Georgia v. Ashcroft

OBERDORFER, District Judge,

concurring in part and dissenting in part.

I am pleased to join in Parts III.C.2 and III.C.S of Judge Sullivan’s opinion. I agree that Georgia has met its burden of proving, by a preponderance of the evidence, that the proposed Congressional and state House redistricting plans have neither a retrogressive purpose nor effect. However, with respect to the state Senate redistricting plan, I give greater credence to the political expertise and motivation of Georgia’s African-American political leaders and reasonable inferences drawn from their testimony and the voting data and statistics than to what I find to be flawed opinions of experts and conflicting lay witness testimony presented by the Department of Justice and intervenors.

As a legal matter, I am not persuaded that a plan that reduces the “probability,” see ante at 81-88, that minority candidates *103of choice will prevail in three of Georgia’s fifty-six Senate districts, yet preserves an equal or fair opportunity for those candidates and other minority candidates statewide, is for that reason alone “retrogressive” in effect in violation of § 5.

The question before us is whether the proposed Senate plan as a whole, has the “purpose or effect of denying or abridging the right to vote on account of race or color.” 42 U.S.C. § 1973c. To resolve that question we must determine the extent to which, if any, Georgia’s minority voters are likely to retain effective “voting” strength equivalent to what they possess under the benchmark Senate plan. As my colleague, Judge Sullivan, puts it, analysis of that question must be “fact-intensive,” requiring us to examine closely “the context in which the voting changes will occur.” Ante at 76. For that task we are not only judges of the law; like jurors we are triers of fact. It is our responsibility to determine the weight, the effect, and the value of the evidence, and the credibility of the witnesses. We are to judge expert testimony “just as any other evidence”; we may “accept it or reject it, or give it as much weight as [we] think it deserves.”48 Like jurors, we may draw from the evidence “any inferences or conclusions that reason and common sense lead [us] to make.”49 To establish a fact by a preponderance of the evidence the proponent of that fact must in essence persuade us “that it is more likely so than not so.”50 In determining whether a fact has been established by a preponderance of the evidence, we “should consider all the evidence bearing upon that fact, regardless of who produced it.”51 Georgia may therefore find support for its case in the testimony of the Department’s expert and lay witnesses.

Georgia has offered two primary sources of support for its argument that African-Americans will enjoy an equal opportunity to win even in districts where barely over 50% of the voting age population is black: testimony from Congressman John Lewis, state Senator Robert Brown, and state Senate majority leader Charles Walker, and Dr. David Epstein’s expert witness report. As Judge Sullivan notes, the testimony of expert witnesses in this case has been unhelpful, with no expert reports focusing “on the question of whether the proposed redistrictihg plans are retrogressive.” Ante at 64. The testimony of Lewis, Brown, and Walker — that African-American candidates can win elections anywhere in the state of Georgia where blacks are a bare majority, or even slightly less, of the voting age population — is to my mind entirely credible and inadequately refuted by the Department’s expert witness or lay witnesses from the three contested state Senate districts. As trier of fact, I would therefore find that it is more likely than not that:

(1) The proposed Senate plan will increase from twelve to thirteen the number of Georgia Senate districts in which African-Americans are a majority of the voting age population.

(2) Under that plan, African-Americans statewide will .retain the power to elect eleven state Senators.

(3) Results from statewide and local elections in the three disputed districts, viewed in the context of the Lewis, Brown, and Walker testimony and reasonable inferences therefrom, establish that over the *104next decade there will be sufficient and increasing white support for minority candidates of choice in the disputed districts and statewide.

4. Black voters in proposed Senate Districts 2 and 26, in tandem with reliable white crossover, will have the necessary voting strength to continue to elect their candidates of choice. In Senate District 12, African-American voters may not be able to defeat the white incumbent to elect their candidate of choice to the Senate seat, but will retain an influential role in elections, representing no change from the status quo in terms of that district or in terms of the number of minority candidates of choice elected to the Senate statewide.

Accordingly, I am persuaded that the proposed Senate plan has neither a retrogressive purpose nor effect, and is entitled to preclearance under § 5.

I. Findings of Fact

Although the facts are well stated in Judge Sullivan’s comprehensive opinion, the findings I make and the inferences I draw on the basis of the extensive record in this case require some repetition and reevaluation.

A. 2001 Reapportionment Process

Georgia’s African-American legislators were key figures in crafting the Congressional, House, and, particularly, the Senate reapportionment plans. The Congressional reapportionment plan was drafted by a conference committee, consisting of three state Senators and three members of the state House. Three of the six were African-American. See ante at 42. Of the twenty-nine members of the state House committee responsible for redistricting, six were African-American. See id. at 40. Twenty-four members of the state Senate, six of whom are African-American, served on the Senate redistricting committee. See id. Senator Brown, who represents Senate District 26, served as the vice-chairman of that committee and chaired the subcommittee responsible for drafting the proposed Senate plan. See id. at 41.

In addition to participation by individual legislators, the voting support of African-American legislators for all three of the proposed plans was overwhelming, and in the case of the Senate plan, necessary for its adoption. No member of the House or Senate Legislative Black Caucuses voted against the proposed Congressional reapportionment plan. See id. at 43. One African-American member of the House, Dorothy Pelote, of the Savannah area, and one African-American member of the Senate, Regina Thomas, also of the Savannah area, voted against the proposed House plan, which passed in both chambers with a comfortable margin. See id. at 46. Representative Pelote and Senator Thomas were also the only African-American legislators to vote against the proposed Senate plan. See id. at 55. Although the House approved the Senate plan by a 101 to 71 vote, the plan was approved in the Senate by a narrow 29 to 26 vote, see id., essentially on political party lines. See E. Johnson Dep. at 27. Ten out of eleven African-American senators voted in support of the redistricting plan. Without the support of at least nine of those African-American senators, the plan would have failed.

B. Senate Redistricting

1. Comparison of the Benchmark and Proposed Plans

Under the benchmark redistricting plan, thirteen of Georgia’s fifty-six Senate districts are majority African-American, measured both by total population and voter registration. African-Americans are a *105majority of the voting age population in twelve of those districts. See ante at 55-56.

The proposed plan, based on 2000 census data, also creates thirteen Senate districts where African-Americans are a majority of the total population. If voter registration is used as a yardstick, the number of districts where more than half of registered voters are African-American drops to eight. However, if voting age population is the relevant measure, the proposed plan creates thirteen districts with a BVAP greater than fifty percent, a net increase of one district.52

Statewide, the proposed Senate plan would “unpack” districts that previously had high concentrations of voting-age African-Americans, ranging from fifty-five percent to more than eighty percent, and would reduce those black majorities to the neighborhood of fifty (and a fraction) to sixty percent. Under the benchmark plan, only two of the twelve relevant Senate districts — District 12, with a 55.43% BVAP, and District 39, with a 54.73% BVAP — have less than sixty percent BVAP.53 Five other benchmark Senate districts have a BVAP between sixty and seventy percent. Four benchmark Senate districts have a greater than 70% BVAP, and in one — Senate District 43 — 88.91% of the voting age population is black. See Def.Ex. 117. Of these twelve majority-minority districts, eleven experience a decline in their BVAP percentage under the proposed Senate plan,54 along with a concomitant drop in the percentage of registered voters who are African-American. See Def.Ex. 118. Despite this overall decrease in arguably overconcentrated black voting majorities, the Department of Justice has focused narrowly on the three proposed Senate districts that would have the lowest BVAP levels under the proposed plan. Using Georgia’s methodology, under the proposed plan Senate District 2 would have a BVAP of 50.31%, Senate District 12, a BVAP of 50.66%, and Senate District 26, a BVAP of 50.80%.

2. Testimony of African-American Political Leaders

Georgia has offered from three of its African-American political leaders testimony that under the proposed plan black candidates, presumably the candidates of choice for the majority of black voters, will retain the ability to be elected in districts with a narrow majority of voting-age African-Americans, such as Senate Districts 2, 12, and 26.

Congressman John Lewis

Congressman Lewis represents the Fifth Congressional District, in the Atlanta metropolitan area. His leadership and courage during the Civil Rights Movement and since have been instrumental in the passage of the 1965 Voting Rights Act and *106its continuing vitality. See Lewis Dep. at 6-11. For decades, he has been intimately involved in, and informed about, the unique politics of the South, and particularly Georgia politics, beginning with his directorship of the Voter Education Project in 1970. Id. at 12. His credibility as an advocate for African-American voting rights, as the Justice Department acknowledges, is “beyond reproach.” Tr. 2/26/02 at 99. The Department and intervenors object that Congressman Lewis’ testimony is irrelevant, because he does not serve in the state Senate and does not reside in any of the three contested districts.55 However, Congressman Lewis based his opinion not only on his experience in campaigning for Congress in the metro Atlanta area, but on his familiarity with Georgia politics at all levels and in all regions of the state. “I’ve spent a great deal of time traveling the length and breadth of the state.... I keep up with what is happening all over the state. It doesn’t matter whether it’s south Georgia, extreme coastal or north Georgia, I try to be responsive and involved.” Id. at 17.

Congressman Lewis testified that, in his political judgment, black candidates have a better than even chance of winning a district with 50% BVAP throughout Georgia:

I think a candidate, a good solid black candidate, would have more than a 50 percent chance of winning with a 50 percent BVAP [district] in Georgia. Whether or not a black candidate wins in a district with that level of BVAP will depend more on the specifics of that particular candidate and his or her campaign. The kinds of things that are important in any campaign, like hard work, putting together a good organization, and so on, will make a difference. But a credible black candidate certainly has a good chance of winning a legislative seat anywhere in the State, I think with a 50% BVAP.

Id. at 18.

On the whole, Congressman Lewis finds the creation of a greater number of districts with a slim minority majority to be preferable to fewer “safe” minority districts. “[G]iving real power to black voters comes from the kind of redistricting efforts the State of Georgia has made with these plans, both for Congress and for the General Assembly, House and Senate.... [W]e don’t need to create these black enclaves politically. It dilutes the ability of minority voters to elect more people, and to affect who has the majority.” Id. at 23.

Senator Robert Brown

Senator Brown is the incumbent state Senator in District 26 and chairs the Senate subcommittee responsible for the proposed Senate redistricting plans. Like Congressman Lewis, he believes black candidates, both incumbents and challengers, can be re-elected and elected in a district with a 50% BVAP:

I think the incumbents in these districts at these BVAP levels are in very *107solid shape. But speaking specifically to the question of an open-seat, I think that an African-American candidate would have a good chance of winning. He or she would have to have a good organization and work hard, but there’s no reason why an African-American can’t win at a 50% BVAP.... And I can tell you this. The nearly unanimous consensus from the Black Caucus in the Senate that voted for the plan would never have been there had that not been a belief shared by those senators.

