Opinion
Civil Action 22-178-SDD-SDJ
11-07-2023
RULING
SHELLY D. DICK, CHIEF DISTRICT JUDGE
Before the Court is a Joint Motion in Limine to Exclude Testimony and Reports of Dr. Lisa Handley filed by Defendant R. Kyle Ardoin, in his official capacity as Secretary of State of Louisiana, and Intervenor-Defendant the State of Louisiana, through Louisiana Attorney General Jeff Landry (collectively, “Movants”). The Plaintiffs have jointly filed a Memorandum in Opposition to the Motion in Limine.
Rec. Doc. 148.
Rec. Doc. 165.
Movants ask the Court to exclude the opinion testimony and reports of Dr. Lisa Handley under Federal Rules of Evidence 702 and 703, as well as Daubert and its progeny. Movants do not challenge Dr. Handley's qualifications as an expert in racially polarized voting. Movants argue that “the methodology she used here are neither reliable nor entirely relevant.” “Dr. Lisa Handley was retained by Plaintiffs to provide opinions about whether voting in the areas of Louisiana where Plaintiffs bring vote dilution claims is racially polarized. An analysis of racially polarized voting (‘RPV') is required to satisfy the Gingles II and III preconditions.”
Daubert v. Merrell Dow Pharms., Inc., 509 U.S. 579 (1993).
Rec. Doc. 148-1, p. 1.
Rec. Doc. 165, p. 1 (citing Thornburg v. Gingles, 478 U.S. 30, 55-56 (1986); Allen v. Milligan, 599 U.S. 1,22-23 (2023)).
LAW AND ANALYSIS
The Court must apply the familiar FRE 702 and Daubert analysis. Notably, a revision to Federal Rules of Evidence 702 is slated to become effective on December 1, 2023. The anticipated change clarifies that expert testimony may not be admitted “unless the proponent demonstrates to the court that it is more likely than not that the proffered testimony meets the admissibility requirements set forth in the rule.” Section 702(d) is being amended to include language that the “expert's opinion reflects a reliable application of the principles and methods to the facts of the case.” The intent of the proposed rule change is to focus and direct district courts to conduct the gate-keeping inquiry enunciated in Daubert and refrain from bypassing the admissibility determination in favor of a question of weight to be decided by a fact finder. The Committee Notes observe that “many courts have held that the critical questions of the sufficiency of an expert's basis, and the application of the expert's methodology, are questions of weight and not admissibility. These rulings are an incorrect application of Rules 702 and 104(a).”
COMMITTEE ON RULES OF PRACTICE AND PROCEDURE, REPORT OF THE JUDICIAL CONFERENCE COMMITTEE ON RULES OF PRACTICE AND PROCEDURE, E-11 (Sept. 2022) (accessible at https://www.uscourts.gov/rulespo|icies/pending-rules-and-forms-amendments).
Id. at E-10-E-11.
Id. at E-11.
Even though the Court, and not a jury, is the fact finder in this case, the Court will undertake the admissibility inquiry required by Rule 702 and will be guided by the imminent revision to Evidence Rule 702 and the Committee Notes.
I. RELIABILITY CHALLENGE
A. Data Sources
Movants contend that Dr. Handley's opinions are unreliable because her “database is derived from unknown sources and relies upon a flawed allocation method.” Movants argue that Dr. Handley relied on undisclosed sources to assist in compiling her database, namely the Voting and Elections Science Team, which assisted with shapefiles, and the ACLU, which assisted with aggregating data. And while they concede that experts may rely upon assistance in gathering underlying data and that “relying on others to assemble data is not a fatal flaw,” Movants submit that reliance on “undisclosed persons with unknown credentials to process data is unreliable.” Movants cite to the Seventh Circuit in Dura Automotive Systems of Indiana, Inc. v. CTS Corp.
Rec. Doc. 148-1, p. 6.
Id. p. 6, 8.
Id.
Id. at 8.
