Ohio A. Philip Randolph Inst. v. Householder

HONORABLE KAREN NELSON MOORE, United States Circuit Judge, HONORABLE TIMOTHY S. BLACK, United States District Judge, HONORABLE MICHAEL H. WATSON, United States District Judge *992 TABLE OF CONTENTS

I. BACKGROUND...994

A. General Overview of the Facts...994

B. Procedural History...1009

II. SUMMARY OF THE EVIDENCE PRESENTED AT TRIAL...1010

A. Plaintiffs' Fact Witnesses...1010

B. Defendants' Fact Witnesses...1019

C. Plaintiffs' Expert Witnesses...1025

D. Defendants' and Intervenors' Expert Witnesses...1049

III. STANDING...1062

A. Vote-Dilution Claims...1063

B. First Amendment Associational Claim...1073

C. Article I Claim...1076

IV. JUSTICIABILITY, THE POLITICAL QUESTION DOCTRINE, AND THE ROLE OF THE FEDERAL COURTS IN REDISTRICTING...1076

A. Justiciability and The Political Question Doctrine...1076

B. Evidentiary Metrics and Statistics...1082

C. Pragmatic and Historical Considerations...1085

V. LEGAL STANDARDS AND APPLICATION...1092

A. Equal Protection Vote-Dilution Claim...1093

B. First Amendment Vote-Dilution Claim...1150

C. First Amendment Associational Claim...1151

D. Article I Claim...1163

VI. LACHES...1165

VII. REMEDY AND ORDER...1168

APPENDICES OF MAPS

Plaintiffs have brought this action alleging that H.B. 369, the redistricting plan enacted by the Ohio General Assembly and signed into law by the Governor in 2011, constitutes an unconstitutional partisan gerrymander under the First and Fourteenth Amendments and exceeds the powers *993 granted to the states under Article I, § 4 of the United States Constitution. As to the First and Fourteenth Amendment district-specific claims, we find that Districts 1-16 were intended to burden Plaintiffs' constitutional rights, had that effect, and the effect is not explained by other legitimate justifications. Moreover, we find that that the plan as a whole burdens Plaintiffs' associational rights and that burden is not outweighed by any other legitimate justification. Finally, we find that the plan exceeds the State's powers under Article I. Therefore, H.B. 369 is an unconstitutional partisan gerrymander. This opinion constitutes our findings of fact and conclusions of law pursuant to Federal Rule of Civil Procedure 52(a)(1).

Due to the length of this opinion, we provide the reader with the following, more concise summary:

"Partisan gerrymandering" occurs when the dominant party in government draws district lines to entrench itself in power and to disadvantage the disfavored party's voters. Plaintiffs in this action are individual Democratic voters from each of Ohio's sixteen congressional districts, two non-partisan pro-democracy organizations, and three Democratic-aligned organizations. They challenge the constitutionality of Ohio's 2012 redistricting map. Defendants are Ohio officials, and Intervenors are Ohio Republican Congressmen; Defendants and Intervenors both argue that the Plaintiffs' claims are not properly before this Court and defend the map's constitutionality on the merits.

In 2011, when Ohio's redistricting process began, Republican dominance in the Ohio State government meant that Republican state legislators could push through a remarkably pro-Republican redistricting bill without meaningful input from their Democratic colleagues. Ohio Republicans took advantage of that opportunity, and invidious partisan intent-the intent to disadvantage Democratic voters and entrench Republican representatives in power-dominated the map-drawing process. They designed the 2012 map using software that allowed them to predict the partisan outcomes that would result from the lines they drew based on various partisan indices that they created from historical Ohio election data. The Ohio map drawers did not work alone, but rather national Republican operatives located in Washington, D.C. collaborated with them throughout the process. These national Republicans generated some of the key strategic ideas for the map, maximizing its likely pro-Republican performance, and had the authority to approve changes to the map before their Ohio counterparts implemented them. Throughout the process, the Ohio and national map drawers made decisions based on their likely partisan effects.

The map drawers focused on several key areas of the Ohio map where careful map design could eke out additional safe Republican seats. They split Hamilton County and the City of Cincinnati in a strange, squiggly, curving shape, dividing its Democratic voters and preventing them from forming a coherent voting bloc, which ensured the election of Republican representatives in Districts 1 and 2. They drew a new District 3 in Franklin County, efficiently concentrating Democratic voters together in an area sometimes referred to as the "Franklin County Sinkhole." This strategy allowed them to secure healthy Republican majorities in neighboring Districts 12 and 15. They paired Democratic incumbent Representatives Kaptur and Kucinich to create the infamous "Snake on the Lake"-a bizarre, elongated sliver of a district that severed numerous counties. They drew a District 11 that departed from its traditional territory to snatch up additional African-American Democratic *994 voters in Summit County, allowing for the creation of a new District 16 in which a Republican incumbent representative could defeat a Democratic incumbent representative. They designed these districts with one overarching goal in mind-the creation of an Ohio congressional map that would reliably elect twelve Republican representatives and four Democratic representatives.

Ohio Republican legislators enacted the first iteration of the 2012 map, H.B. 319, in September 2011. Ohio voters then challenged the map, seeking to subject it to a voter referendum, but their efforts failed. As a result, Ohio Republicans passed a slightly different version of the map, H.B. 369, in December 2011. The changes they made did not materially alter the strong pro-Republican partisan leaning of the map's first iteration. Four cycles of congressional elections have occurred under the map embodied in H.B. 369. Each resulted in the election of twelve Republican representatives and four Democratic representatives. No district has been represented by representatives from different parties during the life of the map.

During a two-week trial, experts testified to the extremity of the gerrymander. They demonstrated that levels of voter support for Democrats can and have changed, but the map's partisan output remains stubbornly undisturbed. The experts used various metrics and methodologies to measure their findings, but several takeaways were universal: (1) the Ohio map sacrifices respect for traditional districting principles in order to maximize pro-Republican partisan advantage, (2) the Ohio map's pro-Republican partisan bias is extreme, compared both to historical plans across the United States and to other possible configurations that could have been adopted in Ohio, and (3) the Ohio map minimizes responsiveness and competition, rendering one consistent result no matter the particularities of the election cycle.

We join the other federal courts that have held partisan gerrymandering unconstitutional and developed substantially similar standards for adjudicating such claims. We are convinced by the evidence that this partisan gerrymander was intentional and effective and that no legitimate justification accounts for its extremity. Performing our analysis district by district, we conclude that the 2012 map dilutes the votes of Democratic voters by packing and cracking them into districts that are so skewed toward one party that the electoral outcome is predetermined. We conclude that the map unconstitutionally burdens associational rights by making it more difficult for voters and certain organizations to advance their aims, be they pro-Democratic or pro-democracy. We conclude that by creating such a map, the State exceeded its powers under Article I of the Constitution. Accordingly, we declare Ohio's 2012 map an unconstitutional partisan gerrymander, enjoin its use in the 2020 election, and order the enactment of a constitutionally viable replacement.

I. BACKGROUND

A. General Overview of the Facts

1. The redistricting process begins

Every ten years, the United States government conducts a census. The census results dictate the size of each state's delegation to the United States House of Representatives because House seats are based on population. Following the release of the census results, state legislatures redraw their United States congressional districts in order to reflect population changes. In Ohio, the 2010 census revealed that the State's comparative population stagnation required reducing the State's previous congressional delegation from *995 eighteen to sixteen. 1 In that same year, Ohioans elected a Republican Governor, elected a Republican majority in the State Senate, and flipped the Ohio House of Representatives to be majority Republican as well. 2 In the State of Ohio, the Ohio General Assembly is responsible for enacting legislation that delineates the federal congressional districts. 3 Both the State Senate and the State House of Representatives must pass such a bill by a simple majority and the Governor must then sign the bill into law. 4 Therefore, when map-drawing activities commenced in 2011, the Republican Party had effective control of all bodies necessary to pass a redistricting bill.

In Ohio, redistricting is facilitated by the Joint Legislative Task Force on Redistricting, Reapportionment, and Demographic Research ("Task Force"). The Task Force is a six-person bipartisan committee. 5 The Task Force does not actually draw the maps. Rather "it is the entity to which the state legislature appropriates money" so that the Task Force can then contract with other entities and individuals to assist in the redistricting process. 6 Prior to the 2011 redistricting, the Task Force requisitioned from Cleveland State University ("CSU") a dataset containing demographic and political data that map drawers of both parties could use in the redistricting process. 7 The practice of the Ohio General Assembly has been to allow the Task Force to allocate separate funds in equal amounts to the Ohio Democratic Caucus and the Ohio Republican Caucus and to allow the parties to conduct much of their redistricting work separately. 8 This is precisely what occurred during the 2011 map-drawing process. 9 Eventually, maps are produced that are then sent for the General Assembly to enact in a bill, which is then sent to the Governor. The Ohio Senate and House of Representatives also established committees on redistricting, chaired by Republicans State Senator Keith Faber and Representative Matthew Huffman, respectively.

2. Logistics of the Republican map drawing

Republican map-drawing planning occurred at both the State and federal levels, and the two levels worked together, collaborated, and consulted one another throughout the process. 10 At the State level, Ray DiRossi and Heather Mann 11 served as the principal on-the-ground map drawers. 12 DiRossi had previously been employed as a staffer for Republican members of the General Assembly and as a fundraiser for the Ohio Republican Senate *996 Campaign Committee. 13 He was also deeply involved in the 2001 redistricting process following the 2000 census. 14 Mann had been working for the Ohio House Republican Caucus since 2004, most recently as Deputy Legal Counsel and Redistricting Director, reporting to Speaker of the Ohio House of Representatives William Batchelder ("Speaker Batchelder"). 15 It was decided that both DiRossi and Mann should formally cease their employment with the Ohio House Republican Caucus and instead conduct their map-drawing work as independent consultants. 16 As a consultant, Mann reported to Speaker Batchelder, 17 and DiRossi reported to State Senate President Tom Niehaus. 18 Troy Judy, Chief of Staff for Speaker Batchelder, was also deeply involved in the map drawing. 19

DiRossi secured a room at the DoubleTree Hotel in Columbus beginning in July 2011 to serve as the base for the map-drawing operations. 20 DiRossi had the hotel move the usual furnishings out of the hotel room and instead had desks and three computers installed. 21 Various Republican legislators, staff members, and operatives visited the DoubleTree room during the map-drawing process. They included Mann, DiRossi, Judy, Speaker Batchelder, 22 President of the Ohio Senate Tom Niehaus, Representative Matt Huffman, State Senator Keith Faber, Chief of Staff in the Ohio State Senate Mike Schuler, Chief Legal Counsel to the majority in the Ohio House of Representatives Mike Lenzo, 23 map-drawing expert John Morgan, 24 head of Team Boehner Tom Whatman, and legal counsel Mark Braden. No Democratic legislator or staffer ever visited. 25

Mann, DiRossi, and Judy each used a computer equipped with a software package called "Maptitude." 26 Various types of demographic data as well as historical election data and compilations of that data can be uploaded into Maptitude. The software then allows map drawers to draw district lines over a map of a state. Map drawers can view and work on maps in very fine detail-down to the census block unit. 27 As the map drawer draws or alters lines, the program will calculate, recalculate, and display the corresponding demographic and historical election data for the newly drawn districts in real time. 28 Map drawers can save their draft maps both as visual depictions and as data files that contain the assignments of each geographical unit *997 to a particular district. 29 Maptitude will also export into Excel spreadsheets the political data that corresponds to the draft maps.

As mentioned above, much of the data that the map drawers used had been furnished to them through a contract that the Ohio General Assembly entered into with CSU. CSU created and provided the Task Force with the Ohio Common Unified Redistricting Database ("Database" or "OCURD"). 30 The Database included many types of geographic, demographic, and historical partisan election data for the State of Ohio, broken down to the split census block level. 31 The Task Force provided this information to both the Democratic and Republican Caucuses. 32 Mark Braden, who was retained by the Ohio Attorney General to represent and advise the General Assembly during the 2011 redistricting process, 33 hired Clark Bensen from the company Polidata to do some additional work with the data sets to make the data more workable and to provide additional historical election data for the Republican map drawers. 34

Mann, DiRossi, and Judy were tasked by the Republican Caucuses with drawing maps that were favorable to Republicans. Many Republican leaders indicated their preference for a 12-4 map. 35 In order to gauge whether their draft maps would achieve this goal, they used partisan indices, created by compiling the historical partisan voting data from certain chosen elections. The indices were then uploaded into Maptitude so that the map drawers could predict how their draft districts would likely perform politically in future elections.

Various indices were used because individuals involved in the map-drawing process preferred different indices. At times they used an index that they created and termed the "Unified Index." 36 The Unified Index averaged the results of five races, overall reflecting a partisan landscape more favorable to the Democratic Party than an index that would have included a fuller set of elections from the decade preceding the redistricting. 37 The map drawers also used the " '08 McCain Index," which also reflected a strong Democratic *998 performance. 38 The map drawers used Maptitude to create spreadsheets by "output[ting] the numbers to show what various indexes, as well as other data, were for all the districts." 39 They sometimes created comparison spreadsheets to allow them to compare the political index scores of different draft maps to one another. Individuals involved in the map-drawing process also used the Partisan Voter Index ("PVI"), which is used in the well-known Cook Political Report. PVI scores classify districts as either Republican leaning (R+) or Democratic leaning (D+). These classifications are accompanied by a score quantifying the strength of such a leaning.

Individuals not involved in the day-to-day map drawing were sometimes shown the draft districts' predicted partisan proclivities as assessed with various indices. 40 The map drawers would also print out spreadsheets that contained the draft districts' predicted partisan leanings using various indices and share them with Republican Party leaders at redistricting meetings. 41 Judy regularly checked in on DiRossi and Mann as they worked, received updates, reviewed draft maps, and relayed information between Batchelder, DiRossi, and Mann. 42 DiRossi and Mann regularly reported developments to and received feedback from Speaker Batchelder and President Niehaus. They also kept Senator Faber and Republican Chief of Staff in the Ohio State Senate Matt Schuler informed as changes were made.

3. National Republican involvement

National Republican operatives supported the State-level map drawers in their work from beginning to end. This collaboration started prior to the map drawing itself, when Ohio Republican staffers such as DiRossi, Mann, Judy, Schuler, and Chief Legal Counsel for the Ohio House Republican Caucus Michael Lenzo, as well as Representative Huffman attended a redistricting conference hosted by the National Conference of State Legislatures ("NCSL") in Washington, D.C. 43 Lenzo had also attended a Redistricting and Election Law Seminar hosted by the Republican National Committee ("RNC") in Washington, D.C., in Spring 2010. At these meetings, the Ohio Republican staffers made contact with national Republican operatives such as Mark Braden, Tom Hofeller, and John Morgan, who later advised them and collaborated with them during the map-drawing process.