Brown Dep. at 30.

As the incumbent from Senate District 26, Senator Brown is familiar with, and singularly qualified to give opinion testimony about, the demographics and white crossover trends in his own jurisdiction. He also expressed the opinion that Senate Districts 2 and 12 specifically, in addition to his own district, would remain competitive for an African-American candidate of choice under the proposed plan. See id. at 40-42.

Senate Majority Leader Charles Walker

Georgia offered similar testimony from the state Senate majority leader, Charles Walker. Senator Walker represents a district centered in Augusta. He was of the opinion that the BVAP levels necessary for an African-American candidate of choice to have a fair opportunity of electoral success were even lower than Congressman Lewis and Senator Brown estimated: “Generally around the state, I would feel comfortable at a 45% BVAP level.... All of the 13 Senate districts in the plan are well above that level.” Walker Dep. at 12.

Senator Brown and Senator Walker, who are both Democrats, may well support the proposed redistricting plan in part because it is likely to preserve a Democratic majority in the state Senate, enabling Senator Brown to maintain his current committee leadership positions and Senator Walker to continue to serve as majority leader. But as a practical matter, Senator Brown and Senator Walker only reap the benefits of a continued Democratic majority in the Senate if they are able to be elected from their newly constituted districts, making it unlikely they would support a redistricting plan that materially reduced their re-election chances. Their willingness to reduce the BVAP levels in their own districts to bare majorities is persuasive evidence that they have such confidence in their estimates of minority political strength that they are willing to stake their own political positions on the accuracy of those estimates.

3. Expert Witness Testimony

Dr. Epstein, Georgia’s expert witness, analyzed the results in all elections in Georgia to the state House, the state Senate, and the United States Congress between 1996 and 2001, using a probit analysis. That analysis purports to calculate the statistical chance of an African-American candidate of choice winning election at varying levels of BVAP. See ante at 64-65. Based on the results in those elections, Epstein estimated that minority voters have an even or better chance of electing their candidate of choice in an open seat election if the BVAP is 44.3%, and a 75% or better chance of electing their candidate of choice when the BVAP rises to 50%. See id. at 66, 67. His analysis assumes that black voters are highly cohesive in supporting their respective candidates of choice, while white voters are relatively less so. “This estimate is reasonable in light of the significant difference in crossover voting; for black voters, it is typically less than 5%, while for white voters in general elections it is generally around 20%.” Pl.Ex. 25, at 17.

*108The Department’s expert witness, Dr. Richard Engstrom, and the intervenors’s expert witness, Dr. Jonathan Katz, criticize Epstein’s methodology, see ante at 67-69, but only Engstrom offers countervailing statistical evidence, which examines the extent to which racially polarized voting exists in Senate Districts 2, 12, and 26.

To assess differences in the voting preferences of white and black voters, Eng-strom looked at three categories of biracial elections in benchmark Senate Districts 2, 12, and 26: (1) elections to the Senate seat in two of the three districts; (2) election returns in each district in contests for statewide office; and (3) local elections in the largest city and/or most populous county in the district. He concluded that there were high levels of racial polarization (or low levels of white crossover voting) in the contested Senate districts. As Judge Sullivan noted, however, “Engstrom does not attempt to predict the effect of this polarized voting on the ability of minority voters to elect candidates of their choice under the proposed redistricting plans,” id. at 69, leaving it for us to determine the reasonable inferences and ultimate conclusions to be drawn from his data patterns in light of all the factual circumstances.

a. Senate District 256

i. Elections for State Senate

Engstrom’s analysis looks at three elections in Senate District 2, two non-partisan and one partisan: a 1999 special non-partisan election in which there were multiple candidates, including four African-Americans; a 2000 special non-partisan runoff between the winner and top African-American vote-getter, Regina Thomas, who won that election to become the incumbent in Senate District 2; and the 2000 general election between Senator Thomas and a white Republican challenger.

Engstrom’s data about the recent elections involving Senator Thomas shows that the willingness of white voters to vote for an African-American, far from being polarized, varies dramatically. In the 1999 special election, 20.4% of whites crossed over and voted for a black candidate, although their votes did not necessarily go to Senator Thomas, who received only 5.0% of the white votes cast in that election. Only 8.9% of white voters crossed over to support her in the special runoff, which she narrowly won against a well-qualified white candidate. However, in the following year, running as an incumbent Democrat in a general election, Senator Thomas won with 43.6% of the white vote and nearly 80% of the total vote.

None of the elections Engstrom examined involved typical circumstances. The 1999 special election and 2000 runoff were non-partisan, meaning that no candidate benefitted from party affiliation. Additionally, there were four black candidates and two white candidates in the special election, dividing the black vote. In the 2000 general election, the white candidate, in addition to being a Republican in a heavily Democratic district, was also a pizza delivery boy with a criminal record. See Thomas Dep. at 32-33.

ii. Statewide Elections

Engstrom’s analysis of votes cast in benchmark Senate District 2 in statewide races shows that, as an empirical matter, significant numbers of white voters crossed over to vote for a black candidate. Engstrom analyzed seven statewide races: the Democratic primary and general elec*109tion for Insurance Commissioner; the Democratic primary, Democratic runoff, and general election for Labor Commissioner; the Democratic primary for Public Service Commissioner; and the general election for Attorney General. In the four Democratic-only races, white crossover for the black candidate ranged from a low of 31.8%, in the runoff for Labor Commissioner, to a high of 58%, in the primary for Public Service Commissioner.57 In the general elections, where black Democrats faced off against white Republicans, 27.8% of white voters in benchmark district 2 voted for the black candidate for Insurance Commissioner, and 44.8% and 44.9%, respectively, voted for the black candidates for Labor Commissioner and Attorney General.58

The probative value of these results is affected by Engstrom’s necessary reliance on elections conducted in benchmark districts. There are significant differences between the political geography of the benchmark and proposed districts, due to the removal of some precincts and the addition of others. See ante at 85-86. To control for this, Engstrom performed a re-aggregation analysis, which examines voting patterns in the precincts that constitute the proposed district. See Def.Ex. 611, Table 4. His re-aggregation analysis shows only the total percentage of the vote a candidate would receive in the proposed Senate district, without breaking down the preferences of black and white voters. However, the analysis indicates that, with respect to proposed Senate District 2, every black candidate who ran in a past statewide election would have carried the proposed district, receiving between 57.4% and 77.7% of the total votes cast, a fact from which results in a Senate race in the new district may be reasonably inferred,

iii. Local Elections

Local races in Savannah, the largest city in Senate District 2, show low levels of white crossover voting, ranging from 2.8% to 10.6%. Engstrom analyzed eight elections, but those elections involved only two black candidates. Additionally, Engstrom codes all of the local elections in Savannah as non-partisan, which is atypical. Eng-strom did not analyze any elections in surrounding Chatham County.

b. Senate District 1259

i. Elections for State Senate

Engstrom analyzed the 1996 and 1998 Democratic primaries for the state Senate race in benchmark District 12. The black candidate of choice, John White, a longtime House member, narrowly lost both primaries, each time to a different white candidate. In 1996, only 10.6% of white voters crossed over to support White. Although he also lost his 1998 Senate bid, white crossover climbed to a more respectable 17.5%.

ii. Statewide Elections

Engstrom’s analysis of the statewide races in District 12 shows a plurality of white voters crossing over to vote for black *110candidates. In thé Democratic primaries and runoff, crossover voting ranged from 33.5%, in the Labor Commissioner primary (where there were more than two candidates), to 58% in the Public Service Commissioner primary. For the two remaining Democrat-only races, 44.1% of whites voted for the black candidate for Insurance Commissioner, and 40.8% voted for the black candidate for Labor Commissioner in the runoff.

In the general elections, the black candidate for Insurance Commissioner received 23.8% of the white votes cast. However, the black candidate for Labor Commissioner received 48.1% of white votes, in a race against a white Republican, and 44.6% of whites crossed over to vote for the black candidate for Attorney General.

These results are based on votes cast within the boundaries of benchmark Senate District 12. Engstrom’s re-aggregation analysis, which looks at votes cast within the boundaries of the proposed district, shows that the black candidate for Insurance Commissioner would have narrowly lost, but all other candidates but one would have received a majority of votes cast.60

iii. Local Elections

With respect to District 12, Engstrom analyzed three types of local elections: partisan races for state House seats, partisan races for elected positions in Dougherty County, and the mayoral race in Albany, which he described as partisan in some years and non-partisan in others.

In the House races, white crossover ranged from 9.9% to 17.1%.61 Three of the four House races involved the same black candidate, Roberts, while John White ran in the fourth race. In the two county-wide races in Dougherty County — Democratic primaries for the chair of the county commission and for county coroner — white crossover was 9.4% and 21.3%, respectively-

Engstrom analyzes four mayoral races in Albany: the partisan 1993 general election,62 the 1995 Democratic primary, the 1997 non-partisan general election, and the 1999 non-partisan general election. In the 1995 primary, Engstrom estimates white crossover votes for the two black candidates totaled 4.8%, with the stronger candidate receiving 4.0%. In the 1997 race, white crossover for the black candidate was 5.4%. In 1999, the four black candidates in the race received 12.8% of white votes cast, with the strongest candidate receiving 9.7%. Even though white support for black candidates remained under 10% in the mayoral elections in Albany, throughout the 1990’s white support for black candidates steadily increased.

c. Senate District 2663

i. Elections for State Senate

There have been no black-white contests for the seat in Senate District 26 under the *111current benchmark plan. Senator Brown was initially elected in a 1991 non-partisan special election, when African-Americans were a minority of registered voters, garnering 56% of the vote to defeat a credible white opponent. See ante at 62.

ii. Statewide Elections

In all seven state-wide races between black and white candidates, white voters in benchmark Senate District 26 demonstrated considerable support for black candidates. In the Democrat-only races, the black candidate for Labor Commissioner received 33.6% of white votes in the multi-candidate primary, and 36.9% in the two-person runoff. The black candidate for Insurance Commissioner garnered 41.5% of white votes cast, and 57.4% of white voters supported the black candidate for Public Service Commissioner.

