[EDITOR'S NOTE: PART 2 OF 2. THIS DOCUMENT HAS BEEN SPLIT INTO MULTIPLE PARTS ON LEXIS TO ACCOMMODATE ITS LARGE SIZE.]
221. The first stage of Professor Estrada's fragmentation analysis was to characterize each census block in the City by its racial or ethnic population majority. Each block was then designated according to its assignment in a white, African-American, Latino, or no voting age population majority under both the Ketchum and the present ward maps. According to this method, under the present ward map: 90% of whites (92% if the 18th ward is viewed as white), 95% of African-Americans (92% if the 18th ward is white), and 74% of the total population of Latinos reside in wards where they comprise the majority voting age population. Under the Ketchum settlement map, as of the 1990 census, only 49.8% of all Latinos were in Latino VAP majority wards. The present ward map, therefore, effects a significant diminution of Latino fragmentation.
222. Professor Estrada's analysis failed to limit itself to legally relevant fragmentation. Given his method, even census blocks which could not be located in a Latino majority ward, because of their remote location from Latino population cores, were counted as fragmented. This runs counter to his own definition of fragmentation. His fragmentation analysis addressed this defect by subjecting the Ketchum map and the Bonilla alternative maps to the same fragmentation analysis. The Bonilla alternative maps result in an average of 86% (86.2% for the first alternative map and 85.5% for the second alternative map) of all Latinos being included in Latino VAP majority wards. Apparently, therefore, since the Bonilla alternative maps were drawn with the intent of placing as many Latinos in Latino majority wards, the majority of the "fragmented" Latino majority census blocks are in areas of the City where they could not be included in Latino majority wards. It is possible to correct a portion of Latino fragmentation as evidenced by the alternative maps. Even under the alternative plans, however, Latinos would remain the most fragmented group in the City. This structural fragmentation is due to the fact that only approximately 61% of the City's Latino community lives in Latino majority census blocks.
223. More telling to the identification of legally relevant fragmentation, is Professor Estrada's analysis of Latino fragmentation on the south side. Under the present ward map, the south side Latino majority wards 12, 22, and 25 all have Latino total population majorities of at least 73%. Under the present ward map, there are also sizeable Latino total population blocks in the adjoining wards: the 11th ward is 28% Latino, the 14th ward is 35% Latino, the 15th ward is 12% Latino, and the 16th ward is 21% Latino. The bulk of this Latino population resides in what Professor Estrada termed the "south side growth area" which encompasses the Back of the Yards, Marquette Park and Gage Park neighborhoods. No other identifiable or similarly-sized population group was divided into as many wards.
224. In order to correct this fragmentation, and to create an additional Latino ward, it is necessary to draw some problematic wards -- in the first Bonilla alternative map, the reconstituted 14th ward which encompasses 97.2% of the "south side growth area" and is 70.03% Latino stretches from approximately 16th Street on the north to 63rd Street on the south. The 14th ward in the second alternative map is much more compact, comprising approximately 98.8% of the south side growth area, however, it has a much smaller Latino majority of only 57.33% TP and a VAP majority of only 51.16%. Thus, while the fragmentation of Latinos on the south and southwest side can be significantly ameliorated by the drawing of an additional Latino majority ward -- it can, apparently, be corrected only by drawing a ward with a bare Latino majority or by drawing a highly elongated Latino majority ward. Nonetheless, at this stage of the analysis, it appears that it is at least theoretically possible to moderate the fragmentation of Latinos on the southwest side of the City.
d) Methods for Estimating Latino Citizenship Rates
225. While this Court ultimately rejects defendants' contention that citizenship rates are relevant to the Bonilla plaintiffs' satisfaction of the first Gingles prong, it is nonetheless worth discussing the competing methodologies for estimating citizenship. Because of the flaws of the offered methods of determining CVAP, and the methodological quagmire in which courts would find themselves if they were to adopt CVAP as a benchmark, this Court finds that VAP is a more adequate indicator of the effectiveness of the alternative Latino wards.
226. Part of the difficulty in using citizenship data lies in the manner in which it was collected by the Census Bureau. Citizenship data was derived from sample data compiled at the block group level. Block groups are clusters of blocks containing between 250 and 550 housing units. Ward boundaries are based upon census blocks rather than block groups. Ward boundaries do not necessarily correspond to the block group boundaries. If ward boundaries indeed corresponded to the boundaries of the census block groups, CVAP might have been a valid population benchmark.
227. Nor was the citizenship data derived from full count data. Only one of every eight households in Chicago received the long form questionnaire from which citizenship data was derived. As with any sample data, there was necessarily the danger of sampling error and, therefore, it is best used for inferring general characteristics of large population groups and it is less reliable when applied to smaller population measures, such as census blocks.
