Bridgeport Guardians, Inc. v. City of Bridgeport, 735 F. Supp. 1126 (D. Conn. 1990)

*1127 Vincent Musto, for plaintiffs.

Thomas Jackson, William Barnes, for defendants.

DALY, District Judge.

In this action plaintiffs claim that the defendants are using a racially discriminatory selection process for the promotion of police officers to the rank of sergeant in the Bridgeport Police Department. Plaintiffs allege that defendants' promotion procedures violate sections 1981 and 1983 of Title 42 of the United States Code and Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-2(a). The parties having fully briefed the legal and factual issues raised at the consolidated preliminary injunction hearing and trial, this matter is now ready for decision.

Plaintiffs are individual Black and Hispanic candidates for the position of police sergeant who took the 1989 Bridgeport Police Sergeant's Examination, and two organizations of minority police officers, the Bridgeport Guardians and the Hispanic Society. They initiated this action on October 2, 1989 seeking a temporary restraining order and preliminary and permanent injunctive relief. The defendants are the City of Bridgeport, the Bridgeport Police Department, and the Bridgeport Civil Service Commission (the "Commission"). On November 15, 1989, the Court allowed thirty-nine individual White police officers and an organization of such officers, the Bridgeport Police for Equal Employment Opportunity, Inc., to intervene as defendants in this matter (the "intervening defendants").

On October 2, 1989, the Court issued a temporary restraining order prohibiting the defendants from making any promotions from the eligibility list drawn up after the completion of the promotion testing and appeals period. On the consent of the parties, the temporary restraining order remains in effect pending this decision. A consolidated trial and preliminary injunction hearing was held on December 27 and 28, 1989.

On February 6, 1989, the Commission announced that it would conduct a promotional examination for the rank of sergeant. The examination notice stated that the Commission required that candidates had served as police officers and/or detectives for one year immediately preceding February 3, 1988, be bona fide residents of the City of Bridgeport^[1], and have "one *1128 year of satisfactory experience as a police officer immediately preceding February 3, 1988"^[2]. Dr. James Outtz, an industrial psychologist and testing expert who previously developed the 1983 Bridgeport Police Sergeant's Examination, was retained to construct the 1989 examination. As a result of the process he developed, a candidate's ranking, for promotion purposes, is based on the following three components: 1) the written examination score accounting for fifty-three percent of one's total score; 2) the oral examination score accounting for forty-two percent of an applicant's total score; and 3) seniority accounting for the remaining five percent.

In the course of developing an examination, in November 1988 Dr. Outtz recommended to the defendants that video simulations be used in place of an oral test in order to improve the examination and reduce the possibility of adverse impact. To encourage the defendants to accept this proposal, he offered to defer his fee until July, 1989.^[3] Despite Outtz's recommendation and the efforts of the Commission personnel director to convince Bridgeport Mayor Thomas Bucci to adopt the video simulations concept, the Mayor's office, while apparently not opposed to the concept in principle, rejected the request to provide additional funds for this proposal and directed the Commission to expend such funds from its own budget to whatever extent it saw fit.

The testimony of Cesar Batalla indicated that in reality the Commission had sufficient funding in its "Special Services" budget (out of which testing costs are normally paid) to pay for the video simulations. Batalla testified that in the 1988-89 fiscal year the Special Services budget had a surplus of $110,000. Though the Commission claims not to have learned of this extra funding until a May 1989 audit turned up an extra $100,000 in the Special Services budget, the Court finds this explanation less than satisfactory. Moreover, regardless of any such confusion about such additional funds, the Commission had available in the Special Services budget large balances of funds (perhaps already committed to other projects) from October 1988 through May 1989. Although the Commission supported the use of the video simulations in attempting to get Mayor Bucci to authorize additional funds for the Special Services budget, for some undisclosed reason it was unwilling to spend any portion of the large balance of funds actually in that budget for that same purpose.

The written examination was administered on April 1, 1989, and the oral examinations were conducted over a three-day period during June 13-15, 1989. One hundred and seventy candidates took the examination, of whom one hundred and fifteen are White, twenty-seven are Black, and twenty-eight are Hispanic.^[4] After the *1129 Commission reviewed appeals, on October 11, 1989 it prepared a pass list^[5] for selecting candidates for promotion. The first nineteen candidates listed are White. The Black candidate with the highest score is ranked twentieth. The Hispanic candidate with the highest score is ranked twenty-second. Of the ninety-nine candidates who passed the examination, seventy-eight are White, eight are Black, and thirteen are Hispanic. Of the seventy-one candidates who failed the examination, thirty-seven are White, nineteen are Black, and fifteen are Hispanic. Thus, sixty-eight percent of the White candidates passed the examination, while thirty percent of the Black candidates and forty-six percent of the Hispanic candidates passed. The defendants presently intend to promote the first twenty-five persons from the promotion eligibility list at the conclusion of this litigation. Over the course of the two-year period that the list will remain in effect, the defendants anticipate making seven to fourteen additional promotions.

