United States v. City of Eastpointe , 378 F. Supp. 3d 589 ( 2019 )

TERRENCE G. BERG, UNITED STATES DISTRICT JUDGE

*593The City of Eastpointe, Michigan uses an "at-large" voting method to elect its City Council. The United States ("the government" or "Plaintiff") has filed a complaint challenging this voting method, claiming that it dilutes the voting strength of African Americans who reside in Eastpointe and thereby violates Section 2 of the Voting Rights Act of 1965, which prohibits any voting practice or procedure that "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. § 10301(a). Named as Defendants in this lawsuit are the City of Eastpointe, the Eastpointe City Council, Mayor Susan Pixley, City Council members Cardi Demonaco, Jr., Michael Klinefelt, Sarah Lucido, and John Marion, as well as City Clerk Steve Duchane (together, "Defendants").

The parties have filed several motions that have been fully briefed, argued, and are now before the Court: Defendants' motion for summary judgment (ECF No. 26 ); Defendants' motion to exclude a certain kind of research methodology and its data, known as Bayesian Improved Surname Geocoding ("BISG"), which was relied upon by one of the government's expert witnesses, Dr. Lisa Handley (ECF No. 25 ); Defendants' motion to strike the government's notice of supplemental exhibit and corresponding exhibit (ECF No. 45 ); and Plaintiff's motion to strike Defendants' purportedly untimely supplemental expert disclosures (ECF No. 42 ). For the reasons set forth in detail below, all of the pending motions will be denied.

BACKGROUND

Eastpointe is a compact municipality of slightly over five square miles located in Macomb County, Michigan, on the northeast border of the City of Detroit. In considering challenges under the Voting Rights Act, the Supreme Court in the seminal case of Thornburg v. Gingles , 478 U.S. 30, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986) emphasized the importance of understanding the historical and social context in which the challenge is being raised, requiring that this Court conduct a "searching practical evaluation of the 'past and present reality' of the electoral system's operation." Id. at 45, 106 S.Ct. 2752 (quoting S. Rep. No. 417, 97th Cong., 2d Sess. 30 (1982) ). Because vote-dilution cases require a holistic rather than formalistic inquiry, the Supreme Court has emphasized the necessity of conducting "an intensely local appraisal of the design and impact" of the challenged electoral structure. Gingles , 478 U.S. at 79, 106 S.Ct. 2752 (quoting Rogers v. Lodge , 458 U.S. 613, 622, 102 S.Ct. 3272, 73 L.Ed.2d 1012 (1982) ).

In accord with the Supreme Court's admonition, the Court will discuss in some detail one of the expert reports before the Court. The report, authored by Dr. *594Thomas J. Sugrue, a noted 20th Century American historian, details how racial segregation and discrimination have been intertwined with Eastpointe's residential patterns, schools, and civic life since at least the early 20th Century. See generally ECF No. 38-13. According to the report, a combination of discriminatory real estate practices and public as well as private discrimination discouraged African Americans from moving to Eastpointe for several decades. Id. at PageID.1775-78. As recently as 1960 Eastpointe had no black residents. Id. at PageID.1773. In 1990, the City's African American population remained below 1%. Id. at PageID.1774. The City's decision in 1992 to change its name from East Detroit to Eastpointe is reported as emblematic of its desire to dissociate the predominantly white municipality from predominantly African American Detroit. Id. at PageID.1784-85.

Beginning in the 2000's, the City's African American population began to grow rapidly-part of a wave of what the expert report calls "black flight" out of the City of Detroit to the inner-ring suburbs. Id. PageID.1784, 1792-93. By 2010, African Americans comprised 30.3% of Eastpointe's population, mostly concentrated on the City's south side, along its Eight Mile Road boundary-line with the City of Detroit. Id. at PageID.1773, 1795-96. Consistent with this trend, information collected by the United States Census Bureau in the 2000 and 2010 censuses demonstrates Eastpointe's black voting age population ("VAP") rose from 4.25% in 2000 to 25.53% in 2010. ECF No. 38-2 PageID.1600 tbl. 2 (Expert Report of Dr. Lisa Handley). Similarly, data on citizen voting age population ("CVAP") by race gathered in the U.S. Census Bureau's American Community Surveys (the sole source of citizenship data published by the Census Bureau), which pool data from five-year periods, shows Eastpointe's black CVAP increasing from 21.45% in 2005-2009 to 36.65% in 2011-2015. Id. By the most recent census estimate, African Americans now comprise 41.2% of Eastpointe residents. ECF No. 38-13 at PageID.1773. The chart below, created by one of the government's experts, Dr. Lisa Handley, illustrates the fast-paced growth of Eastpointe's black voting age population and citizen voting age population.

Table 2: Total and Black Population in Eastpointe Year(s) Total VAP Black VAP Percent Black Census 2000 25744 1095 4.25% Census 2010 24103 6154 25.53% Total CVAP Black CVAP Percent Black ACS 2005-09 24360 5225 21.45% ACS 2006-10 24165 6034 24.97% ACS 2007-11 23590 6349 26.91% ACS 2008-12 23480 6945 29.58% ACS 2009-13 23371 7438 31.83% ACS 2010-14 23567 8034 34.09% ACS 2011-15 23670 8674 36.65%

*595ECF No. 38-2 PageID.1600 tbl. 2; see ECF No. 26 at PageID.614 ¶¶ 6-9 (indicating that Defendants accept these figures for summary judgment purposes).

Despite these rapidly changing demographics, Dr. Sugrue's report recounts that the City's long history of segregation and racial exclusion, as well as indifference towards integrating housing, prevented African Americans from becoming a cohesive part of the Eastpointe community. Complaints of harassment of African American residents remained troublingly common and at times violent. ECF No. 38-13 at PageID.1778-81, 1787-91. Additionally, statistics in the record demonstrate that major disparities exist between the City's black and white residents. Between 2006-2010, more than one in four of Eastpointe's black residents suffered unemployment, for example, compared to one in seven white residents. Id. at PageID.1810. During that same period, only about one in eleven white City residents lived beneath the poverty line, while more than one in five black residents did. Id. This racial poverty gap only widened between 2011 and 2015. Id. at PageID.1811. Income, employment, and poverty levels all affect the likelihood that an individual will vote. Id. at PageID.1820. And lower socioeconomic status is correlated with less participation in the political process. Id.

Eastpointe elects five individuals to serve on its City Council: the mayor and four City Council members. ECF No. 26 PageID.614 ¶ 2. Members of the City Council are elected at-large by all voters in Eastpointe to serve staggered, four-year terms. Id. ¶ 3. Under an at-large voting system, all voters in the jurisdiction can cast ballots for as many seats as there are up for election. The candidates who receive the most votes will represent the entire political subdivision. In contrast, in a single-member system, the jurisdiction is divided into sections and each section elects, and is represented by, a single elected official. As explained by the United States Supreme Court in Rogers v. Lodge , 458 U.S. 613, 616, 102 S.Ct. 3272, 73 L.Ed.2d 1012 (1982), "[a]t-large voting schemes and multimember districts tend to minimize the voting strength of minority groups by permitting the political majority to elect all representatives of the district." While a racial, ethnic, economic, or political minority "may be unable to elect any representatives in an at-large election," the minority "may be able to elect several representatives if the political unit is divided into single-member districts." Id.

At the time this lawsuit was filed, no African American can-didate had ever been elected to the Eastpointe City Council. ECF No. 1 PageID.5 ¶¶ 22-23. This is the case even though at least 39% of the City's total population and 34% of its CVAP are black, and black candidates have frequently run for City Council seats. ECF No. 26 PageID.614 ¶¶ 9-10, 25 (Defendants stipulate to the accuracy of these statistics solely for the purposes of considering this motion for summary judgment and reserve the right to contest them otherwise). Until the election of Council Member Monique Owens in the November 2017 election, after this lawsuit was filed, no black candidate had ever successfully run for Eastpointe's City Council. Id. PageID.614-615 ¶ 10. The next local election for the City of Eastpointe will be held on Tuesday, November 5, 2019 to elect the mayor and two city council members.