In the general elections, 32% of whites voted for the black candidate for Insurance Commissioner, 46.9% voted for the black candidate for Attorney General, and 54.9% voted for the black candidate for Labor Commissioner. In Engstrom’s re-aggregation analysis, looking at how voting patterns would play out in the proposed Senate District 26, black candidates received the majority of votes in every two-person race, and received more than 60% of the vote in five out of the six races. In the multi-candidate primary, the black candidate was the top vote-getter, with 49.6% of white support.

iii. Local Elections

The Department’s expert analysis of local elections in Senate District 26 addressed partisan elections in Bibb County and in Macon. White crossover voting in local elections in Senate District 26 is generally higher than in either Senate District 2 or 12.

In races for seats on the Bibb County Board of Education, two different black candidates each received 34.2% of white votes, in general elections held in 1994 and 1998. A black candidate for district attorney received 16.6% of white votes in the Democratic primary. But in the Democratic primary for the county Board of Commissioners, only 2.7% of white voters supported the black candidate.

In Macon, Engstrom’s data shows that white support for three different black candidates for at-large city council seats ranged from 14.2% in the 1995 general election, to 22.9% in the 1995 general election, to 27.4% in the 1999 general election. However, in the 1999 Democratic mayoral primary, only 10.4% of whites supported the black candidate.

4. Testimony of Lay Witnesses

The Department of Justice has presented testimony, in the form of declarations, from nineteen local public officials and other community leaders who reside in the disputed Senate districts. See Def.Exs. 501-519. On their face, these declarations paint a grim picture of racially polarized voting in the three Senate districts, and cast doubt on the ability of African-Americans to elect candidates of choice in each of the districts as redrawn. However, the credibility and probative value of this testimony is seriously compromised by significant contradictions between these stark declarations, the more nuanced testimony of these witnesses when cross-examined in their depositions, and the actual results in local elections.

a. Senate District 2

The Department offers testimony of ten witnesses from this district, all of whom aver that racially polarized voting will prevent black candidates from succeeding in proposed Senate District 2. The most important testimony comes from Senator *112Thomas.64 As the incumbent in Senate District 2, her views on racial voting patterns, and her potential and the potential of future candidates of choice to be elected in the proposed district, deserve thoughtful consideration. Senator Thomas opposes the Senate plan because it removes parts of her current district, despite its underpopulation, particularly voters who had previously supported her; and also because it divides Chatham County among three senators instead of the present two. Thomas Dep. at 82-83. She further testified to her understanding that proposed Senate District 2 has a BVAP under 50%, and that generally, “when you have your numbers below 50 percent black voting age population then nine times out of ten you’re going to get a white representative or senator.” Id. at 102. Senator Thomas perceives that her BVAP numbers are lower than other African-American senators, which unfairly disadvantages minority voters in her district, but testified she would have opposed the plan even if her numbers were equivalent, due to the addition of a third senator in the county. See id. at 125.

Senator Thomas concedes that her narrow margin of victory in the 2000 special election, which she won by 70-odd votes, was likely for reasons other than racially polarized voting. She was opposed by a popular, well-qualified white candidate, Dana Braun, who had previously served as an at-large alderman in the city of Savannah. Braun was substantially better-funded and was endorsed by African-American elected officials and ministers, while not a single black elected official endorsed Senator Thomas. See id. at 22,142-143.

Despite the reduction in BVAP, Senator Thomas thinks she personally can win the proposed district. See id. at 121. She was less optimistic about the chances of any successor, because “[pjeople may not know them. They may not have any name recognition, and I think for the most part a minority would not win because, as I said earlier whites are more in tune with whites.” Id. at 122.

The Department also presented declarations from four African-American Savannah aldermen: Gwendolyn Goodman, Edna Branch Jackson, Clifton Jones, Jr., and David Jones. Two African-American Chatham County Commissioners, Harris Odell, Jr. and Joe Murray Rivers, also testified in opposition to the plan. Additionally, three local citizens who are active members of the NAACP testified against the plan: Dr. Prince Jackson is a former vice president of the board of education and a former president of Savannah State College, Richard Shinhoster is the acting president of the Savannah branch of the NAACP, and Helen S. Johnson is the CEO of the local civil rights museum and a member of the Executive Committee of the Savannah branch of the NAACP. These African-American community leaders in Senate District 2 are fairly confident that the more progressive white voters in Savannah’s historic neighborhoods will continue to vote for the better candidate, irrespective of race, and are primarily concerned that the addition of heavily white and heavily Republican precincts from the islands area could cause Senator Thomas to be defeated by a Republican challenger. Goodman testified that white voters, at least in her ward, vote primarily on a candidate’s ability. Goodman Dep. at 29. Clifton Jones, in his declaration, refers to a “general rule” that white voters prefer white candidates, C. Jones Decl. ¶ 8, but believes that he has received the support *113of roughly 40% of the white voters in his ward. C. Jones Dep. at 11-12. David Jones affirms that there are white citizens in Savannah who will vote for a black candidate over a white candidate. D. Jones Dep. at 19. Johnson stated in her declaration that “racially polarized voting patterns ... usually” occur in Savannah elections, H. Johnson Decl. ¶ 3, but explained in her deposition that “people don’t really vote because of racial issues ... in the city elections.... [They vote] [o]n the issues and I think the parties, different parties.” H. Johnson Dep. at 41.

There is evidence that black voters in Savannah have been able to form political coalitions with the city’s Jewish community. See Odell Dep. at 26; E. Jackson Dep. at 17. Additionally, African-American candidates who are Roman Catholic draw support from white Catholics in the Chat-ham County area. See Goodman Dep. at 33; P. Jackson Dep. at 8. The city’s white political establishment has shown some willingness to advance African-American candidates, most notably in the endorsement of a black mayoral candidates by Savannah’s long-serving white mayor. See E. Jackson Dep. at 96; H. Johnson Dep. at 40. In two recent city elections, white candidates made racist remarks about their African-American opponents, but those appeals were unsuccessful and rebounded to harm the white candidate among white voters. See E. Jackson Dep. at 78, 80; D. Jones Dep. at 24; P. Jackson Dep. at 30.

In terms of white crossover in the greater Chatham County area, Odell thought that Senator Thomas “would pick up the overwhelming majority of the lower middle income white voters” in an election in the proposed district. Odell Dep. at 25. Rivers testified that he draws white votes in Savannah and Chatham County, and speculated that he could garner a majority even in overwhelmingly white Tybee Island “because I have a lot of people I know out on Tybee.” Rivers Dep. at 42-43. Prince Jackson received white support when he defeated a white candidate in a 1970 election to the Savannah-Chatham County Board of Education. P. Jackson Dep. at 7-8. He estimates that 20% of white voters will typically cross over for a black, candidate and that 70% to 80% of white voters might support a very strong black candidate. See id. at 11, 62. “I would say 20 to 30 percent of people ... haven’t gotten to the point where they can vote for the other race.” Id. at 58.

Blanket statements in the declarations questioning Senator Thomas’ electability in the proposed district are significantly qualified by witnesses’ deposition testimony. In his declaration, Clifton Jones stated that Senator Thomas “probably will not be able to win against a strong white opponent.” C. Jones Decl. ¶ 21. In his deposition, although stating Senator Thomas “would have a better chance under the old plan,” he agreed that she had a fair chance of winning in the proposed district. C. Jones Dep. at 48. In his declaration, Shin-hoster stated that it was “unlikely that Senator Regina Thomas will be reelected from proposed Senate District 2,” Shin-hoster Decl. ¶ 16, but in his deposition expressed the opinion that Senator Thomas “would be a strong candidate for reelection.” Shinhoster Dep. at 28.

Much of the concern about Senator Thomas’ election chances are based on party politics, rather than race. David Jones clarified that he believes the proposed district throws her “into the fire,” D. Jones Decl. ¶ 5, because “Regina runs as a Democrat [and] [t]hey put her in a white Republican district as they extended her district.” D. Jones Dep. at 14. Odell, who stated in his declaration that Senator Thomas “would be beaten badly” by a *114strong white candidate, Odell Decl. ¶ 12, clarified in his deposition that he believed Senator Thomas would win the Democratic primary but lose in the general election to a white Republican.65 Odell Dep. at 24.

In their depositions, several of the local leaders were more sanguine about the prospects of African-American candidates other than Senator Thomas than their declarations had indicated. Goodman believes that a well-known black politician, such as herself or Savannah’s African-American mayor, Floyd Adams, would “get a fair shot” in the proposed Senate district. Goodman Dep. at 32. Rivers seconded Goodman’s opinion that she could be elected to the new Senate seat, although he believed she was an exceptional candidate. Rivers Dep. at 47.

b. Senate District 12

The Department offers the testimony of five black declarants from Senate District 12: Charles Sherrod, John White, Arthur Williams, David Williams, and William Wright. Sherrod is a former Albany city commissioner, who unsuccessfully ran for the District 12 Senate seat in 1992, and acted as White’s campaign adviser in his 1996 and 1998 election efforts for the same seat. White served in the state House for 22 years, and was defeated in the 1996 and 1998 Democratic primaries for the Senate seat by white opponents. Arthur Williams is a member of the Albany city council, representing Ward 3. David Williams is also a city council member, for Ward 6. Wright is a former president of the local NAACP branch, and unsuccessfully ran for election to the county board of commissioners and county board of education.

Senate District 12 presents a somewhat unusual situation. The district is currently represented by a white incumbent, Michael Meyer von Bremen, who was not supported by a majority of black voters in his 1998 election. The concern of local African-American politicians in this district is not the retention of an incumbent, but rather the ability of a minority challenger to defeat Meyer von Bremen.66

Sherrod attributes his defeat and White’s defeat to racial bloc voting. See Sherrod Decl. ¶¶ 6-10. In his declaration, Sherrod stated that “[m]ost white voters in Southwest Georgia simply will not vote for black Senate candidates.” Id. ¶ 11. When pressed to explain the success of two African-American judicial candidates in majority-white Dougherty County, as well as Sanford Bishop’s ability to be re-elected to Congress in a majority-white district which includes Senate District 12, Sherrod fell back on “a phenomenon down here in south Georgia that we can’t explain sometimes,” that occurs when white voters support black candidates. Sherrod Dep. at 97.

*115John White’s explanation of his losses in the Senate primaries, in both his declaration and deposition testimony, focused less on race and more on political factors. ■White faced the current Lieutenant Governor, Mark Taylor, in the 1996 primary, and Taylor substantially outspent him in the campaign. See White Decl. ¶ 12. White also suffered adverse publicity in that race, when the local and Atlanta newspapers reported that he had founded a company called “Connections Unlimited” to capitalize on his 22-year service in the state House. Possibly as a result, White attracted few crossover voters, and more than a third of black voters supported his white opponent. White also cited racial appeals by Taylor in the 1996 primary, see id. ¶¶ 12-18, without quantifying the harm to his campaign. City council member Arthur Williams suggests that White lost in 1996 because Taylor was better-funded and played “the race card.” A. Williams Decl. ¶ 6.

