DECISION and ORDER
TELESCA, District Judge.
Plaintiffs, all former employees of Xerox Corporation (“Xerox”) in the Information Management function of its United States Customer Operations and Corporate Strategic Services groups, bring these actions pursuant to Title VII of the Civil Rights Act, 42 U.S.C. § 2000e-2,
et seq.,
the Age Discrimination in Employment Act (“ADEA”), 29 U.S.C. § 621,
et seq.
and the New York State Human Rights Law, alleging that they were redeployed and dismissed from their positions on the basis of their ages and gender.
On February 1, 1996, defendant filed summary judgment motions seeking dismissal of plaintiffs’ disparate impact and disparate treatment discrimination claims. Thereafter, plaintiffs voluntarily dismissed their disparate treatment claims and instead decided to proceed solely on the theory of disparate impact. The parties agreed to defer further briefing and decision on the summary judgment motions and certified the following question to this Court for decision as the fact-finder:
Did Xerox’ redeployment and involuntary reduction-in-force policy as applied in the United States Customer Operations Group in March 1992 and in the Corporate Strategic Services Group in June 1992 disparately impact males employed in the Information Management subgroup of USCO and employees 40 years of age or older employed in the Information Management subgroup of USCO and CSS?
In order to determine whether Xerox’ actions had a disparate impact upon plaintiffs on the basis of their age and gender, this
Court conducted an evidentiary hearing on July 15 and 16, 1996, during which experts in the field of labor economics testified on behalf of plaintiffs and defendant. Xerox has also moved to strike plaintiffs’ expert’s testimony under the Supreme Court’s decision in
Daubert v. Merrell Dow Pharmaceuticals, Inc.,
509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed.2d 469 (1993).
Pursuant to Federal Rule of Civil Procedure 52(c), the following constitutes this Court’s findings of fact and conclusions of law.
FINDINGS OF FACT
I.
Xerox’ Redeployment and Dismissal Policy
A.
United States Customer Operations Group
In the early 1990s, faced with a sluggish economy and a changing technological environment which demanded enhanced computer use and expertise, Xerox decided to reduce its workforce and upgrade the technical skills of its remaining employees in order to remain competitive during the recessionary economy. A primary target of this corporate reorganization was the United States Marketing Group (“USMG”), which is currently known as the United States Customer Operations (“USCO”) group.
A voluntary reduction in force was implemented in late 1991 during which 385 USCO employees opted to leave Xerox. However, USCO still faced an employee surplus and a skills mix imbalance in some of its subgroups, including the Information Management (“IM”) subgroup. Affidavit of Patricia Wallington, ¶ 3 (“Wallington Aff.”).
Due to the rapidly changing computer technology, Xerox managers perceived that skill deficiencies had developed in USCO/IM to the point where it was necessary to upgrade the skills base in order to meet the changing needs of that division.
Management’s perceptions were confirmed through a skills assessment conducted by the consulting firm of Nolan, Norton and Company, which found, among other things, that the skills of USCO/IM employees were severely lacking in several critical areas important to the organization’s future. Id. ¶ 4. Xerox claims that in order to address these concerns, it attempted to upgrade the skills of the USCO/IM workforce through training and by selectively hiring employees with critical skills, such as software engineering. However, in early 1992, the need to upgrade employees’ skills and to meet budgetary constraints became acute and Xerox determined that 49 exempt employees in USCO/IM would be redeployed. Wallington Aff. ¶ 5. A skills and performance assessment was implemented separately for both managerial employees and exempt non-managers, referred to as individual contributors, to determine which of the USCO/IM employees would be selected for redeployment.
The managerial assessments were based on their last four Performance Appraisals (50%) and Role Model Manager Attributes (50%). The role model portion of the assessment was derived from an evaluation form filled out by the employee’s immediate manager which assessed 19 managerial attributes. This assessment was then reviewed by a second level manager to ensure accuracy and fairness. The points accumulated from the Role Model Attributes assessment were then added to the Performance assessment and all managers were slotted on a matrix based upon the overall assessment point total and tenure. Selection for redeployment was made in a pre-speeified cell order beginning with the lowest cell. According to Xerox, the longer an employee’s tenure, the higher the employee measured on the matrix and the less vulnerable that employee was to being designated for redeployment.
Individual contributor assessments were based on the last four performance appraisals (50%) and a skills assessment (50%). The point total for the skills assessment was derived from a Critical Skills Evaluation form filled out by the employee’s immediate manager. Again, a second level manager reviewed each assessment to ensure accuracy and fairness. The points accumulated from the Critical Skills Evaluation assessment was added to the performance appraisal points and all individual contributors were then slotted on a matrix based upon the overall assessment point total and tenure. As with managerial employees, selection for redeployment was made in a pre-specified cell order beginning with the lowest cell. Xerox claims that the longer an employee’s tenure, the higher the employee placed on the matrix and the less vulnerable that employee was to being designated for redeployment.