The difficulty inherent in evaluating vote-dilution claims is that, because of the secrecy of the ballot, it is not possible to gather precise data showing how voters from different racial or ethnic groups actually cast their individual votes. ECF No. 38-2 PageID.1598. Ballots do not require voters to identify their race, nor does the *596U.S. Census track voting patterns by race, or at all. Id. Because of the dearth of data tracking voting patterns by race, parties typically utilize statistical analysis of available demographic data to make an assessment of whether racial bloc voting exists. Monroe v. City of Woodville , 897 F.2d 763, 764 (5th Cir. 1990) ("Statistical proof of political cohesion is likely to be the most persuasive form of evidence, although other evidence may also establish the phenomenon."); United States v. City of Euclid , 580 F.Supp.2d 584, 596 (N.D. Ohio 2008) ("[C]ourts often begin with a statistical analysis of voting behavior" to assess "whether racial bloc voting exists."); Mallory v. Ohio , 38 F.Supp.2d 525, 537 (S.D. Ohio 1997) ("To determine the degree of racially polarized voting, one must conduct a statistical or other empirical analyses of the election results for each election at issue."), aff'd , 173 F.3d 377 (6th Cir. 1999).

Because no records are kept that specifically track the race of each voter¹ and how that voter casts his or her vote, the only way to estimate whether racial bloc voting is occurring is to gather data from precinct election results, compare that data to the demographic composition of the each precinct's electorate, and then use statistical models to estimate how black and white voters are likely to have voted in that precinct. Though they disagree about which type of data most accurately defines the correct demographic composition of each precinct's electorate, the parties' experts concur with each other that the appropriate methods of statistical analyses to assess racial bloc voting in this case are "ecological regression" and "ecological inference." ECF No. 26 PageID.615 ¶ 12 (explaining that Eastpointe's expert, Dr. John Alford, conducted ecological inference and regression analysis of past elections); ECF No. 38 PageID.1539 ¶ 127 (Pl.'s Counterstatement of Undisputed Material Facts) (stating that Dr. Handley applied these same methodologies).

Applied to voting rights cases, ecological regression estimates to what degree, if any, the vote for a candidate increases in a linear manner as the concentration of voters of a given race in the precinct increases. ECF No. 26-2 PageID.725-26 (Expert Report of Dr. John Alford); ECF No. 38-2 PageID.1605-06. In layman's terms, "regression is a mathematical technique for estimating the single best-fitting straight line that could be drawn to describe the relationship between two variables in a scatter plot." ECF No. 26-2 PageID.725; see Luna v. County of Kern , 291 F.Supp.3d 1088, 1118 (E.D. Cal. 2018) (providing a comprehensive explanation of ecological regression). Here, those variables are the concentration of black voters in the voting precinct, and the number of votes for a given candidate in that same precinct. ECF No. 38-2 PageID.1605-06. In this context, ecological regression assumes that the percentage of African Americans who vote for a given candidate is the same across all voting precincts. Id. at PageID.1605. "[I]f there is a relationship between the race of the voters and the votes cast for candidates then, for example, as the percentage of blacks in the precincts increases, the percentage of votes going to the black-preferred candidate also increases." Id. If there is a strong linear relationship across precincts between the percentage of black (or white) voters in a precinct, and the number of votes a candidate receives in the precinct, that relationship "can be used to estimate *597the percentage of blacks (or whites) voting for each candidate." Id. at PageID.1606.

Ecological inference, the second form of statistical analysis applied by the parties' experts, is largely regarded as an improvement upon ecological regression. City of Euclid , 580 F.Supp.2d at 598 ("King's EI [Ecological Inference] method is an improvement upon HPA [homogenous precinct analysis² ] and BERA [bivariate ecological regression analysis]."); Luna , 291 F. Supp. 3d at 1118 ("Ecological inference ... seeks to overcome some of the shortcomings of ER."). Ecological inference is another method of inferring individual behavior from aggregate data where actual individual-level data is unavailable. But unlike ecological regression, ecological inference does not rely on an assumption of linearity and instead incorporates "maximum likelihood statistics" and the "bounds method" to produce estimates of voting patterns by race. ECF No. 38-2 PageID.1606; ECF No. 26-2 PageID.727-28. The bounds method, according to Dr. Handley, "uses more of the data available from precinct returns as well as providing more of the information about the voting behavior being estimated." Id. PageID.1606. An example of how the bounds method works to preclude estimates of voting patterns by race that are less than 0% or more than 100% would be the case where a precinct has 100 voters, of which 75 are black and 25 are white, and an African American candidate had received 80 votes in that precinct. In estimating how the black voters voted, the bounds method takes into account that, because she received 80 votes, at least 55 (and at most 75) of the black voters must have voted for the African American candidate. Similarly, at least 5 and as many as 25 white voters must have cast their ballots for the African American candidate. Ecological regression, in contrast, can produce estimates of less than 0% or more than 100% because its estimates are not limited by the minimum and maximum percentage of votes a candidate could have received. ECF No. 38-2 PageID.1606 (providing a more detailed description of the ecological inference technique).

While the parties agree that ecological regression and ecological inference are the best techniques available to assess whether Eastpointe's atlarge scheme for electing city council members dilutes black citizens' votes, they disagree about the data sets to which these methodologies should be applied.

Specifically, Defendants take issue with Dr. Handley's use of data gathered using a method called Bayesian Improved Surname Geocoding ("BISG"). Dr. Handley used this data and method to estimate by race the number of Eastpointe voters who have participated in each City election since 2013. ECF No. 38-2 PageID.1602. BISG supplies a method to estimate the racial composition of a group of individuals in the absence of self-reported data on race. Id. PageID.1603. BISG assigns a *598probability of race or ethnicity to an individual based on two data points: surname and residential address. Id. It then updates this probability using the demographic characteristics of the census-block group associated with place of residence. CONSUMER FIN. PROT. BUREAU , USING PUBLICLY AVAILABLE INFORMATION TO PROXY FOR UNIDENTIFIED RACE AND ETHNICITY 6 (2014). Dr. Handley turned to BISG because VAP (again, "Voting Age Population") data from the 2010 census (the most recent census) may be outdated in light of the increases in the black population of Eastpointe since 2010, and neither VAP nor CVAP ("Citizen Voting Age Population") data records Eastpointe's voter turnout or voter registration by race. ECF No. 38-2 at PageID.1598-99.

While VAP data "is often used as a proxy for the demographic composition of the electorate in each precinct," such data does not necessarily match the reality of current demographics. Id. Because Eastpointe's African American population is growing quickly, Dr. Handley suspects 2010 VAP data reflects a black VAP much lower than Eastpointe's black VAP during the 2015 and 2017 elections. Id. Further, because minorities often turn out to vote at lower rates than white voters due to socioeconomic differences and historical discrimination, VAP data may suggest black voters turn out at higher rates than they do in reality. Id. For example, a precinct may be 25% black according to VAP data, but due to the impact of lower voter turnout, the percentage of black voters who participate in elections may be only 10%. Id. Essentially, if voter turnout is indeed lower among black Eastpointe voters than among white voters, using VAP or CVAP data to estimate voter turnout by race without any adjustment for lower turnout rates among black voters would estimate a higher-than-actual level of election participation by black voters as compared to white voters.³ Id. PageID.1599-1601. Defendants' expert, Dr. Alford, acknowledged this problem in another case, stating that "The problem [with relying on CVAP data] ... is that it assumes, without justification" that turnout is equal across different voting groups when, in fact, "Black and Latino populations have significantly lower turnout than White voters." ECF No. 32-8 ¶ 24 (Dr. John Alford Affidavit submitted in NAACP v. E. Ramapo Cent. Sch. Dist. , No. 7:17-CV-08943 (S.D.N.Y. filed Nov. 16, 2017) ).

Dr. Handley used BISG to generate data on voter turnout by race for the 2013 and later elections. Since 2013, Eastpointe has required all voters, including those voting absentee, to fill out applications listing the voter's name, residential address, and precinct. ECF No. 38-2 PageID.1602 n.17. Voters visiting the polls on election day are required to fill out the applications and absentee voters attach completed applications to their absentee ballots. Id. Dr. Handley obtained these applications and extracted voters' surnames and addresses from them. ECF No. 26-2 PageID.844. She then geocoded the addresses to determine which census block group⁴ each voter resided *599in. Id. Finally, Dr. Handley applied BISG to estimate the number of black voters, white voters, and voters of other races who have participated in each Eastpointe election since 2013. ECF No. 38-2 PageID.1602. BISG as applied here thus relies on information about individuals who actually participated in Eastpointe City Council elections. ECF No. 53 PageID.2543 (Aug. 1, 2018 Hearing Tr.). Census data, in contrast, provides only general demographic information about individuals residing in a given block group, who may or may not have participated in elections. Id. at PageID.2543-44.

