# Missouri State Conference of the National Ass'n for the Advancement of Colored People v. Ferguson-Florissant School District

> District Court, E.D. Missouri · August 22, 2016 · 201 F. Supp. 3d 1006

URL: https://www.frixlaw.com/law-library/cases/7239502

## Case

- **Full name:** MISSOURI STATE CONFERENCE OF THE NATIONAL ASSOCIATION FOR THE ADVANCEMENT OF COLORED PEOPLE v. FERGUSON-FLORISSANT SCHOOL DISTRICT
- **Court:** District Court, E.D. Missouri
- **Decided:** August 22, 2016
- **Citations:** 201 F. Supp. 3d 1006; 2016 U.S. Dist. LEXIS 111177; 2016 WL 4429695
- **Precedential status:** Published
- **Opinion:** Opinion of the court by Sippel
- **Judges:** Sippel
- **Cited by:** 10 later opinions in the Frix Law Library

## Citator (automated)

- No negative treatment found by the automated citator. That is not the same as a confirmation that the case is good law; read the citing cases.
- Full citator and citing cases: https://www.frixlaw.com/law-library/cases/7239502

## How later opinions describe it (automated extraction)

- stating that an expert opinion is admissible if there are “sufficient facts already in evidence ... to take his testimony out of the real of guesswork and speculation”
- stating that an expert opinion is admissible if there are “sufficient facts already in evidence...to take his testimony out of the real of guesswork and speculation”
- explaining the court "need not supplement endogenous election data" from five elections where the evidence was sufficient "to discern typical voting behavior and usual results"

## Opinion text

MEMORANDUM OPINION AND ORDER
RODNEY W. SIPPEL, UNITED STATES DISTRICT JUDGE
Plaintiffs Doris Bailey, Redditt Hudson, F. Willis Johnson, and the Missouri State Conference of the National Association for the Advancement of Colored People (“MO NAACP”) bring suit against Defendants Ferguson-Florissant School District (“FFSD” or “the District”) and the St. Louis County Board of Elections Commissioners (“St. Louis BOEC”) under § 2 of the Voting Rights Act of 1965, 52 U.S.C. § 10301 . Plaintiffs contend that the electoral structures used in Ferguson-Floris-sant School Board (“the Board” or “FFSB”) elections interact with historical and socioeconomic conditions to deprive the African American voters in FFSD of an equal opportunity to elect representatives of their choice.
Section 2 of the Voting Rights Act provides that no “standard, practice, or procedure shall be imposed or applied by any State or political subdivision in a manner which results in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color.” 52 U.S.C.A. § 10301 (a). A § 2 violation:
is established if, based on the totality of circumstances, it is shown that the political processes leading to nomination or election in the State or political subdivision are not equally open to participation by members of a class of citizens protected by subsection (a) in that its members have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice. The extent to which members of a protected class have been elected to office in the State or political subdivision is one circumstance which may be considered: Provided, That nothing in this section establishes a right to have members of a protected class elected in numbers equal to their proportion in the population.
52 U.S.CA. § 10301(b).
' [1] To establish a claim of vote 1 dilution under § 2 of the Voting Rights Act, plaintiffs must first establish the following three “preconditions” pursuant to the United States Supreme Court’s decision in Thornburg v. Gingles, 478 U.S. 30 , 106 S.Ct. 2752 , 92 L.Ed.2d 25 (1986):
1. “[T]he minority group.. .is sufficiently large and geographically compact to constitute a majority in a single-member district,”
2. “[T]he minority group.. .is politically cohesive,” and
3. “[T]he white majority votes sufficiently as a bloc to enable it — in the ab *1016 sence of special circumstances...— usually to defeat the minority’s preferred candidate.”
Id. at 50-51 , 106 S.Ct. 2752 .
If all three preconditions are established, then a court must “consider the ‘totality of the circumstances’ and [ ] determine, based ‘upon a searching practical evaluation of the past and present reality,’ whether the political process is equally open to minority voters.” Id. at 79 , 106 S.Ct. 2752 (internal citations and quotations omitted). “ ‘This determination is peculiarly dependent upon the facts of each case,’ and requires ‘an intensely local appraisal of the design and impact’ of the contested electoral mechanisms.” Id. In undertaking this practical evaluation, courts look to the non-exhaustive list of “typical factors” identified in the Senate Report accompanying the 1982 amendments to the VRA (“Senate Factors”), see S. Rep. No. 97-417, at 28-29.
I held a six day non-jury trial in this matter beginning on January 11, 2016. The parties filed post-trial briefs and proposed findings of fact and conclusions of law on April 8,2016. On April 22, 2016, the parties filed responses to each other’s post-trial briefs. Also on April 22, 2016, the District filed a motion to re-open the case for additional evidence, in which it asks me to take judicial notice of the election results from the April 5, 2016 school board election in FFSD. On April 26, 2016, Plaintiffs responded to the District’s motion to reopen the case, stating that they do not oppose the motion, but requesting that I take judicial notice of additional facts related to the 2016 election.
After consideration of the testimony given at trial, the exhibits introduced into evidence, the briefs of the parties, and the applicable law, I make the following findings of fact and conclusions of law. As discussed below, I conclude that Plaintiffs have established a § 2 violation.
I. BACKGROUND
A. Parties
The individual Plaintiffs are U.S. citizens, registered voters, and African Americans who reside in areas of FFSD that could constitute single-member districts in which African Americans are a majority of the voting-age population. Joint Stip. ¶¶ 2-4.
Plaintiff MO NAACP is a state affiliate of the NAACP. The NAACP is the nation’s oldest and largest civil rights organization. The mission of the NAACP is to ensure the political, educational, social, and economic equality of rights of all persons, to eliminate hatred and racial discrimination, and to remove all barriers of racial discrimination through democratic processes. Plaintiff MO NAACP is active in efforts to increase voter registration, education, and turnout, and emphasizes the importance of local elections, including local school board elections. Joint Stip. ¶¶ 5-6. The membership of the MO NAACP includes African Americans who reside, work, and raise families in the District and in areas of the District that could constitute single-member districts in which African Americans are a majority of the voting-age population. See PLTF-117, Hudson Deck, ¶¶ 1, 3-4; PLTF-119, Pruitt Deck, ¶ 4; Trial Tr. vol. 1, 20:28-21:2 (testimony of Adolphus Pruitt); Trial Tr. vol. 4, 86:6-8 (testimony of Redditt Hudson).
Defendant FFSD is a governmental entity that maintains an electoral system based on at-large elections for seven positions on the FFSB. The Board is responsible for the governance and administration of FFSD, a political subdivision of the State of Missouri within the meaning of Article 4, Section 12, of the Missouri Constitution. Joint Stip. ¶ 7.
Defendant St. Louis BOEC is the governmental entity charged with conducting *1017 elections in St. Louis County. It is responsible for conducting elections for positions on the Board in FFSD. Joint Stip. ¶ 8.
FFSD is located in northern St. Louis County, Missouri (“North County”). The District was created by a 1975 desegregation order, which required the then-Ferguson-Florissant School District to annex the primarily African American neighboring school district of KMoch and primarily white neighboring district of Berkeley. Joint Stip. ¶ 9; see also United States v. Missouri, 515 F.2d 1365, 1369-73 (8th Cir.1975). The, District covers all or part of eleven municipalities: Berkeley, Calverton Park, Cool Valley, and Kinloch in their entirety, and parts of Black Jack (one block), Ferguson, Florissant, Dellwood, Hazelwood, Normandy, and Old Jamestown. Joint Stip. ¶ 11. Its headquarters and administration center are in Floris-sant. Id. The City of Florissant comprises the largest portion of the District while the City of Ferguson comprises the second largest portion of the District. Joint Stip. ¶¶ 9,11.
B. Expert Witnesses
The following witnesses provided expert testimony for Plaintiffs. William Cooper 1 testified as to “whether it is possible to create four majority-Black districts,” and as to the historical and current demographics in the District. He also compiled population demographics as reported by the 2010 census. PLTF-44, Cooper Report, ¶9-10. Dr. Colin Gordon testified to Senate Factor 5. PLTF-40, Gordon Report. 2 Dr. David Kimball testified to Senate Factors Two, Three, Four, Five, Seven, Eight and Nine. 3 PLTF-48, Kimball Report, p. 14. Dr. Richard Engstrom testified to the existence of racially polarized voting. PLTF-50, Engstrom Report, ¶5.
FFSD called one expert, Dr. Jonathan Rodden, to testify to all three Cingles preconditions. DEF-A, Rodden Report, p. 2. Rodden provided rebuttal reports to each of Plaintiffs’ four experts including the Gingles preconditions and Senate factors. See DEF-B in response to Eng-strom’s Report, DEF-C in response to Cooper’s Report and DEF-D in response to Kimball and Gordon’s Reports.
C. The FFSD Election System
1. The FFSD election system at the time of trial
The Board is composed of seven members who serve three-year terms and who *1018 are elected through an at-large election system. Board elections are staggered and held off-cycle so that either two or three Board seats are elected every April in accordance with Missouri law. Joint Stip. ¶31; RSMo. §§ 162.0261; 162.341; 162.291. When the number of candidates equals the number of Board seats to be filled, no election is held, and the candidates automatically assume responsibilities as members of the Board. Joint Stip. ¶ 32; RSMo. § 115.124.
• As of the date of trial in January 2016, the seven members of the Board were: Mr. Paul Morris, Mr. Robert Chabot, Mr. Brian Scott Ebert, Ms. Leslie Hogshead, Mr. Keith Brown, Dr. Donna Paulette-Thurman, and Dr. Courtney Graves. Dr. Paulette-Thurman and Dr. Graves are African American. The remaining five Board members were white. Joint Stip. 'X 33.
Each voter has the right to cast up to two votes in a two-seat election and up to three in a three-seat election, but cannot vote more than once for the same candidate in a single election. Joint Stip. ¶ 34. Board seats are awarded to the candidates with the most votes, such that when two seats are contested, the top two vote recipients win the two seats, and when three seats are contested, the top three vote recipients win the three seats. Joint Stip. ¶ 35.
Voters can engage in bullet, or single-shot, voting. Bullet voting is casting just one vote for a single candidate and not using the remaining vote(s) on any other candidate. By engaging in bullet voting, voters increase the likelihood of electing his or her top-choice candidate. If every voter casts all of his or her votes in a two-seat election, i.e., voters do not engage in bullet voting, a candidate can receive at most 50% of all votes. If every voter casts all of his or her votes in a three-se'at election, a. candidate can receive at most 33.33% of all votes. Joint Stip. ¶ 34.
The evidence before me indicates that bullet voting is not very common in FFSD. With the exception of Courtney Graves’s 2015 campaign, the witnesses who had been involved in campaigning in FFSD testified almost uniformly that they rarely, if ever, bullet vote or encourage others to do so, except when they are running for office and bullet vote for themselves. See Henson Testimony, Trial Tr, vol. 2, 19:7-13; Graves Testimony, Trial Tr. vol. 6, 26:2-4; Ebert Dep., June 16, 2015, 19:13-17; Paulette-Thurman Dep., June 17, 2015, 23:3-6; Brown Dep., June 16, 2015, 15:12-16; Hogshead Dep., July 1, ■ 2015, 15:18-16:6; Paul Morris Dep., June 15, 2015, 69:11-20; Chabot Dep., July 2, 2015, 13:22-24; but see Schroeder Dep., July 2, 2015, 19:11-18 (has often voted for only one candidate). Other witnesses suggested that a measurable amount of bullet voting occurred in a small handful of contests. Kim-ball Testimony, -Trial Tr. vol. 2, 180:5-11; Rodden Testimony, Trial Tr. vol. 5, 67:20-68:6, 203:24-204:1; Graves Testimony, Trial Tr. vol. 6, 17:1-10; see also Morris Dep., 69:2-20 (stating that Graves “ran a , bullet vote campaign”); see also Henson Testimony, Vol II, 18:20-19:6; Hogshead Dep., 15:18-16:6; Schroeder Dep., 19:4-18.
2. The District’s motion to re-open the case for additional evidence
After the trial, the District moved to re-open the case for additional evidence. See EOF [#181]. The District seeks to introduce the certified election results for the April 5, 2016, FFSD School Board election, which occurred after the close of evidence in this case.
The certified election results for the April 5, 2016 FFSD School Board election are as follows. There were.four candidates vying for the two Board seats available in the 2016 election, Connie Harge, an African American female, received the most votes, totaling 4,572 votes, or 36.53% of the *1019 votes east. Leslie Hogshead, a white-incumbent female, received the second most votes, with 3,513 votes, or 28.07% of the votes cast. Roger Hines, an African American male, came in third, with 2,907 votes, or 23.23% of votes cast. Donna M. Damer-on, a white female, came in fourth, with 1,357 votes, or 10.84% of votes cast.
The 2016 election resulted in the addition of one African American Board member. As a result, there are currently three African Americans on the seven-member Board.
Plaintiffs do not oppose admission of the certified election results, and agree that I may take judicial notice of the certified election results. Plaintiffs request, however, that if Defendants’ motion to re-open the case for additional evidence is granted, I also take notice of the following facts- for the sake of completeness: (1) the 2016 election included another African American candidate who lost; (2) a white incumbent chose not to run for re-election; and (3) the only incumbent who did run, Leslie Hogshead, who is white, won. Plaintiffs also argue that, in the absence of any meaningful expert testimony, the 2016 election results do not provide probative information about voting patterns, polarization among African American and white voters, or the overall success of African American-preferred candidates in FFSD.
It is within a district court’s discretion to re-open a case to admit new evidence. Dent v. Beazer Materials and Servs., Inc., 156 F.3d 523 , 533 (4th Cir. 1998). When considering whether to admit new evidence, courts consider three factors: (1) whether the evidence sought to be introduced is especially important and probative; (2) whether the moving party’s explanation for failing to introduce the evidence earlier is bona fide; and (3) whether reopening will cause undue prejudice to the nonmoving party. Rivera-Flores v. Puerto Rico Tel. Co., 64 F.3d 742, 746 (1st Cir.1995). Applying these standards, I find that re-opening the case to take judicial notice of the 2016 certified election results as well as the additional facts that Plaintiffs request is proper and I will take judicial notice of these facts. Id.; Fed. R. Evid. 201. However, as Plaintiffs argue, in the absence of meaningful expert testimony regarding these facts, I am unable to draw significant legal conclusions based on these facts.
