Warshawsky disputes this computation and argues that the protected group is all pregnant first-year employees who required sick leave and the non-protected group as all non-pregnant first-year employees who required sick leave. (See table below).
GROUP TOTAL # FIRED % FIRED
who required leave 21 20 95.2%
who required leave 33 33 100.0%
Standard Deviation = 1.27
Warshawsky's model is inappropriate to determine whether a disparate impact in fact exists concerning pregnant employees because it does not take into account the different impact of the leave policy on all pregnant and non-pregnant employees. Disparate impact analysis recognizes that not all employees are similarly situated and, thus, a facially neutral employment policy may effect one group more than another because of the differences in the group. For example, in Griggs v. Duke Power, 401 U.S. 424, 28 L. Ed. 2d 158, 91 S. Ct. 849 (1971), the United States Supreme Court held that an employer's high school diploma requirement, which did not have an adequate business justification, had a disparate impact on the hiring of black applicants because fewer black applicants had high school diplomas. In that case, the disparity focus was between all black applicants and all white applicants -- not between all black applicants who did not have a high school diploma and all white applicants who did not have a high school diploma. Had the analysis been of the latter, as Warshawsky argues it should have been in this case with respect to pregnant and non-pregnant employees requiring leave, there would have been no disparate impact.
The EEOC's model, however, is also incorrect. By comparing "pregnant" and "non-pregnant" employees, disparate impact may be shown when none exists. To illustrate the problem, if Warshawsky had fired only four of the first-year pregnant women, but had allowed sick leave for the other seventeen, the analysis would be as follows:
GROUP TOTAL # FIRED % FIRED
Pregnant Employees 21 4 19.0%
Non-Pregnant Employees 1,858 33 1.8%
Standard Deviation = 5.66 n4
Had 17 of the 21 pregnant employees received sick leave, the facially neutral policy would not have a disparate impact on pregnant women. Yet, applying EEOC's statistical analysis would result in a finding of disparate impact.
EEOC does not cite any case, and this court has not found any, in which a statistical model has been used to show disparate impact in a "pregnancy discrimination" case.
To determine what statistical model is appropriate, it is necessary to examine the PDA.
Congress enacted the PDA for the specific purpose of overruling General Electric v. Gilbert, 429 U.S. 125, 50 L. Ed. 2d 343, 97 S. Ct. 401 (1976), in which the United States Supreme Court held that exclusion of pregnancy, from an otherwise comprehensive employee disability plan, was not sex discrimination actionable under Title VII. The purpose of the PDA was "to change the definition of sex discrimination in Title VII to reflect the common sense view and to ensure that working women are protected against all forms of employment discrimination based on sex." H.R. REP. No. 948, 95th Cong., 2d Sess. 2, 3 (1978) reprinted in 1978 U.S. CODE CONG. & AD. NEWS 4749, 4751. The PDA did not add a cause of action for "pregnancy discrimination" to Title VII. In enacting the PDA, Congress acknowledged that the procreative function of women has been used as a means to prevent full equality for women in the workplace. It made clear that pregnancy was a condition that affected women employees and, as such, was protectable by Title VII as "employment discrimination based on sex."
In the present case, pregnant first-year employees are discharged at a significantly higher rate than non-pregnant first-year employees. This occurs because pregnant employees need more time off from work than non-pregnant employees. Because only women can get pregnant, if an employer denies adequate disability leave across the board, women will be disproportionately affected. n8 Thus, the statistical analysis under a disparate impact theory is the traditional "sex discrimination" n9 analysis. n10 (See table below).
GROUP TOTAL # # FIRED PERCENTAGE
Female Employees 1,105 50 4.5%
Male Employees 773 3 0.4%
Standard Deviation = 5.5
Under this analysis, female first-year employees were eleven times more likely to be fired than male first-year employees because of Warshawsky's first-year sick-leave policy. The likelihood of a male being fired because of the policy was only 8.89% of the likelihood of a female being fired. If there were no disparate impact -- that is females had the same likelihood of being fired as males under the policy -- only 4 to 5 females would have been fired due to the policy (0.004 X 1,105).
