Full opinion text
MEMORANDUM AND ORDER SARIS, District Judge. I. INTRODUCTION In this class action, the plaintiffs allege that the written civil service cognitive ability examination used in 2002 and 2004 to qualify and rank applicants has had a disparate and adverse impact on Black and Hispanic candidates for entry-level firefighter positions in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-2(a), (k) (2006), and the federal consent decree in Boston Chapter, NAACP, Inc. v. Beecher, 371 F.Supp. 507 (D.Mass.1974) (the “Beecher decree”). The defendants are the Human Resources Division of the Commonwealth of Massachusetts (the “HRD”), which develops and administers the examination, the City of Lynn, and various public officials. The HRD argues that because of the statutory veterans preference, residency requirements, and other selection factors, the examination has no disparate impact on the bottom-line hiring of Black and Hispanic candidates for entry-level firefighter positions in Massachusetts. Pursuant to Fed.R.Civ.P. 23(a) and (b)(2), the Court certified the plaintiff firefighter and police officer classes on March 24, 2006. The Court certified the firefighter class as “[a]ll minorities (Black and Hispanic) who took the civil service examination for the position of fire fighter within the Commonwealth of Massachusetts in the years 2002 and 2004.” (Docket No. 81.) A six-day bench trial for the liability phase of the firefighter class began on April 11, 2006, and the parties rested on May 4, 2006. The following witnesses testified for the plaintiffs: Dr. Frank Landy, an expert in industrial psychology and statistics and a former consultant to the HRD; Dr. Joel Wiesen, an expert in industrial psychology and statistics and the HRD Chief of Test Development and Validation between 1977 and 1993; Elizabeth Dennis, the former HRD Director of the Civil Service Unit between the mid-1990s and 2003; and Kevin Bradley, a Lynn firefighter since 1977 and the father of two of the named plaintiffs. The following witnesses testified for the HRD: Dr. Rick Jacobs, an expert in industrial psychology and statistics and a former and current consultant to the HRD; Marc Chavanne, the HRD Deputy Director of Selection and Validation between 1991 and 1996; and Sally McNeely, the current HRD Director of the Organizational Development Group, since 2003. The parties submitted closing briefs, and oral argument was held on June 9, 2006. After trial, oral argument, and review of the post-trial submissions, the Court holds that the written civil service cognitive ability examinations used in 2002 and 2004 have an adverse and disparate impact on the employment opportunities of Black and Hispanic candidates for entry-level firefighter positions, and that the selection process that uses the examination scores to rank candidates is not job related and consistent with business necessity under applicable federal law and the longstanding Beecher decree. The plaintiffs have also demonstrated that there are alternative selection methods with less discriminatory effects that would serve the employer’s legitimate interest in selecting capable firefighters based in part on cognitive ability. Accordingly, I conclude that judgment on liability should enter in favor of the plaintiff firefighter class. II. FINDINGS OF FACT A. Statutory and Administrative Framework The civil service law currently applies to the fire departments of approximately 110 municipalities in Massachusetts, including Boston and Lynn. To become a firefighter in a municipality where the civil service law applies, an individual must first pass a statewide civil service examination. See Mass. Gen. Laws ch. 31, §§ 6, 59. The personnel administrator for the HRD (the “HRD Administrator”) conducts, determines the form, method, and subject matter of, and develops the examinations, id. §§ 5(e), 16; prepares and posts notices of the examinations, id. §§ 18-19; and determines the passing requirements, id. § 22. Based on the examination results, the HRD Administrator ranks the names of those who pass on the “eligible list” based on the following statutory priority. Id. §§ 25, 26. The names of persons who pass examinations for original, appointment to any position in the official service shall be placed on eligible lists in the following order: (1) disabled veterans, in the order of their respective standings; (2) veterans, in the order of their respective standings; (3) widows or widowed mothers of veterans who were killed in action or died from a service connected disability incurred in wartime service, in the order of their respective standings; (4) all others, in the order of their respective standings. Id. § 27. To hire for a firefighter vacancy, a municipality’s appointing authority submits a request to the HRD Administrator, who then certifies “from the eligible list sufficient names of persons for consideration” in rank order. Id. § 6. In addition' to the statutory priority, the appointing authority may have the HRD Administrator rank residents ahead of non-residents, see id. § 58, and may request special certification lists for candidates with certain qualifications, such as Spanish-language abilities (Ex. 9, at 16 (Pers. Admin. Rule 8(4))). The HRD and the HRD Administrator (the “State defendants”) have interpreted the civil service law as giving them discretion to decide how many names should be certified from the eligible list. Once the HRD Administrator certifies a list to a municipality, each candidate must sign the certified list and express a willingness to accept employment in order to be considered for appointment. Mass. Gen. Laws ch. 31, § 25. According to the HRD’s Personnel Administration Rules, the appointing authorities for a municipality “may appoint only from among the first 2n + 1 persons named in the certification willing to accept appointment,” where “n” is the number of vacancies. (Ex. 9, at 16-17 (Pers. Admin. Rule 9(1)).) For five vacancies, for example, a municipality may only appoint from the first eleven named in the certification willing to accept. (See id.; see also Exs. 33P, 33Q (stating on certification lists that selection “must be” within first 2n + 1 “who will accept”).) In evaluating the candidates within the “2n + 1” pool, a municipality may establish its own hiring criteria, such as a drug test, a background check, or an interview. After conditional offers are made by the municipality, the HRD administers a pass/ fail physical abilities test. Some municipalities conduct a full medical or psychological examination. According to Ms. McNeely, some municipalities use these post-certification hurdles to determine the composition of the “2n + 1” pool. None of the post-certification hurdles, however, changes a candidate’s ranking on the list. By statute, to bypass higher-ranked individuals on the certified list to hire lower-ranked individuals, a municipality must submit a written statement to the HRD Administrator justifying the bypass. Mass Gen. Laws ch. 31, § 27. The HRD Administrator has the right to review and withdraw any bypass appointment. MacHenry v. Civil Serv. Comm’n, 40 Mass.App.Ct. 632, 634-36, 666 N.E.2d 1029, 1030-31 (Mass.App.Ct.1996). B. Beecher Firefighter Litigation The civil service examination for firefighters has been the subject of employment discrimination litigation since the 1970’s. The Beecher class action was brought by the Boston Chapter, NAACP, on behalf of a statewide class of Black and Spanish-surnamed applicants for the firefighter position. Beecher, 371 F.Supp. at 509-10. In Beecher, although the available examination statistics were “meager,” after comparing the minority population and employment statistics, the district court concluded that the plaintiffs established a prima facie case that the written examination had a discriminatory effect on Blacks and Spanish-surnamed persons and that the defendants failed to demonstrate that the examination was substantially related to job performance. 