The BISG analytical tool was first published in 2009, in Health Services and Outcomes Research Methodology, a peer-reviewed research journal. ECF No. 26-2 PageID.841 (Dr. Handley Deposition); Marc Elliott et al., Using the Census Bureau's Surname List to Improve Estimates of Race/Ethnicity and Associated Disparities , 9 HEALTH SERV. AND OUTCOMES RESEARCH METHODOLOGY 69 (2009). The Consumer Financial Protection Bureau has used this method to predict the race and ethnicity of mortgage applicants. And the agency reported that, in its experience, BISG produces data "that correlate[s] highly with self-reported race and national origin." CONSUMER FIN. PROT. BUREAU , USING PUBLICLY AVAILABLE INFORMATION TO PROXY FOR UNIDENTIFIED RACE AND ETHNICITY at 3. The Center for Medicare and Medicaid Services and several national health insurance companies also use BISG to refine race and ethnicity information in their data files. See generally Allen Fremont, et al., When Race/Ethnicity Data Are Lacking: Using Advanced Indirect Estimation Methods to Measure Disparities, 6 RAND HEALTH QUARTERLY (2016), https://www.rand.org/pubs/research_reports/RR1162.html; Dzifa Adjaye-Gbewonyo et al., Using the Bayesian Improved Surname Geocoding Method (BISG) to Create a Working Classification of Race and Ethnicity in a Diverse Managed Care Population: A Validation Study , 49 HEALTH SERV. RESEARCH 268 (2014).

While federal agencies and large-scale private sector corporations have found BISG to produce accurate data on race and ethnicity, the parties point out that no federal court has previously relied on this method to assess a claim for vote dilution under Section 2, or any other type of claim. See ECF No. 26-2 PageID.731 (Dr. Alford Expert Report); ECF No. 26-2 PageID.828, 842 (Dr. Handley Deposition). Dr. Handley stated in deposition testimony that a proprietary method similar to BISG, known as "Catalist," has been accepted by courts in vote suppression or voter identification cases. ECF No. 26-2 PageID.841-42. Dr. Eitan Hersch, a political scientist hired by the government to explain the use of BISG and BISG-type methods, explained that Catalist was accepted by the district court in Veasey v. Perry , 71 F.Supp.3d 627 (S.D. Tex. 2014) (crediting testimony and analysis of an expert who relied, in part, on Catalist data). ECF No. 38-6. Veasey was later affirmed in part, vacated in part, and reversed in part by the Fifth Circuit. Veasey v. Abbott , 830 F.3d 216 (5th Cir. 2016) (en banc).⁵ According to Dr. Hersch, Catalist's chief data *600scientist also testified in One Wisconsin Inst. v. Thomsen , 198 F.Supp.3d 896 (W.D. Wis. 2016), but the court's opinion in that case does not specifically reference his testimony or Catalist data. ECF No. 38-6 PageID.1693.

Further, Dr. Handley noted that Dr. Alford, Defendants' expert, in his recent litigation work for the State of Texas, "conducted election analysis using Spanish surname registration data in place of census demographic data.... using the same census surname list probabilities utilized in BISG analysis." ECF No. 38-5 PageID.1665, 1602 (providing explanation of how analysis of Spanish surname data has been used in the State of Texas to estimate the number of Hispanics registered to vote and participating in elections). While the Fifth Circuit has elsewhere expressed skepticism about reliance on Spanish-surname registration data in the place of census data, those concerns appear focused on "its tendency to misidentify Hispanic persons as non-Hispanic and vice-versa." Rodriguez v. Bexar Cnty., Tex. , 385 F.3d 853, 866 n.18 (5th Cir. 2004) (citing United States v. Alamosa Cnty. , 306 F.Supp.2d 1016, 1022 (D. Colo. 2004) (holding that expert testimony based on Spanish-surname data, while probative, should be afforded reduced weight as compared to self-identification data on race and ethnicity) ). It is unclear whether the same concerns would necessarily apply to using BISG to estimate the race of voters based on their names and their residential addresses, as Dr. Handley has done in Eastpointe. Further, the Sixth Circuit has not commented on the relative probative value of BISG or Spanish-surname-generated data.

While Dr. Handley's use of BISG is put at issue by Defendants' Daubert motion to "exclude any evidence, opinion, or testimony related to the government's utilization of [BISG]" it is important to note that Dr. Handley did not rely only on BISG to analyze Eastpointe citizens' voting patterns by race. ECF No. 26 PageID.592. To the contrary, she created separate tables illustrating the results of her ecological regression and inference analyses as applied to CVAP data, and to data created with BISG. Compare ECF No. 38-2 PageID.1615-16 tbl. 4 with PageID.1617 tbl. 4a and ECF No. 38-5 PageID.1680 tbl. 1 with PageID.1681 tbl. 2 (Dr. Handley's Supplemental and Rebuttal Expert Report). Additionally, for purposes of their summary judgment motion, Defendants have decided to accept the BISG results contained in Dr. Handley's expert reports. ECF No. 26 PageID.592. Accordingly, the Court will assume the admissibility of BISG data for purposes of analyzing Defendants' motion for summary judgment and will consider separately the merits of Defendants' motion to exclude the government's BISG data.

DISCUSSION

I. Defendants' Motion for Summary Judgment

In Thornburg v. Gingles , the Supreme Court set out three "pre-conditions" that must be established in order for the government to show a violation of Section 2 of the Voting Rights Act of 1965. First, the party challenging the election system must demonstrate that the minority group purportedly disadvantaged by the system is sufficiently large and geographically compact to constitute a majority in a single-member district. Second, it must be shown that the minority group is politically cohesive. And third, the plaintiff must show that the majority group votes sufficiently as a bloc to enable it, in the absence of special circumstances, "usually to defeat the minority's preferred candidate." Gingles , 478 U.S. at 49-51, 106 S.Ct. 2752. Defendants here contend the government *601cannot establish the first and third preconditions and that summary judgment in their favor is therefore warranted. After carefully reviewing the record and considering the arguments of the parties, the Court finds that material issues of fact preclude summary judgment for Defendants on the first and third preconditions.

Summary judgment is appropriate "if the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law." See Fed. R. Civ. P. 56(a). A fact is material only if it might affect the outcome of the case under the governing law. See Anderson v. Liberty Lobby, Inc. , 477 U.S. 242, 249, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986). On a motion for summary judgment, the Court must view the evidence and any reasonable inferences drawn from the evidence in the light most favorable to the non-moving party. See Matsushita Elec. Indus. Co. v. Zenith Radio Corp. , 475 U.S. 574, 587, 106 S.Ct. 1348, 89 L.Ed.2d 538 (1986) (citations omitted); Redding v. St. Edward , 241 F.3d 530, 531 (6th Cir. 2001).

The moving party has the initial burden of demonstrating an absence of a genuine issue of material fact. See Celotex Corp. v. Catrett , 477 U.S. 317, 325, 106 S.Ct. 2548, 91 L.Ed.2d 265 (1986). If the moving party carries this burden, the party opposing the motion "must come forward with specific facts showing that there is a genuine issue for trial." Matsushita , 475 U.S. at 587, 106 S.Ct. 1348. The trial court is not required to "search the entire record to establish that it is bereft of a genuine issue of material fact." Street v. J.C. Bradford & Co. , 886 F.2d 1472, 1479-80 (6th Cir. 1989). Rather, the "nonmoving party has an affirmative duty to direct the court's attention to those specific portions of the record upon which it seeks to rely to create a genuine issue of material fact." In re Morris , 260 F.3d 654, 655 (6th Cir. 2001). The Court must then determine whether the evidence presents a sufficient factual disagreement to require submission of the challenged claims to the trier of fact or whether the moving party must prevail as a matter of law. See Anderson , 477 U.S. at 252, 106 S.Ct. 2505.