Of those employees assessed, 49 exempt employees in USCO/IM were placed on a redeployment list effective March 2, 1992, including ten managers out of the 67 assessed and 89 individual contributors out of the more than 300 assessed. Those employees chosen for redeployment were given 60 days within which to locate new positions within Xerox or face involuntary termination. At the conclusion of the 60-day period, 17 of the 49 employees selected for redeployment had located new positions within Xerox and the remainder of the redeployed employees were dismissed. Of the 17 plaintiffs in this action, 16 were selected for redeployment in USCO and were dismissed after failing to find alternative employment at Xerox.
B.
Corporate Strategic Services
Xerox’ determined that the Corporate Strategic Services (“CSS”) group also had the same technical deficiencies as USCO in its IM workforce. In 1992, CSS/IM provided computer support for Xerox’ engineering, manufacturing and logistics operations. Affidavit of Patricia Cusick ¶ 3.
However, due to changing information management technologies, CSS management determined that the skills base in its IM subgroup was outdated and. inadequate to meet CSS’ future goals and resource demands. Management’s perception of IM’s skills deficiencies were confirmed by an independent skills assessment conducted by Nolan, Norton- and Company, a 1990 and 1991 corporate audit and various other internal assessments.
Id.
As with ÚSCO, CSS first attempted to upgrade the skills of its existing IM employees. However, it ultimately determined that it would have to redeploy 57 of its exempt workforce and embarked on a selection process. CSS’ selection process differed slightly from USCO’s and in June 1992, 57 CSS/IM employees were selected for redeployment out of 250 total employees assessed. Of the 17 plaintiffs in this action, only one plaintiff, Edna Trumble, was selected for redeployment in CSS. The remainder of plaintiffs were redeployed through USCO’s selection process. Ms. Trumble was a Computer Operator who provided support for a mainframe computer system which was being phased out by CSS and is no longer in use at Xerox.
The CSS employees selected for redeployment were given 60 days within which to find alternative employment at Xerox. Át the conclusion of that 60-day period, 36 of the 57 redeployed IM employees had obtained another position at Xerox. The remaining 21 employees, including plaintiff Trumble, were dismissed after failing to find alternative employment at Xerox.
II.
Plaintiffs’ Expert
In support of their claim that Xerox’ redeployment and reduetion-in-force policy disparately impacted male employees in USCO and employees 40 years of age or older in USCO and CSS, plaintiffs offered the testimony of labor economist Dr. Marjorie Honig, Chair of the Department of Economics at Hunter College, City University of New York. Dr. Honig was tendered and received as an expert in the field Of labor economics without objection.
In conducting her analysis of Xerox’ redeployment and reduction-iri-force policy, Dr. Honig explained that Xerox employed a two-stage process which ultimately resulted in plaintiffs’ dismissal from their positions: (1) USCO/IM and CSS/IM employees were selected for redeployment in March 1992 and June 1992, respectively; and (2) USCO/IM and CSS/IM employees selected for redeployment who could not find alternative employment within Xerox within 60 days of their redeployment date were terminated. She testified that in selecting employees for redeployment, Xerox considered three factors: (1) skills assessments of the workers; (2)their recent and historical performance histories; and (3) tenure.
Dr. Honig identified the relevant pool of workers for statistical purposes as all employees employed in the USCO and CSS groups prior to the redeployments in March and June 1992. With respect to plaintiffs’ claim that Xerox’ policy of selecting employees for redeployment disparately impacted USCO/IM and CSS/IM employees 40 years of age and older, Dr. Honig compared the number of workers employed by USCO and CSS .to the number of employees over the age of 40 in both of those groups who were selected for redeployment. She then statistically analyzed this ratio to determine whether a difference in redeployment rates existed for employees over 40 years of age and those under 40 years of age and whether that difference was statistically significant. She employed the same analysis with respect to plaintiffs’ claims that: (1) Xerox’ dismissal of redeployed employees in USCO and CSS had a disparate impact upon employees 40 years and older; (2) Xerox’ selection of redeployed employees in USCO/IM disparately impacted males; and (3) Xerox’ dismissal of redeployed employees in USCO/IM disparately impacted males.
After conducting statistical analyses with respect to each of these claims, Dr. Honig concluded the following:
(1) the observed differences in the redeployment rates between employees 40 years of age and older and employees under 40 years of age in USCO/IM and CSS/IM would have occurred by chance less than one time in 100 in the absence of age discrimination;
(2) the observed differences in dismissal rates between employees 40 years of age and older and those under 40 years of age in USCO/IM and CSS/IM would have occurred by chance less than one time in 100 in the absence of age discrimination;
(3) the observed differences in redeployment rates between male and female employees at USCO/IM would have occurred by chance less than one time in 100 in the absence of gender discrimination; and
(4) the observed differences in dismissal rates between male and female employees at USCO/IM would have occurred by chance less than five times in 100 in the absence of gender discrimination.
