Denny v. Westfield State College

OPINION FREEDMAN, Chief Judge. I. INTRODUCTION Plaintiffs in these consolidated actions, present and former faculty members of Westfield State College (“WSC” or “West-field State”), allege that they were discriminated against because of their sex in violation of Title VII of the Civil Rights Act of 1964, 42 U.S.C. §§ 2000e et seq. Specifically, plaintiffs allege they received lower salaries than similarly situated male faculty members at WSC. As originally commenced, the plaintiffs in No. 78-2235-F brought their complaint on behalf of themselves and all others similarly situated. On May 30,1986, the Court denied plaintiffs' motion to certify the class and allowed defendants’ motion to dismiss the class allegations from the complaint. *1147 The claims of the three named plaintiffs were tried to the Court on June 2-4, 1986. Pursuant to Fed.R.Civ.P. 52(a), the Court makes the following findings of fact and conclusions of law. II. FINDINGS OF FACT A. Parties 1. Plaintiff Leah Stern is a female who was employed by defendants from August 1973 to August 1977 in the Philosophy Department of Westfield State College. 2. Plaintiff Marilyn Denny is a female who was employed by defendants from December 1971 to August 1977 in the Sociology Department of Westfield State College. 3. Plaintiff Catherine Dower is a female who has been employed by defendants since 1956 in the Music Department of Westfield State College. 4. Defendants Westfield State College and the Board of Regents of Higher Education are agencies of the Commonwealth of Massachusetts, and are “employers” within the meaning of 42 U.S.C. § 2000e. B. Evidence of Discrimination 5. Prior to trial, the parties entered a stipulation providing: The job requirements for all teachers, within ranks, are substantially equal in terms of skill, effort, and responsibility and are performed under similar working conditions. However, the parties acknowledge that different teachers may have different professional qualifications including different skills, experience, education, and disciplinary and technical backgrounds. Nothing contained in this stipulation shall be deemed to mean that the college could not properly have taken these factors into consideration when fixing initial salaries, entitlements to promotions and other benefits affecting the rates of compensation of faculty members, provided that whether the college considered such factors or the legitimacy of the method used to consider such factors are the issues to be determined at trial. 6. Plaintiffs’ evidence at trial consisted of the testimony of plaintiff Dower and Arlene S. Ash, Ph.D., a Professor of Statistics. Dr. Ash undertook a statistical study of wages at WSC, the results of which were admitted at trial. Defendants’ evidence consisted of the testimony of Ernest T. Kendall, Ph.D., a labor economist, who also performed a statistical study of WSC’s system of compensation, and Jean Regan, formerly an employee in the personnel office of the Board of Regents of Higher Education (successor to the Board of Trustees of State Colleges). 1. Statistical Evidence 7. Both Dr. Ash and Dr. Kendall used a statistical method known as multiple re--gression analysis to produce their different opinions. Briefly put, multiple regression analysis is a computer-assisted statistical method by a which a variety of factors (“explanatory variables”) are considered at one time to determine their combined effect on another variable such as salary (“outcome variable”). See generally Pouncy v. Prudential Insurance Co., 499 F.Supp. 427, 499-50 n. 11 (S.D.Tex.1980). In this case, Drs. Ash and Kendall set out to determine how factors such as seniority, departmental affiliation, and prior experience, affected the salaries of WSC faculty members. The theory goes: to the extent salaries at WSC are not fully explained by these factors, an impermissible factor, such as sex, may account for the discrepancy. Moreover, assuming the proper variables are selected and that the result of the analysis is “statistically significant,” the amount of salary differential due to sex discrimination may be estimated. Though both Drs. Ash and Kendall used this basic method, they actually used two different types of regression analysis. Dr. Kendall used what is known as the “dummy variable” technique whereas Dr. Ash performed “men only” regressions. Each criticized the other’s chosen methodology, but generally agreed that neither approach is demonstrably superior to the other. *1148 a. Dr. Ash’s Regression Analysis 8. Relying on data supplied by defendants during discovery, Dr. Ash coded information for all full-time faculty members at WSC during the years 1974 through 1984. Dr. Ash used the following explanatory variables: departmental grouping, seniority at WSC, years of prior tenure-track teaching experience, years of other relevant prior experience, current rank, and the number of years since a terminal degree (e.g., Ph.D. or M.F.A.) was obtained. Using this model, Dr. Ash reported that she was able to “explain” over seventy percent of variation in men’s salaries at WSC in every year studied except one. 1 However, in the latter years of her study, Dr. Ash’s results tended to provide less explanation of salary variance among male faculty members. 