In the 1998 primary, which featured no such distractions, White believes he lost to Meyer von Bremen because of weak turnout among black voters. “I think some black people did not turn out to vote because they figured I was a sure bet to win.” White Deck ¶ 15.

c. Senate District 26

Five witnesses for the Department offered testimony in opposition to the proposed plan as it pertains to Senate District 26: Albert Abrams, the African-American president of the Bibb County Board of Education; William Barnes III, an African-American member of the Bibb County Board of Education; Bert Bivins III, an African-American county commissioner for Bibb County; C. Jack Ellis, the African-American mayor of Macon; and Samuel F. Hart, an African-American county commissioner for Bibb County. In Senate District 26, the declarants’ concerns about racially polarized voting are contradicted by the success of their own candidacies.

Abrams testified that, in his personal experience, “voting patterns are polarized along racial lines” in Bibb County, Abrams Deck ¶ 6. He acknowledged in his deposition that his personal experience includes being elected, with substantial white support, against a qualified white candidate in a county with 43% BVAP. Abrams Dep. at 15-22. Barnes defeated a white Democrat in the primary and a white Republican in the general election, winning nine out of ten precincts. Barnes Deck ¶ 2. In Hart’s election bid, whites not only voted for him, but also sponsored campaign events for him in their homes, although he is not certain he would have enjoyed the same support if his opponent had been white. Hart Dep. at 24-25.

Testimony from these witnesses indicates that it is likely that Robert Brown can be elected in the redrawn district. Hart stated that Senator Brown “is a shoo-in for winning re-election in proposed Senate District 26,” Hart Deck ¶ 9, and other local political figures agree that Senator Brown retains at least a good chance of being elected. See Abrams Deck ¶ 7; Barnes Deck ¶ 7; Ellis Deck ¶ 7. The local political leaders are more concerned that the redrawn district will compromise the ability of African-American candidates other than Senator Brown to be elected. Ellis believes “a non-incumbent minority candidate would have only a 50/50 chance of-winning in proposed Senate District 26.” Ellis Deck ¶ 9.

However, in their deposition testimony, local political leaders from Senate District 26 identified a pool of black candidates who could succeed Robert Brown. Abrams agreed that Ellis, as well as several African-American city council members, would be formidable candidates if any opt*116ed to run for the state Senate. Abrams Dep. at 61. Barnes believes an African-American candidate could be elected from a Macon-based district with a 50% BVAP. Barnes Dep. at 59. Hart described a pool of potential African-American candidates who could, like Abrams, “draw across the board” in Bibb County, Hart Dep. at 41, as well as potentially attract the same level of white crossover that Senator Brown enjoyed in his 1991 election. Id. at 44, 51-52.

II. Analysis

Section 5 requires that certain jurisdictions, Georgia among them, obtain pre-clearance from the Department of Justice or a three-judge panel of the United States District Court for the District of Columbia before implementing any change in a “qualification, prerequisite, standard, practice, or procedure” with respect to voting, including redistricting and reapportionment, to ensure that the proposed change “does not have the purpose and will not have the effect of denying or abridging the right to vote on account of race or color.” 42 U.S.C. § 1973c.

In determining whether Georgia is entitled to preclearance under § 5, we must determine on the basis of the facts “whether the ability of minority groups to participate in the political process and to elect their choices to office is [ajugmented, diminished, or not affected by the change affecting voting....” H.R.Rep. No. 94-196, p. 60, quoted in Beer v. United States, 425 U.S. 130, 141, 96 S.Ct. 1357, 47 L.Ed.2d 629 (1976). There is no retrogression, as defined by the congressional committee and reiterated in Beer and subsequent opinions, where a redistricting plan augments or has no effect on the voting power of a minority group; retrogression occurs only if a plan diminishes “the position of racial minorities with respect to their effective exercise of the electoral franchise.” Beer, 425 U.S. at 141, 96 S.Ct. 1357 (emphasis added). Neither the text of § 5 nor authoritative decisions interpreting it require the preservation of super or “robust” majorities that would guarantee election of the minority candidate of choice; the statute and precedents “merely mandate[ ] that the minority’s opportunity to elect representatives of its choice not be diminished, directly or indirectly, by the State’s actions.” Bush v. Vera, 517 U.S. 952, 983, 116 S.Ct. 1941, 135 L.Ed.2d 248 (1996) (emphasis added).

Opportunity does not necessarily equate to probability, although the majority so holds. See ante at 81, 82-83. Each majority-minority district (and in some cases, districts with a substantial minority population less than a majority), represents an opportunity for a minority candidate of choice to be elected. The majority, rather than comparing the number of majority-minority districts or the number of minority candidates of choice likely to be elected under the benchmark and proposed plans as the measure of opportunity, look to the probability that a minority choice candidate will be elected in each district, specifically whether the chance that a minority candidate of choice will be elected has decreased from a “robust” chance to “a ‘reasonable’ or ‘fair’ chance.” Id. at 77.

There is no legal authority for the majority’s proposition that § 5 requires that a plan preserve a pre-existing probability that a minority choice candidate prevail. To the contrary, the Supreme Court, albeit in the § 2 context, has consistently held that the Voting Rights Act aims to provide nothing more than a fair or equal opportunity, and does not guarantee “safe” seats or a “robust” chance of victory.67 Other *117lower courts have recognized, in the § 5 context, that a plan that preserves or increases the number of districts where minority voters have an equal or reasonable opportunity to elect their candidates of choice is not retrogressive. See Colleton County Council v. McConnell, No. 01-3581-10, slip op. at 95 (D.S.C. Mar. 20, 2002) (three-judge court) (examining the number of majority-minority districts maintained “at a level of equal opportunity”); see also Ketchum v. Byrne, 740 F.2d 1398, 1419 (7th Cir.1984) (defining retrogression as a decrease in “the number of wards in which blacks have a reasonable opportunity to elect a candidate of choice”). This does not conflate a § 5 inquiry with a § 2 inquiry. Rather, it recognizes that a simple comparison of the number of majority-minority districts under the benchmark and proposed plans, although traditionally employed by the courts, is by itself insufficient because it fails to answer the question of whether the majorities are at a level that enables “effective exercise of the electoral franchise,” Beer, 425 U.S. at 141, 96 S.Ct. 1357 (emphasis added).

A majority of the Supreme Court has never definitively answered the question of whether a redistricting plan that preserves or increases the number of districts statewide in which minorities have a fair or reasonable opportunity to elect candidates of choice is entitled to preclearance, or whether every district must remain at or improve on the benchmark probability of victory, even if doing so maintains a minority super-majority far in excess of the level needed for effective exercise of electoral franchise. Cf. City of Richmond v. United States, 422 U.S. at 371, 95 S.Ct. 2296 (in the context of annexation, § 5 does not require the maintenance of the same number of minority-controlled city council seats, when doing so would “permanently overrepresent[ ]” minority voters). However, in United States v. Mississippi, 444 U.S. 1050, 100 S.Ct. 994, 62 L.Ed.2d 739 (1980), the Supreme Court summarily affirmed a decision by a three-judge panel of this court, Mississippi v. United States, 490 F.Supp. 569 (D.D.C.1979) (Wilkey, Pratt and Greene, JJ.),68 granting pre-clearance to Mississippi’s reapportionment plan, with Justice Marshall dissenting. Although the lower court found no retrogressive effect because “both plans had the same number of districts with Negro voting-age populations of 60% or more,” 444 U.S. at 1058, 100 S.Ct. 994,69 Justice Mar*118shall nonetheless would have held that the proposed plan had a retrogressive effect because it decreased the BVAP is certain districts: in Leflore County from 71.72% to 64.26%, in Marshall County from 62% to 56%, and in Adams County from 70% to 63%. See id. at 1058 & n. 6,100 S.Ct. 994. Although Justice Marshall would have required the maintenance of BVAP majorities at a certain percentage, rather than at a certain probability of election, there is a linkage between the two:- a higher BVAP results in at least an incrementally higher probability of electoral success. See ante at 81-82. The outcome in Mississippi, although not binding precedent on this three-judge court, nonetheless squarely rejects the position adopted here by the majority, that a plan may be found retrogressive due to declines in individual districts, even though the plan as a whole, such as the Senate plan here, maintains minority voting strength statewide at levels equivalent to the benchmark plan.

Georgia bears the burden of proving, by a preponderance of the evidence, that its Senate and other redistricting plans are consistent with § 5 of the Voting Rights Act, but its burden is a limited one. A redistricting plan may be pre-cleared under § 5, yet still be “enjoined as unconstitutional,” Shaw v. Reno, 509 U.S. 630, 654, 113 S.Ct. 2816, 125 L.Ed.2d 511 (1993), or result in vote dilution actionable under § 2 of the Voting Rights Act, see Reno v. Bossier Parish Sch. Bd., 528 U.S. 320, 335, 120 S.Ct. 866, 145 L.Ed.2d 845 (2000) (“Bossier Parish //”).70 “In vote dilution cases § 5 prevents nothing but backsliding, and pre-clearance under § 5 affirms nothing but the absence of backsliding.” Id.

It follows from these principles that there is no retrogression where, as here, the facts establish “sidesliding” rather than “backsliding.” No precedent addresses this kind of “sidesliding,” occurring where as here the BVAP majority in a proposed district proves the be less than in the prefiguration, but the new alignment increases the number of BVAP districts statewide and retains the level African-American representation in the Senate. The proposed plan in comparison to the benchmark plan is more likely than not likely (1) to create statewide as many or more majority-minority districts, as measured by BVAP; and (2) to make it reasonable to anticipate that the number of successful minority candidates statewide will equal or exceed the number elected under the benchmark plan.

A. Retrogressive Effect

Today, we face an unprecedented, yet not unforeseen, challenge:. to assess whether a deliberate reduction of black super-majorities, undertaken with the endorsement of African-American legislators with the reasonable expectation that these bare majorities, in combination with mean*119ingful white crossover voting, would have the effect of enhancing or preserving minority voting strength statewide, is nonetheless retrogressive because it reduces the BVAP in three contested Senate districts to just above fifty percent.