In arriving at her conclusions, Dr. Honig relied on the results of two statistical tests, the “t-ratio” test and the “likelihood ratio” test. Both tests were applied separately to the same data set to statistically test two null hypotheses: (1) that there was no age-based disparity in the selection of USCO/IM and CSS/IM employees for redeployment and dismissal; and (2) there was no gender-based disparity in the selection of USCO/IM employees for redeployment and dismissal.
Dr. Honig conceded that her conclusions were based solely on the results of the “t-ratio” and “likelihood ratio” tests and that she did not perform a regression analysis to account for non-discriminatory factors which could have been responsible for the observed disparities,
i.e.,
the results of the employees’ skills assessments and their performance histories. She testified that it would have been improper to perform a regression analysis in this case because the skills assessment and performance history data were subjective opinions made by Xerox managers which were, therefore, inherently unreliable. She also doubted that performance histories were available for all redeployed employees, thus rendering that data set incomplete.
Dr. Honig also criticized the report proffered by Xerox’ expert labor economist, Dr. David Bloom, primarily on the ground that
Dr. Bloom performed a regression analysis using data that Dr. Honig considered to be wholly unreliable and incomplete. She also stated that Dr. Bloom’s identification of the relevant labor pooi was inappropriate. Dr. Honig claims that Dr. Bloom’s use of a data set consisting of only those USCO and CSS employees subject to selection for redeployment (instead of the entire USCO/IM and CSS/IM workforce) adversely affected his analysis with respect to plaintiffs’ age and gender claims.
Dr. Honig also disputed the validity of Dr. Bloom’s conclusions that Xerox’ redeployment and dismissal policies had no disparate impact on age or gender groups since the results of his “t-ratio” and “likelihood ratio” tests were consistent with her findings that the disparities observed could not have occurred by chance but for considerations of age or gender. In those cases where Dr. Bloom could not determine the statistical significance of disparities in dismissal rates, Dr. Honig attributed his failure to report results to his selection of the inappropriate labor pool. Dr. Honig also disagreed with Dr. Bloom’s separate analysis for managers and individual contributors. She testified that the usual division of labor for statistical purposes is based upon the Federal Labor Standard Act’s designation of exempt versus non-exempt workers. She stated that prior to Dr. Bloom’s analysis, she had not previously seen statistical analyses of labor pools divided into management versus non-management personnel.
III.
Defendant’s Expert ■
In support of its argument that its redeployment and reduction-in-force policies had no age or gender-based disparate impact based upon employees in USCO/IM and CSS/IM, Xerox offered the testimony of labor economist, Dr. David Bloom, the Acting Executive Director of the Harvard Institute for International Development and Professor of Economics at Columbia University. Dr. Bloom was tendered and received as an expert in the field of labor economics without objection.
After conducting “t-ratio” and “likelihood ratio” tests and multiple regression analyses, Dr. Bloom concluded that:
(1) Xerox’ selection of USCO/IM and CSS/IM management and non-management employees for redeployment and reduction-in-force had no age-based disparate impact upon those employees;
(2) Xerox’ selection of USCO/IM management and non-management employees for redeployment and reduction in force had no gender-based disparate impact upon those employees; and
(3) Dr. Honig’s conclusions were not supported by sound statistical methodology.
With respect to Dr. Honig’s report, Dr. Bloom testified that she used the wrong data set in that she defined the relevant labor pool as all employees in USCO and CSS and not only those employees subject to selection for redeployment in those groups. Dr. Honig’s data included senior management employees and West Coast employees who were not subject to redeployment and, therefore, inclusion of these employees in her analysis skewed her results. He testified that Dr. Honig also used improper comparisons in that she did not compare similarly situated employees,
i.e.,
those with comparable skills and performance histories.
Dr. Bloom disagreed with Dr. Honig’s conclusion that a regression analysis would have been inappropriate because certain analysis can be used to determine the validity of suspect data. He testified that with respect to skills assessments, he performed more than 20 multiple regression analyses including variables accounting for employee skills and performance histories and then tested the integrity of that data. Before utilizing the data to run his tests, Dr. Bloom examined the skills assessments forms, the role model evaluations forms and the individual contributor skills assessments forms. Satisfied that there were no obvious clerical errors in the data collected, he then measured disparities in redeployment and dismissal rates in USCO based upon age and gender. He compared the skills assessments of employees against their historical performance evaluations and found that those employees with favorable performance histories also had
Mgh skills assessments scores. Since past performance evaluations (for the four years prior to the 1992 redeployment) would not have been influenced by a corporate decision to downsize in 1992, he concluded that the positive correlation between favorable performance histories and high skills assessments indicated that the skills assessments scores were not biased and not inherently unreliable as Dr. Honig assumed.
