Opinion
CIVIL ACTION NO. 4:12-CV-2579
04-25-2014
GILBERT JOE CISNEROS, et al, Plaintiffs, v. PASADENA INDEPENDENT SCHOOL DISTRICT, et al, Defendants.
FINDINGS OF FACT AND CONCLUSIONS OF LAW
The parties in this case are Plaintiffs Gilbert Joe Cisneros, Catherine Garcia Sonnier, Martha Gonzalez, Salvador Martinez, Edward Ybarra, Arnold Hurtado, and Melinda Villarreal (collectively, "Plaintiffs") and Defendants Pasadena Independent School District ("PISD"), Marshall Kendrick, Jerry Ross Speer, Nelda Sullivan, Vickie Morgan, Fred Roberts, Mariselle Quijano-Lerma, and Jack Bailey, all in their official capacities as members of the Board of Trustees of the PISD. Plaintiffs challenge the at-large electoral system used by the PISD to elect the seven members of its Board of Trustees. The case was tried to the Court between November 6 and November 8, 2013. Pursuant to Federal Rule of Civil Procedure 52, the Court's Findings of Fact and Conclusions of Law are set forth below.
I. LEGAL FRAMEWORK
Section 2 of the Voting Rights Act forbids political processes that "result[] in a denial or abridgement of the right of any citizen of the United States to vote on account of race or color." 42 U.S.C. § 1973. The Supreme Court has long recognized that an at-large voting system may violate Section 2 by "operat[ing] to minimize or cancel out the voting strength of racial minorities in the voting population." Thornburg v. Gingles, 478 U.S. 30, 47 (1986) (internal quotation marks omitted). "The theoretical basis for this type of impairment is that where minority and majority voters consistently prefer different candidates, the majority, by virtue of its numerical superiority, will regularly defeat the choices of minority voters." Id. at 48.
As a threshold matter, Plaintiffs must establish three "necessary preconditions" set forth by the Supreme Court in Gingles for a Section 2 violation. Gingles, 478 U.S. at 50. Specifically, "[t]he minority group must demonstrate that: (1) it is sufficiently large and geographically compact to constitute a majority in a single-member district; (2) it is politically cohesive; and (3) the white majority votes sufficiently as a bloc to enable it - in the absence of special circumstances - usually to defeat the minority's preferred candidates." Sensley v. Albritton, 385 F.3d 591, 595 (5th Cir. 2004). Plaintiffs must prove each of these preconditions by a preponderance of the evidence. League of United Latin American Citizens #4552 (LULAC) v. Roscoe Indep. Sch. Dist., 123 F.3d 843, 846 (5th Cir. 1997). Failure to prove any one of these elements defeats a Section 2 claim. Sensley, 385 F.3d at 595.
Plaintiffs must then prove that, "based on the totality of the circumstances, they have less opportunity than other members of the electorate to participate in the political process and to elect representatives of their choice." Sensley, 385 F.3d at 595 (internal quotation marks omitted). When evaluating the totality of the circumstances, courts are guided by nine factors:
1. the extent of any history of official discrimination in the state or political subdivision that touched the right of the members of the minority group to register, to vote, or otherwise to participate in the democratic process;Teague v. Attala Cnty., Miss., 92 F.3d 283, 292-93 (5th Cir. 1996). Although possible, "it will be only the very unusual case in which the plaintiffs can establish the existence of the three Gingles factors but still have failed to establish a violation of § 2 under the totality of the circumstances." Teague, 92 F.3d at 293 (quoting Clark v. Calhoun Cnty., 21 F.3d 92 (5th Cir. 1994)).
2. the extent to which voting in the elections of the state or political subdivision is racially polarized;
3. the extent to which the state or political subdivision has used unusually large election districts, majority vote requirements, anti-single shot provisions, or other voting practices or procedures that may enhance the opportunity for discrimination against the minority
group;
4. if there is a candidate slating process, whether the members of the minority group have been denied access to that process;
5. the extent to which members of the minority group in the state or political subdivision bear the effects of discrimination in such areas as education, employment and health, which hinder their ability to participate effectively in the political process;
6. whether political campaigns have been characterized by overt or subtle racial appeals;
7. the extent to which members of the minority group have been elected to public office in the jurisdiction;
8. whether there is a significant lack of responsiveness on the part of elected officials to the particularized needs of the members of the minority group; and
9. whether the policy underlying the state or political subdivision's use of such voting qualification, prerequisite to voting, or standard, practice, or procedure is tenuous.
"Because the resolution of a voting dilution claim requires close analysis of unusually complex factual patterns, and because the decision of such a case has the potential for serious interference with state functions, . . . district courts [must] explain with particularity their reasoning and the subsidiary factual conclusions underlying their reasoning." Westwego Citizens for Better Gov't v. City of Westwego, 872 F.2d 1201, 1203 (quoting Velasquez v. City of Abilene, 725 F.2d 1017, 1020 (5th Cir. 1989)). The Fifth Circuit has emphasized the district court's responsibility to "specifically state the evidence found credible and the reasons for its conclusions," and to "discuss all 'substantial' evidence contrary to its decision." LULAC, 123 F.3d at 846.
II. FINDINGS OF FACT
A. Background
The Pasadena Independent School District ("PISD") is a governmental unit organized for the operation of schools in the southeast part of Harris County, Texas. The PISD encompasses portions of the City of Pasadena, the City of South Houston, and the City of Houston. Sixty-two schools operate within its boundaries. (Tr. 2 at 448:15-17.)
Seven members on the PISD Board of Trustees govern the school district. Prior to 2011, board members were elected to three-year terms through at-large elections. Today, board members are elected to four-year terms through at-large elections held every other year in May. (Tr. 2 at 432:11-17 and 449:15-17.) The current Board consists of the individual Defendants in this lawsuit - Marshall Kendrick, Jerry Ross Speer, Nelda Sullivan, Vickie Morgan, Fred Roberts, Mariselle Quijano-Lerma, and Jack Bailey.
The demographics of the PISD's student body have changed dramatically over the last twenty years. Between 1993 and 2011, the Hispanic student population increased by 132.4% while the Non-Hispanic White student population decreased by 73.3%. (Pl. Ex. 20 at 19.) In the 1993-1994 school year, PISD enrolled 39,265 students. (Pl. Ex. 18 at 12.) According to the annual report by the Texas Education Agency, the student body was 44.3% Non-Hispanic White, 46.6% Hispanic, 5% Non-Hispanic Black, and 4.1% Asian/Other. (Id.) In the 2011-2012 school year, the student body had grown to include 52,708 students, of whom 7.9% were Non-Hispanic White, 81.8% were Hispanic, 6.5% were Non-Hispanic Black, and 3.8% were Asian/Other. (Id.)
Dr. Richard Murray, an expert for the Plaintiffs, explains that "[t]he enormous change in the student population in the PISD is rooted primarily in the changing racial-ethnic makeup of the district's population, and secondarily in the aging of its resident Anglos, who have many fewer school age children in the 21st century." (Pl. Ex. 18 at 12.) In 2010, the PISD had a total population of 240,506 and a total voting age population of 164,845. (Pl. Ex. 18 at 13.) Hispanic residents accounted for 66.1% of the total population and 60.8% of the total voting age population. (Pl. Ex. 18 at 13.)
Although the Hispanic population of the PISD has increased, Hispanic representation on the Board of Trustees has remained constant. The Board has continuously included one Hispanic member since 1987, when voters elected Carmen Orozco. (Tr. 2 at 433:13-15.) Ms. Orozco served on the Board until 2009, when voters elected Mariselle Quijano-Lerma. (Tr. 2 at 432:18-22 and 433:6-12.)
A group of Hispanic citizens challenged the PISD's at-large electoral system in 1992. Perez v. Pasadena Indep. Sch. Dist., 958 F. Supp. 1196 (S.D. Tex. 1997). In that case, the plaintiffs were unable to form a compact single-member district in the PISD with a majority of Hispanic voting-age citizens using the 1990 census data. Id. at 1210. The court refused to rely instead on an estimated projection of CVAP figures. Id. at 1212-13. The court held that the plaintiffs "ha[d] not established that, based on the 1990 census, the best available data before [the] court, it [was] possible to form a compact single-member district in the PISD that would have a majority of Hispanic voting-age citizens as residents." Id. at 1228-29. The court found in favor of the defendants since the plaintiffs failed to meet the first factor of Gingles, but noted that the plaintiffs "raised valid concerns as to hindrances to Hispanic participation in the PISD political process." Id. at 1230.
Plaintiffs filed this lawsuit in 2012 to challenge once again the PISD's at-large electoral system. Plaintiffs - Gilbert Joe Cisneros, Catherine Garcia Sonnier, Martha Gonzalez, Salvador Martinez, Edward Ybarra, Arnold Hurtado, and Melinda Villarreal - are eligible to vote in the PISD elections. They allege that their votes are diluted by the PISD's at-large electoral scheme and that they are not able to participate equally in the political process.
B. Fact and Expert Witnesses
Plaintiffs offered the testimony of four experts. Dr. Matt Barreto, who presented a statistical analysis of endogenous and exogenous elections, holds B.S. and Ph.D. degrees in political science, as well as a M.S. degree in social science. Dr. Barreto is an Associate Professor of Political Science at the University of Washington. George Korbel analyzed the demographics of the PISD and presented a demonstration plan of single-member districts. Mr. Korbel holds a B.A. degree in government, and a J.D. degree from the University of Minnesota. Mr. Korbel has testified in a number of voting rights cases and has acted as a consultant in the redistricting of several jurisdictions. Richard Murray, a Professor of Public Policy at the University of Houston and the Director of Surveying at the Hobby Center for Public Policy, holds B.A. and M.A. degrees in government, and a Ph.D. degree in political science. Dr. Murray analyzed the history of elections in the PISD and Texas, and presented a statistical analysis of two exogenous elections. Gerald Mark Birnberg holds a B.B.S. degree in accounting and a J.D. degree from the University of Texas College of Law, and held the position of Chair of the Harris County Democratic Party from January of 2003 until December of 2011. Mr. Birnberg testified to the historic and present barriers facing Hispanic voters in the political process in Harris County.
In addition, Plaintiffs offered the testimony of Cody Wheeler, Celestino Munoz Perez, Jr., Orlando Ybarra, and Edward Ybarra. Mr. Wheeler and Mr. Orlando Ybarra are City Councilmen for the City of Pasadena. Mr. Perez, a plaintiff in the 1992 PISD case, previously ran for a PISD board member position. Mr. Edward Ybarra, of no relation to Mr. Orlando Ybarra, worked as an election clerk in the PISD during the 2009 election.
Defendants offered the testimony of Dr. John Alford as an expert in statistical analysis. Dr. Alford holds B.S., M.A., and Ph.D. degrees in political science and a M.P.A. degree in public administration. Dr. Alford is currently an Associate Professor of Political Science at Rice University.