Although a majority of the Supreme Court has not faced the issue, leaving us on uncharted ground, individual Justices have foreseen a situation such as this. When the Supreme Court introduced the concept of retrogression in Beer, Justice Marshall recognized that “it will not always be so easy to determine whether a new plan increases or decreases Negro voting power relative to the prior plan,” anticipating situations where the effectiveness of minority voters could be reduced by “packing” them into districts where their votes would be wasted: “Is it not as common for minorities to be gerrymandered into the same district as into separate ones?” Beer, 425 U.S. at 156 n. 12, 96 S.Ct. 1857 (Marshall, J., dissenting). Justice Thomas, discussing the trade-off between a high percentage of minorities within certain districts and minority influence statewide, stated that:

We have held that a reapportionment plan that “enhances the position of racial minorities” by increasing' the number of majority-minority districts does not “have the ‘effect’ of diluting or abridging the right to vote on account of race within the meaning of § 5.” But in so holding we studiously avoided addressing one of the necessary consequences of increasing majority-minority districts: Such action necessarily decreases the level of minority influence in surrounding districts, and to that extent “dilutes” the vote of minority voters in those other districts, and perhaps dilutes the influence of the minority group as a whole.

Bossier Parish I, 520 U.S. at 490, 117 S.Ct. 1491 (Thomas, J., concurring) (internal citation omitted).

“Retrogression” is an often-used but ill-defined term, not just by the parties in this case, but in the case law as well.71 As Judge Sullivan has pointed out, the number of majority-BVAP districts in the proposed plan “may be a more appropriate number to consider in determining whether a district is properly characterized as a ‘majority-minority’ district,” ante at 79, although it is not a measure to rely on exclusively. See id. Other judges have used the number of elected candidates of choice as a gauge of black voters’ ability to exercise effectively their electoral franchise under the benchmark and proposed plans. See, e.g., Holder, 512 U.S. at 895-908, 114 S.Ct. 2581 (Thomas, J., concurring). These numerical measures, without more, may or may not finally determine whether or not a plan is retrogressive. But, in any event, analysis of these numerical measures is a necessary predicate to the conclusive fact decision: the effect of a proposed plan on the minority’s ultimate ability to exercise its franchise effectively.

1. Defining Effective Electoral Strength

There are three different ways to define a district as majority-minority: by its total population, by its voting-age population, and by registration. Under the first measure, Georgia’s proposed Senate plan is not numerically retrogressive in the sense that the number of districts where Afri*120can-Americans comprise a majority of the total population remains constant at thirteen. If BVAP is used as the relevant measure, the proposed plan is numerically ameliorative, increasing the number of dis- - tricts with a majority BVAP by one, from twelve to thirteen.72 But if black voter registration is the appropriate measure, than the proposed plan is retrogressive, diminishing the number of majority-black districts by five.

The Supreme Court has not expressed a clear preference for any of the three measures. In Beer, the Supreme Court found that a plan that increased the number of districts with a majority black population from one to two, and increased the number of districts with a majority of black registered voters from zero to one, was ameliorative. See Beer, 425 U.S. at 141-142, 96 S.Ct. 1357. Without “expressing] any opinion on the subject,” the Court reiterated the trial court’s dilemma in determining whether registration or voting age population was a superior measure: Registration created a greater likelihood of electoral success, but “ ‘in essence condones voter apathy.’ ” Abrams v. Johnson, 521 U.S. 74, 94, 117 S.Ct. 1925, 138 L.Ed.2d 285 (1997) (quoting Johnson v. Miller, 922 F.Supp. 1556, 1568 n. 18 (S.D.Ga.1995)); see also Johnson v. De Grandy, 512 U.S. 997, 1018, n. 14, 114 S.Ct. 2647, 129 L.Ed.2d 775 (1994).

Here, I would follow the majority of lower courts that have embraced voting age population as the relevant ingredient of minority voting strength in Voting Rights Act litigation. Other courts have “consistently relied” on BVAP percentage in § 5 cases. See ante at 79. “In analyzing the racial fairness factor, the voting age population (VAP) is the relevant number to be used in determining whether minorities in a particular district will be able to elect a candidate of their choice.” DeGrandy v. Wetherell, 794 F.Supp. 1076, 1084 (N.D.Fla.1992); see also NAACP v. Fordice, 252 F.3d 361, 364 (5th Cir.2001); Emery v. Hunt, 272 F.3d 1042, 1044 (8th Cir.2001); Old Person v. Cooney, 230 F.3d 1113, 1122 (9th Cir.2000); Solomon v. Liberty County Comm’rs, 221 F.3d 1218, 1224 (11th Cir.2000). Two other three-judge courts, in determining whether their court-ordered redistricting plans met § 5 requirements, have used voting age population as the relevant statistic. See Smith v. Clark, 189 F.Supp.2d 529, 540-541 *121(S.D.Miss.2002); Colleton County, No. 01-3581-10, slip op. at 96.

Courts have adopted minority voting age population as the relevant measure of electoral strength for both practical and normative reasons. The vast majority of experts, as in this case, base their statistical analyses on voting age population. Voting age population is readily determined from census data, and although variable over time, is not as mutable as voter registration data. See ante at 79.

A state, in drawing its districts, controls the percentage of minorities of voting age population placed in that district. But those minorities, not the state, control, as a matter of individual choice and as a function of political organization through registration drives and the like, the percentage of registered minority voters.73 Section 5 prevents any diminution in the minority’s opportunity to elect representatives of its choice caused “by the State’s actions.” Bush v. Vera, 517 U.S. at 983, 116 S.Ct. 1941. By its terms, the phrase “effective exercise of electoral franchise” implies that it is not the role of Georgia, or any other state, to create districts that minority voters can win even when substantial segments of voting-age adults fail to register or, having registered, stay home on election day. “Accounting for lower voter registration and turnout rates among black citizens when determining what constitutes an ‘equal opportunity to participate in the electoral process’ and creating ‘safe black districts’ to compensate for those rates amounts to an incentive for and institutionalization of black voter apathy.” Johnson v. Miller, 864 F.Supp. 1354, 1392 (S.D.Ga.1994) (quoting Thornburg v. Gingles, 478 U.S. 30, 44, 106 S.Ct. 2752, 92 L.Ed.2d 25 (1986)).

The fact that a district has a certain percentage of voting-age African-Americans does not guarantee that the same percentage of voters will be African-American. The percentage of registered voters who are African-American may be higher or lower than the BVAP, although that variation is generally within a few percentage points, and turnout among black and white registered voters varies from election to election. In ten of the twelve benchmark districts with a BVAP majority, including Senate Districts 2 and 26, the percentage of registered voters who are black is higher than the BVAP percentage. See Pl.Ex. IE. This implies that voting-age African-Americans throughout the state are mobilized, because their registration rates are slightly in excess of white registration rates in ten of the twelve majority-minority districts.

Courts’ concepts of the percentage of minority voting age population necessary to comprise an “effective” majority, cf. Ketchum, 740 F.2d at 1402 n. 2, one that can elect minority candidates of choice, have varied over time. Until the early 1980s, conventional wisdom suggested that African-American super-majorities of 65% were needed to create effective majorities. See Rybicki v. State Bd. of Elections, 574 F.Supp. 1082, 1114 n. 87 (N.D.Ill.1982) (“The 65% figure is a general guideline which has been used by the Department of Justice, reapportionment experts and the courts as a measure of the minority population in a district needed for minority voters to have a meaningful opportunity to *122elect a candidate of their choice.”); cf. UJO, 430 U.S. at 164, 97 S.Ct. 996 (noting that the Justice Department’s conclusion that a non-white population majority in the vicinity of 65%, in order to achieve a nonwhite majority of eligible voters, was not unreasonable). In the intervening twenty years, there are strong indications that progress in race relations has virtually eliminated the rationale for 65% minority super-majorities. One court, applying the 65% guideline, noted presciently that “emerging changes in sociological and electoral characteristics of minority groups and broad changes in political attitudes may substantially alter, or eliminate, the need for a corrective. The 65% figure, in particular, should be reconsidered regularly to reflect new information and new statistical data.” Ketchum, 740 F.2d at 1416.

During the 1990s, courts began to recognize that districts with 55% voting age majorities preserved effective opportunities for minority voters, see Jeffers v. Clinton, 756 F.Supp. 1195, 1199 (E.D.Ark.1990), while minority voters, in certain circumstances, believed they retained effective voting power at even lower levels. See Bush v. Vera, 517 U.S. at 969, 116 S.Ct. 1941 (in the redistricting process, “[t]he community insisted that [a] ‘safe’ black district be drawn that had a total black population of at least 50%”) (internal quotations omitted); DeGrandy, 794 F.Supp. at 1088 n. 5 (African-American intervenors contended that district with less than 50% BVAP nonetheless provided African-Americans with an opportunity to elect candidates of choice). Based on the handful of cases decided thus far in the 2000 redistricting cycle, the trend indicates that courts, under certain factual circumstances, are now willing to accept the proposition that minority voters retain the ability to exercise their electoral franchise effectively in districts with a bare, 50% majority. See Colleton County Council, No. 01-3581-10, slip op. at 88, 96; see also Page v. Bartels, 144 F.Supp.2d 346, 358 (D.N.J.2001).

As a related measure, the Supreme Court has considered, as a de facto standard for effective electoral franchise, the number of seats minority voters control. “If using control of seats as our standard does not reflect a very nuanced theory of political participation, it at least has the superficial advantage of appealing to the ‘most easily measured indicia of political power.’ ” Holder, 512 U.S. at 899, 114 S.Ct. 2581 (quoting Davis v. Bandemer, 478 U.S. 109, 157, 106 S.Ct. 2797, 92 L.Ed.2d 85 (1986) (O’Connor, J., concurring)). Under the benchmark plan, minority voters have elected eleven candidates of choice, all African-American, from the twelve districts with majority BVAP. Black voters in benchmark Senate District 12 have been unable to elect their candidate of choice. Using this measure, the proposed plan would not be retrogressive if African-American voters retained the ability to control eleven or more seats in the state Senate.74

The number of majority-minority districts created by the proposed plan and the number of minority candidates of choice who can reasonably be expected to be *123elected from those districts does not end the inquiry into whether or not the proposed Senate plan satisfies § 5. It is also necessary in most situations to consider whether the reasonably anticipated white crossover vote will enable the numerical strength of black voters, in terms of BVAP, to translate into effective electoral strength, in terms of the ability to elect minority candidates of choice. At the bottom line, African-American voters effectively exercise their electoral franchise when their votes, taking into account the magnitude of the likely white crossover, will enable their candidates of choice to win, regardless of whether that victory is by a landslide or a single vote.