The standard test for determining whether an employment policy has a disparate impact is the eighty percent rule found in EEOC Uniform Guidelines on Employee Selection Procedures, 29 C.F.R. § 1607.4D. See B. SCHLEI & P. GROSSMAN, EMPLOYMENT DISCRIMINATION LAW 74 n. 46 (1976) ("80% rule" used to answer the question: "How much impact is enough?"). The Seventh Circuit summarized the eighty percent rule in United States v. City of Chicago, 663 F.2d 1354, 1357 n. 8 (7th Cir. 1981) (en banc) as follows:
The failure of a sex, race or ethnic group to have a success rate which is at least 80% of the rate of the most successful group is considered evidence of adverse impact.
If the likelihood of being fired for a male were 80% of the likelihood of being fired for a female, 27 to 28 males, as opposed to three, would have been fired due to the policy ((0.045 X.80) X 773). Or stated in terms of the rule, the female success rate would have been 0.005 and only 5 to 6 women would have been fired instead of 50.
Therefore, EEOC has clearly established a prima facie case of sex discrimination under Title VII using a disparate impact theory of liability, and analysis now turns to whether Warshawsky had a "legitimate business reason" for its first-year sick-leave policy.
2. Business Justification
Once a case of disparate impact has been established, the dispositive issue is whether a "challenged practice serves, in a significant way, the legitimate employment goals of the employer." Wards Cove Packing Co., Inc. v. Atonio, 490 U.S. 642, 658-59, 104 L. Ed. 2d 733, 109 S. Ct. 2115 (1989). This involves two considerations: (1) the justification that the employer offers for the employment policy; and (2) the availability of other practices to achieve the same business end with less disparate impact. Id. at 658. Although the burden of persuasion remains with the plaintiff, the employer carries the burden of producing evidence of a business justification. Id. at 659.
Warshawsky justifies its first-year sick-leave policy on the grounds that it used the availability of sick leave as a "reward" for employees who continued with the company for more than one year. Defendant claims a high employee turnover during the first year of employment and, therefore, withheld benefits from first-year employees until they had shown their staying power with the company. Defendant, however, merely recites the above reason as its business justification without providing any evidence for it. Rather, no one in management knew the reason for the policy; the policy just existed. Mr. Warshawsky, the owner of the company, testified that he did not know the origin of the policy. Mr. Bradley testified: "I think it's necessary" without offering any business reason for his opinion. Moreover, the data indicates that approximately 77 percent of the turnover in first-year employees occurred during the first 90 days of employment.
Mere recitation of a policy's necessity is not sufficient to carry defendant's burden of production under Title VII. Wards Cove, 490 U.S. at 659. Without any evidence to justify its policy, Warshawsky cannot overcome EEOC's substantial prima facie evidence of the sick-leave's discriminatory impact on females.
As to the availability of alternative practices, defendant changed its policy to a less discriminatory one in 1989. Defendant changed its first-year sick-leave policy from one year to 60 or 90 days depending on the employee's classification. Because employee turnover occurred most frequently during the first two to three months of employment, the new policy served the same business end: reduction of turnover. EEOC has carried its burden of persuasion that Warshawsky's first-year sick-leave policy did not serve, in a significant way, any legitimate employment goal. Therefore, summary judgment on EEOC's disparate impact count is granted for plaintiff EEOC and against defendant Warshawsky.
B. Disparate Treatment
A disparate treatment suit is one in which an employee alleges that she has been treated less favorably than her peers. Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 256, 67 L. Ed. 2d 207, 101 S. Ct. 1089 (1981); see also McDonnell Douglas Corp. v. Green, 411 U.S. 792, 36 L. Ed. 2d 668, 93 S. Ct. 1817 (1973). Disparate treatment suits require either proof of discriminatory intent or establishment of a prima facie case of disparate treatment without actual proof of intent.