Defendants also offered the testimony of John Hancock, Kirk Lewis, and Gloria Gallegos. Mr. Hancock was the Principal of Pasadena High School in 1987. Mr. Lewis is the Superintendent of the PISD. Ms. Gallegos is the Assistant Superintendent for Special Programs in the PISD.
C. Gingles I
Plaintiffs must first demonstrate that the Hispanic population in the PISD is "sufficiently large and geographically compact to constitute a majority in a single-member district." Gingles, 478 U.S. at 50. The relevant metric for the size of the minority population in any given district is the citizen voting age population ("CVAP"). Perez v. Pasadena Indep. Sch. Dist., 165 F.3d 368, 372 (5th Cir. 1999). The Fifth Circuit has unequivocally held that "courts evaluating vote dilution claims under section 2 of the Voting Rights Act must consider the citizen voting-age population of the group challenging the electoral practice when determining whether the minority group is sufficiently large and geographically compact to constitute a majority in a single-member district." Campos v. City of Houston, 113 F.3d 544, 548 (5th Cir. 1997) (emphasis added).
1. Source of Data
Historically, the United States Census Bureau collected citizenship information on the long form of the decennial census. The short form of the decennial census, which was sent to every household in the United States, included basic questions about age, sex, race, Hispanic origin, and household relationship. The long form of the decennial census, which was sent to only a subset of the population, included more detailed questions about citizenship and socioeconomic status. After the 2000 Census, the Census Bureau decided to discontinue the long form of the decennial census. See Rodriguez v. Harris Cnty., Tex., 964 F. Supp. 2d 686, 727 (S.D. Tex. 2013).
The sole source of citizenship data published by the Census Bureau now comes from the American Community Survey ("ACS"). The ACS is an annual nationwide survey to collect demographic information, including age, race, ethnicity, and citizenship, from a sample of the population. With this data, the Census Bureau is then able to estimate the CVAP of states, counties, cities, census tracts, and block groups. The Census Bureau combines CVAP data over five year periods in order to provide more reliable estimates for small areas, such as census tracts and block groups. See generally U.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data: What General Data Users Need to Know (Oct. 2008), http://www.census.gov/acs/www/Downloads/handbooks/ACSGeneralHandbook.pdf.
Plaintiffs' experts rely instead on a count of the number of registered voters with recognized Spanish surnames. While the ACS's estimate of CVAP projects the number of self-identified Hispanics who could register in a particular area, a tally of Spanish-surname Registered Voters ("SSRVs") attempts to approximate the number of Hispanics who do register in a particular area. (Tr. 2 at 268-69.) To do so, the list of registered voters is compared to a list of Spanish surnames developed by the United States Census Bureau. (Tr. 1 at 21-22.) Neither party has presented any evidence as to the rate of error for this methodology.
The Texas Legislative Council ("TLC") produced the SSRV figures for Plaintiffs' demonstration districts. (See Tr. 2 at 286-87.) Mr. Korbel acknowledged that the TLC could have produced CVAP figures from the 2007-2011 ACS for the districts that he had drawn for this case. (Tr. 2 at 295:9-19, 302:16-17, and 303:9-16.) Mr. Korbel testified that the TLC has "a policy now that they don't like to try and project [CVAP data] at districts this size." (Tr. 2 at 296:2-3.)
2. Demonstration Districts
Mr. Korbel testified that the percentage of registered voters in the PISD with Spanish surnames was 40.54% in 2010. (Tr. 2 at 268:2.) This population of SSRVs was concentrated to the north and west of the PISD. (Tr. 2 at 269:10-14; see also Pl. Ex. 20 at 20.) Based on the numbers and location of SSRVs, Mr. Korbel testified that it is possible to draw "at least four very compact districts [in the PISD] that have more than 50 percent of the Spanish surname registered voters in each one of those districts." (Tr. 2 at 272:2-4.)
Mr. Korbel presented the following demonstration plan:
Image materials not available for display.
Image materials not available for display. (Pl. Ex. 21.) The first four districts, concentrated in the northwestern portion of the PISD, are each above 50% SSRV. (Tr. 2 at 287:4-9.) Specifically, District 1 has 51.7% SSRV, District 2 has 55.6% SSRV, District 3 has 51.6% SSRV, and District 4 has 52.5% SSRV. (Pl. Ex. 21.) Mr. Korbel testified that there was "no question at all" that the same districts could be drawn with greater than 50% Hispanic CVAP. (Tr. 2 at 302-03.) Mr. Korbel explained that "if they are registered voters, they're obviously citizens." (Tr. 2 at 285:22-23.)
Dr. Murray agreed that SSRVs were "sufficiently compact and numerous enough to constitute majorities in two or more single-member districts." (Tr. 2 at 366:13-18.) Dr. Murray provided the SSRV percentage for each individual voting precinct in the PISD as of June of 2012. (Pl. Ex. 18 at 14-15.) Based on this data, Dr. Murray hypothesized that three majority-minority districts could be created:
[A] proto-type District A could be drawn tha[t] would include the voting precincts of 301, 786, 791, 404, 785, 277, and 267, with a(Pl. Ex. 18 at 19 n.20.) Dr. Murray testified that the SSRV rate was "a conservative count" when compared to CVAP, since "you have virtually no [] noncitizens who are registered voters." (Tr. 2 at 367:12-14.)
total population of 35,432 (ideal would be 34,358) that would have a Hispanic VAP of 79.9% and a SSRV of 57.2%. A proto-type District B could be drawn with precincts 181, 527, 777, 190, 188, 394, 221, 278, 708 and two-thirds of 242. Such a district would have a Hispanic VAP of 81.9% and a SSRV of 59.2%. District C could be drawn with voting precincts 029, 093, 329, 308, 770, 762, 763, 280, and 289 that would have a Hispanic VAP of 81.8% and a SSRV of 63.3%.
The precinct data on which these figures are based appears at Appendix 1. There is no accompanying map.
Defendant's expert, Dr. Alford, testified that "[his] impression is you can draw the [majority-minority] district." (Tr. 3 at 604:24-25.) He stressed that he had not personally evaluated the first factor of Gingles, but had reviewed the work of Mr. Korbel. (Tr. 3 at 603-04.) Dr. Alford hesitated, however, to say such a district was certain. He explained: "The only reason that I would hesitate to say that you certainly can draw the district is that I've never seen - in the era since Campos, I've never seen a Gingles 1 district that didn't have a citizen analysis. It's in the computer. It's easy to do." (Tr. 3 at 604-05.) "I personally believe I could draw the district but I haven't seen the district drawn and I don't know why that number hasn't been put in evidence." (Tr. 3 at 605:9-12.)
3. Analysis
Census data, though preferred, is not required to establish CVAP for demonstration districts. See Westwego Citizens for Better Gov't v. City of Westwego, 906 F.2d 1042, 1045 n.3 (5th Cir. 1990); see also Reyes v. City of Farmers Branch, Tex., 2008 WL 4791498 at *9 (N.D. Tex. Nov. 4, 2008). In Westwego, the Fifth Circuit suggested that "other probative evidence," including "registered voter data by race," could be considered when census data proved difficult to obtain. Westwego, 906 F.2d at 1045 n.3. In this case, however, Plaintiffs have shown neither the probative value of SSRV, nor the unreliability of available census data.
Spanish surnames are an imperfect proxy for Hispanic self-identification. See United States v. Alamosa Cnty., Colo., 306 F. Supp. 2d 1016, 1022 (D. Colo. 2004). Errors of omission occur when Hispanic individuals are not counted because they do not have a Spanish surname. For example, Cody Wheeler, a Hispanic Councilman for the City of Pasadena, (Tr. 1 at 130:10-11), would not be counted by the Spanish surname methodology. Conversely, errors of commission occur when non-Hispanic individuals are mistakenly counted because they have a Spanish surname. For example, if a woman married a Hispanic man and took his last name, she would be counted regardless of her own identification. The Fifth Circuit stated that "without a strict showing of its probativeness, Spanish-surname data are disfavored, and census data based upon self-identification provides the proper basis for analyzing Section 2 vote dilution claims in the future." Rodriguez v. Bexar Cnty., Tex., 385 F.3d 853, 866 n.18 (5th Cir. 2004).
In Reyes v. City of Farmers Branch, Tex. , the Northern District of Texas was presented with a choice between census data from 2000 or SSRV estimates from 2006. 2008 WL 4791498 at *8 (N.D. Tex. Nov. 4, 2008). The court found that the SSRV data was sufficiently probative of the first Gingles factor to be considered. Id. at *9. "Although courts have criticized the tendency of this type of data to misidentify Hispanic persons as non-Hispanic and vice versa, the Court finds that it can consider this type of data when, as here, Census data are outdated and therefore less likely to be accurate." Id.
The SSRV data in Reyes, though probative, was insufficient to establish the first factor of Gingles. The defendants introduced evidence of the inaccuracy of the SSRV estimate, which was a majority by only a few percent. Id. at *9-10. Specifically, the defendants' expert testified that splitting precincts could increase inaccuracy, and the TLC itself warned of the accuracy of applying its methodology to small areas. Id. at *9. The court concluded that "the TLC estimate suffers from reliability issues, and does not demonstrate that it is more likely than not that Plaintiffs' proposed district contains a Hispanic citizen voting age majority." Id. at *10. With an unreliable TLC estimate of 52.5% SSRVs in the demonstration district, the court found it "just as likely that Plaintiffs' proposed district has below or equal to 50% [Hispanic CVAP]." Id.
In this case, Plaintiffs chose to present TLC's estimate of SSRVs rather than the ACS's estimate of CVAP. Plaintiffs argued that the ACS's estimate of CVAP was unreliable given the small size of hypothetical districts within the PISD. Mr. Korbel testified that the ACS's sample methodology did not work well for small geographic areas since the sample might have few, or zero, responses from the area. (Tr. 2 at 301-02.) Dr. Murray likewise pointed to "some underlying major methodological problems" when relying on the ACS's CVAP data at the local level. (Tr. 2 at 367:4-5.) Dr. Murray stated that he does not find the ACS data, "particularly as you get down to small geographies, to be very credible and useful." (Tr. 2 at 367:9-11.) Dr. Barreto echoed the other experts' concerns. He testified that the margin of error would be very large for block group data. (Tr. 1 at 102-03.) As such, "we would have far less reliability on those very, very small levels of geography than we would say over an entire county." (Tr. 1 at 103:2-4.)
This Court, like others, finds the ACS's estimates of CVAP sufficiently reliable for use in voting rights litigation. The Census Bureau increases the reliability of its estimates for small political units by pooling together five years of data for any area with fewer than 20,000 people. U.S. Census Bureau, A Compass for Understanding and Using American Community Survey Data: What State and Local Governments Need to Know, 2 (Feb. 2009), http://www.census.gov/acs/www/Downloads/handbooks/ACSstateLocal.pdf. Dr. Barreto in fact used the 2007-2011 five-year estimate from the ACS to provide the Hispanic CVAP population of the City of Pasadena in his expert report. (Pl. Ex. 15 at 2.) In his testimony, he explained that the five-year estimate "gives us a lot more reliability." (Tr. 1 at 20:18-21.) In addition, the ACS publishes accompanying margins of error so that the data can be properly evaluated by statisticians and demographers. The Southern District of Texas found that "[t]he ACS is at present the only reliable source of citizen-voting age population data." Rodriguez v. Harris Cnty., Tex., 964 F. Supp. 2d 686, 728 (S.D. Tex. 2013) (emphasis added). "Inaccuracies, alone, will not render the ACS data inherently unreliable." Id. at 729.