Based upon the results of “t-ratio” and “likelihood ratio” tests, he concluded that statistically significant disparities existed with respect to age and gender and that chance factors could not account for these disparities. However, after conducting regression analyses, he concluded that, within a reasonable degree of statistical certainty, male employees in USCO and employees 40 years of age and older in USCO and CSS were actually given more favorable treatment.
Dr. Bloom also testified that it was not sound statistical methodology to rely solely upon the “t-ratio” or “likelihood ratio” tests since those tests can only determine whether chance factors accounted for a disparity and cannot isolate other criteria possibly responsible for the disparity. Dr. Bloom concluded, therefore, that Dr. Honig could not, within a reasonable degree of statistical certainty, render any competent opinion as to a causal factor for the disparity in redeployment and dismissal rates among IM employees. Although he conceded that his results on the “t-ratio” and the “likelihood ratio” tests were consistent with Dr. Honig’s, he explained that those tests are not designed to rule out neutral factors such as skills and performance histories in accounting for disparities in redeployment and dismissal rates.
With respect to Dr. Honig’s analysis of disparities in dismissal rates, Dr. Bloom noted that she defined the relevant labor pool as all employees employed in USCO and CSS instead of those employees subject to dismissal,
i.e.,
only those employees selected for redeployment. Dr. Bloom testified that Dr. Honig also ignored factors such as whether redeployed employees actually looked for jobs within Xerox, the intensity of their search, and which employees competed for which jobs.
CONCLUSIONS OF LAW
I.
Disparate Impact Claims
Disparate impact claims are those in which it is alleged that a facially neutral test or employment practice impacts a class of employees more harshly than another and cannot be justified by business necessity.
Int’l Brotherhood of Teamsters v. United States,
431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). Proof of discriminatory motive is not required in a disparate impact case.
Griggs v. Duke Power Co.,
401 U.S. 424, 91 S.Ct. 849, 28 L.Ed.2d 158 (1971).
A.
Gender Discrimination
A disparate impact gender claim will lie under Title VII where: (1) the plaintiff demonstrates that an employer uses a particular employment practice that causes a disparate impact on the basis of sex and the employer fails to demonstrate that the challenged policy is job-related for the position in question and consistent with business necessity; or (2) the plaintiff demonstrates that an alternative employment practice was available to alleviate the disparate impact of a policy and the employer has refused to adopt the alternative employment practice. 42 U.S.C. § 2000e-2(k)(l)(A).
Although Title VIPs prohibition against disparate impact discrimination is intended to remedy the discriminatory effects of a facially neutral employment policy, it does not guarantee an equal division of available jobs among various classes of employees. Accordingly, employers are free to employ qualified individuals regardless of their race, gender, sex or national origin, even if one of those groups is represented to a higher degree in the workforce than another group, as long as the differences in employment rates is not attributable to discriminatory conduct. 42 U.S.C. § 2000e-2(h), (j). Legitimate nondiscriminatory business justifications, therefore, may exist to account for disparities in the workplace.
In order to establish a
prima facie
case of disparate impact discrimination,
plaintiffs must identify the specific employment practices at issue and establish that those practices caused a disparate impact upon their group. Only where plaintiffs establish that the elements of an employer’s decision making process are unidentifiable can the entire decision-making process be analyzed as a unit. § 2000e-2(k)(l)(B)(i). Under the disparate impact theory, proof of a disparity is demonstrated through statistical analysis which compares the impact of a particular employment action on a protected class as compared to the impact upon qualified employees in the relevant labor pool. As the Supreme Court stated in
Wards Cove Packing Co. v. Atonio,
490 U.S. 642, 650-51, 109 S.Ct. 2115, 2121-22, 104 L.Ed.2d 733 (1989), which involved a disparate impact race claim brought under Title VII,
It is such a comparison — between the racial composition of the qualified persons in the labor market and the persons holding at-issue jobs — that generally forms the proper basis for the initial inquiry in a disparate-impact case. Alternatively, in cases where such labor market statistics will be difficult if not impossible to ascertain, we have recognized that certain other statistics — such as measures indicating the racial composition of ‘otherwise qualified applicants’ for at-issue jobs — are equally probative for this purpose, (citations omitted).
But, a simple statistical comparison between the number of employees in a protected class affected by an employment policy versus the number of persons employed in the workplace is an inappropriate measure of whether disparate impact exists:
[PJlaintiff does not make out a case of disparate impact simply by showing that, ‘at the bottom line,’ there is racial
imbalance
in the work force.
490 U.S. at 657, 109 S.Ct. at 2124-25 (emphasis supplied).
Once it is established that a statistically valid disparity exists which adversely affects members of a protected class, the plaintiffs must then demonstrate that a particular employment practice at issue caused the disparity:
[T]he plaintiffs’ burden in establishing a prima facie case goes beyond the need to show that there are statistical disparities in the employer’s work force. The plaintiff must begin by identifying the specific employment practice that is challenged ... Especially in cases where an employer combines subjective criteria with the use of more rigid standardized rules or tests, the plaintiff is in our view responsible for isolating and identifying the specific employment practices that are allegedly responsible for any observed statistical disparities.