To survive summary judgment, a plaintiff must present evidence sufficient to allow a finder of fact to conclude that three "necessary preconditions" to a successful Section 2 vote-dilution challenge are present. See Gingles , 478 U.S. at 50, 106 S.Ct. 2752. Once these threshold conditions are met, the court must then determine whether, "based on the totality of circumstances," the challenged electoral process impermissibly impairs the minority group's ability to elect representatives of its choice. Cousin v. McWherter , 46 F.3d 568, 574 (6th Cir. 1995) (discussing Gingles , 478 U.S. at 50-51, 106 S.Ct. 2752 ). In assessing the totality of circumstances, the Supreme Court in Gingles identified several factors relevant to determining whether a Section 2 violation has been established, drawing from the Senate Judiciary Committee Majority Report that accompanied the bill. See S. Rep. No. 417, 28-29, quoted in Gingles , 478 U.S. at 36, 106 S.Ct. 2752. The Senate Report lists seven "typical factors" that are relevant to conducting this totality of the circumstances analysis, as well as additional factors the Supreme Court has found to have probative value in some cases. See Gingles , 478 U.S. at 44-45, 106 S.Ct. 2752 (citing S. Rep. No 417, 28-29). Here, the Court need not address these factors or even discuss the totality of the circumstances inquiry in any detail because Defendants are seeking summary judgment only on the basis of the government's alleged inability to meet the first and third Gingles preconditions. See ECF No. 26 PageID.590-91.

*602A. The First Gingles Precondition-Minority group size and geographic compactness

The first Gingles precondition is straightforward. It requires that the minority group alleging vote dilution "be able to demonstrate that it 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. The reasoning behind this precondition is that unless it can be shown that minority voters have the potential to elect representatives of their choice absent the challenged political structure, they cannot claim to have been injured by that structure. See Mallory , 173 F.3d at 382 (citing Gingles , 478 U.S. at 50, 106 S.Ct. 2752 ). The Court finds that the evidence is sufficient to allow a fact-finder to conclude that the first precondition is established. Consequently, summary judgment on the first precondition in favor of the Defendants is inappropriate.

Satisfying the first Gingles precondition typically requires submitting a hypothetical redistricting plan that includes an electoral district with a greater-than-50-percent voting age minority population. NAACP v. Snyder , 879 F.Supp.2d 662, 671 (E.D. Mich. 2012) (To meet the first Gingles precondition, "the minority group must comprise at least a numerical majority of the proposed district's voting age population."); Gonzalez v. Harris Cnty., Tex. , 601 F. App'x 255, 258 (5th Cir. 2015) (stating that the first precondition is satisfied by supplying a hypothetical redistricting scheme in which at least one district includes enough minority voters to create a majority); Pope v. Cnty. of Albany , 687 F.3d 565, 576 (2nd Cir. 2012) (holding that "the first Gingles question is straightforward and statistical: does the identified minority group form at least a simple majority of the relevant population in the proposed district?").

The compactness inquiry also demands consideration of "traditional districting principles such as maintaining communities of interest and traditional boundaries." League of United Latin Am. Citizens v. Perry , 548 U.S. 399, 433, 126 S.Ct. 2594, 165 L.Ed.2d 609 (2006) (internal quotations and punctuation omitted). Accordingly, to establish the first Gingles precondition the government must also show that black voters live in a geographically compact pattern such that it is possible to draw a district "with a rational shape" that includes a majority of black voters. Mallory , 173 F.3d 377, 382-83 ; Perry , 584 U.S. at 402, 126 S.Ct. 2594 ("A district that reaches out to grab small and apparently isolated minority communities is not reasonably compact.") (internal quotations and citation omitted).

Here, it is undisputed that a single-member Eastpointe district could be drawn in which black residents would constitute the majority of both the VAP and CVAP (as explained above, VAP means "Voting Age Population," and is data gathered as part of the U.S. Census, while CVAP stands for "Citizen Voting Age Population," which is data gathered as part of the Census Bureau's American Community Survey). ECF No. 38-2 PageID.1595-97, 1613. Using data from the 2010 census and 2011-2015 American Community Survey, Dr. Handley created an illustrative four-district plan in which District 1 would include a black VAP and CVAP majority. Id. at PageID.1596-97. According to Dr. Handley's map, District 1 would encompass the area south of Toepfer Drive or east of Kelly Road, along boundaries with the City of Detroit and Interstate 94. ECF No. 38 PageID.1539 ¶ 126.

*603Under this illustrative redistricting plan, Eastpointe's black VAP would comprise 50.02% of District 1, 17.86% of District 2, 15.17% of District 3, and 18.79% of District 4. ECF No. 38-2 PageID.1597 tbl 1. And the City's black CVAP would be divided across the proposed districts as follows: 61.73% in District 1, 34.22% in District 2, 23.85% in District 3, and 26.19% in District 4. Id.

Based on Dr. Handley's map and her analysis of both 2010 census and 2011-2015 American Community Survey data, which Defendants have accepted as true and accurate for summary judgment purposes only, the black voting population in Eastpointe is sufficiently large and geographically compact to constitute a majority in a single-member district. See ECF No. 26 PageID.596 ("For purposes of this motion, Eastpointe accepts the results of the DOJ's statistical analysis as provided in Dr. Handley's expert reports."). The government has therefore cleared the hurdle of the first Gingles precondition by demonstrating a material issue of fact regarding existence of this precondition.

Defendants do not dispute that a geographically compact, single-member district in which black voters are a majority could be drawn in Eastpointe. Id. at PageID.592 ("One of those four district (sic) [in the government's illustrative plan] contains a majority of black voters."). Yet they argue the government cannot establish the first Gingles precondition because "black voters have attained a higher level of political representation from the current at-large system than their relative vote share, and because the government's illustrative redistricting plan would result in fewer black-preferred candidates being elected. Id. at PageID.591-92. Proportionality and efficacy of the illustrative redistricting plan, however, are not factors to be considered in assessing whether the government has met the first Gingles precondition.

"The ultimate end of the first Gingles precondition is to prove that a solution is possible, and not necessarily to present the final solution to the problem." Bone Shirt v. Hazeltine , 461 F.3d 1011, 1019 (8th Cir. 2006). Accordingly, the Supreme Court at this stage requires only that the government establish black voters could be a "simple majority" of voters in a single-member district.

*604Dickinson v. Ind. State Election Bd. , 933 F.2d 497, 503 (7th Cir. 1991) (citing Gingles , 478 U.S. at 50-51, 106 S.Ct. 2752 ); see Anthony v. Michigan , 35 F.Supp.2d 989, 999 (E.D. Mich. 1999) (finding plaintiff had met the first Gingles precondition simply by demonstrating that "over ninety percent of the African Americans in Wayne County live within a legally and politically defined jurisdiction: the City of Detroit."); City of Euclid , 580 F.Supp.2d at 594 (expressly rejecting the effectiveness measure proposed by Defendants and finding the illustrative redistricting plan created by Dr. Handley, an expert in that case as well, sufficient to establish the first precondition). As for proportionality, while it is part of the totality of the circumstances assessment, proportionality does not factor into the first precondition inquiry. See Cousin v. McWherter , 46 F.3d 568, 574 (6th Cir. 1995) (explaining that courts may only employ the totality of the circumstances analysis after determining that the three Gingles preconditions are met).

Defendants further urge that the Supreme Court's recent decision in Abbott v. Perez , --- U.S. ----, 138 S.Ct. 2305, 201 L.Ed.2d 714 (2018) requires this Court, in assessing whether the government can survive summary judgment on the first precondition, to determine whether the government's illustrative redistricting proposal would enhance the ability of Eastpointe's black voters to elect candidates of their choice. See ECF No. 53 PageID.2582 (Sep. 20, 2018 Hearing Tr.). The Court does not, however, read Abbott to expand the first Gingles precondition to require that a plaintiff prove the proposed illustrative redistricting plan would enhance the ability of minority voters to elect their preferred candidates. Rather, Abbott more broadly speculated that it would be difficult to see how a plaintiff alleging violation of Section 2 of the Voting Rights Act could ultimately show that "a districting decision dilutes the votes of minority voters.... if the alternative to the districting decision at issue would not enhance the ability of minority voters to elect the candidates of their choice." 138 S.Ct. at 2332. The Supreme Court was not specifically discussing the first Gingles precondition. Even if the Court credited Defendants' reading of Abbott , it would still find a genuine dispute of material fact regarding whether the government's illustrative redistricting plan would enable black voters in Eastpointe to elect their preferred candidates.