In Fabela v. City of Farmers Branch, Tex. , the Court used SSRV data to corroborate ACS's estimate of CVAP for a small geographic area. 2012 WL 3135545 at *8 (N.D. Tex. Aug. 2, 2012). Like Plaintiffs in this case, the defendants in Fabela argued that "the ACS is unreliable when used for small geographic areas because the small sample size creates very large margins of error." Id. at *6. The court nevertheless found the five-year ACS estimate to be "the most reliable form of readily available citizenship data." Id. at *7 n.14. The court concluded that the plaintiffs "proved that, using the most accurate, readily-available data, a geographically compact demonstration district can be drawn in Farmers Branch in which Hispanics constitute more than 50% of the CVAP." Id. at *8. SSRV data corroborated this CVAP estimate. Id. The court stressed that it was "not relying solely or even primarily on SSRV data to find that plaintiffs have established the first Gingles factor." Id. at *7 n.18.
The plaintiffs in Rodriguez v. Harris Cnty., Tex., attempted to substitute CVAP data with SSRV and VAP data in one demonstration map. 964 F. Supp. 2d 686, 736-37 (S.D. Tex. 2013). This demonstration map "ha[d] a Spanish Surname registration rate of 47.1% and Latinos constitute[d] 69% of the voting age population." Id. at 736. The court recognized that the high VAP figure "lends credence to the inference that Latinos constitute a majority of the citizen voting age population" in the demonstration district. Id. "Nevertheless," the court concluded, "our case law requires plaintiffs to establish each Gingles precondition and these inferences are insufficient to satisfy that burden, especially given the fact that the use of SSRV data is often a crude and imprecise measure of the Latino registered voter population." Id.
Here, Plaintiffs have also presented SSRV and VAP data for each demonstration district in lieu of CVAP data. (Pl. Ex. 21; Pl. Ex. 18 at 19 n.20.) Dr. Murray and Mr. Korbel assumed that the CVAP of the demonstration districts would be above 50% since the SSRV estimate was above 50%. Mr. Korbel testified that the SSRV estimate "[i]s usually less" than CVAP, (Tr. 2 at 268:19-22), and Dr. Murray stated that the SSRV estimate "is almost always going to be less than" CVAP, (Tr. 2 at 368:5-8). The Court credits the testimony of all of the experts that it is theoretically possible to create a majority-minority district in the PISD, but Plaintiffs have not presented sufficient reliable proof to that effect.
There is an important need for flexibility in the face of sparse data for vote dilution claims. If census data were unavailable or unreliable, SSRV data would be an appropriate alternative source of evidence. See, e.g., Reyes, 2008 WL 4791498 at *9 (using SSRV data when census data were outdated). The Court finds no such need in this case. ACS's five-year estimates of CVAP are reliable for the purposes of a Section 2 analysis. Plaintiffs have not provided sufficient evidence that the ACS's estimates are difficult to obtain for their demonstration districts. Since CVAP data from the ACS are reliable and available, there is no need to resort to the "highly problematic" SSRV data. See Rodriguez, 385 F.3d at 866 n.18.
The Court is troubled by some evidence in the record that CVAP data are difficult to obtain for small demonstration districts, like those at issue in this case, but finds such evidence insufficient. Mr. Korbel testified that the TLC has "a policy now that they don't like to try and project at districts this size." (Tr. 2 at 296:2-10.) Although the TLC provided a CVAP estimate for the PISD as a whole, they preferred not to provide a CVAP estimate for smaller districts. (Id.) Mr. Korbel, nevertheless, acknowledged that the TLC might have provided such data if "pressured." (Tr. 2 at 302:16-17.) In fact, he admitted that the TLC system automatically provides CVAP figures for drawn districts when requested. (Tr. 2 at 298:8-12 and 303:9-16.) Moreover, Drs. Barreto and Alford both expressed their belief that CVAP data was available. (Tr. 1 at 72:12 and Tr. 3 at 605:3-4.) The Court does not find this contradictory evidence to be a sufficient showing of the probative value of the SSRV data.
D. Gingles II and III
Together, the second and third factors of Gingles require evidence of "legally significant racial bloc voting." See Gingles, 478 U.S. at 57. "Racial bloc voting," also called "racially polarized voting," exists "where there is a consistent relationship between the race of the voter and the way in which the voter votes." Id. at 53 n.21 (internal quotation marks omitted). For racially polarized voting to be legally significant, minority voters must constitute a "politically cohesive unit" and white voters must vote "sufficiently as a bloc usually to defeat the minority's preferred candidate." Id. at 56. Here, the parties do not dispute that the Hispanic community is politically cohesive, but they diverge as to whether or not racial polarization operates to defeat the Hispanic choice.
Certain elections are more probative in this analysis than others. Recent interracial endogenous elections are the most probative elections. See Bone Shirt v. Hazeltine, 461 F.3d 1011, 1020-21 (8th Cir. 2006). In this case, both parties confined their analyses to elections within the last ten years. (Tr. 1 at 29:14-16 and 68:6-9.) Within this time span, endogenous elections are more probative than exogenous elections. Clark v. Calhoun Cnty., Miss., 88 F.3d 1393, 1397 (5th Cir. 1996). In addition, elections between white and minority candidates are more probative than elections between only white candidates. See Magnolia Bar Ass'n, Inc. v. Lee, 994 F.2d 1143, 1149 (5th Cir. 1993); see also Bone Shirt v. Hazeltine, 336 F. Supp. 2d 976, 1011-12 (D.S.D. 2004).
Endogenous elections are those that are directly related to the contested election practice. (Tr. 1 at 25:10-13.) Exogenous elections are those that take place within the same geography, thus involving the same electorate, but for different types of elections. (Tr. 1 at 25:14-17.)
The parties' experts rely on two main statistical tools to model voting behavior across these elections. (Def. Ex. 29 at 3.) First, Goodman's Ecological Regression estimates the best-fitting straight line to describe the relationship between the predicted vote for the Hispanic candidate to the Hispanic population in the precinct. The intercept of the line, the point at which the line crosses the vertical axis, predicts the share of the vote for the Hispanic candidate when the Hispanic voter population is zero. The slope of the line, the rise over run, predicts the change in the vote for the Hispanic candidate for a one percentage point change in the percent of Hispanic voters. With both the intercept and the slope, the expert can predict the share of the vote for the Hispanic candidate when the Hispanic voter population is 100%. (Def. Ex. 29 at 3-4.) Many courts, including the Supreme Court, have approved the use of a bivariate ecological regression analysis. See, e.g., Gingles, 478 U.S. at 53; Houston v. Lafayette Cnty., Miss., 56 F.3d 606, 612 (5th Cir. 1995); Benavidez v. City of Irving, Tex., 638 F. Supp. 2d 709, 723 (N.D. Tex. 2009).
Second, King's Ecological Inference uses "a method of bounds analysis, combined with a more traditional statistical method, to improve on standard ecological regression." (Def. Ex. 29 at 4.) Dr. Alford explains that "[w]hile the details are mathematically complex, the differences mostly center on utilizing deterministic bounds information contained in individual precinct results that would not be exploited in ecological regression, and by not imposing a linear constraint on the pattern across precincts." (Id.) Ecological inference, which is of a more recent vintage than ecological regression, has been considered by a number of district courts in the Fifth Circuit. See, e.g., Rodriguez v. Harris Cnty., Tex., 964 F. Supp. 2d 686, 759, 767 (S.D. Tex. 2013); Fabela v. City of Farmers Branch, Tex., 2012 WL 3135545 at *9 & n.22. (N.D. Tex. Aug. 2, 2012); Benavidez, 638 F. Supp. 2d at 723-24, 731.
1. Endogenous Elections
Dr. Murray
According to Dr. Murray, PISD Board elections were competitive in the 1960s, 1970s, and 1980s. (Pl. Ex. 18 at 16.) During this time, "[B]oard elections were almost all contested, voter turnout was moderate for a local election scheduled off-cycle from the General Elections for federal, state, and county offices, and many elections were competitive." (Id.) Dr. Murray attributed the competitiveness of past elections to the existence of a conservative business slate and a liberal labor slate. (Tr. 2 at 373:20-25.) The liberal slate effectively collapsed in the 1980s, allowing the conservative slate to dominate subsequent local elections. (Tr. 2 at 370, 372; Pl. Ex. 18 at 17.)
Dr. Murray pinpoints the 1987 election as the critical turning point. (Pl. Ex. 18 at 17; Tr. 2 at 372.) In this election, "[t]wo veteran incumbent Trustees, Cecil Ghormley and B.J. Garner, were defeated by challengers backed by business and conservative interests, and an open seat was won by Carmen Orozco, who was also supported by the same interests." (Pl. Ex. 18 at 17.) Dr. Murray testified that these challengers won with the support of the senior administration of the PISD. (Tr. 2 at 372-73.) As evidence, Dr. Murray quotes the overwhelming margin by which the challengers won the election at the early vote held in the PISD Administration Building. (Id.) On cross-examination, however, Dr. Murray acknowledged that, under the contemporary rules, only voters "who qualified to vote absentee by personal appearance" would have voted at the PISD Administration Building. (Tr. 2 at 400:6-10.) There is no evidence that senior administration officials qualified to vote absentee, or, even if they did, that they voted at this location instead of other voting sites potentially closer to their residences.
John Hancock, Principal of Pasadena High School in 1987, and Kirk Lewis, the current Superintendent of the PISD, provided an alternative explanation for the change on the Board in 1987. Mr. Hancock testified that the prior Board's micromanagement of the day-to-day operations of the school district was the overriding issue of the election. (Tr. 2 at 425:8-17.) Specifically, the prior Board was associated with the summary dismissal of football coaches at some of the high schools. (Tr. 2 at 425-26.) Mr. Lewis similarly testified that "issues surrounding that [1987] election were the termination by the [B]oard of several high school football coaches and the hiring of a superintendent that the community was not real comfortable with." (Tr. 2 at 443:11-14.)
Ms. Orozco became the first Hispanic Board member when she was elected in 1987. Ms. Orozco was reelected in 1990, 1993, 1996, 1999, 2002, and 2005. (Tr. 2 at 432-33.) Her elections in 1996 and 2005 were unopposed. (Id.) In 2009, Ms. Orozco left the Board, and Mariselle Quijano-Lerma filled her position. (Tr. 2 at 433:6-12.) Between Ms. Orozco and Ms. Quijano-Lerma, there has been an Hispanic Board member since 1987. (Tr. 2 at 433:13-15.)