International Bd. of Teamsters v. United States, 431 U.S. 324, 335 n. 15, 52 L. Ed. 2d 396, 97 S. Ct. 1843 (1977). In a government disparate treatment suit, the government must demonstrate "that unlawful discrimination has been the regular policy of the employer, i.e., that 'discrimination was the company's standard operating procedure -- the regular rather then the unusual practice.'" Coates v. Johnson & Johnson, 756 F.2d 524, 532 (7th Cir. 1985); EEOC v. Sears, Roebuck & Co., 628 F. Supp. 1264, 1273-74 (N.D. Ill. 1986), aff'd, 839 F.2d 302 (7th Cir. 1988). Pattern and practice evidence usually consists of statistical evidence of substantial disparities between the protected group and the unprotected group, buttressed by evidence of general discriminatory policies or specific instances of discrimination. Coates, 756 F.2d at 532. Once a prima facie case is established, the burden shifts to the employer to defeat the showing by demonstrating that the government's statistics are inaccurate or by providing a nondiscriminatory explanation for its actions.
As shown above, EEOC has presented statistical evidence of a substantial disparity between female and male employees due to Warshawsky's no sick-leave policy -- 5.5 standard deviations. To prove disparate treatment on summary judgment, EEOC must buttress the above evidence with further evidence of either general discriminatory policies or specific instances of discrimination. Coates, 756 F.2d at 532. To do so, EEOC presents further statistical evidence by analyzing the effect of Warshawsky's lack of a formal sick-leave policy. EEOC presents evidence that many first-year male employees were granted a series of "sick days" or "personal days" and not terminated pursuant to the sick-leave policy, while almost all of the women, pregnant and non-pregnant, who requested a leave of absence were terminated under the no sick-leave policy. For example, EEOC shows that between January 1, 1985 and March, 1989, defendant allowed five out of eight first-year male employees, who needed more than two weeks of sick leave, to take extra days without being terminated pursuant to the first-year sick-leave policy. During the same time period, only 16 out of 66 requests by females were granted without termination under the no sick-leave policy (See table below).
EMPLOYEES REQUESTING LEAVE TOTAL # FIRED % FIRED
Female 66 50 75.8%
Male 8 3 37.5%
Standard Deviation = 2.3
Warshawsky admits the above statistics, arguing only that the comparison is immaterial to this case. As discussed above, however, EEOC brought a sex discrimination suit and the above statistical analysis is appropriate.
EEOC does not provide any evidence of specific instances of discrimination against women. EEOC simply buttresses its earlier disparate impact statistics with a second statistical analysis resulting in 2.3 standard deviations. On the other hand, Warshawsky provides evidence that it treated men and women alike -- whether pregnant or non-pregnant. Furthermore, Warshawsky refutes EEOC's second statistical analysis concerning requests for leave. Warshawsky argues first that the statistics are inaccurate because they are not broken down by department and supervisor and second that an accurate comparison cannot be made between employees who requested one or two extra days and those who requested several extra weeks.
Because EEOC attempts to support its disparate impact claim with a second statistical analysis, its buttressing evidence resulting in 2.3 standard deviations must be carefully examined and defendant's contentions thoroughly considered. EEOC v. Sears, Roebuck & Co., 628 F. Supp. 1264, 1286-87 (N.D. Ill. 1986), aff'd, 839 F.2d 302 (7th Cir. 1988) (standard deviations between two and three must be carefully examined). Warshawsky's first argument is that each department's supervisors made unfettered decisions as to whether a first-year employee was granted sick days or sick leave and, therefore, EEOC's statistics are inaccurate because they do not compare similarly situated employees. Defendant is correct that where an employer's departments are diverse and each department's hiring needs are different and specialized, it is inappropriate to accept a statistical comparison that does not take into account the differences. Sears, Roebuck & Co., 839 F.2d at 311. Warshawsky, however, points to no differences between its departments. That Warshawsky has several departments without more is not evidence that the employees are not similarly situated.