Because it is undisputed that black voters in Eastpointe are a sufficiently large and geographically compact group to constitute a majority in a single-member district, and Defendants' arguments to the contrary are unavailing, the government has met its burden of proof on the first Gingles precondition sufficient to survive summary judgment.

B. The Second Gingles Precondition-Political cohesion

The second Gingles precondition mandates that "the minority group must be able to show that it is politically cohesive." Gingles , 478 U.S. at 51, 106 S.Ct. 2752. The reasoning behind this precondition was that "if a minority group does not tend to vote together, the challenged electoral system cannot be responsible for the group's alleged inability to elect candidates of its choice." Mallory , 173 F.3d 377, 383.

To meet this condition at trial, the government will have the burden of establishing that black voters in Eastpointe have a candidate that they as a minority group would prefer to elect. Mallory , 173 F.3d 377, 383. For purposes of summary judgment, however, Defendants have conceded the second Gingles precondition is met. ECF No. 26 PageID.607-08 ("Defendants only concede the second Gingles precondition *605for purposes of this motion."). Accordingly, the Court will not undertake an analysis of the second Gingles precondition at this time. For summary judgment purposes, this condition is considered established.

C. The Third Gingles Precondition-Majority bloc-voting

The final Gingles precondition requires that "the minority must be able to demonstrate that the white majority votes sufficiently as a bloc to enable it-in the absence of special circumstances, such as the minority candidate running unopposed-usually to defeat the minority's preferred candidate." Gingles , 478 U.S. at 51, 106 S.Ct. 2752 (internal citation and citations omitted). While the second precondition asks only whether minority voters generally vote as a cohesive group, the third precondition assesses whether " 'a bloc-voting majority can routinely outvote' the minority, thereby 'impair[ing] the ability of a protected class to elect candidates of its choice.' " Cousin v. Sundquist , 145 F.3d 818, 823 (6th Cir. 1998) (quoting Johnson v. De Grandy , 512 U.S. 997, 1007, 114 S.Ct. 2647, 129 L.Ed.2d 775 (1994) ).

Critically, the inquiry under the third Gingles precondition "is not whether white candidates do or do not usually defeat black candidates, but whether minority-preferred candidates, whatever their race, usually lose." Cousin , 145 F.3d at 825 (emphasis added). Under Sixth Circuit precedent, the Court must consider not only Eastpointe City Council elections in which black and white candidates ran but also elections that involved only white candidates. Cousin , 145 F.3d at 825 ; see Rural W. Tennessee African-American Affairs Council , 209 F.3d at 840 ("This court has made clear that white-white elections are relevant in the analysis of a voting dilution claim."). Contests involving only white candidates will, however, be less probative than those involving black and white candidates-in large part because the Voting Rights Act's guarantee of equal opportunity is not met when "candidates favored by blacks can win, but only if the candidates are white." Id. (citing Smith v. Clinton , 687 F.Supp. 1310, 1318 (E.D. Ark. 1988) (three judge panel) ); see Nipper v. Smith , 39 F.3d 1494 (11th Cir. 1994) (explaining that elections involving only white candidates may be considered but are less probative than those that also include white candidates).

In light of Monique Owens's election to the Eastpointe City Council in 2017, it is notable that the Gingles Court specified that "the success of a minority candidate in a particular election does not necessarily prove that the district did not experience polarized voting in that election." Gingles , 478 U.S. at 31, 106 S.Ct. 2752. This is because special circumstances, such as the absence of an opponent or incumbency, "may explain minority electoral success in a polarized contest." Id. at 57, 106 S.Ct. 2752. When such special circumstances are present in an election, "there is less probative value" in it because the election is necessarily "not representative of the typical way in which the electoral process functions." Miss. State Conference of the Nat'l Ass'n for the Advancement of Colored People v. Ferguson-Florissant Sch. Dist. , 201 F.Supp.3d 1006, 1040 (E.D. Mo. 2016) (quoting Ruiz v. City of Santa Maria , 160 F.3d 543, 557-58 (9th Cir. 1998) ). While the Gingles Court did not provide more specific metrics to inform what types of situations may constitute special circumstances, the Sixth Circuit has suggested a special circumstance must be a dynamic that plays an "unusually important role in the election at issue."

*606Clarke v. City of Cincinnati , 40 F.3d 807, 813 (6th Cir. 1994) (discussing incumbency as a special circumstance).

Material issues of disputed fact make summary judgment on the third Gingles precondition inappropriate in this case. Defendants on the one hand say they have accepted the government's version of the material facts and the analysis of its expert, Dr. Handley, for summary judgment purposes. ECF No. 26 PageID.592 ("For purposes of this motion, Eastpointe accepts the DOJ's analysis. It accepts their facts."); ECF No. 53 PageID.2544-45 (reiterating that Defendants accept the government's facts and use of BISG data for the summary judgment phase). On the other hand, however, Defendants take issue with the government's analysis of key elections and in their motion for summary judgment urge the Court to credit the findings of their expert over the government's when it comes to assessing whether black voters were able to elect their preferred candidates in some of the most recent and probative elections. Compare, e.g., ECF No. 27 PageID.620 ¶ 57 ("Eastpointe's ecological inference analysis using ACS [American Community Survey] data indicates the white candidate [DeMonaco] received the highest point estimate from black voters in the February 2015 special election.") with ECF No. 38 PageID.1542-53 ¶ 146 ("Black voters supported Owens [the black candidate]" in the February 2015 special election.). Because there are factual disputes about whether, and in which elections, black voters in the City of Eastpointe have been able to elect their preferred candidates, summary judgment on the final Gingles precondition is inappropriate. The Court will discuss these factual disputes in greater detail below.

1. Endogenous elections in which both black and white candidates ran.

Both parties agree that "endogenous elections" with black and white candidates on the ballot are the most probative in assessing whether the government can meet the third Gingles precondition. ECF No. 38-2 PageID.1598; ECF No. 53 PageID.2546. "Endogenous elections" are those for the office at issue in the challenged voting process, here-elections for the Eastpointe City Council. ECF No. 38-2 PageID.1598. There have only been six Eastpointe City Council contests since 2007 in which both black and white candidates ran. ECF No. 26 PageID.617 ¶ 26.

Based on their respective experts' analyses of CVAP data (and Dr. Handley's BISG analysis), the parties agree black and white voters generally voted for the same candidates in the 2009 and 2011 Eastpointe City Council elections, and that those candidates won the open city council seats. ECF No. 26 PageID.617 ¶¶ 27, 30-31, 34; ECF No. 38 PageID.1543 ¶ 147; ECF No. 38-2 PageID.1616 tbl. 4. According to the government, while these elections included black candidates, those candidates were relatively unknown and did not garner significant support from black or white voters. ECF No. 38 PageID.1543 ¶ 147; ECF No. 38-2 PageID.1616. Accordingly, in at least two of the six elections considered most probative by the parties, black voters were able to elect their preferred candidates; but those candidates were white.

In the four more recent contested City Council elections, which took place between 2015 and 2017, black and white voters may have supported different candidates-the parties are not entirely in agreement. Compare ECF No. 38 PageID.1540 ¶ 136 ("... black voters have cohesively supported black candidates.... White voters have cohesively supported white candidates, and each white candidate has finished with more votes than any black candidate.") with ECF No. 41 PageID.2115 *607¶ 136 (disputing that the government's data supports the conclusion that white and black voters supported candidates of their own race in these elections, and that white candidates received more votes).

In the February 2015 special election, one white candidate, Cardi DeMonaco, Jr., and two black candidates, Monique Owens and Serina Pinkston, ran for the single open seat. ECF No. 38-2 PageID.1609, 1615 tbl. 4. DeMonaco was the incumbent. ECF No. 38-14 PageID.1882 (Cardi DeMonaco Deposition). The government's expert, Dr. Handley, found that most black voters supported Owens, who received 53.4% of black votes according to CVAP demographic data and 94.7% of black votes based on ecological inference analysis of BISG data.⁶ ECF No. 38-2 PageID.1615 tbl. 4, PageID.1617 tbl. 4a. Owens ran a sophisticated campaign to increase turnout among black voters and garner their support in the election. ECF No. 38-15 PageID.1886-88 (Monique Owens Deposition). But white voters overwhelmingly supported DeMonaco in the election; she received at least 81.7% of white votes according to Dr. Handley's analysis of CVAP data, and at least 76.8% when BISG is used. Id. at PageID.1615 tbl. 4, PageID.1617 tbl. 4a.