At the Spring 2010 seminar, Morgan gave a presentation on map drawing, advising map drawers to keep the process secret and to score the maps to determine the likely partisan outcome. 44 In 2011, Morgan conducted a follow-up visit to Ohio, where he presented on map-drawing *999 tactics to DiRossi, Mann, and Judy. 45 Speaker Batchelder and President Niehaus also attended a redistricting meeting in Washington, D.C. in the spring of 2011 with Whatman and Republican members of the U.S. congressional delegation. 46

At the time of the census and redistricting, Congressman John Boehner of Ohio was the Speaker of the United States House of Representatives. Ohio Republicans understood that Speaker Boehner would have considerable input in the 2012 map and were committed to enacting a map that he supported. 47 Batchelder spoke with Boehner about once each month during the creation of the 2012 map and met with Boehner twice. 48 Boehner employed Tom Whatman as the head of his "Team Boehner." Boehner tasked Whatman with liaising between Republican members of the congressional delegation and the Ohio map drawers; 49 Whatman began working on the redistricting process at the federal level in December 2010 or January 2011. 50

Whatman employed Adam Kincaid, the Redistricting Coordinator of the National Republican Congressional Committee ("NRCC"), to assist in the redistricting efforts. Kincaid drafted proposed maps and district lines that incorporated Whatman's requests and sent them to DiRossi and Mann and, on occasion, Braden. 51 Kincaid also met repeatedly with members of Ohio's congressional delegation throughout the redistricting process to hear their concerns and keep them abreast of developments. 52 As the districts were drawn, Kincaid updated Whatman and the Republican congressmen about the political leanings of their new districts based on the historical election data, producing spreadsheets with partisan index information for the various draft districts. 53 In the final days of the drafting, state and national Republicans tweaked the map, mindful of the partisan consequences of very minor tweaks. 54 In *1000 some cases, it was clear that national Republican operatives had the authority to "sign off" on changes before they were implemented by the State-level team. 55

4. Major features of H.B. 319

Because of the stagnation in Ohio's population compared to other states, two districts had to be eliminated. This meant that if all incumbents were to run for office, at least two sets of incumbents would have to be paired. The Republicans decided to pair two Republican representatives and two Democratic representatives. 56 Whatman made the decision to pair Republican Congressmen Turner and Austria; Speaker Boehner approved the pairing. 57 Whatman also spoke to both Austria and Turner about the decision. 58 Speaker Batchelder was not involved in the decision to pair those two Republican congressmen. 59

As for the Democratic pairing, the map drawers paired Representative Marcy Kaptur of former District 9 and Representative Dennis Kucinich of former District 10; Kaptur won the Democratic primary that ensued. Kaptur testified that she did not want to be paired with Kucinich, 60 but she was not consulted by the Republican *1001 map drawers on the matter. 61 She saw the map embodied in H.B. 319 for the first time in media reports around the time of the bill's introduction. Kaptur was "astonish[ed]," upset, and offended by the map, which she understood to break up communities of interest and involve unnatural groupings of communities with diverging interests. 62

The map drawers also paired Republican Representative Jim Renacci of the former District 16 and Democratic Representative Betty Sutton of the former District 13 to run against each other in the new District 16. DiRossi testified that the third pairing was necessitated by: drawing District 11 to include portions of Akron, population loss in Northeast Ohio, "two congresspeople who were living very close together," and the creation of the new District 3 in Franklin County. 63

The map drawers drew District 11 to include some portions of the City of Cleveland in Cuyahoga County and a thin strip dropping southward into Summit County where it incorporated sections of the City of Akron. Representative Marcia Fudge, who had represented District 11 under the previous map prior to the 2011 redistricting, was not consulted by Republican map drawers and did not learn of District 11's new boundaries until around the time that H.B. 319 was introduced in the legislature. 64 She was displeased with the new shape of the district, particularly the extension of the district into Summit County and Akron, areas with which she was not familiar and that she had not previously represented. 65 District 11 had historically been a majority-minority district that elected African-American congressional representatives by large margins. Some map drawers expressed that it "was a consideration for us in a proposed map to make sure it remained a majority-minority district." 66

*1002 The map drawers created a new district, District 3, in Franklin County, where the City of Columbus is located. Columbus had been experiencing population growth while metropolitan areas in northern Ohio had been losing population. 67 It is an urban center that is the home of The Ohio State University, and it contains many Democratic voters. Whatman and Kincaid had the idea to create the new District 3 in Columbus that would concentrate many of Columbus's Democratic voters into one district. 68 One spreadsheet sent among those involved in the map-drawing process referred to the new District 3 as the "Franklin County Sinkhole," but it is unclear who exactly included that term. 69 The draft map creating the new District 3 allowed for safe quantities of Columbus's Democratic voter bloc to be absorbed by the neighboring Districts 12 and 15 such that those districts could maintain or achieve safe Republican majorities. 70

State-level and national Republican operatives emailed back and forth sharing and consulting on plans for this new district. Kincaid created a proposed map that included such a district, which scored as D+15 using his PVI metric, and shared the draft map with DiRossi and Mann. 71 Braden asked Hofeller to consult on one draft of the map created by Kincaid, including the new district. Hofeller approved it after removing from District 15 some territory that Kincaid had allocated to it. Hofeller noted that this " 'downtown' area" was " 'dog meat' voting territory" and "awful" in explaining why it should not be included in the Republican-assigned District *1003 15. 72 Kincaid followed up with minor tweaks of the Columbus area division, but the general contours, as tweaked by Hofeller, remained the same. The 2012 map, which placed downtown Columbus in District 3, uses irregular lines to divide Franklin County and Columbus into three districts-3, 12, and 15. In every election under the 2012 map, the Democratic candidate has won District 3 while Districts 12 and 15 have elected Republican representatives. 73

For a time, the Republicans considered drawing a map that would include "13 'safe' seats" for their party rather than twelve. 74 In order to accomplish this, Franklin County and the City of Columbus would be split into four different districts rather than the three they were split into under the 2012 map. 75 Kincaid developed such a map and calculated the PVI scores of the resulting districts. Although such a map could have secured the election of thirteen Republican representatives, the map drawers believed that the margins of victory would have been tighter, as evidenced by lower R+ PVI scores. 76 The Republicans eventually opted for the map *1004 that promised one less Republican seat, but in which those twelve Republican seats were safer.

The map drawers sometimes rejected specific requests from Republican members of the Ohio General Assembly, instead prioritizing maintaining the partisan balance of the draft map. For example, State Senator Christopher Widener requested that the map keep Clark County whole. 77 DiRossi and the other map drawers rejected Widener's request in part because unifying Clark County would have negative consequences for the partisan scores of District 15-making the Republican seat there less secure. 78

The resulting map featured twelve districts likely to elect a Republican representative (Districts 1, 2, 4, 5, 6, 7, 8, 10, 12, 14, 15, and 16) and four districts likely to elect a Democratic Representative (Districts 3, 9, 11, and 13).

5. Secrecy surrounding the map

The Republican map drawers did not share plans for the map with either the public or Democratic legislators or staffers prior to introducing it in the Ohio House of Representatives. 79 Although the State Senate's and State House's committees on redistricting, chaired by Senator Faber and Representative Huffman, respectively, held five public hearings in different locations across Ohio in July and August of 2011 while the maps were being drafted, their members did not share drafts of the maps or political indices at the hearings. 80 The Republican map drawers shared the map with Representative Armond Budish, the Democratic Minority Leader in the Ohio State House of Representatives, only just immediately before the bill was introduced. 81 The map drawers even declined to share information with other Republican members of the Ohio General Assembly prior to the formal introduction of the bill. For example, State Senator Faber saw the map just shortly before its introduction as a bill. 82

6. Passage of H.B. 319

The Ohio Republicans first introduced a 2012 redistricting map in the form of H.B. 319 on September 13, 2011 in the House State Government and Elections Committee. The Committee referred the bill to the House, and it was debated on the floor of *1005 the House on September 15, 2011. 83 Representative Huffman, the sponsor of the bill, spoke on the House floor about the map-drawing process and the factors that the map drawers had considered in drawing the new district lines. 84 Democratic Minority Leader Budish spoke on the floor of the House, criticizing the secrecy of the map-drawing process and the Republicans' failure to take outside input into account. 85 House Democrats also complained that the bill was being rushed through the General Assembly and that the accelerated timeframe for its passage prevented serious scrutiny and critique. 86 The bill passed in the House of Representatives that same day by a vote of fifty-six to thirty-six. 87

On September 19, 2011, H.B. 319 was introduced in the Ohio State Senate. The Senate Committee on Government Oversight and Reform, chaired by Senator Faber, then held hearings on the bill. 88 The Committee amended the bill to include a $ 2.75 million appropriation for local boards of elections in an attempt to make the bill immediately effective and shield it from a voter referendum. 89 The Committee referred the amended bill to the Ohio Senate. 90 On the floor of the Senate, some Democratic State Senators, including Senator Nina Turner, a member of the Black Caucus, opposed the bill and argued that it "lays out 12 Republican districts and four Democratic districts." 91 The bill passed in the Senate by a vote of twenty-four to seven on the same day it was referred. The amended H.B. 319 then returned to the House of Representatives where it passed by a vote of sixty to thirty-five. 92 It was signed into law on September 26, 2011, by Republican Governor John Kasich.

7. Referendum and negotiations

Despite the appropriation amendment intended to insulate the map from a voter referendum, Ohio voters sought to mount such a referendum. A group of Ohio voters filed a petition for a writ of mandamus with the Supreme Court of Ohio. They sought an order declaring that H.B. 319 could indeed be subjected to a voter referendum. State ex rel. Ohioans for Fair Dists. v. Husted , 130 Ohio St.3d 240 , 957 N.E.2d 277 (2011). The Ohio Supreme Court granted the writ of mandamus on October 14, 2011; voters could seek a referendum and the bill could not immediately go into effect. Id. In order to put the referendum on the ballot, Ohio voters would have to gather the signatures of 6% of state electors in slightly over two *1006 months. 93

This also meant that H.B. 319 would not take effect until December 25, 2011, after the December 7, 2011 candidate filing deadline set for the March 2012 primaries. 94 In response, Republican legislators passed H.B. 318, which split the Ohio primaries. The local, state, and U.S. Senate primaries would still occur in March 2012, but the U.S. presidential and U.S. House of Representatives primaries were pushed back to June 2012. 95 This split primary would cost the State of Ohio $ 15 million. 96

In the shadow of the possible referendum and split primaries, Ohio Republican and Democratic legislators attempted to negotiate some alterations to H.B. 319 that could be enacted as a new bill-H.B. 369. 97 This openness to feedback from the Democrats had not been present in the drawing of H.B. 319. 98 Some Republican map drawers testified that Bob Bennett, the chairman of the Ohio Republican Party and a member of the RNC, 99 served as a go-between for the Republicans and Democrats during this period, communicating Democratic requests to the Republican map drawers. 100 The Republicans, although making small concessions and alterations to their original map to cater to Democratic desires, 101 refused to make changes that would alter the likely partisan outcome of the map. 102 Speaker Batchelder commented that the Democratic legislators' "theory was somehow or another that they could overcome a majority of people who were in the other party, and I don't know how that would have happened." 103

*1007 DiRossi, Mann, and Judy worked with Maptitude at their office at the Ohio House of Representatives to draw minor changes into the redistricting map in the period between the passage of H.B. 319 and H.B. 369. 104 For example, DiRossi testified that he made changes based on his belief that Representative Kaptur and others had requested that additional territory in Lucas County and Toledo be added and territory in Cleveland be removed from District 9 so that Kaptur would have a better chance of defeating Kucinich. 105 The changes also included the unification of Clark County. 106

On November 3, 2011 Representative Huffman introduced the new Republican redistricting bill, H.B. 369, in the House Rules and Reference Committee; he gave sponsor testimony in the committee on November 9. H.B. 369 would eliminate the newly split primary. 107 Republican State Representative Lou Blessing sought to push H.B. 369 through the General Assembly by suspending the normal rules mandating that bills be considered by each legislative house on three separate days. 108 Representative Blessing did not have sufficient votes to achieve this result. 109 Around this time it became clear that the Ohio voter referendum challenging H.B. 319 would not be successful; the required votes would not be collected in time. This meant that Democrats had a weaker bargaining position in their efforts to convince Republicans to make further changes to H.B. 369.

8. Passage of H.B. 369

On December 14, 2011, both the Ohio House of Representatives and the Ohio Senate passed an amended version of H.B. 369, over vigorous opposition from some Democrats. 110 The bill passed in the House by a margin of seventy-seven to seventeen (including twenty-one Democratic votes in favor) and in the Senate by a margin of twenty-seven to six (including four Democratic votes in favor). 111 Not only was the amended H.B. 369 nearly identical in terms of partisan leanings to H.B. 369 as it was first introduced, 112 but it was also highly similar to H.B. 319, the first redistricting *1008 plan that the General Assembly had passed. 113 It was signed into law by Governor Kasich the following day. Because the partisan metrics of the map did not change, the new congressional districting map passed as H.B. 369 was just as likely as H.B. 319 to result in the election of twelve Republican representatives and four Democratic representatives.

Following the passage of H.B. 369, Kincaid created a spreadsheet that documented his analysis of the partisan outcomes of the newly enacted map. 114 The spreadsheet featured four D+ districts, with their numerical scores ranging from D+12 to D+29. It also featured twelve R+ districts, with all but one of their numerical scores ranging from R+2 to R+9, and the outlier measuring at R+14. 115 Kincaid prepared a presentation in which he showed how the redistricting efforts had shored up Republican support in three previously competitive districts-Districts 1, 12, and 15, rendering them safe for Republican Representatives Chabot, Tiberi, and Stivers, thereby taking them "out of play." 116 By Kincaid's calculations, District 1 had moved seven PVI points in favor of Republicans by including Warren County and removing portions of Democratic Hamilton County. District 12 had moved nine PVI points in favor of Republicans because portions of Democratic Columbus had been removed from the district and into District 3. Similarly, District 15 had moved seven PVI points in favor of Republicans, as the new District 3 now also contained many of District 15's former Democratic constituents. Kincaid's presentation also noted that Districts 6 and 16 were "Competitive R Seats Improved" because their PVI scores had become more pronouncedly pro-Republican as a result of the redistricting, District 6 by three points and District 16 by one point. 117 Kincaid continued to praise the results of his map-drawing collaboration with the Ohio Republicans, representing that the "new [Ohio] map should be a 12-4 map," that it "eliminat[ed] Ms. Sutton's seat," and that it "created a new Democrat seat in Franklin County." 118 He stated elsewhere that the Ohio "Republican map shored up multiple seats for the decade." 119

U.S. Representative Stivers's communications with his staff reflected his similar *1009 belief that various previously competitive districts had been made solidly Republican as a result of the redistricting. For example, he stated that "[t]he redistricting in Ohio did shore up some of the toss-up districts" based on the changes in the PVI scores for Districts 1, 6, and 15. 120 He acknowledged that U.S. Representative Chabot of District 1 "probably won't have a close race for the next decade" based on the changes the redistricting wrought on that district's PVI score and the fact that his district contained many more Republican voters following the redistricting. 121

9. Congressional elections under the 2012 Map

As predicted by Kincaid, the same four Ohio congressional districts (Districts 3, 9, 11, and 13) have elected Democratic representatives, and the same twelve districts (Districts 1, 2, 4, 5, 6, 7, 8, 10, 12, 14, 15, and 16) have elected Republican representatives in every election since the enactment of the 2012 map.

B. Procedural History

Plaintiffs include seventeen individual Ohio residents, who collectively reside and vote in each of Ohio's sixteen congressional districts, and five organizations based in Ohio. The individual Plaintiffs are: Linda Goldenhar, Douglas Burks, Sarah Inskeep, Cynthia Libster, Kathryn Deitsch, LuAnn Boothe, Mark John Griffiths, Lawrence Nadler, Chitra Walker, Tristan Rader, Ria Megnin, Andrew Harris, Aaron Dagres, Elizabeth Myer, Beth Hutton, Teresa Thobaben, and Constance Rubin. The organizational Plaintiffs, which include nonpartisan groups as well as groups affiliated with the Democratic Party, are: the Ohio A. Philip Randolph Institute ("APRI"), the League of Women Voters of Ohio ("The League"), The Ohio State University College Democrats ("OSU College Democrats"), the Northeast Ohio Young Black Democrats ("NEOYBD"), and the Hamilton County Young Democrats ("HCYD").

Defendants are State Representative Larry Householder, Speaker of the Ohio House of Representatives; State Senator Larry Obhof, President of the Ohio State Senate; and Ohio's Secretary of State, Frank LaRose. All Defendants are sued in their official capacities.

Plaintiffs filed this lawsuit on May 23, 2018. Dkt. 1 (First Compl.). This three-judge panel was then convened pursuant to 28 U.S.C. § 2284 . See Dkt. 28. Plaintiffs twice amended their complaint and, as relevant here, filed their second amended complaint on July 11, 2018, seeking declaratory and injunctive relief and the enactment of a new congressional districting plan. See Dkt. 37 (Second Am. Compl. at 50-52). On August 15, 2018, we denied Defendants' motion to dismiss. See Ohio A. Philip Randolph Inst. v. Smith , 335 F.Supp.3d 988 (S.D. Ohio 2018). After that, we granted the Intervenors' motion to intervene, and they joined the litigation. See Dkt. 64. 122

The case then proceeded through discovery, and on January 8, 2019, Defendants moved for summary judgment. See Dkt.

*1010 136 (Mot. for Summ. J.); Dkt. 140, 140-1 (Intervenors' Suppl. Mot. for Summ. J. & Mem.). After a round of briefing, we denied the motion for summary judgment. See Ohio A. Philip Randolph Inst. v. Householder , 367 F.Supp.3d 697 , 2019 WL 652980 (S.D. Ohio Feb. 15, 2019). 123 Trial commenced on March 4, 2019 and lasted eight days, concluding on March 13. 124

Since the trial, the parties have filed post-trial briefs with proposed conclusions of law, and separately, proposed findings of fact. The parties have also finalized their objections to the other side's evidence, responded to each other's objections, and submitted additional briefs on those objections. 125 This briefing schedule concluded on April 7, 2019.