II. CENSUS DATA AND THE FFSD POPULATION
A. Student Population
Based on data provided to the U.S. Department of Education for the 2011 survey year, the District public schools serve 13,-234 students from preschool through 12th grade. 77.1% of the students are African American and 15.6% are white. Joint Stip. ¶ 12. The enrollment demographics of District schoolchildren are different from the population demographics of District residents at least in part because African American families and students rely more heavily on public education than do white families and students. Gordon Testimony, Trial Tr. vol. 1, 172:2-10; see also PLTF-41, Colin Gordon, Response to Report of Jonathan Rodden, June 30, 2015 (“Gordon Resp.”), ¶ 3, Fig. 1.
B. Total Population and Voting-Age Population of the District
One of the central issues raised in this case has been whether African Americans constitute a numerical majority of the voting-age population in the District and if so, whether a numerical majority status is a bar to relief under Gingles I. Resolving these questions raises another question: which datasets and" methodologies are appropriate bases for determining the makeup of the FFSD voting-age population?
*1020 Plaintiffs contend that the only-appropriate method for determining the breakdown of the FFSD voting-age population is by using the 2010 Decennial Census, which reports that the “any-part Black” 4 (“AP Black”) voting-age population in FFSD is 24,466, or 48.19% of the total voting-age population. Courts consider Decennial Census data “presumptively accurate.” See Valdespino v. Alamo Heights Indep. Sch. Dist., 168 F.3d 848 , 853-64 (5th Cir.1999). To show that there has been a substantial population shift rendering the actual count of the Decennial Census data inaccurate, the party challenging the use of the Census data must “thoroughly document[ ]” changed population figures with “a high degree of accuracy,” a showing that must be “clear, cogent and convincing.” Perez v. Pasadena Indep. Sch. Dist., 958 F.Supp. 1196, 1210 (S.D.Tex.1997) (citation omitted), aff'd, 165 F.3d 368 (5th Cir.1999)
The District contends that I should rely on the results of the 2011-2013 American Community Survey (“ACS”), which provides that the “single-race Black” voting-age population in FFSD is 24,313 +/- 1,513 (22,800 to 25,826), or 48.94% +/- 3.05% (45.89% to 51.99%) of the total voting-age population. The District also argues that I should consider the estimates of its expert, Dr. Jonathan Rodden, which are based on the Decennial Census data, ACS data, and trend lines. Under these analyses, the District contends that the current any-part Black voting-age population in FFSD is 24,994, or 51.0% of the total voting-age population.
A summary of the demographic data and data sources offered by the parties is presented below:
[[Image here]]
1. 2010 Decennial Census
a) Total population according to the 2010 Decennial Census
*1021 The U.S. Census Bureau measures the total population of each jurisdiction in the United States every ten years through the Decennial Census. According to the 2010 Decennial Census, the District has a total population of 68,663, with an any-part Black population of 36,967 (53.84%), and a white population of 29,581 (43.08%). Joint Stip. ¶ 13.
The boundaries of FFSD as recorded by the Census Bureau are slightly larger than the boundaries of FFSD as recorded by the St. Louis BOEC. For this reason, the most accurate way to count the population is by adding together the populations of the individual census blocks that comprise the District as recorded by the St. Louis BOEC. Calculating the FFSD population in this manner has a net effect of there being 387 fewer persons within the voting district than as recorded by the Census Bureau. 5 PLTF-44, Cooper Decl, ¶ 13.
b) Voting-age population according to the 2010 Decennial Census
The voting-age population (“VAP”) of a jurisdiction includes all'individuals age 18 and over. See, e.g., Gordon Testimony, Trial Tr. vol. 1, 137:1-10. The 2010 Decennial Census is the most recent “complete count” of the VAP of a jurisdiction. PLTF-45, Cooper Suppl. Decl, July 2, 2015, ¶¶ 3, 6; see also Rodden Testimony, Trial Tr. vol. 5,160:13-17.
According to the 2010 Decennial Census, the total VAP in the District is 50,771, of whom 24,466 (48.19%) are any-part Black; 24,030 (47.33%) are single-race Black; and 24,852 (48.95%) are non-Hispanic single-race white. Joint Stip. ¶ 13. There are 1,324 (1.93% of the total population) Latinos in the District, of whom 824 (1.62% of the VAP) are voting-aged. Joint Stip. ¶¶ 13,16.
In 2010, there were only 386 more Non-Hispanic white voters than AP Black voters in the District. Joint Stip. ¶14. Non-Hispanic white voters were not a majority in the District in 2010. Instead, they held a slight plurality of 0.76%. Joint Stip. ¶13. A plurality is the largest group of voters. Rodden Testimony, Vol. V, 19:2-3.
2. The American Community Survey or ‘'ACS” Data
a) Characteristics of the American Community Survey
In addition to the Decennial Census, the Census Bureau publishes one- and five-year estimates, and previously published three-year estimates, 6 of population demographics based on the American Community Survey. The ACS is a rolling sample survey based on responses from one in about forty persons on an annual basis. Joint Stip. ¶ 21; see also Cooper Testimony, Trial Tr. vol. 1, 202:13-203:4.
“A Compass for Understanding and Using American Community Survey Data,” issued by the United States Census Bureau, states, “The ACS is a nationwide, continuous survey designed to provide communities with reliable and timely demographic, housing, social and economic data every year.” PLTF-133, p.l. The Census Bureau specifically reports “age” and “race” in the ACS. Id. According to the Census Bureau, the three-year ACS data is available for populations greater than 20,000. Id. at 3. It is more precise than the one-year ACS, more current than the five-year ACS, and it has a larger sample size *1022 than the one-year ACS. Id. The District’s population was 68,663 according to the 2010 Decennial Census and 66,768 according to the 2011-2013 ACS. Joint Stip. ¶16, ¶23.
Because ACS population estimates are based on a sample, they are subject to sampling bias, ie., error margins or confidence intervals. Joint Stip. ¶ 22; Gordon Testimony, Trial Tr. vol. 1, 135:9-21; Cooper Testimony, Trial Tr. vol. 1, 202:13-203:4; see also Joint Stip. ¶ 24, 2011-2013 3-Year ACS Table B01001: Sex by Age, FFSD; PLTF-41, Gordon Besp., at 1. All ACS population estimates are published with margins of error “based on a 90 percent confidence level.” PLTF-136, ACS Office, U.S. Census Bureau, American Community Survey Multiyear Accuracy of the Data, Sept. 24, 2014, at 11. This means that a reader can be 90 percent confident that the true value lies within the confidence interval provided by the ACS. See Engstrom Testimony, Trial Tr. vol. 4, 14:12-23. ACS error margins are larger for smaller sample sizes, smaller geographic units, and smaller demographic groups within geographic areas. See Gordon Testimony, Trial Tr. vol. 1, 135:2-8, 174:19-175:20; Engstrom Testimony, Trial Tr. vol. 4, 71:1-72:6; Rodden Testimony, Trial Tr. vol. 5,164:5-20.
Statisticians routinely employ a rule of thumb that when the error margins of two population estimates overlap, there is no statistically significant difference between those two numbers. See Gordon Testimony, Trial Tr. vol. 1, 135:22-136:2 (Gordon testifying that “if you’re comparing two numbers and the confidence intervals overlap, then there’s no statistical significance between those two numbers. There’s no basis for claiming that one number is bigger than the other, because the supposed difference between the numbers is completely swallowed by the margin of error as reported by ACS”); Rodden Testimony, Trial Tr. vol. 5, 6:17-7:6.
The Census Bureau cautions that ACS data are estimates and recommends that users turn to other Census products for population counts. It includes a similar warning on the top of every ACS population table it provides. See PLTF-132, U.S. Census Bureau, Comparing ACS Data, July 14, 2015 (“Use numbers from the 2010 Census to obtain counts of the population and their basic characteristics (sex, age, race, Hispanic origin, and homeowner status)”); Joint Stip. ¶24, 2011-2013 3-Year ACS Table B01001: Sex by Age, FFSD; see also Gordon Testimony, Trial Tr. vol. 1,157:17-20 (“I would not draw conclusions about voting-age population based on the ACS because it’s nqt designed for that purpose, for a threshold of the population, and because the margins of error are too wide”), id. at 133:24-134:21; Cooper Testimony, Trial Tr. vol. 1 203:9-17.
b) VAP demographics in FFSD as estimated by the ACS
The 2011-2013 ACS estimates for Sex by Age by Veteran Status for the Civilian Population 18 Years and Over in FFSD are as follows:
*1023 [[Image here]]
Source: U.S. Census Bureau, 2011-2013
American Community Survey
PLTF-44, Cooper Decl., at Ex. D, p. 29 of 46.
According to the 2011-2013 three-year ACS, the VAP in the District is 49,679. Joint Stip. ¶ 28. The 2011-2013 ACS estimates that the single-race white VAP (including Hispanic whites) is 23,740, or 47.24% of the District’s VAP, and the non-Hispanic white VAP is 23,242, or 46.78% of the District’s VAP. Joint Stip. ¶25. According to the 2011-2013 ACS, there are an estimated 1,316 (1.97%) single-race individuals in FFSD who identify as neither Black nor white, 86 of whom identify as American Indian or Alaska Native; 644 of whom identify as Asian; and 686 of whom identify as “some other race.” See Joint Stip. ¶ 23 (table); PLTF-30, 2011-2013 3-Year ACS Table C02003: Detailed Race, FFSD; PLTF-41, Gordon Resp., at 2. The 2011-2013 ACS estimates that the FFSD single-race Black VAP is 24,313, or 48.94% of the District’s VAP. Joint Stip. ¶ 25. The margin of error for the 2011-2013 ACS estimate for the single-race Black VAP is ±1,513, meaning that the 2011-2013 ACS estimates that the single-race Black VAP could be as low as 22,800 or as high as 25,826. PLTF-44, Cooper Decl., at Ex. D, p. 29 of 46.
Unlike the Decennial Census, the ACS does not publish estimates for the “AP Black” category or estimate the margin of error associated with the AP Black fraction of the VAP. Joint Stip. ¶¶ 25, 26; see also Gordon Testimony, Trial Tr. vol. 1, 136:5-12. As a result, the 2011-2013 three-year ACS estimates do not establish to a degree of statistical significance that the single-race Black VAP has grown since the 2010 Census. Rodden Testimony, Trial Tr. vol. 5,177:1-21.
The 2010 Census totals for single-race Black VAP is 24,030, which falls well within the margin of error for the 2011-2013 ACS estimate for the single-race Black VAP of 22,800 to 25,826. In other words, according to the 2011-2013 ACS estimate, *1024 the single-race Black YAP could be as small as 22,800, which is less than the 2010 Census count.
Comparing population sizes as estimated by the 2011-2013 ACS alone, the confidence intervals for the single-race Black VAP and the non-Hispanic white YAP overlap. The 2011-2013 ACS survey estimates state with 90% certainty that the single-race BVAP is between 22,800 and 25,826 and that the single-race non-Hispanic white VAP is between 21,829 and 24,655. PLTF-44, Cooper Deck, at Ex. D, p. 29 of 46. Because these confidence intervals overlap, the 2011-2013 ACS estimates do not establish to a degree of statistical significance that the single-race Black VAP is greater than the non-Hispanic white VAP within FFSD. See PLTF-41, Gordon Resp., at 1-2; see Deft-FFSD C, Supplemental Report of Jonathan Rodden & Jowei Chen: Assessment of Plaintiffs’ Redistricting Proposals, July 2, 2015 (“Rodden Redistricting Rep.”), ¶7; see also generally Gordon Testimony, Trial Tr. vol. 1,134:22-136:12.
Additionally, the Black and white VAP estimates from the ACS are independent of one another. As a result, it is not appropriate to assume that any sampling bias in the two numbers “will work in the same direction.” Gordon Testimony, Trial Tr. vol. 1,175:15-20.
3. The District’s population estimates
a) Calculating an AP Black VAP figure from the 2011-2013 ACS estimates
Although the 2011-2013 ACS estimates do not include an estimate of the AP Black VAP, the District’s expert, Dr. Jonathan Rodden, attempted to extrapolate that figure, which he estimated as 51.0% of the FFSD VAP. Deft-FFSD C, Rodden Redistricting Rep., Tbl. 1 (p. 3). Dr. Rodden came to this figure by dividing the number of individuals who identified as two or more races with no-part African American (632) by those who identified as two or more races with some-part African American (1,564), to determine the ratio of individuals in the 2011-2013 ACS that identify as any-part African American. Then Dr. Rodden applied the same ratio for individuals that identify as any-part African American in the total population to the number of voting-age individuals identified in the 2011-2013 ACS.
Dr. Rodden calculated that the total number of voters that identify as any-part Black was 24,994, which according to Dr. Rodden is 51% of the District’s voters. Rodden Testimony, Vol. V, 15:23-16:16; DEF Ex. C, Table 1. However, Dr. Rod-den did not use the total VAP of the FFSD as estimated by the 2011-2013 ACS, which leads him to inflate his calculation of the percentage of the VAP in FFSD that is Black. The 2011-2013 ACS estimates the total VAP for the District as 49,679, see Joint Stip. ¶ 28, but because Dr. Rodden excludes persons reporting two or more races, he underreports the total VAP of the District as 49,009. See Deft-FFSD C, Rodden Redistricting Rep., ¶¶ 5-6, Tbl. 1 (p. 3). When the correct denominator for the total VAP of the District is used (49,-679, as reported by the ACS), the single-race Black VAP is 48.9% (24,313 / 49,679) of the total VAP and the any-part Black VAP (as estimated by Dr. Rodden) should be 50.3% (24,994 / 49,679) of the total VAP. See id.; Joint Stip. ¶ 28.
b) Demographic trends in the Ferguson-Florissant School District
Dr. Rodden also created linear projections from the three-year ACS estimates to demonstrate a trend of steady growth in the African American population in FFSD since 2000. Rodden Report, DEF-A, Figure 3. To project this trend, Dr. Rodden plotted the average population estimates from the midpoint year of four sets of ACS three-year estimates, 2008-2010, 2009-2011, 2010-2012, and 2011-2013, and extended that line into 2015. DEF-A, Rodden Rep., *1025 ¶¶ 12-13, Fig. 3 (p. 6); see also Rodden Testimony, Trial Tr. vol. 5,160:2-7.