Dr. Alford's analysis of CVAP data yielded somewhat different results for the February 2015 election (although he describes them as "extremely similar to the [CVAP] estimates reported by Dr. Handley."). ECF No. 26-2 PageID.732. Applying ecological regression to CVAP data, Dr. Alford determined that 52.8% of black voters cast ballots for Owens while 40.5% voted for DeMonaco, the winning candidate. ECF No. 26-2 PageID.733 tbl. 1. Yet his use of ecological inference yielded opposite results, suggesting 60.1% of black voters actually voted for DeMonaco while only 29.7% voted for Owens. Id. Dr. Alford's results are thus inconclusive while Dr. Handley's CVAP results show black voters preferred Owens, and her BISG methodology indicates black voters overwhelmingly voted for Owens, who was ultimately unsuccessful in her race. DeMonaco received 60.2% of the overall vote, winning the open seat. ECF No. 38 PageID.1542 ¶ 146; ECF No. 38-2 PageID.1616 tbl. 4 cont'd. Owens won just 28.5% of the overall vote. Id.

In addition to the fact that Dr. Alford's and Dr. Handley's studies of the February 2015 election yielded largely inconsistent results, the parties differ in their conclusions on the critical question of which candidate black voters supported in this election. Defendants urge the Court to accept their ecological inference estimates, which "indicate the white candidate, DeMonaco, was the black-preferred candidate."⁷ ECF No. 27 PageID.606. And the government contends "[b]lack Voters preferred Monique Owens." ECF No. 38 PageID.1580 ¶ 55. Because this February 2015 election is one of only four endogenous elections in which both black and white candidates ran, and in which the government alleges Eastpointe's at-large voting system enabled the *608white majority voting bloc operated to prevent black voters from electing their preferred candidate, the facts surrounding which candidate black voters preferred in this election are material. They are also plainly in dispute.⁸

The parties also disagree to some extent on whether black voters were able to elect their preferred candidate in the November 2015 City Council election for two open full-term seats. Three candidates ran in that election: Sarah Lucido and John Marion, who are white, and Alexandria Bibb Williams, who is black. ECF No. 38 PageID.1541 ¶ 142. According to the government, Williams was black voters' preferred candidates and Lucido their second choice. Id. ¶ 144. Defendants, however, assert "[b]lack voters' first choice for the November 2015 elections [was] Lucido." ECF No. 26 PageID.616 ¶ 20. Lucido, who won 47.7% of the overall vote, and Marion, who won 31.8%, were elected to the city council. ECF No. 38-2 PageID.1615 tbl. 4. Williams received only 20.5% of the overall vote and accordingly did not win a City Council seat. Id.

Both Dr. Handley and Dr. Alford's ecological inference and ecological regression analyses of CVAP data show Lucido received the most support among black as well as white voters, while Williams received the second-most support from black voters (but minimal support from white voters). Id. ; ECF No. 2602 PageID.733 tbl. 1. Specifically, Dr. Handley's application of ecological inference and ecological regression to CVAP data shows Lucido received between 42.3% and 44.6% of black votes, and Williams between 30.5% and 32.2%. ECF No. 38-2 PageID.1615 tbl. 4. But Dr. Handley's BISG analysis suggests Williams-not Lucido-was black voters' first-choice candidate, and that Williams received between 79.2% and 87.1% of black votes while Lucido received between 66.9% and 70.6%. ECF No. 38-2. PageID.1617 tbl. 4a.

Defendants' analysis of the November 2015 election equivocates between the ability of Eastpointe's black voters to elect their first-choice as opposed to second-choice candidate. While Defendants say they have accepted the results of Dr. Handley's BISG methodology for summary judgment purposes, their analysis largely ignores the fact that BISG shows Williams, the black candidate, (not Lucido) was black voters' preferred candidate. ECF No. 26 PageID.618 ¶ 44, PageID.619 ¶ 49. Specifically, Defendants conclude that because Lucido was elected to the City Council, black voters were able to elect their preferred candidate in the November 2015 election. But this fails to consider the BISG data showing that Lucido was black voters' second-choice candidate, not their first. ECF No. 38-2 PageID.1617 tbl. 4a.

Throughout their analysis of the third Gingles precondition, Defendants lean heavily on their assertion that "black-preferred candidates win 60% of the time in the most probative elections - elections involving black and white candidates for the Eastpointe City Council. Id. PageID.590. Closer examination of this figure reveals that it in fact includes this February 2015 election in which, based on BISG data, black voters were only able to elect their second-choice candidate, Lucido, who happened to also be white voters' first-choice candidate. ECF No. 38-2 PageID.1615 tbl. 4; ECF No. 26-2 PageID.733 tbl. 1; ECF No. 38-2 PageID.1617 tbl. 4a. Marion, who was white *609voters' second-choice candidate, won the other open seat. ECF No. 38-2 PageID.1615 tbl. 4. Accordingly, based on both CVAP and BISG data, white voters successfully elected both their first and second-choice candidates to the City Council in this election while black voters, as indicated by BISG data, were only able to elect their second-choice candidate to one of the available seats; the other seat was won by black voters' last-preferred candidate, Marion. Id. PageID.1617 tbl. 4a. Disagreement between the parties as to whether black voters were able to elect their preferred candidate in the November 2015 election is another dispute of material fact that weighs in favor of denying summary judgment on the third Gingles precondition in this case.

The most recent endogenous elections in which both black and white candidates ran are the November 2017 Eastpointe City Council elections to fill one partial-term seat, and two full-term seats. ECF No. 38-5 PageID.1658 (Supplemental and Rebuttal Expert Report of Dr. Lisa Handley). Defendants' expert has not provided any analysis of the 2017 elections, nor do Defendants address those elections in detail in their brief. ECF No. 26. Because Defendants have accepted Dr. Handley's analysis for summary judgment purposes, the Court relies only on the data set forth in her reports concerning the 2017 election. See ECF No. 26 PageID.596.

Two candidates competed for the available partial-term seat in the 2017 election: Tonia Gladney, who is African American, and Michael Klinefelt, who is white. ECF No. 38-5 PageID.1658. It is undisputed that black voters supported Gladney over Klinefelt. ECF No. 26 PageID.619 ¶ 46; ECF No. 38 PageID.1542 ¶ 141. Dr. Handley's analysis of CVAP data shows between 52.6% and 60% of black voters supported Gladney while white voters overwhelmingly supported Klinefelt. ECF No. 38-5 PageID.1680 tbl.1. BISG data indicates as many as 79.3% of black voters supported Gladney. Id. PageID.1681 tbl. 2. Both Dr. Handley's CVAP and BISG results thus demonstrate Gladney was black voters' preferred candidate in the 2017 election for a single partial-term seat. Yet Klinefelt received 65% of the overall vote, and Gladney only 35%. Id. PageID.1680 tbl.1. The results of this 2017 election therefore suggest that Eastpointe's white voters were able to elect their preferred candidate over black voters' preferred candidate in the 2017 election for one partial-term seat.

Six candidates initially entered the competition for two fullterm seats up for election in the general November 2017 City Council race. ECF No. 38 PageID.1541-42 ¶ 137. Only two candidates, Cardi DeMonaco, Jr. and Michael Klinefelt, are white, and Klinefelt dropped out of the race to run instead for the open partial-term seat. Id. The remaining candidates: Clarence Duren, Edward Williams, Monique Owens, and R.J. Johnson, are all black. Id. Dr. Handley's CVAP analysis shows black voters' first-choice candidate was Owens, and their second choice Johnson. ECF No. 38-5 PageID.1680 tbl. 1. Those two candidates received the least support among white voters, whose first-choice candidate was DeMonaco, the only white candidate left in the race after Klinefelt dropped out. Id. BISG analysis yields similar conclusions for this election but suggests black voters slightly preferred Johnson to Owens. Id. PageID.1681 tbl. 2. Analyses of both data sets establish Owens and Johnson were the two preferred candidates among black voters while white voters by a great margin supported DeMonaco, the only remaining white candidate. It was DeMonaco who ultimately received the most overall votes, winning 30% of the electorate in Eastpointe. Id. Because there were two City *610Council seats up for election, and only one white candidate in the running, the election of an African American candidate to one of the two seats was essentially a foregone conclusion. Owens came in second in the election, winning 23.1% of the vote and thereby securing the second available City Council seat. Id. With two open seats, and only one white, but three black, candidates running, Monica Owens became the first African American to be elected to the Eastpointe City Council. ECF No. 38 PageID.1562.