Dr. Murray explains the continuous election of Hispanic candidates to the PISD Board based on their connection to the conservative slate. (Pl. Ex. 18 at 19.) "The controlling clique recruited a suitable Hispanic candidate (Carmen Orozco) in the mid-1980s and supported her easy reelections until she recently retired. This 'Hispanic' seat was then filled in 2009 by Mariselle Quijano-Lerma, with the same base of establishment support that Ms. Orozco had enjoyed." (Id.) Likewise, Mr. Celestino Perez, Jr. testified that "[t]here has never been an independent Hispanic person represented or selected by a Hispanic community to run and be successful in any PISD Board of Trustee[s] election." (Tr. 1 at 166:17-19.) By independent, Mr. Perez explained, he meant able to run without the support of the slating group. (Tr. 1 at 176:1-2.)
Dr. Murray performed no statistical analysis of endogenous elections. Instead, Dr. Murray urges the Court to look to exogenous elections, given "the extra-ordinary effectiveness of the Anglo-controlled [B]oard in diminishing competition." (Pl. Ex. 18 at 20.) "With most elections in the district now cancelled, and few independent candidacies by Hispanics, there is a glaring absence of relevant data from recent [B]oard contests." (Id.)
Dr. Matt Barreto
Dr. Barreto co-authored a report with Dr. Francisco Pedraza which analyzed the results of the two endogenous interracial elections in the last ten years for PISD Board positions. (Pl. Ex. 15 at 5; Tr. 1 at 35:6-9.) Specifically, Ms. Orozco's election in 2002 and Ms. Quijano-Lerma's election in 2009 were the only contested endogenous elections over the last ten years that included an Hispanic candidate. (Pl. Ex. 15 at 5.)
Drs. Barreto and Pedraza relied on Latino voting age population ("VAP") data to provide the demographics of each voting precinct. (Pl. Ex. 15 at 6.) In total, there are 58 Voting Precinct Districts ("VPDs") within the PISD boundaries, but these are consolidated amongst many fewer polling places for PISD elections. (Tr. 1 at 27:2-10.) The PISD had ten polling places during the 2002 election, and eleven during the 2009 election. (See Pl. Ex. 15 at 10-11.) Using the demographic information from the precincts and the election results from the polling places, Drs. Barreto and Pedraza ran an ecological regression to determine the voting patterns in the PISD.
The record is unclear as to the exact number of polling places. Dr. Barreto testified that there were "11 or 9 polling locations." (Tr. 1 at 27:7-10.) In their expert report, Drs. Barreto and Pedraza explain that there are only 11 polling places for PISD elections. (Pl. Ex. 15 at 6.) In their results, Drs. Barreto and Pedraza note 10 observation units for the 2002 election and 11 for the 2009 election. (Pl. Ex. 15 at 10-11.) The Court relies on the numbers found in the results themselves.
The ecological regression produced the following results for the two endogenous elections involving an Hispanic candidate:
CONTESTED ENDOGENOUS ELECTIONS WITH AN HISPANIC CANDIDATE
| Table 1: Goodman's Ecological Regression: Vote for Orozco | |
| (1) 2002 PISD Board of Trustee | |
| pct_latino | 0.059G (0.172) |
| Constant | 0.683*** (0.0777) |
| Observations | 10 |
| R2 | 0.015 |
| Adjusted R2 | -0.108 |
| Standard errors in parentheses | |
| * p < 0.05, ** p< 0.01, ** p < 0.001 |
| Table 2: Goodman's Ecological Regression: Vote for Quijano | |
| (1) 2002 PISD Board of Trustee | |
| pct_latino | -0.0370 (0.132) |
| Constant | 0.628*** (0.0850) |
| Observations | 11 |
| R2 | 0.009 |
| Adjusted R2 | -0.101 |
| Standard errors in parentheses | |
| * p < 0.05, ** p< 0.01, ** p < 0.001 |
(Pl. Ex. 15 at 10-11.) The constant, or intercept, represents the estimated vote given to the Hispanic candidate by non-Hispanic White voters. (Pl. Ex. 15 at 7.) Adding the coefficient for "percent Latino" to the constant estimates the vote given to the Hispanic candidate by Hispanic voters. (Id.) Using these directions, the Court calculates that Ms. Orozco received 68.3% of the non-Hispanic vote, and 74.3% of the Hispanic vote. Seven years later, Ms. Quijano-Lerma received 62.8% of the non-Hispanic vote, and 59.1% of the Hispanic vote.
Drs. Barreto and Pedraza concluded that there was "no evidence of racially polarized voting" in these two endogenous elections. (Pl. Ex. 15 at 6.) Dr. Barreto explained that he and Dr. Pedraza "were able to find that both candidates did receive support from Hispanic voters, and that there was Hispanic political cohesion." (Tr. 1 at 37:6-9.) In addition, both candidates received support from non-Hispanic voters. (Tr. 1 at 37:9-12.)
Dr. Barreto warned against drawing broad conclusions from these two elections. "[T]hese two elections, first of all, are not frequent or numeric enough to draw conclusions about the current state of the environment in Pasadena." (Tr. 1 at 35:14-16.) Furthermore, these two elections were not particularly competitive. Ms. Orozco was an incumbent candidate, who enjoyed the support of the slating group. (Tr. 1 at 37:14-20; Pl. Ex. 15 at 5.) Ms. Quijano-Lerma ran with a white incumbent, and was endorsed by Ms. Orozco when she retired. (Tr. 1 at 37-38; Pl. Ex. 15 at 5.) Dr. Barreto testified that, "because of the incumbency advantage and the slating association of the candidates, they maybe don't represent good examples of the types of elections, and the types of voting options in Pasadena." (Tr. 1 at 35:18-21.)
In addition to questioning whether these results are representative of broader trends, Dr. Barreto questioned whether these results are reliable due to limited data. Drs. Barreto and Pedraza reported that "the very small number of voting precincts makes it more difficult to analyze voting patterns and make determinations of racial bloc voting." (Pl. Ex. 15 at 5.) The limited polling places provide few data points for the statistical analysis, and the data points themselves are less representative of the community since the polling places amalgamate potentially different voting precincts. (See id.; Tr. 1 at 27-28.) On top of this problem, the low voter turnout makes the data "much, much more sensitive to any errors." (Tr. 1 at 41:5-6.) Dr. Barreto explained that "if you only have a group of 10 people, and you miscalculate two of them, that's a 20 percent error rate. Whereas [] if you miscalculate two out of a thousand, it's miniscule." (Tr. 1 at 42:5-8.)
Given the limitations present for endogenous elections, Drs. Barreto and Pedraza turned to exogenous elections to more fully analyze voting patterns in the PISD. (Tr. 1 at 36.) The voters of the PISD vote in many other elections than those for the PISD Board. (See id.) By expanding the scope of their analysis to include other elections, Drs. Barreto and Pedraza hoped to capture "relevant information about the patterns of candidate preferences among the same voters." (Pl. Ex. 15 at 3.) They explained that data for exogenous elections "[was] available at a more granular level" since votes were distributed across 58 polling places. (Id.) "If patterns of clear racial polarization exist across a variety of exogenous elections among voters in a given jurisdiction, we can reasonably expect those same voters to bring their same predispositions and preferences into endogenous elections." (Pl. Ex. 15 at 7-8.)
Dr. John Alford
While Drs. Barreto and Pedraza considered only endogenous elections involving an Hispanic candidate, Dr. Alford analyzed all contested endogenous elections since 2000. This approach provided Dr. Alford with six contested elections to consider. (Def. Ex. 29 at 8.) Namely, the 2000 election between Jerry Ross Speer and Bob Blair, the 2002 election between Carmen Orozco and Doris Barnes, the 2004 and 2008 elections between Vickie Morgan and Randy Smith, the 2009 election between Mariselle Quijano-Lerma and Neel McGovern, and the 2009 election between Nelda Sullivan, Terry Robinson, and Randy Smith.
For these elections, Dr. Alford considered an expanded universe of polling places by treating early voting separately from election day voting. There were ten polling places across the PISD until 2008, and eleven afterwards. (Def. Ex. 29 at 6.) Each polling place reported the early vote results separately from the election day results, "in effect creating two distinct election reporting units for analytical purposes." (Id.) In 2009, six of the eleven polling places also reported early votes by mail separately, creating an additional six reporting units. (Def. Ex. 29 at 6-7.) Each of these 28 data points provided separate vote totals and sign-in sheets. Dr. Alford calculated the Hispanic voting population at each "polling place" - whether voting early, by mail, or on election day - by counting the number of voters with recognized Spanish surnames on the sign-in sheets. (Def. Ex. 29 at 7.)
The Court's refutation of Spanish-surname data to analyze the first factor of Gingles might appear to contradict the Court's acceptance of Spanish-surname data to analyze the third factor of Gingles. Here, a tally of Spanish surnames on sign-in sheets is an effective, albeit imperfect, way to estimate the Hispanic voters who voted early or on election day. The SSRV count is necessary to treat the results of these elections as essentially different "polling places." (See Def. Ex. 29 at 7.) As such, Defendants have illustrated the probative value of this specific form of evidence in a way lacking for the first factor.
Dr. Alford summarized his results, using ecological inference and ecological regression, for the six endogenous elections in the following table. Plaintiffs' results, using only ecological regression, are included for comparison and highlighted in the chart:
CONTESTED ENDOGENOUS ELECTIONS
| King's Ecological Inference | Goodman's Ecological Regression | ||||
| Percent of Hispanics Supporting Candidate | Percent of Non-Hispanics Supporting Candidate | Percent of Hispanics Supporting Candidate | Percent of Non Hispanics Supporting Candidate | Number of Precincts Used in Estimation | |
| 2009, Position 2 | |||||
| Quijano-Lerma | 72% (2.26) | 57% (1.17) | 71% (7.39) | 54% (3.16) | 28 |
| (Barreto-Pedraza) | 59% | 63% | 11 | ||
| McGovern | 28% (2.26) | 43% (1.12) | 29% (7.39) | 46% (3.16) | 28 |
| 2009, Position 3 | |||||
| Sullivan | 38% (2.36) | 67% (1.30) | 37% (7.39) | 70% (3.16) | 28 |
| Smith | 14% (1.53) | 12% (0.84) | 13% (5.64) | 16% (2.41) | 28 |
| Robinson | 47% (2.94) | 21% (1.63) | 50% (6.36) | 15% (2.72) | 28 |
| 2008, Position 7 | |||||
| Morgan | 75% (0.98) | 77% (0.41) | 75% (5.56) | 74% (2.30) | 22 |
| Smith | 25% (0.90) | 23% (0.38) | 25% (5.56) | 26% (2.30) | 22 |
| 2004, Position 7 | |||||
| Morgan | 73% (9.23) | 65% (1.00) | 78% (13.83) | 68% (3.69) | 20 |
| Smith | 28% (9.06) | 34% (0.98) | 22% (13.83) | 32% (3.69) | 20 |
| 2002, Position 2 |
| Orozco | 92% (4.43) | 68% (0.61) | 89% (12.77) | 64% (3.39) | 20 |
| (Barreto-Pedraza) | 74% | 68% | 10 | ||
| Barnes | 8% (4.50) | 32% (0.61) | 11% (12.77) | 36% (3.39) | 20 |
| 2000, Position 5 | |||||
| Speer | 97% (2.64) | 50% (0.22) | 65% (46.17) | 49% (9.25) | 20 |
| Blair | 1% (0.52) | 42% (0.04) | 12% (43.97) | 43% (8.81) | 20 |
| Others | 0.2% (0.01) | 10% (0.001) | 23% (10.37) | 8% (2.08) | 20 |
Numbers in parentheses are standard errors. (Def. Ex. 29 at 10.)