371 F.Supp. at 514, 517. As a result, the district court issued a consent decree, which established certification quotas for minorities in “all cities and towns subject to Civil Service law” until “a city or town achieves a complement of [firefighter] minorities commensurate with the percentage of minorities within the community.” Id. at 522-23. Importantly, the decree ordered: The Massachusetts Division of Civil Service shall cease using written firefighter entrance examinations of the type administered by the Division of Civil Service in August 1971, for the purpose of determining qualifications for the selection of firefighters. Should the Division of Civil Service desire to utilize entrance examinations in the future for the purpose of selecting firefighters, such examinations shall be demonstrably job-related and validated in accordance with the “Guidelines on Employees Selection Procedures” issued by the Equal Employment Opportunity Commission, 29 C.F.R. § 1607.1 et seq., or otherwise shown to have no discriminatory impact. If the parties disagree as to whether a written examination has been shown to be valid within the meaning of the Guidelines, the question of their validity and job relatedness shall be resolved by the Court, and such resolution, whether by the parties’ agreement or by the Court, shall be accomplished before any such test is put into use for the purpose of qualifying or selecting. As with the instant study, the Court will scrutinize closely a future study which shows only a minimal level of job-relatedness. Id. at 521. The First Circuit affirmed the district court’s disparate impact findings in Boston Chapter, NAACP, Inc. v. Beecher, 504 F.2d 1017, 1021, 1026 (1st Cir.1974). With respect to the prima facie burden, the First Circuit stated: “Plaintiffs usually meet their initial burden by demonstrating that minority candidates have a higher test failure rate.” Id. at 1019. In addition, the First Circuit found that the validation study did not survive close scrutiny and affirmed the Beecher decree’s implementation of certification quotas that remained in effect for each local fire department “until that department attains sufficient minority fire fighters to have a percentage on the force approximately equal to the percentage of minorities in the locality.” Id. at 1024-28. Over the past thirty-plus years, municipalities have been released from the Beecher decree as their fire departments achieved racial parity with their populations. As of March 14, 2006, only nine of the 110 municipalities subject to the civil service law remain under the decree. (Ex. 8.) Since the Beecher decree ended in Lynn in 1986, Mr. Bradley has estimated that only four of the 106 entry-level firefighters hired in Lynn have been Black or Hispanic. Since the First Circuit held that the Beecher decree should no longer apply to Boston in Quinn v. City of Boston, 325 F.3d 18, 37 (1st Cir.2003), as of the time of trial, only seven of the 105 entry-level firefighters hired in Boston from the 2004 examination have been reported to be Black or Hispanic. (See Ex. 33D, at 2.) C. Firefighter Hiring from the 2002 and 2004 Civil Service Examinations The HRD administered a civil service examination for firefighters on April 27, 2002 (Ex. 28) and another on April 24, 2004 (Ex. 30). Both examinations contained one-hundred multiple choice questions testing only cognitive ability. The 2002 examination tested 4543 applicants, and the 2004 examination tested 2447 applicants. (See Ex. 5, Table 1; Ex. 6, Table 1.) The HRD used the passing point of seventy, an arbitrary number that has been used since at least 1971. See, e.g., Beecher, 371 F.Supp. at 511-12. The HRD adjusted scores after administering the examination both by removing questions and by crediting multiple answers as correct on questions so that the passing point of seventy produced no adverse impact on minorities under the EEOC Guidelines. (Ex. 10, at 3; Ex. 12, at 4-5; Trial Tr. 23, May 3, 2006; Trial Tr. 144-46, May 4, 2006.) Before putting the 2002 and 2004 examination results into use for hiring, the HRD provided adverse impact analyses to the court-appointed NAACP monitor for the Beecher decree. (Exs. 10, 12.) The analy-ses for both examinations showed no adverse impact at the passing point of seventy under the EEOC Guidelines; however, the analyses did show that minorities were adversely impacted at every score above seventy. (Ex. 10, Attach. I; Ex. 12, Attach. G.) While stating to the HRD that the “examination results reflect a significant adverse impact on Black and Hispanic candidates at scores higher than 70,” the NAACP monitor did not object before the HRD put the 2002 and 2004 examination results into use for firefighter hiring. (Ex. 13.) Based on the certified lists the HRD Administrator provided in response to requests from municipalities, 311 candidates were hired from the 2002 examination, and thus far, 200 candidates have been hired from the 2004 examination. {See Ex. 5, Table 1; Ex. 6, Table 1.) The overall hiring numbers in Massachusetts indicate that minority candidates have been hired less frequently than non-minority candidates: Civil Service Exam 2002 2004 Number of Minority Takers 655 502 Number Minorities Appointed 19 16 Minority Appointment Rate 3.4% 3.2% Number of Non-Minority Takers 3988 1945 Number of Non-Minorities 292 194 Appointed Non-Minority Appointment Rate 7.3% 10% Ratio of Appointment Rates 47% 32% Between Minorities and Non-Minorities {See Ex. 5, Table 1; Ex. 6, Table 1.) There will be further hiring from the pool of candidates who passed the 2004 examination. For example, Boston, which hired a class in June 2005 and January 2006, recently requested a new list of candidates based on the 2004 examination to hire for fifty vacancies. The HRD certified a list of candidates to Boston on April 11, 2006, and expanded it on April 21, 2006, providing a total of 156 candidates. The expanded list was produced to the plaintiffs on May 4, 2006, the last day of testimony. (Exs. 33M, 33Q.) Also on May 4, 2006, the HRD certified thirty-seven candidates in response to Lynn’s request to fill four vacancies. (Ex. 33P.) D. Creation of the Entry-Level Civil Service Examination for the Rank Ordering of Candidates Beginning in 1992 and continuing throughout the 1990’s, the HRD hired Landy, Jacobs and Associates, Inc. (“Lan-dy-Jacobs”) to develop the written cognitive examinations. Dr. Landy is now the plaintiffs expert; Dr. Jacobs is the HRD’s expert. While no longer affiliated, they remain good friends. In June 1992, Landy-Jacobs completed the Massachusetts Firefighter Final Validation Report (the “1992 Report”). (Ex. 27.) The 1992 Report was done under the direction of Dr. Landy and concluded that the written and physical examinations proposed by Landy-Jacobs for use in the selection of firefighters in Massachusetts were valid. Criterion-related