II. SUMMARY OF THE EVIDENCE PRESENTED AT TRIAL

A. Plaintiffs' Fact Witnesses

1. Individual Plaintiffs

Individual Plaintiffs Douglas Burks, Mark Griffiths, Aaron Dagres, and Elizabeth Myer testified at trial. They live in District 2, District 7, District 12, and District 13, respectively. The remainder of the individual Plaintiffs, who reside in the rest of the congressional districts, testified via deposition. All individual Plaintiffs testified to their affiliation with the Democratic Party and/or that they consistently vote for Democratic candidates. See infra Sections III.A.1.-16. In addition to being Democratic voters, the individual Plaintiffs are politically active in supporting, volunteering for, and working for Democratic candidates and causes. 126 Collectively, they have engaged in a variety of activities, including door-to-door canvassing, calling other voters to support candidates, writing campaign postcards, fundraising for and donating to candidates, writing letters to representatives and opinion pieces, and protesting. Several of the Plaintiffs have also worked on Democratic campaigns and served on boards of groups or political committees affiliated with the Democratic Party. Finally, the individual Plaintiffs testified, based on their direct lay experiences *1011 of engaging in political activity, to the burdens that they themselves have experienced in translating their Party's political efforts in the electorate into political power in the U.S. House of Representatives. 127 The individual Plaintiffs testified that their efforts included candidate recruitment, fundraising, and get-out-the-vote activities.

2. Organizational Plaintiffs

APRI, the League, and HCYD each testified at trial through a representative, and some additional members of the organizations supplemented the testimony. Several themes ran throughout this testimony. First, the organizations actively engage in politics by encouraging citizens to vote, registering and educating voters, and in the case of HCYD, advocating on behalf of Democratic candidates. Second, in their experience, voter outreach and engagement work was made more difficult by continuously encountering significant voter apathy. They heard voters state their beliefs that their votes did not matter; voters believed that the outcome of any given election was preordained and that the same Republican or Democrat would be elected regardless of whether they voted. Third, the organizational plaintiffs encountered voter confusion-voters did not know to which district they belonged, who represented them, or who was running for office in their districts. Fourth, the organizational plaintiffs testified that they were forced to divert resources from their other work to address this voter apathy and confusion. Individual members of the organizations testified about their involvement with their organizations and their own political work supporting the elections of Democratic candidates. They testified that in their experience, they found their Republican congressional representatives unresponsive to them and not engaged in their communities. They also explained how their communities had been split into different districts under the 2012 map.

Andre Washington, the president of APRI, testified at trial on the organization's behalf. 128 Washington is a Democrat who votes regularly and resides in District 12. 129 Under Plaintiffs' Proposed Remedial Plan, Washington would reside in the reconfigured District 12. 130 APRI is a nonpartisan organization but supports civil rights and labor issues. 131 Its activities center around voter education, registration, and outreach. 132 APRI has eight chapters across Ohio, seven of which are currently active, and has between 150 and 200 members spread throughout nearly every congressional district in Ohio. 133 It is a volunteer-run *1012 organization, funded by membership dues. 134

Washington testified that he has personally witnessed voter apathy-people feeling like their vote does not matter-while attempting to engage voters in his own district. 135 He testified that because of the way the lines are drawn, voters do not know where to vote or who is running in their district. 136 Washington testified that APRI must deploy some of its limited resources to combat voter apathy and confusion rather than spending these resources on its other work. 137

Stephanie White, the vice president of APRI's Toledo chapter, also testified at trial. 138 White is a Democrat who votes regularly and resides in District 5. 139 White believes that District 5 "is not part of the Lucas County community," but rather that "it's part of the Fulton County, Defiance, Williams County area, which is predominantly Republican." 140 She is represented by Republican Congressman Bob Latta. 141 White testified that she has spent time in her political work with ARPI addressing Toledo voters' confusion about their assigned congressional districts. 142 She also conducts partisan political activities such as door-to-door canvassing, phone banking, voter registration drives, and get-out-the-vote ("GOTV") work to help elect Democratic candidates such as James Neu and John Galbraith, who ran for Congress against Representative Latta in the 2016 and 2018 elections, respectively. 143

Jennifer Miller, the Executive Director of the League testified at trial on the organization's behalf. 144 The League is a nonpartisan organization that hosts candidate forums, publishes voter education materials, registers voters, and participates in GOTV activities. 145 It has around 2,800 members across Ohio, living in all of Ohio's congressional districts. 146 The League has a long history of attempting to reform the districting process and Ohio's district lines. 147 For example, it commissioned and published a report criticizing the process through which the 2012 map was drawn, and in 2011 it hosted a competition in which members of the public could submit redistricting map drafts that comported with non-partisan traditional redistricting principles. 148

Miller testified that the League spends resources combating voter apathy and confusion due to the 2012 map that it then cannot spend on its other initiatives such as voter registration and education. 149 For example, during the 2018 special election in District 12, the League had to divert significant resources to fielding voters' calls inquiring about their assigned congressional districts. Miller has also observed political candidates' unresponsiveness *1013 to the League's attempts to plan candidate forums, particularly in Republican-dominated areas. She testified that Congressmen Jordan, Stivers, and Joyce have all been unresponsive to the League's requests that they participate in candidate forums. 150 The League cannot hold a candidate forum in which only one party is represented, and therefore must cancel the planned forums if the candidate from one party declines to participate. 151

John Fitzpatrick, a member of the League and a voter in District 14 also testified at trial. 152 Fitzpatrick lives in Stow, Ohio, which is a northern suburb located about ten minutes from downtown Akron. 153 Under Plaintiffs' Proposed Remedial Plan, Fitzpatrick would live in the new District 16. 154 He is a Democrat who votes regularly, has informal conversations with friends to encourage them to vote and vote for particular candidates, has contributed financially to Democratic candidate Betsy Rader's congressional campaign, and has canvassed and phone banked in other elections. 155 Fitzpatrick is currently represented by Republican Congressman David Joyce. 156 Fitzpatrick considers himself a part of the Akron community because he and his wife spend most of their time, recreate, and are involved in the community there. 157 He has been involved in League activities such as planning candidate nights, voter education, and anti-gerrymandering activities such as working to get Ballot Initiative 1 on the Ohio ballot. 158 Fitzpatrick stated that in the year and a half prior to the passage of Initiative 1, 80% of his work with the League was dedicated to anti-gerrymandering work. 159

Fitzpatrick also testified about voters in the Akron area being confused about the district in which they live. He himself attempted to use a "congressional house finder" tool to determine his congressional district, but typing in his zip code produced two possible districts. 160 He stated that because Summit County encompasses four different congressional districts, "before [he] got super-involved in [his] district, there [were] more than a few times when [he] had to look it up because [he] had a hard time just remembering exactly which district [he] was in." 161

Nathaniel Simon, the outgoing president of the HCYD, testified on the organization's behalf. 162 Simon lives and votes in District 2 and is represented by Republican Congressman Brad Wenstrup. 163 Under Plaintiffs' Proposed Remedial Plan, Simon would live in the new District 1. 164 HCYD is a volunteer organization that educates and registers voters and supports Democratic candidates by canvassing and *1014 conducting GOTV efforts on their behalf. 165 HCYD has between 100 and 150 members who vote, identify as Democrats, and live in Districts 1 and 2. 166 Simon testified that HCYD has to expend additional resources fighting voter apathy and confusion. 167 He testified that he felt voters were apathetic because, while canvassing for Democratic candidates Aftab Pureval and Jill Schiller, he encountered voters who "refuse[d] to engage in politics because they felt like there was no point, just being that a Republican is always going to win with the way the lines are drawn." 168 Simon testified that the voter confusion in Hamilton County was due in large part to the current map, in particular the manner in which Districts 1 and 2 "wrap[ ] around each other" and the splitting of the City of Cincinnati itself into two districts. 169 For example, Simon testified that he worked at a polling place in Silverton and that:

many people who came out of the polling booth asked why wasn't Aftab Pureval on my ballot ... I had to explain to them that they are in the 2nd Congressional District, but to the east and west of Silverton is the 1st Congressional District. Also, in my neighborhood, which is in the 2nd Congressional District, there were Aftab Pureval signs, and he is the candidate for the 1st district. 170

Simon also testified that the district lines have made it more difficult for HCYD to attract and retain members. 171

NEOYBD and OSU College Democrats' testimony was introduced through designated depositions. NEOYBD is a Democratic group that "looks to mentor, empower and recruit the next generation of young people of color who want to be involved in the political process." 172 It has around sixty Democratic members who vote regularly and live in Districts 9, 11, 13, and 14. 173 Gabrielle Jackson, the president of the organization, was its Rule 30(b)(6) representative. 174 The organization canvasses, runs phone banks, educates people on "why [their] vote matters, why [they] should be voting," and "concrete issues that are on the ballot," and advocates on behalf of the candidates that the organization supports. 175 Jackson testified that her group fundraises both for candidates and for itself. 176 She stated that "it's been challenging based on the way this map is currently drawn, because folks have been feeling like, you know, [their] voices aren't being heard. So it's causing us to use more of our resources, when we have a hard time bringing in resources." 177 Jackson testified that while canvassing and phone-banking with her organization, she spoke with people who expressed apathy about voting and said that they did not believe that their votes mattered. 178

Alexis Oberdorf is the President of the OSU College Democrats and was the *1015 group's Rule 30(b)(6) representative. 179 The OSU College Democrats "advocate, educate, and engage people at OSU in alignment with the Democratic Party's platform." 180 The organization has around 55 members who regularly attend meetings but hosts events throughout the year that around 100 people attend. 181 OSU College Democrats canvasses and runs phone banks in support of Democratic candidates and has held fundraisers for Democratic candidates such as Danny O'Connor. 182 Oberdorf testified that OSU students who live near campus reside in Districts 3, 12, and 15 and that the organization must therefore "spread[ ] [its] capital among three different areas on campus." 183 The majority of OSU College Democrats vote "on campus in their district." 184 She testified that she worked a poll in District 12 during an election and witnessed students coming to vote in the incorrect district "because they assumed seeing that they're ... in this campus area, they are all going to vote in the same area. So that creates confusion. And part of what we do as a club is aim to educate people." 185 She also testified that her organization has "done coordinated call campaigns for bills that [it] oppose[s]" to representatives from those districts and has found "it challenging especially to contact or get ... a response from those individuals." 186

3. Congresswoman Marcia Fudge

Congresswoman Marcia Fudge, representative to the United States House of Representatives from Ohio's Congressional District 11, testified for Plaintiffs at trial. 187 She testified that District 11 has been represented by three different representatives in Congress: Lou Stokes, Stephanie Tubbs Jones, and herself. 188

Congresswoman Fudge described the historical contours of District 11. When Congresswoman Fudge took office in 2008, District 11 "was primarily a little better than two-thirds of the city of Cleveland and most of the southeast suburbs." 189 The district was entirely contained within Cuyahoga County. 190 When Stephanie Tubbs Jones took office in 1999, District 11 included "most of the city of Cleveland, the lower west side all the way to the east and the southeast suburbs of Cuyahoga County," and was again entirely within Cuyahoga County. 191 The district that Congressman Stokes represented was "pretty much the same," again, entirely within Cuyahoga County. 192 Congresswoman Fudge contrasted that historical District 11 with the version of District 11 that she currently represents: "[T]he first major difference is that [her district] go[es] from Cuyahoga down to Summit County" via a "narrow *1016 strip." 193

Congresswoman Fudge unequivocally stated that she "didn't have any role" in the drawing of the new congressional map in 2011. 194 She first learned that the new District 11 would extend into Summit County and include parts of Akron "around the time that the map was made public." 195 Armond Budish, the Democratic minority leader of the Ohio House of Representatives, was the one to first show her the map "pretty much so [she] wouldn't get caught off guard." 196 She stated that she was "surprise[d], obviously" by the new District 11 and had "no idea that [she] would ever go down into Summit County." 197 She was not "pleased" by the new design, she "would not have chosen it," and she "was not happy about it." 198 Congresswoman Fudge stated that she "didn't know anything about Summit County" at the time and that her lack of familiarity with the new area made it "an uncomfortable place to be." 199 She stated that due to Ohio's losing two congressional seats and the inevitable changes that that would necessitate, she thought that the new District 11 would most likely include the entire City of Cleveland and its southeast suburbs. 200

Congresswoman Fudge stated that after learning of the new map, the only complaint that she voiced was her belief that allocating "Summit County or that portion of Akron" to the new District 11 "would make it almost impossible" for Democratic Representative Sutton to win an election in the new District 16. 201 Congresswoman Fudge stated that she got together with Congresswoman Sutton and Congresswoman Kaptur to contact Armond Budish to "ask him was there any way to give Betty back Akron so she would have a fighting chance at keeping her seat." 202 She testified that she "may have" spoken with U.S. House of Representatives Speaker Boehner in 2011 about the redistricting "in passing" but recalls nothing about such a conversation. 203 She spoke to "[l]ots of people" about the shape of her district in 2011, including Republican Congressman Steve LaTourette, who she believed was "kind of the point person for John Boehner." 204 She also spoke to Representatives Sutton and Kucinich, first attempting "to see if we could get [the shape of the district] changed because we wanted to try to see if we could help protect Betty [Sutton]. We couldn't." 205 She then "made sure they knew [she] was not pleased." 206

Congresswoman Fudge admitted that she did not tell any of the people that she *1017 spoke with in 2011 about District 11 that she did not want District 11 to be a majority-minority district. 207 She did not advocate the drawing of District 11 with less than 50% BVAP ("Black Voting Age Population"). 208 She testified that in 2011 she did not view the new district as a violation of the Voting Rights Act ("VRA"). 209 Congresswoman Fudge stated that she was not concerned about being paired with another incumbent in the redistricting because she "felt if they were to pair me with somebody, I felt that I was strong enough to win." 210 She expressed no concern to anyone about being paired with Congressman Kucinich. 211 On cross-examination, Congresswoman Fudge stated that since Stokes's time as the congressman for the district, it has been a majority-minority district. 212

4. State Senator Nina Turner

State Senator Nina Turner, a former Democratic member of the Ohio State Senate, testified for Plaintiffs as a fact witness. Senator Turner served Ohio's 25th State Senate District from 2008 to 2014. At the time of the 2011 redistricting, Senator Turner testified that the State Senate was comprised of ten Democratic Senators, five of whom were African American, and twenty-three Republican Senators. 213 As a result of being in the "deep minority," Senator Turner testified that she had no involvement in the drawing of the current map and that the Democratic Caucus as a whole "didn't have the power to draw the map" because "Republicans could hold business on the [Senate] floor without really having Democrats there." 214 When she first learned of the map presented in H.B. 319, Senator Turner testified that she was "outraged" and that her Caucus tried to "introduce a map that was a fairer reflection of the will of the people." 215 As to H.B. 319, Senator Turner stated that only two Democratic State Senators voted for the bill and that she voted no. 216 Senator Turner believed that the map presented in H.B. 319 would be a 12-4 map. 217

Senator Turner also gave a floor speech against H.B. 319, in part addressing the justification that the District 11 was drawn to comply with the VRA. 218 At trial, Senator Turner explained her belief that the way District 11 was drawn harmed the voters the VRA sought to protect by "hurt[ing] the[ir] voting prowess" and decreasing their "influence that they would have through representative democracy by stripping or combining portions of the 11th *1018 Congressional District in ways that representatives could not focus purely on Cleveland and/or Cuyahoga County." 219 Senator Turner also noted that Congresswoman Marcia Fudge and former Congressman Louis Stokes "never had a problem winning elections in that district." 220 She further testified that the way District 11 was drawn harmed both the greater Cleveland and the greater Akron communities because she believed that the two communities have separate needs and "deserve to have a representation that can really focus in on their needs." 221

As recounted above, after H.B. 319 was enacted into law, Democratic state legislators sought a referendum to overturn the law, which required a certain number of signatures. 222 This referendum failed because not enough signatures were collected, and Republican state legislators then went forward with H.B. 369. 223 Senator Turner testified that she had no input on the map presented in H.B. 369, that she believed that the map was still 12-4 in favor of Republicans like H.B. 319, and that she and a majority of the Democratic Caucus in the State Senate (as well as a majority of the African-American State Senators) voted against H.B. 369. 224