9. Applying the formula for men’s salaries to women faculty members in each year, Dr. Ash predicted what salary comparable women would be expected to receive at WSC in each year. For the years studied, Dr. Ash concluded that on average, women faculty members at WSC were subject to salary deficits (presumably because of sex) in the following amounts: YEAR SALARY DEFICIT 1974 $ 702 1975 676 1976 704 1977 578 1978 485 1979 725 1980 750 1981 707 1982 1354 1983 1996 1984 1861 In every year except for 1978 these results were thought to be “statistically significant” as measured by their “P-values.” 2 The results for 1978 were considered statistically borderline. 10.Extrapolating from the model to the three named plaintiffs, Dr. Ash estimated that in each year at WSC they were subject to the wage disparities as compared to the predicted similarly qualified male professors in the following amounts: YEAR DENNY STERN DOWER 1974 $1,637 $1,187 $ 4,578 1975 1,373 1,265 4,346 1976 1,582 1,582 4,321 1977 4,314 1978 3,572 1979 3,425 1980 3,056 1981 3,252 1982 3,065 1983 2,765 1984 3,449 TOTALS $4,592 $4,034 $40,143 11.Dr. Ash also performed analysis using the “dummy-variable” approach favored by Dr. Kendall. The results obtained were similar to her male-only regression studies. b. Defendants’ Criticisms of Dr. Ash’s Report 12. Defendants’ expert, Dr. Kendall, raised several objections to the appropriateness of Dr. Ash’s methodology and the validity of her results. The Court will discuss these criticisms in some depth. 13. The first set of criticisms levelled at Dr. Ash’s report concerns problems believed to be inherent in the data set from WSC. Specifically, Dr. Kendall testified that the results obtained from regression analysis are suspect because of the presence of “multicollinearity,” “nonrandomness in the distribution of the residuals,” and “heteroscedasticity.” *1149 14. Multicollinearity exists when two or more of the explanatory variables tend to move in a parallel fashion. For example, a faculty member’s age and seniority at Westfield State have an obvious relationship to one another. Multicollinearity is present to some degree in all of the regression studies performed by Drs. Ash and Kendall. Drs. Ash and Kendall both agreed that the presence of multicollinearity would produce results that are less precise than they would otherwise be. However, from the evidence presented, the Court accepts the testimony of Dr. Ash that the presence of multicollinearity would merely tend to overestimate the amount of error associated with the estimate of difference in salary by sex. In other words, P-values will be artificially higher than they would be if there were no multicolli-nearity present. Thus, the presence of multicollinearity means that results found to be significant in the analysis are likely to be even more significant than they appear. Because Dr. Ash’s results demonstrated statistic significance for each year studied except one, the Court does not find that the presence of multicollinearity detracts from the validity of her findings. 15. Heteroscedasticity is the lack of common or constant variance. In Dr. Ash’s model, for example, this would occur if the wage disparity were greater for more senior faculty than less senior faculty. In fact, Dr. Ash admitted that this defect was probably present to some degree, but suggested that its effect was not significant. The Court does not find that the possible presence of heteroscedasticity itself detracts from the validity of Dr. Ash’s study. 16. Dr. Kendall also criticized the non-randomness in the distribution of residuals in Dr. Ash’s model. In other words, Dr. Kendall hypothesized that the function between sex-neutral qualifications and salary was not linear as Dr. Ash’s model assumed, and that, as a result, her model might tend to report women as being underpaid when in reality they were being paid commensurately with comparable male faculty members. 3 Dr. Ash testified that she examined the salary data and found that the relationship between qualification and salary was not exactly linear, but more parabolic. The implication of the parabolic function, the data revealed, was precisely the opposite of what Dr. Kendall suggested. Thus, the straight-line curve employed in Dr. Ash’s regression model actually tended to underestimate the salaries of women who occupy the segment of the faculty with intermediate levels of experience in greater numbers than men. Dr. Ash testified that she performed analyses which took this parabolic function into account (she substituted the term seniority squared for seniority). This regression analysis yielded results with a somewhat better fit (a higher R2) and an increase in the extent of salary underpayment for women. Despite these results, Dr. Ash chose not to include a seniority-squared term for the sake of greater simplicity. The Court does not find that the presence of nonrandomness of the residuals is a significant problem with Dr. Ash’s study. 