American Bar Association. Section of Labor and Employment Law 1999Annual Meeting AUGUST 7-11, 1999 ATLANTA, GEORGIA CLASS ACTIONS IN EMPLOYMENT LAW

Transcription

1 American Bar Association Section of Labor and Employment Law 1999Annual Meeting AUGUST 7-11, 1999 ATLANTA, GEORGIA CLASS ACTIONS IN EMPLOYMENT LAW by Hunter R. Hughes, III Terry L. Houser Rogers & Hardin LLP Atlanta, Georgia

2 TABLE OF CONTENTS Page I. INTRODUCTION... 1 II. RECENT CHANGES IN EMPLOYMENT DISCRIMINATION CLASS ACTIONS...2 A. MORE CLASS ACTIONS ARE BEING FILED Publicity of Large Class-Action Settlements... 2 B. CHANGES IN THE BURDEN OF PROOF: APPLICATION OF DISPARATE IMPACT THEORY TO SUBJECTIVE SELECTION PRACTICES Overview of Disparate Impact Analysis The Law Prior to Watson, Wards Cove and the Civil Rights Act of C. THE APPLICATION OF THE CIVIL RIGHTS ACT OF a. Civil Rights Act of Punitive Damages... 6 III. PROCEDURAL ISSUES... 7 A AMENDMENT TO RULE 23 ALLOWING DISCRETIONARY APPEAL... 7 B. PROPOSED CONGRESSIONAL AMENDMENTS TO RULE 23 AND DIVERSITY JURISDICTION STATUTE... 8 C. COMMUNICATIONS WITH CLASS MEMBERS... 9 D. ARBITRATION ISSUES E. DEFINITION OF CLASS IV. USE OF STATISTICAL EVIDENCE IN EMPLOYMENT DISCRIMINATION CLASS ACTIONS A. REQUIREMENTS FOR USE OF STATISTICS B. EXAMPLES OF THE USE OF STATISTICS IN EMPLOYMENT DISCRIMINATION CASES C. USING STATISTICAL EVIDENCE TO INFER CLASS-WIDE INTENTIONAL DISCRIMINATION D. REBUTTING STATISTICAL EVIDENCE V. COMMON CHALLENGES TO EMPLOYMENT PRACTICES A. CHALLENGES TO HIRING PRACTICES Class Certification Issues Pattern or Practice Claims EEOC Actions Class Challenges to Subjective Decisionmaking Using Anecdotal Evidence To Attack An Employer s Employment Practices Whether Opt Outs Will be Permitted i

3 B. CHALLENGES TO PROMOTION PRACTICES The Effects Of Subjective Decisionmaking On Promotions C. CHANNELING CLAIMS D. COMPENSATION CLAIMS E. TERMINATIONS VI. REDUCING THE RISK OF CLASS-ACTION LITIGATION AND LIABILITY A. AVOID EXCESSIVELY SUBJECTIVE EMPLOYMENT SELECTION PRACTICES B. PROVIDE EEO TRAINING C. ADOPT AND IMPLEMENT AN EFFECTIVE AND CREDIBLE INTERNAL COMPLAINT RESOLUTION PROCESS D. MINORITIES AND FEMALES SHOULD BE REPRESENTED IN MANAGEMENT AND SUPERVISORY DECISIONS E. REQUIRE HUMAN RESOURCES OR LEGAL REVIEW OF DISCIPLINE AND DISCHARGE DECISIONS BEFORE THEY ARE IMPLEMENTED ii

4 CLASS ACTIONS IN EMPLOYMENT LAW by Hunter R. Hughes, III Terry L. Houser Rogers & Hardin LLP Atlanta, Georgia I. INTRODUCTION After a steady decline for years in the number of employment discrimination class actions filed, class action cases are on the upswing again. A number of high profile cases have been brought in the last several years, most notably the litigation involving claims of racial discrimination against Texaco, Smith, Barney and a host of large retailers. There seem to be at least several reasons for this increase in employment class actions: The Civil Rights Act of 1991 allows compensatory and punitive damages, dramatically increasing the potential exposure and burden of litigation for defendants in Title VII cases; The magnitude of the potential recovery and the burden of defense of a class action, plus the complexity of class action procedures encourage large settlements; The prospect of large attorneys fees provisions have encouraged plaintiff s lawyers to make the substantial up-front out-of-pocket investment necessary to file and prosecute class actions to the point where the burdensomeness and potential for adverse outcome often will cause the employer to consider a substantial settlement; and The Equal Employment Opportunity Commission has stepped up its charges and intervention in large cases, particularly class actions. The reemergence of discrimination class actions can be seen in recent headlines, such as: Sexual Harassment: EEOC Suit Charges That Dial Corp. Failed To Act On Harassment Charges, BNA Employment Policy & Law Daily News (May 25, 1999). Race Discrimination: Suit Charges Coca-Cola Uses Classwide Race Bias In Pay, Promotions, Discipline, BNA Employment Policy & Law Daily News (April 27, 1999). Race Discrimination: Judge Approves $12.1 Million Settlement In Part-Timeworkers Lawsuit Against UPS, BNA Employment Policy & Law Daily News (April 15, 1999). 1