On August 15, 1989, Dr. Outtz recommended in writing that the Commission use "banding" in making promotions from the pass list rather than promoting in "strict rank-ordered fashion" in recognition of the fact "that certain differences in test scores may not be significant." Plf. Exh. 4. Based on that premise, the banding technique selects a range of scores whose differences are not statistically significant and then provides for promotions from a band range, considering matters such as race or ethnicity, gender, work experience, past job dependability, and other factors that the hiring authorities deem pertinent and worthy of consideration. Outtz stated in his August 15, 1989 letter that the tests he developed "cannot be justified as valid for the specific use of strict rank ordering." Id. At trial, however, Outtz testified that his letter should not be construed as an admission of adverse impact because he regularly recommends banding to make the test procedure as valid as possible. On August 16, 1989, the Commission voted (with four members present) two to zero to use banding for promotion purposes. On August 21, 1989, the full Commission revisited the issue and voted three to two to promote in strict rank order. Commission Chairman John Fabrizi testified at trial that if the examination or selection procedure had a disparate impact on minority candidates or otherwise violated equal employment laws, he would vote to use banding. He further testified that when he cast his vote against banding on the second occasion he and the Commission had not reviewed any analysis of the test results to determine if there was a disparate impact on minority groups.

The Court's remaining findings of facts are incorporated in its discussion concerning its conclusions of law and primarily in its discussion of whether the plaintiffs have succeeded in proving their disparate impact claim.

Title VII of the Civil Rights Act of 1964 proscribes employer discrimination against any person with respect to hiring or the terms and conditions of employment because of such individual's race, color, religion, sex, or national origin. 42 U.S.C. § 2000e-2(a). Also proscribed is the limitation, segregation or classification of employees in ways that adversely affect any employee on account of race, color, religion, sex, or national origin. Id. In Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S. Ct. 849, 853, 28 L. Ed. 2d 158 (1971), the Supreme Court held that Title VII prohibits not only intentional discrimination, but also discrimination resulting from practices which, though apparently "fair in form", are discriminatory in fact. *1130 Basing their claim on this theory of liability, plaintiffs argue that the sergeant's promotion process had a "disparate impact", i.e., it adversely impacted upon minority promotion candidates.

To prove this claim, the plaintiffs' must first establish a prima facie case of liability by showing that the promotion disparity is the result of certain, specific aspects of the promotion process having a significantly disparate impact upon them. Wards Cove Packing Co. v. Atonio, ___ U.S. ___, 109 S. Ct. 2115, 2124-25, 104 L. Ed. 2d 733 (1989); Equal Employment Opportunity Commission v. Joint Apprentice Committee of the Joint Industry Board of the Electrical Industry, 895 F.2d 86, 90-91 (2d Cir. 1990) ("plaintiff must establish that the challenged employment practice caused the statistical disparity") ("Joint Apprentice"). If plaintiffs succeed in establishing disparate impact, then the burden of production shifts to the defendants to offer justifications for the practices at issue. Wards Cove, 109 S. Ct. at 2125-26. If plaintiffs are unable to prove that such practices are unjustified, then plaintiffs' claims will fail, unless they can demonstrate that alternative selection techniques (without the same undesirable disparate effect) exist that are equally as effective in meeting the employer's goals as those presently used. Id. at 2126-27.

Initially, the plaintiffs bear the burden of proving that the promotion process had a disparate impact on minorities. The parties dispute the frame of reference the Court should use in examining this issue. The defendants and intervening defendants, through the testimony of their experts, Drs. Outtz and Veres, assert that the Court should focus on the selection rates from the promotion list at the various expected levels of promotions over the next two years. Plaintiffs, through the testimony of one of their experts, Dr. Gelfand, assert in part that the Court should resolve the question of disparate impact by relying on an analysis of the entire distribution of candidates' test scores, regardless of whether they passed or failed and regardless of the rates at which different groups will be selected for promotion.

Though there is merit in both approaches, as a starting point the Court finds the latter to be more appropriate here. The analysis proposed by the defendants and intervening defendants dilutes the disparate impact evident in the entire range of exam scores by requiring specification of an arbitrary score for dividing the entire distribution of scores and in doing so ignores the information provided by examination scores falling outside the sample derived in part from the setting of the passing score.^[6]Cf. Bushey v. New York State Civil Serv. Comm'n, 733 F.2d 220, 224 (2d Cir.1984) (reviewing both the distribution of minority and nonminority candidates' scores and rates at which they passed the examination at issue), cert. denied, 469 U.S. 1117, 105 S. Ct. 803, 83 L. Ed. 2d 795 (1985). Moreover, the importance of reviewing the examination results as a whole is particularly significant since the examination accounts for 95% of a candidate's promotion ranking. Using the Mann-Whitney procedure to assess the relative distributions of scores of White, Black, and Hispanic candidates, plaintiffs' expert, Dr. Gelfand, demonstrated that the probability that differences in the combined scores occurred by chance in the examination scores of Whites verses Blacks was 0.0001 and in the case of Whites verses Hispanics was 0.0002. These results indicate that the differences in the distributions of scores are extremely unlikely to have occurred by chance and are statistically significant. Similar results are obtained whether one views the examination as a whole or analyzes separately its written and oral components. Thus, the Court concludes that the examination itself had a disparate impact on Black and Hispanic candidates, significantly and disproportionately reducing their promotional opportunities.