validity evidence existed for both • the written and physical examinations from a study performed in 1986 for the Columbus, Ohio Fire Department. (Id. at 7.) The 1992 Report also documented a “high degree of demonstrated similarity between the job of firefighter in Massachusetts and the job of firefighter • in Columbus,” indicating the transportability of the written and physical examinations developed in Columbus, Ohio to Massachusetts. (Id. at 1-2.) In addition, the 1992 Report found that the written and physical examinations could validly be used for rank-order selection. (Id. at 5.) Importantly, Landy-Jacobs did not validate the written cognitive examination for rank ordering as a stand-alone mechanism; rather, the 1992 Report validated rank ordering only when the written examination constituted 40% and the physical examination constituted 60% of the overall composite score. (See Trial Tr. 5-9, Apr. 13, 2006 (“[W]hat we did was to Validate the procedure, which was a combination of a weighted cognitive ability and a weighted physical ability exam.”); Trial Tr. 8-9, May 4, 2006 (“[T] he weighting [Landy-Jacobs] recommended should be 60 percent physical and 40 percent cognitive.”).) Criterion-related validity studies, such as the 1986 Columbus, Ohio study, determine whether “the selection procedure is predictive of or significantly correlated with important elements of job performance.” 29 C.F.R. § 1607.5(B). The magnitude of the relationship is measured by calculating a correlation coefficient. Id. § 1607.14(B)(6). In the 1986 Columbus, Ohio study, Lan-dy-Jacobs found the correlation coefficient of the cognitive ability exam to be between 0.2 and 0.3 (Ex. 24, Tab 6, at 2; Ex. 27, Attach. A.) The physical abilities test was found to have a correlation coefficient between 0.3 and 0.4. (Ex. 27, Attach. 9.) These correlation coefficients were one of the reasons why Landy-Jacobs believed that the physical agility test should comprise 60% of the total score. Landy-Ja-cobs never validated the cognitive examination to be the exclusive basis for rank ordering. The HRD purchased the written cognitive examination from Landy-Jacobs, administered it on May 22, 1993, and weighted it at 40% of the overall score. (Ex. 24, Attach. 6, at 1, 9.) The other 60% of the overall score came from the physical examination, which the HRD developed with the help of Landy-Jacobs. E. Physical Examination The physical examination, which was paired with the 1993 written civil service examination, consisted of several timed events, such as a stair climb event that required candidates to make six trips up and down two flights of stairs carrying different pieces of equipment, a ladder event where a pulley mechanism replicated raising a ladder, and a rescue event that replicated crawling into a dark area to save a victim. (Trial Tr. 8-9, 14-16, May 4, 2006.) During this selection process, two factors delayed the HRD’s ability to issue certified lists for municipal vacancies. First, by placing the physical examination at the beginning of the hiring process as an initial screen rather than at the end as a final pass/fail hurdle, the number of applicants tested physically increased to almost 5000. This increase required the HRD to secure and construct additional testing sites, which it did. Second, the HRD had to suspend the administration of the physical examination after receiving medical complaints from eight candidates. An expert panel convened by the HRD concluded that the physical examination could resume with additional safeguards, including making water more accessible and giving more discretion to on-site emergency medical technicians to screen out candidates based on high blood pressure or heart rate. When the HRD resumed administering the physical examination, no further medical problems were reported. The average physical examination score of non-minority candidates was 95.55 and of minority candidates was 92.43. While 78.6% of non-minority candidates achieved one of the top two possible scores of 100 or 95, only 61.3% of minority candidates scored as well. (Ex. 35B.) Based on these results, the HRD concluded that administering the physical examination at the beginning of the hiring process did not “mitigate the adverse impact [of the written cognitive examination] as much as [the HRD] had hoped it would.” The HRD thus “saw no advantage in continuing this beyond 1994” and instead, decided to test for physical abilities at the end when making conditional job offers, enabling full-fledged medical evaluations and avoiding administrative time and costs. (Trial Tr. 68, May 3, 2006; Trial Tr. 24-27, May 4, 2006; Ex. 24, Attach. 6, at 9.) Dr. Landy testified, however, that the difference in physical examination scores was not statistically significant. As such, the use of the physical examination at the beginning of the hiring process, which decreased the weight of the written examination to 40% of the overall score, had the effect of diluting the adverse impact of the written examination. (Trial Tr. 44-47, Apr. 13, 2006.) Regardless, after 1993, the HRD ultimately decided to use only the written cognitive examination for rank ordering and to administer the physical examination at the end of the hiring process as one of the final pass/fail hurdles for entry-level firefighter candidates who already received conditional offers. F. Subsequent Job Analyses and Validation Studies In December 1995, Landy-Jacobs completed the Massachusetts Fire Departments Job Analysis and Physical Fitness Standards Test Development Report (the “1995 Report”). (Ex. 36.) The 1995 Report documented a job analysis and validation of physical fitness and medical standards for firefighters of all ranks. (Id. at 1.) The job analysis consisted of (a) developing a preliminary list of tasks by using job analyses conducted of fire departments in other cities; and (b) surveying a sample of Massachusetts firefighters to determine whether each task was performed by them, whether a firefighter would be responsible for performing the task on their first day, and the task’s importance and frequency. (Id. at 4-5.) The 1995 Report concluded that the physical fitness standards test was content-valid. (Id. at 11-12.) The development of the physical fitness test included using a sample of incumbent firefighters to set reasonable cut times for each event. (Id. at 9.) In June 2002, SHL USA, Inc. (“SHL”), the firm that purchased Landy-Jacobs, completed the Commonwealth of Massachusetts 2002 Entry-Level Firefighter and Police Officer Job Analysis Report (the “2002 Job Analysis”). (Ex. 38.) The 2002 Job Analysis updated and revised the job analysis from the 1996 Report. (Id. at 1.) SHL used the findings from the 1996 Report to form a preliminary task list, which SHL enhanced by referencing previously conducted job analyses for entry-level firefighters in other jurisdictions and by interviewing incumbents and supervisors from Massachusetts. (Id. at 5-6.) SHL matched the tasks with a set of cognitive and motor abilities based on a published ability taxonomy. (Id. at 7-8.) The HRD then administered a SHL-developed survey to determine each task’s importance and frequency and each ability’s relative importance. (Id. at 9-14.) The 2002 Job Analysis concluded that “the basic areas of responsibility for firefighters have changed very little over the years in terms of the importance of the duties relative to one another” and