17. A much more significant criticism of Dr. Ash’s study has to do with the manner in which she accounted for departmental affiliation in constructing her model. This criticism is potentially critical because departmental affiliation represents a major way in which larger market forces affect the wage structure at an institution. For example, one would expect higher salaries for faculty members in disciplines in which there is a keen competition for qualified individuals such as computer science and business administration, than for many liberal arts fields such as philosophy and ancient languages. See generally Lamphere v. Brown University, 685 F.2d 743, 748 (1st Cir.1982) 18. During the years studied, Westfield State had nineteen separate departments. *1150 Unlike some educational institutions that organize departments into larger departmental groups or divisions, each department at Westfield State was independent. Because some WSC departments were extremely small, assigning a variable to each individual department would not permit reliable detection of the effect of market forces on salary. It is very difficult to extrapolate information from a data set consisting of one person. To avoid the problems inherent in analyzing nineteen sui generis departments, Dr. Ash grouped them into five “departmental groups” for purposes of her study, relying on her understanding of how universities commonly aggregate departments. The groupings, each designated with a two-or three-letter code, are as follows: (1) Applied (APP) consisting of mass media, criminal justice, economics and business administration, and geography and regional planning; (2) Humanities (HUM) consisting of English, music, history, art, modern foreign languages and philosophy; (3) Social Sciences (SOC) consisting of sociology, psychology and political science; (4) Natural Sciences (SCI) consisting of physical science, math, computer and information science and biology; (5) Education (ED) consisting of education and health and physical education. 19.Using the male-only faculty data from Westfield State, Dr. Ash calculated the affect of departmental affiliation on wages, using the Social Sciences as a bench mark. The results of these calculations are indicated below: Year APP HUM SCI ED 1974 -298 264 -904 -1071 1975 -45 171 -773 -883 1976 58 209 -792 -1098 1977 -27 81 -764 -946 1978 -38 135 -1107 -1303 1979 -483 118 -640 -920 1980 -270 420 -511 -405 1981 -274 632 -653 -661 1982 -946 391 -745 -468 1983 -1369 109 -965 -416 1984 -1292 506 -835 -610 In 1974 for example, a professor in one of the constituent departments within the Applied Group, such as business administration, could be expected to earn $298 less than a similarly situated professor in a Social Science field such as political science. Similarly, in 1974 a professor in one of the Humanities fields such as English would be expected to earn $264 more than the Social Science professor. 20. Dr. Kendall’s main criticism of Dr. Ash’s departmental groupings is that they are counterintuitive. Dr. Kendall testified that, from his understanding of the national labor market, he knows that the Applied fields such as computer science are far more competitive than the Social Science fields. Thus, he believes it illogical for Dr. Ash’s study to present science professors consistently earning less than social scientists. According to defendants, By its inability to control for departmental affiliation, Dr. Ash’s regression analysis has failed to take cognizance of market forces that operate at the departmental level and that are not random in their effect on the salaries of male and female members of the faculty at West-field State College.... The probative value of her analysis is much diminished by this, and it calls seriously into question the ability of Dr. Ash’s model to generate realistic estimates of male-equivalent salaries for sociologists such as Professor Denny or for any other faculty member. Defendants’ Proposed Findings at 24 (citations omitted). Plaintiffs counter by asserting that if Dr. Ash’s model yielded results different from what one would expect from one’s knowledge of the larger labor market, it is only because Westfield State’s actual salary practices depart from these expectations. 21. The Court is not persuaded that, simply because Dr. Ash’s model fails to parallel precisely national labor statistics, it is inaccurate. Given the nearly limitless variance among local economies and how different institutions tend to specialize in different fields, it would not be surprising, for example, for a professor of art at a *1151 given institution to earn more than a professor of business administration if that school’s art program emphasized highly marketable training in graphic arts (as opposed to less marketable art history) and the business administration program emphasized less marketable specialties such as management of not-for-profit entities. 