Plaintiffs also assert that the strict rank order selection procedure defendants intend to use in making promotions will *1131 cause additional disparate impact upon minorities. The competing statistical analyses offered by the defense make this contention somewhat more difficult to resolve. At present, the defendants seek to promote twenty-five candidates from the promotion list. During the two-year life of the list, between thirty-two and thirty-nine promotions are expected. Using the benchmark known as the "four-fifths" rule of the 1978 Uniform Guidelines on Employee Selection Procedures ("Guidelines"), 29 C.F.R. part 1607, § 1607.4(D), plaintiffs argue that an examination of selection ratios, which defendants implicitly agree gets closer to the proper focus of disparate impact analysis, evidences adverse impact. As the Guidelines explain, "[a] selection rate for any race, sex, or ethnic group which is less than four-fifths ( 4/5 ) (or eighty percent) of the rate for the group with the highest rate will generally be regarded ... as evidence of adverse impact...." Id.; see also Griggs, 401 U.S. at 432, 91 S. Ct. at 854 (stating that the Guidelines are entitled to "great deference"); Guardians Ass'n of the New York City Police Dep't v. Civil Service Comm'n, 630 F.2d 79, 88 (2d Cir. 1980), cert. denied, 452 U.S. 940, 101 S. Ct. 3083, 69 L. Ed. 2d 954 (1981). The regulations that establish the four-fifths rule state, however, that "[g]reater differences in selection rate[s] may not constitute adverse impact where the differences are based on small numbers and are not statistically significant...." 29 C.F.R. § 1607.4D; accord NAACP v. City of Mansfield, 866 F.2d 162, 168-69 (6th Cir. 1989) (noting that the four-fifths computation should be treated carefully when dealing with small sample sizes); see also Clady v. County of Los Angeles, 770 F.2d 1421, 1427 (9th Cir.1985) (noting that four-fifths computation, or 80% rule, has been sharply criticized by some courts and commentators and used by others), cert. denied, 475 U.S. 1109, 106 S. Ct. 1516, 89 L. Ed. 2d 915 (1986).

In this case, the selection ratios for Blacks as opposed to Whites and for Hispanics as opposed to Whites at twenty-five, thirty-two, thirty-nine, forty-five, fifty, fifty-five, and sixty promotions indicate that strict rank order promotions will also have a substantial adverse impact upon Black and Hispanic candidates and confirm Dr. Gelfand's findings regarding the examination as a whole. The highest Black/White selection ratio at any of the promotion levels discussed above is only 45.3% at forty-five promotions. The highest Hispanic/White selection ratio at any of those same promotion levels is 39.1% at thirty-nine promotions.^[7]

Drs. Outtz and Veres dispute the significance of the four-fifths rule analysis relying on their own statistical analyses, respectively the Chi-square and the Fisher's exact test, to prove that no adverse impact results from the use of strict rank-ordered promotions. These analyses are persuasive. They indicate that at twenty-five, thirty-two, and thirty-nine promotions there is no evidence of disparate impact because the difference in the actual results from the expected results, when comparing promotion levels of Blacks and Whites and Hispanics and Whites, is not statistically significant.^[8]

*1132 Though plaintiffs argue that these results are highly susceptible to error because the sample sizes involved are small, they ignore the utility and sensitivity of the Fisher's exact test in giving exact results, rather than the Chi-square's approximations (even when used with the Yates correction). Dr. Gelfand admitted as much. Similarly, plaintiffs' other expert Dr. Barrett, partly disagreeing with Dr. Gelfand on another point, candidly testified that the Chi-square analysis is "particularly sensitive to small samples" and that such analysis is useful in this type of case because "it gets at the heart of the matter [by computing] the likelihood that the number of promotions is the result of random chance."^[9] In light of this testimony, the inherent descriptive weakness of the four-fifths rule in the case of small samples^[10], and the nature of defendants's expert testimony, the Court concludes that disparate impact has not been sufficiently proven to result from the use of strict rank ordering in promoting.^[11]

