-leaning voters. The

redistricting created a maximum population deviation of 9.98%

and disproportionately protected Democratic incumbents. Id. at

1328–31. The Supreme Court (with only Justice Scalia

12
dissenting) affirmed the district court’s rejection of the

redistricting. Larios, 542 U.S. 947.

As the Supreme Court has explained, in Larios, “those

attacking the plan had shown that it was more probable than not

that the use of illegitimate factors significantly explained

deviations from numerical equality among districts.” Harris,

136 S. Ct. at 1310. The Supreme Court noted the “many examples

showing that population deviation as well as the shape of many

districts did not result from any attempt to create districts

that were compact or contiguous, or to keep counties whole, or

to preserve the cores of prior districts.” Id. (quotation marks

and citation omitted). The Supreme Court contrasted the Larios

plaintiffs’ successful showing with that of the failed

plaintiffs in Harris, stating “[i]t is appellants’ inability to

show that the present plan’s deviations and boundary shapes

result from the predominance of similarly illegitimate factors

that makes [Larios] inapposite here.” Id.

Looking at Larios and Harris, we conclude that, to succeed

on the merits, plaintiffs in one person, one vote cases with

population deviations below 10% must show by a preponderance of

the evidence that improper considerations predominate in

explaining the deviations. This is just such a case, and that

legal standard therefore applies.

B.

13
 1.

The law in this area is challenging. In the earlier

appeal of this matter, we sought to clarify some points to ease

the burden on the district court. Nonetheless, there were

numerous instances in which the law we set out in Wright was not

adhered to. For example, in evaluating Plaintiffs’ one person,

one vote claim, the district court did not properly characterize

what Plaintiffs must show to succeed. The district court

stated, for example, that “in order to prove a prima facie case

in a one person one vote challenge, plaintiffs must at least

negate the most common legitimate reasons that could explain the

legislature’s action.” Raleigh Wake Citizens Ass’n, 2016 WL

1060378, at *22 (quotation marks and citation omitted). The

district court indicated that “any conceivable legislative

purpose is sufficient” to support the redistricting plan and

that those “attacking the rationality” thereof “have the burden

to [negate] every conceivable basis which might support it.”

Id. at 27 (alteration in original) (quotation marks and citation

omitted).

Contrary to the district court’s characterization, what

Plaintiffs must actually show to succeed with their one person,

one vote claims is that it is “more probable than not that a

deviation of less than 10% reflects the predominance of

illegitimate reapportionment factors.” Harris, 136 S. Ct. at

14
1307. This specific, deviation-focused inquiry differs markedly

from the district court’s rational-basis review of whether a

rational state policy could explain the redistricting generally.

2.

Further, in Wright, we emphasized the importance of the

Supreme Court’s affirmance of Larios for this case. Thus, we

made it clear that Larios was more than a mere summary

affirmance holding little sway. Raleigh Wake Citizens Ass’n,

2016 WL 1060378, at *18. Instead, with Wright, we set forth

precedent binding on the district courts of this Circuit making

clear that Larios constitutes persuasive authority generally, as

well as analogous authority in this concrete case. Wright, 787

F.3d at 267. The district court’s heavy emphasis on Justice

Scalia’s Larios dissent—an opinion with no precedential value—is

thus squarely at odds with Wright. See, e.g., Raleigh Wake

Citizens Ass’n, 2016 WL 1060378, at *18-19 (“According to

Justice Scalia, ‘politics as usual’ is a ‘traditional

redistricting criterion,’ and ‘a constitutional one,’” and

“‘[f]erreting out political motives in minute population

deviations seems to me more likely to encourage politically

motivated litigation than to vindicate political rights.’”).

Moreover, the district court misapplied the core principles

of Larios. The district court stated, for example, that, in

contrast to Larios, Plaintiffs here did not prove “that the

15
General Assembly disregarded all districting principles in

creating the 2013 Wake County School Board Plan, or that the

2013 Wake County School Board Plan is not rationally related to

a permissible, rational state policy of improving School Board

representation.” Raleigh Wake Citizens Ass’n, 2016 WL 1060378,

at *36. The district court thus concluded that “unlike Larios,

plaintiffs have failed to prove that the 2013 Wake County School

Board Plan resulted from a desire to favor suburban and rural

voters over urban voters.” Id.

Crucially, neither the three-judge district court in

Larios, nor the Supreme Court in affirming and later discussing

Larios, ever suggested that plaintiffs in such cases need to

show that “all districting principles” were “disregarded.” Id.

Further, neither court focused on the challenged redistricting

plans as a whole. Instead, the focus, in Larios as well as,

Harris, was whether “deviation[s] of less than 10% reflect[ed]

the predominance of illegitimate reapportionment factors.”

Harris, 136 S. Ct. at 1307 (emphasis added); Larios, 300 F.

Supp. 2d at 1338 (holding that “population deviations . . . not

supported by . . . legitimate interests . . . cannot withstand

constitutional scrutiny” (emphasis added)). In Larios, the

state legislature’s attempt to privilege rural and urban

Democrats at the expense of suburban Republicans explained the

deviations in population, not the redistricting plan generally,

16
did not constitute a legitimate apportionment factor, and was

prohibited. Larios, 300 F. Supp. 2d at 1338.

3.

Additionally, in evaluating the evidence Plaintiffs

proffered to support their one person, one vote claims, the

district court improperly discounted every single one of

Plaintiffs’ fifteen trial witnesses. For example, it

discredited all the testifying legislators because of their

“strong legislative opposition to the 2013 Wake County School

Board Plan. [The pertinent] testimony at trial fits within the

line of precedent giving no weight to statements made by

opponents of legislation.” Raleigh Wake Citizens Ass’n, 2016 WL

1060378, at *29.

The only analogous case in the purported “line of

precedent,” Veasey v. Abbott, 796 F.3d 487 (5th Cir. 2015), has

been vacated and is thus no longer good law, 815 F.3d 958 (5th

Cir. 2016) (granting rehearing en banc and vacating the panel

opinion). The other cases the district court cited—cases

dealing with statutory interpretation—stand for the unremarkable

and inapposite proposition that courts usually do not “accord

much weight to the statements of a bill’s opponents [when

interpreting the words of the bill]. The fears and doubts of

the opposition are no authoritative guide to the construction of

legislation.” Shell Oil Co. v. Iowa Dep’t of Revenue, 488 U.S.

17
19, 29 (1988) (quotation marks, citation, and brackets omitted)

(holding that one passing reference to preemption in a speech by

an opponent of a law cannot properly guide the court’s

interpretation of that law); see also Schwegmann Bros. v.

Calvert Distillers Corp., 341 U.S. 384, 394 (1951) (noting that

“doubts of the opposition” do not guide “the construction of

legislation”); NLRB v. Fruit & Vegetable Packers & Warehousemen,

Local 760, 377 U.S. 58, 66 (1964) (same).