Dr. Alford concurs with Dr. Barreto's analysis that racial polarization is not evident in the two contested elections involving Hispanic candidates. In 2002, Ms. Orozco was clearly the candidate of choice for Hispanic voters in the PISD, receiving 92% of the Hispanic vote according to ecological inference, 89% of the Hispanic vote according to Dr. Alford's analysis under ecological regression, and 74% of the Hispanic vote according to Dr. Barreto's analysis under ecological regression. Ms. Orozco also received the majority of non-Hispanic votes, receiving 68% of the non-Hispanic vote according to ecological inference, 64% of the non-Hispanic vote according to Dr. Alford's analysis under ecological regression, and 68% of the non-Hispanic vote according to Dr. Barreto's analysis under ecological regression.
The same pattern reoccurs for Ms. Quijano-Lerma in 2009, when she was the clear candidate of choice for Hispanic voters and non-Hispanic voters alike. Ms. Quijano-Lerma received 72% of the Hispanic vote according to ecological inference, 71% of the Hispanic vote according to Dr. Alford's analysis under ecological regression, and 59% of the Hispanic vote according to Dr. Barreto's analysis under ecological regression. Ms. Quijano-Lerma also received 57% of the non-Hispanic vote according to ecological inference, 54% of the non-Hispanic vote according to Dr. Alford's analysis under ecological regression, and 63% of the non-Hispanic vote according to Dr. Barreto's analysis under ecological regression.
Another three endogenous elections involve the same pattern whereby the preferred candidate of Hispanic voters is also the preferred candidate of non-Hispanic voters. Jerry Ross Speer, the victor in the 2000 contest for Position 5, enjoyed the support of Hispanic and non-Hispanic voters. (Def. Ex. 29 at 9.) Vicki Morgan, the victor in 2004 and 2008 for Position 7, was also the preferred candidate of Hispanic and non-Hispanic voters. (Id.)
The 2009 election of Nelda Sullivan for Position 3 is the only endogenous election that does not support this pattern. In that contest, Terry Robinson was the "plurality preferred candidate of Hispanic voters," with 47% of the Hispanic vote. (Def. Ex. 29 at 9.) Nelda Sullivan, who was the clear candidate of choice for non-Hispanic voters, still garnered 38% of the Hispanic vote. With this sole exception, each of the Hispanic preferred candidates won the contested endogenous elections since 2000. (Id. at 11.)
Dr. Alford notes that "the standard errors of these estimates . . . are very small, particularly for the more efficient EI technique." (Def. Ex. 29 at 8.) As explained above, Dr. Barreto raised the concern that there was not enough data across ten or eleven polling places to rely on the results of the ecological regression. (See Tr. 1 at 36:16-22.) Dr. Alford argues that Plaintiffs' complaint as to the lack of evidence is "self-imposed." (Def. Ex. 29 at 6.) According to Dr. Alford, the small standard errors in his analysis are "the empirical indication that the 'granularity' issue raised by Professors Barreto and Pedraza has indeed been effectively addressed by employing more and better data and adding a more efficient analytical technique." (Id. at 8-9.) The results show that, in the PISD Board elections, Hispanic voting is cohesive and successful.
2. Exogenous Elections
Dr. Richard Murray
Dr. Murray compared the voting patterns in "two recent high profile elections involving Hispanic candidates." (Pl. Ex. 18 at 20.) Dr. Murray first chose the 2010 race for Harris County Commissioner, between Sylvia Garcia, the Democratic candidate, and Jack Morman, the Republican candidate. (Id.) Dr. Murray also selected the 2012 race for Harris County Sheriff, between Adrian Garcia, the Democratic candidate, and Louis Guthrie, the Republican candidate. (Id.) Dr. Murray hypothesized that these races were "situation[s] where voters had more information and a lot of people were voting" due to the increased media coverage. (Tr. 2 at 383:12-15.)
For these two races, Dr. Murray compared the returns from precincts that were largely Hispanic or white. (Tr. 2 at 383:15-21.) His results are summarized as follows:
DR. MURRAY'S EXOGENOUS ELECTIONS
| 2010 County Commissioner | 2012 Sheriff | |||||
| Hispanic Precincts | Votes for Garcia | Votes for Morman | Percentage for Garcia | Votes for Garcia | Votes for Guthrie | Percentage for Garcia |
| 029 | 387 | 193 | 66.7 | 673 | 273 | 71.1 |
| 093 | 480 | 331 | 58.5 | 907 | 439 | 67.4 |
| 188 | 248 | 162 | 60.5 | 437 | 215 | 67.0 |
| 190 | 267 | 142 | 65.3 | 343 | 141 | 70.9 |
| 301 | 377 | 226 | 62.5 | 699 | 348 | 66.8 |
| Mean | 62.7 | 68.6 | ||||
| Anglo Precincts | Votes for Garcia | Votes for Morman | Percentage for Garcia | Votes for Garcia | Votes for Guthrie | Percentage for Garcia |
| 348 | 224 | 561 | 28.5 | 303 | 670 | 31.1 |
| 419 | 422 | 745 | 36.2 | 564 | 958 | 37.1 |
| 534 | 507 | 1154 | 30.5 | 660 | 1573 | 29.6 |
| 679 | 392 | 1007 | 28.0 | 556 | 1466 | 27.5 |
| 752 | 175 | 488 | 26.4 | 240 | 566 | 29.8 |
| Mean | 29.9 | 31.0 |
(Pl. Ex. 18 at 21.) In the 2010 race for County Commissioner, 62.7% of votes from "Hispanic precincts" were cast for Sylvia Garcia, compared to 29.9% of votes from "Anglo precincts." In the 2012 race for Sheriff, 68.6% of votes from "Hispanic precincts" were cast for Adrian Garcia, compared to 31% of votes from "Anglo precincts."
According to Dr. Murray, these two elections are representative of "the reality that Hispanic voters in the PISD, as in Texas, prefer different candidates than do Anglo voters." (Pl. Ex. 18 at 20.) He concluded that "there is a significant amount of voter cohesion, both among Hispanics and Anglos, and that they're cohering behind different candidates." (Tr. 2 at 383-84.) He acknowledges that the difference in voting patterns tracks the difference in political parties, with the "Hispanic precincts" voting for the Democratic candidate and "Anglo precincts" voting for the Republican candidate. (Pl. Ex. 18 at 20.) Dr. Murray explains that Hispanic voters prefer different candidates than white voters because "the two major parties have assumed very different positions on a range of issues that are important to voters of different races and ethnicity both locally and statewide." (Id.)
Dr. Matt Barreto
Drs. Barreto and Pedraza analyzed the results of 26 exogenous elections, as spread across 58 polling places in the PISD. (Pl. Ex. 15 at 6.) Dr. Barreto acknowledged that the 2010 race for County Commissioner of Precinct 2 "only had 30 or so of the available 58 precincts." (Tr. 1 at 43:12-19 and 87:3-5.) In its analysis, the Court considers only the 25 elections that included the same set of PISD voters who were eligible to participate in PISD Board elections. (See Pl. Ex. 15 at 3.)
Using ecological inference and ecological regression, Drs. Baretto and Pedraza estimated the percentage of support for Hispanic candidates from Hispanic, non-Hispanic white, and non-Hispanic black voters. The Court summarizes these results for Hispanic and non-Hispanic white voters in the following table:
DR. BARRETO'S EXOGENOUS ELECTIONS
| King's Ecological Inference | Goodman's Ecological Regression | ||||
| Percent of Hispanic Voters Supporting Candidate | Percent of White Voters Supporting Candidate | Percent of Hispanic Voters Supporting Candidate | Percent of White Voters Supporting Candidate | ||
| 2002 | County Treasurer | ||||
| Richard Garcia (D) | 77.2% | 2.6% | 78.6% | 6.8% | |
| 2002 | Judge, County Crim. Ct. No. 2 | ||||
| Silvia Pubchara (D) | 78.1% | 3.2% | 80.4% | 7.8% | |
| 2002 | Judge, County Crim. Ct. No. 5 | ||||
| Blanca Lopez (D) | 80.0% | 2.7% | 80.8% | 7.2% | |
| 2006 | Lieutenant Governor | ||||
| Maria Alvarado (D) | 60.7% | 3.7% | 59.7% | 14.1% | |
| 2006 | County Treasurer | ||||
| Richard Garcia (D) | 65.9% | 10.9% | 66.4% | 21.9% | |
| Orlando Sanchez (R) | 33.8% | 89.3% | 33.6% | 78.1% | |
| 2006 | Judge, County Crim. Ct. No. 2 | ||||
| Silvia Pubchara (D) | 67.6% | 7.7% | 68% | 18.9% | |
| 2006 | District Judge, 183rd Judicial Dist. | ||||
| Vanessa Velasquez (R) | 34.7% | 87.3% | 31.9% | 77.8% | |
| 2008 | Family District Judge, 312th Judicial Dist. | ||||
| Robert Hinojosa (D) | 70.3% | 0.9% | 67.9% | 12.6% | |
| 2008 | U.S. Senator | ||||
| Richard J. Noriega (D) | 69.7% | 2.1% | 67.3% | 14.3% | |
| 2008 | District Judge, 55th Judicial Dist. | ||||
| Dion Ramos (D) | 69.8% | 0.8% | 67.3% | 11.8% | |
| 2010 | Lieutenant Governor | ||||
| Linda Chavez-Thompson (D) | 58.8% | 0.6% | 59.5% | 5.2% | |
| 2010 | Commissioner of the General Land Office | ||||
| Hector Uribe (D) | 59.5% | 0.5% | 61.3% | 2.8% | |
| 2010 | Justice, Supreme Ct., Place 9 | ||||
| Eva Guzman (R) | 38.6% | 99.1% | 39.2% | 92.0% | |
| 2010 | Justice, 1st Ct. of Appeals Dist., Place 4 | ||||
| Michael Gomez (D) | 62.1% | 1.0% | 63.5% | 3.9% | |
| 2010 | District Judge, 55th Judicial Dist. | ||||
| Dion Ramos (D) | 61.7% | 0.7% | 63.5% | 5.1% | |
| 2010 | District Judge, 183rd Judicial Dist. | ||||
| Vanessa Velasquez (R) | 39.5% | 98.9% | 38.2% | 96.0% | |
| 2010 | Family District Judge, 312th |
| Judicial Dist. | |||||
| Robert Hinojosa (D) | 63.0% | 0.7% | 63.7% | 5.8% | |
| 2010 | Family District Judge, 314th Judicial Dist. | ||||
| David Longoria (D) | 64.1% | 0.8% | 64.4% | 6.4% | |
| 2010 | Judge, County Crim. Ct., No. 2 | ||||
| Mary Connealy Acosta (D) | 62.4% | 0.8% | 63.3% | 5.3% | |
| 2010 | Judge, County Crim. Ct., No. 4 | ||||
| Alfred G. Leal (D) | 63.4% | 0.9% | 63.8% | 5.7% | |
| 2010 | Judge, County Crim. Ct., No. 5 | ||||
| Alfred Valdez (D) | 61.5% | 0.9% | 63.1% | 3.2% | |
| 2010 | Judge, County Crim. Ct., No. 7 | ||||
| Shelia Acosta (D) | 62.7% | 0.9% | 64.2% | 4.0% | |
| 2010 | Judge, County Crim. Ct., No. 11 | ||||
| Mark Diaz (D) | 62.3% | 0.7% | 63.8% | 3.6% | |
| 2010 | County Treasurer | ||||
| Orlando Sanchez (R) | 39.9% | 99.5% | 39.9% | 93.9% | |
| 2012 | Justice, 14th Ct. of Appeals, Place 8 | ||||
| Julia Maldonado (D) | 67.6% | 0.6% | 66.3% | 8.5% |
(See Pl. Ex. 15 at 12-118.)