Since the plaintiffs have succeeded, in part, in establishing a prima facie case of disparate impact, the Court must next consider whether the defendants have met their burden of producing evidence to rebut the prima facie case by showing that legitimate, nondiscriminatory reasons exist for the use of the examinations in question. Joint Apprentice, 895 F.2d at 91 (citing Watson v. Fort Worth Bank and Trust, 487 U.S. 977, ___, 108 S. Ct. 2777, 2784, 101 L. Ed. 2d 827 (1988)). In this phase of disparate impact analysis, the Court must review "the justifications an employer offers for [using the practice in question] and ... the availability of alternate practices to achieve the same business ends, with less racial impact." Wards Cove, 109 S. Ct. at 2125. Though in substantial justifications will not suffice to meet an employer's burden, the defendants need not show that the challenged practice is essential or indispensable for it to pass muster. Id. at 2126.

To facilitate this analysis, the Court will review the written and oral portions of the examination separately. The written examination, the defendants contend, arose out of the process agreed to by the plaintiff Bridgeport Guardians and approved by this Court in 1982 in connection with litigation surrounding the 1983 Bridgeport sergeant's promotional process. Prior to 1983, there were no minorities above the rank of patrolman in the Bridgeport Police Department. Today, minorities comprise 30.7% of those officers ranked as sergeants. Defendants argue that by agreeing to the 1983 process, about which no *1133 claim of discrimination has been lodged with this Court, plaintiffs are barred from challenging this examination process. See Gilty v. City of Oak Park, F.E.P. Cas. (BNA) 1388, 1394, 1989 WL 84398 (N.D.Ill. July 18, 1989) (defendant who implemented promotional process agreed upon by plaintiff as part of settlement of prior litigation has articulated legitimate business reason for using same).

This argument, though not without some illusory appeal, is entitled to little weight for the following reasons. First, it ignores the fact that, with the exception of the Bridgeport Guardians, many of the plaintiffs in this action were not parties to the November 3, 1982 consent decree. Cf. Martin v. Wilks, ___ U.S. ___, 109 S. Ct. 2180, 104 L. Ed. 2d 835 (1989) (holding that nonparties to a consent decree resolving a Title VII case may fully challenge, beyond the narrow grounds normally available for such a collateral attack, employment decisions taken pursuant to it). Second, the consent decree specifically does not include the suggestion that the plaintiffs would not challenge the selection procedure if it disproportionately impacted upon minority candidates. Exh. 502, ¶ 5 (crossed-out language). Nor does the consent decree purport to restrict the ability of plaintiffs to challenge the improper development or administration of any promotional process. And finally, after its implementation, the validity of the 1983 sergeant's promotional process has never been determined in any legal proceeding.

The defendants next argue that Outtz demonstrated that the examination is both reliable and valid.^[12] The written portion of the examination was developed largely from the 1983 written examination. Knowing that the Bridgeport police force had rewritten its policies and procedures in 1987 and attempting to build upon the prior written test, Dr. Outtz updated his job analysis of the knowledges needed for job success by sampling twenty-six current Bridgeport police sergeants. He factored this updated data into a chart^[13] relating job duties, knowledges, skills, and abilities and then later drafted multiple choice examination questions from the six categories of documents identified as the sources for the knowledges deemed important to those officers who were sampled.

Additionally, Dr. Outtz attempted to improve the 1989 process and reduce the possibility of any adverse impact by creating a study manual to assist candidates preparing for the written test. The manual consisted of open-ended questions derived from those source documents identified in the job analysis as important to job success. All of the written multiple-choice test questions were taken from questions in the study manual, given to each candidate eight weeks prior to the test. This allowed candidates to prepare themselves for the multiple-choice written test by answering the questions in the study manual.

In further attempting to produce evidence that the written portion of the examination was reliable, the defendants also offered evidence that the written examination's reliability coefficient was 0.85 for all candidates and 0.79 for minority candidates, *1134 in both instances exceeding the ranges normally deemed acceptable.

In response to defendants' assertions that the written examination was reliable and valid and that, therefore its use as one component of the promotion process is justified, plaintiffs initially contend that the job analysis and, hence, the resulting examination was flawed. This argument attacks the validity of both the oral and written portions of the test. Specifically, plaintiffs criticize Outtz for: 1) failing to update his analysis of the skills and abilities needed for the job from 1983 to 1989; 2) determining only a single average importance rating for each knowledge, skill, and ability; and 3) relying on the misleading "cross-products" obtained by multiplying the average importance of each knowledge, skill, and ability by the average importance of each job duty.

These criticisms are insubstantial. After interviewing four to five sergeants on the Bridgeport police force and developing nine sergeant's examinations for a number of municipalities, Dr. Outtz determined that the skills and abilities need for the job remained constant and decided to rely on his 1983 calculations identifying and assigning average importance ratings to the various skills and abilities needed for the job. Other than attacking his failure to take further steps, the plaintiffs have not proven to the Court that additional skills and abilities had arisen during the intervening years or that the mean importance of the already identified skills and abilities had changed since 1983.