Dr. Rodden attributes most of the difference in demographics to a decline in the white population. See DEF-A, Figure 3; Rodden Testimony, Trial Tr. vol. 5, 8:15-9:9. Plaintiffs’ expert Dr. Gordon corroborates that trend, stating, “Rodden and I agree substantially on the basic demographic trends. The' district is in the midst of an ongoing racial transition marked by white flight to the outer suburbs. Since 2000, the white population of the Ferguson-Florissant School District (FFSD) has fallen by almost half (50,000 to 28,160) while the black population has grown steadily. And, there is a stark disparity in population by age, with the share of the African-American population much higher among school-age residents.” Gordon Rebuttal Report, PLTF-41, p. 1; Rodden Testimony, Trial Tr. vol. 5,11:4-12.
Dr. Rodden testified that trends in the voting-age population in the District mirror the District’s overall population trends. Rodden Testimony, Trial Tr. vol. 4, 213:18-23. Dr. Rodden testified that the trend is due in part to the demographic differences between African Americans and whites. The white population is considerably older than the African American population, as African American families tend to be younger. In the 2010 Decennial Census, there was a large number of African Americans on the cusp of turning 18, whereas there were far fewer in that category for whites. Rodden Testimony, Vol. IV, 214: 1-17. Dr. Gordon corroborates that finding in his report. Gordon Supplemental Report, PLTF-41, Figure 1: Enrollment in FFSD by Race, 1991-2014.
Based on his trend analysis, Dr. Rodden concluded that African American voters, who were at near parity with white voters at the time of the'2010 Census,' are now a majority of the District’s VAP. Rodden Testimony, Trial Tr. vol. 5, 8:3-13:17; Rod-den Testimony, Trial Tr. vol. 4, 214:25-215:12; DEF-X, “Share of Voting-Age Population.”
k- Impediments to voting faced by African Americans in FFSD
The population that can vote in any given election is a subset of the VAP because the VAP includes some individuals who are old enough to vote, but who cannot vote for some other reason. See, e.g., Gordon Testimony, Trial Tr. vol. 1, 136:23-141:13; PLTF-41, Gordon Resp., at 2. Impediments that may prevent or hinder particular groups from exercising their voting rights include felony disenfranchisement, the lack of homeownership, and other socioeconomic indicators that are barriers to registration.
a) Felony disenfranchisement
In Missouri, individuals who are part of the VAP but who are imprisoned, or who are on probation or parole as part of their sentence for commission of a felony, are not eligible to vote. Gordon Testimony, Trial Tr. vol. 1, 136:23-141:13; PLTF-41, Gordon Resp., at 2; see also RSMo. §§ 115.133, 115.195; 561.026; Mo. Const, art. 8, § 2. About one-third of the approximately 106,000 Missourians who have lost the right to vote are serving a sentence in a state prison. Gordon Testimony, Trial Tr. vol. ’ 1, 163:21-164:2. The remaining more than two-thirds, or 75,000, are held in local jail, or have returned to their home communities on probation or parole. Id.
In Missouri, the rate of felony disenfranchisement for the population as a whole is' 2.32% of VAP. Gordon Testimony, Trial Tr. vol. 1, 139:10-17; PLTF-41, Gordon Resp., at 2. The rate of felony disenfranchisement for the African American population is almost triple the overall statewide rate, at 6.88%. Gordon Testimony, Trial Tr. vol. 1, 139:10-17; PLTF-41, Gordon Resp., at 2.
Dr. Colin Gordon, a historian and tenured professor, testified on behalf of Plain *1026 tiffs on the effect of felony disenfranchisement in FFSD. Dr. Gordon is nationally recognized as an expert in urban history, specifically in the history of development, decline, residential patterns, and segregation in the St. Louis metropolitan area. See Trial Tr. vol. 1, 87:4-94:23; PLTF-40, Colin Gordon, Segregation and Uneven Development in Greater St. Louis, St. Louis County, and the Ferguson-Florissant School District, May 26, 2016 (“Gordon Rep.”), at 62 (Gordon CV); PLTF-42, Colin Gordon, Mapping Decline: St. Louis and the Fate of the American City (U. of Penn. Press 2008).
Dr. Gordon testified that, although there is no reported data specific to FFSD showing felony disenfranchisement rates by race, one can reasonably conclude that, in FFSD, African Americans are disenfranchised due to a criminal conviction at higher rates than whites. See Gordon Testimony, Trial Tr. vol. 1, 138:3-4, 166:1-6. Dr. Gordon noted the large racial disparities statewide in felony disenfranchisement rates, and also noted that felony disenfranchisement rates are generally higher in urban áreas than elsewhere. Trial Tr. vol. 1,138:5-19. Based on these facts, and “given what we know about patterns of policing in North County in the wake of the Department of Justice report after Ferguson,” Dr. Gordon opined that “an application of the [statewide] African-American rate [of disenfranchisement] to voting-eligible African Americans in Ferguson-Flor-issant would be a conservative estimate of the rate of disenfranchisement [among African Americans there].” Id.
.Although Dr. Gordon did not provide District-specific felony disenfranchisement testimony, I find Dr. Gordon’s testimony regarding the applicability of the statewide disenfranchisement rate to the District credible and I find that African Americans in FFSD are disproportionately affected by felony disenfranchisement as compared to non-Hispanic whites. However, because there are no prisons in the District, it would not be appropriate to conclude that the same percent of African Americans who are disenfranchised statewide applies equally to the District. Dr. Gordon testified that approximately one-third of the disenfranchised population is located in prisons. As a result, only two-thirds of Dr. Gordon’s estimated rate of 6.88% of African American felony disenfranchisement should be applied to the FFSD population, a rate of approximately 4.59%.
b) Voter registration
In Missouri, individuals who are part of the VAP but who fail to register by the operative election registration deadline— which is 27 days before a given election— cannot vote in that election. RSMo. §§ 115.135, 115.139; Mo. Const. art. 8, § 2; Marre v. Reed, 775 S.W.2d 951, 955-56 (Mo. banc 1989). Additionally, individuals who are part of the VAP and who are registered to vote, but who then move from one jurisdiction to another before an election registration deadline, cannot vote in that election unless they re-register at their new address before that deadline. See RSMo. §§ 115.275, 115.277, 115.135; Mo. Const. art. 8, § 2. Other restrictions may occur, for example, to individuals who are part of the VAP but who move within a jurisdiction. These individuals can change their address on Election Day, but only at their new polling place or the central polling location of the county, and not at their old polling place. RSMo. §§ 115.135.1-3, 115.165. Additionally, individuals who move from one jurisdiction to another after an election registration deadline cannot vote for local offices in that election. RSMo. §§ 115.275.4, 115.277.4.
There is undisputed evidence from the Census Bureau that, single-race African Americans in Missouri register to vote at a lower rate (67.1%) than non-Hispanic *1027 whites (72.2%). Kimball Testimony, Trial Tr. vol. 2, 119:14-18; 7 Trial Tr. vol. 3, 41:6-10 (same); see also PLTF-63, Reported Voting and Registration, by Sex, Race, and Hispanic Origin, for States (Nov. 20H) (Ex. 5 to Rodden Dep., Aug. 20, 2015), at 7; Rodden Testimony, Trial Tr. vol. 5, 71:1-4,147:12-16.
Plaintiffs, however, did not produce any statistical evidence showing the voter registration rates of African American and white voters within the District. Rodden Testimony, .Trial Tr. vol. 5, 70:24-71:4; Kimball Testimony, Trial Tr. vol. 2, 185:6-9; Gordon Testimony, Trial Tr. vol. 1, 173:16-24. Because Plaintiffs did not provide any such evidence, the District argues that Plaintiffs have failed to meet their burden in establishing a statistically significant difference in voter registration rates in the District.
Moreover, according to the District, the evidence shows that there is little to no difference in registration rates between African Americans and whites in FFSD. To support this argument, the District’s expert, Dr. Rodden, analyzed voter turnout in the District. Dr. Rodden examined voter turnout by race using two methods: Ecological Inference and scatterplots for the 2012-2015 elections. Rodden Report, DEF-A, Figure 5, 6. Based on these analy-ses, Dr. Rodden concluded that there was “not much of a relationship or any relationship here between turnout and African-American share,” particularly in 2012, 2013, and 2014. Rodden Testimony, Trial Tr. vol. 5, 77:11-78:3. Rodden did note that there was evidence of á slight difference in turnout in 2011 and a significant difference in turnout in 2015. Id. at 77:18-78:3.
Turnout, however, is different from registration, Dr. Rodden’s calculations of turnout merely measure turnout rates of already registered voters. Because Dr. Rodden did not first examine or determine voter registration rates by race, his conclusions regarding turnout merely suggest, at best, that turnout to the polls is substantially similar within the registered voter population. Rodden Testimony, Trial Tr. vol. 5, 146:23-25. As a result, Dr. Rodden’s analysis does not inform the inquiry into whether initial registration rates are as disparate in FFSD as they are throughout Missouri, nor does it overcome the Plaintiffs evidence that some difference in registration rates exists.
I find that there is sufficient evidence in the record to conclude that African Americans in FFSD are likely registered to vote at a lower rate than white residents. Neither party produced statistical evidence regarding the voter-registration gap in FFSD in particular. However, FFSD experiences substantial racial disparities along a range of socioeconomic factors. See, e.g„ infra Section II.B.4(c). These disparities tend to correlate with lower registration rates, a proposition with which the District’s own expert concurred. See Rod-den Testimony, Trial Tr. vol. 5, 117:14-118:1. Additionally, there is no credible evidence to suggest that the statewide gap disappears in FFSD. As a result, it is reasonable to conclude that the statewide disparities in registration rates are similar to those in the District.
Accordingly, based on the credible testimony of Plaintiffs’ experts regarding state *1028 wide disparities in registration rates and the existence of other racial disparities in FFSD that are likely to impact voter registration rates, I find that African Americans in FFSD register to vote at a rate lower than non-Hispanic whites.
c) Homeownership
Homeownership is a strong predictor of voting in local elections. Kimball Testimony, Trial Tr. vol. 2, 147:17-24. Dr. Gordon testified that individuals who do not own their homes experience residential instability and move more frequently than homeowners, which impacts the ability to efficiently and effectively register and vote. See Trial Tr. vol. 1, 173:7-14; 130:21-313:1; see also Hudson Testimony, Trial Tr. vol. 4, 90:10-91:11; supra at Section II.B.4(b).
Within FFSD, African Americans own their homes at a significantly lower rate than whites. Gordon Testimony, Trial Tr. vol. 1, 173:7-11. In FFSD, the homeowner-ship rate among African American residents is 50.7%, which is more than 30 percentage points lower than the home-ownership rate among whites (82.8%). Id. at 174:7-9; 176:11-13; see also PLTF-45, Cooper Suppl. Decl., ¶ 17 (reporting that the FFSD African American homeowner-ship rate as estimated by ACS has dropped from 61% to 50.7% since the 2005-2007 ACS).
Based on thé credible testimony of Plaintiffs’ experts, I find that the lower rate of homeownership among African Americans in FFSD, as compared to white residents, contributes to the depressed levels of voter registration in the African American community.
DISCUSSION
III. GINGLES I
A. Gingles I Legal Standards
To satisfy the requirements of the Gingles I precondition, Plaintiffs must show that African Americans in the District are “sufficiently large and geographically compact to constitute a majority in a single-member district.” Gingles, 478 U.S. at 50 , 106 S.Ct. 2752 . This straightforward threshold requirement is satisfied by the creation of an illustrative plan containing a single-member district in which Black voters constitute a bare majority of the VAP. See Bartlett v. Strickland, 556 U.S. 1, 18 , 129 S.Ct. 1231 , 173 L.Ed.2d 173 (2009) (stating that Gingles I asks a simple threshold question: whether the protected group of voters can make up “more' than 50 percent of the voting-age population in the relevant geographic area?”). “[T]he Supreme Court [at this stage] requires only a simple majority of eligible voters in the single-member district. The court may consider, at the remedial stage, what type of remedy is possible.... But this difficulty should not impede the judge at the liability stage of the proceedings.” Bone Shirt v. Hazeltine, 461 F.3d 1011 , 1019 (8th Cir. 2006) {“Bone Shirt II”) (alteration in original) (quoting Dickinson v. Ind. State Election Bd., 933 F.2d 497 , 503 (7th Cir.1991)).
The plan must also (1) satisfy the one person, one vote constitutional requirements, ie., approximate population equality across all districts in the plan, see Abrams v. Johnson, 521 U.S. 74, 98 , 117 S.Ct. 1925 , 138 L.Ed.2d 285 (1997); and (2) be composed of geographically compact districts that comply with traditional redistricting principles, including: contiguity; minimizing the splits of counties, municipalities, and precincts; recognizing communities of interest; and avoiding multimem-ber districts, see id. at 92 , 117 S.Ct. 1925 (“[Section] 2 compactness inquiry should take into account ‘traditional districting principles such as maintaining communities of interest and traditional boundaries.’”) (citation omitted);Bone Shirt II, 461 F.3d at 1019, and state and county *1029 districting requirements, see Voinovich v. Quilter, 507 U.S. 146, 156 , 113 S.Ct. 1149 , 122 L.Ed.2d 500 (1993) (in area of voting and apportionment, “federal courts are bound to respect the States’... choices unless those choices contravene federal requirements”).
B. Gingles I Findings of Fact
Mr. William Cooper, Plaintiffs’ expert demographer, testified at trial. Since 1986, Mr. Cooper has prepared redistricting plans for approximately 700 jurisdictions. PLTF-44, Cooper Decl., at Ex. A; see Cooper Testimony, Trial Tr. vol. 1, 179:23-180:3 (testifying that he has drawn statewide plans in approximately forty states, but his main focus is on local jurisdictions). Since 2011, he has prepared more than 150 redistricting plans for local jurisdictions using the 2010 Decennial Census. PLTF-44, Cooper Decl., at Ex. A.