2. White Crossover Voting Versus Racial Polarization in the Proposed Senate Districts

Fifty percent is not a “magic number” for defining effective electoral strength. The extent to which minority voters join forces with voters of other races to elect a mutual candidate of choice is a key variable. Majority-minority districts are a necessary remedy in circumstances where white voters refuse to support minority candidates, but “in those communities in which minority citizens are able to form coalitions with voters from other racial and ethnic groups ... minority voters are not immune from the obligation to pull, haul, and trade to find common ground.” De Grandy, 512 U.S. at 1020, 114 S.Ct. 2647. A redistricting plan that requires minority candidates to draw support from white voters is not offensive to the purposes of the Voting Rights Act, and indeed has a virtue “which is not to be slighted in a statute meant to hasten the waning of racism in American politics.” Id. However, a redistrieting plan that lowers BVAP majorities to the point where minority candidates of choice will likely need at least some white support to prevail effectuates the purpose of the Voting Rights Act only where, as here, racism is waning and minority candidates can draw the necessary level of white support. See infra at 50-52.

In litigation under § 2 of the Voting Rights Act, the Supreme Court refers to racially polarized voting as voting patterns which produce elections in which a majority of whites and a majority of African-Americans each support a different candidates. This racial polarization is legally significant in a § 2 case only if “a white bloc vote ... normally will defeat the combined strength of minority support plus white ‘crossover’ votes.” Gingles, 478 U.S. at 56, 106 S.Ct. 2752. It should follow, on the opposite side of this coin, that in a § 5 context, racially polarized voting is legally and practically insignificant if minority candidates of choice will likely be able to attract sufficient white crossover votes to win. Where, as here, the proposed Senate plan creates a number of districts equal to or greater than the benchmark plan in which it is likely that a sufficiently large and cohesive African-American cohort, see id. at 50-51, 106 S.Ct. 2752, will combine with sufficient numbers of white crossover voters to create an absolute voting majority, the plan is not retrogressive in violation of § 5.

I agree that “[i]f voting patterns are not marked by racially polarized voting,” — in other words, if there is sufficient white crossover — reductions in the percentage of minority voters “may have little or no effect on their ability to elect preferred candidates” and would evidence the absence of backsliding or retrogression. Ante at 78. However, they invoke the testimony from the Department’s expert witness, Eng-strom, that African-American candidates will be unable to attract the necessary level of white crossover votes to win in the proposed districts. See id. at 88. Eng-*124strom has presented data from Senate races, statewide elections, and municipal and/or county elections. Although Eng-strom’s study uses well-accepted statistical methods, there is a large gap between levels of white crossover voting in different types of elections that he is unable to explain. Even without considering which type of election is most probative, the mere fact that “the degree of racial bloc voting varied widely from election to election” argues against a finding that legally significant racially polarized voting exists. Mallory v. Ohio, 173 F.3d 377, 382 (6th Cir.1999).

Senate elections provide the most probative results, because “elections involving the particular office at issue will be more relevant than elections involving other offices,” Magnolia Bar Ass’n v. Lee, 994 F.2d 1143, 1149 (5th Cir.1993), but the data sample is severely limited.75

The majority finds that Engstrom’s analysis of local elections, which indicates only a small percentage of whites will support a black candidate, is more probative than his analysis of the district-wide returns in statewide elections. See ante at 86. Although neither the statewide nor local races are for the relevant office, the results in the state-level races reflect the voting patterns of the Senate district as a whole, rather than a discrete and possibly unrepresentative part, such as its largest city or most populous county. Additionally, municipal and county boundaries do not necessarily overlap with the boundaries of the Senate district. See ante at 85-86. The use of local election results to predict the level of white crossover in future Senate races is therefore highly questionable, because the former differ from the latter not only in the type of election, but also involve an incomplete, and in some cases extraneous, group of voters. See Clark v. Calhoun County, 21 F.3d 92, 96 (5th Cir.1994) (rejecting the use of municipal elections as a predictor of equal opportunity to elect minority candidates of choice to county-wide office).

The statewide election results in the disputed Senate districts show a much higher rate of white crossover support for black candidates than is the case in local elections in those districts.76 When white voters in Senate Districts 2, 12, and 26 voted in statewide elections, between 23.8% and 58% cast their ballots for a black candidate over a white candidate. White crossover in statewide elections averaged 40.3% in Senate District 2, 41.8% in Senate District 12, and 43.3% in Senate District 26. At *125trial, Engstrom conceded that African-American candidates who receive this level of white crossover support have a “good chance” of winning election. See ante at 87. The Department challenges the relevance of crossover votes in statewide elections in those districts to Senate elections, arguing that white crossover voting in Senate races is more akin to voting patterns seen in local elections.

Engstrom’s analysis of state-level election results in each of the contested districts, particularly his re-aggregation analysis indicating the returns in each of the statewide elections as if they had been run in the proposed districts,77 is highly probative evidence of sufficiently high white crossover voting to enable candidates of choice to prevail. The lower rates of white crossover in local elections may be specific to those pockets of the Senate district, may be specific to the office,78 or may be less statistically accurate due to smaller number of precincts available for analysis.79 Any of these explanations are more plausible than a theory that white voters who refuse for racist reasons to support black candidates for local office would nonetheless vote to elect black candidates to more powerful and prestigious statewide positions.80

To the limited extent they are relevant, local election results tend to disprove the Department’s contention that levels of white crossover voting have been, and will continue to be, so low as to defeat qualified minority candidates in the contested Senate districts. In Savannah, the largest city in Senate District 2, African-Americans were a minority of registered voters during the 1995 and 1999 election cycles. See Shinhoster Dep. at 36. White crossover rates, according to Engstrom, were no greater than 10.6%. Nonetheless, African-Americans won two of the three runoff *126elections where they competed head-to-head against white candidates.81

In Senate District 12, African-American candidates won both elections to the state House included in Engstrom’s expert report.82 In the two Dougherty County races, one African-American candidate won, and the other lost.83 There is no record evidence that reveals the racial makeup of the House districts. Dougherty County was majority white in 1996, but the percentage of African-American residents is increasing. See Sherrod Dep. at 48; White Dep. at 16; D. Williams Dep. at 66. There is no evidence in the record to indicate whether the African-American candidates in the Albany mayoral races won or lost.

In Bibb County, the dominant county in Senate District 26, at least three African-Americans have won county-wide elections.84 Bibb County is majority-white, with approximately 43% BVAP and black voter registration of roughly 40%. See ante at 62. African-American candidates also won at least three of the four Macon city elections.85

These are not isolated victories, to be dismissed as aberrations, see ante at n. 45; they demonstrate that the relative lack of white crossover voting that Engstrom discerns in local elections is legally and practically insignificant. To the extent these local elections are worthy of consideration, they tend to support Georgia’s basic position: that African-Americans have a fair opportunity to elect candidates of their choice even in districts where the black voting age population or registered voters numbers fifty percent or less.

3. Fair Opportunity to Elect Minority Candidates of Choice

It is my view that § 5 does not prevent a state from adopting a redistricting plan, with the blessing of African-American legislators, that reduces “packed” concentra*127tions of black voters so long as it preserves equal or fair opportunities for minorities to elect candidates of choice.86 It may well be that super-majorities of black voters under the benchmark plan create “robust” opportunities to elect a candidate of choice. But under the law of unintended consequences, they may also create conditions that are “unfair,” “unreasonable,” and “unequal,” to both minority voters in those districts whose votes are “wasted,” to the point that they may find it unnecessary to turn out and vote, and to non-minority voters in those districts whose voting interests might well be “submerged” by the super-majority to the point that they turn away from the political process. Cf. Gingles, 478 U.S. at 68, 106 S.Ct. 2752. The Voting Right Act does not countenance, let alone require, such a result.

The Constitution and the Voting Rights Act do not guarantee victory to minority candidates, but only equal opportunity. A state’s maintenance of minority super-majorities within a particular district is required by § 5 only when necessitated by legally significant racially polarized voting, large numbers of ineligible minority voters, or other barriers to the effective exercise of electoral franchise that are outside the control of the minority group. There is “no vested right of a minority group to a majority of a particular magnitude unrelated to the provision of a reasonable opportunity to elect a representative.” Ketchum, 740 F.2d at 1418. Moreover, the continuation of super-majorities, even when progress has been made sufficient so that minority voters are no longer “fenced out of the electoral process,” Rybicki, 574 •F.Supp. at 1189 (dissenting opinion), and no longer need the edge those super-majorities provide, diminishes their opportunity to influence elections elsewhere and “threatens to carry us further from the goal of a political system in which race no longer matters.” De Grnndy, 512 U.S. at 1029, 114 S.Ct. 2647 (Kennedy, J., concurring). A proposed plan that provides a fair opportunity to elect the same or greater number of candidates of choice than the benchmark plan provides is entitled to § 5 preclearance.

4. Relative Weight Accorded to Expert and Lay Testimony

It is our responsibility, as the triers of fact, to determine the relevance, credibility, and proper weight of the evidence presented. This case need not be decided solely on the basis of expert testimony. As this trial has amply demonstrated, statistics is an inexact science, made more so by the “inherently uncertain” nature of voting behavior data, Easley v. Cromartie, 582 U.S. 234, 247, 121 S.Ct. 1452, 149 L.Ed.2d 430 (2001), and already-outdated census data, see Abrams, 521 U.S. at 100-101, 117 S.Ct. 1925. The Supreme Court has previously noted, in a related context, that results in another Georgia election “underscore the weakness of the Justice Department’s methodology of calculating the likelihood of a black-preferred candidate winning based on strict racial percentages.”87 Id. at 93, 117 S.Ct. 1925.

*128Courts have no obligation to accept statistical evidence as conclusive. See Magnolia Bar Ass’n, 994 F.2d at 1149 (“[T]he plaintiffs have not offered any authority, and we can find none, for their assertion that the district court may rely only on expert conclusions in determining whether white bloc voting is legally significant.”). This does not mean that statistical evidence should be rejected out of hand, particularly when it is weighty and uncontra-dicted by testimony from other experts or lay witnesses. “In the face of a strong statistical case ... general statements that race played no role at the polls carry little weight.” Teague v. Attala County, 92 F.3d 283, 291 (5th Cir.1996). While the statistics qua statistics in the record are meaningful, the expert opinions are flawed, contradictory, and highly varied in their consistency with the lay testimony and their own statistics; and the African-American legislators who testified in favor of the proposed plan rendered much more probative opinions.