that “the abilities needed to succeed as a firefighter or police officer within the Commonwealth of Massachusetts have not changed substantially over time.” (Id. at 24.) “This suggests that the existing cognitive and physical examinations should continue to have great relevancy in their current form with only slight enhancements based on the 2002 job analysis results.” (Id.) Also in June 2002, SHL completed the Commonwealth of Massachusetts Entry-Level Firefighter and Police Officer Medical Standards Update Report (the “2002 Medical Update”). (Ex. 37.) The 2002 Medical Update documented an update and a content-validation of the then existing medical standards based on the 2002 Job Analysis. (See id. at 1-3.) G. Creation of the 2002 and 2004 Civil Service Examinations The HRD designed the 2002 and 2004 examinations in-house with the goal that they would be equivalent or comparable to the Landy-Jacobs examinations. (Ex. 10, at 1; Ex. 12, at 3.) No outside consultants assisted, and during that time, the HRD employed no industrial psychologists. Dr. Jacobs believes that the 2002 and 2004 examinations are similar to the examination created by Landy-Jacobs for Massachusetts in 1992, and Dr. Landy agrees that the 2002 and 2004 examinations appear to follow the “spirit” of the Landy-Jacobs 1992 examination. Dr. Landy also points out, however, that there is no evidence of the HRD analyzing the 2002 and 2004 examinations to ensure, for example, that the reading level of the new examinations did not exceed the reading level of the job or that the new questions were linked with the constructs intended to be tested. (See Ex. 1, at 22-23.) Beginning in June 2006, the HRD administered a new test, which it claims is state-of-the-art and will be validated. III. CONCLUSIONS OF LAW A. Title VII 1. Statutory Framework Title YII of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-2(a), prohibits not only overt and intentional discrimination but also “more subtle forms of discrimination, known as disparate impact discrimination.” EEOC v. Steamship Clerks Union, Local 1066, 48 F.3d 594, 600-01 (1st Cir.1995). The “disparate impact approach roots out ‘employment policies that are facially neutral in their treatment of different groups but that in fact fall more harshly on one group than another and cannot be justified by business necessity.’ ” Id. (citations omitted). Stated another way, the disparate impact approach prohibits employment “practices that are fair in form, but discriminatory in operation.” Griggs v. Duke Power Co., 401 U.S. 424, 431, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971). Section 2000e-2(a)(2) provides in relevant part: It shall be an unlawful employment practice for an employer — ... (2) to limit, segregate, or classify his employees or applicants for employment in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s race, color, religion, sex, or national origin. The Civil Rights Act of 1991 added a provision to make the burden of proof in disparate impact cases explicit: An unlawful employment practice based on disparate impact is established under this subchapter only if— (i) a complaining party demonstrates that a respondent uses a particular employment practice that causes a disparate impact on the basis of race, color, religion, sex, or national origin and the respondent fails to demonstrate that the challenged practice is job related for the position in question and consistent with business necessity; or (ii) the complaining party makes the demonstration described in subpara-graph (C) with respect to an alternative employment practice and the respondent refuses to adopt such alternative employment practice. 42 U.S.C. § 2000e-2(k)(l)(A). “The term ‘demonstrates’ means meets the burdens of production and persuasion.” Id. § 2000e(m). “The demonstration referred to by subparagraph (A)(ii) shall be in accordance with the law as it existed on June 4, 1989, with respect to the concept of ‘alternative employment practice,’ ” id. § 2000e-2(k)(l)(C), which the courts had defined as another selection device without a similar discriminatory effect that would also serve the employer’s legitimate interest. The plaintiff bears the burden of demonstrating the existence of alternative business practices. Int’l Bhd. of Elec. Workers v. Miss. Power & Light Co., 442 F.3d 313, 317-18 (5th Cir.2006). 2. Shifting Burden The First Circuit set forth the legal framework that applies to disparate impact cases in Steamship Clerks. [I]t is incumbent upon the plaintiff to demonstrate a prima facie case of discrimination. In the disparate impact milieu, the prima facie case consists of three elements: identification, impact, and causation. First, the plaintiff must identify the challenged employment practice or policy, and pinpoint the defendant’s use of it. Second, the plaintiff must demonstrate a disparate impact on a group characteristic, such as race, that falls within the protective ambit of Title VII. Third, the plaintiff must demonstrate a causal relationship between the identified practice and the disparate impact. When the plaintiff rests, declaring herself satisfied that she has established a prima facie case of disparate impact discrimination, the ball bounces into the defendant’s court. At that point, the defendant has several options. First, it may attack the plaintiffs proof head-on, debunking its sufficiency or attempting to rebut it by adducing countervailing evidence addressed to one or more of the three constituent strands from which the prima facie case is woven, [ ] asserting, say, that no identifiable policy exists, or that the policy’s implementation produces no disparate impact, or that the plaintiffs empirical claims-such as the claim of causation-are insupportable. Alternatively, the defendant may confess and avoid, acknowledging the legal sufficiency of the prima facie case but endeavoring to show either that the challenged practice is job-related and consistent with business necessity, [ ] or that it fits within one or more of the explicit statutory exceptions covering bona fide seniority systems, veterans’ preferences, and the like. In all events, however, a defendant’s good faith is not a defense to a disparate impact claim. If the defendant fails in its efforts to counter the plaintiffs prima facie case, then the factfinder is entitled-though not necessarily compelled, [ ] -to enter judgment for the plaintiff. On the other hand, even if the defendant stalemates the prima facie case by elucidating a legitimate, nondiscriminatory rationale for utilizing the challenged practice, the plaintiff may still prevail if she is able to establish that the professed rationale is pretextual. The plaintiff might demonstrate, for example, that some .other practice, without a similarly undesirable side effect, was available and would have served the defendant’s legitimate interest equally well. Such an exhibition constitutes competent evidence that the defendant was using the interdicted practice “merely as a ‘pretext’ for discrimination.” 