Similarly, the Court also finds unpersuasive plaintiffs' criticisms of Outtz's chart of job duties, knowledges, skills, and abilities. Dr. Veres testified that conducting a job analysis is an art. The Court agrees. Dr. Outtz's methodology is well within the realm of acceptability. Though plaintiffs' criticism that the cross-products distort the true relationships between certain job duties and specific knowledges, skills, or abilities is well-taken, it is of little significance because this error does not impact upon the relative weights ascribed to knowledges, skills, and abilities for purposes of determining how to weight the written and oral tests.

Through the frequently unspecific testimony of Dr. Barrett, plaintiffs also criticized the use of written examinations in general, the failure of the written test to measure a candidate's interpersonal relations and ability to integrate information, and the derivation of knowledges from some 3000 pages of material in 1983. In light of Dr. Barrett's candid admission that the written test had "some limited validity" and might be useful as an initial screening device, and because Dr. Outtz developed the study manual to focus candidates' study of knowledges needed for the test and to remove emphasis on rote memorization, the Court deems these criticisms insufficient.

Defendants also seek to justify the reliability and validity of the oral test. Each candidate who sat for the written examination was allowed to take the oral examination. For the oral assessment process, officers from other police departments who agreed to act as raters helped Dr. Outtz developed ten scenarios designed to further test a candidate's skills and abilities not measured by the written test. Before a panel of three raters each candidate was presented with the scenarios and then asked how he or she would respond. Candidates were allotted three minutes to respond to each scenario. Each of the twenty-seven panel members had law enforcement experience and had attained the rank of sergeant or higher. None were minorities or were affiliated with the Bridgeport police department. To structure their scoring of candidates, the raters were trained beforehand and provided with a scoring key for evaluating answers. With the exception of the chairperson, panelists were also rotated to prevent the raters from becoming too accustomed to one another. To alleviate any problems resulting from disclosure of the scenarios over the course of the two and one-half day oral test period, alternative wording was used to vary the scenarios without changing their substantive content. Finally, for appeal purposes and to keep a check on the raters, candidates' responses were audiotaped.

*1135 Plaintiffs attack the oral assessments because: 1) the defendants failed to obtain minority panelists as raters; 2) the questions and answers from the oral examination panels were leaked by officers who took the examination on the first or second day and then discussed with or around those who had yet to sit for it; 3) Dr. Outtz failed to weight the answers to the oral test so as diminish the fact that certain scenarios and a candidate's responses to them accounted for a larger portion of a candidate's score than others, e.g., a correct response to scenario three accounted for 16% of the test while a correct response to scenario five accounted for only 5.3% of the test score; and 4) oral examiners were allowed to collaborate in rating each candidate.

The failure to obtain minority panelists, in spite of Outtz's recommendation to do so, is disturbing. The City of Bridgeport's Affirmative Action plan requires that minorities be included on all oral panels. Exh. 36 at 42. The defendants sent letters to other departments requesting assistance in connection with the oral examination. These letters, however, did not specifically state that minority panelists were needed. Assistant Personnel Director John Colligan testified that efforts were made to get minority raters by telephone contact with other police departments in Connecticut and New York, but that these efforts were unsuccessful largely due to the fact that the raters were requested to set aside eight days for training and development of the scenarios and for examination and grading. He further testified that the letters were sent to only those police departments who requested a written follow-up to the initial telephone contacts described above.

Though the Court finds Colligan's testimony considerably less than convincing, the question of what, if any, effect the omission of minority panelists had on the oral test scores is the Court's main focus. In this regard, Dr. Barrett testified that a thirty-four year old study supported his conclusion that the omission of minority panelists had a "chilling effect" on minorities taking the oral test. Testimony from some of the minority sergeant's candidates also indicated that they felt "intimidated" by the fact that were no minority panelists. The three officers who so testified at trial, however, scored above the nonminority mean score on the oral test.

More significant, however, are the results of Dr. Outtz's multiple regression analysis used to determine which variables, race, gender, the day the oral test was taken, or the written examination score, were most predictive of the oral test score. This analysis showed that the written score was most predictive of one's oral examination score and that the date the oral examination was taken was also of some lesser significance. The other two variables, race and gender, did not add anything to the relative predictability of the oral score. The significance of this result, as testified to by Dr. Outtz, is that candidates with higher written scores did better on the oral test, regardless of the panel before whom they appeared and regardless of their race. Though Dr. Gelfand criticized Dr. Outtz's failure to use "dummy coding" in connection with the "day of the oral exam" variable, he did not perform his own multiple regression analysis to counter Outtz's analysis. Moreover, Outtz noted that the use of scoring keys structured the rating panels to reduce the possibility that impermissible considerations, such as race or ethnicity, would enter into the scoring decision. Plaintiffs have failed to offer sufficient evidence to rebut such proof.