Mr. Cooper’s election plans have been precleared and/or adopted in state or local jurisdictions on at least ten occasions. Id. Illustrative or proposed plans created by Mr. Cooper have also been ordered adopted as Section 2 remedies by the District of South Dakota in Bone Shirt v. Hazeltine, 387 F.Supp.2d 1035 (D.S.D. 2005), aff'd, 461 F.3d 1011 (8th Cir.2006), and by the Eastern District of Washington in Montes v. City of Yakima, No. 12-CV-3108 TOR, 2015 WL 11120964 (E.D.Wash. Feb. 17, 2015). Over the course of his career, Mr. Cooper has testified at trial as an expert witness on redistricting and demographics in 36 federal voting rights cases and been deposed or entered declarations in an additional 33 voting rights cases. PLTF-44, Cooper Decl., at Ex. A. He has never been deemed unqualified to provide expert testimony. Trial Tr. vol. 1, 181:3-5.
Mr. Cooper drew two illustrative seven-district plans on behalf of Plaintiffs to show that it is possible to create at least one subdistrict in which African Americans are a majority of the VAP. Joint Stip. ¶ 36; PLTF-44, Cooper Decl., ¶¶ 38-57. In each of the two illustrative plans drawn by Mr. Cooper, there are four districts in which Blacks are a majority of the VAP. PLTF-44, Cooper Decl., ¶¶ 38-57. Mr. Cooper used generally accepted demographic methods for creating the two illustrative plans. See Joint Stip. ¶¶ 37, 39, 40, 44, 45, 47; PLTF-44, Cooper Decl., ¶¶38, 39-42, 43-46, 51, Fig. 10 (p. 23), 55, Fig. 12 (p. 26); Cooper Testimony, Trial Tr. vol. 1, 181:16-185:8,188:9-12.
For each of Plaintiffs’ illustrative plans, all parts (i.e., census blocks) of each component election district are connected at some point with the rest of the district, so each component district is contiguous. Joint Stip. ¶49; PLTF-44, Cooper Decl., ¶ 60; Cooper Testimony, Trial Tr. vol. 1, 182:3-4; Gordon Testimony, Tr. vol. 1, 127:8-10.
The “compactness” of a district can be measured using the Reock test, which compares the shape of a district to a circle, considered to be the most compact shape possible. The illustrative plans’ Reock scores show the component election districts are compact. PLTF-44, Cooper Decl., ¶¶ 58 n.11, 59; Cooper Testimony, Trial Tr. vol. 1, 195:7-20.
In Plaintiffs’ Illustrative Plan 1, three of the 96 precincts in the District are split by district boundaries with a few additional precincts split by 2010 Census blocks, but no census blocks are split by district boundaries. Public schools are balanced across all seven districts with each district having two to five schools. Based on the addresses, at the time of election, of the individuals who were Board members at the time of trial, PLTF-44, Cooper Decl., at Ex. E-7, Ex. F-7 (addresses), three incumbents are paired in District 2 and two in District 7. Joint Stip. ¶ 46; PLTF- *1030 44, Cooper Decl., ¶¶ 51-53; Cooper Testimony, Trial Tr. vol. 1, 193:13-14, 195:2-6, 199:3-6, 199:23-200:5.
In Plaintiffs’ Illustrative Plan 2, nine of the 96 precincts in the District are split by district boundaries with a few additional precincts split by 2010 Census blocks, but no census blocks are split by district boundaries. Public schools are balanced across all seven districts with each district having two to four schools. Based on the addresses, at the time of election, of the individuals who were Board members at the time of trial, PLTF-44, Cooper Decl., at Ex. E-7, Ex. F-7 (addresses), two incumbents who reside in District 2 are paired, but no other incumbents are paired. Joint Stip. ¶ 48; PLTF-44, Cooper Decl., ¶¶ 55-57. Because two incumbents live within the same Census block, there is no way to draw single-member districts that will separate those incumbents while also complying with traditional redistricting principles. See Cooper Testimony, Trial Tr. vol. 1, 197:4-11.
Plaintiffs’ illustrative plans use the 2010 Decennial Census total population for the apportionment base. See Joint Stip. ¶ 43; PLTF-44, Cooper Decl. ¶ 40. Mr. Cooper’s testimony that illustrative plans must be drawn from the Decennial Census data, rather than the ACS estimates, was credible because it is undisputed that only the Decennial Census data goes down to the census block level. Because the ACS does not report census block-level data, it does not allow a demographer to avoid splitting census blocks, making it is impossible to measure precisely the population in a given voting district. See Trial Tr. vol. 1, 200:2-5, 203:5-204:17; see also RSMo. § 1.100.1; Trial Tr. vol. 1, 192:17-194:25.
The addresses of the individuals who were members of the Board at the time of trial are all within the municipalities of Ferguson and Florissant, which together make up about two-thirds of the District spatially. See PLTF-44, Cooper Decl, at Ex. E-7, Ex. F-7; Cooper Testimony, Trial Tr. vol. 1, 197:21-198:3. It is undisputed that no current Board member resided in Berkeley, Kinloch, Cool Valley, or Normandy, which are more predominantly African American areas of the District. See Cooper Testimony, Trial Tr. vol. 1,198:1-7.
C. Gingles I Conclusions of Law
I conclude that Plaintiffs have established Gingles I, i.e., that the African American population in the District is “sufficiently large and geographically compact to constitute a majority in a single-member district.” Gingles, 478 U.S. at 50 , 106 S.Ct. 2752 . Plaintiffs’ expert demographer, William Cooper, presented two illustrative plans in which Black voters constitute a majority of the VAP in not just one but four of seven districts.
As described above, Mr. Cooper is a recognized expert demographer in the context of § 2 analysis. For the very first time at trial, the District objected to Mr. Cooper being an expert. I overrule this objection and conclude that Mr. Cooper qualifies as an expert demographer and Plaintiffs have established by a preponderance of the evidence that his testimony is admissible. See Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 , 592 & n. 10, 113 S.Ct. 2786 , 125 L.Ed.2d 469 (citing Rule 104(a)) (1993); Fed. R. Evid. 702; see also Pis.’ Post-Trial Br. at 2-4. I also find that Mr. Cooper’s testimony is reliable, and that Mr. Cooper testified credibly to the data and methodology used to develop Plaintiffs’ two illustrative plans. As a result, Plaintiffs have established by a preponderance of the evidence that his testimony is admissible. Daubert, 509 U.S. at 592 & n. 10, 113 S.Ct. 2786 (citing Rule 104(a)).
*1031
1. Plaintiffs’ illustrative plans comply ■ with constitutional requirements
Mr. Cooper’s illustrative plans comply with one-person, one-vote constitutional requirements because they properly apportion the District’s total population across the districts with minor deviation, ie., total population deviations of less than 10%. Evenwel v. Abbott, — U.S. -, 136 S.Ct. 1120, 1124 , 194 L.Ed.2d 291 (2016); White v. Regester, 412 U.S. 755, 764 , 93 S.Ct. 2332 , 37 L.Ed.2d 314 (1973) (holding that plans with total deviation of less than 10% presumptively comply with one person, one vote). The plans also comply with traditional redistricting principles including contiguity, compactness, appropriate population-size deviations, keeping census blocks together, and incumbency protection. The District did not present any evidence to contradict either that these plans include majority-Black single-member districts or that the plans comply with constitutional requirements and redistricting principles.
2. Gingles I does not require an effectiveness analysis
The District argues that Plaintiffs’ illustrative plans do not provide an effective remedy. In fact, the District contends that the implementation of either of Plaintiffs’ illustrative single-member district plans will impede, rather than improve, African American voters’ ability to effectively elect candidates of their choice in FFSD Board elections. 8
Gingles I, however, only asks a simple threshold question: whether the protected group of voters can make up “more than 50 percent of the voting-age population in the relevant geographic area?” Bartlett v. Strickland, 556 U.S. 1, 18 , 129 S.Ct. 1231 , 173 L.Ed.2d 173 (2009). “[T]he Supreme Court [at this stage] requires only a simple majority of eligible voters in the single-member district. The court may consider, at the remedial stage, what type of remedy is possible... .But this difficulty should not impede the judge at the liability stage of the proceedings.” Bone Shirt v. Hazeltine, 461 F.3d 1011 , 1019 (8th Cir.2006) (“Bone Shirt IF) (alteration in original) (internal quotation omitted). As a result, the District’s predictions about the ultimate effectiveness of Plaintiffs’ illustrative single-member district plans is irrelevant to the Gingles I threshold determination.
S. Plaintiffs’ claim is not barred because of the size of the African American ■ voting-age population
The District also argues that Plaintiffs have not satisfied Gingles I because African Americans purportedly constitute a majority of the VAP in FFSD, and therefore already possess an opportunity to elect their preferred candidates under the existing at-large system. This argument fails for several reasons.
As an initial matter, this argument is irrelevant because Gingles I requires no additional analysis of the demographics of the District or the illustrative plan districts beyond a showing that African Americans in the District are “sufficiently large and geographically compact to constitute a majority in a single-member *1032 district.” Gingles, 478 U.S. at 50 , 106 S.Ct. 2752 .
Moreover, even if this argument were relevant to the Gingles I inquiry, it is unpersuasive for two reasons. First, the District has not provided reliable data sufficient to overcome the presumptive validity of the 2010 Decennial Census, which indicates that African Americans are a minority of the FFSD VAP and outnumbered by whites. See Valdespino v. Alamo Heights Indep. Sch. Dist., 168 F.3d 848 , 853-54 (5th Cir.1999); United States v. Vill. of Port Chester, 704 F.Supp.2d 411, 439 (S.D.N.Y.2010). Second, even if Black residents were to constitute a bare majority of the VAP in FFSD, vote dilution claims may go forward where, as here, the racial minority group in question has suffered a history of discrimination that inhibits their participation in the political process, such that they are disadvantaged by an at-large electoral arrangement. See Pope, 687 F.3d at 575 n. 8; Kingman Park Civic Ass’n v. Williams, 348 F.3d 1033, 1041 (D.C.Cir.2003); Meek v. Metro. Dade Cty., 908 F.2d 1540, 1545-46 (11th Cir.1990);Mowroe v. City of Woodville, 881 F.2d 1327, 1332-33 (5th Cir.1989).
a) African Americans do not currently constitute a majority of FFSD’s VAP
The Decennial Census population data is the appropriate metric by which to determine the demographics of FFSD’s VAP. The Decennial Census is a full count of the nation’s entire population every ten years and provides an accurate and complete count of FFSD’s population and demographics. See Evenwel v. Abbott, — U.S. -, 136 S.Ct. 1120, 1140-41 , 194 L.Ed.2d 291 (2016) (Alito, J., concurring in judgment). According to the 2010 Census, African Americans are neither a majority nor a plurality of the total VAP of the District.
Courts consider Decennial Census data “presumptively accurate.” See Valdespino v. Alamo Heights Indep. Sch. Dist., 168 F.3d 848 , 853-54 (5th Cir.1999); United States v. Vill. of Port Chester, 704 F.Supp.2d 411, 439 (S.D.N.Y.2010). As a result, courts resolving Voting Rights Act claims in the Eighth Circuit regularly rely on Decennial Census data for determining the demographics of a jurisdiction. 9 See, e.g., Harvell v. Blytheville Sch. Dist. No. 5, 71 F.3d 1382 ,1385 n. 1 (8th Cir.1995); Clay v. Bd. of Educ. of St. Louis, 90 F.3d 1357, 1359 (8th Cir.1996) (“Clay II”); African Am. Voting Rights Legal Def. Fund, Inc. v. Villa, 54 F.3d 1345 , 1347-48 (8th Cir. 1995); Jeffers v. Beebe, 895 F.Supp.2d 920, 925-26 (E.D.Ark.2012); Bone Shirt v. Hazeltine, 336 F.Supp,2d 976, 983 (D.S.D. 2004) (“Bone Shirt I”).
Courts presume the continued accuracy of the most recent Decennial Census figures absent a party meeting the burden of proving otherwise. See McNeil v. Springfield Park Dist., 851 F.2d 937, 946 (7th Cir.1988). To show that there has been a substantial population shift rendering the actual count of the Decennial Census data inaccurate, the party challenging the use of the Census data must “thoroughly docu-mente]” changed population figures with “a high degree of accuracy,” a showing that must be “clear, cogent and convincing.” Perez v. Pasadena Indep. Sch. Dist., 958 F.Supp. 1196, 1210 (S.D.Tex.1997) (citation omitted), aff'd, 165 F.3d 368 (5th Cir.1999); Vill. of Port Chester, 704 F.Supp.2d at 439 ; see also Kirkpatrick v. *1033 Preisler, 394 U.S. 526, 535 , 89 S.Ct. 1225 , 22 L.Ed.2d 519 (1969).
The District presents three sets of data that it argues should be used to determine the Black YAP: (1) 2011-2013 3-year American Community Survey (“ACS”) population estimates, (2) Dr. Rodderis estimates of the “any part Black” VAP, and (3) Dr. Rodderis population projections based on his trend analysis. As explained below, none of these data sets are sufficient to overcome the presumptive accuracy of the 2010 Decennial Census count, which states that Blacks are a minority of the VAP in the District.
i. The 2011-2013 ACS estimates
The ACS publishes no data that overcomes the presumptive validity of the 2010 Decennial Census’s count of the Black VAP in the District. As an initial matter, there is no information published by the ACS that, on its face, indicates that African Americans are a majority of the VAP in the District. Although the 2011-2013 ACS population estimates for the District suggest that single-race African Americans may now be a plurality of the VAP in the District at 48.9%, they do not show that single-race African Americans are a majority of the VAP. And the ACS reports no estimate for the percentage of the VAP in the FFSD that is “any part” Black.
But even considering the ACS estimates of the single-race Black VAP in the District, it is crucial to recognize that the margins of error in the 2011-2013 ACS for the District’s VAP are too wide to establish that the demographics of the FFSD population have substantially changed since 2010. The ACS provides a population estimate with only a sufficient degree of certainty to say that the actual number is within a particular range. Because ACS population estimates are based on a sample, and not a complete count like the Decennial Census, they are subject to sampling bias, la, error margins or confidence intervals. The Census Bureau itself cautions against using ACS estimates rather than the Decennial Census complete count to determine the population of a given geographic area. 10
Here, the 2011-2013 ACS estimates fail to establish with statistical confidence that the single-race Black VAP has grown larger than the white VAP in the District. The 2011-2013 ACS survey estimates state with 90% certainty that the single-race BVAP is between 22,800 and 25,826 and that the single-race non-Hispanic white VAP is between 21,829 and 24,655. As estimated by the 2011-2013 ACS, the margins of error for the single-race BVAP and white VAP overlap, which means that the difference in the two population figures is not statistically significant. See Gordon Testimony, Trial Tr. vol. 1,135:22-136:2.