4 Of course, this is not to imply that this was generally the case at Westfield State, but only to suggest that the fact that a study of a particular educational institution does not comport precisely with national labor statistics does not, without more, detract from the validity of its results. 22. Another difficulty with Dr. Ash’s grouping of departments at Westfield State has not been commented upon by the parties. According to plaintiffs, Dr. Ash grouped departments using “a logical grouping based on subject content.” Plaintiffs’ Post-Trial Memorandum at 21. This seems a rational solution to the problem of a data set consisting of many small departments. However, the Court is concerned about the consequences of lumping together within one group departments which, though logically related based on their subject matter, have very different wage structures. For example, in 1984, a year in which Dr. Ash’s results are disclosed for particular departments, the Social Science Department Group (SOC) contained one department with the highest anticipated wage — psychology—and one near to the lowest — sociology. How this affects the validity of Dr. Ash’s analysis is unclear; as noted, neither party has commented on the issue. Nevertheless, it may cause the Court to pause before it is willing to adopt completely Dr. Ash’s findings. 23. Finally, Dr. Kendall criticized Dr. Ash’s results because she failed to include the device by which Westfield State compensated highly productive faculty members — Distinguished Service Awards (“DSAs”) — in her model. These awards were not included, according to Dr. Ash, because they might be tainted; that is, awarded in a sexually discriminatory manner. Dr. Ash discusses the notion of “tainted variables” as follows: Subjective criteria, such as merit evaluation by the employer are clearly suspect, although if the plaintiff is willing to accept the employer’s evaluations as fair and not “tainted” by the same prejudice that is alleged to be present in the setting, then such evaluation may be used. Included variables should be part of the publicly accepted reasons for making distinctions among faculty salaries. (Thus, though short people may be paid less than tall ones, height should not be used.) While salary upgrades which have been distributed by the administration (such as Distinguished Service Awards and Compensatory Awards at Westfield) will usually be helpful in achieving a better fitting formula, they are not faculty characteristics, and are suspect of possible tainting, unless demonstrated otherwise. Plaintiffs’ Exhibit 6 at 20, 25-26. Dr. Ash admitted during cross examination that rank carried the same possibility of taint but was nevertheless used after rank data were examined and the possibility of taint eliminated. Dr. Ash’s failure to similarly examine DSAs was unexplained, despite her testimony that using DSAs, would have improved the fit of her model in the later years. This omission is notable because more female faculty members at WSC (including plaintiff Dower) received DSAs than males. On redirect, Dr. Ash explained that she did perform some regressions including DSAs which did not yield significantly different results. The Court credits Dr. Ash’s testimony and does not find that the omission of DSAs in her primary analysis detracts from the validity of her results. *1152 c. Dr. Kendall’s Regression Analysis 24. In addition to making specific criticisms of Dr. Ash’s study, Dr. Kendall performed his own regression analysis from the data supplied by Westfield State. He produced the results of three regression analyses which were included as an Appendix to Dr. Ash's Report (Exhibit 6). The parties agreed that the results of two of the analyses, identified as R-2 and R-3, are unreliable. The Court, therefore, will give them no consideration. The remaining analysis labeled R-l which was referred to by Dr. Kendall as “the best of a bad lot,” revealed a “statistically significant” wage differential for women in five of the eleven years studied (1976 and 1981 through 1984). Plaintiffs criticize some of the technical aspects of R-l, suggesting that flaws in this analysis tend to underestimate the degree of underpayment to female faculty members at WSC. 5 The Court generally agrees with plaintiffs’ criticism. 