Outtz's regression analysis did show, notwithstanding Gelfand's criticism, that the day the oral examination was taken was of slight statistical significance as a predictive factor for the oral score. At trial, Officer Carwin Spearman testified that he was approached by another officer who had yet to take the oral assessment but stated that he knew all the questions on it anyway. That officer, as it turns out, failed the examination. Officer Spearman also testified that he heard several other conversations in which the oral test was discussed by unidentified other persons who had taken or were anticipating taking it. Given the results of Outtz's regression analysis, the time and financial constraints *1136 involved in obtaining twenty-seven raters to administer a half-hour oral test to one hundred and seventy candidates and then to grade such candidates, and the fact that the defendants required each candidate to sign an identification sheet forbidding disclosure of the test materials and indicating penalties for the same, the Court is not persuaded the defendants could have or realistically should be expected to prevent such talk.

Nor is the Court persuaded that the plaintiffs' remaining two criticisms of the oral test, involving weighing of the responses to the scenarios and rater collaboration, withstand the justifications offered by the defendants. The Commission, not Dr. Outtz, decided that there would no adjustment for the fact that some scenarios would account for a greater portion of the oral test score apparently to avoid the perception that candidates' scores were being unfairly adjusted after the fact. The Court also notes that the number of responses appropriate for any given scenario accounts for the reason that the scenarios are of different weight. Similarly, the Court accepts the justification offered for why the rater collaboration was allowed in scoring the oral examination. Dr. Outtz testified that he introduced this feature to reduce individual rater bias. The effect of such collaboration also was lessened by the fact that the panelists rotated among panels. In light of the insubstantial nature of plaintiffs' criticisms and the reasons offered by the defendants for these practices, the Court finds them justified.

Accordingly, the Court holds that the justifications offered by the defendants for the use of the written and oral examination components as a part of the promotional process are convincing and such use serves, in significant fashion, the defendants' legitimate employment goals. Plaintiffs have failed to meet their burden of proving the contrary is true, in spite of the disparate impact analysis shown by Dr. Gelfand's analyses. Therefore, the Court must next consider whether alternative practices are available to achieve the same business goals, namely promoting qualified persons to the rank of sergeant, with less racial impact. See Wards Cove, 109 S. Ct. at 2125.

"If [plaintiffs], having established a prima facie case, come forward with alternatives to [defendants'] hiring practices that reduce the racially-disparate impact of practices currently being used, and [defendants] refuse to adopt these alternatives, such a refusal would belie a claim by [defendants] that their incumbent practices are being employed for nondiscriminatory reasons." Id. at 2126-27. In considering such alternatives, cost factors and other burdens resulting from the use of such alternatives may be relevant to the Court's determination of whether such practices would be equally as effective in serving legitimate employment goals. Id.

Plaintiffs propose two alternatives to reduce the combined examination scores' disparate impact. First, they contend that the defendants should have implemented the video simulations recommended by Dr. Outtz in lieu of the oral assessment procedure. The Court is not persuaded by this argument for two reasons. Outtz testified that he has typically found that, though the use of such procedures leads to a more valid examination process, typically it has not increased the relative standing of minorities on the tests with which he has been associated. Furthermore, this procedure substantially adds to the cost of the promotion process. Though it appears that the Commission had such funds to spend, perhaps unbeknownst to even itself, the cost of the video simulation would nearly have doubled the cost of the examination to the defendant City. Given such doubts about the ultimate utility of such a procedure and its relatively high cost, the Court holds that the plaintiffs have not proven that this alternative testing procedure would be equally or more effective than those actually utilized.

The second alternative procedure plaintiffs proffer is banding. Plaintiffs argue that banding will enable promotions to be made from a larger portion of the promotion *1137 list, thereby enhancing the promotion opportunities of Blacks and Hispanics. As noted previously, Dr. Outtz recommended in August 1989 that banding be used, stating that "employment tests ... cannot be justified as valid for the specific use of strict rank ordering." During trial, Outtz backed away from that statement, testifying that because under his analysis no disparate impact had been shown, banding, while desirable and the best way to use the test results, was not needed to correct any disparate impact. Dr. Outtz further testified that banding might put one or two minorities in a better position to be promoted.

Defendants and intervening defendants respond that the outcome of the kind of banding Dr. Outtz proposes is unclear at best, along with the parameters of what factors should be appropriately considered for promotion within any specific band. Their experts both agree, however, that they regularly recommend banding. Defendants and the intervenors further argue that the City Charter provides that promotions shall be made according to the "rule of one" (i.e., in strict rank order). Cf. Bridgeport Firebird Society v. City of Bridgeport, 686 F. Supp. 53, 58 (D.Conn. 1988) (noting that civil service charter provisions may not override the goals of Title VII). In light of these uncertainties and in view of the additional burdens that implementing such a banding procedure would entail, they argue that the plaintiffs have not met their burden of showing that this alternative selection technique will reduce the adverse impact resulting from the examination while providing an equally as effective means of meeting the employer's hiring goals.