228. Professor Estrada's method of estimation involved the application of the sample data for citizenship to those portions of an actual ward which contained a particular block group. In short, Professor Estrada applied the citizenship rate of a particular block group to the actual full count population of the portion of the ward which was contained in the block group straddling ward boundaries. This method had the benefit of using the actual, full-count census data for total and voting age population and using sample data only for the citizenship estimate. This method also had the benefit of establishing a citizenship estimate for an actual ward.
229. Professor Rives, on the other hand, relied upon sample data to determine voting-age and total population even though it was possible to obtain that information from the full-count data. (DX 509,866). This use of sample data, when full data was available, had the effect of understating Latino voting age population by approximately 1%.
231. Professor Estrada's method for estimating citizenship itself contained some problematic assumptions. His method assumed that each block within a block group had the same citizenship rate. This is precisely the same assumption, but on a smaller level, upon which the sample data is based -- that deviations of this nature will tend to cancel each other out throughout the statistical sample. The danger of citizenship rate variations across block groups points to the problematic applicability of citizenship rates at the census block level, the data is increasingly unreliable as it is applied to a decreasing population base. Applied to the City as a whole the sample data is reasonably reliable, but at the ward or block level it is not. Sample data is by its nature subject to sampling error, requiring an analysis of point estimates and confidence intervals for the citizenship estimates, statistical devices which courts are notoriously ill-equipped to handle. Not only is this Court uncomfortable with the notion of applying statistical methodologies with which lawyers and courts are often less than adroit, it is equally hesitant to rely upon estimates that are intrinsically inaccurate and subject to various sampling errors.
232. The redistricting process is intrinsically political, and for all the testimony concerning statistical sciences which this Court heard at trial, it is not convinced that the participants in the process were less cognizant of the population needed to constitute an effective ward than the various experts, even though they did not have the benefit of the citizenship sample data.
233. It is also inappropriate to use citizenship as a population benchmark for VRA § 2 because that data, besides being subject to sampling error, was not even available at the time the ward maps were being prepared. The redistricting process is difficult and drawn out enough as it is without forcing the persons drawing a ward map to perform a metaphysical impossibility and to use citizenship population figures that were not yet even available when the ward map was being prepared.
234. It is unclear in any event that CVAP, rather than VAP, would be a more accurate indicator of the effectiveness of a ward, since by Professor Rives' estimation method, the present 12th ward does not have a Latino CVAP majority, yet it operates as an effective Latino majority ward.
3. Professor Allan Lichtman
235. Professor Allan Lichtman, a professor of History at American University, testified on behalf of the Bonilla plaintiffs concerning: 1) polarized voting between Latinos and non-Latinos in the City of Chicago; 2) the effect of racially polarized voting on the capacity of the Latino electorate to elect candidates of their choice; 3) voter registration and turnout levels; and 4) the percentage of Latino voters necessary to give Latino voters an opportunity to elect their candidates of choice in aldermanic elections. (BOPX 74, 76, 180).
a) Latino Political Cohesiveness and Polarized Voting
236. Professor Lichtman analyzed 28 elections in an effort to determine the extent to which Latinos tend to vote for Latino candidates and the extent to which non-Latinos tend to vote for non-Latino candidates (and correspondingly the extent of any crossover voting for Latino candidates by non-Latinos). This analysis pertains to the second and third Gingles preconditions -- that is, whether minority voters are politically cohesive, and whether the white majority votes sufficiently as a bloc to defeat usually the minority's preferred candidate.
238. Professor Lichtman's analysis was based solely upon data retrieved from election returns and census data. All the voting behavior experts relied upon the same information, in fact the parties created a joint computer base containing this information. Defendants criticize Professor Lichtman because he did not attempt to go "behind the numbers" to account for such variables as a candidate's gravitas, charisma, funding, experience, incumbency or endorsements. The relevant issues to Professor Lichtman's analysis were simply whether minority voters are cohesive and whether whites vote as a bloc to defeat the minority candidate of choice. If plaintiffs establish evidence of a pattern of polarized voting, defendants may attempt to overcome the inference of racially polarized voting by presenting evidence that whites vote as a bloc for reasons other than race or that minority candidates have lacked whatever it takes to prevail in the normal course of politics. UNO, 72 F.3d 973, 981. Defendants, however, have not come forward with enough evidence to disprove the historical trend of polarized voting established by Professor Lichtman.
239. While this Court is somewhat perplexed by Professor Lichtman's exclusion of non multi-ethnic elections, it nonetheless concludes that there is adequate evidence of polarized voting between Latinos and whites in the City of Chicago to satisfy the second and third Gingles prongs. By examining only Latino versus white elections, Dr. Lichtman apparently presumed that there was no identifiable Latino-preferred candidate in those elections. It is troubling to suppose that Latino voters could never prefer or be adequately represented by non-Latino candidates. The opportunity to participate in the political process is not reducible to being ensured of electing candidates of one's own race or ethnicity. Focusing on multi-ethnic elections, however, highlights the evidence of polarized voting. See, LULAC v. Clemens, 999 F.2d 831, 864 (5th Cir. 1993), cert. denied, 114 S. Ct. 878 (1994).