25. Dr. Kendall “explained” the statistically significant findings in the latter years by attempting to demonstrate that the sex differentials could be accounted for by differences in starting salaries of new faculty members in two particularly competitive fields — business administration and computer sciences. To reach this result, defendants begin with a basic assumption: “Sex-related differentials in salary could have arisen at Westfield State College during the period in question through one or more of four possible avenues. Males could have been favored in (1) promotions, (2) starting salaries, (3) distinguished service awards, or (4) starting salaries.” Defendants’ Proposed Findings and Decision at 28. From there, they analyze data concerning each of these four factors and conclude that “[ajnalysis of the first three have revealed absolutely no evidence that males were given preferential treatment with respect to any of these determinants of salary_” Id. This suggests, according to defendants, that the source of salary differential must be starting salaries in the period 1980 to 1984. Defendants state: It was during this period that West-field State College sought to expand its program in computer science and business administration. Between 1980 and 1984, the College hired a total of 48 new faculty members, 21 of whom were in the fields of computer science and business. Of these 21 new faculty, only 2 were women. In contrast 14 of the 27 faculty hired into the other departments were women. What happened at Westfield is now readily understood. Faculty in the computer science and business administration are in two fields where the competition for qualified personnel is very strong, both inside and outside of academia. In order to staff these two growing departments, it was necessary for the College to add substantial market adjustments to starting salaries in order to compete in a market-place that is dominated by private industry. Because most of those hired were males, many males were given substantial salary adjustments in response to these market forces; as a direct result the average male starting salary rose substantially above that of females. Id. at 30-31. 26.At first glance, this appears to be a persuasive explanation for the seemingly discriminatory wage structure at WSC. Closer scrutiny belies this, however. For one thing, the Court does not find that an adequate foundation was laid for these results. As defendants elsewhere argue “because of the sophistication and complexity of many of the statistical models being used in discrimination cases by professional econometricians, courts must give ‘close scrutiny [to the] empirical proof’ on which the models are ereet- *1153 ed ... in order to guard against the use of statistical data that may have been ‘segmented and particularized and fashioned to obtain a desired result.’ ” E.E.O.C. v. Federal Reserve Bank of Richmond, 698 F.2d [633,] 646 [(4th Cir.1983)]. Defendants’ Proposed Findings and Decision at 11. By looking at these four factors in isolation, perhaps defendants are guilty of the very segmenting and particularizing they caution against. Defendants have not shown that these analyses are based on anything but gross statistics of particular variables, the relationship of which to the entire data set is unclear at best. 27.Apart from this general criticism, even assuming Dr. Kendall was correct that every valid explanation for wage disparity could be eliminated except for starting salaries in two departments, this approach hardly accounts for the entire variance revealed in the studies of Drs. Ash and Kendall. Thus, whatever may have occurred in the period in which WSC was attempting to “beef up” its business administration and computer science departments, this does not explain the statistically significant results Dr. Ash uncovered for pre-1980 years. It is also misleading to assume that the special cases of the business administration and computer science departments are relevant for the entire period of 1980 to 1984 because, as plaintiffs state, no bonuses were given in the business administration department in the years 1980, 1981 and 1982 and none given in the computer science department in 1980 and 1981. Plaintiffs’ Post-Trial Memorandum at 28-29. In addition, defendants seem to be double counting the computer science department. As plaintiffs note, Residual R-l already contains a marker for a professor’s membership in the computer science department. Having already accounted for the effect of service in this department in his regression, Dr. Kendall cannot also point to the computer science to explain statistically significant results his analysis revealed. 2. Non-Statistical Evidence 28. Through the evidence at trial consisted mostly of statistical evidence, both sides presented some non-statistical evidence to prove or rebut discrimination. 29. Plaintiff Dower testified that in 1967 she had a conversation with Dr. Leonard Savignano, then President of Westfield State. 6 Plaintiff Dower approached Dr. Sa-vignano because she felt her salary was low compared to comparable males at WSC. Dr. Savignano reportedly told Dr. Dower that men needed higher salaries than women because they had families to support and that the only way she could improve her salary was by getting her Ph. D. so she could be promoted from assistant professor to the rank of associate professor. Based on this conversation, plaintiff Dower eventually obtained her Ph.D. and was promoted. According to Jean Regan, an employee of the Board of Regents of Higher Education, the requirement of a Ph.D. for promotion to associate professor had in fact been adopted by the Board of Trustees of State Colleges in 1967. Thus, even if one assumes that Dr. Savignano did make the discriminatory statement, his advice to Dr. Dower seems to have been quite legitimate. 