This is not a case about what a particular word in a

statute means. Rather, at the heart of this case is whether

illegitimate factors predominated the General Assembly’s

supplemental redistricting of Wake County such that illegitimate

factors explain the population deviations in the redistricting

plan. While we recognize that a trial judge generally may

consider “bias or prejudice” when “assessing witness

credibility,” United States v. Muse, 83 F.3d 672, 676-77 (4th

Cir. 1996), the district court discredited categorically the

legislators’ testimony, even regarding objective facts. Yet the

district court has cited, and we see, no controlling precedent

suggesting that their testimony should simply have been

discounted wholesale and “giv[en] no weight.” Raleigh Wake

Citizens Ass’n, 2016 WL 1060378, at *29.

Similarly, the district court completely rejected as

“materially flawed and unhelpful,” id. at *32, the analysis of

18
Plaintiffs’ expert Dr. Jowei Chen, a political science professor

from the University of Michigan. Upon closer inspection,

however, it is the district court’s own analysis of Dr. Chen’s

analysis that is materially flawed.

Dr. Chen analyzed whether the population deviations in the

seven single-member district plans and the two super districts

plans were motivated by a partisan purpose using computer

simulation programming techniques that allow him to generate

randomly a large number of alternative redistricting plans

created subject to traditional redistricting criteria. The four

traditional redistricting criteria Dr. Chen used were:

population equality; keeping municipalities intact; keeping

precincts whole; and geographic compactness. Dr. Chen’s

computer simulations are based on the logic that if a computer

randomly draws five hundred redistricting plans following

traditional redistricting criteria, and the actual enacted plans

fall completely outside the range of what the computer has

drawn, one can conclude that the traditional criteria do not

explain that enacted plan.

The computer simulations led Dr. Chen to just that

conclusion: that the “enacted districting plans create a

partisan distribution of seats falling completely outside the

range of outcomes that are possible under a non-partisan

districting process that creates equally populated districts

19
while maximizing compactness and preserving precinct and

municipal boundaries.” J.A. 768. Dr. Chen thus concluded “with

extremely high statistical certainty, beyond any sort of doubt

here” that “the only way to draw districts as extreme in

partisanship as the legislature’s B and A districts is to use

population deviations” that are high. J.A. 463. In other

words, Dr. Chen testified that he could conclude with certainty

from his simulations that the deviations at issue here are the

result of using partisanship in apportioning the districts.

In critiquing Dr. Chen’s analysis, the district court

seized on the fact that certain criteria accounted for in the

computer simulations—such as setting maximum population

deviation at 2% or less or “completely . . . ignor[ing]

partisanship,” Raleigh Wake Citizens Ass’n, 2016 WL 1060378, at

*30, are required by neither state nor federal law. This

critique misses the point: The point is not that the simulated

plans are legally required, but rather that they help

demonstrate what might explain the population deviations in the

enacted plan.

The district court went on to “find[] that Dr. Chen’s

simulations simply show that ‘better’ . . . redistricting plans

were possible, but ‘better’ plans do not equate to the

unconstitutionality of the 2013 Wake County School Board Plan.”

Id. With that finding, the district court again missed the

20
point: The import of Dr. Chen’s simulations was not to produce

better plans, but rather to hold several legitimate

apportionment considerations constant so that Dr. Chen could

assess whether the population deviations in the challenged plans

could have been the product of something other than partisan

bias. He concluded “with extremely high statistical certainty,

beyond any sort of doubt here” that they could not have. J.A.

463. The district court clearly and reversibly erred in

rejecting Dr. Chen’s expert testimony. Easley v. Cromartie, 532

U.S. 234 (2001) (reversing a three-judge district court panel in

a racial gerrymandering case in which the district court clearly

erred in rejecting expert evidence).

4.

We could go on detailing the errors in the opinion below.

Suffice it to say that the legal analysis of what Plaintiffs

needed to show as well as the evaluation of the evidence

Plaintiffs proffered to make that showing are fundamentally

flawed.

C.

1.

When, as here, the district court applies the wrong

standards, we tend to remand to allow “the trier of fact to re-

examine the record” using the correct standards. Kelley v. S.

Pac. Co., 419 U.S. 318, 332 (1974). However, when “the record

21
permits only one resolution of the factual issue,” Pullman–

Standard v. Swint, 456 U.S. 273, 292 (1982), remand is

unnecessary, and we may rule based on the record before us.

Thus, for example, in the recent Class v. Towson University

opinion, this Court, based on the record before it, straight-out

reversed the district court, which had applied the incorrect

legal standard following a bench trial. 806 F.3d 236 (4th Cir.

2015). And in Cromartie, 532 U.S. 234, the Supreme Court

outright reversed a three-judge district court panel in a racial

gerrymandering case because, among other things, the district

court had clearly erred in rejecting pertinent expert evidence.

Likewise, here, we deem remand unnecessary. At trial, in

addition to copious documentary evidence, Plaintiffs presented

fifteen live witnesses—two experts, four legislators, four

county elected officials, and five plaintiffs and lay witnesses. 5

These witnesses and documents presented abundant support for

Plaintiffs’ one person, one vote claims within the nine-hour

total that the district court allowed Plaintiffs for presenting

their case.

Defendant, by contrast, offered not even one witness.

Instead, Defendant expressly disclaimed any stake in

5 The district court did not deem any of Plaintiffs’
witnesses to be untrustworthy. Raleigh Wake Citizens Ass’n,
2016 WL 1060378.

22
“representing the political interests of the General Assembly,”

Trial Tr. vol. I, 13:24-25, and essentially passed on defending

the General Assembly’s redistricting. Even the legislative

proponents of the challenged redistricting laws refused to

defend their actions, instead claiming legislative immunity.

The resulting record, discussed in more detail below,

permits only one resolution of Plaintiffs’ one person, one vote

claims: Plaintiffs have proven that it is more probable than

not that the population deviations at issue here reflect the

predominance of a illegitimate reapportionment factor, Harris,

136 S. Ct. at 13—namely an “intentional effort” to create “a

significant . . . partisan advantage,” Larios, 542 U.S. at 947-

49 (Stevens, J., concurring). In other words, Plaintiffs have

successfully made their case.

2.

First putting the challenged plans in context, the evidence

at trial showed that that Wake County’s population generally,

and the overall population deviation amongst the School Board

districts in particular, swelled significantly by the time of

the 2010 decennial census. Accordingly, the School Board

redrew its election maps. The resulting 2011 redistricting plan

reduced maximum population deviation down to 1.75%, with no

single district deviation reaching even 1% from the ideal. The

districts were “vetted” by county residents and the members of

23
the School Board, and were considered relatively compact,

contiguous, and respectful of communities of interest. J.A.

210. The Board of County Commissioners also redrew its

residency districts after the 2010 decennial census.

Despite the fact the 2011 redistricting had been shepherded

by a “Republican School Board” and that a “Republican lawyer”

had drafted the districts, J.A. 420, the 2011 elections, the

first administered under the new plan, resulted in a “shift[]

from the Republicans to the Democrats.” J.A. 200. The

Republican-controlled General Assembly then intervened with the

redistricting plans that are the subject of this action.