240. Looking only at multi-ethnic elections has flaws, however. Where minority voters supported a particular white candidate, and white voters supported another white candidate that information could be relevant to the vote dilution analysis. Professor Lichtman's analytical method failed to recognize that communities within this City might develop multi-ethnic or multi-racial coalitions, in which social or economic interests rather than racial or ethnic factors dominate an electoral choice. By neglecting to look at elections in multi-ethnic wards which did not have elections with multi-ethnic candidates, Professor Lichtman's analysis failed to account for the possibility that voters could coalesce around issues other than race or ethnicity.
241. While Professor Lichtman's analysis was not entirely complete because it failed to consider non multi-ethnic elections, it still established that polarized voting exists within the City of Chicago. Since an issue in this case is whether the VRA requires the drawing of additional Latino majority wards which would have lower Latino majorities than the present wards, the outcome of multi-ethnic elections is relevant to this Court.
243. Professor Lichtman's analysis of polarized voting combined candidates by race or ethnicity. Rather than analyzing the level of support for each candidate, Professor Lichtman looked at the level of support for Latino candidates and non-Latino candidates as a group. In elections with multiple candidates, Professor Lichtman did not consider the level of support for any individual candidate. This methodology presented the polarization issue in its clearest terms and allows for the identification of electoral polarization where Latinos tend to vote predominantly for Latinos and non-Latinos for non-Latinos. As the leading scholarly literature has observed, the failure to look at the combined candidate data often results in a misleading picture of minority political cohesion. Grofman, Handley, Niemi, Minority Representation and the Quest for Voting Equality, at 99. The practice of combining candidates permits a clear view of the level of racially or ethnically polarized voting. It is, however, independently instructive to look at the election data for individual candidates in order to identify the extent to which minority voters have coalesced around a particular candidate of choice. In the end, however, requiring minority voters to coalesce around a single minority candidate in order to demonstrate their cohesiveness, when they have clearly demonstrated their preference to vote for minority candidates would be a strange way of encouraging minorities to participate in the political process.
244. Professor Lichtman made extensive use of exogenous elections, that is elections for other offices, in his analysis. Professor Lichtman himself recognized that endogenous (here, aldermanic) elections are generally preferred. Endogenous elections have the benefit of having a similar pool of voters, of having the same size pool of voters, and of having the same election rules -- all of which help ensure the reliability of the analysis. The pool of multi-racial or ethnic aldermanic elections, which satisfied Professor Lichtman's criteria was very small -- consisting of only 6 elections.
Professor Lichtman, therefore, also reviewed the results of non-aldermanic elections including judicial elections, democratic ward committeeman elections, and democratic primaries for city clerk, state representative, county commissioners, congressional representatives, and city treasurer. The same general pool of candidates is eligible for the non-partisan aldermanic elections as for the democratic primaries since more than 90% of the Chicago electorate is Democratic. (Tr. 3042). The review of additional elections, in this instance, also permitted a broader indication of the existence of polarized voting since aldermanic races in the Latino majority wards have not tended to be competitive insofar as competitive white candidates have tended not to run in these wards. The absence of competitive multi-ethnic elections races in the Latino majority wards skews the polarization analysis, while the more competitive non-aldermanic elections better indicate the extent to which polarized voting takes place in elections with significant candidates. With four exceptions, Latino candidates have prevailed only in Latino-majority districts. (Tr. 3102).
246. Additionally, while the pool of the eligible electorate might be the same for democratic primaries as for non-partisan aldermanic races, it is equally likely that the character, education or motive of the voters who vote for less visible offices in party primaries will likely be different in many ways from voters in local aldermanic elections.
247. The pattern of polarization for non-aldermanic and aldermanic elections differed, but aldermanic elections were still for the most part polarized (other than the 1991 aldermanic election in the 33rd ward in which Alderman Mell was the Latino candidate of choice), although "Anglos" and Latinos tended to vote more similarly in aldermanic elections. This phenomenon can likely be explained by the large Latino supermajorities in the Latino majority wards -- assuming the validity of an explicitly racial calculus, it would make more sense for non-Latino voters to throw their support behind a palatable Latino candidate than to mount what would likely be a futile non-Latino candidacy. The one aldermanic election with a clear showing of racially polarized voting took place in the 10th ward in 1991, a ward that had a Latino plurality rather than a supermajority.