Dr. Barreto concluded that the exogenous elections provide evidence of both political cohesion and racial polarization. Dr. Barreto testified that "[t]here is a very strong degree of political cohesion" present among Hispanic voters. (Tr. 1 at 52:1-2.) In each exogenous election, the majority of Hispanic voters and the majority of white voters support opposite candidates. Dr. Barreto noted that "the gaps that we demonstrate between Hispanic and Anglo voting preferences are among the strongest and largest that I have documented," pointing to the 77 point gap present in Blanca Lopez's election in 2002. (Tr. 1 at 52:10-16.) Dr. Barreto reports that these exogenous elections provide "strong, and statistically significant evidence of racially polarized voting." (Pl. Ex. 15 at 6.)
The expert report of Drs. Barreto and Pedraza highlights two cases in which Hispanic candidates were not preferred by Hispanic voters. (Pl. Ex. 15 at 6.) This Court counts five - Vanessa Velasquez in 2006 and 2010, Eva Guzman in 2010, and Orlando Sanchez in 2006 and 2010, highlighted in the above chart. These Hispanic candidates, who all ran as Republicans, garnered no more than 40% of the Hispanic vote under either analysis, but received overwhelming support from white voters.
In two of these five contests, Hispanic candidates ran against each other. (Pl. Ex. 15 at 6.) Orlando Sanchez ran against Richard Garcia in 2006 for County Treasurer, and Vanessa Velasquez ran against Michael Gomez in 2010 for District Judge in the 183rd Judicial District. In both cases, Hispanic voters preferred the Hispanic Democratic candidate rather than the Hispanic Republican candidate. (Id.) Drs. Barreto and Pedraza concluded that Ms. Velasquez and Mr. Sanchez were "significant outliers according to the data," but failed to explain the three other races in which an Hispanic candidate lost the Hispanic vote to a white opponent. (See id.)
When questioned on the partisan nature of these elections and their results, Dr. Barreto explained the need and utility of partisan elections. First, Dr. Barreto stated that there are very few non-partisan elections in Harris County to analyze. (Tr. 1 at 53:11-16.) Second, Dr. Barreto explained that "the partisan loyalties of voters in southern states, and especially in a state like Texas, are often driven by racial attitudes and racial prejudice." (Tr. 1 at 54:2-7.) "[I]t's not enough to say this is just partisan politics," Dr. Barreto testified, "because the research underlying that partisan ide[o]logy suggests that voters are relying on their racial attitudes when they pick their party identification." (Tr. 1 at 55:7-12.)
Dr. John Alford
Dr. Alford contests the comparability of Dr. Barreto's chosen exogenous elections to the PISD's endogenous elections. Dr. Alford notes the differences in the positions themselves. The exogenous elections are all for full-time, salaried, statewide or countywide positions. (Def. Ex. 29 at 13.) The PISD Board positions, in contrast, are part-time, inherently local, and without salary. (Id. at 13-14.) Dr. Alford further explains that the elections themselves are very different. The exogenous elections are all "on the November even-numbered year ballot with contests for federal political office at the top of the ballots, partisan affiliation of the candidates identified, and the option of casting a straight party ticket, which about two-thirds of the voters in these elections take." (Id. at 13.) PISD elections occur in May, without primaries, and "there are no partisan elections on the ballot, no party labels, and no straight ticket voting." (Id. at 14.)
Dr. Alford nevertheless considered the statistical analysis provided by Drs. Baretto and Pedraza. Using the estimates found by Drs. Barreto and Pedraza with ecological inference, Dr. Alford produced the following summary of exogenous elections, formatted slightly differently by the Court. In addition to the percentage of support from Hispanic and white voters for each Hispanic candidate, Dr. Alford included the percentage of support from Hispanic and white voters for each Democratic candidate.
DR. ALFORD'S EXOGENOUS ELECTIONS
| Year | Office | Party ID | Candidate | Percent of Anglo Vote for the Hispanic Candidate | Percent of Hispanic Vote for the Hispanic Candidate | Percent of Anglo Vote for the Democratic Candidate | Percent of Hispanic Vote for the Democratic Candidate |
| 2002 | County Treasurer | Dem. | Richard Garcia | 2.6% | 77.2% | 2.6% | 77.2% |
| Rep. | Jack Cato | ||||||
| 2002 | Judge, County | Dem. | Silvia Pubchara | 3.2% | 78.1% | 3.2% | 78.1% |
| Crim. Ct. No. 2 | Rep. | Michael Allen Peters | |||||
| 2002 | Judge, County | Dem. | Blanca E. Lopez | 2.7% | 80.0% | 2.7% | 80.0% |
| Crim. Ct. No. 5 | Rep. | M. Stewart Harris | |||||
| 2006 | Lieutenant | Dem. | Maria Luisa Alvarado | 3.7% | 60.7% | 3.7% | 60.7% |
| Governor | Rep. | David Dewhurst | |||||
| 2006 | County Treasurer | Dem. | Richard Garcia | 10.9% | 65.9% | 10.9% | 65.9% |
| Rep. | Orlando Sanchez | 89.3% | 33.8% | ||||
| 2006 | Judge, County | Dem. | Silvia Pubchara | 7.7% | 67.6% | 7.7% | 67.6% |
| Crim. Ct. No. 2 | Rep. | Bill Harmon | |||||
| 2006 | District Judge, | Dem. | Robert Voigt | 12.7% | 65.3% | ||
| 183rd Judicial Dist. | Rep. | Vanessa Velasquez | 87.3% | 34.7% | |||
| 2008 | Family District | Dem. | Robert Hinojosa | 0.9% | 70.3% | 0.9% | 70.3% |
| Judge, 312th | Rep. | David Farr | |||||
| Judicial Dist. | |||||||
| 2008 | U.S. Senator | Dem. | Richard J. (Rick) Noriega | 2.1% | 69.7% | 2.1% | 69.7% |
| Rep. | John Cornyn | ||||||
| 2008 | District Judge, 55th Judicial Dist. | Dem. | Dion Ramos | 0.8% | 69.8% | 0.8% | 69.8% |
| Rep. | Jeff Shadwick | ||||||
| 2010 | Lieutenant Governor | Dem. | Linda Chavez-Thompson | 0.6% | 58.8% | 0.6% | 58.8% |
| Rep. | David Dewhurst | ||||||
| 2010 | Commissioner of the General Land Office | Dem. | Hector Uribe | 0.5% | 59.5% | 0.5% | 59.5% |
| Rep. | Jerry Patterson | ||||||
| 2010 | Justice, Supreme Ct., Place 9 | Dem. | Blake Bailey | 0.9% | 61.4% | ||
| Rep. | Eva Guzman | 99.1% | 38.6% | ||||
| 2010 | Justice, 1st Ct. of Appeals Dist., Place 4 | Dem. | Michael Gomez | 1.0% | 62.1% | 1.0% | 62.1% |
| Rep. | Evelyn Keyes | ||||||
| 2010 | District Judge, 55th Judicial Dist. | Dem. | Dion Ramos | 0.7% | 61.7% | 0.7% | 61.7% |
| Rep. | Jeff Shadwick | ||||||
| 2010 | District Judge, 183rd Judicial Dist. | Dem. | Michael Gomez | 1.1% | 60.5% | ||
| Rep. | Vanessa Velasquez | 98.9% | 39.5% | ||||
| 2010 | Family District Judge, 312th Judicial Dist. | Dem. | Robert Hinojosa | 0.7% | 63.0% | 0.7% | 63.0% |
| Rep. | David Farr | ||||||
| 2010 | Family District Judge, 314th Judicial Dist. | Dem. | David Longoria | 0.8% | 64.1% | 0.8% | 64.1% |
| Rep. | John F. Phillips | ||||||
| 2010 | Judge, County Crim. Ct. No. 2 | Dem. | Mary Connealy Acosta | 0.8% | 62.4% | 0.8% | 62.4% |
| Rep. | Bill Harmon | ||||||
| 2010 | Judge, County Crim. Ct. No. 4 | Dem. | Alfred G. "Al" Leal | 0.9% | 63.4% | 0.9% | 63.4% |
| Rep. | John Clinton | ||||||
| 2010 | Judge, County Crim. Ct. No. 5 | Dem. | Alfred "Bud" Valdez | 0.9% | 61.5% | 0.9% | 61.5% |
| Rep. | Margaret Stewart Harris | ||||||
| 2010 | Judge, County Crim. Ct. No. 7 | Dem. | Shelia Acosta | 0.9% | 62.7% | 0.9% | 62.7% |
| Rep. | Pam Derbyshire | ||||||
| 2010 | Judge, County Crim. Ct. No. 11 | Dem. | Mark Diaz | 0.7% | 62.3% | 0.7% | 62.3% |
| Rep. | Diane Bull | ||||||
| 2010 | County Treasurer | Dem. | Billy Briscoe | 0.5% | 60.1% | ||
| Rep. | Orlando Sanchez | 99.5% | 39.9% | ||||
| 2012 | Justice, 14th Ct. of Appeals, Place 8 | Dem. | Julia Maldonado | 0.6% | 67.6% | 0.6% | 67.6% |
| Rep. | John Donovan |
(Def. Ex. 29 at 16.)