Senator Turner spoke against H.B. 369 in a floor speech similar to the one she made against H.B. 319. In this floor speech, Senator Turner stated that "[t]o say that this map is bipartisan is laughable" because, as she stated at trial, she believed that "the mere fact that some Democrats, for whatever reason, decided to vote for the bill does not make it bipartisan." 225 At bottom, Senator Turner maintained her belief that H.B. 369 had a clear partisan effect. 226

Finally, on cross-examination, Senator Turner admitted that she considered running against Congresswoman Fudge in the 2012 Democratic primary, but she dropped out because she believed that the redistricting process was manipulated to guarantee the reelection of incumbent politicians. 227 Senator Turner also acknowledged that it "might be possible" that she received proposals from Democratic map drawers that incorporated, among other things, a majority African-American district in northeast Ohio. 228 But such a district existed previously (with different boundaries, limited to the greater Cleveland area), and Senator Turner maintained that the enacted map did not contain any of the Democratic suggestions. 229

5. Congresswoman Marcy Kaptur

Plaintiffs called Congresswoman Marcy Kaptur, a Democratic member of the U.S. House of Representatives, as a rebuttal witness. Representative Kaptur won election to Congress in 1982 and has served *1019 Ohio's Congressional District 9 since 1983. She is the most senior member of Ohio's congressional delegation. 230 Representative Kaptur testified that she did not play any part in creating the map that was submitted with H.B. 319, the initial redistricting bill, and she first learned about the shape of the new District 9 in the newspaper after H.B. 319 became public. 231 Representative Kaptur testified that, after learning about the map presented in H.B. 319, she called then-Governor John Kasich's office to object to the fact that her church and the cemetery where her family is buried were cut out of District 9; 232 moreover, she had conversations with a Democratic state legislator after the release of H.B. 319 to "try[ ] to piece [Toledo] back together." 233 Representative Kaptur did not want to be paired with then-Congressman Kucinich, a Democratic colleague of Kaptur's, because he had "run for president" and she believed that the proposed District 9 was drawn to favor Representative Kucinich over her if they ran against each other. 234 On cross-examination, Representative Kaptur acknowledged that, due to population loss, her district's geography would have to expand, but she stated that she "hop[ed] it would be in the economic region that [she] represented" such as Wood or Fulton Counties. 235

B. Defendants' Fact Witnesses

1. Raymond DiRossi

Raymond DiRossi testified at trial for Defendants as a fact witness, and he was one of the principal map drawers during the 2011 redistricting process. He also played a role in the 2001 redistricting process. 236 Starting in 2001, DiRossi became involved with the Task Force and "was very involved in the creation of [the] legislative districts and also the congressional districts ...." 237 DiRossi testified that he worked out of the DoubleTree hotel in Columbus during both the 2001 and 2011 redistricting processes. 238

DiRossi testified that, in 2011, he was "very prominent" in the congressional redistricting process and that "basically, the process was the same" as in 2001. 239 According to DiRossi, the main issues in the 2011 redistricting process were that Ohio lost two congressional seats, the State had experienced population shifts, District 11 was majority-minority in the past and in 2011 "great care was ... taken to ... make sure that [District 11] was going to be created in a way that would be satisfactory," and he also understood that there was a "desire to make a new district in Franklin County that would have the ability to elect, for the first time ever," a minority candidate to Congress. 240

To deal with the loss of two incumbents (because Ohio lost two congressional seats), DiRossi testified that "the decision was made to pair two Republicans together and two Democrats together. So we would have ended up with" twelve Republicans *1020 and four Democrats. 241 In terms of how to handle which Democratic incumbents to pair, he stated that it was his belief that "nobody thought it was a good idea to pair" Representative Fudge with another incumbent because she represented a majority-minority district. 242 In the end, Representatives Kaptur and Kucinich were selected as the paired Democratic incumbents. DiRossi testified that he drew the current District 9 the way it is based on what various other Republican legislators and political officials had said various Democrats wanted (these other Republicans were purportedly in conversation with the Democrats). 243

DiRossi further testified to changes made to various other districts, purportedly at the request of (occasionally unspecified) Democrats, and to the effects those changes had on the map as a whole. 244

*1021 Negotiations between state legislative Democrats and state legislative Republicans began around the time of the attempted petition drive (after H.B. 319). 245 As to District 11, for example, DiRossi asserted that he "wanted to take great care to make sure the district was drawn the way that the incumbent [Representative Fudge] wanted it." 246 At trial, DiRossi did not mention any concerns about VRA compliance, but at his deposition, he stated that he was concerned about majority-minority districts, including District 11, because of the VRA. 247 At his deposition, he further stated that, in 2001, District 11 was drawn with more than a 50% BVAP, so in 2011, "one of the first things that [DiRossi] was looking at was ... was it possible to still draw a district that would be more than 50 percent non-Hispanic voting age African American population." 248 It was DiRossi's "understanding that the maps were going to make their way to Congresswoman Fudge," but he clarified that, "obviously, [he] was not present for that." 249

With respect to District 3, DiRossi similarly testified that a "back and forth" occurred between Bob Bennett, Republican legislative leaders, "some other people," and Joyce Beatty and her husband Otto. 250 At that time, now-Congresswoman Beatty was not yet a Congresswoman and did not hold any position in government, though DiRossi testified that "a number of people, including myself who had worked with ... Joyce Beatty .. thought that she would be an ideal candidate" for the new District 3. 251

Some changes did, in fact, occur between H.B. 319 and H.B. 369. DiRossi testified to these changes and explained an exhibit that illustrates them. 252 And again, he asserted at trial that many of these changes were made in response to what he believed were requests of various Democrats. 253 For H.B. 319, he worked out of the DoubleTree Hotel and did not work with any Democrats; he also admitted that he received requests from Tom Whatman (from Team Boehner). 254 For H.B. 369, *1022 DiRossi stated that he worked out of the State House, and, for that bill, he asserted that Republicans "were working with the Democrats ...." 255

As to the logistics of the actual map-drawing process, DiRossi testified to that he used Maptitude and the Unified Index that he created. 256 Along with the Unified Index that he created and additional political indices that others wanted him to use, his computer also displayed the population of each district, the African-American voting-age population, the non-Hispanic voting-age population, and the Hispanic voting-age population as he drew draft maps. 257 "[W]henever [he] would make a change on the ... screen, all of that would automatically change ...." 258 The other political indices included presidential election results, as well as the "D+1, D+2, R+1, R+2 system" (often referred to as the D+1, R+1, or PVI) from "the D.C. folks." 259

DiRossi admitted that in 2011 he worked with Adam Kincaid, from the RNC, and that Kincaid "was one of a number of people that would send ideas or [DiRossi] could bounce ideas off." 260 In a September 10, 2011 email exchange between DiRossi and State Senator Faber, DiRossi wrote, "DC is increasingly pushing to put the lid on this [i.e., the map]." 261 DiRossi also admitted that the changes supposedly requested by now-Congresswoman Beatty (who, again, was not yet a Congresswoman) to draw a potential opponent out of District 3 affected a fairly trivial number of voters. 262 Finally, DiRossi admitted that he did not calculate compactness scores for the districts in either H.B. 319 or H.B. 369. 263

2. Speaker William Batchelder

Former Speaker of the Ohio House of Representatives William Batchelder testified for Defendants at trial, explaining how Districts 11 and 3 came to be. 264

*1023 a. District 11

Speaker Batchelder testified that he knew George Forbes, the former president of the city council of Cleveland "very well" and would occasionally discuss "matters that were coming before the house" with Forbes. 265 Speaker Batchelder stated that District 11 "had changed in its nature, which we knew from the census, and [he and Forbes], therefore, were concerned about its continuance as an African-American district." 266 Therefore, Speaker Batchelder believed "[t]here would have to be a change in the district so that there would be a balance so that it would continue as an African-American district." 267 Speaker Batchelder testified that he had discussions with Forbes about District 11 "extending down into Summit County" because "we ... did not have the makings, under the census, of a district that would be African American" and "there were sufficient African-Americans in Summit County to undertake that alteration." 268 Speaker Batchelder testified that he "asked [Forbes] what he thought of that, and he was amenable." 269 Speaker Batchelder "ultimately approve[d] a District 11 that started in Cuyahoga County and went down into Summit County." 270 He agreed that he did this "in part, based on [his] understanding and belief of how Mr. Forbes felt about that." 271

On cross-examination, Speaker Batchelder admitted that he "never personally had communications with Representative Fudge" about the composition of District 11. 272 Speaker Batchelder also stated that he and Representative Stokes "did communicate, but not on that issue." 273

b. District 3

Speaker Batchelder testified about the creation of the new District 3 in the Columbus area. He stated that he "first had consulted with the chairman of the Republican Party there, and he indicated that there was not going to be a viable candidate for his party." 274 Speaker Batchelder went on to explain that he was close friends with Otto Beatty and had served in the Ohio House of Representatives with *1024 his wife, Joyce Beatty. 275 Speaker Batchelder agreed that he "intend[ed] to draw a district that [Joyce Beatty] could potentially win." 276 Speaker Batchelder stated that he had never referred to the Franklin County district as a "sinkhole" nor had he referred to voters as "dog meat." 277

3. Troy Judy

Troy Judy had a long history of working for the Ohio House of Representatives and served as the Chief of Staff to Speaker of the Ohio House of Representatives William Batchelder during the redistricting process. 278 He testified about the various people who played a role in the redistricting. 279 He also testified about the map-drawing process, both before and after the passage of H.B. 319, and offered reasons that certain congressional districts in the 2012 map were drawn as they are. 280

Judy testified that "[a]fter [H.B.] 319 was passed, the Democrats, of course, announced a referendum on the bill and began collecting signatures.... And with the overarching pressure of a referendum, it led us to begin conversations with members of the Democratic caucus." 281 Speaker Batchelder asked Judy and Representative Huffman "to begin very quiet conversations with the Democrats to see what changes they would like to see in a map in order to garner bipartisan support of a bill, a new bill." 282 Judy testified that in this context he conversed directly with three Democratic members of the Ohio House of Representatives who communicated to him "some of the changes [they] would like to see." 283 Some of these changes were incorporated into new map drafts and Judy and Keary McCarthy, the minority Democratic Chief of Staff exchanged map files including such changes. 284 Judy stated that in the back-and-forth between himself and McCarthy, McCarthy never proposed a District 11 or District 3 "that was materially different from the one proposed by the Republicans." 285 Judy testified that at this stage, the now-deceased Bob Bennett, "the outgoing chairman of the state Republican party," was involved in communications between the Republican map drawers and Democratic players. 286

Judy testified that District 3 had been a "priorit[y]" of Speaker Batchelder's. 287 He testified that Speaker Batchelder's "relationship with Congresswoman Beatty and her husband Otto Beatty led him to have a priority to create a central district in Franklin County encompassing Columbus and having representation specifically for Congressman [sic] Beatty." 288 He also testified that population shifts toward Franklin County and Ohio's loss of *1025 two congressional seats following the 2010 census were factors in the drawing of District 3. 289

Judy testified that District 9 was drawn in response to the Democratic leadership's desire that Representative Marcy Kaptur and Representative Dennis Kucinich be the two Democratic incumbents paired. 290 Judy stated that Bob Bennett "was also in contact with a Democratic leader from the Toledo region, Jim Ruvolo, 291 who then communicated to us about what the shape of the Kaptur district should look like and what Democrats should be paired together, actually." 292 Judy stated that he was "not sure who else [Bennett] was speaking with." 293

Judy also testified about the contours of District 11. He stated that Speaker Batchelder had relationships with members of the African-American community in Cleveland, including George Forbes, and has "consulted" for many years with these individuals "with respect to any issues that would affect the African-American community." 294 This was the only testimony that Judy related regarding the involvement of leaders of Northeast Ohio's African-American community in the redistricting of District 11.

Judy testified that when the Republican map drawers began negotiations with Democratic individuals in an effort to pass the second iteration of the map, Bob Bennett played a key role in these communications, serving as a "back channel to Congresswoman Fudge ... to communicate with us about the shape of [District 11]." 295 Judy testified that Bennett "communicated to [Judy] that he was in contact with Representative Fudge" and that Fudge "was pleased with the configuration [of District 11] that was in 369" after the Republican map drawers had "ma[d]e changes and incorporate[d] things that the Democrats wanted to see." 296

On cross-examination, Judy admitted that despite changes that were made to H.B. 369 prior to its passage, it looked "substantially similar" to the initial version of H.B. 369 introduced by the Republicans members of the General Assembly. 297

C. Plaintiffs' Expert Witnesses

1. Dr. Christopher Warshaw

Dr. Christopher Warshaw testified at trial for Plaintiffs as an expert witness. Dr. Warshaw is a tenure-track assistant professor of political science at the George Washington University, teaching courses on political science, elections, public opinion, statistical methodology, and political representation. 298 His research has been published extensively in prestigious peer-reviewed publications and he has published *1026 specifically on the topic of partisan gerrymandering. 299 Dr. Warshaw has also served as an expert witness in two other partisan-gerrymandering cases; no court has ever failed to credit his testimony. 300 The Court qualified Dr. Warshaw as an expert in the fields of elections, partisan gerrymandering, polarization, and representation and found his testimony highly credible. 301

a. Partisan-bias metrics

Dr. Warshaw testified at length about four 302 specific partisan-bias metrics that he used to evaluate the 2012 map. He defines partisan bias broadly as "the idea of trying to quantify whether one party or another has an advantage in the translation of votes to seats." 303 Successful partisan gerrymanders efficiently translate votes for the favored party into seats for that same party. "In practice, this entails drawing districts in which the supporters of the advantaged party constitute either a slim majority ... or a small minority." 304 Map designers accomplish the former by cracking voters from the opposition party into different districts so that they are highly unlikely to break the 50% mark in a given district and are therefore unable to elect the candidate of their choice. They accomplish the latter by packing voters from the opposition party into districts such that they have an unnecessarily large margin of victory.

The concept of "wasted" votes underlies both of these strategies. 305 In cracked districts, the votes of the losing disfavored party are all wasted because they were allocated to a race that the disfavored party did not win. The closer the margin of victory in cracked districts, the more disfavored party votes are wasted. In packed districts, many votes of the winning disfavored party are wasted because there are many excess votes beyond those needed for victory. A party designing a partisan gerrymander will attempt to waste few of its own supporters' votes and waste many of the opposing party's supporters' votes. Partisan bias, an asymmetry or advantage in the efficiency of vote-seat translation, results.