Even defendants' polarized voting expert, Professor Weber, analysis looking solely at aldermanic elections revealed some polarized voting. (DX 506). The exogenous elections merely strengthened the ambiguity of the polarization finding derived from endogenous elections. Thus even though there were significant problems with Professor Lichtman's selection of elections, those problems discredited only a portion of his analysis but did not discredit the central finding that in competitive elections there is polarized voting between Latinos and "Anglo" voters and there is polarized voting, albeit less extreme, between Latinos and African-Americans.
248. While there have been many elections in which Latino candidates have won, those elections have generally been in Latino VAP supermajority districts. Where Latinos have won in other districts, they have often won due to the effects of multiple-winner election rules or because of a fractured field of candidates. (Tr. 3100, 3102). There have been two elections in which Latino candidates have prevailed in non-Latino districts with a majority of the votes, namely in the 1991 and 1995 Democratic primaries for City Treasurer. Also of some relevance in those elections was the fact that Miriam Santos was the preferred candidate of the Democratic Party leadership, which likely contributed to the level of white support she received.
249. Professor Lichtman relied upon three chief forms of statistical methodology to analyze voter behavior: multi-variate ecological regression, extreme case analysis, and an examination of the actual outcomes of elections. Professor Lichtman also relied upon the R2 values (or squared-correlation coefficient) to assess the reliability of his regression analysis.
250. Courts, and respected experts in the field have been reluctant to accord much value to multi-variate regression analysis which is prone to extensive manipulation. Grofman, Handley, Niemi, Minority Representation and the Quest for Voting Equality, at 100. One accepted use of multi-variate regression analysis is to analyze voter behavior in areas with multiple racial or ethnic groups (as is the situation in Chicago in which there are Latino, African-American, and white residents in many wards) and in which there are also homogenous precincts. Multi-variate regression permits the simultaneous analysis of the voting behavior of all three relevant racial or ethnic groups.
251. By adding a third variable, however, the complexity of the regression equation is increased geometrically. Using trivariate regression it is no longer as readily possible to make use of scatterplots of the regression equation in order to double-check the neatness of fit of the regression line. Nor is it as easy to identify "outliers", that is, individual precincts in which voters behave outside of the trend of the other voters and which might, by their aberrant result, affect the average outcome of the regression equation.
252. Based on his analysis, Professor Lichtman found that Latino voters tended to vote for Latino candidates at an overall average of 86%. The mean "Anglo" (which meant all non-Latino and non-African American voters) level of support for Latino candidates was 33%, meaning that 67% of "Anglo" voters did not support Latino candidates. The mean African-American level of support for Latino candidates was 55%. There were, however, several elections in which there were significant exceptions to this average, for example in the 1991 aldermanic election in the 33rd ward, Alderman Mell received the plurality of Latino votes, and in the 1991 and 1995 Democratic primaries for the City Treasurer, a Latino candidate, Miriam Santos, received the majority of "Anglo" and Latino votes, additionally in several Democratic primaries for County Commissioner and Sanitary Commissioner, Latino candidates received the majority of "Anglo" votes.
b) Latino Majority Necessary to Constitute an Effective Ward
253. Although the parties disagree on the possible cause, they concur that the registration and turnout rates of Latinos are significantly lower than those of whites. Professor Lichtman's analysis of turnout rates, however, indicated that the lower Latino turnout cannot be attributed solely to citizenship. According to Professor Lichtman's data, in the south side wards, the Latino turnout in terms of VAP is frequently one half to a quarter of the turnout rate of whites. On the north side, the differential in turnout rates is generally not as large.
254. This Court is not satisfied that Professor Lichtman's analysis supports his conclusion that the Bonilla alternative wards would be effective Latino wards. The lowest Latino VAP majority in any of the current aldermanic wards is 59.6% in the 35th ward; all of the other aldermanic wards have VAP majorities of at least 62.7%. The state representative districts all have VAP majorities of at least approximately 65%. The county commissioner districts have Latino VAP majorities of approximately 56.7%. Additionally, in the primary elections analyzed for these exogenous offices, it is not necessary to obtain a majority vote to prevail. Only the judicial districts have a Latino VAP majority lower than a sizeable majority with a Latino VAP majority of approximately 50.8% in one of the districts. These judicial elections, however, have multiple winners making it easier for Latino candidates to prevail. No aldermanic election has ever been won by a Latino candidate with less than a 59.5% VAP majority.
255. Under the Bonilla alternative maps, the southwest side alternative wards would have lower Latino VAP majorities than the present southwest side Latino majority wards. Under the current ward maps, the Latino majority wards have sizeable Latino VAP supermajorities (the 12th ward has a VAP majority of 65.44%, the 22nd ward has a VAP majority of 74.49%, and the 25th ward has a VAP majority of 68.153%). Under Bonilla alternative map I the Latino VAP majorities are lower (12th ward-- VAP majority of 63.35%, 14th ward -- VAP majority of 64.23%, 22nd ward--VAP majority of 65.10%, 25th ward -- VAP majority of 64.71%). The margins are even smaller in some of the Bonilla alternative II maps (12th ward -- 66.05% Latino VAP majority, 14th ward -- 51.16% Latino VAP majority, 22nd ward -- 68.14% Latino VAP majority, and 25th ward -- 71.90% Latino VAP majority).