Wards Cove,
490 U.S. at 656, 109 S.Ct. at 2124 (quoting
Watson v. Fort Worth Bank & Trust,
487 U.S. 977, 994, 108 S.Ct. 2777, 2788, 101 L.Ed.2d 827 (1988));
see NAACP v. Town of East Haven,
70 F.3d 219, 225 (2d Cir.1995)
{prima facie
case of disparate impact discrimination can be established through statistical evidence which reports a disparity that cannot be attributable to chance and is sufficiently substantial to raise an inference of causation).
B.
Age Discrimination
1.
Disparate Impact Claims Under the ADEA
Xerox argues that plaintiffs’ disparate impact age claim should be dismissed because such claims are not cognizable under the ADEA. Although the Second Circuit has held that disparate impact claims are cognizable under the ADEA,
see Maresco v. Evans Chemetics,
964 F.2d 106, 115 (2d Cir.1992);
Lowe v. Commack Union Free School District,
886 F.2d 1364, 1369 (2d Cir.1989),
cert. denied,
494 U.S. 1026, 110 S.Ct. 1470, 108 L.Ed.2d 608 (1990), Xerox argues that these holdings are questionable in light of a recent Supreme Court decision,
Hazen Paper v. Biggins,
507 U.S. 604, 113 S.Ct. 1701, 123 L.Ed.2d 338 (1993).
In
Hazen,
the plaintiff alleged that his termination violated the ADEA under a disparate treatment theory of discrimination, in part because it interfered with the vesting of his pension rights. In discussing disparate treatment claims under the ADEA, the Court stated that “[disparate treatment ... captures the essence of what Congress sought to prohibit in the ADEA.” 507 U.S. at 610, 113
S.Ct. at 1706. However, the Court declined to decide whether disparate impact claims are cognizable under the ADEA, stating that:
We have never decided whether a disparate impact theory of liability is available under the ADEA ... and we need not do so here. ■
Id.
Relying upon that language, some circuits have held that disparate impact claims are not cognizable under the ADEA.
Furr v. Seagate Technology, Inc.,
82 F.3d 980 (10th Cir.1996);
Ellis v. United Airlines,
73 F.3d 999 (10th Cir.),
cert. denied,
— U.S. -, 116 S.Ct. 2500, 135 L.Ed.2d 191 (1996);
DiBiase v. SmithKline Beecham Corp.,
48 F.3d 719, 732-34 (3d Cir.),
cert. denied
, — U.S. —, 116 S.Ct. 306, 133 L.Ed.2d 210 (1995);
EEOC v. Francis W. Parker School,
41 F.3d 1073 (7th Cir.1994),
cert. denied,
— U.S. -, 115 S.Ct. 2577, 132 L.Ed.2d 828 (1995). However, others have specifically held or have implied that disparate impact claims are allowed under the ADEA.
Graffam v. Scott Paper Co.,
870 F.Supp. 389, 393-94 (D.Me.1994), aff
'd,
60 F.3d 809 (1st Cir.1995);
Lyon v. Ohio Education Association & Professional Staff, 53
F.3d 135, 139 n. 5 (6th Cir.1995);
Houghton v. SIPCO, Inc.,
38 F.3d 953, 958 (8th Cir.1994);
Abbott v. Federal Forge, Inc.,
912 F.2d 867, 872 (6th Cir.1990).
The Second Circuit has not directly addressed the issue of whether a disparate impact claim is allowable under the ADEA in light of
Hazen
and the district courts in this Circuit to reach the issue have followed the Circuit’s precedent allowing such a claim.
See, e.g., Krueger v. New York Telephone Co.,
1993 WL 276058 (S.D.N.Y. July 21, 1993). Until this Circuit rules otherwise, this Court is bound to follow Second Circuit precedent allowing disparate impact claims under the ADEA. Accordingly, Xerox’ motion to dismiss plaintiffs’ disparate impact age claims on this basis is denied.
2.
Analysis of ADEA Disparate Impact Claims
In order to establish a
prima facie
case of disparate impact age discrimination, plaintiffs must demonstrate that a specific employment practice caused a disparate impact on their class,
i.e.,
employees in USCO/IM and CSS/IM, 40 years of age or older. As with disparate impact claims under Title VII, the plaintiffs’ proof on causation must demonstrate a disparity between the impact of a specified employment practice upon them versus the impact of the policy on qualified individuals in the relevant labor market.
See Wards Cove,
490 U.S. at 650, 109 S.Ct. at 2121;
see also Lowe,
886 F.2d at 1370 (discussing requirements for
prima facie
case in disparate impact claim under the ADEA). Statistical proof of disparate impact is required and the Supreme Court’s discussion of the probative value of statistical proof as set forth in
Wards Cove,
is equally applicable to disparate impact claims brought under the ADEA.