Dr. Warshaw used the efficiency gap, symmetry in the vote-seat curve, the mean-median difference, and the declination metric to measure partisan bias in the 2012 map. 306

i. Efficiency Gap

The efficiency gap compares the wasted votes for each party by calculating "the difference between the parties' respective wasted votes, divided by the total number *1027 of votes cast in the election." 307 The efficiency gap reflects "the extra seats one party wins over and above what would be expected if neither party were advantaged in the translation of votes to seats (i.e., if they had the same number of wasted votes)." 308

Dr. Warshaw surveyed historical efficiency gaps across the country and found that they were generally quite small. Around 75% were between -10% and 10%, and only around 4% had an efficiency gap of greater than 20% in either direction. 309 He demonstrated that Ohio's 2012 efficiency gap of -22.4% was a historical outlier-"more extreme than 98% of previous plans in states with more than six seats over the past 45 years, and ... more Republican-leaning than 99% of previous congressional redistricting plans." 310 It also reflected a major increase from Ohio's efficiency gap prior to the 2011 redistricting efforts. 311 Ohio's efficiency gaps in 2014 and 2016 were -9% and - 8.7%, respectively, "imply[ing] that Republicans in Ohio won 1-4 more seats in these elections than they would have won if Ohio had no partisan bias in its efficiency gap." 312 Ohio's efficiency gap in the 2018 election was -20%, more extreme than 96% and more pro-Republican than 98% of previous comparable plans. 313

ii. Partisan symmetry in the vote-seat curve

Symmetry in the vote-seat curve compares how both parties' seat shares change as their vote shares increase or decrease. 314 Dr. Warshaw explained that in an unbiased districting scheme, if Democratic candidates receive 52% of the votes and earn 60% of the seats, then when Republican candidates receive 52% of the votes, they should also earn 60% of the seats. One can measure symmetry by applying a counterfactual uniform swing in vote shares from 45% to 55% and measuring departures from parity in seat share between the parties. 315 One applies a uniform swing by increasing the vote share of a given party by a fixed percentage across all districts. 316 Symmetry can also be measured simply by comparing the seat share that each party achieves when it receives 50% of the vote. Applying uniform swings, the level of partisan asymmetry in Ohio's 2012 election was "more extreme than 96% of previous elections and more pro-Republican than 97% of previous U.S. congressional elections over the past 45 years." 317 The result was the same when the symmetry analysis was conducted using the method that compares seat shares when each party earns 50% of the vote. 318 With uniform swings, the 2018 elections were more asymmetric than 92% of previous elections *1028 and more pro-Republican than 94% of the comparison group. 319

iii. Mean-median gap

The mean-median gap reflects "the difference between a party's vote share in the median district and their average vote share across all districts. If the party wins more votes in the median district than in the average district, they have an advantage in the translation of votes to seats." 320 Dr. Warshaw found that Ohio's mean-median gap jumped from 1.7% in 2010 to 7.8% in 2012, following the redistricting. 321 He also found that the 2012 mean-median gap was more extreme than that in 83% of prior elections and more pro-Republican than that in 92% of prior elections. 322 The 2018 mean-median gap was 5%, more extreme than in 62% of previous elections and more pro-Republican than in 81% of previous elections. 323

iv. Declination

Lastly, the declination metric involves graphically plotting the districts in a plan from least Democratic to most Democratic and then measuring and comparing the angles formed by best-fit lines for each party's seats measured from the 50% Democratic vote share line. 324 The calculations result in a score between -1 and 1, which indicates the size and direction of the partisan bias of the map. 325 Ohio's 2012 declination score of -0.77 was "more extreme than 99% of previous elections and more pro-Republican than any previous U.S. congressional election over the past 45 years." 326 Ohio's 2018 declination score of -0.69 "was more extreme than 98% of previous elections and more pro-Republican than 99% of previous U.S. congressional elections." 327

v. Strengths and weaknesses of the metrics

Dr. Warshaw highlighted some of the strengths and weaknesses of each partisan-bias metric. For example, a strength of the efficiency gap is that it "can be calculated directly from observed election returns even when the parties' statewide vote shares are not equal." 328 However, the efficiency gap can also be a more volatile metric than some of the others, and it is not recommended for use in smaller states with relatively few congressional districts. 329 A strength of the symmetry metric is that it is far less volatile over time and has been widely used and accepted in academic work on partisan gerrymandering.

*1029 330 One weakness of both symmetry metrics is that they involve the calculation of counterfactual elections. 331 The mean-median gap is easy to apply, but it is "sensitive to the outcome in the median district." 332 For its part, the declination measure "is somewhat unstable when a party holds a very small number of seats in the legislature." 333 Dr. Warshaw explained that all these metrics are "closely related both theoretically and empirically, but nonetheless, there's small differences between them ... [and] looking at a suite of different metrics in concert gives us greater confidence in any conclusion that we ... draw." 334 Looking across all the metrics, Dr. Warshaw concluded that "Ohio's recent elections [under the 2012 plan] display a larger partisan bias in favor of Republicans than most previous plans in Ohio or in other states." 335

b. Requirements of a partisan gerrymander

Dr. Warshaw testified about how he determines in his academic work whether a redistricting plan is a partisan gerrymander. According to Dr. Warshaw, to qualify as a partisan gerrymander, a districting plan must satisfy four different elements. First, a single party must have controlled the redistricting process-meaning that in a state with a bicameral legislature, it must have had control of both houses and the governorship-and that same party must be favored by the map. 336 Under Dr. Warshaw's criteria, whether members of the disfavored party cast roll-call votes in support of the redistricting plan is meaningless in determining whether the plan was a gerrymander. 337 Second, all partisan-bias metrics that Dr. Warshaw employs (efficiency gap, symmetry in the vote-seat curve, mean-median gap, and declination) must "indicate [that] the same party that controlled the redistricting process was actually advantaged in the translation of votes to seats." 338 Third, the map must be an outlier in terms of its partisan-bias metrics when compared to historical elections across the country in the last forty-five years. 339 Fourth, all four partisan-bias metrics measuring a given map must point in the same direction. 340

*1030 Dr. Warshaw found that under this rubric, the 2012 plan was a partisan gerrymander because: (1) the Republican Party controlled the redistricting process and the map favored the Republican Party; (2) all four of his partisan metrics indicated that the Republicans were actually advantaged in the translation of votes to seats; (3) the map was an outlier when compared to the dataset of hundreds of historical maps; and (4) all four partisan metrics pointed in the same direction-toward a pro-Republican bias.

c. Responsiveness, competitiveness, and durability

Dr. Warshaw also evaluated the responsiveness and competitiveness of the 2012 map. Responsiveness measures "how insulated a plan is from changes in voter preferences" or, conversely, "how likely the election results are to change due to changes in voter preferences." 341 A map is more responsive if it yields different seat shares when there are swings in voter preferences from year to year. Dr. Warshaw measures responsiveness in two ways: (1) determining how many districts with competitive seats exist and (2) applying a uniform swing of vote shares between 45% and 55% across all districts and measuring how the seat-share outcome changes. 342

Dr. Warshaw concluded that Ohio's present map "has led to historically uncompetitive elections." 343 First, in 2012, Ohio had only two competitive congressional seats. 344 In both 2014 and 2016, not a single congressional district in Ohio saw a competitive election. 345 In 2018, Ohio again had only two competitive seats. 346 The uniform swings also demonstrated that the 2012 map is highly unresponsive. 347 Applying uniform swings to the 2012 election results, he found that Democrats would win the same 25% of the congressional seats if they won anywhere from 30% to 52% of the statewide vote. To advance to holding 37.5% of seat-share, they had to win 55% of the statewide vote. 348 Dr. Warshaw determined that 2018 was a more responsive year than earlier years according to the uniform swing analysis. However, "most of this responsiveness occurs at the very upper end of the range of plausible statewide vote shares for democrats"; Republicans would still win "75% of the seats across most of the range of plausible election swings," even if 50% of the vote share was Democratic. 349

Dr. Warshaw also found that the effects of the 2012 map are durable throughout time. 350 Although the partisan-bias metrics generally became somewhat less extreme as time went on, the level of partisan bias *1031 in 2012 under each metric was a "powerful and statistically significant predictor" of the same metric's level in 2016 and 2018. 351

d. Polarization, representation, and trust in representatives

Dr. Warshaw testified about political polarization and its impact on representation. He defined polarization as "the distance between the average preferences of members of the two parties." 352 He concluded that due to increased ideological polarization between Democratic and Republican members of Congress, Ohio Democratic voters who are disadvantaged by the districting scheme and represented by Republican congressmen are unlikely to have their views represented by their representatives in Congress; gerrymandering therefore negatively affects representation. He also found that "voters in gerrymandered states ... trust their representatives less than voters in non-gerrymandered states." 353

e. Proposed Remedial Plan

Dr. Warshaw used the same data to analyze the Proposed Remedial Plan as he did with the 2012 map and found that the Proposed Remedial Plan had far lower levels of partisan bias and higher levels of responsiveness than the 2012 map; it "had no substantial partisan bias." 354

2. Dr. Wendy K. Tam Cho

Dr. Wendy K. Tam Cho testified at trial for Plaintiffs as an expert witness. Dr. Cho is a full professor at the University of Illinois at Urbana-Champaign, and she holds appointments in several departments, including political science, statistics, and mathematics. 355 Dr. Cho is also a Senior Research Scientist at the National Center for Supercomputing Applications at the University of Illinois. 356 She has studied redistricting for thirty years and written extensively on the topic through the lens of multiple academic disciplines. 357 Dr. Cho previously testified as an expert in a partisan-gerrymandering case on behalf of defendants in Pennsylvania who were defending a map enacted by the Republican legislature in the Commonwealth; the court in that case qualified her as an expert. 358 This Court qualified Dr. Cho as an expert in political science, political geography and redistricting, statistics and applied statistics, statistical modeling and sampling from unknown distributions, and the design of algorithms. 359

Dr. Cho testified about her analysis of the current map and its partisan characteristics as compared to a set of simulated maps that she generated. Dr. Cho used an Evolutionary Markov Chain Monte Carlo ("EMCMC") algorithm 360 to run a simulation *1032 on a supercomputer, and the algorithm generated 3,037,645 simulated maps. 361 These maps incorporated only neutral redistricting criteria and no partisan data (she analyzed partisanship after generating the maps). 362 Through this analysis, Dr. Cho was "trying to understand what would be a typical map that would emerge from a non-partisan [map-drawing] process." 363 Specifically, her analysis sought to determine whether neutral factors, primarily political geography, could explain the 12-4 outcome of the current map.

Dr. Cho's simulations can be analogized to a coin toss. For example, if you toss a coin 1,000 times, and the coin lands on heads 582 times, that is one datapoint. If you flip the coin another 1,000 times, and the coin lands on heads 602 times, that is another datapoint. Running through this process many times (e.g., 3 million) provides a fuller picture of the typical outcomes. With a fair coin, outcomes of around 500-heads and 500-tails would be typical; 950-heads or even 1,000-heads out of 1,000 flips are also theoretically possible, but such outcomes would be surprising if the coin tosses were done with a fair coin. In this redistricting context, Dr. Cho generated over 3-million simulated maps and then analyzed the seat share between the parties under each. This process allowed her to compare how typical a 12-4 seat share between Republicans and Democrats would be under a neutral map-drawing process and, thus, to analyze whether it is likely that the 12-4 seat share can be explained by factors such as Ohio's natural political geography. 364 In short, Dr. Cho's simulated maps are meant to provide a nonpartisan baseline against which to compare the current map.

Dr. Cho's methodology includes several key and related components. 365 Dr. *1033 Cho's EMCMC algorithm, which she used to generate the simulated maps, is grounded in the Markov Chain Monte Carlo ("MCMC") theorem. 366 MCMC algorithms are a commonly used technique for sampling. 367 In the redistricting context, a Markov Chain randomly walks from one simulated map to another, different simulated map. 368 In Dr. Cho's EMCMC, each movement of the Markov Chain is guided by optimization heuristics, which improve the Markov Chain's "efficiency and effectiveness in the traversal of the search space." 369 The MCMC theorem, meanwhile, ensures a representative sample of the massive universe of possible maps. 370 Lastly, Dr. Cho ran the algorithm on the University of Illinois's Blue Waters supercomputer, which enabled the algorithm to output the sample of over 3-million simulated maps relatively quickly. 371 All these *1034 components worked together to allow for the drawing of "a random and large sample of feasible electoral maps," out of the much larger universe of feasible alternative maps. 372

Dr. Cho built in several constraints when she produced her simulated maps, and those constraints are what define a map as "feasible" in her simulation. Dr. Cho testified that she arrived at the constraining criteria by "look[ing] at the legislative record to see what the legislature was applying." 373 Primarily, Dr. Cho looked at State Representative Huffman's statements in support of H.B. 319. 374 Representative Huffman explained that the map considered compliance with the VRA, equal population, and "several other traditional redistricting principles": "compactness, contiguity, preservation of political subdivisions, preservation of communities of interest, preservation of cores of prior districts, and protection of incumbents." 375 In regards to incumbent protection, Representative Huffman described that criterion as "a subservient one to the other ones that [he] listed" 376 and further explained that, "[n]obody has a district.... There's nobody that owns a piece of land in Congress. People elect them." 377 From this record, Dr. Cho decided to employ the following constraints: the creation of a minority district, 378 county and city preservation, 379 population equality, 380 and compactness.

*1035 Because she concluded from State Representative Huffman's statement that incumbent protection was not a goal of the legislature when drafting the enacted map, Dr. Cho did not include as a constraint the avoidance of pairing incumbents. 381

After generating the 3,037,645 simulated maps based on only neutral criteria, Dr. Cho engaged in two overarching analyses using partisan data. Again, this use of partisan data came into play only after the simulated maps were produced. First, she engaged in a Plaintiff-specific analysis. Second, she examined the partisan unfairness of the map as a whole by comparing its partisan characteristics to the partisan characteristics of the set of simulated maps.

a. Plaintiff-specific analysis

Dr. Cho was given the home addresses of each individual Plaintiff, which allowed her to determine where each Plaintiff would live in each simulated map and to compare each Plaintiff's current district with each Plaintiff's set of simulated districts. Dr. Cho "compute[d] the average Democratic vote share for the plaintiff's current district by calculating the average Democratic vote share in that district for congressional races from 2012 to 2016 ...." 382 For the simulated maps, Dr. Cho "calculate[d] the average Democratic vote share for the plaintiff's [simulated] district ... with the 2008-2010 statewide election data." 383 These data included eight statewide races: the 2008 presidential race, the 2010 U.S. Senate race, the 2008 and 2010 Attorney General races, and the 2010 Governor, Auditor, Secretary of State, and Treasurer races. 384 Dr. Cho used statewide races to "avoid issues with district-specific factors and provide[ ] greater comparability across the state as a whole." 385 From there, Dr. Cho compared the likelihood of electing a Democratic candidate in each Plaintiff's simulated districts with the likelihood of electing a Democratic candidate in their current district. 386 We provide a fuller discussion of these findings in Section III.A., but we will provide two illustrative examples here. Some Plaintiffs, such as Plaintiff Goldenhar, live in allegedly cracked districts. Dr. Cho's analysis showed that "[a]mong the set of simulated maps, 95.68% of them would have placed Plaintiff Goldenhar in a district that would have provided a higher likelihood of electing a Democrat." 387 That, is 95.68% of the simulated maps placed Plaintiff Goldenhar in a district with a higher average Democratic *1036 vote share. Other Plaintiffs, such as Plaintiff Inskeep, live in allegedly packed districts. Dr. Cho's analysis showed that "none of [the simulated maps] would have placed Plaintiff Inskeep in a district that would have provided a higher likelihood of electing a Democrat." 388 That is, 0% of the simulated maps placed Plaintiff Inskeep in a district with a higher average Democratic vote share.

b. Partisan unfairness analysis

In addition to her Plaintiff-specific analysis, Dr. Cho examined the partisan outcomes of her simulated maps as compared to the current map, which allowed her to assess partisan effect. At a high-level, Dr. Cho assessed competitiveness 389 and partisan bias using multiple metrics. 390

i. Competitiveness

Dr. Cho "consider[ed] a district to be competitive if the margin of victory, or the difference between the Republican two-party vote share and the Democratic two-party vote share, is 1) within 5 percentage points and 2) within 10 percentage points." 391 Dr. Cho concludes that "[a]t the 5% margin of victory, the simulated maps generally have between 2-6 competitive seats," and that "[f]or both parties, [winning] 1-3 seats with a margin of victory within 5% [is] not unusual." 392 Meanwhile, the current map produced three competitive elections within a 5% margin of victory, one in 2012 (District 16) and two in 2018 (Districts 1 and 12), and the Republican won each. 393 Additionally, one other election in 2012 (District 6) was competitive at the 10% margin of victory. 394 Under the simulated maps, "often, 9 of the seats are competitive at the 10% margin of victory"; the next most common result was 8 competitive seats. 395 Three or four of these competitive seats (at the 10% margin of victory) generally favor Republicans, and four to six generally favor Democrats. 396 In her supplemental report, Dr. Cho provides further analysis of competitiveness based on the 10% margin of victory. "For the 2012-2014 data, 2-3 of the competitive seats were commonly Republican while 3-5 of the competitive seats were commonly Democratic." 397 In 2018, that number remained the same for Republicans, but competitive seats that leaned Democratic decreased to three or four. 398

Based on her analysis of competitiveness, Dr. Cho concludes that "[t]he Republican margins across the entire set of districts [in the current map] are large enough that they are sufficiently insulating to produce an enduring effect." 399 Moreover, she concludes that because of "the difference in the competitiveness, via several *1037 different measures, 400 of the simulated maps versus the current map, it seems that competitiveness was almost a non-existent factor if one at all in the construction of the enacted map since the current districts lean so heavily toward one party." 401

ii. Responsiveness and bias

In her initial report, which utilized 2008-2010 election data, Dr. Cho assessed the responsiveness and bias in the simulated maps compared to the current map using two measures based on the seats-votes curve (which shows how, as the proportion of votes a party receives increases, so too should that party's seat share). 402 When Dr. Cho measured responsiveness, she produced her results in a histogram in which, as the values along the x-axis increase (from left to right), the responsiveness increases; thus, maps falling along the right of the x-axis are more responsive than those on the left. 403 Dr. Cho concludes that the current map is "less responsive than almost all of the simulated maps." 404