48 F.3d at 601-02 (citations and footnotes omitted). While Steamship Clerks addressed the legal framework as it existed prior to 1991, the First Circuit has applied the same framework in the context of the Civil Rights Act of 1991. See Donnelly v. R.I. Bd. of Governors for Higher Educ., 110 F.3d 2, 4 (1st Cir.1997). As part of the prima facie case, the plaintiffs must demonstrate that the civil service examinations have both an adverse and disparate impact. Specifically, the plaintiffs must demonstrate that the adverse’ effects of the practice fall more heavily on members of the protected class than they fall on nonmembers who are similarly situated. Steamship Clerks, 48 F.3d at 601. With respect to the causation prong, the Supreme Court has stated that: [T]he plaintiff must offer statistical evidence of a kind and degree sufficient to show that the practice in question has caused the exclusion of applicants for jobs or promotions because of their membership in a protected group. Our formulations, which have never been framed in terms of any rigid mathematical formula, have consistently stressed that statistical disparities must be sufficiently substantial that they raise such an inference of causation. Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 994-95, 108 S.Ct. 2777, 101 L.Ed.2d 827 (1988) (plurality); see Wessmann v. Gittens, 160 F.3d 790, 804 (1st Cir.1998) (“Even strong statistical correlation between variables does not automatically establish causation.”). The Second Circuit has summarized: Because statistical analysis, by its very nature, can never scientifically prove discrimination, a disparate impact plaintiff need not prove causation to a scientific degree of certainty. Accordingly, this Court has held that a plaintiff■ may establish a prima facie case of disparate impact discrimination by proffering statistical evidence which reveals a disparity substantial enough to raise an inference of causation. That is, a plaintiffs statistical evidence must reflect a disparity so great that it cannot be accounted for by chance. EEOC v. Joint Apprenticeship Comm. of the Joint Indus. Bd. of the Elec. Indus., 186 F.3d 110, 117 (2d Cir.1999) (citations omitted). However, causation need not “invariably include a formal statistical analysis.” Steamship Clerks, 48 F.3d at 606. Where the employment practice at issue dispositively excludes individuals, some courts have observed that the disparate impact and causation elements appear to merge. See Nash v. Consol. City of Jacksonville, 905 F.2d 355, 358 (11th Cir.1990) (finding that “the fact that an examinee’s failure to pass the examination absolutely bars promotion satisfies the Court’s ‘specific causation’ requirement”); cf. Phillips v. Cohen, 400 F.3d 388, 398 n. 8 (6th Cir.2005) (“If the employee challenges the employer’s promotion process as a whole, however — as is the case here — then the disparate impact and causation elements merge.”). 3. Testing Caselaw The starting point for analysis of the Title VII claim is the seminal case Connecticut v. Teal, 457 U.S. 440, 102 S.Ct. 2525, 73 L.Ed.2d 130 (1982), in which the Supreme Court reaffirmed that Title VII prohibits “procedures or testing mechanisms that operate as ‘built-in headwinds’ for minority groups.” Id. at 448-49, 102 S.Ct. 2525 (quoting Griggs, 401 U.S. at 432, 91 S.Ct. 849). Commenting on congressional concern about the widespread use by state and local governmental agencies of invalid selection techniques that had a discriminatory impact, the Supreme Court stated: In considering claims of disparate impact under [Title VII], this Court has consistently focused on employment and promotion requirements that create a discriminatory bar to opportunities. This Court has never read [Title VII] as requiring the focus to be placed on the overall number of minority or female applicants actually hired or promoted. Id. at 450, 91 S.Ct. 849. In Teal, the pass-fail examination for a public agency had a disparate impact on minorities’ eligibility for promotion but no effect on the bottom-line promotion statistics because of an affirmative action program for promoting those who passed. See id. at 443-44, 102 S.Ct. 2525. Rejecting the agency’s “bottom-line defense,” the Supreme Court admonished, “[t]he suggestion that disparate impact should be measured only at the bottom line ignores the fact that Title VII guarantees these individual respondents the opportunity to compete equally with white workers on the basis of job-related criteria.” Id. at 451, 102 S.Ct. 2525; cf. Donahue v. City of Boston, 304 F.3d 110, 119-20 (1st Cir.2002) (stating that standing to assert non-Title VII equal protection claim for prospective relief is established if plaintiff is denied “the opportunity to compete on equal footing in the [ ] hiring process on account of his race”). When the Supreme Court first held that the Civil Rights Act proscribes disparate impact discrimination in Griggs, the focus was on barriers to employment rather than who was hired. [T]he Act does not command that any person be hired simply because he was formerly the subject of discrimination, or because he is a member of a minority group.... What is required by Congress is the removal of artificial, arbitrary, and unnecessary barriers to employment when the barriers operate invidiously to discriminate on the basis of racial or other impermissible classification. Griggs, 401 U.S. at 430-31, 91 S.Ct. 849. Thus, under Teal and its progeny, individual components of a hiring process may constitute separate and independent employment practices subject to Title VII even if the overall decision-making process does not disparately impact the ultimate employment decisions involving a protected group. See, e.g., Stout v. Potter, 276 F.3d 1118, 1122 (9th Cir.2002) (“The no-nadverse results of the ultimate promotion decisions cannot refute a prima facie case of disparate impact at the dispositive interview selection stage.”); Smith v. Xerox Corp., 196 F.3d 358, 370 (2d Cir.1999) (“Even if the overall decision-making process did not create an adverse impact on a protected group, that group still has a cause of action if it can show that some component of the decision-making process caused a disparate impact.”); Newark Branch, NAACP v. City of Bayonne, N.J., 134 F.3d 113, 124 (3d Cir.1998) (“Teal suggests that a subsequent affirmative action program cannot ‘redeem’ discriminatory conduct that produces disparate results.”). Courts have applied Teal to reject examinations used to rank-order candidates. For example, in Waisome v. Port Auth. of N.Y. & N.J., 948 F.2d 1370, 1378 (2d Cir.1991), involving a composite score of tests to rank police officers for promotion, the Second Circuit held: Moreover, our prior case law lends support to the use of the [effective cutoff score]. Where a written test served, as here, both as a passing “gate” to further consideration for promotion, and as a major component of the ultimate score required for promotion, we indicated there was no disparate impact in the pass rate, but the disparity in actual promotions established that the written test had a prohibited disparate impact. In [Kirkland v. New York State Dep’t of Correctional Servs., 711 F.2d 1117 (2d Cir.1983)], as in the present case, evidence demonstrated that, though there was no disparity in the rate at which minority candidates for promotion passed an examination, their representation on the eligibility list was disproportionately low at the top of the list and high at its bottom. Hence, remand is required for the district court to develop a full record against which to evaluate the evidence of bunching and to determine whether the written examination had a disparate impact when these statistics and all the surrounding facts and circumstances are considered. Id. at 1378 (citations omitted) (finding that