This Court disagrees. There is no evidence before the Court that any added burdens that a banding analysis would impose upon the defendants are more than minimal, either financially or temporally. While the Court accepts the proposition that rankings near the top of the promotion list indicate higher qualifications than those at the bottom of the list, lock-step reliance upon a system that promotes one person over another when the difference between their rankings is not statistically significant is unwarranted, particularly in a case where the examination that accounts for 95% of one's ranking on that list has been shown to disproportionately affect Blacks and Hispanics. Nor is the Court persuaded that uncertainty about the ultimate results of banding should preclude its implementation. Banding results in a more valid promotional procedure and may lead to the promotion of additional minorities. Finally, while it might seem anomalous to consider banding as an alternative selection device in light of the Court's conclusion that strict rank order promotions will not add to the disparate impact previously identified to have occurred, the use of banding is entirely justified to help overcome the disparate impact resulting from the examination itself. Plaintiffs have met their burden of proving that this alternative selection device will assist to offset the examination's disparate impact.

At this time, the Court is no position to mandate what type of banding procedure is appropriate. However, the Court rejects Dr. Barrett's suggestion that a separate race norm promotion process be implemented, dividing the candidates into three groups on the basis of race and ethnicity and promoting a certain percentage of top candidates within each group. The Court's view, which the parties will be given further opportunity to influence, is that banding should take into account a number of factors not necessarily accounted for in the examination process including a candidate's past work dependability, job experience, race, ethnicity, gender, and any other factors shown to be relevant and worthy of consideration by a similar employer. Rather than promoting simply on the basis of a candidate's race or ethnicity, the Court contemplates merely that such factors are among those to be considered in promoting within a specific band. Ultimately, however, a decision regarding the most appropriate type of banding procedure, is a matter for the proceedings that will follow.

Plaintiffs' complaint also alleges violations of 42 U.S.C. §§ 1981 & 1983. In their *1138 pre- and post-trial briefs, however, plaintiffs have failed to include any conclusions of law relating to such claims. Accordingly, the Court deems them abandoned and dismisses them with prejudice. Moreover, were such claims not abandoned they could not proceed here because plaintiffs have failed to prove the requisite discriminatory intent necessary to support such claims. General Building Contractors Ass'n, Inc. v. Pennsylvania, 458 U.S. 375, 382-91, 102 S. Ct. 3141, 3145-50, 73 L. Ed. 2d 835 (1982) (§ 1981); Washington v. Davis, 426 U.S. 229, 239-41, 96 S. Ct. 2040, 2047-48, 48 L. Ed. 2d 597 (1976) (fourteenth amendment).

Judgment shall enter in plaintiffs' favor on their Title VII claims to the extent of and in accordance with this opinion. Plaintiffs' remaining two claims are hereby DISMISSED with prejudice. Counsel for all parties shall submit by April 6, 1990 written proposals regarding how banding should be accomplished for the Court's review.

SO ORDERED.

[1] During the trial there was some dispute about whether certain officers had met this requirement. The defendants offered evidence that the Commission requested and received verification from the defendant police department that all of the candidates were residents of the City of Bridgeport. After the examination was completed, certain candidates complained about, and on October 13, 1989 the Commission asked Mayor Bucci to have the Office of Internal Affairs investigate, the residency of four individuals. At the time of trial, that investigation was still ongoing. Without any evidence of its results and without any substantial basis for concluding to the contrary, on the basis of the present record the Court deems those four officers to be residents of Bridgeport.

[2] This requisite has been referred to as an officer's "service rating." On a pass/fail basis each candidate's service rating was reviewed by his or her commanding officer. Though plaintiffs contend that Officers Christy and Halpin should have been given negative service ratings and offered evidence via Superintendent Izzo that they were previously disciplined in connection with an incident of alleged brutality, this incident is not claimed to have occurred within the time frame for which the service rating was given. Additionally, plaintiffs offered evidence that in June 1987 a supervisor expressed concern about Officer Christy's progress after that incident. Though this came within the one-year period "preceding February 1988", defendants argue that it was not within the one-year period prior to the exam that the Commission intended to apply with respect to the service rating and also that the letter expressing concern need not necessarily have led to a negative service rating. The Court finds these arguments persuasive and finds that there is no evidence that the service ratings given were improper or intentionally or negligently manipulated for promotion purposes.

[3] The examination ultimately used by the defendants cost between $40,000 and $50,000. The use of the video simulations would have required an additional $46,000 expenditure.