Dr. Cho employed a symmetry measure to assess biasedness. This measure is grounded in the concept that "both parties should expect to receive the same number of seats given the same vote proportion." 405 Dr. Cho again produces her results in a histogram. "Here, a value of zero [in the middle of the x-axis on the histogram] is unbiased." 406 Positive values to the right of zero indicate a Republican bias, and negative values to the left indicate a Democratic bias. 407 Dr. Cho finds that, although most of the simulated maps "have a Republican tilt[,] ... the tilt toward Republicans is larger in the current map than it is for the simulated maps." 408 Indeed, some of the simulated maps were neutral and some even had a Democratic tilt; at any rate, H.B. 369 is far to the right of the simulated maps' Republican tilt as presented in figure 26. 409

iii. Seat share

Dr. Cho also compared the seat share between the parties from the current map *1038 to the seat share in her simulated maps. Based on the use of 2008 and 2010 election data, "none of the [simulated] maps in [Dr. Cho's] sample had the same 12-4 seat share as in the challenged map." 410 Furthermore, figure 19 of Dr. Cho's initial report shows that the most common outcome in the simulated maps was eight or nine Republican seats, at about 1.3 million and 1.2 million respectively. 411 Just over 250,000 of the simulated maps produced a 10-6 seat share in favor of Republicans, 412 and some of the simulated maps even produced six or seven Republican seats. 413 Very few of the simulated maps produced an 11-5 seat share, but that outcome is barely visible in figure 19. 414

Dr. Cho performed the same analysis using 2012-2014 data and 2018 data in her supplemental report. This analysis shows that over the decade, a 9-7 seat share in favor of Republicans became the most common partisan outcome in the simulated maps. 415 An 8-8 seat share is the second most common outcome, but by 2018, the number of 8-8 outcomes was about equal to the number of 10-6 outcomes. 416 "Eleven [Republican] seats occurred 0.12% of the time in the 2008-2010 analysis, 0.20% of the time in the 2012-2014 analysis, and 1.88% of the time in the 2018 analysis." 417 Finally, using the 2018 data, "a small number of maps, 1,445 out of more than 3 million total maps (0.046%) had, like the current map, 12 Republican seats." 418

3. Dr. J. David Niven

Dr. J. David Niven testified at trial for Plaintiffs as an expert witness. Dr. Niven is a tenured associate professor of political science at the University of Cincinnati, and he has a doctorate in political science from The Ohio State University. 419 He teaches a variety of classes, including on the U.S. Congress and congressional elections, government and politics in Ohio, and political parties, among others. 420 Dr. Niven's scholarship focuses on questions of congressional representation and elections, public opinion, and voting preferences, and he has published in peer-reviewed journals and book chapters on these topics but not on redistricting and gerrymandering specifically. 421 Before writing his reports in this case, Dr. Niven had never used census tracts specifically, though he had "used a variety of census data points in understanding the makeup of districts as a whole." 422 Also before writing his reports in this case, Dr. Niven had never tried to identify boundaries for communities of interest. 423 This Court admitted Dr. Niven as *1039 an expert in political science, subject to Defendants' Daubert motion. 424

Dr. Niven's report and testimony assessed the current map's makeup and the degree to which the districts divide communities of interest and reflect the political preferences of local residents. He undertook this examination by analyzing census tracts 425 that were either kept intact or split and by using the election data contained in "the 2010 Ohio Common and Unified Redistricting Database ('OCURD')" that was available to the map drawers during the 2011 redistricting. 426 Dr. Niven used census tracts as a basis for his analysis because they represent "a compact delineation of people who live in common geographic, cultural, and economic circumstance." 427

Dr. Niven finds that between the 2002 redistricting plan and the 2012 redistricting plan, the number of census tracts split between multiple congressional districts rose from 209 to 332 (out of approximately 3,000 census tracts). 428 Dr. Niven further finds that census tracts kept intact had an average Republican composition of 52.14%, whereas split census tracts had a higher composition of Democratic voters, with Republicans averaging 49.25% in split census tracts. 429 We note that Dr. Thornton reaches slightly different results on the partisan makeup of these census tracts and that there is a debate about the statistical significance of these results. See infra Section II.D.2.b. (discussing this issue). Nevertheless, both experts agree that split census tracts lean Democratic and intact census tracts lean Republican, and both agree that the number of census splits increased in the current map from the prior one.

We credit Dr. Niven's census-tract analysis to the extent that it shows some differential treatment between Republican and Democratic voters, and we observe that this difference is consistent with the nature of other splits (not involving census tracts) present in the current map. We do not give any significant weight to just the raw number of splits, without any further context. For example, census tracts could contain more than one municipality, so a split census tract could nonetheless keep *1040 its component municipalities intact. 430

In his response to Dr. Thornton, Dr. Niven also shows that, using a four-election index, 431 9.4% of Republican census tracts and 13.8% of Democratic census tracts were split among multiple congressional districts. 432 Using an eight-election index, 433 9.7% of Republican census tracts and 13.5% of census tracts were split. 434 In sum, split census tracts leaned Democratic, and census tracts with more Democratic voters were also more likely to be split into multiple congressional districts than census tracts with more Republican voters. 435

After his statewide analysis, 436 Dr. Niven discussed particular districts. His report focuses on Hamilton County (Districts 1 and 2), District 9, Franklin County (Districts 3, 12, and 15), and Summit County (Districts 11, 13, 14, and 16). Dr. Niven's report also surveys political science literature that shows that, when neighborhoods are divided into different districts, campaign efforts become "more complicated and less efficient ...." 437 Dr. Niven similarly testified at trial that "the political science literature is very clear that the more you subject a neighborhood to political splitting, ... it has a demobilizing effect.... It's harder for parties and other entities to go into a neighborhood and activate voters when those voters live in separate districts and, therefore, are responding to separate candidates." 438

a. Hamilton County: Districts 1 and 2

Dr. Niven began his analysis of Hamilton County with District 1. He notes that District 1 swung back and forth between electing Republicans and Democrats under the prior map and that one "academic analysis deemed [District 1] a 'textbook example of a marginal district.' " 439 After redistricting, that has not been the case. Dr. Niven's analysis shows, for example, that in 2008 President Obama won the old District 1 with 55.17% of the vote compared to Senator John McCain's 44.83%. By contrast, the same election under the current District 1, which splits Cincinnati and more of Hamilton County than under the old District 1, results in a 52.3% to 47.7% win for Senator McCain. 440 The new District 1 both split Hamilton County and *1041 added the whole of Warren County, which votes heavily Republican (and voted heavily for Senator McCain in the 2008 presidential election). 441 Using an index that incorporates a wider array of elections ("Dr. Niven's index"), 442 he found that Republican candidates averaged 42.07% of the vote in the old District 1, but that index percentage increased to 51.89% in the new District 1. 443

Meanwhile, District 2 was and remains safely Republican, but fourteen Cincinnati neighborhoods are divided between Districts 1 and 2. 444 Dr. Niven explains that "Cincinnati is unusual in its commitment to formally recognizing and building policy around the city's 52 neighborhoods. Indeed, the city's economic development strategy is built around the individual needs and assets of individual neighborhoods ...." 445 He notes that "while the rest of Hamilton County gave 52.19% of its vote" to President Obama in 2008, "the Cincinnati neighborhoods divided between the 1st and 2nd districts gave 59.37% of their vote" to President Obama in that election. 446 Looking at those same neighborhoods under Dr. Niven's index, the "split neighborhoods gave more than 75% of their vote to Democratic candidates" and the percentage for the rest of Hamilton County was about 45%. 447 Dr. Niven testified that "the 2nd District becomes something of a donor district. It had more Republicans than was needed to ensure a safe district." 448 In short, Cincinnati and these neighborhoods supported Democratic candidates, and they are split between Districts 1 and 2; District 2 already contained a large Republican majority, and thus it could take on those Democratic voters without putting a Republican candidate at any material risk of losing.

Throughout his report, Dr. Niven highlighted certain district boundary lines in which the lines divide census tracts populated by Democratic voters. In the case of his example for Hamilton County, the split census tract "is overwhelmingly populated by Democrats" per Dr. Niven's index. 449

b. District 9

Dr. Niven emphasizes that "[o]ne of the defining aspects of the 9th Congressional district is its comprehensive propensity to divide communities." 450 In fact, District 9 contains no whole counties and five partial counties-Cuyahoga is split between District 9 and three other districts, Lorain is split between District 9 and two other districts, and Erie, Lucas, and Ottawa are split between District 9 and one other district. 451 Dr. Niven further explains that "[i]n its economic development efforts, the state of Ohio places Cleveland and Toledo in separate regions," and thus, in combination *1042 with other cultural differences between Cleveland and Toledo, District 9 "combines quite disparate communities." 452 Dr. Niven's illustrative example of a suspect boundary for District 9 is in Lorain County, and the boundary divides a census tract that is heavily Democratic and more Democratic than the rest of Lorain County. 453 Moreover, each county in District 9 voted Democratic in the 2008 presidential election and leaned Democratic under Dr. Niven's index. 454

c. Franklin County: Districts 3, 12, and 15

Dr. Niven finds that Franklin County both packs (District 3) and cracks (Districts 12 and 15) Democratic voters. 455 Dr. Niven ultimately concludes that "what was achieved in these rather odd-looking districts is that a very Democratic County [Franklin County] winds up with two Republican representatives ... out of its three members of Congress." 456 On cross-examination, Dr. Niven acknowledged that under the prior map, Franklin County was split into three districts and that Republican candidates for Congress usually won, with some exceptions, the elections in those districts. 457 As will be discussed in more detail in the analysis, although this redrawing seemingly adds a Democratic district where there previously was not one, it was part of an overall strategy to solidify Republican districts and reduce the statewide number of Democratic districts.

He begins his analysis with District 15, a District which was competitive in 2006 and was won by a Democratic candidate for Congress in 2008. 458 Dr. Niven's analysis shows President Obama carried the old District 15 by about 29,000 votes, but the same election in the new District 15 would result in Senator McCain winning by 21,000 votes; under Dr. Niven's index, the old District 15 was nearly evenly split between Democratic and Republican supporters, with a very slight Democratic lean, and the new District 15 leans Republican. 459 Dr. Niven notes that nine out of the ten counties added to District 15 in the 2011 redistricting process "were inclined to support Republican candidates." 460 Additionally, the portions of three of the four split counties within District 15 leaned heavily Republican in the prior decade, except for the portion of Franklin County in District 15, which voted 50.52% in favor of Senator McCain and scored a 0.5237 (leaning Republican) under Dr. Niven's index. 461 The portions of those same counties not within District 15, however, had: a less-strong Republican tilt (Fayette County), were competitive (Ross County), or leaned heavily Democratic (Franklin County). 462 He also finds that the new District 15 split seventy-two census tracts (with fifty-eight in Franklin County), but the old District 15 split forty-one (all in Franklin County). 463 In sum, Dr. Niven concludes that *1043 Democratic-leaning areas were removed from the old District 15, while Republican-leaning areas were added, together resulting in a "net gain of more than 40,000 votes for the Republicans." 464

District 12 under either the 2008 presidential election results or Dr. Niven's index went from a leaning-Democratic district in the prior decade to a strongly-Republican district under the current map. 465 Dr. Niven's analysis shows that Democratic-leaning voters in Franklin County were removed from District 12 and Republican-leaning voters were added, resulting in a new gain of 60,518 Republican voters (using the 2008 presidential election data). 466 He further finds that census tract splits increased from forty-eight to sixty-one between the prior map and the current map. 467

District 3 is the final Franklin County district addressed by Dr. Niven. He concludes that District 3 "is a classic packing example" because it received Democratic voters from Districts 12 and 15. 468 Dr. Niven emphasizes the odd, jagged shape of District 3, and he testified that he included specific, street-level examples of odd lines in his report because "when we look statewide, ... it's hard to appreciate in the most granular detail the number of cuts necessary to achieve these effects." 469 Overall, he found that "14 out of 16 cities in Franklin County are split between multiple [congressional] districts." 470 In responding to Intervenors' expert Dr. Brunell's view that "funny shaped districts are inevitable," see infra Section II.D.3., Dr. Niven testified that, in this case, the "funny shapes" were "a strategic choice" and that they are "an illustration of division ... imposed with a partisan tinge such that democrats are far more likely to have found themselves in the midst of these cuts and divides." 471

Dr. Niven explained how gerrymandered district lines can cause confusion. For example, Dr. Niven found that in Franklin County, voters showed up to the polls for the 2018 special election, only to find out that they did not in fact live in District 12. 472 As it turned out, election officials had mis-assigned more than 2,000 people to the wrong congressional district, and the Franklin County Board of Elections took more than 4,000 calls (and received hundreds of emails) from confused voters who could not cast a ballot or whose polling locations were closed. 473

d. Summit County: Districts 11, 13, 14, and 16

Summit County's population is small enough such that it could be placed within a single congressional district-yet *1044 Summit County is divided into four congressional districts. (The prior map split Summit County into three districts.) Using either the 2008 presidential election or Dr. Niven's index, Dr. Niven's analysis shows that Summit County leaned Democratic. 474 He also finds that census tract splits increased from twenty-seven under the prior map to fifty-five under the current map. 475

As for the particular districts in Summit County, Districts 11 and 13 have consistently elected Democratic candidates to Congress under the current map, whereas Districts 14 and 16 have consistently elected Republican candidates. Consistent with these results, using either 2008 presidential election data or Dr. Niven's index, Dr. Niven's analysis shows that voters placed into Districts 11 and 13 leaned heavily in favor of Democratic candidates; meanwhile, voters placed into Districts 14 and 16 were almost evenly divided in the 2008 presidential election, and under Dr. Niven's index, the voters placed in these Districts leaned Republican. 476 Lastly, Dr. Niven finds that split census tracts leaned more Democratic than census tracts kept intact in Summit County, and he therefore concludes that "Summit County residents were not equally apt to have their neighborhoods divided between districts - as more heavily Democratic areas were more likely to be divided." 477

4. Dr. Lisa Handley

Dr. Lisa Handley, an election consultant who works on voting rights and redistricting, testified for Plaintiffs as an expert witness. 478 She has taught and lectured on voting rights and redistricting and has published articles and books on these subjects. 479 She has served as a redistricting consultant, aiding jurisdictions to draw lines in compliance with the VRA. 480 She has also served as an expert witness performing racial bloc voting analyses in cases in which districting plans are challenged under Section 2 of the VRA. 481 She has been hired as an expert by the Department of Justice in five cases and has provided expert testimony in over twenty cases throughout her career. 482 The Court qualified Dr. Handley as an expert in the VRA, including on racially polarized voting and analysis of such voting patterns. 483

*1045 District 11 has consistently elected African-American representatives to Congress since 1968, when it was first drawn as a majority Black district. 484 Handley's report indicated that since 2002, the Black-preferred congressional candidate (whether or not that candidate was African American) has won District 11 by a considerable margin. 485 This is true of elections both before and after the 2011 redistricting. 486 In fact, the tightest congressional race since 2002 in District 11 was won by Stephanie Tubbs Jones in that year with 76.3% of the total vote. 487 Prior to the 2011 redistricting, District 11 had a BVAP of 57.7%, although it was originally drawn in 2001 with a BVAP of 52.3%. 488 After the redistricting, its BVAP was 52.4%. 489

Dr. Handley conducted a "district-specific, functional analysis of voting patterns by race to ascertain the black voting age population necessary to provide black voters with an opportunity to elect their candidates of choice in the vicinity of the 11th Congressional District of Ohio." 490 The analysis must be district specific because the BVAP required to elect the Black-preferred candidate differs from jurisdiction to jurisdiction based on factors such as the type of election (e.g., federal versus local), turnout and voting patterns of African Americans and whites, the *1046 cohesiveness of African-American voters in supporting particular candidates, and "crossover" voting patterns of whites who also support Black-preferred candidates. 491 Dr. Handley's analysis estimated the vote share that Black-preferred candidates would have received had District 11 been configured as 55%, 50%, 45%, or 40% Black. 492 She conducted this analysis using data from statewide and federal elections from 2008 through 2016 occurring within the vicinity of the current District 11.