C.
Xerox’ Motion in Limine
Xerox argues that this Court need not determine whether plaintiffs have established a
prima facie
case of disparate impact age discrimination since the testimony of their statistical expert, Dr. Marjorie Honig, should be ruled inadmissible under the Supreme Court’s decision in
Daubert v. Merrell Dow Pharmaceuticals, Inc.,
509 U.S. 579, 113. S.Ct. 2786, 125 L.Ed.2d 469 (1993).
In
Daubert,
the Supreme Court defined a trial court’s role with respect to determining whether expert testimony should be deemed admissible during trial:
Faced with a proffer of expert scientific testimony, then, the trial judge must then determine at the outset, pursuant to Rule 104(a), whether the expert is proposing to testify to (1) scientific knowledge that (2) will assist the trier of fact to understand or determine a fact in issue. This entails a preliminary assessment of whether the reasoning or methodology underlying the testimony is scientifically valid and of whether that reasoning or methodology properly can be applied to the facts in issue.
Id.
at 591, 113 S.Ct. at 2796. The Supreme Court found that trial judges should act as “gatekeepers” in preliminarily assessing the admissibility of expert opinion to protect juries from considering proof which is unrelia
ble, confusing or irrelevant to fact finding.
Id.
at 589, 597, 113 S.Ct. at 2795, 2799. In this ease, however, the parties have agreed that this Court, and not a jury, shall act as the fact-finder and determine whether plaintiffs have established a
prima facie
case.
Under the standards set forth in
Daubert
and Federal Rule of Evidence 702, the relevancy of Dr. Honig’s testimony to the facts in issue is clear. Dr. Honig testified regarding the statistical analyses she performed and her conclusion that Xerox’ redeployment and reduction-in-force policies disparately impacted USCO/IM and CSS/IM employees on the basis of their age and gender. Her testimony, therefore, is admissible on relevancy grounds.
However, Xerox argues that Dr. Honig’s statistical methodology was so flawed that it should be deemed inadmissible as a matter of law. This Court disagrees. The question is whether Dr. Honig’s analysis is sufficient to establish a
prima facie
case of disparate impact age and gender discrimination.
See
discussion
infra; see also Tyler v. Bethlehem Steel Corp.,
958 F.2d 1176, 1188-89 (2d Cir.),
cert. denied,
506 U.S. 826, 113 S.Ct. 82, 121 L.Ed.2d 46 (1992) (soundness of methodology can go to weight, not admissibility, of expert opinion). Therefore, Xerox’
in limine
motion precluding the use of Dr. Honig’s testimony is denied.
D.
Plaintiffs’ Prima Facie Case
In order to establish a
prima facie
case of disparate impact discrimination, plaintiffs must demonstrate that Xerox utilized a specific employment practice which disparately impacted them on the basis of their age and gender.
Wards Cove,
490 U.S. at 656, 109 S.Ct. at 2124;
Maresco,
964 F.2d at 115 (discussion of
prima facie
case of disparate impact age discrimination).
Plaintiffs are required to identify the “specific employment practice” utilized by Xerox which caused a disparate impact upon them as members of a protected class. 42 U.S.C. § 2000&-2;
Wards Cove,
490 U.S. at 656, 109 S.Ct. at 2124;
Lowe,
886 F.2d at 1370. In their written submissions, plaintiffs identified Xerox’ assessment of their skills during the redeployment selection process as the “specific employment practice” which disparately impacted them on the basis of their age and gender. However, their expert, Dr. Honig, rendered no expert opinion identifying the skills assessments as the employment practice responsible for the disparity in redeployment and dismissal rates among the older, male workforce. In fact, Dr. Honig
refused
to consider the skills assessments in the context of her statistical analysis of plaintiffs’ claims because she deemed that data to be incomplete and inherently inaccurate.
Dr. Honig’s failure to analyze the impact of a specific employment practice upon plaintiffs is alone a sufficient basis upon which to find that plaintiffs have failed to establish a
prima facie
ease of discrimination.
Wards Cove,
490 U.S. at 656, 109 S.Ct. at 2124;
Lowe,
886 F.2d at 1371. However, even assuming that Dr. Honig
had
identified and analyzed a specific employment practice, her methodology and conclusions are insufficient to establish a
prima facie
case of disparate impact discrimination as required by
Wards Cove.