30. Dr. Regan also testified to the effect that the Board of Regents of Higher Education or any of its previous incarnations 7 had the final say on the question of faculty salaries. Though this was literally true, it seems that the recommendation from a president of one of the state colleges was accorded great weight. 31. The remainder of the evidence was of the pseudo-statistical variety comparing each one of the plaintiffs to one or another of the male faculty members. This evi *1154 dence not only swings both ways but it is also singularly unconvincing. The Court, therefore, gives it very little weight. III. CONCLUSIONS OF LAW 1. This Court has jurisdiction over this matter pursuant to 28 U.S.C. §§ 1331, 1343(a)(3) and 2201 and under sections 706(e) and (f) of Title VII of the Civil Rights Act of 1964 as amended, 42 U.S.C. § 2000e-5(f)(l), (3) (“Title VII”). 2. Venue is properly laid in this District pursuant to 28 U.S.C. § 1391(b) and section 706(f) of Title VII, 42 U.S.C. § 2000e-5(f)(3). 3. The parties agree that all jurisdictional prerequisites for instituting this action have been met under section 706(e) of Title VII, 42 U.S.C. § 2000e-5(f)(l). 4. Section 703 of Title VII, 42 U.S.C. § 2000e-2(a), provides, inter alia: It shall be an unlawful employment practice for an employer— (1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions or privileges of employment, because of such individual’s race, color, religion, sex, or national origin.... (Emphasis added). See generally Annotation, Wage Differentials as Violative of Those Provisions of Title VII of the Civil Rights Act of 1964, as amended (42 U.S.C. §§ 2000e et seq.), which Prohibit Sex Discrimination in Employment, 62 A.L.R. Fed. 33 (1983). 5. In order to make out a claim for disparate impact in a traditional Title VII case, plaintiffs must first establish a prima facie case of discrimination by a preponderance of the evidence. The burden then shifts to the defendants to articulate a legitimate, nondiscriminatory reason for their challenged actions. If defendants do so, plaintiffs are then given an opportunity to establish by a preponderance of the evidence that the reasons asserted by the defendants are a mere pretext for unlawful discrimination. Texas Department of Community Affairs v. Burdine, 450 U.S. 248, 252-53, 101 S.Ct. 1089, 1093, 67 L.Ed. 2d 207 (1980); McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-05, 93 S.Ct. 1817, 1824-25, 36 L.Ed.2d 668 (1973); Lamphere v. Brown University, 685 F.2d at 748. However, “[t]he ultimate burden of persuading the trier of fact that the defendant intentionally discriminated against the plaintiff ... 'remains at all times with the plaintiff.’ ” Cooper v. Federal Reserve Bank of Richmond, 467 U.S. 867, 875, 104 S.Ct. 2794, 2799, 81 L.Ed.2d 718 (1984), quoting United States Postal Service Board of Governors v. Aikens, 460 U.S. 711, 716, 103 S.Ct. 1478, 1482, 75 L.Ed.2d 403 (1983). 6. Though this action was tried solely as a Title VII case, plaintiffs argue that the burdens of proof applicable to actions under the Equal Pay Act of 1963, 29 U.S.C. § 206(d) (“EPA”), should govern. The Equal Pay Act provides as follows: No employer having employees subject to any provisions of this section shall discriminate, within any establishment in which such employees are employed, between employees on the basis of sex by paying wages- to employees in such establishment at a rate less than the rate at which he pays wages to employees of the opposite sex in such establishment for equal work on jobs the performance of which requires equal skill, effort, and responsibility, and which are performed under similar working conditions, except where such payment is made pursuant to (i) a seniority system; (ii) a merit system; (iii) a system which measures earnings by quantity or quality of production; or (iv) a differential based on any further factor other than sex. 77 Stat. 56, 29 U.S.C. § 206(d)(1). 7. Generally, EPA is a narrower remedial statute than Title VII. Unlike Title VII, EPA does not prohibit discrimination in hiring, firing or promoting employees and it is applicable to a much narrower range of employers. However, because of the way burdens of proof are allocated in an Equal Pay Act case, a plaintiff may be able to prevail under that Act where she *1155 could not under Title VII. This is because, unlike an ordinary Title VII case, the defendant in an Equal Pay Act has the burden of persuasion as well as production on the four enumerated defenses. Corning Glass Works v. Brennan, 417 U.S. 188, 196, 94 S.Ct. 2223, 2229, 41 L.Ed.2d 1 (1974). 