Uncontroverted testimony and evidence adduced at trial

showed that the legislative process relating to Session Law

2013-110 was truncated by, for example, not having “community

hearings and participation of the affected parties,” J.A. 211,

and failing to incorporate “any of the ideas that people . . .

proffered,” id., without even “discussing it amongst the [Wake

County] delegation first,” a “stark departure” from common

practice, J.A. 419. As School Board Member Bill Fletcher, a

registered Republican, put it, “nothing was discussed. There

was no opportunity to provide input, to have a debate or

discussion about different election strategies, it was simply

drafted in a bill and presented and passed with little

opportunity for rational thought.” J.A. 263.

24
 3.

Moving on to the showing Plaintiffs needed to make on their

one person, one vote claims, uncontroverted evidence at trial

showed that the deviations resulting from the latter-day

redistricting more likely than not reflected the predominance of

illegitimate reapportionment factors.

Plaintiffs proffered uncontroverted evidence of an

illegitimate factor predominating in the skewed, unequal

redistricting: an attempt to guaranty Republican victory through

the intentional packing of Democratic districts. Various

witnesses testified that “the true motivation[]” for the

redistricting was to “ensure Republican control . . . at the

expense of Democrats.” J.A. 364. The “real reason” behind the

redistricting was “[t]o ensure a Republican majority . . .

despite the vote totals,” J.A. 405, a “kind of punitive and

retributive effort to punish the Democrats for winning,” J.A.

392.

Plaintiffs’ expert Anthony Fairfax analyzed the challenged

redistricting plans and reported, among other things, that

“[t]here was a marked pattern of overpopulation in Democratic-

performing districts, and underpopulation in Republican-

performing districts.” J.A. 805. And as Mr. Fairfax noted in

his testimony, “by overpopulating you obviously minimize the

25
Democratic performance in other districts, other surrounding

districts.” J.A. 305. 6

Plaintiffs’ second expert, Dr. Chen, conducted an analysis

showing that “[t]he General Assembly’s enacted districting plans

create a partisan distribution of seats falling completely

outside the range of outcomes that are possible under a non-

partisan districting process that creates equally populated

districts while maximizing compactness and preserving precinct

and municipal boundaries.” J.A. 768. In other words, as Dr.

Chen testified at trial, “the only way to achieve a districting

plan that allowed for such an extreme partisan Republican

control over four districts out of seven, the only way to create

such an extreme partisan plan was to deviate from population

equality to a great extent.” J.A. 466-67.

6
The district court discounted Mr. Fairfax’s testimony just
as it did every single one of Plaintiffs’ other witnesses. And
in the case of Mr. Fairfax, as with the others, the bases for
that discounting fall apart upon careful inspection. For
example, the district court faulted Mr. Fairfax for using
election results data, asserting that he “failed to analyze
voter registration data in Wake County.” Raleigh Wake Citizens
Ass’n, 2016 WL 1060378, at *34. Yet in focusing on election
results instead of registration data, Mr. Fairfax followed
precisely what the Supreme Court has instructed those analyzing
redistricting plans to do. See, e.g., Cromartie, 532 U.S. at
239 (noting its instruction that courts should look to “data
showing how voters actually behave, not data showing only how
those voters are registered”).

26
 The legislators who hatched the redistricting plans claimed

legislative immunity. Absent from the record, therefore, is any

trial testimony confirming (or denying) a partisan motive behind

the redistricting and its deviations. 7 The record does, however,

contain several e-mails including third parties, the only

category of e-mails Plaintiffs managed to obtain, that indeed

suggest a partisan motive behind the redistricting and its

deviations. For example, the Wake County Republican Party Chair

exchanged several e-mails with, and apparently met with, key

legislators involved in the redistricting, with a focus on “how

we would take 5 of the 9 seats.” J.A. 1114.

We do not doubt that some amount of partisan politics is

par for the course in redistricting generally. For example, in

Gaffney v. Cummings, a case on which the district court relied

here, the Supreme Court upheld a redistricting plan drawn based

on partisan considerations. 412 U.S. 735 (1973). But the facts

in and consequences of Gaffney differ markedly and tellingly

7 Both the district court and Defendant make much ado of the
admissions the legislators made in Larios, noting the direct
evidence that legislators purposefully skewed district
deviations along urban, suburban, and rural divides to achieve
partisan goals. See, e.g., Appellee’s Br. at 41; Raleigh Wake
Citizens Ass’n, 2016 WL 1060378, at *18. Both Defendant and the
district court contrast those facts with this case, with its
lack of such direct evidence. But here, the lack of direct
evidence may have its roots in the legislators’ avoiding
discovery through claims of legislative immunity. Moreover,
direct evidence is simply not required.

27
from those here. In Gaffney, a state legislature had drafted a

redistricting plan following a decennial census; in doing so, it

followed a “policy of ‘political fairness.’” Id. at 738. The

plan, which exhibited less than 2% overall deviation in the

state senate and less than 8% overall deviation in the state

house, sought “proportional representation of the two major

political parties. . . . [T]he Board took into account the party

voting results in the preceding three statewide elections, and,

on that basis, created what was thought to be a proportionate

number of Republican and Democratic legislative seats.” Id.

In this case, by contrast, rather than seeking proportional

representation of the two main political parties, the evidence

shows that the challenged plans under-populated Republican-

leaning districts and over-populated Democratic-leaning

districts in order to gerrymander Republican victories. 8 In

other words, the challenged redistricting here subverts

8 The district court played up the fact that District 5 and
District 6 constitute exceptions to the rule that Democratic-
leaning districts were over-populated and Republican-leaning
districts were under-populated. Raleigh Wake Citizens Ass’n,
2016 WL 1060378, at *35. According to the district court,
“[t]his evidence belies a systematic under-population of
districts to harm incumbents . . . who are registered Democrats
who support ‘progressive’ education policies.” Id. What the
evidence actually belies is the tenuousness of the district
court’s analysis—because both District 5 and District 6 exhibit
only negligible deviations from ideal population—both less than
0.2%.

28
political fairness and proportional representation and

sublimates partisan gamesmanship. Gaffney simply cannot

reasonably be read as supporting that; if anything, it does the

opposite. Indeed, the Supreme Court suggested that partisanship

is not a legitimate reason to weight some votes more than

others, and the Gaffney Court itself underscored that

redistricting so as to “minimize” the “political strength” of a

party or group would be constitutionally “vulnerable.” Id. at

754.

Further, the Supreme Court rejected just such partisan

deviation games in Larios, 542 U.S. 947, indicating that “if a

plan contains any population deviations, a court may decide that

the deviations are caused by impermissible partisanship and

strike the plan down . . . for failure to comply with one

person, one vote.” Samuel Issacharoff & Pamela S. Karlan, Where

to Draw the Line?: Judicial Review of Political Gerrymanders,

153 U. Pa. L. Rev. 541, 567-68 (2004); see Larios, 300 F. Supp.

2d at 1338 (holding that pertinent “population deviations” were

“not the result of an effort to further any legitimate” policy

but were instead “systematically and intentionally created” to

“protect Democratic incumbents” and holding that that did not

“withstand[] Equal Protection scrutiny”).