5 CSX Settles Discrimination Suit, Charleston Gazette, Jan. 30, 1999, at 7A, available in 1999 WL Airline Industry Information: Boeing to Settle Class-Action Lawsuits Over Discrimination, Airline Indus. Info., Jan. 28, 1999, available in 1999 WL Boeing Plans to Pay $15 Million to Settle Suits Alleging Job Bias, Wall St. J., Jan. 25, 1999, at C19, available in 1999 WL-WSJ UPS to Pay $12 Million in Bias Suit Settlement, L.A. Times, Jan. 19, 1999, at C3, available in 1999 WL L. M. Sixel, Pennzoil Settles Racial Bias Lawsuit - Black Plaintiffs to Share $6.75 Million, Hous. Chron., Nov. 11, 1998, at 1, available in 1998 WL Patrick McGeehan, Judge Approves Class-Action Settlement For Sex Harassment at Smith Barney, Wall St. J., July 27, 1998, available in 1998 WL-WSJ Bias Case Could Cost Utility $65 Million, L.A. Times, Feb. 11, 1998, at D3, available in 1998 WL II. RECENT CHANGES IN EMPLOYMENT DISCRIMINATION CLASS ACTIONS A. More Class Actions are Being Filed 1. Publicity of Large Class-Action Settlements a. The increase in class-action discrimination lawsuits is attributable, in part, to publicity in recent years about large class-action settlements. For example, in Roberts v. Texaco, S.D.N.Y., No (March 4, 1997), 1,350 black salaried employees sued Texaco for racial discrimination. It was widely publicized that company officials allegedly destroyed certain documents to avoid having to produce them to the plaintiffs. Soon thereafter a settlement was reached where the company agreed to pay $115 million to plaintiffs, $26.1 million for pay increases and $35 million in programs, to encourage equal opportunity. According to one plaintiff s lawyer, the settlement would work out to approximately $60,000 per plaintiff. BNA s Employment Discrimination Report No Class Members Chose to Opt Out of Texaco Settlement, Attorney Says, Vol. 8, at p Other examples of widely publicized class actions include Kraszewski v. State Farm General Ins. Co. (reported $245 million settlement); Haynes v. Shoney s, Inc. (reported $132.5 million settlement); and Stender v. Lucky Stores, Inc. (reported $107 million settlement). 2

6 b. The certification of a large class in and of itself creates exposure to enormous potential litigation expenses, such as attorneys fees (including payment of plaintiff class counsel s fees), expert witness fees, and other costs. Employment class actions are often divided into two stages : (1) a trial to determine whether there is any class-wide liability, and (2) a process to determine the amount of damages to be awarded to individual class members. Both stages can involve huge litigation expenses. B. Changes in the Burden of Proof: Application of Disparate Impact Theory to Subjective Selection Practices 1. Overview of Disparate Impact Analysis a. The disparate impact analysis applies to employment practices that are facially neutral but which have a greater impact on a protected group (i.e., older workers are affected in significantly greater percentages by the practice). b. To establish a prima facie case of disparate impact, plaintiff must: (1) identify a specific employment practice; (2) show the practice has an adverse impact on the basis of a prohibitive factor; and (3) and prove causation. See, e.g., Wards Cove Packing Co. v. Atonio, 490 U.S. 642 (1989)(Title VII case). The impact upon the protected group must be statistically significant in order to show disparate impact. See Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988)(Title VII case). c. After plaintiff shows a prima facie case, the burden shifts to the employer to show that the practice is job-related and consistent with business necessity. Civil Rights Act of The Law Prior to the Civil Rights Act of 1991 a. Traditionally, the disparate-impact theory had applied only to identifiable, objective selection devices. See, e.g., Griggs v. Duke Power Co., 401 U.S. 424 (1971) (general aptitude tests and high school diploma requirement); Dothard v. Rawlinson, 433 U.S. 321 (1977) (height and weight requirements); Connecticut v. Teal, 457 U.S. 440 (1982) (written examination). As a practical matter, this resulted in abandonment by many employers of most pass/fail tests. The cost of proving the business necessity of such tests through validation studies was often considered too great, and the use of such tests without validation was considered to have too much risk. The lower courts were divided on the question whether disparate impact theory could be applied to challenge subjective selection processes. 3