Warshawsky's second argument is that an accurate statistical comparison cannot be made between employees who needed one extra day of sick leave and those who needed several extra weeks. Warshawsky presents evidence of several occasions where women were given an extra day or two of sick leave and occasions where males were denied additional sick leave and terminated. Warshawsky argues that these decisions turned upon the length of time requested, not upon whether the employee was male or female, pregnant or not pregnant. EEOC's evidence, statistical and otherwise, fails to take into account whether employees who were granted leave as extra sick days had requested one or two extra days as opposed to several extra weeks. EEOC presents no evidence as to whether male employees were given several "weeks" of sick or personal "days" rather than being terminated under the first-year sick-leave policy. Genuine issues of material fact exist as to whether Warshawsky had a discriminatory policy concerning female employees. Although EEOC's evidence is sufficient to hold that Warshawsky's policy had a disparate impact on women, without more it is not sufficient on summary judgment to hold that Warshawsky intentionally discriminated against women. Thus, partial summary judgment is inappropriate on a theory of disparate treatment, and both parties' motions are denied as to the disparate treatment count.
Warshawsky argues that EEOC waited too long to file this suit and, thus, is barred by the equitable doctrine of laches. Title VII does not contain an express limitation on the time within which EEOC may bring an enforcement suit. The doctrine of laches may bar EEOC, however, if it has "delayed inexcusably and the defendant was materially prejudiced by its delay." EEOC v. Massey-Ferguson, Inc., 622 F.2d 271, 275 (7th Cir. 1980). Defendant bases its affirmative defense solely upon the fact that four years elapsed between January of 1986, when EEOC received a complaint against Warshawsky, and March of 1990, when EEOC filed this suit. Warshawsky presents no evidence on the issue of delay. Accordingly, summary judgment as to defendant's affirmative defense of laches is denied.
IV. EMPLOYMENT APPLICANTS
Defendant grounds its motion for partial summary judgment on the argument that this court is without subject matter jurisdiction to entertain claims other than those claims specified in EEOC's August 19, 1988 Determination Notice. The only issue to which this argument applies is whether EEOC may press its claim on behalf of female applicants. EEOC's Determination Notice refers only to employees and does not mention applicants. This lack, however, does not affect this court's subject matter jurisdiction to hear Title VII claims. Jackson v. Seaboard Coast Line Ry., 678 F.2d 992, 1002 (11th Cir. 1982) (citing Zipes v. TWA, Inc., 455 U.S. 385, 393, 71 L. Ed. 2d 234, 102 S. Ct. 1127 (1982)) ("Congress did not intend the conditions precedent to a Title VII action to constitute jurisdictional prerequisites.").
EEOC's complaint brings suit on behalf of "a class of aggrieved female employees and potential employees." "Potential employees" are presumed to be applicants for employment. Parties use interchangeably the terms "potential applicants" and "potential employees." Both terms suggest an indeterminate class and neither term is appropriate. An issue exists as to whether "applicants" are to be included during this suit's remedy phase, as opposed to only employees. This issue must be addressed as a part of the parties' proposals concerning relief.
IT IS THEREFORE ORDERED that:
(1) Plaintiff's motion for partial summary judgment is granted on the disparate impact claim.
(2) Plaintiff's motion for partial summary judgment is denied on the disparate treatment claim.
(3) Defendant's motion for partial summary judgment is denied as to all claims.
(4) Defendant's motion to strike is denied.
(5) Plaintiff's motion to strike is denied.
(6) The parties are to submit separate statements concerning proposed relief based on liability on the disparate impact claim by Monday, July 22, 1991. No extensions will be granted.
(7) Status hearing set for Friday, July 26, 1991 at 2:00 p.m.