We recognize that the Supreme Court has not yet clarified

when exactly partisan considerations cross the line from

29
legitimate to unlawful. See, e.g., Harris v. McCrory, No. 1:13-

CV-949, 2016 WL 3129213, at *2 (M.D.N.C. June 2, 2016) (citing

Larios, 542 U.S. 947, for the proposition that redistricting

plans may be challenged “when partisan considerations go ‘too

far,’” while citing Vieth v. Jubelirer, 541 U.S. 267 (2004), for

the lack of “judicially discernible and manageable standards for

adjudicating political gerrymandering claims”). Yet it is

important to bear in mind that only a plurality (i.e., not a

controlling majority) of the Supreme Court has suggested that

partisanship-based redistricting claims should be considered

nonjusticiable. 9 Id. And shortly after Vieth, a nearly

unanimous Supreme Court, including three Justices from the Vieth

plurality, affirmed Larios, in which the lower court struck down

a redistricting plan with population deviations under 10% as a

9 The district court incorrectly suggested that “[i]n Vieth,
the Supreme Court rejected as nonjusticiable a political
gerrymandering claim.” Raleigh Wake Citizens Ass’n 2016 WL
1060378, at *19, n.11. On the contrary, as we noted in Wright,
“a majority of the (Vieth) Supreme Court refused to deem
political gerrymandering claims to be per se nonjusticiable.
And the Court has since recognized as much.” 787 F.3d at 269
(citing League of United Latin Am. Citizens v. Perry, 548 U.S.
399, 414 (2006) (“A plurality of the Court in Vieth would have
held [political gerrymandering] challenges to be nonjusticiable
political questions, but a majority declined to do so.”)).

30
blatant and unlawful attempt at partisan favoritism. Larios,

542 U.S. 947. 10

4.

Not only did the uncontested record evidence demonstrate

that illegitimate reapportionment factors predominated,

resulting in an overall deviation of barely under 10%; the

evidence also exposed the stated reasons for the redistricting

as pretextual. For example, one stated goal of the School

Board’s redistricting was to increase the alignment between

citizen’s voting districts and their assigned schools.

Uncontroverted testimony at trial indicated that the

redistricting resulted in the opposite, “mak[ing] alignment

worse.” J.A. 235. Indeed, “[j]ust a perfect downtown example

is Daniels Middle School and Broughton High School[, which] are

in the same feeder pattern, they were in the same district under

the 2011 maps . . . but they were in different districts under

the [new] map” challenged here. J.A. 424. Further, even if

10
Stated different, “barely two months [after Vieth], three
of those Justices were part of an eight-Justice majority that
affirmed the judgment in Larios, a case in which the lower court
struck down a plan [with] relatively minuscule population
deviations . . . because they reflected ‘blatantly partisan and
discriminatory’ attempts to protect Democratic incumbents while
undermining Republican-held seats. As Sister Maria says in The
Sound of Music, ‘When the Lord closes a door, somewhere He opens
a window.’” Issacharoff & Karlan, 153 U. Pa. L. Rev. at 542.

31
increased alignment were indeed a goal, it need not necessarily

have resulted in population deviations amongst the districts.

A second stated rationale for the redistricting debunked at

trial: reducing campaign costs. As trial testimony

demonstrated, “the proponents of this legislation said that they

were concerned about the cost of campaigning and that these

districts would make it cheaper to run. . . . That is either

inaccurate or deceptive, because Wake County is a media market

and if you’re going to run in any of these widespread districts

here or if you’re going to run all in the entire county you are

still going to be advertising in the Raleigh/Wake media market,

[and] it’s still expensive.” J.A. 395-96. Further, moving

down-ballot races like those for School Board members to even

years that include congressional and presidential races is

“going to dramatically increase the costs of running” in those

elections, J.A. 420, even simply for candidates “to have any

visibility in a Presidential election cycle.” J.A. 258. And,

again, nothing about this stated rationale, cost reduction,

explains the population deviation amongst the districts.

Another stated goal of the redistricting legislation—

increasing voter turnout—also has nothing to do with re-drawing

districts, much less re-drawing them unequally. The district

court noted that “Plaintiffs do not dispute the other

legislative goal of increasing voter turnout by having . . .

32
elections in even-numbered years.” Raleigh Wake Citizens Ass’n,

2016 WL 1060378, at *27 n.18. But they did not need to dispute

that goal, because it has no logical connection to, and does not

justify, re-drawing districts, much less districts with

population deviations.

A further rationale given for the redistricting: allowing

voters greater representation. Yet the redistricting of the

County Commission arguably reduced citizens’ opportunity to cast

votes for their preferred commissioners by moving away from an

all at-large system. As testified at trial, voters “had the

ability to elect all seven members . . . . As it stands with the

maps that were passed by the House and the Senate, [they] will

be able to exercise [their] vote on only two of those members,

so with every -- everything that I know about the word

representation, that’s less.” J.A. 387-88. And again, nothing

about this goal explains the population deviations of the

districts as drawn.

Moreover, alternatives were suggested that would have

achieved, even more effectively, the stated rationale of

increased representation without resulting in such great

population deviations. For example, Representative Darren

Jackson proposed an amendment to create two purely at large

districts instead of the donut and donut hole districts, while

maintaining the 2011 single-member districts. Such a plan would

33
have “accomplish[ed] both of the Republicans’ stated goals, to

give you more representation on the School Board and to make

sure that you had a School Board member who represented your

child’s school, and it accomplished both of those goals.” J.A.

354. That amendment, which would have achieved greater

representation on the School Board, was rejected—yet more

evidence that the stated rationales were pretextual and fail to

justify the population deviations in the challenged

redistricting.

The legislators pushing the redistricting also sought to

ground it in administrative ease, having the School Board and

Board of County Commissioners fall under the same plan. Again,

that goal is wholly unrelated to, and plainly fails to justify,

the deviations in population amongst the districts. Somewhat

relatedly, and certainly breathtakingly under the circumstances,

the Board of County Commissioners’ redistricting was ostensibly

intended to “avoid litigation.” Raleigh Wake Citizens Ass’n,

2016 WL 1060378, at *37. Yet the School Board redistricting was

being actively litigated and was in fact pending before this

Court at that time. The litigation rationale is thus utterly

irrational and, further, has no logical connection to the

deviations at issue.

Moving beyond the pretextual rationales, the record

evidence demonstrates that traditional, legitimate apportionment

34
factors did not predominate. On the contrary, the redistricting

resulted in: “a total of 31 [split] precincts” (as opposed to 12

split precincts under the 2011 plan), J.A. 805; bizarrely shaped

districts, including “donut[s]” and “donut munchkin[s],” J.A.

432, “crab claw[s]” and “pincer[s],” J.A. 212; and obviously

non-compact districts that make it harder, for example, for

School Board members “to have more detailed knowledge about

[their] own districts,” J.A. 280. 11 Indeed, Plaintiffs’ expert

Dr. Chen considered several traditional, legitimate

reapportionment criteria, i.e., population equality, community

and precinct boundaries, and geographical compactness, and found

that the redistricting “create[d] a partisan distribution of

seats falling completely outside the range of outcomes that are

possible under a non-partisan districting process that creates

equally populated districts while maximizing compactness and

preserving precinct and municipal boundaries.” J.A. 768.