written test served as pass-fail mechanism requiring score of sixty-six to move on in hiring process and as ranking mechanism requiring score of seventy-six to be hired, both of which constituted employment practices under Teal). Thus, when an examination is a ranking mechanism that dictates whether and when passing candidates are reached for consideration, the Court must determine whether it is a gateway that has a disparate impact on minority hiring. Teal does not necessarily require that courts deconstruct employment practices into their individual components and evaluate each for disparate impact. Recognizing the potential burden of such a requirement, some courts have interpreted Teal as applying Title VII protection to a component of an employment practice regardless of the “bottom-line” only if that component is an identifiable and dispositive barrier that denies an employment opportunity by preventing an individual from proceeding to the next step in the employment process. See, e.g., District Council 37, AFL-CIO v. N.Y. City Dep’t of Parks & Recreation, 113 F.3d 347, 352-54 (2d Cir.1997) (finding that plaintiffs could challenge only “dispositive step” under Teal when bottom-line was nondiscriminatory); City of Chicago v. Lindley, 66 F.3d 819, 829 (7th Cir.1995) (finding Teal applicable only where “one ‘step’ disparately excluded minority individuals from moving on to the next step and, in turn, deprived them of any opportunity for benefits”); Reynolds v. Ala. Dep’t. of Transp., 295 F.Supp.2d 1298, 1315 (M.D.Ala.2003) (finding Teal inapplicable because examinations did not work as absolute barrier to further hiring considération and because there was no Title VII issue before court). Similarly, the EEOC Guidelines provide: C. Evaluation of selection rates. The “bottom line”.... If this information shows that the total selection process does not have an adverse impact, the Federal enforcement agencies, in the exercise of their administrative and prosecutorial discretion, in usual circumstances, will not expect a user to evaluate the individual components for adverse impact, or to validate such individual components, and will not take enforcement action based upon adverse impact of any component of that process, including the separate parts of a multipart selection procedure or any separate procedure that is used as an alternative method of selection. See 29 C.F.R. § 1607.4(C) (emphasis added). 4. Statistical Evidence a. Prima Facie Case In evaluating statistical evidence, “[t]he Supreme Court has said that no single test controls in measuring disparate impact.” Langlois v. Abington Hous. Auth., 207 F.3d 43, 50 (1st Cir.2000) (citing Watson, 487 U.S. at 995-96 n. 3, 108 S.Ct. 2777 (plurality) (“[W]e believe that such a case-by-case approach properly reflects our recognition that statistics ‘come in infinite variety and ... their usefulness depends on all of the surrounding facts and circumstances’ ” (citation omitted).)). In this case, the plaintiffs offer the “four-fifths rule” and chi-square analysis as the statistical benchmarks. The “four-fifths rule” comes from the EEOC’s Uniform Guidelines on Employee Selection Procedures (1978) (the “EEOC Guidelines”), which provide: A selection rate for any race ... 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 by the Federal enforcement agencies as evidence of adverse impact, while a greater than four-fifths rate will generally not be regarded by Federal enforcement agencies as evidence of adverse impact. 29 C.F.R. § 1607.4(D). For example, if a government agency promoted female candidates 20% of the time but male candidates 40% of the time, the selection rate for females would be 50% (or half) of the selection rate for males. This 50% ratio is less than 80% and thus, would violate the four-fifths rule and demonstrate adverse impact. The four-fifths rule is a pertinent benchmark in the employment context. See Langlois, 207 F.3d at 50. The Supreme Court has cautioned, however, that the rule “has not provided more than a rule of thumb for the courts.” Watson, 487 U.S. at 995 n. 3, 108 S.Ct. 2777 (plurality) (citations omitted). Indeed, the EEOC Guidelines limit the rule’s applicability in two ways. First, “[s]maller differences in selection rate may nevertheless constitute adverse impact, where they are significant in both statistical and practical terms or where a user’s actions have discouraged applicants disproportionately on grounds of race, sex, or ethnic group.” 29 C.F.R. § 1607.4(D). Second, “[g]reater differences in selection rate may not constitute adverse impact where the differences are based on small numbers” that “are too small to be reliable.” Id. When numbers are “too small to be reliable,” the federal agencies that issued the EEOC Guidelines provided the following guidance: Generally, it is inappropriate to require validity evidence or to take enforcement action where the number of persons and the difference in selection rates are so small that the selection of one different person for one job would shift the result from adverse impact against one group to a situation in which that group has a higher selection rate than the other group. On the other hand, if a lower selection rate continued over a period of time, so as to constitute a pattern, then the lower selection rate would constitute adverse impact, warranting the need for validity-evidence. 44 Fed.Reg. 11996,11999 (Mar. 2,1979). At times, courts have also used chi-square calculations in determining the existence of disparate impact. See, e.g., NAACP v. City of Mansfield, Ohio, 866 F.2d 162, 167-69 (6th Cir.1989). Without describing the details of the calculation, chi-square is a statistical calculation that examines differences between what is expected and what is observed. For example, if a government agency hired 85 of 600 non-minority candidates (~ 14.2% hiring rate) and 15 of 400 minority candidates (~ 3.8% hiring rate), a chi-square calculation would indicate that the probability of the difference in hiring rates being due to chance is 0.236%. Because the probability due to chance is 0.236% and less than 5% in this example, there is “a very powerful indication that the process of selecting new hires acts disparately on [minorities] in the applicant pool.” See Walter B. Connolly, Jr., David W. Peterson & Michael J. Connolly, Use of Statistics In Eqtial Employment Opportunity Litigation § 4.05[4][a] (1996); see also id. § 8.04[l][b] (applying chi-square calculation to exam scores). If the probability due to chance is 5% or less, the difference is said to rise to the 0.05 level of statistical significance (i.e., 95% certainty that difference is due to nonrandom factors). See, e.g., Xerox, 196 F.3d at 366-67. b. Job Related and Consistent With Business Necessity In this case, the HRD asserts that a criterion-related study validates the written cognitive examination. Criterion-related studies examine the relationship between selection procedure scores and job performance. The selection procedures are valid if “the selection procedure is predictive of or significantly correlated with important elements of job performance.” 