[4] There was some dispute at trial about whether Officer James Viadero should be treated as White or Hispanic for purposes of promotion. Viadero has intervened as a defendant in this action and stated in an affidavit in support of the motion to intervene that his color and race is White. The Court finds, based on Viadero's testimony that all four of his grandparents came from Spain, that he is Hispanic. See 29 C.F.R. § 1607.4B ("Hispanic [includes] persons of Mexican, Puerto Rican, Cuban, Central or South American, or other Spanish origin or culture regardless of race").

[5] For purposes of creating the pass list, the passing point was established to be seventy percent of the highest grade.

[6] Stated another way, the results of the entire range of examination scores cannot be discounted away because it is primarily from such results that defendants' selection rates flow.

[7] Dr. Outtz criticized the reliance on the "four-fifths" rule in part because it considers all candidates without giving consideration to whether they passed or failed, to whether they were sufficiently qualified, motivated, or capable, and to whether they had adequately prepared. In view of this criticism, it might be argued that Dr. Barrett's selection ratio analysis overstates the disparate impact. By the Court's own calculations, however, the application of the four-fifths rule only to those ninety-nine candidates attaining ranking on the promotion list, as opposed to the entire group who took the exam, yields similar results. At levels of twenty-five, thirty-two, and thirty-nine promotions, the four-fifths rule evidenced disparate impact at all levels except vis-a-vis Blacks and Whites when thirty-nine promotions are assumed. At the thirty-nine promotion level, the selection ratio for Blacks/Whites is 89%.

[8] In other words, the probability values derived from the Fisher's exact test at these various promotion levels in all instances but one exceeded 0.05 and the number of standard deviations at the same levels were always less than two. See Castaneda v. Partida, 430 U.S. 482, 497 n. 12, 97 S. Ct. 1272, 1281 n. 12, 51 L. Ed. 2d 498 (1977) (noting general rule that standard deviations greater than two or three necessarily exclude chance as a cause of underrepresentation); see also Kirkland v. New York State Dept. of Correctional Services, 711 F.2d 1117, 1131 n. 17 (2d Cir.1983) (noting that the "likelihood that the actual results will fall more than one standard deviation beyond the expected results is about 32% ... [while] [f]or more than two standard deviations, it is about 4.6% and for more than three standard deviations, it is about [0].03%.") (citation omitted), cert. denied, 465 U.S. 1005, 104 S. Ct. 997, 79 L. Ed. 2d 230 (1984). At twenty-five promotions, comparing Whites and Hispanics, the difference between the expected and observed number of Hispanic promotions was within the range of statistical significance for probability values but not for the number of standard deviations.

[9] Barrett also testified that the use of the Fisher's exact test reduces that possibility of the so-called "type II" error possibly associated with Chi-square analysis and small sample sizes.

[10] This weakness can be demonstrated by considering selection ratios at twenty-five and thirty-two promotions by way of the following example. Assume that these selection ratios will be calculated considering only those persons who passed the examination. Further assume that while the number of Blacks who passed the examination remains constant at eight, the relative ranking of two officers are switched, thereby switching the positions of the Black candidate ranked forty-fifth and the White candidate ranked twenty-third. In this instance, the selection ratios derived from comparing Blacks and Whites at twenty-five and thirty-two promotions are 89% and 108% percent respectively; neither of which is statistically significant under the four-fifths rule.

[11] Though plaintiffs take a "shotgun approach" to attacking many other parts of the examination and promotion process, in light of this Court's prior findings of fact and in view of plaintiffs' failure to offer proof regarding the specific disparate impact resulting from such other selection devices, see Wards Cove, 109 S. Ct. at 2125, the Court will not consider such claims to the extent they are offered to establish disparate impact. The Court, when appropriate, will consider such evidence in connection with its discussion concerning defendants' justifications for the promotion process.

[12] Outtz testified that reliability is an index used to measure the internal consistency of an examination. Validity refers to whether a test reasonably measures the skills needed for the job. He further stated that a reliable test is not necessarily valid.

[13] Though the Court will also discuss this chart later, a brief explanation of its utility will suffice now. On the y-axis Dr. Outtz charted the knowledges, skills, and abilities deemed sufficiently critical to the job by the subject-matter experts for the purposes of the test. A candidate's comprehension, understanding, memory, and application of the knowledge are tested for on the written component of the test. The oral test targets a candidate's skills and abilities. On the x-axis, Outtz listed various job duties that sergeants must perform deemed important by the subject-matter experts. Numerical values of average rates of importance were assigned to each knowledge, skill, ability, and job duty based on Outtz's surveys of sergeants, whether in 1983 or 1989. Then by multiplying the average importance of each knowledge, skill, and ability by the average importance of each job duty and assessing the relative weight of each product, Dr. Outtz determined the relative weights to be given to the written (knowledges) and oral portions (skills and abilities) of the examination.