Because there were fewer white candidates than open seats in the 2017 general City Council election, the government contends Owens won election under "special circumstances" and therefore that her election should not weigh against finding that Eastpointe's white majority votes sufficiently as a bloc to enable it usually to defeat the minority-preferred candidate. See Gingles , 478 U.S. at 90, 106 S.Ct. 2752. The Court agrees. Gingles expressly contemplated that an election with a "minority candidate running unopposed" would be a special circumstance that mandates giving that election less probative weight in assessing whether the third precondition is met. Gingles , 478 U.S. at 51, 106 S.Ct. 2752 ; see Harvell v. Blytheville Sch. Dist. No. 5 , 71 F.3d 1382, 1389 (8th Cir. 1995) ("Even in an extreme case of total vote dilution a candidate running in the face of no opposition is ensured success."). Yet because Owens was black voters' first-choice candidate in this election, Defendants consider the 2017 general City Council election one in which black voters were able to elect their preferred candidate. ECF No. 26 PageID.597-98. They also counted this election in determining that black voters successfully elected their preferred candidate in 60% of endogenous elections in which both black and white candidates ran. Id. Defendants failed in any way to differentiate this election from one in which election of an African American candidate was not predetermined, further calling into question their 60% figure. Common sense counsels that conditions in which an African American candidate was guaranteed election to the City Council should not be given great weight when trying to prove or disprove the existence of vote dilution. Further, under Gingles , the availability of more seats than white candidates in the 2017 general City Council election demands that results of that election be given less probative weight in determining whether Eastpointe's white majority votes sufficiently as a bloc to enable it usually to defeat black voters' preferred candidate.

Factual disputes about whether black voters were able to elect their preferred candidates in recent endogenous elections involving both black and white candidates-undisputedly the most probative here-preclude summary judgment on the third Gingles precondition. The Court's finding that Owens, the only black candidate to successfully run for the Eastpointe City Council, was elected in an election characterized by special circumstances also weighs in favor of denying summary judgment on the final Gingles prerequisite.

2. Endogenous elections in which only white candidates ran.

The parties agree that Eastpointe City Council contests involving only white candidates generally have not been racially polarized. ECF No. 38-2 PageID.1612; ECF No. 26 PageID.623 ¶¶ 78-79. These elections are, however, less probative because the fact that black voters also support white candidates acceptable to the majority does not negate instances in which a white voting majority operates to defeat the candidate preferred by black voters when that candidate is a minority. See Clarke , 40 F.3d at 812 (citing *611Citizens for a Better Gretna v. City of Gretna, La. , 834 F.2d 496, 502 (5th Cir. 1987) ); see also Rural W. Tenn. African-Am. Affairs Council , 209 F.3d at 841 (expressing disapproval of notion that elections involving both black and white candidates, and elections involving only white candidates, have equal probative value).

Since 2007, there have been three contests for City Council seats and three City mayoral elections where only white candidates competed. ECF No. 26 PageID.623 ¶ 76. According to Dr. Handley, black and white voters supported the same candidates in the 2007, 2001, and 2015 mayoral contests, and also appear to have supported the same two candidates in the November 2007 and 2013 general City Council elections,⁹ and 2009 City Council election to fill a partial-term seat. ECF No. 38-2 PageID.1612, 1626-29 tbls. 9-9a.

3. Exogenous Elections

While not as instructive as endogenous elections, "exogenous elections"-those elections for offices other than those which are at issue in this lawsuit-still hold some probative value. See Cousin , 145 F.3d at 824 (finding expert analyses of exogenous elections at least somewhat probative where there was a "dearth of especially relevant contests."). Because there are only six elections for the Eastpointe City Council in which both black and white candidates ran, the government chose also to analyze Eastpointe voting patterns for recent local and state elections in which black candidates competed. ECF No. 38-2 PageID.1618-25 tbls. 5-8. Defendants maintain that exogenous elections have minimal probative value and so did not conduct their own analysis of those elections. See ECF No. 26 PageID.601. They did, however, analyze Dr. Handley's analyses of various exogenous elections,

Dr. Handley analyzed the following exogenous elections: 2009 and 2014 East Detroit school board elections; August 2016 primary election and November 2016 general elections for Macomb County Circuit Court Judge; 2010, 2012, and 2014 elections for the Michigan Supreme Court; and the 2012 Macomb Community College Board contest. ECF No. 38-2 PageID.1618-25 tbls. 5-8. Exogenous elections that "occurred on the same day as City Council elections, involved the same citywide electorate, and featured African-American candidates" are more probative than other exogenous elections. See City of Euclid , 580 F.Supp.2d at 598. Of the elections Dr. Handley studied, only the 2014 East Detroit Public Schools election involved nearly the same electorate as a City Council race. ECF No. 38 PageID.1568. The other races encompassed the entire county or state's electorate and are therefore less helpful.

The November 2009 East Detroit School Board election was the only exogenous contest to include a black candidate that was held contemporaneously with a City Council election and involved the same electorate. ECF No. 38 PageID.1545 ¶ 156. Dr. Handley's analysis of this election using CVAP data and applying both ecological regression and ecological inference shows black voters' preferred candidate was Calvin Washington, an appointed incumbent and the only black candidate in the race. ECF No. 38-2 PageID.1618 tbl. 5. But white voters preferred the other four candidates, all of whom are white. Id. Washington finished last among the five candidates, receiving only 15% of the overall *612vote. Id. In the November 2014 East Detroit School Board election, three candidates ran for three open seats. ECF No. 38 PageID.1545 ¶ 158. While this election is not especially probative given that all three candidates were elected by default, white voters preferred the two white candidates over Charley Jackson Jr., the third candidate, who identifies as black or mixed-race. Id. ; ECF No. 38-2 PageID.1618 tbl.5. Notably, black voters appear to have slightly favored Julie DeVita, one of the white candidates, over Jackson when CVAP data is analyzed but Jackson to DeVita when BISG is used. ECF No. 38-2 PageID.1618 tbl.5, PageID.1619 tbl.5a. Results of the November 2009 and November 2014 East Detroit School Board elections further suggest there is, at minimum, a triable issue of fact as to whether white voters in the City of Eastpointe vote sufficiently as a bloc in a manner sufficient usually to defeat black voters' preferred candidate.

The Court has examined the parties' respective analyses of the remaining exogenous elections examined by Dr. Handley but finds that material issues of fact remain as to whether African American voters in Eastpointe were able to elect their preferred candidates in exogenous elections. For instance, Dr. Handley, the government's expert, asserts that all six of the contested exogenous elections that included black candidates were racially polarized and that, while black-preferred candidates were more successful in these exogenous contests than in Eastpointe City Council races, these were elections that coincided with a presidential race. ECF No. 38-2 PageID.1612. In those elections, she contends that black voters made up a larger share of the electorate than in other elections. Id. Defendants, in turn, claim "the Court can objectively quantify that black-preferred candidates win 66% of the time in exogenous elections involving both white and black candidates." ECF No. 26 PageID.595-96.

There are genuine disputes between the parties as to numerous facts material to assessing the third Gingles precondition. Principally, while Defendants say they accept the analysis of the government's expert, Dr. Handley, for summary judgment purposes, including her use of BISG data, they repeatedly dispute her factual findings, or rely on different factual findings that contravene Dr. Handley's analysis. This is particularly true for endogenous elections involving black and white candidates. Because the parties disagree about material facts central to the elections undisputedly most probative here, the Court declines to enter summary judgment in favor of Defendants on the third Gingles precondition.

II. Defendants' Motion to Exclude BISG Data and Methodology

Defendants next ask the Court "to exclude any evidence, opinion, or testimony related to [the government's] use of the BISG method in this case." ECF No. 25 PageID.161. Specifically, Defendants argue BISG-created data is inadmissible under Rule 702 of the Federal Rules of Evidence because Dr. Handley's analysis relies on "insufficient, untested, and unreliable methods to arrive at its conclusions." Id.