285 F.3d 609 (7th Cir. 2002).
The Court finds the Dura Automotive Systems case inapposite. In that case, the testifying expert's “assistants did not merely collect data . . . or otherwise perform routine procedures.” Furthermore, the court found that the proposed expert “lack[ed] the necessary expertise to determine whether the techniques were appropriately chosen and applied.” There is no evidence that the compilation of data by others, relied upon by Dr. Handley, was anything more than just that, a compilation of data. While the data may have been gathered by others, that does not render the data insufficient. Dr. Handley testified that she directed the gathering of the data and its compilation. There is no evidence before the Court that others performed analysis. There is no evidence that Dr. Handley relied on the opinions or expertise of undisclosed experts which lie beyond Dr. Handley's scope of expertise. A review of Dr. Handley's reports reveals that she performed the data analysis to reach her conclusions, and there is no challenge to Dr. Handley's expertise in analyzing voting patterns by race. The plaintiffs were not required to disclose the underlying data sources and gatherers under Rule 26(a)(B)(2) of the Federal Rules of Civil Procedure or the Court's scheduling order.
Id. at 615.
Id.
Rec. Doc. 165-1.
In the Dura case, undisclosed experts “constructed the model, and the ‘iterative process,'” which the Court found was beyond the scope of the testifying expert's expertise. 285 F.3d at 615.
B. Allocation of Vote Methodology
“Early and absentee votes are reported only at the parish level in Louisiana-they are not allocated back to the precinct where the voter resides. Rather than simply ignore these votes, they have been allocated to the parish precincts proportionally based on the votes received by each of the candidates on Election Day.” “Because the Louisiana Secretary of State website only reports candidate-specific early and absentee votes at the parish-wide level, Dr. Handley had to disaggregate the data down to the precinct level to perform her RPV analysis.” Movants urge exclusion of Dr. Handley's opinions on the grounds that this allocation of early and absentee votes is flawed and unreliable. Movants point out that owing to Dr. Handley's method of allocating the parish-wide early voter data proportionally to the precincts, total votes for certain candidates were overestimated in some precincts and underestimated in other precincts.
Rec. Doc. 165-1, p. 7.
Rec. Doc. 148-1, p. 9.
The simple fact is that early voting in Louisiana represents a statistically significant percentage of the total vote which must be accounted for. The data on early votes per candidate are collected by parish, but the scientifically accepted method for analyzing whether there is racially polarized voting (“RPV”) is the ecological inference analysis (“EI”), which requires precinct-level voting data-hence, why assumptions are made. In this case, Dr. Handley assumed the same allocation of election day votes per candidate per precinct to allocate the early votes per candidate per precinct. As explained by Dr. Handley in her report: Hypothetical parish Z is comprised of two precincts, Precincts A and B. Candidate #1 receives 80% of her votes on election day in Precinct A and 20% of her election day votes in precinct B. Hence, without specific early voting data available by precinct, Dr. Handley allocates 80% of the early election votes for Candidate 1 to Precinct A and 20% of the early election votes for Candidate 1 to Precinct B. Movants and Secretary of State Ardoin's expert, Dr. Tumulesh Solanky, point out that this allocation method results in over and underestimating the number of votes in some precincts.Movants argue that the resulting over and underestimated vote allocation represents bias and renders Dr. Hadley's opinions unreliable.
Rec. Doc. 165-1, p. 7, n.8.
For example, for the 2020 presidential election, Hadley allocated 191 votes to President Biden in Precinct 1 in Caddo Parish, but the entire voter turnout for that precinct was only 182 voters. See Rec Doc. 148-1, p.10.
The Court disagrees. Movants do not contend that EI is an improper analysis to evaluate RPV. Movants do not dispute that precinct-level data is necessary to run the EI analysis. The dispute is how best to de-aggregate or allocate the available parish-wide data down to usable precinct-level data. “Some challenges to expert testimony will raise matters of weight rather than admissibility.” The Court is persuaded that the slight over and underestimate of votes per precinct resulting from the subject allocation method is statistically insignificant and thus does not render the conclusions unreliable. “EI analysis is done using proportions of the vote share that each candidate received, . . . not raw total numbers that are input into the EI algorithm.”
COMMITTEE ON RULES OF PRACTICE AND PROCEDURE, REPORT OF THE JUDICIAL CONFERENCE COMMITTEE ON RULES OF PRACTICE AND PROCEDURE, E-12 (Sept. 2022) (accessible at https://www.uscourts.gov/rulespolicies/pending-rules-and-forms-amendments).
Rec. Doc. 165, p. 4 (citing Dr. Handley's deposition, Rec. Doc. 165-3).