256. By the calculations provided by Bonilla counsel, guided by Professor Lichtman's estimates, it would be possible that Latino-preferred candidates could prevail in the alternative southwest side wards, although by those calculations these elections would be extremely close. Given the very close margin for error, it is not difficult to discern why representatives of the Latino community preferred that there be three Latino supermajority wards on the southwest side with high supermajorities in order to cement Latino political gains for the present time rather than seek four wards in which electoral success was not assured.
257. Dr. James Lewis, Vice-President for Research and Planning for the Chicago Urban League, testified on behalf of the Barnett plaintiffs concerning: 1) racial divisions in recent City Council votes, and 2) disparities in the fund-raising opportunities available to African-American and white candidates. (BAPX 109). In his professional capacity, Dr. Lewis has extensive experience at conducting and overseeing research on public policy issues and, despite defendants' contentions to the contrary, was qualified to testify as an expert for the purpose of assembling and analyzing the public records which were compiled in his report and rendering an opinion regarding the impact of racial divisions on City Council votes and the financing of aldermanic elections.
258. While Dr. Lewis is indeed qualified to testify as an expert regarding public policy concerns, his testimony was not entirely credible. This Court was particularly troubled by Dr. Lewis' repeated reliance upon head signals from counsel during his cross-examination. (Tr. 2345).
a) City Council Voting and Race
259. In total, Dr. Lewis' testimony was not very instructive. The methodology and selection criteria in his analysis of city council votes left much to be desired. Dr. Lewis' analysis of city council voting patterns consisted of three steps: 1) he examined public opinion by selecting several survey questions from the Metro Chicago Information Center expressing public opinion on a variety of public policy questions; 2) he selected several City Council votes which he believed showed racial divisions on important public policy issues; and 3) he attempted to correlate public opinion to the City Council votes. Each of these steps contained choices which Dr. Lewis could not explain to an extent precluding this Court from giving great credence to his analysis. Nonetheless, Dr. Lewis' report and testimony does tend to indicate that as a general matter there are slight, general differences of opinion concerning issues such as housing, health services and affirmative action and fair housing, between the African-American and white communities in Chicago which are reflected in City Council votes. Dr. Lewis' report and testimony established these divergences in such a cursory and superficial manner that this Court can give Dr. Lewis' findings only little weight.
261. In many cases, the differences in the survey answers were not significant -- for example: 72% of whites and 92% of African-Americans believed that government spending on health care is very important; 64% of whites and 82% of African-American favored affirmative action set asides for college or employment; or 23% of whites and 20% of African-Americans have gone to a legal casino in the last year while African-Americans tended to spend twice as much as whites on the lottery each week. Reliance upon survey responses such as those reduced Dr. Lewis's conclusions to nothing more than a statement of superficial generalities.
262. The problematic selection criteria used by Dr. Lewis point to the chief problem with this section of Dr. Lewis' report -- his entire report presumed that any difference in white and African-American responses to the survey questions were necessarily due to racial differences. Given Dr, Lewis' selection of particular survey questions in isolation this Court must query whether in their entirety, the differences in the survey responses were due to non-racial factors such as income or religious affiliation. By taking survey responses in isolation, Dr. Lewis presupposed that his hypothesis was correct and selected survey questions which tended to support his position without accounting for other factors which may or may have not supported his views.
Therefore, this Court concludes that, while the report indicated that there were some differences in the opinions of the white and African-American communities concerning housing, education, fair housing issues, affirmative action and health services, the extent to which these differences can indeed be attributed to divergent communities of interest is not at all clear.
263. The second step in Dr. Lewis' analysis, the analysis of City Council votes, was equally troubling. Once again, Dr. Lewis apparently selected data that would prove his hypothesis, disregarding, without any justification, data that would not support his theories. Nor was the procedural posture or the eventual outcome of any particular piece of legislation plain from Dr. Lewis' report and testimony -- for example, there were apparently several South African disinvestment ordinances contemporaneously before the City Council and the support for a more moderate proposal by white aldermen did not necessarily indicate racial division in the City Council. It was impossible to tell how often there were similar situations in which there were competing pieces of legislation before the Council one of which was viewed as an administration proposal and one which was viewed as an independent proposal, or one of which might require modifications to the budget or other appropriations.