Dr. Handley used three different statistical techniques to complete this analysis: homogeneous-precinct analysis, ecological-regression analysis, and ecological-inference analysis. 493 Both homogenous-precinct analysis and ecological-regression analysis were used in Thornburg v. Gingles , 478 U.S. 30 , 106 S.Ct. 2752 , 92 L.Ed.2d 25 (1986), the Supreme Court's seminal Section 2 case. 494 Ecological-inference analysis developed later to address a shortcoming of ecological-regression analysis but has subsequently been widely accepted. 495 All three statistical techniques yielded similar results. 496

Dr. Handley concluded that with a 45% BVAP in District 11, African-American voters would have a realistic opportunity to elect their candidate of choice with a "comfortable margin." 497 In fact, even with a BVAP as low as 40%, African-American voters would have elected the Black-preferred candidate in the elections studied. 498 She concluded that there is no need to draw a majority African-American District 11 in order to allow African-American voters to elect their candidate of choice there. 499

5. Mr. William Cooper

William Cooper, a mapping consultant, testified as an expert witness at trial. 500 Over the course of his career, Mr. Cooper has drawn plans for about 750 jurisdictions, many of which were statewide plans and around six of which were congressional districting plans. 501 Mr. Cooper has also previously drawn plans specifically for partisan-gerrymandering cases. 502 Mr. Cooper generally submits illustrative or remedial districting plans, and courts have implemented several of his remedial plans. 503 This Court qualified Mr. Cooper as an expert in the fields of redistricting, map drawing, and demography 504 and found his testimony and reports credible and reliable.

Mr. Cooper used census data and mapping software "to reexamine the plan that was adopted in 2012 and apply traditional redistricting principles to result in a map that was a little more fair for Democratic voters and at the same time visually more appealing" and also "undid ... [the] partisan *1047 gerrymander." 505 He used Maptitude software, the same kind used by the map drawers in 2011, to do this work. 506 Mr. Cooper relied upon traditional redistricting principles (equipopulation, contiguity, compliance with the VRA, and preserving communities of interest) to craft his Proposed Remedial Plan and also made sure that it would satisfy the requirements of Ballot Initiative 1. 507 He "did not pair incumbents except when in direct conflict with the other factors." 508 Mr. Cooper had the CSU dataset used by the map drawers available to him while he was drawing his Proposed Remedial Plan and "occasionally glanced at it" although he "was not constantly monitoring every little-every little change." 509 The Proposed Remedial Plan that he created was intended to be a forward-looking plan that avoided the pairing of the current congressional officeholders. 510

Mr. Cooper explained the traditional redistricting factors that drove his maps and the manner in which those factors are measured. Equipopulation means that a district is the exact population of the ideal district size, plus or minus one. 511 Contiguity means that a district is entirely contiguous with itself; there are no severed sections. Compactness can be measured with an "eyeball test ... just take a look at it and see if it makes sense visually" or with mathematical tests such as the Reock and Polsby-Popper measures, both of which can be run using Maptitude. 512 The Polsby-Popper and Reock metrics measure compactness on a scale of zero through one; the closer to one, the more compact the district. The "Polsby-Popper score is a perimeter score over area of a district"-the ratio of the perimeter and the area of a district generates the score. A low score is "an indication that it's not a very compact district." 513 The Reock score is "a ratio of an area for a circle drawn around the district." Mr. Cooper testified that "districts that start getting below .20 are somewhat problematic, generally speaking." 514

Mr. Cooper defined a community of interest as "an area or a region where there are certain cultural or socioeconomic ties, historical ties." 515 He testified that minority populations can be considered communities of interest and that counties or municipal *1048 subdivisions are "a more objective way to identify communities of interest." 516 Maptitude allows users to monitor how many counties and metropolitan civil divisions are split as a plan is drawn. 517 He stated that, generally, maps with fewer districts overall should contain fewer county splits if traditional districting principles are being applied. 518

Mr. Cooper also compared the shapes of several districts from the 2012 map to his Proposed Remedial Plan, commenting on the 2012 districts' irregular shapes and frequent splits of county lines and municipal boundaries. 519 The Proposed Remedial Plan splits fourteen counties and twenty-seven political subdivisions. 520 In contrast, the 2012 map splits twenty-three counties and seventy-three political subdivisions, fifty-five of which are populated. 521 Mr. Cooper also compared the compactness of the districts in the 2012 map with those in his Proposed Remedial Plan. The Proposed Remedial Plan "score[d] significantly higher on Polsby-Popper in terms of minimums and maximums as well as the overall mean" than the 2012 map. 522

Mr. Cooper's Proposed Remedial Plan was conscious of advancing minority voting power in various districts. First, it included a minority-opportunity district contained entirely within Cuyahoga County with a 47% BVAP, higher than the 45% that Dr. Handley calculated was necessary to allow minorities in the district to elect a candidate of their choice. 523 Mr. Cooper testified that simply by keeping the City of Cleveland whole in District 11 and including "a couple of suburbs," achieving this 47% BVAP "just happened" without "trying to max it out in any way." 524 Second, Mr. Cooper's Proposed Remedial Plan included a District 1 with a higher percentage BVAP than the 2012 map's District 1. The Proposed Remedial Plan's District 1 has a 26.74% BVAP; the 2012 map's District 1 has a 21.30% BVAP. 525 He testified that this increase of over five percentage points resulted "because [he] left Cincinnati in a single district rather than splitting it into part of District 2 as well as District 1." 526 Third, the District 3 included in his Proposed Remedial Plan had roughly the same BVAP as was present in the 2012 map. 527

Mr. Cooper also responded to the report of Defendants' expert Dr. Hood. 528 Dr. Hood had challenged the Proposed Remedial Plan, arguing that it would not have been politically viable had it been implemented *1049 in 2012 because it would have paired many incumbents. Mr. Cooper maintained in his response that the Proposed Remedial Plan was "presented for future use, not solely as a point of comparison to the 2012 plan." 529 He also drew and demonstrated the feasibility of two hypothetical plans that shared many features with his Proposed Remedial Plan but could have been implemented in 2011 without pairing more incumbents than the adopted 2012 map did. 530

D. Defendants' and Intervenors' Expert Witnesses

1. Dr. M.V. Hood III

Dr. M.V. Hood III, a tenured professor of political science at the University of Georgia, testified as an expert for Defendants at trial. 531 Dr. Hood has taught courses in Southern politics, American politics, research methods, election administration, and the legislative process. 532 His work has appeared in peer-reviewed journals between forty and fifty times and he has published four articles "directly related to redistricting in one way or another" in peer-reviewed journals. 533 Dr. Hood has testified as an expert witness in several cases involving redistricting. 534 We qualified Dr. Hood as an expert in "American politics and policy, quantitative political analysis and election administration, including redistricting." 535 We, however, can draw limited inferences from his testimony and report due to some inapt comparisons, unexplained and apparently meaningful exclusions of certain elections in his partisan indices, and admitted failures to account for certain confounding variables in some of his analyses. 536

a. Incumbent pairing, core retention, compactness, and county and municipality splits

Dr. Hood's report stated that the 2012 map paired three sets of incumbents. 537 He also testified that the 2012 map's core retention level, the "percentage of a member's constituents [who] were carried over from their previous district," was "55.7% across the 16 districts." 538 Dr. Hood concluded, based on the number of incumbents who were paired and the core-retention rate, "that at least some weight was given in the plan to the ... criteria protecting incumbents to the extent possible." 539 Dr. Hood, however, agreed that *1050 "there is no agreed-upon standard for what levels of core retention indicates that the goal of a districting map is to protect incumbents." 540 He also acknowledged that in a previous academic article, he had concluded that "a core retention level of 68.7 percent greatly altered the relationship between representatives and constituents." 541

Dr. Hood compared the 2012 map with the 2002 map. He testified that the 2012 map was "on par with the 2002 plan in terms of compactness" measured with both the Polsby-Popper and Reock tests. 542 He stated that the 2002 plan split twenty-one counties and the 2012 plan split twenty-three counties. 543 He found that the 2002 plan split 4.3% of Ohio's municipalities while the 2012 plan split 4.5% of Ohio's municipalities. From this data, he concluded that the 2012 map "is on par with the 2002 benchmark plan" in terms of its adherence to traditional redistricting criteria. 544

Dr. Hood also compared the 2002 map to Plaintiffs' Proposed Remedial Plan in terms of compactness and splits of communities of interest, defined here as counties and municipalities. He found that the Proposed Remedial Plan had "slightly higher" compactness scores than the 2002 map measured by both the Polsby-Popper and Reock tests. He also testified that Mr. Cooper's hypothetical plans, which were designed as alternatives that could have been enacted in 2012, also had higher compactness scores than the adopted 2012 map. 545 The Proposed Remedial Plan splits fourteen counties while the 2002 map split twenty-one. 546 The Proposed Remedial Plan splits 1.7% of Ohio's municipalities while the 2002 map split 4.3% of them. 547 Mr. Cooper's hypothetical plans also split fewer counties and municipalities than the enacted 2012 map. 548

Dr. Hood also demonstrated that, had Plaintiffs' Proposed Remedial Plan been enacted in 2012, it would have resulted in the pairing of six sets of incumbents, the majority of which would have been Republican pairings. 549 Dr. Hood calculated that had the Proposed Remedial Plan been enacted in 2012, its mean core-retention figure would have been 39.5%. 550 As is discussed in the summary of Mr. Cooper's testimony, the Proposed Remedial Plan was designed principally as a forward-looking map to be implemented today, using the 2012 map rather than the 2002 map as a baseline. It designed its incumbent pairings based off where current representatives live under the 2012 map. This makes it an inapt comparison to count incumbent pairings that would have resulted had it been implemented in 2012, when *1051 a different set of representatives would have been the affected incumbents. Similarly, the implementation of the 2012 map shifted the district lines and assigned constituents to new districts. Therefore, it is odd to conduct core-retention analysis of the Proposed Remedial Plan against the baseline of the 2002 district lines when it was designed with the 2012 lines as its baselines. On cross-examination, Dr. Hood acknowledged that Mr. Cooper's hypothetical plans, which were designed as alternatives that could have been enacted in 2012, had core retention rates that were "highly similar" to those of the actually-enacted 2012 map. 551

b. Political geography

Dr. Hood also discussed Ohio's political geography-"the spatial distribution of partisans in Ohio." 552 He created a partisan vote index using fifteen statewide contested elections from four election cycles prior to the 2011 redistricting. 553 He then used this partisan vote index to color code and plot areas of Democratic, strong Democratic, Republican, and strong Republican support on several maps of Ohio. 554 Based on these maps, Dr. Hood concluded that "there's a much larger Republican footprint outside of urban areas. Much of the Democratic footprint during this time is inside urban areas, like Cleveland and Columbus, Cincinnati." 555 He calculated that "about 78.5% of Ohio's land area" leans Republican, and 21.5% of its land area leans Democratic. 556

Dr. Hood then calculated a Moran's I statistic to determine that from 2004 to 2010 "Republican VTDs tend[ed] to be located proximate to other Republican VTDs, and Democratic VTDs tend[ed] to be located proximate to other Democratic VTDs" in Ohio. 557 Dr. Hood acknowledged on cross-examination that this analysis did not "indicate that Democrats are differentially clustered than Republicans"-that they cluster with other members of their own party at higher rates than Republican voters do. 558 His analysis also demonstrated that "Democratic VTDs are more likely to be located in urban areas" than Republican VTDs. 559

c. Partisan leanings

Dr. Hood then used his first partisan index to analyze the partisan leaning of Ohio's congressional districts as drawn under the 2012 map. 560 He determined that six were safe Republican districts, five were competitive, Republican-leaning districts, four were safe Democratic districts, and one was a competitive, Democratic-leaning district. 561

*1052 Dr. Hood did the same analysis applying the partisan index to the Plaintiff's Proposed Remedial Plan and found that the only differences between it and the 2012 map were that under the Proposed Remedial Plan there would be "on[e] less safe Republican district and one additional competitive district leaning Democratic." 562 On cross-examination, Dr. Hood conceded that his "index state[s] a lower Republican percentage as compared to [an index that includes] the full set of elections based on the statewide contested elections for the decade preceding the 2010 redistricting cycle, including 2002." 563 When the 2002 congressional election results are included in the index, there are no competitive districts, rather than the six competitive districts that Dr. Hood indicated. 564 Such an index predicts voting outcomes that more reliably correspond to the actual electoral outcomes observed in the elections since the 2012 redistricting. 565

Dr. Hood created another partisan index using elections from 2012, 2014, and 2016, and then used the same process described earlier to color code the partisan leanings of VTDs on a map of Ohio. 566 Comparing that map to the color-coded map he produced of Ohio using elections from the preceding decade, he concluded that Ohio has become increasingly Republican over time. 567

Finally, Dr. Hood used this latter partisan index to evaluate the partisan leanings of each individual Plaintiff's new district under the Proposed Remedial Plan compared to the partisan leanings of their current district under the 2012 map. 568 He concluded, based on this analysis, that two of the seventeen individual Plaintiffs would have a better chance of electing a Democratic representative under the Proposed Remedial Plan versus under the current map-Plaintiff Griffiths in District 7 and Plaintiff Hutton in District 14. 569

d. Other influences on electoral success

Dr. Hood also testified about various factors that "influence the outcome of congressional races"-"[f]undraising, media attention, name recognition, incumbency," as well as "candidates and campaigns." 570 He testified that there is a strong trend of incumbents being reelected to office that is recognized in the political science literature and was observable in Ohio after the 2011 redistricting-all of the unpaired incumbent congressional representatives were reelected in 2012 and in every congressional election in Ohio since then. 571 Relatedly, Dr. Hood testified about challenger quality, which he measures by whether the challenger has held prior *1053 elective office. 572 He concluded that "[t]ypically, more often than not, the challengers" of incumbents in Ohio from 2012 through 2018 were "political novices" without prior elective officeholding experience. 573 Dr. Hood admitted on cross-examination that he did nothing "to assess whether the district lines themselves prevented the recruitment of experienced candidates" and that it was possible that they had. 574

Dr. Hood also examined "the amount of campaign contributions that were collected by the Republican and Democrat" in each election because fundraising is helpful in winning elections. 575 He concluded that, in Ohio between 2012 and 2016, the incumbents had "outraised challengers by about $ 1.2 million on average." 576 On cross-examination, Dr. Hood admitted that he did nothing "to determine that the district lines themselves did not cause Democratic challengers to fail to raise comparable funds" and admitted that it was possible that the lines themselves affected challenger fundraising abilities. 577

e. Efficiency gap and seat-share relationship

Dr. Hood plotted the efficiency gap numbers for Ohio from 1992 to 2016 against the seat share of the congressional delegation. 578 He concluded based on the regression from this plot that the efficiency gap is "closer to zero as the seat share is more evenly balanced" between the parties and increases "as the seat share tilts one way or another." 579

2. Dr. Janet Thornton

Dr. Janet Thornton testified at trial for Defendants as an expert witness. Dr. Thornton is currently the managing director and an economist and applied statistician at Berkeley Research Group, LLC, a consulting firm located in Florida. 580 Dr. Thornton has a doctorate and master's degree in economics from Florida State University, as well as a bachelor's degree in economics and political science from the University of Central Florida. 581 Dr. Thornton's fields of specialization in her academic background were labor economics and applied statistics. 582 Additionally, Dr. Thornton has "been working with census data since the early 1980s" and has also "work[ed] with data from the 1960 d[e]cennial census all the way up to the current time period ...." 583 Although Dr. Thornton has prepared statistical analyses and served as an expert in voting cases related to, for example, the effect of voter-identification laws on voter-participation rates by race and minority status, Dr. Thornton has never served as an expert in *1054 a redistricting case. 584 And although Dr. Thornton has never been precluded from testifying as an expert, at least one court found her analysis "simplistic and not credible." See Democratic Nat'l Comm. v. Reagan , 329 F.Supp.3d 824 , 838 (D. Ariz. 2018). Dr. Thornton has also not published any articles related to voting. 585 This Court qualified Dr. Thornton as an expert in economic and statistical analysis, subject to Plaintiffs' Daubert motion. 586

Dr. Thornton's report and testimony are offered to rebut Plaintiffs' experts Dr. Cho and Dr. Niven. As to Dr. Cho, Defendants presented Dr. Thornton's report and testimony to critique the underlying data and assumptions in Dr. Cho's report. 587 As to Dr. Niven, Defendants offered Dr. Thornton's report and testimony to rebut Dr. Niven's conclusion that the splitting of census tracts in the current plan is correlated with the political composition of census tracts. 588 Before turning to Dr. Thornton's critique of each of these Plaintiffs' experts, two preliminary matters need to be addressed.