The admissibility of expert testimony under Rule 702 "entails a flexible inquiry, and is addressed to the trial judge's discretion." Rochow v. Life Ins. Co. of N. Am. , No. 04-73628, 2011 WL 13208692, at *1 (E.D. Mich. Mar. 4, 2011) (citations omitted). The Rule provides that:

If scientific, technical, or other specialized knowledge will assist the trier of fact to understand the evidence or to determine a fact in issue, a witness qualified *613as an expert by knowledge, skill, experience, training, or education, may testify thereto in the form of an opinion or otherwise, if (1) the testimony is based upon sufficient facts or data, (2) the testimony is the product of reliable principles and methods, and (3) the witness has applied the principles and methods reliably to the facts of the case.

Rule 702 thus allows the trial court to act as a gatekeeper in barring unsupported, unreliable and speculative expert opinion from trial. See Daubert v. Merrell Dow Pharmaceuticals , 509 U.S. 579, 597, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993) ; Kumho Tire Co. v. Carmichael , 526 U.S. 137, 152, 119 S.Ct. 1167, 143 L.Ed.2d 238 (1999).

"The task for the district court in deciding whether an expert's opinion is reliable is not to determine whether it is correct, but rather to determine whether it rests upon a reliable foundation, as opposed to, say, unsupported speculation." In re Scrap Metal Antitrust Litigation , 527 F.3d 517, 529-30 (6th Cir. 2008). Because the gatekeeper doctrine was designed to protect juries, the Sixth Circuit has described it as "largely irrelevant in the context of a bench trial." Rochow , 2011 WL 13208692, at *2 (quoting Deal v. Hamilton Cnty. Bd. of Educ. , 392 F.3d 840, 852 (6th Cir. 2004) ). A trial judge is "presumed capable of weighing evidence to sift the important from the unimportant, and even the admissible from the inadmissible when those are intertwined in a way that might counsel excluding the same evidence from consideration by a lay jury." UAW v. Gen. Motors Corp. , 235 F.R.D. 383, 387 (E.D. Mich. 2006).

Because the parties have agreed to a bench trial of this matter, traditional gatekeeping concerns that might arise in a jury trial are diminished. Dr. Handley, who applied the BISG data-creation method at issue in Defendants' motion to exclude, has provided sufficient facts and data to support the reliability of BISG data in this case, and to show that it was applied in a reliable manner. See, e.g. , ECF No. 38-2 PageID.1632-34 (Appendix to Dr. Handley Expert Report). The explanation contained in her three expert reports, as well as the peer-reviewed analysis of BISG contained in articles she cites, satisfies the Court that her testimony will be the product of reliable principles and methods, applied reliably to the facts of this case. See generally ECF Nos. 38-2, 38-5, 38-7; see also infra at 12-19. Similarly, Dr. Hersch's report on the increasing prevalence of methods of predicting racial identity to reduce uncertainty in racial bloc voting analyses addressed some of the Court's concerns about reliance on BISG. See ECF No. 38-6.

Because the Court is well equipped to weigh Dr. Handley's application of BISG in relation to other evidence submitted in this matter, and finds that the parties have thoroughly informed it of the benefits and risks of BISG, the Court denies without prejudice Defendants' motion to exclude evidence, opinion, and testimony related to the methodology and data, including Defendants' alternative request to appoint an independent expert to evaluate the government's BISG methodology and data. See ECF No 53 PageID.2538 (denying without prejudice Defendants' motion to exclude BISG methodology and testimony in open court). Prior to trial, should they wish to do so, Defendants may refile their Daubert motion in limine.

III. Defendants' Motion to Strike the Government's Notice of Supplemental Exhibit and Corresponding Exhibit

The Court denies with prejudice Defendants' motion to strike Plaintiff's supplemental exhibit, which is a copy of *614the court-appointed expert's report in Anthony v. Michigan , 35 F.Supp.2d 989 (E.D. Mich. 1999). See ECF Nos. 44, 44-1. The Court rejects Defendants' characterization of the notice and accompanying exhibit as an unauthorized sur-reply. Accordingly, there is no violation of Rule 56(d) of the Federal Rules of Civil Procedure or Local Rule 7.1(c)(3). The Court further determines that accepting the supplemental exhibit will serve the interest of developing a complete record and cause no prejudice to Defendants, who cite liberally to the Anthony case throughout their summary judgment briefing. See Johnson v. Cnty. of Wayne , No. 08-CV-102009, 2008 WL 4279359, at *8 (E.D. Mich. Sep. 16, 2008) (accepting supplemental exhibit in the interest of developing a full record where it caused no prejudice to parties).

IV. The Government's Motion to Strike

The Court also denies without prejudice Plaintiff's motion to strike Defendants' supplemental expert disclosures and briefing that Plaintiff contends were untimely filed. See ECF No. 53 PageID.2540 (statement by the Court during Aug. 1, 2018 hearing that Plaintiff's motion to strike is denied without prejudice). Again, in the interest of developing a complete record, the Court will consider Defendants' supplemental expert disclosures and briefing.

CONCLUSION

For the foregoing reasons, this Court DENIES Defendants' Motion for Summary Judgment (ECF No. 26 ), DENIES without prejudice Defendants' Motion to Exclude BISG Data and Methodology (ECF No. 25 ), DENIES with prejudice Defendants' Motion to Strike Notice of Supplemental Exhibit and Supplemental Exhibit (ECF No. 45 ), and DENIES without prejudice Plaintiff's Motion to Strike (ECF No. 42 ).

SO ORDERED.

1 Dr. Handley points out that "only a handful of Southern states" collect pre-cinct-by-precinct data showing voter turnout by race and voter registration data by race. No such data is available for Eastpointe. ECF 38-2 PageID.1598.

2 A third statistical methodology, homogenous precinct analysis, is frequently applied in vote-dilution cases and has also been relied on by courts. See, e.g. , Rural W. Tenn. African-Am. Affairs Council v. Sundquist , 209 F.3d 835, 839 (6th Cir. 2000) (approving district court's reliance on homogenous precinct and regression analysis); City of Euclid , 580 F. Supp. 2d at 596 (crediting homogenous precinct analysis, ecological regression, and ecological inference in vote-dilution case). But homogenous precinct analysis compares the percentage of votes received by candidates in precincts that are considered racially homogenous, that is, where 90% or more of the voting age population (VAP) is all of one race. In Eastpointe, none of the individual precincts have a VAP that is more than 90% black or white. Consequently, this method cannot be applied here. ECF No. 38-2 PageID.1605.

3 Dr. Handley provides at least one concrete reason why turnout may have been lower among black voters in Eastpointe. Michigan law until 2019 limited no-excuse absentee voting to registered voters 60 years or older. 25.8% of Eastpointe's white population is over the age of 60 as compared to 11.2% of the city's black population. ECF No. 38-2 PageID.1601-02 n.13. Accordingly, a greater percentage of the City's white voters benefitted from easier access to absentee voting during the elections at issue in this case and thus may have been more likely to vote in those elections. ECF No. 38-2 PageID.1601 n.11.

4 Census-block groups are statistical divisions of census tracts. "A block group consists of clusters of blocks within the same census tract that have the same first digit of their four-digit census block number." U.S. Census Bureau , Geographic Terms and Concepts - Block Groups (2010), https://www.census.gov/geo/reference/gtc/gtc_bg.html.

5 On appeal, the State of Texas did not dispute the underlying data or methodologies relied on by the district court. Abbott , 830 F.3d 216. Additionally, the Fifth Circuit did not take issue with the district court's crediting of expert analysis that relied partly on Catalist data. 830 F.3d 216.

6 Plaintiff's expert only conducted ecological regression analysis of American Community Survey data for the February 2015 election because she determined ecological inference estimates based on American Community Survey data alone were not plausible. Defendants' expert conducted both ecological regression and inference analyses of the February 2015 election; the results of the former indicate Owens was the black-preferred candidate while the latter suggested DeMonaco was. ECF No. 26-2 PageID.733 tbl. 1.

7 Defendants provide no explanation as to why the Court should give greater weight to Dr. Alford's ecological inference results than his ecological regression results.

8 At another point in their brief, Defendants state they will accept that Owens was black voters' preferred candidate in the February 2015 election for purposes of their summary judgment motion. ECF No. 26 PageID.601.

9 Dr. Handley expressed that there is a lack of clarity regarding whether white-preferred candidates won both City Council seats in the November 2013 election. ECF No. 38-2 PageID.1612; ECF No. 38 PageID.1583 ¶ 82.