264. Dr. Lewis chose 12 votes which took place between 1989 and 1994, leaving this Court to wonder how many other votes were held which might also have been relevant. Many of the votes referred to by Dr. Lewis were procedural votes, and Dr. Lewis could not identify the final outcome on these substantive issues. Two of the Council votes, South African disinvestment and the establishment of a Fred Hampton Day, did not have any relation to the MCIC Survey data results but were selected for their symbolic value. Fred Hampton was also a political and social lightning rod about whom well-intentioned people could have very different opinions.
While the votes selected by Dr. Lewis indicated that at least 12 City Council votes between 1989 and 1994 broke down along racial lines, the report did not lead to a broader conclusion than that there were 12 divided votes concerning various issues in the City Council.
265. The final step of Dr. Lewis' analysis was to attempt to match the MCIC survey data to City Council votes. The match between the two was often not entirely clear. The purpose of this analysis was to show that the votes of African-American aldermen tended to track the opinions of African-American community more closely than the votes of white aldermen. There were no survey questions related to Fred Hampton Day or South African disinvestment and with respect to these two votes, Dr. Lewis simply intuited the interests of the African-American and white communities. Dr. Lewis' conclusions in this phase of his analysis were beset by the difficulty of tracking the generalized interests identified in the survey data to particular pieces of legislation. For example, most respondents to the survey stated that it was very important to spend state/local taxes on health (72% of whites and 92% of African-Americans responded "yes") yet it was not clear how those opinions related to a resolution "urging the Department of Health to postpone negotiations, contracts or studies affecting privatization of public health services." Professor Lewis' report is rife such grapes and raisins comparisons.
266. For example, it was unclear how the survey response indicating that more African-Americans than whites (32% over 13%) fell behind on housing or utility bills related to the extension of Commonwealth Edison's utility license agreement. Nor was it obvious how a survey question asking why a person has been unable to find work (to which 6% of whites and 13% of African-Americans responded that they were not working because they could not find work) related to a proposed ordinance which would have required a public hearing before waiving the City residency requirement on City contracts.
267. In summary, Dr. Lewis' report did not lend itself to anything. more than the most general conclusions concerning differences in opinion on public issues between the African-American and white communities and the corresponding policy differences in the City Council. Dr. Lewis set out to find racial cleavages and he evidently disregarded any information which did not support his conclusions. The report and testimony was filled with oversimplifications and generalities which often approached characterization and stereotyping. Dr. Lewis's report and testimony was little more than anecdotal evidence dressed up in the garb of expert opinion.
b) Campaign Fund Raising and Race
268. Nor did Dr. Lewis' analysis of disparities in access to campaign financing permit anything more than the most generalized and superficial of conclusions. Once again, his report amounted to a type of stereotyping or self-fulfilling prophecy insofar as he disregarded any other factors which may have had an impact of fund-raising by a candidate such as incumbency or the character of the ward.
269. Dr. Lewis' compilation of fund-raising records made no effort to account for several factors which the data itself indicated had a serious impact on fund-raising abilities -- incumbency and the seriousness of the candidate. While the data indicated that on average white aldermanic candidates raised more than African-American candidates, Dr. Lewis' report made no effort to go beyond the most superficial statement of that fact. There was no attempt to separate out candidates who were not serious contenders. For example, in 1991 several of the African-American wards had elections with as many as 5 candidates running against an incumbent -- in each of these elections the incumbent raised substantially more than any of the other candidates. According to Dr. Lewis' data, African-American incumbents generally, with a few exceptions, raised as much as their white counterparts. Incumbency is a particularly important consideration in fund-raising abilities: as an incumbent in 1995 Alderman Murphy of the 18th ward raised twice as much as he had four years previously. Another indicator of the importance of incumbency is found in the fund-raising figures for Alderman Streeter of the 17th ward, in 1991 when facing no opponents who had submitted financial disclosures, he raised only $ 16,373, while in 1995 facing several adversaries he raised more than twice as much. Dr. Lewis' report and testimony made no effort to account for the effect of incumbency on fund-raising ability.
270. The financial disclosure statements from the 1995 18th ward aldermanic election are of some interest. This was the only election included in Dr. Lewis' report in which a white candidate ran against African-American candidates. In that election the incumbent white candidate raised nearly 10 times as much as the combined total for the three African-American candidates. How much of this disparity was due to incumbency, or the skill of the candidates and how much was due to racial disparities in fund-raising access was left to speculation but both factors were likely at play. This Court is leery, however, of making broad conclusions based upon one election.
271. While there are many non-racial reasons for disparities in fund-raising ability that were not touched upon in Dr. Lewis' report or testimony, on average, African-American candidates tended to raise less than their white counterparts. Dr. Lewis' data also included, though, several African-American aldermen who, for whatever reason, were able to raise substantially more than any other aldermen, white or black. This indicates that fund-raising ability is not entirely race-specific.