First , we give no weight to Dr. Thornton's finding that "Dr. Cho failed to provide all of the underlying code and output sufficient to replicate all of her findings." 589 This finding is entirely off base. Dr. Thornton admitted that she is not an expert in C++ and that she cannot read it without the help of a manual; 590 and again, Plaintiffs offered to provide Defendants with the code. See supra Section II.C.2. More importantly, the code is not the algorithm; the code simply implements the algorithm. Consequently, nothing prohibited Dr. Thornton from critiquing the MCMC algorithm used by Dr. Cho if she had been qualified to do so. 591

Second , Dr. Thornton is an expert in statistics generally, not in political science or redistricting, and she has never run an MCMC algorithm or, prior to this case, reviewed, evaluated, or assessed an MCMC algorithm. 592 We consider her findings with that backdrop. Ultimately, we give some weight to her critiques of the underlying data that Dr. Cho used as a basis for assessing her simulated maps, *1055 but several of Dr. Thornton's other critiques miss the mark and are not credible.

a. Rebuttal to Dr. Cho

Dr. Thornton opines that "the manner in which [Dr. Cho] generates new maps (i.e., simulations) is biased towards selecting half of the districts in which the Republican votes outnumber the Democratic votes and half of the districts in which Democratic votes outnumber the Republican votes." 593 In other words, Dr. Thornton's opinion is that the process Dr. Cho used to produce the simulated maps was biased toward creating an 8-8 map. This is wrong. As explained earlier, Dr. Cho analyzed the competitiveness and partisan outcomes of the simulated maps only after the simulated maps were generated. See supra Section II.C.2. 594 Dr. Thornton offered no evidence to rebut this sequence of events.

In a similar manner, Dr. Thornton criticizes the election data that Dr. Cho used to assess the partisanship of the simulated maps as compared to the current map. This criticism, however, is distinct in an important way because it goes to Dr. Cho's after-the-fact assessment of partisanship and not the creation of the simulated maps. The general thrust of Dr. Thornton's critique on this front is that the 2008-2010 data used by Dr. Cho contains higher Democratic vote totals than in the 2012-2016 data. 595 Further, Dr. Cho never used the 2016 statewide Democratic vote share for her analysis, which Dr. Thornton computed as 42.4% (lower than the other indices used by Dr. Cho). 596 Dr. Thornton concludes that Dr. Cho's selection and use of election data "is faulty, misleading, and unreliable." 597

We give some weight to this particular conclusion-Dr. Cho's omission of the 2016 election data (which was less favorable to the Democratic Party) and use of 2008-2010 data to assess the partisan effect of the 2012 plan raises some concern. At the same time, Dr. Thornton's critique on this point does not significantly undermine Dr. Cho's conclusions. After all, the 2008-2010 election data were part of the data available to the map drawers, so that data is not irrelevant to assessing whether different districts could have been drawn. It is true, however, that the Democratic vote shares have decreased in the present decade as compared to the last, and this waning in support is relevant to partisan effect. In response, Dr. Cho provided an updated analysis in her supplemental report that incorporated the 2012-2014 and 2018 election data; that analysis showed the most common Republican vote share as nine seats, and eight and ten Republican seats were also not uncommon. See supra Section II.C.2. This cures at least part of Dr. Thornton's critique, specifically that using the 2008-2010 data misleadingly resulted in eight Republican seats being most common. In any event, Dr. Cho's supplemental report further shows that incorporating *1056 recent election data does not significantly alter her conclusions on partisan effect-a 12-4 map is still a highly unusual outlier under all her analyses. In sum, although we give some weight to Dr. Thornton's critique on Dr. Cho's selection and use of data, hence rendering Dr. Cho's findings less probative than they otherwise could be, we do not find that Dr. Thornton has significantly undermined Dr. Cho's conclusions.

Dr. Thornton also performed her own analysis using a binomial distribution, but we do not give any weight to that analysis. Dr. Thornton's analysis used the Republican statewide vote share in congressional races "to predict the number of Republican seats." 598 As an example, in 2016, the Republican vote share was 58.2%, and Dr. Thornton multiplied that number by 16 (i.e., the number of seats) to arrive at 9.31 as the expected number of seats (2.69 fewer seats than the actual outcome of 12). 599 Dr. Thornton then calculated "the number of standard deviations associated with the difference between the actual and predicted number of Republican seats." 600 When the difference is less than two standard deviations, whether positive or negative, the difference is not considered statistically significant. 601 From this analysis, Dr. Thornton concludes that for 2012, 2014, and 2016, "the difference between the actual and predicted number of Republican seats using the Republican vote share are not statistically significant." 602

Several factual and legal problems are apparent in Dr. Thornton's analysis. Factually, under the binomial distribution, the expected number of Republican seats unquestionably reflects proportional representation-Dr. Thornton multiplied the statewide vote share by the number of seats. Legally, proportional representation is not required. See infra Section IV.B. For this reason, Dr. Cho does not assume proportional representation. 603 The analysis incorporates yet another faulty assumption that each district has a 51% chance of being won by a Republican because Republicans won 51% of the congressional vote across the State; this assumption does not comport with basic understandings of congressional elections, i.e., that although some districts may be competitive (a 51% Republican to 49% Democrat district), other districts lean heavily in favor of one party or the other. Finally, Dr. Thornton's analysis has nothing to do with whether Republicans and Democrats are statistically treated similarly or differently under the current map-she assesses only whether the actual number of Republican seats differs in a statistically significant way from the expected number of Republican seats. This analysis, without more, says nothing about how the current map affects Democratic voters compared to Republican voters. For all of these reasons, we give no weight to her statistical significance analysis.

Additionally, Dr. Thornton applied a similar analysis comparing the difference between the number of Republican seats in 2010 and the number of Republican seats *1057 in 2012. 604 Again, she concluded that this difference was not statistically significant. 605 We find that this analysis is simplistic and not particularly helpful. To be sure, the Republicans flipped the congressional delegation in 2010 from one that was a Democratic majority to one that was a Republican majority, and this Republican majority has been maintained. But that simply shows part of the problem with the 2012 map: Despite fluctuating vote shares, the seat share has remained 12-4; under the prior plan, the seat share fluctuated as did the vote share. Indeed, the fact that a political party that controlled the redistricting process maintained (or slightly improved) their seat-share percentage from before redistricting to after is not surprising if they have drawn an effective partisan gerrymander.

Lastly, Dr. Thornton critiqued Dr. Cho for not considering incumbency in her analysis, and Dr. Thornton herself observed the success of incumbent candidates under the current map. 606 This critique holds some weight, but Dr. Cho's analysis still permits an inference, albeit less strong, on the partisan effect of the current map. See infra Section V.A.2.d. (addressing the problems with the incumbent-protection justification as applied to this case).

b. Rebuttal to Dr. Niven

Defendants also offered Dr. Thornton to rebut some of Dr. Niven's findings. According to Dr. Thornton, she performed analyses similar to Dr. Niven's but reached different results. 607 First, her "attempt to replicate Dr. Niven's finding [on the political orientation of census tracts left intact versus those which were split] result[ed] in an estimate that 50.48% of census tracts left intact are Republican in contrast to 48.18% among those that were split under the current plan" using the same election data as Dr. Niven. 608 The corresponding numbers from Dr. Niven were 52.14% (or 0.5214) and 49.25% (or 0.4925). 609 Dr. Thornton further critiques Dr. Niven's failure to perform the same calculations for the prior plan, which according to Dr. Thornton shows "a 0.4% increase in the percentage Republican among census tracts left intact" between the 2002 plan and the 2012 plan "and a 2.4% decrease in the percentage Republican among census tracts that were split between the two plans ...." 610 Second, Dr. Thornton "prepared correlation statistics to determine if the splitting of a census tract is correlated with the percentage Republican" using the same election data as Dr. Niven. 611 She concludes that "split census tracts are, statistically speaking, not correlated with the percentage Republican in the census tract as measured by Dr. Niven under either the prior plan or the current plan." 612 At trial, Dr. Thornton further testified that "there is no statistically significant difference in the proportion Republican and *1058 whether or not a census tract is split." 613

As an initial matter on this issue, we credit Dr. Niven's census tract analysis for a limited purpose. See supra Section II.C.3. Debates about the strength of various correlations aside, each expert's calculations are close to 50%, and both experts agree that split census tracts lean slightly Democratic. Moreover, Dr. Thornton's analysis is not entirely clear-she measured whether "the splitting of a census tract is correlated with the percentage Republican ...." 614 Dr. Niven, on the other hand, seems to have tested the statistical significance of the difference between census tracts that were left intact (which lean Republican) and those that were split (which lean Democratic). 615 An analysis of this differential treatment between Republican and Democratic voters seems to be absent from Dr. Thornton's report.

3. Dr. Thomas Brunell

Dr. Thomas Brunell testified at trial for the Intervenors as an expert witness. Dr. Brunell is a tenured professor of political science at the University of Texas at Dallas. 616 He received his bachelor's, master's, and doctorate, all in political science, from the University of California, Irvine. 617 Dr. Brunell teaches classes on Congress, political parties and interest groups, campaigns and elections, redistricting, and statistics, among others. 618 He has published books and articles in peer-reviewed journals on redistricting, elections, issues of representation in government, and party polarization. 619 Dr. Brunell has served as an expert witness in several other redistricting and VRA cases. 620 This Court qualified Dr. Brunell as an expert in the fields of redistricting, elections, the VRA and representation, and statistics. 621

Dr. Brunell's report and testimony is offered to rebut Plaintiffs' experts Dr. Cho, Dr. Warshaw, Dr. Niven, Dr. Handley, and Mr. Cooper.

a. Rebuttal to Dr. Cho

Dr. Brunell questions whether Dr. Cho's simulated maps "serve as a good basis for comparison to the actual map." 622 For various reasons, Dr. Brunell opines that Dr. Cho's maps cannot serve as a good comparison to the current map. He asserts that "all of Professor Cho's maps would likely be tossed" because they *1059 do not perfectly equalize population. 623 For the reasons we explained earlier, see supra Section II.C.2., we do not find this critique persuasive. In brief, Dr. Cho's 1% population deviation does not alter or undermine her analysis of partisan outcomes. We further note this criticism, along with others, offered by Dr. Brunell seems to miss the point of Dr. Cho's simulated maps. 624 Dr. Cho's simulated maps are not offered as examples of maps that should be enacted by the State per se; rather, the simulated maps provide a baseline to compare the partisan outcomes between the current map and maps that incorporate only neutral criteria. Moreover, Dr. Brunell critiques Dr. Cho's failure to consider incumbent protection, and he testified that protecting incumbents is "automatically going to make all of her districts different from ... one of the main stated goals by the legislature here in Ohio." 625 We address this point in the context of evaluating the proof of partisan effect and considering Defendants' justifications for the map, see infra Section V.A.2.d. (addressing the problems with the incumbent-protection justification as applied to this case), and we observe again that Representative Huffman described incumbent protection as "subservient" to other criteria in the process of creating H.B. 319. 626

Dr. Brunell incorrectly reads Dr. Cho's histograms to "suggest[ ] that there are just a handful of different maps in Prof. Cho's exercises, each with hundreds of thousands of repetitions." 627 Dr. Cho responds in her rebuttal report that Dr. Brunell's inference is unsupported by the data provided: "none of these histograms can suggest anything about how many different maps are represented since two drastically different maps can have the same metrics.... The number of bars in the histograms has no relationship with the similarity of the maps." 628 Accordingly, we reject this critique by Dr. Brunell.

Next, Dr. Brunell disagrees with Dr. Cho's conclusion that the current map is not responsive to voters. Instead, he "would characterize Prof. Cho [sic] simulated maps as hyper-responsive." 629 He *1060 further offered his normative view that responsiveness is not necessarily a positive feature of a map because "[m]assive volatility in the seat shares of the two parties is probably not conducive to good public policy." 630 As a basis for his conclusions on responsiveness, Dr. Brunell partly relied on "an old article by Edward Tufte, who was one of the first people to ... talk about these two metrics of swing ratio and bias" (which are related to responsiveness). 631 In fact, the article is from the early 1970s, and the data provided are for Great Britain, New Zealand, the United States generally, Michigan, New Jersey, and New York. 632 Importantly, much of the data precede the one-person, one-vote cases decided in the early-to-mid-1960s-an era in which districts were malapportioned. Tufte also used a linear fit of the data, not a seats-votes curve like Dr. Cho, which is a different model with different underlying assumptions. 633 Because Dr. Brunell's critique is based in on an inapt comparison, we give it little to no weight.

Lastly, Dr. Brunell misunderstands the point of Dr. Cho's individual Plaintiff-specific analyses. He takes issue with the fact that, because some Plaintiffs end up in the same district under Dr. Cho's simulated maps, "we cannot know what the partisanship of all 16 of the districts looks like" in the simulated maps. 634 As Dr. Cho responds, this specific analysis "was never intended for this purpose, and [she] never suggested that the plaintiff data could be or should be used in this way." 635 We agree. Dr. Cho's Plaintiff-specific analysis provides a comparison between each Plaintiff's current district and each Plaintiff's set of simulated districts, and this analysis is thus some evidence of whether Plaintiffs currently live in a packed or cracked district. The Plaintiff-specific analysis is just that, Plaintiff-specific; it does not compare the current map as a whole to the set of simulated maps as a whole. Dr. Cho has made such a comparison in a separate analysis.

b. Rebuttal to Dr. Warshaw

Dr. Brunell's critique of Dr. Warshaw's metrics focuses only on the efficiency gap. First, Dr. Brunell points out supposed issues with using actual congressional elections to calculate the efficiency gap, including uncontested elections and the variability of candidates. 636 Dr. Warshaw acknowledges some drawbacks in his report, but he also explains that "[i]n practice, ... both legislative races and other statewide races produce similar efficiency gap results for modern elections where voters are well sorted by party and ideology." 637 We do not find unreasonable Dr. Warshaw's use of actual congressional election results to calculate the efficiency gap in congressional elections. 638 Second, Dr. Brunell quibbles with the efficiency gap's definition of wasted votes, *1061 stating that "[i]t is not clear why all votes for the winning candidate greater than the total number of votes for the losing candidate are not classified as wasted." 639 Dr. Warshaw, however, explains in his rebuttal the logic behind the definition of wasted votes, a term of art in the context of the efficiency gap-only "50%+1 of the total votes, rather than 1 more vote than the losing candidate's current vote tally, are needed to win a counter-factual election" and therefore the efficiency gaps considers wasted votes for the winning candidate beyond that 50%+1. 640 Dr. Brunell's critique does not thread the needle, telling us why the generally-accepted definition of wasted votes from the efficiency gap literature poses a problem for measuring the extent of a partisan gerrymander. Accordingly, it does not impact our view of the helpfulness of the efficiency gap as a tool. Third, according to Dr. Brunell, "[i]t is hard to say how much of a gap is too much. Is five too much, or seven, or ten?" 641 Furthermore, he criticizes the metric's variability across elections. 642 While these criticisms have some merit, they do not overcome Dr. Warshaw's use of other metrics and how Dr. Warshaw holistically determines whether a map is a gerrymander (e.g., a map must also be an outlier). See supra Section II.C.1. Accordingly, we find that Dr. Brunell does not undermine Dr. Warshaw's conclusions or the usefulness of the efficiency gap.

c. Rebuttal to Dr. Niven

The thrust of Dr. Brunell's response to Dr. Niven is that "when electoral boundaries are being drawn some cities, counties, communities, neighborhoods have to be divided" and "[t]he boundaries have to go somewhere ...." 643 Although that may be true as a general proposition, it does not respond to Dr. Niven's findings that the divisions imposed by the current map are more likely to be imposed on Democratic voters than Republican voters. See supra Section II.C.3. Dr. Brunell also comments on some conceptions of communities of interest used by Dr. Niven, noting that "[t]here is no clear definition of what constitutes a community of interest, but cities and counties are generally characterized as such[.]" 644

d. Rebuttal to Dr. Handley

Dr. Brunell's rebuttal to Dr. Handley does not contain any criticisms. His report simply states: "It is interest

This text is long and has been trimmed here. Open the source document for the complete record.