29 C.F.R. § 1607.5(B). The magnitude of the relationship is determined by calculating a correlation coefficient. Id. § 1607.14(B)(6). “A correlation coefficient of +1.0 indicates a complete identity between relative test scores and relative job performance.” Williams v. Ford Motor Co., 187 F.3d 533, 540 (6th Cir.1999). As Dr. Landy explained, a correlation coefficient of 0.0 indicates that the examination has no relationship with job performance. Correlation coefficients “around .30 — .40 are considered acceptable (by testing professionals) for tests used in selecting employees” (Ex. 24, Tab 6, at 2). See also Beecher, 371 F.Supp. at 516 (“[A]s a ‘rule of thumb’ a coefficient of .3 would be the minimum level to indicate a satisfactory relationship. A lower coefficient would not be practically significant and would not justify use of the test.”). Therefore, for a selection procedure to be valid under a criterion-related study, the correlation coefficient must generally exceed a 0.3 threshold. In addition to the 0.3 threshold, in measuring the correlation coefficient, the calculation must be “statistically significant at the 0.05 level of significance.” 29 C.F.R. § 1607.14(B)(5). Lastly, as explained by Dr. Jacobs, correlation coefficients of different selection procedures do not necessary add to one another. This is because there may be overlap when the different selection procedures correlate with overall job performance. For example, cognitive examinations and physical ability tests are both correlated with certain entry-level firefighting duties, such as search and rescue. (See Ex. 1, at 14.) Therefore, the correlation coefficient of a selection procedure that uses both a cognitive examination and a physical ability test cannot be calculated by simply adding the correlation coefficients of each component. B. Prima Facie Case The plaintiffs identify the HRD’s use of the entry-level civil service examination as the employment practice subject to their challenge under Title VII, asserting that it has an adverse and disparate impact with respect to the selection and consideration of minority candidates for hire as firefighters in specific municipalities and statewide. The HRD asserts that the Court should not consider the disparate impact of the examination on the scores of minorities because the plaintiffs have not demonstrated that the examination has caused a discriminatory impact on the hiring of minorities. 1. Examination Scores The statistical evidence shows clear differences in scores as a function of race for both the 2002 and 2004 examinations. On average, minority candidates score lower than non-minority candidates: African _Whites Americans Hispanics 2002 Examination 89.10 77.80_78.44 2004 Examination 88.10_77.00_78.70 (Ex. 33A.) The examination scores of minority candidates are disproportionately lower at all scores above the nominal passing score of seventy. (See Ex. 1, Tables 6-7.) In addition, all parties agree that to be hired in most communities, which are not applying for a special certification (e.g., Spanish-language ability) or subject to the Beecher certification quotas, non-veteran candidates must obtain a score above an effective cutoff score of ninety. (Ex. 24, at 68-69.) In other words, in most municipalities, any non-veteran candidate with a score under ninety who is hired is a rare bird. Accordingly, the following table demonstrates the passing rates and ratio of passing rates for the 2002 examination at the scores of seventy, which is the nominal passing score, and ninety, which is the effective cutoff score for non-veteran candidates in most communities. African Majority American 2002 Score Passing Rate Passing Rate Score_Rate_Rate 90_60.20%_19.93% 70_96.22%_76.20% Ratio of Ratio of African Other American Minority and Other and Majority Minority Majority Passing Rate Passing Rate Passing Rate Rates_Rate_Rates 33.11%_22.63%_37.60% 79.19%_77.96%_81.03% (See Ex. 1, Table 6; Ex. 10, Attach. I.) For the 2004 examination:_ Ratio of African American Non- and Non-protected African Protected Group American Group Hispanic 2004 Passing Passing Rate Passing Rate Passing Rate Score Rate Rate__Rates_Rate Ratio of Hispanic and Non-protected Group Passing Rates 90 52.47% 17.69% 3.71% 19.12% 36.45% 70 96.52% 75.55% 78.27% 80.36% 3.26% (See Ex. 1, Table 7; Ex. 12, Attach. G.) Under the four-fifths rule, there is no adverse impact on minorities (Hispanics and African-Americans) at the nominal passing score of seventy. (See Trial Tr. 39^40, Apr. 14, 2006.) It is undisputed, however, that the four-fifths rule is violated for every score greater than seventy. (See Ex. 1, Tables 6-7.) The examinations thus have a severe adverse and disparate impact on non-veteran minority candidates, who must score above the effective cutoff score of ninety to be hired in most communities. The chi-square calculations support these findings by demonstrating that the differences in scores by race are statistically significant. (See Ex. 1, Tables 8-9.) 2. Hiring Statistics Examination statistics are not determinative of the critical issue of whether the examination disparately precludes minority candidates from being hired. The plaintiffs argue that the 2002 and 2004 examinations have an adverse and disparate impact on the employment opportunities of these minority candidates, as demonstrated by an overall comparison of the percentage of candidates who passed the examination to the percentage of those hired. In rebuttal, the HRD urges the Court to take a more nuanced approach, insisting that these overall statistics improperly aggregate numbers statewide and gloss over the major impact of preferences for veterans and residents on minority hiring. The HRD believes that adverse impact must be determined by municipality because each is a separate hiring unit with different requirements and separate appointing authorities. The Court finds merit in some of the HRD’s arguments. The statutory framework in Massachusetts gives candidates with veteran status priority over those without, regardless of examination score. This distinction between veterans and non-veterans is important because there is a lower percentage of minorities in the veteran candidate pool. 2002 2004 _Exam Exam Minority Candidates With Veteran 83 78 _Status_ _Minority Candidates Total_1845 864 % of Minority Candidates With 4.5% 9.0% _Veteran Status_ Non-Minority Candidates With 824 822 _Veteran Status_ Non-Minority Candidates Total_10052 6165 % of Non-Minority Candidates With 8.2% 13.3% _Veteran Status_ (Ex. 331, at 5-6.) Therefore, to ensure that apples are compared with apples, the Court evaluates veterans and non-veterans separately. In addition, the Court finds that the statewide aggregated approach advocated by the plaintiffs and the disaggregated municipality approach advocated by the HRD are both useful in this case. As Dr. Jacobs observes, determining the appropriate level of hiring analysis “is a complicated question that statisticians would argue about, but my opinion is that when we look at what happens in hiring, we try and replicate the decision process.” (Trial Tr. 100-02, Apr. 13, 2006.) The decision process in hiring firefighters in Massachusetts, however, contains multiple steps that span both the state and municipal levels. See Bradley v. City of Lynn, 403 F.Supp.2d 161 (D.Mass.2005) (holding HRD to be Title VII employer). Therefore, the Court considers hiring statistics on both statewide and municipality levels. Lastly, the Court does not compare hiring data from municipalities subject to the Beecher certification quotas with those that have been released from the decree. Indeed, the parties focused on the hiring statistics of only non-Beecher municipalities, and I