In fact, given the margins of error, the 2011-2013 ACS estimates do not establish that the Black VAP of the District has substantially changed since 2010. The 2011-2013 ACS survey estimates state with 90% certainty that the single-race Black VAP is between 22,800 and 25,826, 11 a margin of error that includes the Decennial Census single-race Black VAP count of 24,030. Put another way, according to the *1034 ACS confidence interval, the single-race Black VAP could be as small as 22,800, which is actually less than the 2010 Census estimate of 24,080. As a result, given the large margin of error, the most recent three-year ACS estimate of the District’s single-race Black VAP is not statistically distinguishable from the single-race Black VAP reported by the Decennial Census. See Rodden Testimony, Trial Tr. vol. 5, 177:5-21. And the ACS estimated change in the single-race non-Hispanic white VAP is not substantial. 12
Accordingly, because the error margins in the 2011-2013 ACS are too large to constitute statistically significant evidence that the Black VAP outnumbers the white VAP in the District, or that the demographics in the District have substantially changed since 2010, the District has failed to meet its burden of proving with “clear, cogent and convincing” evidence, and to a “high degree of accuracy,” that there has been a substantial population shift rendering the actual count of the 2010 Decennial Census data inaccurate such that I should reject the 2010 Decennial Census’s count of African Americans as less than a majority of the District’s VAP. Perez v. Pasadena Indep. Sch. Dist., 958 F.Supp. 1196, 1210 (S.D.Tex.1997) (citation omitted), aff'd, 165 F.3d 368 (5th Cir.1999).
ii. The District’s population estimates of the any-part Black VAP
As noted above, the ACS reports that, at most, single-race African Americans are estimated to be a plurality but not a majority of the District’s VAP. Accordingly, there are no published government data stating that African Americans — including individuals who are two or more races and “some part” Black — are the majority of the VAP. The only figures that even purport to show that African Americans are a majority of the District’s VAP are Dr. Rodden’s estimates of the number of voting-age individuals in FFSD who are “any-part” Black. See Rodden Testimony, Trial Tr. vol. 5, 174:9-15. These estimates, however, suffer from three limitations that diminish their reliability, leading me to find that these estimates are not sufficient to overcome the presumptive accuracy of the 2010 Decennial Census.
First, Dr. Rodden’s calculations are based on ACS data, which, for the reasons discussed above, cannot overcome the Decennial Census’s presumptively accurate complete count.
Second, in his estimate of the number of multi-race individuals in FFSD who are any-part Black and 18 years of age or older, Dr. Rodden cannot report any error margins, Rodden Testimony, Trial Tr. vol. 5, 170:2-8, which renders this estimate inconsistent with generally accepted professional standards in political science. When making estimates, the smaller the survey sample, the larger the error margin. Here, the ACS survey samples only about 2% of the total population. See Gordon Testimony, Trial Tr. vol. 1, 133:18-134:5. Dr. Rod-den’s attempt to produce an estimate of the any-part Black VAP in FFSD entailed extrapolating from a tiny slice of that 2% sample (ie., mixed-race individuals who are part-Black and live within FFSD). See Gordon Testimony, Trial Tr. vol. 1, 136:5-12. Although Dr. Rodden described his calculations here as relatively simple, the expert demographers at the Census Bureau declined to publish estimates for the category of any-part Black for the VAP of *1035 FFSD in the 2011-2013 three-year ACS. See id., 174:9-15. Notably, other courts have declined to use sources of population data that do not contain the relevant VAP figures. Cf. Blytheville, 71 F.3d at 1385 n. 1 (using 1980, rather 1990, Census figures because the panel did not have “the relevant voting-age populations” from the 1990 Census before it).
Third, Dr. Rodden’s estimates tend to overestimate the District’s Black VAP. For instance, Dr. Rodden excludes persons reporting two or more races from the District’s full VAP in his calculations, thereby using a smaller denominator and overstating the Black VAP proportion of the population. Dr. Rodden also does not use the BOEC’s FFSD boundaries in his population estimates, because the ACS does not provide tract data, and the exact geographic area of BOEC and the corresponding population cannot be reproduced. See Cooper Report at ¶ 13.
Accordingly, I find that Dr. Rodden’s calculations estimating the AP Black VAP based on the 2011-2013 ACS are insufficient to overcome the presumptive accuracy of the 2010 Decennial Census’s count Of the AP Black VAP in the District.
Hi. The District’s trend analysis and population projections
Next, in an attempt to determine the VAP in the District by race in 2015, Dr. Rodden created a linear projection by plotting the average population estimates from the midpoint year of four sets of ACS three-year estimates, 2008-2010, 2009-2011, 2010-2012, and 2011-2013; see DEF-A, Rodden Rep., ¶¶ 12-13, Fig. 3 (p. 6). 13
As Dr. Rodden credibly testified, trend lines are an acceptable practice among social scientists. Rodden Testimony, Trial Tr. vol. 5, 10: 12-17. And Dr. Rodden demonstrates an undeniable trend in his report, showing that the African American population has been growing steadily since 2000 and the white population has been decreasing rapidly. Rodden Report, DEF-A, Figure 3; Rodden .Testimony, Trial Tr. vol. 5, 8:15-9:9; see also Gordon Rebuttal Report, PLTF-41, p. 1 (corroborating Rod-den’s findings).
Overall, however, Dr. Rodden’s linear projection of FFSD’s VAP population by race does not provide the legally required clear, cogent, or convincing evidence that has a high degree of accuracy necessary to overcome the presumptive accuracy of the Decennial Census. First, Dr. Rodden’s calculations are precisely the kind of “analy-ses'that are [ ] too inaccurate to servó as a basis for changing the basis of conducting elections” that courts have repeatedly rejected. Benavidez v. Irving Indep. Sch. Dist., 690 F.Supp.2d 451, 459 (N.D.Tex.2010) (citation omitted) (rejecting expert’s projected population estimate based on uncertain ACS estimates as “correspondingly unreliable”); Perez, 958 F.Supp. at 1212-13 ; see also Dixon v. Hassler, 412 F.Supp. 1036, 1041 (W.D.Tenn.1976) (rejecting estimates prepared by the National Planning Data Corporation as insufficient proof that the Decennial Census figures are not the best evidence of the current population).
Second, Dr. Rodden’s calculations were based on ACS estimates, but the Census Bureau warns against using uncertain ACS estimates to project population forward. Moreover, Dr. Rodden does not provide margins of error for his projections, which prevents the Court from determining whether an estimated population change since 2010 is a substantial change rendering the 2010 population count obsolete. See McNeil, 851 F.2d 937, 946 (1988) *1036 (even if Census data' does not reflect present population precisely, it does not require courts to use a different data set); Perez, 958 F.Supp. at 1212-13 .
In addition, Dr. Rodden’s trend analysis does not address factors that might affect demographic change in the District and impede voting. For example, property ownership and the housing market can affect migration into or out of the District and implicate voting rights. Local events, including the high profile events in Ferguson the past two years, may affect residents’ decisions to move to or remain in the district. See Cooper Testimony, Trial Tr. vol. 1, 205:14-24 (testifying that the Michael Brown incident in Ferguson “could have completely changed the trend line” and “[t]here’s no way to know how that may, affect future population changes in the school district.”). 14
As a result, I find that Dr. Rodden’s trend analysis does not overcome the presumptive validity of the 2010 Decennial Census, and I accept the Decennial Census figures as the data upon which to assess Plaintiffs’ dilution claim, under which African Americans remain a minority of the VAP.
b) No preclusion even if African Americans were a majority of the VAP
Finally, even if I were to find that African Americans constitute a majority of the District’s VAP, that would not be the end of the analysis. Racial minorities do not suddenly lose the broad protections of the VRA at the moment that they surpass 50% of a jurisdiction’s VAP. A racial minority group that is a bare majority of a jurisdiction’s VAP may still suffer from actionable vote dilution where its members have suffered from a history of discrimination and persistent socioeconomic inequalities that hinder their participation in the political process, such that they remain disadvantaged by a traditional at-large electoral arrangement or other unequally open political processes.
A “racial minority,” as it is used in the § 2 context, refers not to a numerical minority, but rather to members of a protected class. Section 2, subsection (a), prohibits voting practices that result in an abridgement of the right to vote “on account of race or color.” 52 U.S.C. § 10301 (a). Subsection (b) establishes a violation of § 2 if the political processes are not equally open to “members of a class of citizens protected by subsection (a),” and uses “members of a protected class” two other times in the subsection. 52 U.S.C. § 10301 (b). Cingles states that § 2(a) concerns “member[s] of a protected class of racial and language minorities.” 478 U.S. at 43 , 106 S.Ct. 2752 ; see also Salas v. Sw. Tex. Junior Coll. Dist., 964 F.2d 1542, 1548 (5th Cir.1992) (“The Act was aimed at measures that dilute the voting strength of groups because of their race, not their numerical inferiority.”); id. at 1547 (noting that “the plain text of [§ 2 of the VRA], as affirmed by case law, makes clear that the Act is concerned with protecting the minority in its capacity as a national racial or language group,” not in its capacity as a numerical minority in any particular jurisdiction). And the Act’s legislative history makes clear that Congress passed the law to effectuate the guarantees of the Fifteenth Amendment and remedy “the systematic exclusion of [Black people] from the polls that characterizes certain regions of this Nation.” H. Rep. No. 89-439, 1965 U.S.C.C.A.N. 2437, 2440. As a result, racial *1037 minorities may maintain VRA protection even if they reach a bare numerical majority of a jurisdiction.
Indeed, the Supreme Court has made clear that “it may be possible for a citizen voting-age majority to lack -real electoral opportunity.” League of United Latin Am. Citizens v. Perry, 548 U.S. 309, 428 , 126 S.Ct. 2594 , 165 L.Ed.2d 609 (2006). 15 Consistent with this Supreme Court guidance, four of the five Courts of Appeals that have considered this question (the Second, Fifth, Eleventh, and D.C. Circuits) have rejected the per se rule prohibiting vote dilution claims where a racial minority constitutes a numerical majority of a jurisdiction. See Kingman Park Civic Ass’n v. Williams, 348 F.3d 1033, 1041 (D.C.Cir.2003) (“Vote dilution claims must be assessed in light of the demographic and political context, and it is conceivable that minority voters might have less opportunity., .to elect representatives of their choice even where they remain an absolute majority in a contested voting district.”) (internal quotation marks and citation omitted);Monroe v. City of Woodville, 881 F.2d 1327, 1332-33 (5th Cir.1989) (“Unimpeachable authority from [the Fifth Circuit] has rejected any per se rule that a racial minority that is a majority in a political subdivision cannot experience vote dilution”), cert. denied, 498 U.S. 822 , 111 S.Ct. 71 , 112 L.Ed.2d 45 (1990); 16 Meek v. Metro. Dade Cty., 908 F.2d 1540, 1545-46 (11th Cir.1990) (holding that • a claim brought by minority voters who constitute a numerical majority could be viable due to “functional effect” of existing system, and that the district court “properly rejected the county’s contention that Cingles could not apply at all in a setting where the Non Latin White bloc did not constitute a majority of the total population”); Pope v. Cty. of Albany, 687 F.3d 565 , 575 n. 8 (2d Cir.2012) (approvingly citing Salas's conclusion that majority-minority vote dilution claims are not barred as a matter of law). 17
Additionally, courts have found § 2 liability where a racial minority group comprised a majority or near-majority of the jurisdiction’s population. See, e.g., Martin v. Allain, 658 F.Supp. 1183, 1188-91, 1204-05 (S.D.Miss.1987) (finding § 2 violations in certain at-large, multimember districts with- majority Black populations and BVAPs); United States v. Dallas Cty. Comm’n, 636 F.Supp. 704, 710 (S.D.Ala.1986) (finding at-large system for county commissioners violated § 2 where Blacks comprised 49.8% of the VAP, an almost 5% increase from four years prior); Windy Boy v. Cty. of Big Horn, 647 F.Supp. 1002, 1004, 1023 (D.Mont.1986) (finding § 2 violation where American Indians comprised 46.2% of the population and whites 52.1%); Jordan v. City of Greenwood, 599 F.Supp. 397, 400, 404-05 (N.D.Miss.1984) (finding *1038 § 2 -violation where Blacks constituted 52% of the total population and 46.2% of the VAP); N.A.A.C.P. v. Gadsden Cty. Sch. Bd., 691 F.2d 978 , 980, 988 (11th Cir.1982) (finding at-large system had effect of diluting Black voting power where African Americans constituted a majority of the population and over 49.36% of the registered voters).
Indeed, courts in the Eighth Circuit have recognized in crafting remedial plans that minority voters often require more than a bare numerical majority in order to elect their preferred candidates. See Smith v. Clinton, 687 F.Supp. 1361, 1362 (E.D.Ark.1988) (three-judge court), aff'd mem., 488 U.S. 988 , 109 S.Ct. 548 , 102 L.Ed.2d 576 (1988); see id. at 1363 (ordering Board to implement plaintiffs’ plan providing for single-member majority-Black district with a 60.55% BVAP to “give blacks a fair opportunity to elect the candidate of their choice... and help to eradicate the effect of the dual-member, at-large system on participation by blacks in the political process”); see also Bone Shirt II, 461 F.3d at 1023 (applying 65% minority population as a “guideline” to consider in fashioning remedial relief and correctly considering turnout rate and incumbency in formulating a districting plan).
As a result, even if African Americans were a bare majority of the VAP, Plaintiffs’ claim would not be barred per se.
c) There is evidence that African Americans lack an equal opportunity to elect their preferred candidates in the District, even if they are just over 50% of the VAP.
In addition to finding there is no per se bar to Plaintiffs’ claim, I also conclude that even if the AP BVAP in FFSD were a slight numerical majority of the population, Plaintiffs’ claim in this case would still succeed because a range of socioeconomic and political factors negatively impact their ability to effectively participate in the electoral process.