Representative Rosa Gill also proposed an alternative

redistricting during the legislative process. Her proposal

demonstrated that it was entirely possible to meet all of the

stated rationales for the skewed redistricting—including giving

voters the opportunity to elect two school board members,

11
No party has made an argument regarding Voting Rights Act
compliance, also recognized as a legitimate apportionment
factor. We therefore do not address it.

35
providing district representation for the County Commissioners,

moving school board elections to even numbered years to increase

turnout, reducing voter confusion by using the same districts

for both the School Board and the Board of County Commissioners,

and reducing costs—while creating only miniscule deviations.

Representative Gill’s plan divided no precincts and had overall

deviations in the single-member and super districts of less than

0.5%. J.A. 795-96.

The trial court dismissed the evidence of Representative

Gill’s alternative plan because it “simply shows that ‘better’

plans can be drawn, but ‘better’ plans do not equate to

unconstitutionality.” Raleigh Wake Citizens Ass’n, 2016 WL

1060378, at *33. In fact, what the alternative plan shows is

that legitimate considerations, including the stated rationales

for the redistricting, utterly failed to explain or justify the

high levels of deviation in the enacted plans—because those

rationales could have been accomplished by a plan with virtually

no population deviations.

5.

At the end of the day, when we review the evidentiary

record, we can reach only one conclusion: that Plaintiffs, the

only parties to make their case at trial, successfully showed it

to be more probable than not that the deviations at issue here

reflect the predominance of an illegitimate reapportionment

36
factor rather than legitimate considerations. Harris, 136 S.

Ct. at 1307. We recognize that, generally, “attacks on

deviations under 10% will succeed only rarely, in unusual

cases.” Id. But after reviewing this matter closely, and for

the reasons discussed above, we are convinced that these mid-

decade, partisan redistricting plans constitute just such an

unusual case. The district court therefore committed reversible

error in granting judgment in Defendant’s favor.

6.

In addition to improper partisanship, Plaintiffs claimed

improper regional favoritism as an illegitimate factor behind

the deviations in the challenged reapportionments. Because we

have already ruled in Plaintiffs’ favor based on partisanship,

we need not reach this related but separate basis. We

nevertheless note that “[a] citizen, a qualified voter, is no

more nor no less so because he lives in the city or on the farm.

This is the clear and strong command of our Constitution’s Equal

Protection Clause.” Reynolds, 377 U.S. at 568. Therefore,

“[i]n Larios, a federal court struck down a [state] legislative

redistricting plan . . . . The plaintiffs there alleged that the

plan . . . under-populat[ed] districts in the urban Atlanta

region and the rural south-Georgia area—both Democratic

strongholds—while over-populating districts with Republican-

leaning voters.” Wright, 787 F.3d at 266-67. In Wright, we

37
left no doubt that, as in Larios, Plaintiffs here claim that “a

state legislature designed a redistricting plan with a maximum

deviation in population of just under 10%, designed to pit rural

and urban voters against one another” and that “[e]ven if Larios

does not control this case . . ., we nevertheless find it” and

its rejection of regional favoritism as a basis for deviating

from ideal population by such margins “persuasive.” Id. at 267.

Moreover, the district court held that “the General

Assembly rationally considered the communities of interest

within Wake County’s urban areas and within Wake County’s rural

and suburban areas in adopting” the challenged redistricting

plans. Raleigh Wake Citizens Ass’n, 2016 WL 1060378, at *40.

But the pertinent inquiry is not whether it was “rational” to

“consider” communities of interest in adopting the plans

generally; instead, the proper inquiry is whether the

redistricting’s deviations more likely than not reflect the

predominance of illegitimate reapportionment factors. Harris,

136 S. Ct. at 1307. The district court plainly engaged the

wrong legal standard in its analysis of this factor. But

because we rule on the basis of partisanship, we need go no

further of the regional favoritism issue.

D.

In addition to their federal constitutional one person, one

vote claim, Plaintiffs brought a similar North Carolina state

38
claim. Under the North Carolina Constitution, “[t]he right to

vote on equal terms in representative elections—a one-person,

one-vote standard—is a fundamental right.” Blankenship v.

Bartlett, 681 S.E.2d 759, 762–63 (N.C. 2009). A North Carolina

analysis of the state’s “Equal Protection Clause generally

follows the analysis of the Supreme Court of the United States

in interpreting the corresponding federal clause.” Id. at 762.

If anything, North Carolina’s one person, one vote principle

applies with even more force than its federal counterpart. See,

e.g., id. at 763 (deeming the one person, one vote principle

applicable in North Carolina’s election of superior court judges

even though “federal courts have articulated that the ‘one-

person, one-vote’ standard is inapplicable to state judicial

elections”); Stephenson v. Bartlett, 562 S.E.2d 377, 397 (N.C.

2002) (requiring legislative districts to be within plus or

minus five percent of ideal population). Accordingly, for the

same reasons that Plaintiffs succeed with their federal claim,

so, too, do they succeed with their North Carolina state one

person, one vote claim.

IV.

In addition to their one person, one vote claim, Plaintiffs

have also brought a racial gerrymandering claim regarding the

39
Board of County Commissioners’ District 4. 12 Plaintiffs contend

that race predominated in determining the boundaries, shape, and

composition of that district without narrow tailoring to serve a

compelling state interest. As explained below, the district

court did not commit clear error in rejecting this claim.

A.

To successfully challenge the constitutionality of an

electoral district under the Equal Protection Clause, a

plaintiff must “show, either through circumstantial evidence of

a district’s shape and demographics or more direct evidence

going to legislative purpose, that race was the predominant

factor motivating the legislature’s decision to place a

significant number of voters within or without a particular

district.” Ala. Legislative Black Caucus v. Alabama, 135 S. Ct.

1257, 1267 (2015) (quotation marks and citation omitted).

Such a showing requires proof that “the legislature

subordinated traditional race-neutral districting principles . .

. to racial considerations.” Miller v. Johnson, 515 U.S. 900,

916 (1995). Traditional race-neutral principles include

“compactness, contiguity, and respect for political subdivisions

or communities defined by actual shared interests,” id.,

12
Even though the corresponding School Board district is
identical, Plaintiffs in Wright made no such claim. We, like
the district court, therefore do not address that issue.

40
incumbency protection, and political advantage, Bush v. Vera,

517 U.S. 952, 964, 968 (1996). And evidence that such

traditional principles took a back seat to racial considerations

may include direct and circumstantial evidence of legislative

intent, indications that a racial percentage within a given

district was non-negotiable, bizarre or non-compact district

shapes, and district lines that cut through traditional

geographic boundaries or election precincts. See, e.g., Vera,

517 U.S. at 970-71; Miller, 515 U.S. at 917-18; Shaw v. Reno,

509 U.S. 630, 646–48 (1993).

If a plaintiff successfully shows racial predominance in

drawing the lines of a district, the court must apply “strictest

scrutiny,” Miller, 515 U.S. at 915, that is, it must determine

whether the design of the challenged district was narrowly

tailored to advance a compelling state interest—a burden the

state must bear, Shaw v. Hunt, 517 U.S. 899, 908 (1996). If the

answer to that question is no, the district must be struck as

unconstitutional.