The first two columns of results, detailing the support of white and Hispanic voters for Hispanic candidates, are taken from Dr. Barreto's own results. As seen before, less than 1% of white voters, compared to 60 to 70% of Hispanic voters, typically support Hispanic candidates. There are, however, five "jarring exceptions" to this pattern when white voters heavily supported Hispanic candidates, whom Hispanic voters have not favored. (Def. Ex. 29 at 15.) First, in the 2006 race for County Treasurer, Orlando Sanchez won 89.3% of the white vote and 33.8% of the Hispanic vote. Second, in the 2006 race for District Judge of the 183rd Judicial District, Vanessa Velasquez won 87.3% of the white vote and 34.7% of the Hispanic vote. Third, in the 2010 race for a position on the Texas Supreme Court, Eva Guzman won 99.1% of the white vote and 38.6% of the Hispanic vote. Fourth, in the 2010 race for District Judge of the 183rd Judicial District, Vanessa Velasquez won 98.9% of the white vote and 39.5% of the Hispanic vote. Fifth, in the 2010 race for County Treasurer, Orlando Sanchez won 99.5% of the white vote and 39.9% of the Hispanic vote.
Dr. Alford added the last two columns, detailing the support of white and Hispanic voters for the Democratic candidates, to explain these five anomalies. Without exception, the white vote for the Democratic candidate is minimal and the Hispanic vote for the Democratic candidate is above 50%. In sum, Dr. Alford states that predictions of voting behavior rely on the party, rather than the race or ethnicity, of the candidate. (Def. Ex. 29 at 18.) "What the exogenous election analysis shows then, is that partisanship, and not the race or ethnicity of the candidates, is driving polarization in these contests." (Id. at 19.)
Dr. Alford argues that the evidence of partisan polarization in exogenous elections is not informative of voting patterns in the non-partisan endogenous elections. (Def. Ex. 29 at 19.) Partisanship is "built in to the exogenous elections from the very start in the form of a party nomination system, incorporated into the ballot itself with party indications for each candidate, and ultimately mechanically connected to the majority of votes cast through the mechanism of straight-party ticket voting." (Id. at 20.) The PISD elections, in contrast, are not partisan. (Id. at 19.)
"In the end," Dr. Alford explains, "the best proof of the irrelevance of the partisan exogenous elections for making inferences about likely voting behavior in PISD elections is the fact that the voting patterns of Anglos and Hispanics in the PISD elections over the last decade don't look anything like the voting patterns of Anglos and Hispanics in the partisan exogenous election[s] over the last decade." (Def. Ex. 29 at 20.) In the partisan exogenous elections, white and Hispanic voters always give majority support to different candidates. In the non-partisan endogenous elections, white and Hispanic voters have given majority support to the same candidate, regardless of race, in five out of six contested elections.
3. Recent City of Pasadena Election
On November 5, 2013, the City of Pasadena voted on a proposition ("Proposition 1") that would change the electoral system for places on the City Council from eight single-member districts to six single-member districts and two at-large districts. Dr. Murray testified that 27 of the 37 precincts that voted in this election "are wholly or largely in the Pasadena Independent School District." (Tr. 2 at 387-88.) Analyzing the unofficial election results for the proposition therefore provides additional information in the analysis of racially polarized voting.
The experts for both parties agree that the results of the recent election demonstrate racially polarized voting. Using ecological regression to analyze the voting results from the 37 precincts, Drs. Barreto and Pedraza found that 73.8% of the white voters and 9.2% of the Hispanic voters supported the proposition. (Pl. Ex. 22.) Dr. Alford testified that he "replicate[d] the same analysis using ecological inference and using Spanish surname registered voter[s] as the measure for the polling place" and reached the same result. (Tr. 3 at 597-98.) Dr. Alford summarized that the results showed "strong Hispanic opposition," and "some Anglo crossover." (Tr. 3 at 598:16-21.)
Plaintiffs' Proposed Findings of Fact state that "the analysis of racial polarization performed by the experts included only PISD voting precincts." (Doc. No. 28 at 8.) The data produced by Drs. Barreto and Pedraza, however, state that there are 37 points of observation, which would correspond to the City of Pasadena as a whole. (See Pl. Ex. 22.)
Dr. Murray stated that this data "strongly confirms [his] initial conclusion that there is racially polarized voting in this area." (Tr. 2 at 391:6-8.) In the City as a whole, the proposition passed by a slim majority of 87 votes. (Tr. 2 at 388:20-21.) When broken down into individual precincts, however, the proposition passed or failed "by a pretty sizeable margin." (Tr. 2 at 388:21-23.) According to Dr. Murray's review of the data, "[t]he race ethnicity makeup of the district[s] are largely driving the voting pattern." (Tr. 2 at 389:3-4.)
Although the proposition passed narrowly in the City of Pasadena, Dr. Alford hypothesized that it would have lost within the PISD. The following map helps to visualize the overlap between the two:
Image materials not available for display. (Def. Ex. 31.) In the portion of the City that falls outside of the PISD's boundaries, in which the voting age population is 30.63% Hispanic and 60.39% Non-Hispanic White, the proposition won by 373 votes. (Def. Ex. 32-33.) In the portion of the City that falls within the PISD's boundaries, in which the voting age population is 62.31% Hispanic and 33.41% Non-Hispanic White, the proposition lost by 286 votes. (Def. Ex. 32-33.) There are obviously no election results for the districts within the PISD that fall outside of the City. The voting age population of these PISD-only districts is 58.97% Hispanic and 21.62% Non-Hispanic White. (Def. Ex. 32.) Assuming that these districts were as racially polarized as those within the City of Pasadena, Dr. Alford predicted that the proposition would have lost. (Tr. 3 at 602:2-7.)
Dr. Alford's results can be summarized as follows:
PROPOSITION ONE RESULTS
| Summarized Blocks Location | Total VAP | Percent Hispanic VAP | Percent Anglo VAP | Percent Black VAP | Percent Asian VAP | Percent Other VAP | Proposition Result |
| In City Only | 21,315 | 30.63% | 60.39% | 2.02% | 1 5.48% | 1.47% | Win by 373 |
| In Both City and School District | 82,168 | 62.31% | 33.41% | 1.90% | 1.52% | 0.86% | Lose by 286 |
| In School District Only | 84,320 | 58.97% | 21.62% | 11.78% | 6.60% | 1.04% | (Predicted Loss) |
(Def. Ex. 32-33.) As applies to the third factor of Gingles, Dr. Alford concluded that the recent election shows that voting was polarized, but white voters did not vote sufficiently as a bloc to defeat the preference of Hispanics within the school district. (Tr. 3 at 602:13-19.) Dr. Alford explained, "in part, that's simply . . . because of the demographics and partly because of the 25 percent plus crossover." (Tr. 3 at 602:17-19.)
4. Analysis
The Court concludes that there is no evidence of racially polarized voting in the recent endogenous elections for the PISD. Plaintiffs' experts concede as much, but claim that a reliable conclusion requires more data. (Pl. Ex. 15 at 6; Tr. 1 at 36:16-22.) Drs. Barreto and Pedraza analyzed the results of two endogenous elections across ten or eleven polling places. They found "no evidence of racially polarized voting" in these elections since the Hispanic candidates in both races won with the support of Hispanic and non-Hispanic voters. (Pl. Ex. 15 at 6.) For a fuller analysis, Drs. Barreto and Pedraza "would have wanted to see more data, more precincts made available to us, and we would have also wanted to see more elections." (Tr. 1 at 39:2-4.)
Dr. Alford analyzed more "polling places" and more endogenous elections and also concluded that "the voting in PISD school board elections is not polarized." (Def. Ex. 29 at 6.) To increase the number of reporting units, Dr. Alford separately considered the results of early voting and election day voting. (Def. Ex. 29 at 6-7.) The Court finds this to be a reliable technique. To increase the number of endogenous elections, Dr. Alford also considered contested endogenous elections that did not include an Hispanic candidate. Endogenous elections that include an Hispanic candidate are more probative in a Section 2 analysis, but endogenous elections between white candidates should not be ignored. In three additional endogenous elections, Hispanic and non-Hispanic voters coalesced behind the winning candidate. In total, Hispanic-preferred candidates won five of the six contested endogenous elections over the past ten years. (Id. at 11.)
Analyses of exogenous elections can supplement, though not replace, those of endogenous elections. Rodriguez v. Harris Cnty., Tex., 964 F. Supp. 2d 686, 759 (S.D. Tex. 2013). The Fifth Circuit has "repeatedly endorsed the analysis of exogenous elections in Section 2 vote dilution claims." Rodriguez v. Bexar Cnty., Tex., 385 F.3d 853, 860 n.5 (5th Cir. 2004). The Court must, however, carefully evaluate the probative value of exogenous elections to the particular election at issue. See Westwego Citizens for Better Gov't v. City of Westwego, 872 F.2d 1201, 1208 n.8 (5th Cir. 1989) (noting that "evidence from other elections should not be deemed irrelevant per se to plaintiffs' claims, but must be evaluated according to its particular probative value").
In this case, Dr. Murray selected "two recent high profile elections involving Hispanic candidates." (Pl. Ex. 18 at 20.) In both, voters in "Hispanic precincts" largely supported the Hispanic Democratic candidates, who did not garner majority support from "Anglo precincts." (Id. at 21.) Dr. Murray highlighted the different voting patterns between Hispanic and white voters, although he acknowledged that the differences could be attributed to race or political party. (Id. at 20.)
Drs. Barreto and Pedraza selected 26 exogenous elections, including races for County Treasurer, Judge, and U.S. Senator. (See Pl. Ex. 15 at 4.) The same electorate for PISD Board elections was eligible to vote in 25 of these exogenous elections. In 20 of these elections, the Hispanic candidate won a majority of Hispanic support, but minimal white support. In stark contrast, five Hispanic candidates, who ran as Republicans, were overwhelmingly supported by white voters instead of Hispanic voters. Nevertheless, for all 25 elections, Hispanic and white voters supported opposite candidates.
Dr. Alford demonstrated that the voting patterns in these exogenous elections better tracked the party, not the race or ethnicity, of the candidate. In all 25 elections, Hispanic voters preferred the Democratic candidate and white voters preferred the Republican candidate, regardless of the candidate's race. (See Def. Ex. 29 at 16.) Strikingly, in the 2006 race for District Judge, the 2010 race for Justice, and the 2010 race for County Treasurer, Hispanic voters supported the white Democrat and white voters supported the Hispanic Republican.
Plaintiffs' experts responded that party affiliations cannot be disentangled from racial politics. (Tr. 1 at 54:8-13.) According to Dr. Barreto, the racial politics of both parties continues to dominate party identification in the south. (Tr. 1 at 54-55.) Dr. Murray testified that "the principal driver[s] of the increased racial and ethnic polarization are the policy decisions made by political leaders, particularly Republican political leaders." (Tr. 2 at 358:18-24.) He explains that the "Republican Party has abandoned the outreach efforts toward Hispanics" and moved "significantly to the right" on issues such as health care and comprehensive immigration reform. (Pl. Ex. 18 at 20-21.)