In reaching her conclusion that Xerox’ redeployment and reduction-in-force policy disparately impacted plaintiffs on the basis of their ages and gender, Dr. Honig relied on the results of two statistical tests, the “t-ratio” and “likelihood ratio” tests. It is undisputed that these tests alone can only measure whether or not a specific outcome is attributable to chance factors. Although both experts concluded that the redeployment and dismissal rates among certain portions of the USCO/IM and CSS/IM workforce who were 40 years of age or older and USCO/IM male employees were greater than those of other employee groups which disparities could not be caused by chance factors, Dr. Honig failed to conduct further statistical tests to determine what other factors could have accounted for those disparities. Instead, she concluded that since the disparities were not caused by chance, the high probabilities (1 in 100) demonstrated that they were caused by age or gender. Without conducting any other statistical tests to rule out factors other than age or gender and by relying solely on age and gender ratios, Dr.
Honig’s testimony is fatally flawed pursuant to
Wards Cove.
Dr. Honig candidly testified that she refused to conduct regression analyses because plaintiffs’ skills assessments and performance history data were conducted by Xerox managers who, in her opinion, were predisposed to evaluate older and male employees more harshly than other employees. Her bias against using this data was obvious from her testimony:
Q. Dr. Honig, your assumption that a manager who was faced with a layoff would take age or gender into account in doing so is based upon what?
A. Is based on — on the evidence I see in society that age or gender discrimination does seem to exist, and on the basis of statistical analyses that have been done by labor economists would suggest that — whether it’s personal bias or some other matter — discrimination on the — on the basis of age or gender does appear to exist.
Q. And so you had a preconceived notion in this case that the managers in question at Xerox would take age or gender into account?
A. No, that’s not true. All I’m saying is that there is a likelihood — -and I believe a strong likelihood, but there is a likelihood — that they may have. Whether they did or did not, I can’t say anything about.
Dr. Honig readily admitted that she conducted no independent statistical analyses of the data relating to skills assessments or performance evaluations in support of her assumption that such data was inherently unreliable. Her testimony on direct examination was very clear when she stated:
Q. Dr. Honig, with respect to Dr. Bloom’s statistical report submitted in these matters, did Dr. Bloom do a regression analysis as part of his statistical analysis set forth in these reports?
A. Yes, he did.
Q. In your expert opinion was it correct for Dr. Bloom to do a regression analysis for these reports?
A. Absolutely not.
Q. And what is the basis for your opinion in that regard?
A. Because I believe that the data he used for the most important issue, the most important factor involved in the redeployment and RIF at USCO and CSS, which was the skills of the workers employed, I believe that that data was not objective. I believe that it is — it was not available for all workers, and it is therefore unbiased and if I may say so dangerous data.
Q. And, Dr. Honig, why do you believe that that data was not objective?
A. This was a skills assessment done not by somebody outside of USCO or CSS so that it might be considered to be an objective skills assessment. This skills assessment was done by the employee’s immediate supervisors or managers or superiors in organizations which had undergone downsizing in years before where there was a certain amount of information that another downsizing would occur. Given that these assessments were done by people in these organizations involved with the downsizing, in my expert opinion they do not constitute objective unbiased data.
In contrast, Dr. Bloom testified that several analytical tools were available to Dr. Honig to test the reliability of the skills assessment and employee performance data. He compared the employees’ skills assessment scores with their performance histories and found that those employees with favorable past performance histories also tended to have high skills assessment scores. This positive correlation indicated to him that the skills assessment scores were not tainted or inherently unreliable as claimed by Dr. Hon-ig. Dr. Bloom then conducted numerous regression analyses both including and excluding the skills assessment data and found that the results were consistent — consideration of employee performance and skills assessment scores (whether considered alone or together), together with the employees’ tenure actually
favored
older workers in the redeployment assessment process.
When asked by this Court whether a regression analysis using only employee tenure and performance evaluations might be useful in determining whether the observed disparities were caused by a neutral, nondiscriminatory factor, Dr. Honig responded that it would be an “incomplete” regression analysis which would not be helpful. However, Dr. Bloom testified that the probative value of a regression analysis is not eliminated simply because the analysis uses fewer than all critical variables. In
Bazemore v. Friday,
the Supreme Court reached the same conclusion:
Importantly, it is clear that a regression analysis that includes less than all measurable variables may serve to prove a plaintiff’s ease. A plaintiff in a Title VII suit need not prove discrimination with scientific certainty; rather, his or her burden is to prove discrimination by a preponderance of the evidence.
478 U.S. 385, 400, 106 S.Ct. 3000, 3009, 92 L.Ed.2d 315 (1986). Rather than exhausting the analytical tools at her disposal to determine the usefulness of a regression analysis, Dr. Honig simply concluded that regression analyses would not be helpful. While a regression analysis in this case might have been difficult, it was not impossible. As the Second Circuit noted in the context of a gender discrimination case:
We also reject [defendant’s] attack on the multiple regression technique as a general matter when applied to the complex and diverse context of a medical school faculty. While it is true that the relative uniqueness of each faculty member, and the subjectivity of many of the determinants of salary, make a regression analysis difficult, these problems are not insurmountable. Indeed, as a device designed to sift through various factors in order to assess as accurately as possible the influence of any one of them, the multiple regression analysis is the accepted means for performing this difficult task.