B.

In contrast to its one person, one vote analysis, the

district court did not miscomprehend the applicable law.

Accordingly, while we were “not bound by the clearly erroneous

standard” regarding the one person, one vote findings, Inwood

Labs., 456 U.S. at 855 n.15, the same cannot be said here.

41
Here, we must affirm if “the district court’s account of the

evidence is plausible,” even if we are “convinced that we would

have decided the question of fact differently.” TFWS, Inc. v.

Franchot, 572 F.3d 186, 196 (4th Cir. 2009) (quotation marks and

citation omitted).

While we might have decided this matter differently in the

first instance, we cannot say that the district court’s account

of the evidence is not plausible; it is. For example, the

district court considered legislator comments indicating that

race was a consideration in the redistricting process, such as a

representative’s observation “that at-large electoral systems

submerge the views of various minorities, ‘whether it’s racial,

gender, political, rural, urban or whatever.’” Raleigh Wake

Citizens Ass’n, 2016 WL 1060378, at *46. While such comments

evidence the fact that race was a consideration in the

redistricting process, doing so is not unlawful. See, e.g.,

Miller, 515 U.S. at 916 (“Redistricting legislatures will, for

example, almost always be aware of racial demographics; but it

does not follow that race predominates in the redistricting

process.”). We cannot fault the district court for determining

that the comments here did not constitute direct evidence that

race predominated in the drawing of District 4, i.e., of racial

gerrymandering.

42
 Further, in the racial gerrymandering context, partisan

advantage may be considered a traditional redistricting

criterion, and evidence that politics was the primary motivation

for the drawing of a district can defeat an allegation that race

predominated. See, e.g., Cromartie, 532 U.S. at 257–58; Vera,

517 U.S. at 968. The district court recognized this, noting

that the fact that District 4 is majority-minority “alone does

not mean that the General Assembly racially gerrymandered

District 4,” Raleigh Wake Citizens Ass’n, 2016 WL 1060378, at

*47, and that evidence supports the district’s having been drawn

with a focus on partisanship rather than race. For example, in

evaluating the expert support for Plaintiffs’ racial

gerrymandering claim, the district court noted that the expert’s

“partisan neutral” analysis did not help answer the question of

whether politics or race led to District 4’s boundaries. Id.

Here, too, we cannot disagree.

In sum, even if we might have found otherwise in the first

instance, it was not implausible for the district court to

determine that Plaintiffs had fallen short of proving that

traditional districting criteria were subordinated to race in

the drawing of District 4. Accordingly, because the district

court’s analysis of Plaintiffs’ racial gerrymandering claim is

not clearly erroneous, we affirm on that issue.

43
 V.

For the reasons discussed above, we reverse the district

court’s judgment in Defendant’s favor as to Plaintiffs’ one

person, one vote claims. We remand with instructions to enter

immediately 13 judgment for Plaintiffs, granting both declaratory

relief and a permanent injunction, as to the one person, one

vote claims. However, we affirm the district court’s judgment

for Defendant as to Plaintiffs’ racial gerrymander claim.

REVERSED AND REMANDED IN PART
AND AFFIRMED IN PART

13 We see no reason why the November 2016 elections should
proceed under the unconstitutional plans we strike down today.

44
DIANA GRIBBON MOTZ, Circuit Judge, dissenting:

With respect, I dissent from the majority’s holding that

the district court erred in rejecting Plaintiffs’ equal

protection challenge to twin presumptively constitutional

redistricting plans. Plaintiffs’ one person, one vote claim

rests on their contention that improper “partisanship” rendered

the challenged redistricting plans unconstitutional, even though

those plans have population deviations of less than 10%. 1 If

such a claim is justiciable, and it is not clear that it is, the

showing necessary to prove such a claim is extremely demanding.

The Supreme Court explained only a few weeks ago that such

challenges “will succeed only rarely, in unusual cases.” Harris

v. Ariz. Indep. Redistricting Comm’n, 136 S. Ct. 1301, 1307

(2016). The challenge here, like that in Harris, is not that

1
In their amended complaint, Plaintiffs also alleged that
the plans impermissibly favored rural voters over urban voters.
At trial, however, they focused on assertedly improper
“partisanship” and produced scant evidence that the State sought
to advantage rural over urban voters. Plaintiffs did not even
offer evidence as to which districts they considered “urban” or
“rural.” Their experts testified that assertedly illegitimate
“partisan” motivations, not regional favoritism, predominately
motivated the challenged plans. Unsurprisingly, the district
court found that Plaintiffs “failed to prove” that either plan
“impermissibly favors suburban and rural voters over urban
voters or substantially dilutes the individual voting strength
of Wake County’s urban voters.” Raleigh Wake Citizens Ass’n v.
Wake Cty. Bd. of Elections, No. 5:15-CV-156-D, 2016 WL 1060378,
at *40 (E.D.N.C. Feb. 26, 2016). On appeal, Plaintiffs provide
no basis on which to disturb that finding.
“unusual case.” For this reason, I would affirm in its entirety

the judgment of the district court rejecting Plaintiffs’

challenges to the redistricting plans.

I.

The Equal Protection Clause requires a State to “make an

honest and good faith effort to construct [state legislative]

districts . . . as nearly of equal population as is

practicable.” Reynolds v. Sims, 377 U.S. 533, 577 (1964). But,

the Reynolds Court itself recognized that, in determining what

is “practicable,” the Constitution permits some deviations from

perfect population equality when justified by “legitimate

considerations incident to the effectuation of a rational state

policy.” Id. at 579; accord Harris, 136 S. Ct. at 1306.

In a long line of cases decided in the wake of Reynolds,

the Court has held that districts, like those at issue here,

with a “maximum population deviation under 10%” are

presumptively constitutional. See, e.g., Brown v. Thomson, 462

U.S. 835, 842 (1983); accord Harris, 136 S. Ct. at 1307 and

cases cited therein. These “minor deviations from mathematical

equality do not, by themselves, make out a prima facie case of

invidious discrimination under the Fourteenth Amendment so as to

require justification by the State.” Harris, 136 S. Ct. at 1307

46
(quoting Gaffney v. Cummings, 412 U.S. 735, 745 (1973))

(internal quotation marks omitted).

It was because of “the inherent difficulty of measuring and

comparing factors that may legitimately account for small

deviations from strict mathematical equality” that the Supreme

Court recently reiterated that “attacks on deviations under 10%

will succeed only rarely, in unusual cases.” Harris, 136 S. Ct.

at 1307. To prevail on such claims, the Harris Court held that

a challenger “must show that it is more probable than not that a

deviation of less than 10% reflects the predominance of

illegitimate reapportionment factors rather than the ‘legitimate

considerations’” that the Court had identified in previous

cases. Id.