Review of this record confirms the conclusions of other courts that the testimony of minority political leaders, particularly incumbents regarding their own districts, is highly probative evidence of the minority percentages necessary to ensure continued success in electing minority candidates of choice. See, e.g., Colleton County Council, No. 01-3581-10, slip op. at 110 (testimony of Congressman James Clyburn); Ketchum, 740 F.2d at 1415 (testimony of Congressman Harold Washington); Rybicki, 574 F.Supp. at 1114-1115 (crediting testimony of black aldermen that the inclusion of white neighborhoods in their wards would jeopardize their re-election chances); cf. Wright v. Rockefeller, 376 U.S. 52, 84 S.Ct. 603, 11 L.Ed.2d 512 (1964) (intervention of Congressman Adam Clayton Powell).88 I therefore accord great weight to Senator Brown’s support for his own district, as well as Senator Thomas’ concession that she will probably win re-election in her proposed district.89 More broadly, I consider Lewis, Brown, and Walker to be witnesses with unmatched knowledge of Georgia politics and African-American voting strength, with Senator Brown and Senator Walker being especially attuned to the level of minority voting strength necessary for minority candidates of choice to win a Senate seat. I consider Epstein’s statistics as reinforcement of this assessment of the political situation on the ground in Georgia, rather than the main support for the plaintiffs argument.

I have also considered the testimony of the Department’s lay witnesses, although I believe it pales in importance to the testimony of Lewis, Brown, and Walker and the expert witnesses. To the extent that discrepancies exist between declarations and depositions of the Department’s lay witnesses, I accept the latter as more credible, because it represents the witnesses’ own words, rather than the adoption of *129statements at least partially prepared by the counsel. The deposition testimony is also more comprehensive, and permits the witnesses to explain and elaborate on statements contained in the declarations.

B. Purpose of Georgia’s Senate Plan

I agree with the majority that the Senate’s purpose in advancing this redistricting plan is non-retrogressive. See ante at 96. Georgia’s legislators had a dual purpose: to maintain existing minority voting strength but avoid the “waste” of black votes, and to maintain a Democratic majority in both houses of the State Assembly. See Meggers Dep. at 20-21. The role that African-American legislators played in drafting redistricting plans for the House, Senate and Congress, and their near-unanimous support in voting for those plans, is a reliable indication that Georgia has no retrogressive purpose.

The desire to strengthen the position of one political party relative to the other is not a retrogressive purpose, although there are circumstances, not present here, where it might have a retrogressive effect. With respect to redistricting, the interests of the Democratic party and Georgia’s primarily Democratic voters are largely in tandem. Although that does not immunize the proposed redistricting plans from complying with § 5, it argues against a finding of discriminatory purpose.

III. Conclusion

I find that the testimony of Lewis, Brown, and Walker provides the State with a preponderance of evidence that outweighs the expert testimony and the testimony of the lay witnesses who testified for the Department and intervenors. Congressman Lewis has devoted his life, and risked it more than once, to advance the cause of African-American voting rights. He would not advocate a redistricting plan that would jeopardize what he has struggled so hard to win. It is inherently incredible that Senator Brown, Senator Walker, and all but one of the African-American members of the Georgia Senate would invite this court to place their Senate seats at serious risk, or cause minority voters — who, after all, are a-majority of their constituents — to lose an equal opportunity to elect the candidates of their choice. The judicially noticeable changes in the political landscape in the South in general, and Georgia in particular, from the days of “Massive Resistance” to the present, corroborate the expressed confidence of Georgia’s African-American legislative leaders that the steady rise in white voter crossover from zero to substantial numbers will continue. This confirms to me that the proposed redistricting of Senate Districts 2, 12, and 26 would not cause retrogression in the ability of African-American voters in those districts and statewide to exercise their franchise effectively.

I have read the majority opinion and concurrence with care and an open mind. However, I am not persuaded by Judge Sullivan’s comprehensive and well-written majority opinion or by the concurrence. Neither addresses four points I consider to be crucial to the resolution of this case.

First, a three-judge court is a trier of fact. As with a jury, judicial triers of fact may reach different conclusions about the probative value of items of evidence, including expert testimony.

Second, there is no persuasive response to my observations that the number of majority-minority districts, measured by BVAP — which I and most courts to have considered the issue find to be a highly probative gauge of minority voting strength — increases by one from the benchmark plan, and the number of minority choice candidates likely to be elected *130under the proposed plan is the same as under the benchmark plan.

Third, I have not discovered any legal authority in support of the majority proposition that § 5 requires that a plan preserve a “robust” (whatever that means), pre-existing probability that a minority candidate of choice prevail, and the majority cites none. Other courts have held that a plan that preserves or increases the number of districts where minority voters have a “fair” or “equal” opportunity to elect their candidates of choice are not retrogressive. See supra at 43^44.

Fourth, the notion that a trial court is bound by the arguments of counsel, which may be perceived in this case as principally concerned with racial polarization, will come as a great surprise to most trial court judges. We instruct jurors that the “[statements and arguments of the lawyers ... are not evidence.”90 Although it may be easier to focus our attention where the lawyers direct it, on the expert witnesses who testified live at trial, the voluminous written testimony from other witnesses is equally a part of the evidentiary record in this case. Our responsibility is to review the evidence in toto and assign it the weight it deserves.

Therefore, I respectfully dissent from the majority decision with respect to the Senate redistricting plan.

JUDGMENT

Pursuant to Federal Rule of Civil Procedure 58 and for the reasons stated by the court in its Opinion and Order docketed this same day, it is hereby

ORDERED and ADJUDGED that the Clerk shall enter final judgment in favor of plaintiff with respect to Georgia’s State House reapportionment plan, Act No. 2EX23, and Georgia’s United States Congressional reapportionment plan, Act No. 2EX11, and against defendants; and it is

FURTHER ORDERED and ADJUDGED that the Clerk shall enter final judgment in favor of defendants with respect to Georgia’s State Senate reapportionment plan, Act No. 1EX6, and against plaintiff.

IT IS SO ORDERED FOR THE THREE-JUDGE COURT.

. Standardized. Civil Jury Instructions for the District of Columbia at 3-3 (1998).

. Id. at 2-3.

. Id. at 2-8.

. Id. at 2-9.

.I use the plaintiff's methodology, which I found more convincing. See infra at n. 25. When the Department of Justice's methodology is employed, the number of districts in the proposed plan that are majority-minority as measured by BVAP falls to eleven. See ante at 56. According to Georgia, the BVAP in proposed Senate Districts 2 and 34 is 50.31% and 50.54%, respectively, while the Department of Justice calculates the BVAP at 49.81% in District 2 and 49.53% in District 34.

. Senate District 12 is represented by a white incumbent, Michael Meyer von Bremen, while the incumbent in Senate District 39, Vincent Fort, is African-American.

. The BVAP in Senate District 39 increases by 1.81% from the benchmark plan. Additionally, the BVAP in proposed Senate District 34, which represents a net addition to the number of majority-minority districts (by Georgia's calculations), increases by 16.58%. See Def.Ex. 118.

. The intervenors further argue that Congressman Lewis is biased because it is in his "best interest[] to have Democrats remain in control.” Tr. 2/26/02 at 158. The interve-nors misapprehend his testimony. As a member of the United States Congress, it is in his personal self-interest to have a Democratic majority in the House of Representatives, because it would place him (and other minority Members of Congress) in line to chair a House committee or subcommittee. See Lewis Dep. at 20. The Georgia Senate operates on a different political plane than the United States Congress, and Congressman Lewis realizes no personal benefit from a Democratic majority in the former. It is reasonable to infer that he would prefer control of the Georgia Senate by the Democratic party, as it is now constituted, because he considers it more responsive to the interests of his constituents than is its opposition.

. Statistics in this section are taken from the King's Ecological Inference methodology column in Def.Ex. 611, Table 1.

.The results of the four Democratic-only races disprove the hypothesis that some white voters are so politically partisan that they would support a black Democratic over a white Republican, but also so racist that they would never vote for a black Democrat if a white Democrat were running. Cf. Smith v. Allwright, 321 U.S. 649, 64 S.Ct. 757, 88 L.Ed. 987 (1944) (holding white-only primaries unlawful); Terry v. Adams, 345 U.S. 461, 73 S.Ct. 809, 97 L.Ed. 1152 (1953).

. The African-American candidate for Attorney General, Thurbert Baker, prevailed in that election and represents the state of Georgia in this litigation.

. Statistics in this section are taken from the King's Ecological Inference methodology column in Def.Ex. 611, Table 2.

. In the multi-candidate primary race for Labor Commissioner, the black candidate would have received 42.6% of the votes, more than any other candidate in the primary but less than a majority.

. John White received 17.1% of the vote when he ran for election to the House in 1994, predating the negative publicity he received in 1996 for forming a lobbying company to trade on his political connections. This suggests that 17% white crossover may be typical for candidates like White, and the 1996 Senate primary aberrational.

. The 1993 election is outside the normal scope of Engstrom’s analysis. In that race, a black candidate ran as a Republican and received 3.1% of white votes and 42.1% of the black vote. In that election, the white Democratic opponent was the candidate preferred by both black and white voters.

. Statistics in this section are taken from the King’s Ecological Inference methodology column in Def.Ex. 611, Table 3.

. Senator Thomas did not provide a declaration, but was deposed in person by the parties. See Def.Ex. 704.

. Proposed Senate District 2 would be heavily Democratic. Data from all of the precincts that will comprise the proposed district indicates that 64.41% of those voters supported Al Gore, 67.84% voted for Governor Barnes, and more than 70% supported the African-American Democratic candidates in the elections for statewide offices discussed supra. See Pi. Ex.2D. In a district so heavily tilted towards Democratic candidates, it may reasonably be inferred that Senator Thomas is likely to face a stiffer challenge in the primary than in the general election.

. Although two of the Department's witnesses from Senate District 12 complained Meyer von Bremen did not adequately represent the interests of the African-American community, a Clark University study based on the voting records of Senate members found that Meyer von Bremen consistently voted with the Senate Black Caucus. See United States’ Response to Plaintiff's Post-Trial Proposed Findings of Fact and Conclusions of Law ¶ 489.