8. Attempts at “integrating” Title VII and EPA have their origin in the so-called Bennett Amendment to Title VII. Section 703(h) of the Civil Rights Act of 1964, 42 U.S.C. § 2000e-2(h). The amendment provides: It shall not be an unlawful employment practice under this title for an employer to differentiate upon the basis of sex in determining the amount of the wages or compensation paid or to be paid to employees of such employer if such differentiation is authorized by the provisions of section 6(d) of the Fair Labor Standards Act as amended (29 U.S.C. § 206(d)). 9. In County of Washington v. Gunther, 452 U.S. 161, 101 S.Ct. 2242, 68 L.Ed.2d 751 (1981), the Court explained to what degree the Bennett Amendment actually limited a Title VII plaintiff. In Gunther, the Court considered whether, by virtue of the Bennett Amendment, EPA’s “equal work” requirement was applicable to Title VII plaintiffs. After reviewing the legislative history, the Court concluded it did not. The Court held that “only differentials attributable to the four affirmative defenses of the Equal Pay Act are ‘authorized’ by that Act within the meaning of [the Bennett Amendment].” Id. at 171,107 S.Ct. at 2249. The Court left unanswered questions about “how sex-based wage discrimination litigation under Title VII should be structured to accommodate” the affirmative defense of EPA. Id. at 171, 101 S.Ct. at 2249. Relying on Gunther, the Ninth Circuit in Kouba v. Allstate Insurance Co., 691 F.2d 873 (1982), held that EPA’s order of proof was applicable to Title VII claims. As the court stated: [E]ven under Title VII, the employer bears the burden of showing that the wage differential resulted from a factor other than sex.... Nothing in Burdine converts this affirmative defense which the employer must plead and prove under Coming Glass, into an element of the cause of action, which the employee must show does not exist. 691 F.2d at 875. Accord Plemmer v. Parsons-Gilbane, 713 F.2d 1127, 1136-37 (5th Cir.1983); Odomes v. NuCare, Inc., 653 F.2d 246, 250-51 (6th Cir.1981); Chang v. University of Rhode Island, 606 F.Supp. 1161, 1181 (D.R.I.1985); Melanson v. Rantoul, 536 F.Supp. 271, 286 (D.R.I.1982). This approach is rejected in EEOC v. Sears, Roebuck & Co., 628 F.Supp. 1264, 1328-32 (N.D.I11.1986). In that case the district court analyzed the legislative history to the Bennett Amendment and concluded that “Congress did not intend the Bennett Amendment to impose Equal Pay Act burdens and elements of proof to pay claims brought under Title VII.” Id. at 1330. See also Crockwell v. Blackmon-Mooring Steamatic, Inc., 627 F.Supp. 800, 806 (W.D.Tenn.1985). To date, the First Circuit has not ruled on the question of whether EPA’s burdens of proof are applicable to Title VII equal pay claims. The issue was discussed in Marcoux v. State of Maine, 797 F.2d 1100 (1st Cir.1986), in which the district court adopted the Kouba approach, but because neither party had raised an objection below, the Court of Appeals declined to rule. 10. This Court concludes that Kouba and its progeny (which constitute the clear weight of authority) represent the better view. The Equal Pay Act and Title VII “are in pari materia and must be construed in harmony with one another.” Miller v. Kansas Power & Light Co., 585 F.Supp. 1509 (D.Kan.1984). The Supreme Court’s holding in Gunther demonstrates that the Bennett Amendment is not simply, as Sears, Roebuck would have it, a “protection to] employers in Title VII cases by requiring them to meet a greater burden of proof.” Sears, Roebuck, 628 F.Supp. at 1330 n. 88. Rather, the Bennett Amendment is a device intended to harmonize the protections of two, somewhat overlapping *1156 statutory schemes by incorporating EPA’s four affirmative defenses into Title VII. Indeed, the very notion of an “affirmative defense” strongly suggests that, as with other affirmative defenses, the party claiming its benefit has the burden, not just of pleading, but also of persuading the trier of fact by a preponderance of the evidence that the party should prevail on the defense. Accordingly, the Court will proceed to analyze plaintiffs' claim under the EPA approach. 11. In light of this conclusion, plaintiffs are required initially to establish that they received lower salaries than men for comparable work. Once this is established, defendants have the burden of persuading the Court, by a preponderance of evidence, that the wage disparity was the result of (i) a seniority system; (ii) a merit system; (iii) a system which measures earnings by quantity or quality of production; or (iv) a differential based on any further factor other than sex. 12. The Court concludes plaintiffs have established their initial burden of demonstrating a prima facie case. The parties have stipulated to the comparable nature of the work performed by male and female faculty members at WSC. Supra, Paragraph 5. Plaintiffs have also established discriminatory wage differentials by a preponderance of the evidence through the statistical analyses presented at trial. Such evidence alone is sufficient to establish plaintiffs’ prima facie case. Hazelwood School District v. United States, 433 U.S. 299, 307-08, 97 S.Ct. 2736, 2741, 53 L.Ed.2d 768 (1977); International Brotherhood of Teamsters v. United States, 431 U.S. 324, 97 S.Ct. 1843, 52 L.Ed.2d 396 (1977). Though the Court has expressed criticism for some aspects of Dr. Ash’s report, on the whole the Court finds that the report is sufficient to demonstrate that female faculty members at WSC received significantly lower salaries than male faculty members given equivalent experience, rank, years since receiving doctorate and departmental affiliation. 