A local appraisal of FFSD (see infra in Section IV.B.3 (regarding the existence of racially polarized voting), and Section V.B. (regarding the historical and ongoing effects of discrimination)), demonstrates precisely why the per se rule urged by the District would frustrate the purpose of § 2: it is clear that voting in the District is racially polarized and Black-preferred candidates have a much lower rate of success than white-preferred candidates, which indicates that, regardless of the precise size of Black VAP in the District, the existing at-large arrangement dilutes Black voting power.
Critically, a range of ongoing disparities hinder African Americans’ present ability to participate equally in school board elections. See supra, Section II.B.4. Members of the African American community in FFSD are historically disadvantaged and face functional barriers to electoral participation as a result of the ongoing socioeconomic effects of discrimination, as well as electoral processes that, in practice, favor the status quo. See infra, Sections V.B; V.F. These barriers have likely lead to decreased rates of registration and turnout among African Americans in FFSD. See supra Section II.4; see also PLTF-49, Rebuttal Report of David Kimball, July 2, 2015 (“Kimball Rebuttal”), at 6 (during the last five contested elections, African American turnout has been lower than white turnout four times).
Importantly, “Low minority turnout does not militate against finding a Section 2 violation .... Instead, ‘low voter turnout has often been considered the result of the minority’s inability to effectively participate in the political process.” United States v. City of Euclid, 580 F.Supp.2d 584, 604 (N.D.Ohio 2008) (citing Blytheville, 71 F.3d at 1388 ) (emphasis in origi *1039 nal); see also Benavidez v. City of Irving, 638 F.Supp.2d 709, 725 (N.D.Tex.2009); Engstrom Testimony, Trial Tr. vol. 4, 82:8-11. Where, as in FFSD, many African-American voters believe their diluted votes are futile in electing their candidates of choice, turnout may remain low. But registration and turnout rates often improve once a court finds and remedies the § 2 violation. See, e.g., City of Irving, 638 F.Supp.2d at 725 ; Vill. of Port Chester, 704 F.Supp.2d at 451 .
As a result, even if the any-part Black VAP were slightly over 50% of the VAP, there is no evidence that such a distinction makes a functional difference in terms of how the political processes operate to the detriment of African Americans in FFSD. Under the facts of this case, a bare numerical majority of the VAP is insufficient to translate into meaningful electoral opportunity. See League of United Latin Am. Citizens v. Perry, 548 U.S. 399, 428 , 126 S.Ct. 2594 , 165 L.Ed.2d 609 (2006) (“it may be possible for a citizen voting-age majority to lack real electoral opportunity”).
For all of these reasons, I conclude that Plaintiffs have established Gingles I.
IV. GINGLES II AND III
A. Gingles II and III Legal Standards
Together, Gingles II and Gingles III ask first, whether African American and white voters tend to “vote differently,” ie., whether there is racially-polarized voting, Gingles, 478 U.S. at 53 n. 21, 106 S.Ct. 2752 ; and second, whether the candidates preferred by Black voters “usually” lose to candidates preferred by white voters, Blytheville, 71 F.3d at 1385 .
Gingles II is satisfied where the minority group is politically cohesive. See Gingles, 478 U.S. at 61, 56 , 106 S.Ct. 2752 . Cohesiveness exists where “a significant number of minority group members usually vote for the same candidates,” Bone Shirt II, 461 F.3d at 1025 (citation omitted), and can be established by demonstrating the existence of racially polarized voting, ie., “a consistent relationship between [the], race of the voter and the way in which the voter votes, or to put it differently, where black voters and white voters vote differently,” Gingles, 478 U.S. at 53 n. 21, 106 S.Ct. 2752 (alteration in original) (citations and internal quotation marks omitted). This “consistent relationship” does not require completely divergent racial preferences, as “the Gingles standard presupposes the existence of crossover voting.” Jenkins v. Red Clay Consol. Sch. Dist. Bd. of Educ., 4 F.3d 1103 , 1123 (3d Cir.1993); see Sanchez v. Colorado, 97 F.3d 1303, 1319 (10th Cir.1996) (same).
Gingles III, meanwhile, is satisfied where “the white majority votes sufficiently as a bloc to enable it, in the absence of special circumstances, usually to defeat the minority’s preferred candidate.” Blytheville, 71 F.3d at 1385 (citing Gingles, 478 U.S. at 50-51 , 106 S.Ct. 2752 ). There is no requirement that white voters have “an unbending or unalterable hostility” to minority-preferred candidates such that those candidates always lose. Jenkins, 4 F.3d at 1123; see also Blytheville, 71 F.3d at 1389 (marginal minority electoral success “fit[s] precisely in the Gingles test as to whether the white majority does indeed vote sufficiently as a bloc to enable it.. .usually to defeat the minority’s preferred candidate”) (alteration in original) (citation and internal quotation marks omitted).
As the Supreme Court observed, “no simple doctrinal test” applies to the third Gingles factor because racial bloc voting can “vary according to a variety of factual circumstances.” Gingles, 478 U.S. at 57-58 , 106 S.Ct. 2752 . Gingles III is determined through three inquiries: (1) identifying the minority-preferred candi *1040 dates; (2) assessing whether “the white majority vote as a bloc to defeat the minority preferred candidate”; and (3) resolving whether “there [were] special circumstances such as the minority candidate running unopposed present when minority-preferred candidates won.” Bone Shirt II, 461 F.3d at 1020 (alteration in original) (citation omitted).
1. Relevant elections for evaluating Gin-pies II and III
Before a court can identify the minority-preferred candidates, it must determine which elections to analyze. Not all elections are equally probative in assessing Gingles II and III. In particular, contested elections have much more probative value than uncontested elections because only if elections are held can the outcomes provide relevant information about the ability of Black voters to elect candidates of their choice. See Engstrom Testimony, Trial Tr. vol. 4, 45:22-46:10. Here, the 2005, 2007, 2008, and 2010 elections in FFSD were uncontested and therefore they have little to no probative value in the Gingles preconditions analysis. Id,.; see also Gingles, 478 U.S. at 57 , 106 S.Ct. 2752 . 18
Moreover, for purposes of analyzing the Gingles preconditions, not all contested elections have equal probative value. More recent elections are generally more probative. Bone Shirt II, 461 F.3d at 1020-21. Additionally, “endogenous” elec tions — ie., those elections for the offices at issue, here, the FFSD Board — are more probative than the results of “exogenous” elections — ie., contests for other offices, such as Congress or President, id.; see also Clay II, 90 F.3d at 1362 . While exogenous elections may still inform the inquiry, the Court need not supplement endogenous election data where, as here, there is sufficient evidence from endogenous elections from which to discern typical voting behavior and usual results. See, e.g., Westwego Citizens for Better Gov’t v. City of Westwego, 872 F.2d 1201 , 1208 n. 8 (5th Cir.1989). Likewise, “interracial elections are the best indicators of whether the white majority usually defeats the minority candidate,” Bone Shirt II, 461 F.3d at 1020-21, 19 because “[a] system that works for minorities only in the absence of white opposition is a system that fails to operate in accord with the law,” Blytheville, 71 F.3d at 1389-90 ; see Jenkins, 4 F.3d at 1128. And finally, there is less probative value in an election that is marked by special circumstances that suggest that the election “was not representative of the typical way in which the electoral process functions.” Ruiz, 160 F.3d at 557-58; see also Gingles, 478 U.S. at 75-76 , 106 S.Ct. 2752 ; Blytheville, 71 F.3d at 1389 .
2. Methods for identifying Black-preferred candidates
[36] To analyze whether Plaintiffs have met Gingles II and III, the Court must determine whether Black voters have candidates of choice, and if so, identify those candidates. This requires courts to conduct “an intensely local appraisal” that accurately reflects voters’ preferences. Gingles, *1041 478 U.S. at 78 , 106 S.Ct. 2752 (citation omitted).
Identifying Black voters’ candidates of choice is not a mechanical task. Rather, a court must employ a contextual case-by-case approach that, as required by the Eighth Circuit, establishes, “the preferences of the minority voters... on an election-specific basis, viewing all the relevant circumstances.” Blytheville, 71 F.3d at 1386 (citing Jenkins, 4 F.3d at 1126); see also Collins v. City of Norfolk, 816 F.2d 932 , 937 (4th Cir.1987) (“Collins /”). “There is no blanket definition of ‘minority preferred candidate.’ ” Clay II, 90 F.3d at 1361 . Answering the Gingles II and III threshold questions “typically requires a statistical and non-statistical evaluation” of the voting behavior and election results in the relevant elections. Bone Shirt II, 461 F.3d at 1020.
B. Gingles II and III Findings of Fact
1. Relevant elections
From 2000 through 2015, there have been twelve contested Board elections. Joint Stip. ¶ 53. By statute, where the number of candidates who have filed for a Board election equals the number of seats to be elected, no election is held, and the candidates automatically assume Board member responsibilities. From 2000 through 2015, there were four such non-elections (2005, 2007, 2008, 2010). Joint Stip. ¶ 54; RSMo. § 115.124.1.
The Plaintiffs’ expert Dr. Engstrom analyzed the elections from 2011 through 2015. Dr. Engstrom focused on the 2011 to 2015 elections because he wanted to analyze at least two three-vote, three-seat elections and because-the four-year period immediately preceding 2011 saw no contested interracial elections. In Thornburg v. Gingles, 478 U.S. 30 , 106 S.Ct. 2752 , 92 L.Ed.2d 25 (1986), the Supreme Court found an analysis of three elections to be sufficient. See also Engstrom Testimony, Trial Tr. vol. 4, 17:5-18:4. Like the Court did in Gingles , Dr. Engstrom focused on endogenous interracial contested elections.
The District’s expert Dr. Rodden analyzed each of the past twelve contested Board elections. Joint Stip. ¶ 55; see also Engstrom Testimony, Trial Tr. vol. 4, 11:15-25; Rodden Testimony, Trial Tr. vol. 5, 36:22-37:5.
2. Methods for identifying Black-preferred, candidates
In the multi-seat contests at issue here, the identification of minority-preferred candidates is complex. Because a voter can cast more than one vote, minority voters may (but will not necessarily) have a second (or third) candidate of choice. See Engstrom Testimony, Trial Tr. vol. 4, 18:18-19:16. A critical question, then, is how to identify whether minority voters in fact have a second or third candidate of choice in a given election.
a) Ecological Inference analysis
Both Dr. Engstrom and Dr. Rodden analyzed FFSD Board elections using Gary King’s Ecological Inference (“El”) procedure. Joint Stip. ¶¶ 55-56; PLTF-52, Eng-strom Rep., ¶¶ 15-39, Tbl. 1 (pp. 17-19); PLTF-53, Rebuttal Report of Richard L. Engstrom, July 2, 2015 (“Engstrom Rebuttal”), Tbls. 1-5 (pp. 10-13); Rodden Testimony, Trial Tr. vol. 5, 36:22-37:5.
El is used widely by expert witnesses in assessing racially' polarized voting in voting rights cases. It was developed subsequent to the U.S. Supreme Court’s 1986 decision in Gingles for the explicit purpose of improving estimates of candidate preferences between or among groups of voters. See Joint Stip. ¶ 59. El provides estimates of the true value of a group’s support of a candidate, i.e., the actual percent of votes cast by African American voters for a candidate. This estimate is expressed in terms of a point estimate and confidence interval. Engstrom Testimony, Trial Tr. vol. 4,14:1-15:23. The point estimate is the value that *1042 is closest to the true value as one can get with the data, or statistically the best estimate of true value. Id, at 14:24-15:10.
The confidence interval is a measure of uncertainty. Id. at 14:11-15:23. It provides a range of estimates around the point estimate so that political scientists can establish with 95% confidence that the true value (the candidates’ level of support) is within that range. Joint Stip. ¶ 57; PLTF-52, Engstrom Rep., ¶ 10; Engstrom Testimony, Trial Tr. vol. 4, 14:13-23. A statistical estimate with a 95% degree of confidence employs the level of statistical certainty that is consistent with generally accepted standards for El in the field of political science, and is consistent with peer-review standards for research in that field. Joint Stip. ¶ 58; Engstrom Testimony, Trial Tr. vol. 4, 14:13-23, 15:24-16:1; Rodden Testimony, Trial Tr. vol. 5, 151:13-16.
El estimates must include a confidence interval to determine whether the difference in the estimated support among candidates can be considered statistically significant. Engstrom Testimony, Trial Tr. vol. 4, 16:2-12; Rodden Testimony, Trial Tr. vol. 5, 154:2-21. As a “rule of thumb,” when the point estimate for the level of support for one candidate lies within the confidence interval of support for another candidate, one cannot say that there is a statistically significant difference between the two candidates. See Engstrom Testimony, Trial Tr. vol. 4, 18:8-17, 19:4-16; Rodden Testimony, Trial Tr. vol. 5, 154:2-11.
b) Homogenous-precinct analysis
Homogenous-precinct analysis is another tool that can be used to confirm the existence and extent of racially polarized voting. Engstrom Testimony, Trial Tr. vol. 4, 12:15-21. Although El is a superior tool for racial polarization analysis, homogenous-precinct analysis may be a helpful supplement to El when there are many precincts in a particular area with similar proportions of African American and white adults. Id. Homogenous-precinct analysis examines the few but highly segregated precincts and therefore “reduces uncertainty in estimating voting behavior of each racial group.” Kimball Rebuttal at 4, n.3.
c) Top-ranked candidate approach
Dr. Rodden used two methods for identifying Black-preferred candidates: the top-ranked candidate approach, in which he identified only the candidate whose point estimate indicated the highest level of support among Black voters; and the point estimate approach, in which he identified the candidates whose point estimates indicated the highest and second-highest estimated levels of support among Black voters in a two-seat contests, and the highest, second-highest, and third-highest estimated levels of support among Black voters in three-seat contests. See Rodden Testimony, Trial Tr. vol. 5, 38:24-40:14.