. In the context of § 2, it is clear that the purpose of the Voting Rights Act is to provide *117an "equal” or "fair” opportunity. "The most natural reading of that language would suggest that citizens have an equal 'opportunity’ to participate in the electoral process and an equal 'opportunity' to elect representatives when they have been given the same free and open access to the ballot as other citizens and their votes have been properly counted. The section speaks in terms of an opportunity — a chance' — -to participate and to elect, not an assured ability to attain any particular result.” Holder v. Hall, 512 U.S. 874, 925, 114 S.Ct. 2581, 129 L.Ed.2d 687 (1994) (Thomas, J., concurring); see also Miller v. Johnson, 515 U.S. 900, 927, 115 S.Ct. 2475, 132 L.Ed.2d 762 (1995) (the Act seeks to provide "equal opportunity to gain public office regardless of race”); Voinovich v. Quilter, 507 U.S. 146, 155, 113 S.Ct. 1149, 122 L.Ed.2d 500 (1993) (“Only if the apportionment scheme has the effect of denying a protected class the equal opportunity to elect its candidate of choice does it violate § 2.”).

. Judge Harold Greene, as chief of the appellate section of the Civil Rights Division of the Department of Justice, was a principal draftsman of the Voting Rights Act as well as the Civil Rights Act of 1964.

. The three-judge court made a finding that a “Negro voting-age population of 60% was necessary in order for Negroes to have a fair opportunity to elect a candidate of their choice.” 444 U.S. at 1056, 100 S.Ct. 994 (emphasis added) (Marshall, J. dissenting). Justice Marshall did not disagree with the fair opportunity analysis, but thought the BVAP *118percentage necessary to provide a fair opportunity in rural counties was in excess of 65%. See id. at 1057, 100 S.Ct. 994.

. In the Bossier Parish decisions, the Supreme Court emphasized that a jurisdiction seeking preclearance under § 5 need not prove that its plan satisfy § 2’s more stringent vote dilution requirements. “To require a jurisdiction to litigate whether its proposed redistricting plan also has a dilutive ‘result’ before it can implement that plan — even if the Attorney General bears the burden of proving that ‘result’- — is to increase further the serious federalism costs already implicated by § 5.” Reno v. Bossier Parish Sch. Bd., 520 U.S. 471, 478, 117 S.Ct. 1491, 137 L.Ed.2d 730 (1997) (internal citation omitted) ("Bossier Parish I”). The Supreme Court did not consider the converse issue of whether a plan that is not vote-dilutive under § 2, because it preserves a fair or equal opportunity for minority candidates of choice to prevail, may nonetheless be retrogressive under § 5.

. It may be easier to define what retrogression is not. For example, § 5 does not require, as intervenors suggest, that retrogression be evaluated on a district-by-district as well as statewide basis, with any decrease in minority population or BVAP of an individual district deemed fatally retrogressive. See Ketchum, 740 F.2d at 1414 (rejecting a similar approach as "too inflexible an approach to the practical needs of redistricting”).

. Georgia and the Department of Justice have proposed two competing methods for calculating BVAP. With all due deference to the Department's guidelines, see Bossier Parish I, 520 U.S. at 483, 117 S.Ct. 1491, the Department's definition of black voting age population is unreasonable. For the first time, respondents to the 2000 census were permitted to identify themselves as belonging to more than one racial or ethnic group. The Department counts towards BVAP those who self-identify as black, or black and white, but excludes those who self-identify as black in combination with a racial or ethnic group other than white. Georgia includes all of those who self-identify as black, whether exclusively or in combination with any other racial or ethnic group, in its BVAP figures. Georgia presented unrefuted testimony from Dr. Roderick Joseph Harrison, former chief of the Racial Statistics Branch at the U.S. Census Bureau, that the Department's allocation rule is not justified as a matter of statistics, empirical evidence, or other considerations, such as communities of interest. See Pl.Ex. 26.

Resolution of this methodological dispute is critical to this analysis, because the exclusion or inclusion of the small group of citizens who provided multi-racial responses on their census forms directly impacts on whether Georgia's proposed Senate plan, in terms of the number of districts it creates with a majority BVAP, is numerically retrogressive or numerically ameliorative. I find that Georgia's method for calculating BVAP is superior to the Department’s.

. Testimony from witnesses resident in the contested Senate districts indicates that registration is a variable well-within the control of the African-American community. Community leaders in Savannah were able to register 5,000 new African-American voters. See Shinhoster Dep. at 36. Dougherty County has been targeted for a registration campaign that aims to increase the percentage of African-American voters by two to three percent. See White Dep. at 98, 100.

. The' fact that black voters in Senate District 12 may not be able to defeat the current white incumbent is not an indication of retrogression, because it represents only a continuation of the status quo — white incumbency. As another three-judge district court recently observed, "We must remember that the question is not what BVAP would be necessary to defeat a popular incumbent; the question is what BVAP is required to insure that the minority population has an equal opportunity to elect a minority candidate of choice in an open election.” Colleton County Council, No, 01-3581-10, slip op. at 101 (emphasis in original).

. Engstrom provided no data from Senate District 26, and the multiple elections in Senate Districts 2 and 12 each involved a single African-American candidate, making it impossible to discern whether the crossover rates in those elections were a race-based response or specific to those candidates.

. Plaintiff relies on a theory of “preference differentials” to argue that the voting patterns in statewide and local elections are not dissimilar. See ante at 87-88. Like Judge Sullivan, I am not persuaded by that theory, and find it tangential to our task in weighing the evidence, here by determining the relative value of statewide versus local elections as a predictor of racial voting patterns in the proposed Senate districts.

As an aside, I note that the differential percentages provided in footnote 46, ante, for both statewide and local elections, which range from 28% to 63%, are at a level that tends to disprove the existence of racially polarized voting. See Clarke v. Cincinnati, 40 F.3d 807, 816 (6th Cir.1994) (Boggs, J., concurring) (“One excellent measure of polarized voting or white bloc voting is the difference between the percentage of whites who vote for a given candidate and the percentage of blacks who vote for the same candidate. In the classically polarized races in most of the southern voting-rights cases, this figure has tended to be 80 percent or more for almost all candidates.”).

. Engstrom's re-aggregation analysis indicated that out of twenty-one elections, the African-American candidate won eighteen outright, received substantial pluralities in two primary races with multiple candidates, sufficient to advance to the runoff for the top two vote-getters, and lost a single election. See Def.Ex. 611 at 11-12.

. I note that, as a general rule, the lowest crossover rates are seen in elections to local executive office, in mayoral races or elections to the county board of commissioners.

. Engstrom’s data indicates that Floyd Adams, the incumbent African-American mayor of Savannah, received only 2.8% of white votes in the election and 8.7% of white votes in the runoff when he won in 1995. The Department has also provided testimony from local politicians and community leaders in Savannah, who testified that Adams in fact received support from roughly 20% of white voters in his 1995 election. See P. Jackson Dep. at 29 (“[W]e calculated white crossover to be somewhere in the neighborhood of 20 percent.”); H. Johnson Dep. at 36; Shinhoster Dep. at 41.

At the time Adams was elected in 1995, Savannah’s BVAP was less than 52% and whites were a majority of registered voters. See Shinhoster Dep. at 25, 36. This undercuts the Department’s statistical evidence in two possible ways: either Adams in fact received 20% of the white crossover vote and Engstrom's local election data is statistically questionable, or Adams was able to win election in a city where black voters were a numerical minority with minimal white crossover.

.A comparison of the 16.6% white crossover rate in the district attorney election in Bibb County, and the 46.9% white crossover rate in the election for state Attorney General is illustrative. It may be that there is a cohort of white voters in Georgia who will not support a black candidate for county district attorney simply because they are racists. However, it is implausible to assume that a significant portion of those white racists would turn around and support a black candidate for state Attorney General.

. Floyd Adams won the mayor's race in 1995. Edna Jackson lost the election for an at-large city council seat in 1995, but won in 1999. See E. Jackson Dep. at 8. Interestingly, Engstrom’s report indicates that she received an identical percentage of white votes in both races, meaning that her victory in 1999 was the likely result of relatively higher African-American turnout, and did not depend on a shift in the white voting preferences. See Def.Ex. 611 at 3-4 (Savannah Council At-Large, 1995 and 1999 Runoffs).

. White testified that he won his 1994 election to the state House with more than 70% of the vote. White Dep. at 63. I take judicial notice that Lawrence R. Roberts, the African-American candidate in House District 162, has served in the State House since 1992. See <http://www. legis.state.ga.us/Legis/1995— 96lhouselgahml62.htm > (visited March 22, 2002).

. Jane Taylor won the coroner’s race and Wright lost the election to chair the county commission. See White Dep. at 18; Wright Dep. at 14. Wright’s support among African-American voters was relatively soft, at 72.2%. In contrast, Taylor was supported by 96% of black voters. See Def.Ex. 611 at 7.

. Two of the Department's witnesses, Abrams and Barnes, currently serve on the Bibb County Board of Education and defeated white candidates to win election. Engstrom's data indicates that a third black candidate, Hutchings, received 34.2% of the white vote in his 1994 election. African-Americans therefore won two out of four elections analyzed in Engstrom's report, and won an additional countywide election not included in the data set.

. Jack Ellis, Macon’s incumbent mayor, won the 1999 primary by six percentage points, receiving 10.4% of the white crossover voter. (Ellis Decl. ¶ 2). I take judicial notice that Brenda Youmas was elected to the City Council in 1995 and that James Timley prevailed in his 1999 council race. <http://www.cityofma-con.net/CityDept/council/members.htm> (visited March 22, 2002). There is no evidence to indicate whether the African-American candidate in the fourth Macon race won or lost.

. Indeed, if Georgia had maintained the heavy concentrations of African-American voters in certain of its Senate and House districts, particularly in the metropolitan Atlanta area, black voters in those districts may have a had a cognizable § 2 claim based on dilution of their votes through packing. See Quilter, 507 U.S. at 154, 113 S.Ct. 1149.

. In Abrams, the Justice Department had predicted that a minority candidate of choice would not be able to prevail in Georgia’s Fourth Congressional district, which had a BVAP of 33%. Cynthia McKinney, who is African-American, nonetheless won the election.

. The Rybicki Court speculated that a three-judge district court and the Supreme Court may have upheld a redistricting plan that divided Manhattan into three heavily white Congressional districts and one heavily minority Congressional district because Powell, the African-American incumbent, intervened as a defendant in support of the proposed plan. See Rybicki, 574 F.Supp. at 1118 n. 97.

. Based on Senator Thomas' testimony, I infer that her opposition to proposed Senate District 2 is based less on retrogression, and more on garden-variety political concerns: the proposed plan impinges on her political turf in Chatham County by bringing in a third senator, and makes it likely that she will have to campaign harder than in her- most recent election, when she received more than 77% of the vote. Her objections are legitimate and understandable from a political standpoint, but do not implicate the Voting Rights Act.

. Standardized Civil Jury Instructions for the District of Columbia at 2-5.