13. In Title VII terms, defendants have articulated a legitimate, non-discriminatory reason for the salary differential at WSC; in EPA terms, defendants have come forward with some evidence to sustain the defense that female faculty members at WSC received lower salaries than equivalent males because of “a differential based on any further factor other than sex,” 29 U.S.C. § 206(d)(l)(iv): according to defendants’ expert, Dr. Kendall, the salary differential uncovered by Dr. Ash can be explained by two variables not fully accounted for in Dr. Ash’s study — market forces as reflected in departmental affiliation and exceptional performance as reflected in distinguished service awards. However, as the Court noted above, supra at Paragraphs 23 & 25, apart from serious statistical short-comings in defendants’ argument, defendants are unable to account for salary discrepancies in all years studied. 14. The Court finds, however, that while defendants have satisfied their burden of production, they have failed to sustain their burden of proving, by a preponderance of the evidence, that the salary differentials for faculty members at WSC in the years studied were due to factors other than sex. IV. CONCLUSION The Court finds that the named plaintiffs have sustained their burden of proving a violation of 42 U.S.C. §§ 2000e, et seq., with respect to their claim that they received lower salaries than similarly situated males at WSC, and that this differential was due to illegal sex discrimination. Moreover, the Court finds that plaintiffs are entitled to an award of backpay limited by the periods in which the administrative complaints were filed as measured by the extent of the wage disparities discovered in Dr. Ash’s results. Supra at Paragraph 9. See 42 U.S.C. § 2000e-5(g); EEOC v. Enterprises Association, Steamfitters Local No. 638, 542 F.2d 579 (2d Cir.), cert. denied, 430 U.S. 911, 97 S.Ct. 1186, 51 L.Ed. 2d 588 (1976). The parties are ordered to attempt to stipulate to a proposed form of judgment within thirty days. Should stipulation prove impossible, the parties may *1157 submit separate proposed judgments within this time period. Footnotes [1] The degree to which the model is capable of accounting for all of its variability is referred to as its "fit.” Statisticians quantify the degree of fit as the model’s sum of squared residuals ("R-Squared” or "R2”). A perfect fitting model would have an R 2 of 1.00. Dr. Ash testified that an R 2 of 0.8 is indicative of a well-fitting model. This means that the model is able to "explain” 80% of the variability of salaries at the institution under study. [2] As explained in Dr. Ash's report, a P-value is a measure of the likelihood that something is the result of pure randomness. In this context, it is the likelihood that "a totally sex neutral situation would ‘just happen’ to look this bad, by chance alone.” Exhibit 6 at 32. The nearer a P-value is to 0, the more statistically reliable the result; the closer to 1.00, the less reliable. A P-value of .05 or less is often considered to be statistically significant. [3] Significantly, Dr. Kendall did not make a study of whether there was nonrandomness in the distribution of residuals in Dr. Ash's model; he merely assumed, from his experience as a labor economist, that it probably was. [4] At Westfield State in one year studied — 1984— the relative wages in the art and business administration reflected a wage disparity in this direction. In that year, the art department had an average wage $787 more than the economics and business administration department. Plaintiffs’ Post-Trial Memorandum at 21. [5] Specifically, plaintiffs contend that the presence of multicollinearity in R-l (which Dr. Kendall acknowledged) tended to inflate P-values thus making results that should be significant appear non-significant. In addition, Dr. Kendall omitted a variable for degrees at the time of analysis. Since women were more likely than men to have terminal degrees, the sex deficit was also probably underestimated. [6] Dr. Savignano is now deceased. [7] This body was at one time known as the Department of Higher Education and then as the Board of Trustees of State Colleges until 1981 when it was supplanted by the Board of Regents. See St. 1980, ch. 329, §§ 111 & 112. See also Defendants’ Proposed Findings of Fact and Decision at 34.