Dr. Barreto relied on an almost identical set of exogenous elections and the same argument in an earlier challenge to the redistricting of Harris County commissioner precincts. See Rodriguez v. Harris Cnty., Tex., 964 F. Supp. 2d 686, 761-63, 775-77 (S.D. Tex. 2013). In that case, the district court found that the regression results of the two endogenous elections "indicate[d] that Anglos vote as a bloc to defeat the Latino-preferred candidate." Id. at 764. "This evidence is further supported by the regression analysis of the exogenous elections." Id. (emphasis added). Although the defendant presented some evidence that the voting pattern could be explained by partisanship, the court found that the "[p]laintiffs have produced countervailing evidence that race is a consideration for voters in deciding the candidate for whom they will cast a ballot." Id. at 777. The court concluded that the plaintiffs had established racially polarized voting as required by the third Gingles factor. Id.
The list of exogenous elections for the two is identical but for two elections. The earlier case did not include data for Richard Garcia's race for County Treasurer in 2006, or Julia Maldonado's race for Justice on the 14th Court of Appeals in 2012.
--------
These exogenous elections, well-suited to analyze the partisan elections at issue in Rodriguez, are of limited probative value in this case. In Rodriguez, Dr. Barreto "only selected races that were sufficiently similar to the endogenous elections [of County Commissioners] in terms of the candidates, the level of office, and the partisan nature of the race" in order "[t]o maximize the probative value of the exogenous races." 964 F. Supp. 2d at 760. The results of these exogenous elections were used to support similar results from the endogenous elections. Id. at 764. In this case, the exogenous elections chosen by Dr. Barreto are not similar to the endogenous elections in a critical respect, and the results from the exogenous elections run counter to the results from the endogenous elections. The exogenous elections chosen by Dr. Murray suffer from the same flaws. These partisan exogenous elections cannot be used to overcome the evidence supplied by the non-partisan endogenous elections.
The Court finds the recent vote in the City of Pasadena to be more probative of the voting patterns within the PISD than the exogenous elections provided by Drs. Barreto or Murray since neither is driven by partisan politics. Nevertheless, the results from Proposition 1 do not help Plaintiffs to carry their burden for the third factor of Gingles. Plaintiffs must present evidence not only that white voters and Hispanic voters hold different preferences, but that white voters are able "usually to defeat the minority's preferred candidate." See Gingles, 478 U.S. at 51. "[T]o prove legally significant white bloc voting," the Fifth Circuit explained, "minority plaintiffs must present evidence of 'a white bloc vote that normally will defeat the combined strength of minority support plus white 'crossover' votes.'" Rangel v. Morales, 8 F.3d 242, 245 (5th Cir. 1993) (quoting Gingles, 478 U.S. at 56).
In the recent City of Pasadena election, there is no evidence of legally significant white bloc voting in the PISD. Hispanic voters overwhelmingly opposed the proposition, joined by approximately 25% of white voters. Both experts agreed that success at the individual district level can be predicted by the district's demographics. Given the demographics of the districts within the PISD, Dr. Alford hypothesized that the Hispanic voters' preference would have carried the day. (Tr. 3 at 660:6-11.)
In sum, the only evidence of legally significant racially polarized voting in the record comes from the two exogenous elections chosen by Dr. Murray and the 25 exogenous elections chosen by Drs. Barreto and Pedraza. As explained, the Court does not find these partisan exogenous elections to be particularly probative of voting patterns for the non-partisan PISD Board elections. The Court refuses to rely on these exogenous elections as the only evidence of legally significant racial polarization in the PISD Board elections. See Citizens for a Better Gretna v. City of Gretna, La., 834 F.2d 496, 502 (5th Cir. 1987) ("Although exogenous elections alone could not prove racially polarized voting in Gretna aldermanic elections, the district court properly considered them as additional evidence of bloc voting - particularly in light of the sparsity of available data.").
III. CONCLUSIONS OF LAW
A. Gingles I: A Sufficiently Large and Geographically Compact Group
Plaintiffs have failed to demonstrate that the Hispanic population in the PISD is "sufficiently large and geographically compact to constitute a majority in a single-member district." Gingles, 478 U.S. at 50. The Fifth Circuit has made clear that, "without a strict showing of its probativeness, Spanish-surname data are disfavored, and census data based upon self-identification provides the proper basis for analyzing Section 2 vote dilution claims." Rodriguez, 385 F.3d at 866 n.18.
In this case, Plaintiffs provided demonstration districts with more than 50% SSRVs, and assumed that the CVAP figure would be higher. Plaintiffs' expert conceded that five-year estimates of CVAP from the ACS were available if requested, but did not provide them. The Court finds no reason to replace available census data with unreliable Spanish-surname data. As such, Plaintiffs have not met their burden of proving the first factor of Gingles.
B. Gingles II: Political Cohesion of the Minority
Plaintiffs have proven that Hispanic voters in the PISD are politically cohesive. Endogenous elections show, and exogenous elections support, that Hispanic candidates win the majority of Hispanic votes.
C. Gingles III: Legally Significant Bloc Voting
Plaintiffs have failed to show that the white voters in the PISD vote sufficiently as a bloc to generally defeat the Hispanic voters' preference. There is no evidence of racially polarized voting in the endogenous elections. In fact, Hispanic voters and white voters have generally supported the same candidates in PISD Board elections.
Plaintiffs rely heavily on the evidence of racially polarized voting in exogenous elections. In two high-profile exogenous elections selected by Dr. Murray and 25 exogenous elections selected by Drs. Barreto and Pedraza, white voters voted cohesively against the clear Hispanic choice. The Court finds that these exogenous elections are of minimal probative value since the exogenous elections are of a critically different nature than the endogenous elections. The Court cannot find the presence of racially polarized voting in the non-partisan PISD elections based solely on the evidence of racially polarized voting in the partisan exogenous elections.
The recent election in the City of Pasadena regarding electoral systems carries more probative weight in the present inquiry. While the election results show racially polarized voting, it does not reach the level of legal significance required by the third factor of Gingles. Plaintiffs did not rebut Dr. Alford's contention that white voters would have failed to vote sufficiently as a bloc to defeat the Hispanic preference within the PISD's boundaries.
D. Totality of the Circumstances
Because the Plaintiffs have not met the three preconditions required by Gingles, the Court does not move to the next step of a Section 2 analysis. The Supreme Court recognized that Plaintiffs might be able to establish other factors for the totality analysis, but still not sustain a Section 2 claim if the three most important factors - those that make up the preconditions - have not been met. Gingles, 478 U.S. at 48 n.15. "Minority voters may be able to prove that they still suffer social and economic effects of past discrimination, that appeals to racial bias are employed in election campaigns, and that a majority vote is required to win a seat, but they have not demonstrated a substantial inability to elect caused by the use of a multimember district." Id.
IV. CONCLUSION
The PISD Board has included an Hispanic member since 1987. This is far from proportional with the rapidly growing Hispanic population in the community and the schools, but the lack of proportional representation in office does not prove a lack of opportunity to participate in the electoral process. Plaintiffs have not met their burden of proving that the at-large electoral mechanism employed by the PISD is unconstitutional. The Court must find in favor of Defendants since Plaintiffs have not proven two of Gingles' preconditions.
IT IS SO ORDERED.
__________
KEITH P. ELLISON
UNITED STATES DISTRICT JUDGE
Appendix 1
Table Four: Hispanic Registered Voters by Precincts to the Pasadena Independent
School District as of June, 2012
| Pet | All Registered Voters | Hispanic Registered Voters | % Hispanic Registered Voters |
| 029 | 2,637 | 2,040 | 77.4 |
| 076 | 2,326 | 747 | 32.1 |
| 077 | 558 | 148 | 26.5 |
| 093 | 3,248 | 2,222 | 68.4 |
| 134 | 0 | ||
| 170 | 1,676 | 805 | 48.0 |
| 181 | 5 | 4 | 80.0 |
| 188 | 1,594 | 993 | 623 |
| 190 | 1,716 | 1,204 | 70.2 |
| 191 | 1,364 | 899 | 65,9 |
| 211 | 0 | ||
| 221 | 960 | 592 | 60.9 |
| 242 | 1,552 | 730 | 47.0 |
| 257 | 2,339 | 1,033 | 44.2 |
| 267 | 1,332 | 694 | 52.1 |
| 275 | 23 | 6 | 26.1 |
| 277 | 1,467 | 808 | 55.1 |
| 278 | 2,369 | 1,196 | 50.5 |
| 279 | 937 | 350 | 37.4 |
| 280 | 1,500 | 732 | 48.8 |
| 289 | 1,505 | 821 | 54.6 |
| 301 | 2,377 | 1,552 | 65.3 |
| 302 | 264 | 114 | 43.2 |
| 308 | 568 | 238 | 41.9 |
| 328 | 1,344 | 696 | 51.8 |
| 329 | 105 | 53 | 50.4 |
| 347 | 3,257 | 1,156 | 35.5 |
| 348 | 1,721 | 329 | 19.1 |
| 394 | 844 | 484 | 57.3 |
| 404 | 2,565 | 1,135 | 44.2 |
| Pct | Reg Voters | SSRV | SSRV % |
| 417 | 1,953 | 531 | 27.2 |
| 418 | 3,356 | 1,087 | 32.4 |
| 419 | 2,675 | 784 | 29.3 |
| 475 | 1,610 | 461 | 28.5 |
| 476 | 1,904 | 562 | 29.5 |
| 527 | 556 | 368 | 66.2 |
| 534 | 2,467 | 619 | 25.1 |
| 536 | 4,466 | 1,954 | 43.8 |
| 537 | 150 | 15 | 10.0 |
| 545 | 279 | 55 | 19.7 |
| 654 | 3,297 | 1,026 | 31.1 |
| 679 | 3,276 | 715 | 21.8 |
| 696 | 1,551 | 442 | 28.5 |
| 708 | 561 | 336 | 59.9 |
| 715 | 1,368 | 272 | 19.9 |
| 718 | 2 | 0 | |
| 752 | 1,522 | 353 | 23.2 |
| 755 | 3,460 | 1,514 | 43.8 |
| 762 | 0 | ||
| 763 | 1,161 | 794 | 68.4 |
| 770 | 247 | 130 | 52.6 |
| 774 | 229 | 27 | 11.8 |
| 777 | 634 | 430 | 67.8 |
| 785 | 229 | 145 | 63.3 |
| 786 | 1,288 | 848 | 65.8 |
| 793 | 2,386 | 926 | 38.8 |
| 796 | 467 | 213 | 45.6 |
| 842 | 3,032 | 910 | 30.0 |
| 849 | 0 | ||
| Total | 86,279 | 37,295 | 43.2 |
Source: Hispanic Summary Report, June 4, 2012, Office of the Secretary of State, State of Texas, pp. 109 - 137.
(Pl. Ex. 18 at 14-15.)