Movants cite Overton v. City of Austin, a case in which the district court found the voting rights expert's RPV analysis seriously flawed and unreliable. The expert in Overton used “differing measures” for “the ethnic composition of precincts.” He used data from different sources to measure the size of different groups of voters. This flaw is not present here. Dr. Handley's data sets are consistent. The argument that Handley assumed homogeneity in voting across precincts is also unavailing. Day-of-voting data was available at the precinct level. The actual precinct-level voting is a direct measure of the votes cast and thus reflective of the different voting patterns among precincts. The allocation of early votes among precincts proportionally is logical, and Dr. Handley tested for confidence of the data. Finally, addressing Dr. Solanky's criticism of the allocation method, she tested for bias and found none. The Overton case is wholly distinguishable.
871 F.2d 529, 539 (5th Cir. 1989) (per curiam).
Id.
The expert in Overton used two different data sets to extrapolate the number of voters in each precinct: he used census data for Black voters and Spanish surnames on precinct voter registration lists for Hispanic voters. Id.
Rec. Doc. 165-1, at p. 6-7 (race data derived from Louisiana's voter registration data).
Rec. Doc. 165-6, p. 3, Handley's Supplemental Report (“To be certain that my opinion about the lack of bias is correct, I examined the possibility of allocation bias using two different approaches: I examined whether the voters of one political party were more likely to vote early than the other party; and I analyzed the voting patterns of early voters and election day voters separately to see if the degree of polarization among the two sets of voters differed substantially.”).
Movants further argue that the early vote allocation method has not been peer reviewed. But Dr. Handley testified in her deposition that other experts use the same methodology for allocation of early and absentee voting. Notably, even though Dr. Solanky disagreed with Dr. Handley's allocation methodology, he used Dr. Hadley's database when running his EI analysis. Additionally, Dr. Handley submitted a rebuttal report in which she performed two additional analyses to test the appropriateness of her early voting allocation method. These additional evaluations were consistent with and provided further support to her RPV conclusions. The Court does not find that the method employed by Dr. Handley to de-aggregate parish-wide numbers was the result of bias, and there is no evidence that it rendered the analysis infirm or the conclusions unreliable. The allocation assumptions can be challenged by cross-examination.
Rec. Doc. 165-3, Handley Dep. at 161:9-162:17.
Rec. Doc. 148-5, p. 13.
Rec. Doc. 165-6.
C. DISTRICT-SPECIFIC RPV
Movants argue that Dr. Hadley's opinions are irrelevant or not helpful to the trier of fact because she did not perform a “district-specific RPV analysis” but focused only on seven “areas of interest.” Movants submit that the EI analysis should have been performed state-wide.
Rec. Doc. 148-1, p. 2.
Movants argue that Dr. Handley classifies districts as either “effective” or not, without opining as to the level of Black Voting Age Population (“BVAP”) needed to be effective. In other words, Movants argue that Dr. Hadley's opinion is irrelevant because it fails to express the “threshold level of BVAP” necessary to provide black voters with a realistic opportunity to elect the candidate of their choice.
Id.
Rec. Doc. 148-1, p. 14.
Plaintiffs allege that “[t]he State Legislative Maps are dilutive” in part because “the Black Population in Louisiana is ‘sufficiently large and geographically compact to constitute a majority' in six to nine additional single-majority House districts and three additional single-member Senate districts.” Vote dilution claims are “district-specific.”
Rec. Doc. 14, p. 2.
Gingles, 478 U.S. at 103 (O'Connor, J., concurring).
RVP analysis must be specific to the areas of the state where the vote dilution is alleged to occur. The Fifth Circuit directs that it is error to “rely on statewide voting statistics to establish legally significant white bloc voting.” Dr. Handley's methodology included using election results from 16 different statewide elections, confining her EI analysis to the specific election data for the voters who live within each of the seven areas of interest. The Court finds that Dr. Handley performed a sufficiently local analysis of the challenged districts.
Magnolia Bar Ass'n, Inc. v. Lee, 994 F.2d 1143, 1151 (5th Cir. 1993).
Rec. Doc. 165-1, pp. 8-9. The seven areas of interest are the proposed new Black-majority districts created by Mr. Cooper's illustrative plans.
The Court finds that the Plaintiffs have demonstrated by a preponderance that Dr. Handley's opinion testimony will assist the Court as the trier of fact; her opinions are based on sufficient facts and data, the product of reliable principles and methods; and she reliably applied the principles and methods to the facts of the case.
The Motion in Limine (Rec. Doc. 148) is hereby DENIED.
IT IS ORDERED.