272. Dr. D. Garth Taylor was called by the Barnett plaintiffs to present expert testimony concerning the differing communities of interest between the white and African-American communities. He is the executive director of the Metro Chicago Information Center ("MCIC"), a non-profit research organization. His report and testimony focussed on several areas: 1) an examination of city-wide patterns of per-capita income, public school enrollment, families in poverty and other socio-economic indicators which Dr. Taylor testified are closely correlated to race; 2) his historical discussion of racial hostility by whites in the 18th and 19th wards on the southwest side of the City, 3) a socioeconomic analysis focusing on the southwest and southeast sides, and 4) his city-wide canvassing of the divergences of African-American and white opinion on a spectrum of public policy issues including the ward map referendum and an analysis of the incidence of hate crimes in the City. (BAPX 78-107). The bulk of Dr. Taylor's analysis was presented in the form of maps illustrating the various socio-economic indicators and public opinions selected by Dr. Taylor.
a) City-wide Socio-economic Communities of Interest
273. This Court's criticism of Dr. Taylor's analysis of city-wide socioeconomic data is similar to that of Dr. Lewis' analysis. Dr. Taylor's analysis begs the question of the correlation between the socio-economic data he selected and race thanks to Dr. Taylor's selection of data which only proved his hypothesis. Again, as was the case with Dr. Lewis' analysis, Dr. Taylor's opinions merely stated the obvious in a manner placing an expert patina on what was essentially subjective or anecdotal evidence.
274. Dr. Taylor's analytical method consisted of mapping socio-economic data derived from either census data or the MCIC survey and then attempting to draw conclusions concerning the correlation between race and communities of interest based upon the patterns revealed on the maps. The chief defect of Dr. Taylor's city-wide socioeconomic analysis was in his selection of the parameters for his illustrative charts. The break-points in his illustrative maps were puzzling and were never adequately explained. His city-wide socio-economic maps merely established that based upon per capita income, percent public school enrollment, percent single parents with children, and percent of families in poverty with children there were differing core socio-economic communities of interest with poorer areas centered around the west-side and the center of the south side, and more affluent areas in the periphery of the City with the boundary areas being somewhere in the middle of the two extremes.
276. Despite these difficulties, Dr. Taylor's analysis still demonstrated that there are differences, at the extremes, between the two communities. It is nearly impossible, given Dr. Taylor's extremely selective use of the data, to determine precisely how pervasive those differences are.
b) Communities of Interest in the Southwest and Southeast Sides
277. The first phase of Dr. Taylor's analysis of the southwest side was a non-quantitative analysis of historical events concerning housing and education in the 18th and 19th wards intended to illustrate cleavages between the white and African-American communities in this area of the City. Dr. Taylor did not perform a similar analysis for any other part of the City -- this was likely due to the history of this area of the City and the focus of the Barnett case on the 18th and 19th wards.
278. While much of Dr. Taylor's analysis of the 18th and 19th wards concerned incidents which occurred 20 or more years ago, the report also discussed more recent incidents which implied continuing racial tensions in this area of the City. While the racial divergences in this part of the City are neither as virulent nor as obvious as they were in response to the school desegregation efforts, including bussing, and the fair housing initiatives of the 1960's and 1970's, the African-American and white communities still have different interests with respect to education and affordable housing policies.
279. Dr. Taylor's contemporary analysis revolved around: 1) the debate to expand the High School for Agricultural Sciences, a successful magnet school with a largely African-American student body situated in the majority white Mt. Greenwood neighborhood of the 19th ward, and 2) the controversy surrounding the development of low-income and affordable housing projects in the 19th ward. While this Court does not question the good faith of the elected representatives in the 19th ward in their opposition to these projects due to fears of overdevelopment and overcrowding, the relevant issue is whether whites tended to have a different valuative calculus than their African-American counterparts and whether white elected representatives tended to favor the interests of their white constituents to the detriment of their African-American constituents. Well-intentioned people could have had radically different opinions, which were entirely non-racial, with respect to each of the recent incidents discussed by Dr. Taylor. What is relevant is whether these differences tracked differing communities of interest between the white and African-American communities.
280. In order to illustrate the existence of disparate communities of interest in the southeast and southwest sides, Dr. Taylor made use of socio-economic data and maps illustrating that data. Dr. Taylor's illustrative maps were intended to indicate the extent to which residents of the relevant areas had similar needs and desires and therefore constituted a community of interest. It is likely, of course, an oversimplification to equate a person's socio-economic condition with his or her opinions on public policy matters. In order to attempt to establish his hypothesis, Dr. Taylor selected six socio-economic factors: households per census block, percent rental units, per capita income, percent vacant dwelling units, percent public school enrollment, percent of families in poverty with children, percent single parents with children, and percent African-American. Why these data were selected instead of other, or additional, socio-economic data was never explained. Apparently this data was selected because it captured the differences Dr. Taylor sought to portray. This, of course, does not confront the issue of whether self-fulfilling indicators were selected to the exclusion of data that would not have confirmed Dr. Taylor's hypothesis. Nor was it clear that these indicators are ones which were relevant in the redistricting process. Given the trial record, the factors relevant to redistricting were not necessarily socioeconomic or racial but were the preservation of traditional ward boundaries, the preservation of neighborhoods and parish boundaries, and population equalization.