In earlier cases the Supreme Court had identified numerous

“legitimate considerations” justifying a State’s reapportionment

plan. Among them are a State’s valid interests in: maintaining

the competitive balance among political parties, Gaffney, 412

U.S. at 752-53; accord Harris, 136 S. Ct. at 1306, avoiding

contests between incumbents as long as incumbents of one party

are not favored over those of another, Karcher v. Daggett, 462

U.S. 725, 740 (1983), and recognizing communities of interest,

Evenwel v. Abbott, 136 S. Ct. 1120, 1124 (2016). Indeed, in

League of United Latin American Citizens v. Perry, the Supreme

Court characterized “avoiding the pairing of incumbents” as a

47
“‘neutral’ redistricting standard[ ]” and “maintaining

communities of interest” as a “traditional districting

principle[ ].” 548 U.S. 399, 412 (2006) (plurality opinion)

(“LULAC”); Id. at 433 (majority opinion).

Thus, notwithstanding Plaintiffs’ apparent belief, the

Court has expressly recognized that a redistricting plan can in

these ways legitimately take account of political

considerations. The Court has never suggested that doing so

constitutes reliance on an “illegitimate reapportionment

factor.” Harris, 136 S. Ct. at 1307. This approach necessarily

follows from the fact that “[p]olitics and political

considerations are inseparable from districting and

apportionment” and so “districting inevitably has and is

intended to have substantial political consequences.” Gaffney,

412 U.S. at 753.

If those attacking a redistricting plan prove that a State

has abused legitimate political considerations by systemically

over- or under-populating districts to benefit one party at the

expense another, then the challengers may be able to prevail as

they did in Larios v. Cox, 300 F. Supp. 2d 1320, 1325 (N.D.

Ga.), aff’d, 542 U.S. 947 (2004) (mem.). Plaintiffs lean

heavily on Larios. Their reliance is misplaced.

First, Plaintiffs ignore the very different factual record

developed in that case. In Larios, the challenged plan paired

48
in the same district, and thus pitted against each other, 37 of

the 74 incumbent Republicans but only 9 of the 105 incumbent

Democrats. 300 F. Supp. 2d at 1326. In Larios, Georgia

legislators admitted before the district court that they had

intentionally drawn legislative districts to favor incumbents of

one party over those of the other. Id. at 1325. Thus, in

Larios, the state legislators conceded that they had not made

the “good faith effort” to draw equal districts that Reynolds

requires. The record in this case contains no such evidence.

In addition to ignoring the very different evidentiary

record in Larios, Plaintiffs turn a blind eye to the Court’s

subsequent treatment of that case. In LULAC, the Court

explained that Larios “does not give clear guidance” in

“addressing political motivation as a justification for an

equal-protection violation.” 548 U.S. at 423 (plurality

opinion). And in Harris, the unanimous Supreme Court expressly

reserved the question of whether the sort of abusive

partisanship at issue in Larios even constitutes “an

illegitimate redistricting factor.” Harris, 136 S. Ct. at 1310. 2

2 Tellingly, the Court has never addressed the alternative
holding by the lower court in Larios invalidating the challenged
plans on the basis of regional favoritism. That alternative
holding has little precedential or persuasive value given, as
the Supreme Court has explained, “a summary affirmance is an
affirmance of the judgment only,” not the rationale of the lower
court, which “should not be understood as breaking new ground.”
(Continued)
49
Despite Plaintiffs’ protestations to the contrary, the

foundations of Larios as persuasive authority rest on shaky

ground.

Equally significantly, Plaintiffs take no notice of the

holding in Harris that, even if abusive partisanship did

constitute an illegitimate factor, those challenging the

redistricting plan before it had “not carried their burden.”

Id. This holding is particularly significant given that the

Harris plaintiffs had made a much stronger evidentiary showing

than Plaintiffs do here. For example, the Harris plaintiffs

offered direct evidence of a Republican-leaning district made

“more competitive” at the request of a Democratic redistricting

commissioner by “hyperpacking Republicans into other districts.”

Id. at 1309 (internal quotation marks omitted). The

redistricting commission in Harris had overpopulated almost all

the Republican-leaning districts in the thirty-district plan

while underpopulating almost all the Democratic-leaning

Mandel v. Bradley, 432 U.S. 173, 176 (1977). And invalidating a
redistricting plan because it allegedly favors “rural” or
“urban” voters would break new ground. The Supreme Court has
never before or after Larios suggested that considering the
urban or rural characteristics of a district is an illegitimate
apportionment factor. In fact, statements in several cases
suggest that these are the quintessential types of communities
of interest a State may consider when redistricting. See, e.g.,
Dusch v. Davis, 387 U.S. 112, 117 (1967).

50
districts. Id. at 1309-10. Even in the face of this evidence,

the district court did not find the redistricting plan

unconstitutional -- and the Supreme Court agreed. Id. at 1309.

Furthermore, in explaining its rejection of the Harris

plaintiffs’ claims, the Supreme Court distinguished Larios in

ways that apply with equal force here. The Harris Court held

that in Larios, unlike in the case before it (and unlike in the

case at hand), “the district court found that those attacking

the plan had shown” that no legitimate factors explained the

deviations in the plan. Id. at 1310 (emphasis added). The

Harris Court explained: “It is appellants’ inability to show”

that illegitimate factors predominated “that makes [Larios]

inapposite here.” Id. Thus the Court emphasized and re-

emphasized that those attacking a presumptively constitutional

redistricting plan, like Plaintiffs here, must prove that

illegitimate factors predominated.

In sum, even if abusive partisanship claims are

justiciable, and do provide the basis for a one person, one vote

claim, Plaintiffs had to prove at trial that the State relied on

this consideration in redistricting, and that this reliance took

precedence over all legitimate considerations, including

maintaining political balance among political parties, avoiding

contests between incumbents of both parties, and recognizing

communities of interest. The State, on the other hand, did not

51
need to offer any justification for its presumptively

constitutional redistricting plans. See, e.g., Harris, 136

S. Ct. at 1307.

A fair review of the factual record seems to me to

demonstrate that, as in Harris, Plaintiffs here failed to meet

their burden and so, as the Supreme Court did in Harris, we

should affirm the district court’s rejection of their challenge. 3

II.

In attempting to meet their substantial burden, Plaintiffs

principally rely on the trial testimony of their expert,

Dr. Jowei Chen. On the basis of statistical models that he had

created, Dr. Chen opined that deviations in the challenged

redistricting plans were motivated entirely by a desire to

obtain “Republican partisan control over four of the” seven

numbered districts and over one of the two lettered super-

districts. But, as the district court found, Dr. Chen’s model

3
The district court also rejected Plaintiffs’ one person,
one vote claim under the North Carolina Constitution. Because
North Carolina courts “generally follow[ ] the analysis of the
Supreme Court of the United States” when interpreting the
State’s corresponding Equal Protection Clause, I would affirm
the district court’s finding that Plaintiffs failed to carry
their burden on their state law claims for the same reasons that
apply to their federal claims. Blankenship v. Bartlett, 681
S.E.2d 759, 762 (N.C. 2009).

52
simply does not prove either conclusion. Dr. Chen’s analysis

suffers from two critical flaws.