The top-ranked candidate approach identifies only a single Black-preferred candidate in each election; it does not consider the possibility that minority voters have clear second- and third-choice candidates in multi-seat elections; and it also does not account for relative levels of support for each candidate among Black voters, treating the single candidate with the highest estimated level of support among Black voters-as the only Black-preferred candidate, even if another candidate received a substantially similar or even statistically indistinguishable level of support from Black voters. Rodden Testimony, Trial Tr. vol. 5, 95:23-96:3; 97:5-17.
The following table summarizes the preferred candidates among Black and white voters using the top-ranked candidate approach, respectively, and their success rates:
*1043 [[Image here]]
PLTF-75, Top-Ranked Candidate Ap--proach Worksheet (Ex. 18 to Dep. of Jonathan Rodden, Aug. 20, 2015); Rodden Dep., 260:21-272:5.
According to Dr. Rodden’s top-ranked candidate approach, Black and white voters in FFSD diverge in every election since 2000 in terms of candidate, preference. In the twelve contested elections from 2000 through 2015, Black voters and white voters never preferred the same candidate as their top choice. Joint Stip. ¶ 189; Rodden Testimony, Trial Tr. vol. 5, 100:1-6.
According to Dr. Rodden’s top-ranked candidate approach, Black and white voters display a consistent preference for candidates from their own racial group. In the twelve contested elections from 2000 through '2015, every top-ranked candidate among white voters was white (twelve of twelve). See Joint Stip. ¶¶ 64, 71, 79, 88, 96, 103, 112, 119, 130, 138, 146, 167. In the twelve contested elections from 2000 through 2015, all but one of the top-ranked candidates among Black voters were Black (eleven of twelve). Joint Stip. ¶ 190. The only election in which Black voters’ top-ranked candidate was not Black was the 2009 election, which featured no Black candidates. See Joint Stip. ¶ 190.
The top-ranked candidate approach also shows that Black-preferred candidates lose much more frequently than do white-preferred candidates. In the twelve contested elections from 2000 through 2015, every top-ranked candidate among white voters was elected (twelve of twelve), as compared to six out of twelve top-ranked candidates among Black voters. Joint Stip. ¶ 191. In the five contested elections over the last five years (i.e., from 2011 through 2015), every top-ranked candidate among white voters was elected (five out of five), as compared to only two out of five top-ranked candidates among Black voters. Joint Stip. ¶ 193.
d) Point estimate approach
Dr. Rodden’s point estimate approach considers a candidate preferred if they receive the highest or second-highest (in a two-seat election), or highest, second-highest, or third-highest (in a three-seat election) estimated level of support from a group of voters in a given election, depending on the number of seats up for election. This approach does not consider relative levels of support for each candidate among African American voters — for example, in *1044 a two-seat election, the two candidates with the highest and second-highest estimated levels of Black support are both considered Black-preferred candidates to the same degree, even if one candidate had a much lower level of support among Black voters than the other.
The point estimate approach also does not consider overlapping confidence intervals among candidates, meaning in comparing two candidates, one could be considered Black-preferred while the other is not simply because the first candidate has a higher point estimate of Black support, even if the two candidates’ respective levels of Black support are in fact statistically indistinct from one another. See, e.g,, Eng-strom Testimony, Trial Tr. vol. 4, 29:7-30:2, 30:3-11, 32:22-33:12. Under the point estimate approach, Dr. Rodden identified the following candidates as Black-preferred:
[[Image here]]
PLTF-69, Point Estimate Approach Worksheet (Ex. 12 to Dep. of Jonathan Rodden, Aug. 20, 2015); Rodden Dep., at 136:22 — 137:6, 356:19 — 357:9.
Applying the point estimate approach reveals frequent divergence between Black and white voters in FFSD in terms of candidate preference, and also reveals voters’ consistent preference for candidates from their own racial groups. Using this approach, white voters preferred white candidates for 25 of the 27 total seats up for election from 2000 to 2015 (93% of the time), while Black voters preferred Black candidates for 17 of the 27 seats up for election (63% of the time). See Rodden Testimony, Trial Tr. vol. 5, 107:11-17. Black and white voters preferred the same candidates only one-third of the time. See Deft FFSD-A, Rodden Rep., ¶45, Fig. 8 (p. 25); see also generally id. ¶¶ 47-68; Rodden Testimony, Trial Tr. vol. 5, 107:18-108:25. 20
*1045 Under the point estimate approach, it is evident that Black-preferred candidates are usually defeated, while white-preferred candidates usually win. Since 2000, 24 out of 27 white-preferred candidates were elected (88.9%), as compared to 13 out of 27 Black-preferred candidates (48.1%). See Bodden Testimony, Trial Tr. vol. 5,110:18-111:9. This gap in success rates has grown in more recent years: during the last five years, 11 out of 12 white-preferred candidates were elected (91.7%), as compared to 4 out of 12 preferred candidates among Black voters (33.3%).
3. Findings of fact concerning racially polarized voting
a) Dr. Engstrom’s testimony toas reliable and highly credible
For the following reasons, I find that Dr. Engstrom’s testimony was reliable and highly credible, and I will adopt his findings concerning racially polarized voting.
Dr. Richard Engstrom is a political science professor and social science researcher at Duke University with thirty years of experience related to electoral systems and minority electoral participation. Dr. Engstrom is an expert in electoral systems, minority politics, racial polarization analysis, and statistical methods related to racial polarization analysis. More specifically, he is a nationally recognized expert in the relationship between election systems and the ability of minority voters to participate fully in the political process and to elect representatives of their choice. Trial Tr. vol. 4, 6:21-25, 9:23-10:3 (admitting Engstrom as expert without objection); Joint Stip. ¶ 56; Engstrom Testimony, Trial Tr. vol. 4, 5:4-10:3.
Dr. Engstrom has published a peer-reviewed book on minority politics and election systems, and many peer-reviewed articles on the structure of electoral contests. See PLTF-52, Engstrom Rep., at Appendix (CV). Dr. Engstrom has testified at trial or in depositions in approximately 100 Voting Rights Act cases, including as a court-appointed expert, a special master, and on behalf of both plaintiffs and defendants, and he has never been deemed not qualified as an expert. See Engstrom Testimony, Trial Tr. vol. 4, 5:6-9:22.
Dr. Engstrom’s El analysis was accurate, complete, and confirmed by similar outcomes from Dr. Engstrom’s HP analysis, as well as Dr. Rodden’s El analysis.
Based on the El analysis, Dr. Engstrom determined for each of the five elections from 2011-2015 whether African American voters expressed a preference for a particular candidate or candidates and identified Black voters’ candidates of choice. Dr. Engstrom’s focus on the last five years of endogenous elections was appropriate in the particular circumstances of FFSD. See Trial Tr. vol. 4, 17:3-18:4. The 2011-2015 elections are all recent, endogenous, 21 and interracial elections, 22 and *1046 reviewing five contested elections satisfies the three elections threshold the Supreme Court considered sufficient in Gingles, 478 U.S. at 61 , 106 S.Ct. 2752 ; see also Gingles, 478 U.S. at 57 n. 25, 106 S.Ct. 2752 . The five-year period also includes two three-seat elections, to allow the Court at least two observations of that particular electoral context. Analysis of the endogenous Board elections from 2011 to 2015 is sufficient and appropriate to determine the District’s liability.
Dr. Engstrom applied a case-by-case method for identifying candidates of choice of Black voters. Evaluating Black-preferred candidates on an election-by-election basis, Dr. Engstrom identified eight candidates preferred by Black voters between 2011 and 2015. He identified a candidate as a Black-preferred candidate based on the three following principles. Dr. Engstrom’s reliance on these three principles was appropriate and reliable.
First, Dr. Engstrom considered statistically significant differences in levels of support for the candidates in characterizing a second- or third-place finisher among Black voters as a Black candidate of choice. PLTF-52, Engstrom Rep., ¶ 16 (2011); ¶ 20 (2012), ¶24 (2013); ¶28 (2014); ¶ 34 (2015); see also ¶ 17 (statistical significance in 2011); ¶ 20 (statistical significance in 2012); ¶28 (statistical significance in 2014); ¶ 34 (statistical significance in 2015). When the point estimate of one candidate is within the confidence interval of the other, Dr. Engstrom testified that one cannot determine which candidate received a higher level of support. See Engstrom Testimony, Trial Tr. vol. 4,16:2-12. By considering statistically significant differences in levels of support, Dr. Engstrom followed generally-accepted standards in social science.
Second, a second- or third-choice candidate is not necessarily minority-preferred if that candidate received significantly less support than the top-choice candidate. See, e.g., Engstrom Testimony, Trial Tr. vol. 4, 18:5-21, 19:4-6; see also id. 26:2-20. Dr. Engstrom testified that Black voters’ second- or third-most-favored candidate should be deemed a candidate of choice only if he or she had at least approximately two-thirds of the support of the most-favored candidate. See, e.g., NAACP v. City of Thomasville, 401 F.Supp.2d 489 , 498 & n. 3 (M.D.N.C.2005). As other courts have observed, and as the facts of this case demonstrate, the identification of a second- or third-choice candidate, if any, requires a detailed and nuanced analysis. See, e.g., Lewis v. Alamance Cty., 99 F.3d 600 , 614 (4th Cir.1996); N.A.A.C.P. v. City of Niagara Falls, 65 F.3d 1002 , 1017-19 (2d Cir.1995); Askew, 127 F.3d at 1379; N.A.A.C.P. v. City of Thomasville, 401 F.Supp.2d at 498 & n. 3.
Third, Dr. Engstrom testified that, where the clear first-choice candidate among Black voters is a Black candidate who has lost, courts should be skeptical of attempts to characterize a winning white candidate as a candidate of choice of minority voters, particularly where Black voters show a preference for candidates within their racial group. Engstrom Testimony, Trial Tr. Vol. 4, 19:20-20:15; id., 30:12-23 (testifying to Black voters’ preference in 2012); id., 33:13-22 (testifying to Black voters’ preferences in 2013); see also, e.g., Collins I, 816 F.2d at 937 & n. 6 (casting doubt on whether elected white candidates who received some support from Black voters could “be fairly considered as representatives of the minority community” where those candidates were *1047 elected over Black candidates who were clearly the most minority-preferred candidates). .
An electoral system does not provide equal opportunity if Black voters cannot elect their top candidate(s) of choice and can only elect lesser preferred candidates, and only if they are white. If Black voters prefer Black candidates, they must have an opportunity to elect those candidates. See,' e.g., Collins I, 816 F.2d at 937 & n. 6; Citizens for a Better Gretna, 834 F.2d at 502; see also Aldasoro v. Kennerson, 922 F.Supp. 339, 374 (S.D.Cal.1995) (citing Collins v. City of Norfolk, 883 F.2d 1232, 1238 (4th Cir.1989) (“Collins IF)). “Gingles addresses not only a group’s ability to elect a satisfactory candidate (that is a candidate for whom the minority voter is willing to cast a vote), but the group’s ability to elect its 'preferred candidate.” Meek, 908 F.2d at 1547 . In FFSD, “African Americans have clearly shown, through their voting behavior, a preference to be represented on that Board by African American candidates.” PLTF-52, Expert Report of Richard L. Engstrom, May 27, 2015 (“EngstromRep”), ¶43.
Notably, this principle does not assume that a candidate is preferred because of their race. Blytheville, 71 F.3d at 1386 . Rather, it acknowledges legal guidance that, although the race of a candidate is not dispositive, an electoral system does not truly provide equal opportunity if Black voters cannot elect members of their own group who are their clearly most-preferred candidate(s), and can only elect lesser-preferred candidates who are white. See Engstrom Testimony, Trial Tr. vol. 4, 20:3-10; see also Aldasoro, 922 F.Supp. at 374 (citing Collins II, 883 F.2d at 1238 ).
Using this case-by-case method, Dr. Engstrom identified eight candidates preferred by Black voters between 2011 and 2015.
As discussed above, in addition to using El, Dr. Rodden offered two methods for identifying candidates. of choice:, -the top-preferred candidate approach and the point-estimate approach. Both of Dr. Rod-den’s approaches employ a singular mechanical rule for identifying the Black-preferred candidate. As the Supreme Court observed, however, “no simple doctrinal test” applies to legally significant bloc voting because “the degree of racial bloc voting that is cognizable as an element of a § 2 vote dilution claim will vary according to a variety of factual circumstances.” Gingles, 478 U.S. at 57-58 , 106 S.Ct. 2752 . The Eighth Circuit has also made clear that minority candidates of choice should be identified on an- election-specific basis, viewing all the relevant circumstances. See Blytheville, 71 F.3d at 1386 ; see also Jenkins, 4 F.3d at 1126; Collins I, 816 F.2d at 937.
Dr. Rodden’s approaches have limitations. Although the top-ranked candidate method provides relevant information — in that, as a general matter, a candidate who receives the greatest number of votes from minority voters (the “top-ranked candidate” of minority voters) is appropriately described as a Black-preferred candidate, see, e.g., Lewis, 99 F.3d at 614, looking only at the top-ranked candidate does not capture the full voting preference picture in the context of a multi-seat election because it disregards the fact that multiple seats are available in each election, and with that the possibility that minority voters prefer more than one candidate. As explained below, this flaw is particularly on display when analyzing the 2011 and 2014 elections. In those elections, Black voters showed a strong preference for two or three candidates.
Dr. Rodden’s other approach, the point estimate approach, takes the two or three candidates with the highest estimates for Black voters’ support (depending on *1048 whether there are two or three seats available), and designates them all as Black-preferred candidates, regardless of the comparative levels of support. This mechanical approach fails to identify Black-preferred candidates “on an election-specific basis,” with consideration of “all the relevant circumstances,” Blytheville, 71 F.3d at 1386 , and ignores two important factors in identifying candidates of choice. First, this approach ignores the relative levels of support among first-, second-, and third-choice candidates, treating all three equally as candidates of choice in a three-seat election, even where one or two candidates clearly stand out as preferred in terms of support from Black voters. In so doing, this approach ignores case law that warns against treating the election of a second-choice candidate as a victory for minority voters, particularly if minority voters show a clear preference for a candidate from within their racial group who loses and their second-choice candidate, who wins, is white. Second, the point estimate approach draws a distinction between candidates even when their levels of support are statistically indistinguishable from each other. In this respect, the point estimate approach is inappropriate in light of generally accepted principles of statistical analysis. As explained below, these flaws are most clearly on display when analyzing the 2012 and 2013 elections.