do so as well. a. Statewide Aggregated Hiring Data i. Candidates With Veteran Status The following table summarizes the statewide veteran hiring statistics for non-Beecher municipalities, excluding data from municipalities that did not appoint anyone or had no minority candidates. Number of Candidates Taking Examination Number Appointed Selection Ratio Minorities (2002) 37 4 10.8% Non-Minorities (2002) 187 67 35.8 Minorities (2004) 45 12 26.7% Non-Minorities (2004) 275 127 46.2% (See Ex. 33E, Table 2; Ex. 33F, Table 2.) Under the four-fifths rule, the disparate impact ratio is 30% for the 2002 examination and 58% for the 2004 examination, evidencing adverse impact (i.e., less than 80%). The chi-square calculations support these findings by demonstrating that the differences in scores by race are statistically significant. (See Ex. 33E, at 2; Ex. 33F, at 3.) ii. Candidates Without Veteran Status The following table summarizes the statewide non-veteran hiring statistics for non-Beecher municipalities, excluding data from municipalities that did not appoint any non-veterans. Number of Candidates Taking Examination Number Appointed Selection Ratio Minorities (2002) 210 2.9% Non-Minorities (2002) 2561 152 5.9% Minorities (2004) 4.8% Non-Minorities (2004) 456 41 9.0% Under the four-fifths rule, the disparate impact ratio is 49% for the 2002 examination and 54% for the 2004 examination, evidencing adverse impact (i.e., less than 80%). (See Ex. 5, Table 3; Ex. 6, Table 3.) b. Disaggregated Hiring Statistics The HRD argues that statistical evidence aggregated on a statewide basis is flawed because hiring decisions are made individually by each municipality at the local level and because a large municipality can distort the whole adverse impact analysis. Dr. Jacobs asserts that aggregation “allows one single jurisdiction to have a major influence on the entire system leading to a conclusion of adverse impact for the entire Commonwealth when a single community is responsible for the outcome.” (Ex. 33S.) Specifically, Dr. Jacobs argues that Boston’s hiring distorts the calculation, because if it is remedied, the state passes the four-fifths rule. Because this seems reasonable, the Court examines the disaggregated hiring data of Boston. The Court also examines the disaggregated hiring data of Lynn, the municipality of the four named plaintiffs. The data shows that the examination has a disparate impact on the hiring of minorities in both of these municipalities. i. Boston The following table summarizes the Boston hiring statistics for the 2004 examination of candidates with veteran status. _Non-Minorities Minorities Number of Candidates_177_27 Number Appointed_90_7 Appointment Ratio_51%_26% (See Ex. 33D, at 2.) Under the four-fifths rule, the disparate impact ratio is 51%, evidencing adverse impact (i.e., less than 80%). The chi-square calculations support these findings by demonstrating that the differences in scores by race are statistically significant. (See id.) It is also important to note that Boston has hired two separate classes of entry-level firefighters from the 2004 examination-one in June 2005 and one in January 2006. The following table summarizes in which class candidates were hired as a function of race._ June January 2005 2006 _Class Class _Non-Minorities_50_40 Non-Minorities Hired in Class / 55% 45% Non-Minorities Hired Total_ _Minorities_2_5 Minorities Hired in Class / 28% 72% _Minorities Hired Total_ (See Ex. 33D, at 3.) Under the four-fifths rule, the disparate impact ratio for the first June 2005 class is 51%, evidencing adverse impact (i.e., less than 80%). Stated plainly, white veterans were twice as likely to be appointed in the first June 2005 class as minority veterans. (Id.) ii. Lynn The data from Lynn, where the named plaintiffs reside, are less clear. Since 1986, when Lynn was released from the Beecher decree, it is estimated that only four of the 106 entry-level firefighters have been minorities. The following table summarizes the Lynn non-veteran hiring statistics for the 2004 examination. _Non-Minorities Minorities Number of Candidates_69_15 Number Appointed_12_0 Appointment Ratio_17.4%_0.0% (See Ex. 33, at 17.) As a statistical matter, the experts disagree on whether a disparate impact ratio can be calculated under the four-fifths rule when Lynn hired zero minorities. That may be an interesting mathematical quandary, but in this case, the “fine tuning of the statistics could not have obscured the glaring absence of minority” hires. Int’l Bhd. of Teamsters v. United States, 431 U.S. 324, 342 n. 23, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). The Court concludes that the 2004 examination has had a disparate impact on the hiring of non-veterans in Lynn. However, the record is poor on the impact of the 2004 examination on veterans. The record does not contain a summary of Lynn hiring for the 2004 examination of candidates with veteran status. Thus, the record is too sparse to tell if there has been a disparate impact on hiring veteran minority candidates in Lynn from the 2004 examination. The record is equally unclear with respect to all candidates from the 2002 examination. According to Dr. Jacobs’ summary statistics, Lynn hired eight candidates from the 2002 examination; all were paramedics without veteran status, but none were minorities. (Ex. 32, at 15.) Since all were non-veteran paramedics, this would appear to be a class hired pursuant to a special certification request for emergency medical technicians. However, nothing in the record indicates whether the class was the result of a special certification, and the raw certification data is again unhelpful as it conflicts with Dr. Jacobs’ summary data. According to the raw certification data, Lynn hired twelve candidates from the 2002 examination, some of whom were not veterans or paramedics. (See Ex. 17.) Thus, the record is too sparse to evaluate the disparate impact on hiring in Lynn from the 2002 examination. iii. Other Municipalities The HRD contends that in most communities, the number hired is too small to determine whether there has been a disparate impact under the four-fifths rule or to be statistically significant. Disparate impact in hiring is not found in some municipalities because the numbers are “too small to be reliable” under the four-fifths rule. (See Ex. 331, at 3-4.) For example, if Haverhill had hired one additional minority from the 2002 examination, its hiring statistics would change from the presence to the absence of adverse impact under the four-fifths rule. (See id. at 3 (showing minority shortfall under four-fifths rule to be less than one for Haver-hill).) While it is true that in some municipalities there is no disparate impact, based on the testimony of Dr. Landy and Dr. Wiesen, the plaintiffs have demonstrated a reasonable statistical basis to address the “small numbers” problem by aggregating non-Beecher communities with minority applicants to determine whether there was an adverse impact on minority hiring. Given the statutory framework mandating the HRD’s involvement in the hiring process across municipalities, the aggregation approach is supported in this case by the EEOC Guidelines and the caselaw. 29 C.F.R. § 1607.4(D); Vulcan Pioneers, Inc. v. N.J. Dep’t of Civil Serv., 625 F.Supp. 527, 534-35, 544-45 (D.N.J.1985) (finding aggregation across municipalities and across years appropriate where State administered firefigh