First, in his model, Dr. Chen pegged the maximum tolerable

level of population deviation between districts at 2%. In doing

so he held the State to a standard not required by law. Of

course, a State must make a “good faith effort” to draw equal

districts. Reynolds, 377 U.S. at 577. But neutral factors may

cause population deviations well above 10% without running afoul

of the Constitution. See, e.g., Mahan v. Howell, 410 U.S. 315,

328 (1973). Moreover, Dr. Chen’s arbitrary 2% threshold seems

particularly unwarranted in light of the Supreme Court’s

repeated characterization of deviations below 10% as “minor” and

its admonition that such minor deviations do not “substantially

dilute the weight of individual votes in the larger districts so

as to deprive individuals in these districts of fair and

effective representation.” White v. Regester, 412 U.S. 755, 764

(1973).

The second fatal flaw in Dr. Chen’s analysis is his failure

to look beyond what he considered to be the only four legitimate

or “traditional” districting factors -- population equality,

intact municipal boundaries, intact precincts, and geographic

53
compactness. 4 Dr. Chen ignored the many apolitical and political

factors States may consider during redistricting (like striking

a competitive balance among political parties, avoiding contests

among incumbents, and recognizing communities of interest), even

if pursuing these goals causes minor population deviations.

This is particularly troubling because it is undisputed

that two of the legitimate districting factors Dr. Chen failed

to consider -- incumbency protection and grouping communities of

interest -- actually motivated the legislature here. The

parties stipulated to the accuracy of transcripts of the

legislative debate and those transcripts reveal that state

legislators altered the district lines in the final version of

the School Board redistricting bill to protect two incumbents --

one registered Democrat and one registered Republican. Further,

the Democratic incumbent, Christine Kushner, testified at trial

that “Ms. Prickett, who is a registered Republican, had been

placed into a Democratic leaning district,” but “was moved out

4 Plaintiffs actually concede the limited reach of
Dr. Chen’s analysis, noting that his analysis “shows that the
partisanship of the enacted districts does not happen when
traditional redistricting criteria are followed.” See
Plaintiffs’ Rep. Br. at 21 (emphasis added). Of course, as
explained above, the Supreme Court has repeatedly recognized
numerous legitimate ‘redistricting criteria’ other than those
that Dr. Chen considers “traditional.” And in LULAC, 548 U.S.
at 433, the Court expressly included “maintaining communities of
interest” among “traditional” redistricting criteria.

54
of that district and put into a Republican leaning district, and

I [Ms. Kushner] was switched out of District 2 into District 5,”

which she admitted was a “more favorable district” for her.

Accommodating the legitimate interest in protecting incumbents

of both parties had a demonstrable impact on the population

deviations across four of the seven numbered districts in the

plan. District 1 swung from 2.76% overpopulated to -0.41%

underpopulated. District 2 swelled from -4.19% underpopulated

to just -1.05% underpopulated. District 5 dipped from 0.19%

overpopulated to -1.53% underpopulated. Finally, District 6

grew from -0.14% underpopulated to 1.6% overpopulated.

Dr. Chen’s model does not in any way account for these

population deviations. As a result, Dr. Chen’s view that

nothing but improper “partisanship” could explain the population

deviations in the twin redistricting plans completely ignores

the undisputed impact that the legislative effort to protect the

two incumbents had on the plans. In light of that omission, I

cannot agree that the district court clearly erred in concluding

that Dr. Chen’s testimony did not demonstrate that the

legislature deviated from population equality only for the

predominant purpose of creating four safe Republican seats out

of seven.

Dr. Chen committed the same sort of analytic error in

considering the two lettered super-districts. One of the stated

55
purposes for the super-districts was to improve representation

for voters in rural areas. Without challenging the State’s

consideration of communities of interest generally, Plaintiffs

argue that “[t]here is no possible way that [the stated]

rationales explain why Super District A needs to be 44,117

people larger than Super District B.” Plaintiffs’ Rep. Br. at

12. According to Dr. Chen, again improper “partisanship” is the

only explanation.

And again, Dr. Chen’s model does not support this

conclusion. To be sure, the State could have overpopulated

District A, an area of the County that has historically voted

for Democratic candidates, to increase a Republican candidate’s

odds of winning in District B. But the State also could have

deviated from population equality to group more urban areas in

District A based on their shared interests. This, after all,

was the purpose for having the super-districts in the first

place, and of course it constitutes a clearly valid State

interest. See Evenwel, 136 S. Ct. at 1124. Or the State could

have had the dual motivation to accomplish both. Dr. Chen’s

model tells us nothing about how grouping together communities

of interest motivated the legislature because it a priori

excludes any consideration of that legitimate redistricting

consideration.

56
 Plaintiffs’ remaining evidence also falls far short of

meeting their burden of proving that illegitimate partisan

considerations predominated here. Plaintiffs’ expert Anthony

Fairfax concluded that the legislature desired to “minimize the

Democratic performance” in certain districts by overpopulating

the “Democratic performing districts.” That opinion rests on

his view that correlation between overpopulation and Democratic

performance in the districts in and of itself demonstrates

legislative intent -- i.e., that the numbers speak for

themselves. The district court concluded that they do not.

The record here provides no basis for holding that finding

clearly erroneous. Of the four districts assertedly favorable

or competitive for Democrats, three are overpopulated. Of the

five districts assertedly favorable or competitive for

Republicans, only three are underpopulated by more than 1%. One

of these three districts, District 2, is underpopulated by just

-1.05%. Thus, the asserted correlation between population and

Democratic performance is, to say the least, minimal. This

minimal correlation limits the strength of any inference that

can be drawn. Cf. Harris, 136 S. Ct. at 1309-10 (refusing to

infer predominance of illegitimate partisanship over a thirty-

district plan where every district underpopulated by more than

1% (nine total) favored Democrats and every district

overpopulated by more than 1% (twelve total) favored

57
Republicans). At the very least, the district court did not

clearly err when it declined, as the Supreme Court did in Harris

in the face of stronger evidence, to make an inference of

unconstitutional motivation.

Plaintiffs also offered the lay testimony of members of the

state legislature who opposed the redistricting plans. I agree

with the majority that the district court erred in categorically

rejecting this testimony as irrelevant. But, despite this

error, the testimony does not move the needle far on the issue

of intent of those voting to adopt the redistricting plans

because, to a person, Plaintiffs’ lay witnesses disclaimed any

knowledge of the sponsors’ motivations. 5

In sum, faced with the heavy burden of proving that

assertedly illegitimate “partisanship” constituted the

predominant motivation for the presumptively constitutional

redistricting plans, Plaintiffs failed to offer any evidence

truly probative of legislative intent. Plaintiffs’ experts

tendered conclusions that their analyses could not support.

5More probative are emails from Wake County Republican
Chairwoman Donna Williams to Republican members of the state
legislature and School Board. Williams expressed concern that
the proposed map would not be sufficiently favorable to
Republicans to permit them to “take 5 of the 9 seats.” However,
the record does not contain requests for information or
responses from State officials or any indication that
Ms. Williams’ lobbying efforts had any effect on the
legislation.

58
Plaintiffs’ remaining evidence proved little. The district

court refused to draw Plaintiffs’ preferred inference. In doing

so, the court did not clearly err. To the contrary, given the

weakness of Plaintiffs’ case, Defendants would have had strong

grounds to appeal had the district court ruled otherwise.

I would affirm the judgment of the district court in its

entirety.

59