281. This Court is troubled by Taylor's analytical method in this phase of his report. Dr. Taylor's method consisted of mapping the socio-economic data and then drawing conclusions concerning the correlation between race and communities of interest based upon a review of the patterns revealed on the maps. The keys prepared by Dr. Taylor significantly effected the overall patterns identified on the maps and rendered Dr. Taylor's identification of communities of interest problematic.
282. The statistics displayed in Dr. Taylor's southwest side and southeast side maps were exceptionally skewed by his selection parameters. A recital of the keys contained in Dr. Taylor's maps illustrates the problem:
a) in his households per census block graph, Dr. Taylor's parameters were 0-15, 16-30, 31-40 and 41-366 households per block, a statistical division which made little sense in a city with as many apartment buildings as Chicago;
b) in his percent of rental units graph, Dr. Taylor's parameters were 0-9.99%, 10-19.99%, 20-29.99%, and 30-100% rental units, once again no reasonable justification was offered;
c) in his per capita income graph, Dr. Taylor's parameters were $ 0-$ 12,000, $ 12,001-$ 17,000, and $ 17,001-$ 50,281 per capita income, a set of parameters exhibiting a rather bewildering selection of break-points insofar as a community where the average per capita income is approximately $ 17,500 would have little in common with one where the average per capita income is $ 50,000, yet in Dr. Taylor's analysis these two communities would be identical;
d) in his graph of percent vacant dwelling units, Dr. Taylor's parameters were 0-.00%, 1-2.99%, 3-5.99%, and 6-100% vacant rental units;
e) in his graph of public school enrollment, Dr. Taylor's parameters were 0-50%, 50.01-80% and 80%-100%;
f) in his graph of the percent of families in poverty with children, Dr. Taylor's parameters were 0-.99%, 1%-3.99%, 4%-7.99%, and 8%-100%;
g) in his graph of the percent African-American population, Dr. Taylor's parameters are 0-50%, 50.01-80%, and 80.01-100%.
These particular break points were so extreme, and were apparently calculated for the purpose of creating the starkest possible differences in the communities by concealing the more nuanced nature of the socio-economic spectrum.
283. Even accepting these fragmented socio-economic data, only the most generalized conclusions concerning socio-economic communities of interest could be reached based upon them. While there were some differences between the western and the eastern ends of the 18th and 19th wards, these areas were not so disparate that they could confidently be characterized as containing different communities of interest. Nor were the eastern sections of the 18th and 19th wards significantly more similar to the areas to their east than they were to the areas to their immediate west, except in terms of being super-majority African-American areas. If Dr. Taylor's socio-economic analysis of the southwest side established anything, it was that the eastern ends of the 18th and 19th wards are transitional areas with mixed socio-economic characteristics which are not readily stereotyped or diagramed.
285. Dr. Taylor's 'hate crimes' map, however, indicated that the southwest side, particularly the 18th and 19th wards, and the Marquette Park areas, as well as Hyde Park and Kenwood, Bridgeport and portions of the downtown area and the far north side and Uptown are sites of some tension. While the 'hate crimes' included in Dr. Taylor's map were not limited to racially based 'hate crimes', the areas of the highest incidence of hate crimes on the southwest side corresponded roughly to those areas experiencing racial change or which could be characterized as border areas. While this data indicated that these neighborhoods were unfortunately unstable, the conclusion that the eastern ends of the 18th and 19th wards should be joined with the wards to their east does not necessarily follow.
3. Professor Richard Engstrom
286. Professor Richard Engstrom, a Professor of political science at the University of New Orleans, testified on behalf of the Barnett plaintiffs concerning: 1) racially polarized voting between African-Americans and non-African-Americans; 2) alternative ward maps intended to establish that additional African-American wards could be drawn; 3) differential rates of packing and fracturing between African-Americans and whites under the present ward map. (BAPX 13, 14A, 38-57, 70, 149, 150).
287. Professor Engstrom is a seasoned veteran at considering the validity of districting schemes under the VRA. He is generally considered one of the foremost experts in the field of studying voting behavior. He has also published numerous articles on election systems. Even the defense experts recognized that he is one of the foremost experts in the field whose reputation for integrity is beyond cavil. (Tr. 7838-Weber). This Court mentions Professor Engstrom's professional resume in light of defendants' attempts to characterize him as an unreliable "professional" witness with little practical experience, and to imply that Professor Engstrom's testimony and reports were authored by other persons. This Court finds no ...