For all of these reasons, I find that Dr. Engstrom’s El analysis was accurate, complete, and reliable. His testimony was highly credible, and his case-by-case method for identifying Black-preferred candidates was done in accordance with relevant legal standards and was superior to the other methods proposed by Dr. Rodden. See Gingles, 478 U.S. at 57, 75-76, 78 , 106 S.Ct. 2752 (setting out standards for determining relevant elections to analyze and requiring “intensely local appraisal” to determine candidates of choice); see also Bone Shirt II, 461 F.3d at 1020. Accordingly, the Court credits Dr. Engstrom’s testimony.
b) Candidates of choice by election year
In the following section, I have identified candidates of choice based on Dr. Eng-strom’s case-by-case approach. Each of the elections held since 2011 is described briefly below. For contested elections, tables are provided that identify each candidate’s name, race, total number of votes received, and level of support (the “point estimate”) among Black and white voters with confidence intervals for each estimate as calculated by Dr. Engstrom and Dr. Rodden, and the winning candidates. Percentages reported are the percentage of votes received by a candidate from each group, not the percentage of voters supporting from each group supporting each candidate. 23 Joint Stip. ¶ 61; see Deft-FFSD A, Rodden Rep., ¶ 45, Fig. 8 (p. 25); PLTF-67, Underlying Data for Fig. 8 in Rodden Report (Ex. 10 to Dep. of Jonathan Rodden, Aug. 20, 2015); PLTF-68, Rodden’s Underlying Data Reformatted (Ex. 11 to Dep. of Jonathan Rodden, Aug. 20, 2015); PLTF-52, Engstrom Rep., ¶¶ 15-39, Tbl. 1 (pp. 17-19); PLTF-53, Engstrom Rebuttal, Tbls. 1-5 (pp. 10-13).
The parties’ experts agreed that in FFSD Board elections, African Americans were more likely to vote for African American candidates and whites were more likely to vote for white candidates for the past sixteen years. Deft-FFSD B, Supplemental Report of Jonathan Rodden: Racial Bloc Voting and Cohesion, July 2, 2015 (“Rod-den Bloc Voting Rep.”), at ¶ 2; Engstrom *1049 Testimony, Trial Tr. vol. 4, 19:17-20:10; see also Dep. of Jowei Chen, Aug. 19, 2015 (“Chen Dep”), at 54:23-56:3; PLTF-52, Engstrom Rep., ¶ 43 (regarding 2011-2015).
The parties’ experts, both using El analysis, did not materially disagree as to the various levels of support each Board candidate received among different racial groups in each of the elections in 2011-2015, and report consistent estimates of the respective levels of support received by the candidates from Black and white voters. Joint Stip. ¶ 60; see PLTF-53, Eng-strom Rebuttal, ¶¶3-4, Tbls. 1-5 (pp. 10-13).
i. The 2011 election
In 2011, nine candidates (incumbents Graham, Clark, and Lentz and six challengers, Chris Martinez, Paul Morris, Robert Chabot, Vanessa Hawkins, Brian Scott Ebert, and Joseph Hosea) ran for three seats. Joint Stip. ¶ 118. Martinez, a white-Hispanic candidate, and Chabot and P. Morris, both white, were successful. Joint Stip. ¶¶ 118-119; PLTF-53, Engstrom Rebuttal, ¶ 9 n.5; PLTF-8, Official 2011 FFSD Election Results.
Dr. Rodden’s and Dr. Engstrom’s El estimates and confidence intervals are as follows:
[[Image here]]
* Indicates winner
Joint Stip. ¶ 119. Because this was a three-seat election, the highest percentage of votes that any candidate could receive in absence of bullet voting is 33%.
Graham received the highest estimated level of support, and Hawkins the second-highest estimated level of support, among Black voters in the 2011 election (24.1% and 21.5%, respectively, according to Eng-strom). Joint Stip. ¶¶ 119-21. Both of these candidates received cohesive support from Black voters. Engstrom Testimony, Trial Tr. vol. 4, 24:25-26:1.
*1050 Although Clark received the third-highest estimated level of support among Black voters, Joint Stip. ¶ 120, his percentage of the Black vote (13.5% according to Engstrom) was substantially lower, and statistically significantly lower than the estimated level of support for Graham or Hawkins. Engstrom Testimony, Trial Tr. vol. 4, 26:2-20 (“[t]here’s no cohesive vote for Clark. He may come in third, but that doesn’t mean he — I wouldn’t call him a black-preferred candidate. He’s so far behind the two leading candidates among black voters.”).
Graham and Hawkins together received an estimated 45.6% of the total votes cast by African Americans in a race with nine candidates. Joint Stip. ¶ 119. Graham and Hawkins, who are Black, were the only Black-preferred candidates in 2011. Based on the level of support he received,' Clark, who is white, was not a Black-preferred candidate. Engstrom Testimony, Trial Tr. vol. 4, 25:24-26:7.
Graham and Hawkins received minimal support from white voters. Engstrom Testimony, Trial Tr. vol. 4, 25:18-23 (white crossover voting for Black-preferred candidates in 2011 was “minimal,” at 6.1% for Graham and 7.4% for Hawkins). Neither Graham nor Hawkins was elected to one of the three available seats on the Board in 2011. Joint Stip. ¶ 123.
ii. The 2012 election
In 2012,' three candidates (incumbent Schroeder and two challengers, Ebert and Barbara Morris) ran for two seats. Ebert and Schroeder, both white candidates, were successful. Joint Stip. ¶ 129.
Dr. Rodden’s and Dr. Engstrom’s El estimates and confidence intervals are as follows:
[[Image here]]
* Indicates winner
Joint Stip. ¶ 130. Because this was a two-seat election, the highest percentage of votes that any candidate could receive in absence of bullet voting is 50%.
B. Morris, who was the lone Black candidate, was. the top-ranked candidate among Black voters, and received the highest-estimated level of support among Black voters. Joint Stip. ¶ 131. In fact, she received over 50% of votes cast by Black voters, indicating essentially unanimous Black support for her candidacy and that at least some bullet-voted for her. Rodden Testimony, Trial Tr. vol. 5, 67:20-25, 203:19-204:5; see also id, at 203:19-204:5; Engstrom Testimony, Trial Tr. vol, 4, 27:24-28:4. ' ■
Schroeder and Ebert,' who. are both white, received the second-highest and third-highest estimated levels of support, 25.0 percent and 23.1 percent respectively, among Black voters. Joint Stip. ¶¶ 130, 132.
Dr. Rodden, employing the point estimate approach, identified Schroeder, who is white and who was elected, as a Black-preferred candidate, because Schroeder *1051 was estimated to have received the next-highest number of Black votes. See Deft-FFSD A, Rodden Rep., ¶ 57. But as Dr. Engstrom explained, Schroeder’s level of support among Black voters (25.0% was “statistically significantly lower” (in fact less than half)), than that of B. Morris, wlio received 51.9%. Engstrom Testimony, Trial Tr. vol. 4, 29:15-30:2; PLTF-52, Eng-strom Rep., Table 1 (p. 18); see also Rod-den Testimony, Trial Tr. vol. 5, 205:10-17; PLTF-68, Rodden’s Underlying Data Reformatted; PLTF-58, Engstrom Rebuttal, Tbl. 2 (p. 11). Moreover, Black voters’ estimated support for their last-choice candidate, Ebert (23.1%), was statistically indistinguishable from that of Schroeder. Engstrom Testimony, Trial Tr. vol. 4, 30:3-11; Rodden Testimony, Trial Tr. vol. 5, 206:1-12; Joint Stip. ¶ 132. In essence, B. Morris had universal support from Black voters, who (when not bullet-voting) split their remaining votes roughly equally between the two remaining candidates. Because Black voters’ support for B. Morris was cohesive and about double the support for the next-highest candidate, Schroeder, whose vote total was not statistically different from Ebert’s, Dr. Engstrom appropriately identified B. Morris as the only Black-preferred candidate, and declined to characterize the election of Schroeder as -a win for Black voters. Engstrom Testimony, Trial Tr. vol. 4, 30:3-23. I find that there was one Black-preferred candidate in the 2012 two-seat election: B. Morris, who lost. Schroeder and Ebert were elected. Joint Stip. ¶ 129; Trial Tr. vol. 4, 27:22-23. '
Additionally, B. Morris received 12.8 percent of white voters’ votes, which is “minimal,” given that there were only three candidates, running for two seats. PLTF-52, Engstrom Rep., ¶ 21; Trial Tr. vol. 4, 28:5-10.
iii. The 2013 election
In 2013, four candidates (incumbents Hogshead and Henson, and two challengers, Keith Brown and Larry Thomas) ran for two seats. Brown and Hogshead, both white, were successful. Joint Stip. ¶ 137.
Dr. Rodden’s and Dr. Engstrom’s El estimates and confidence intervals are as follows:
[[Image here]]
* Indicates winner
Joint Stip. ¶ 138.
Henson, who is African American, received 43.7 percent of Black voters’ votes, the highest estimated level of support among Black voters in the 2013 election. See Joint Stip. ¶¶ 138, 140. Hogshead, who is white, received 24.2 percent of Black *1052 voters’ votes, the second-highest estimated level of support among Black voters. See Joint Stip. ¶¶ 138, 139. Henson received a statistically significant and substantially greater estimated portion of Black voters’ votes than Hogshead. Engstrom Testimony, Trial Tr. vol. 4, 31:25-32:8, 32:25-33:12.
As estimated by Dr. Rodden and Dr. Engstrom, the difference in the levels of support among Black voters received by Hogshead and the third-place candidate, Brown, is not statistically significant. See Joint Stip. ¶ 138; Rodden Testimony, Trial Tr. vol. 5,197:21-198:5.
Because he was the only candidate to receive cohesive support from Black voters in 2013, and because his level of support was substantially higher than the candidate with the second-highest level of support, Henson was the only Black-preferred candidate in the 2013 election. See Eng-strom Testimony, Trial Tr. vol. 4, 31:25-33:9.
Henson was the third-place candidate among white voters, receiving 17.0 percent of their votes, which is a “small” portion of white voters’ votes. Engstrom Testimony, Trial Tr. vol. 4, 32:9-18; see Joint Stip. ¶ 138. Henson was not elected to one of the two available seats in 2013. Joint Stip. ¶ 141.
Dr. Rodden, employing the point estimate approach, characterized Hogshead as the second Black-preferred candidate in this election. But Hogshead received less than two-thirds of the support from Black voters than Henson received. Moreover, because Hogshead’s level of support among Black voters was statistically indistinguishable from that of another candidate, Brown, one cannot even state with statistical confidence that she in fact received the second-highest number of votes from Black voters. In these circumstances, it is inappropriate to characterize Hogshead’s election as a victory for Black voters, particularly in light of Henson’s loss. 24 Engstrom Testimony, Trial Tr. vol. 4, 33:12-33:22. I find that there was one Black-preferred candidate in the 2013 two-seat election: Henson, who lost.
iv. The 2014 election
In 2014, eight candidates (two incumbents, Rob Chabot and Paul Morris, and six challengers, Donna Paulette-Thurman, James Savala, Kimberly Benz, F. Willis Johnson, LaWanda Wallace, and Larry Thomas), ran for three seats. Joint Stip. ¶ 145. Chabot and P. Morris, both white, and Paulette-Thurman, one of five African American challengers, were successful. Joint Stip. ¶ 145.
Dr. Rodden’s and Dr. Engstrom’s El estimates and confidence intervals are as follows:
*1053 [[Image here]]
* Indicates winner
Joint Stip. ¶ 146.
Paulette-Thurman, Johnson, and Savala, who are all African American, received the highest, second-highest and third-highest estimated level of support among Black voters, respectively, in the 2014 election. See Joint Stip. ¶¶ 146, 148. The estimated levels of support among Black voters for Paulette-Thurman, Johnson, and Savala are each more than double that for the candidate with the fourth-highest estimated support, Wallace, who is also African American. Joint Stip. ¶ 150.
Black voters cohesively supported each of these three candidates. Engstrom Testimony, Trial Tr. vol. 4, 34:24-35:28 (Black voters “cohesively supported each of those three candidates... .if you add them together, that’s 75 percent of the votes in a nine-candidate contest. That’s cohesion. I don’t see how you could describe it as anything else.”). Because of the cohesive support they received from Black voters, and the substantial difference in support between Black voters’ support for those three candidates and the rest of the pool, Paulette-Thurman, Johnson, and Savala were Black voters’ preferred candidates in 2014.
Each of the three Black-preferred candidates received minimal white support, with less than 10 percent of white voters’ votes. Paulette-Thurman received 8.4 percent, Savala received 7.0 percent, and Johnson received 2.8 percent. Engstrom Testimony, Trial Tr. yol. 4, 36:2-14. None of these three candidates were preferred by white voters, leading Dr. Rodden to concede that this election was polarized. See Rodden Testimony, Trial Tr. vol. 5, 62:8-14.
I find that there were three Black-preferred candidates in the 2014 three-seat election: Paulette-Thurman, Savala, and Johnson. Paulette-Thurman was elected, along with Chabot and Paul Morris, but Johnson and Savala were not elected. Joint Stip. ¶¶ 145,152. 25
*1054 Special Circumstances in the 2014 Election
The 2014 election took place less than one month after the “controversial” resignation of Dr. Art McCoy, the first African American District Superintendent. Joint Stip. ¶ 160; Deft-FFSD A, Rodden Rep., ¶¶39, 50; PLTF-5B, Engstrom Rebuttal, ¶ 15. Dr. McCoy is highly respected within the African American community, whose members viewed him as a very successful and effective superintendent. Joint Stip. ¶ 162; PLTF-117, Hudson Decl., ¶8; PLTF-119, Pruitt Decl., ¶ 22; PLTF-116, Deck of Charles Henson, Sept. 15, 2015 (“Henson Deck”), ¶ 15; PLTF-116, Deck of Doris Graham, Sept. 23, 2015 (“Graham Deck”), ¶ 18; Paulette-Thurman Dep., 82:1-3; Trial Tr. vol. 1, 45:8-47:12 (Pruitt testi

[Text truncated at 120,000 characters. The full text is on the page linked above.]

---

Source: Frix Law Library, https://www.frixlaw.com/law-library/cases/7239502. Public record. Not legal advice.