Sobel v. Yeshiva University,
839 F.2d 18, 35 (2d Cir.1988),
cert. denied,
490 U.S. 1105, 109 S.Ct. 3154, 104 L.Ed.2d 1018 (1989). Dr. Honig’s categorical refusal to test the reliability of the skills and performance data undermines her conclusions and renders them inadequate as a matter of law. Wards
Cove,
490 U.S. at 656, 109 S.Ct. at 2124.
See, e.g., New York Urban League, Inc. v. State of New York,
71 F.3d 1031, 1038 (2d Cir.1995) (in a disparate impact case, criticizing a district court’s reliance on statistics which did not account for other factors possibly responsible for a statistical disparity);
Lopez v. Metropolitan Life Ins. Co.,
930 F.2d 157, 160 (2d Cir.),
cert. denied,
502 U.S. 880, 112 S.Ct. 228, 116 L.Ed.2d 185 (1991) (“The causal requirement of [Title VII] recognizes that underrepresentation [of a protected class] might result from any number of factors, and it places an initial burden on the plaintiff to show that the specific factor challenged under the disparate impact model results in the discriminatory impact.”) (citations omitted).
Additional flaws in Dr. Honig’s methodology are also apparent. In rendering her opinion that the dismissal rates for redeployed USCO/IM and CSS/IM employees were higher for older and male employees, she failed to account for the fact that some of the employees included in her data set had failed to seek alternative employment within Xerox or that differences existed between employees as to the intensity of their searches or their qualifications for alternative positions in Xerox. In explaining her failure to consider these differences in compiling her data set, Dr. Honig testified that she assumed that redeployed personnel would be rehired within Xerox at a lower grade level and salary, which would not mitigate the discriminatory impact of their selection for redeployment. Dr. Honig’s assumption was based on her observation that 9 of 35 redeployed CSS employees were rehired' at a lower grade level. However, without further investigation into the types of positions those employees were hired to fill, there is no proof that those employees lost salary, benefits or were otherwise disadvantaged.
This Court finds that plaintiffs’ expert has failed to establish that Xerox’ redeployment and reduetion-in-force policy disparately impacted USCO/IM employees on the basis of their age or gender of CSS/IM employees on the basis of their age. Accordingly, plaintiffs have failed to prove a
prima facie
ease of
disparate impact age or gender discrimination and their complaints are dismissed.
III.
Remaining Motions to Dismiss
Xerox also moved to dismiss these actions on the ground that the disparate impact age claim was not properly pled in plaintiffs’ complaints. Xerox clearly believed that disparate impact age claims were at issue in this case in light of Dr. Bloom’s testimony and it has' suffered no prejudice by inclusion of those claims in this case.
Xerox also argues that the ADEA claims brought by plaintiff Earl May must be dismissed because he failed to file an EEOC complaint and receive a right-to-sue letter before he filed his action. Although it is undisputed that May did not file an administrative charge of age discrimination, his claims arise out of the same redeployment and reduction-in-force initiative as do the claims of the remaining plaintiffs who did file EEOC age charges and receive right-to-sue letters. Accordingly, May’s age claims “piggyback” onto those administrative claims and are timely.
Tolliver v. Xerox Corp.,
918 F.2d 1052, 1057 (2d Cir.1990),
cert. denied,
499 U.S. 983, 111 S.Ct. 1641, 113 L.Ed.2d 736 (1991);
see also Bagg v. Xerox Corp.,
94-CV-6261T, Decision and Order filed May 5, 1995;
Bowers v. Xerox Corp.,
94-CV-6093T, Decision and Order filed May 5, 1995. Xerox’ motion to dismiss plaintiff May’s claims is denied.
WHEREFORE, based upon the parties’ submissions and upon the testimony of plaintiffs’ expert Dr. Marjorie Honig and Xerox’ expert Dr. David Bloom, proffered at an evidentiary hearing held by this Court on July 15 and 16, 1996, and the arguments of counsel for both sides, this Court finds that plaintiffs have failed to establish a
prima facie
ease of disparate impact age discrimination under the ADEA or disparate impact gender discrimination under Title VII of the Civil Rights Act of 1964, as amended. Because plaintiffs’ federal discrimination claims have been dismissed, their state law age and gender discrimination claims brought pursuant to the New York Human Rights Law, and plaintiff Tuttle’s state law disability claim, are also dismissed for lack of supplemental jurisdiction.
See
28 U.S.C. § 1367.
Xerox’ motion in limine to exclude the expert opinion of Dr. Honig, its motion to dismiss plaintiffs’ complaints for failure to properly plead a disparate impact age discrimination claim and its motion to dismiss plaintiff Earl May’s complaint for failure to file an EEOC age discrimination charge are all denied.
Plaintiffs’ complaints are dismissed.
ALL OF THE ABOVE IS SO ORDERED.