7 b. Many employers assumed the U.S. Equal Employment Opportunity Commission s ( EEOC ) Uniform Guidelines on Employee Selection Procedures ( Guidelines ), 29 C.F.R et seq., did not apply to subjective selection processes (despite language in the Guidelines that suggests they did apply), and narrowly interpreted the EEOC s requirement that they maintain... records or other information which will disclose the impact which its tests and other selection procedures have upon employment opportunities of persons by identifiable race, sex or ethnic group[s]. 29 C.F.R (A) (emphasis added). 3. Watson, Wards Cove and the Civil Rights Act of 1991 a. In Watson v. Fort Worth Bank & Trust, 487 U.S. 977 (1988), the United States Supreme Court unanimously held that disparate impact theory could be applied to subjective selection devices. b. However, a plurality of the Supreme Court in Watson (Justice O Connor, joined by Chief Justice Rehnquist and Justices White and Scalia) were concerned about the practical implications of this expansion of disparate impact theory to subjective selection devices. Specifically, they were concerned that if it were too easy for an employee to launch a broad-scale disparate impact challenge against all kinds of selection processes, the costs and risks of litigation might force the employer to resort to use of quotas: We recognize that today s extension of [disparate impact] theory into the context of subjective selection practices could increase the risk that employers will be given incentives to adopt quotas or to engage in preferential treatment. Because Congress has so clearly and emphatically expressed its intent that Title VII not lead to this result, 42 U.S.C. 2000e-2(j), we think it imperative to explain in some detail why the evidentiary standards that apply in these cases should serve as adequate safeguards against the danger that Congress recognized. 487 U.S. at 993. c. The following year in Wards Cove Packing Co. v. Atonio, 490 U.S. 642 (1989), the Supreme Court held that the employer was only required to present evidence that a selection practice with a disparate impact serves, in a significant way, the legitimate employment goals of the employer. 490 U.S. at 659 (emphasis added). The Court further held that the burden of proof remains with the disparate-impact plaintiff. This shift in the burden of proof became the focus of attack by the proponents of the Civil Rights Act of

8 d. The Civil Rights Act of 1991 shifted the burden of proof back to the employer in all disparate impact cases. Where disparate impact of any selection practice, subjective or objective, is shown by the plaintiff, the employer now has the burden of proving that the practice is job-related and consistent with business necessity. The proponents of the Civil Rights Act of 1991 (and most employers) never fully appreciated that the change in the burden of proof in Wards Cove was the result of concerns about the practical implications of the Supreme Court s expansion of disparate impact theory to apply to literally all forms of employee selection processes. The provisions in the 1991 Civil Rights Act about disparate impact must be understood in this context. They did not simply restore the disparate impact theory of Griggs that existed prior to 1988, because prior to Watson and Wards Cove, the disparate impact theory had not been applied to the whole universe of employee selection processes. C. The Application of the Civil Rights Act of Compensatory Damages a. Civil Rights Act of 1991 i. Prior to the Civil Rights Act of 1991, compensatory and punitive damages were not available under Title VII. See Walker v. Ford Motor Co., 684 F.2d 1355 (11th Cir. 1982); Protos v. Volkswagen of America, Inc., 797 F.2d 129 (3d Cir. 1986); Johnston v. Ethyl Corp., 683 F. Supp (M.D. La. 1988). ii. Under the Civil Rights Act of 1991, subject to caps, compensatory and punitive damages now are recoverable in actions under Title VII where the employer has engaged in intentional discrimination and the plaintiff cannot recover under 42 U.S.C U.S.C. 1977A. In addition, compensatory and punitive damages now are available under the Rehabilitation Act of 1973 (the Rehabilitation Act ) and the Americans with Disabilities Act ( ADA ). Id. Subsection (a)(3) provides a defense in actions under the Rehabilitation Act and ADA where the employer has made a good faith effort to accommodate the plaintiff s handicap. iii. The statute provides that compensatory damages do not include backpay, interest on backpay, or any other type of relief authorized under section 706(g) of the Civil Rights Act of iv. However, most federal courts are requiring more than ordinary hurt feelings to support a substantial award of compensatory damages. In Annis v. County Of Westchester, 136 F.3d 239 (2d Cir. 1998), for instance, the court held that a female police officer s testimony that she suffered 5

9 emotional distress as result of sex discrimination was insufficient to sustain award of compensatory damages, where she had not alleged any physical manifestations of her emotional distress, she remained on job, and she introduced no affidavit or other evidence to corroborate her testimony that she needs and has had counseling. v. See Lindale v. Tokheim Corp., 145 F.3d 953 (7th Cir. 1998) (an award of $100,000 in compensatory damages could not be upheld even if violation of Title VII had been proven by female mechanical engineer who alleged that she was constructively discharged when she quit after it became evident that she was not going to be given promotion for which she had been recommended, where one quarter of award was for distress caused by constructive discharge, but this amount would have been excessive by at least factor of two even if there had been outright discharge in absence of any evidence of significant emotional injury or other circumstances of aggravation, and most of rest of award was for conjectured reduction in present value of her future earnings as result of failure to be promoted). 2. Punitive Damages a. Punitive damages were not recoverable under Title VII prior to the Civil Rights Act of See Walker v. Ford Motor Co., 684 F.2d 1355 (11th Cir. 1982); Johnson v. Al Tech Specialties Steel Corp., 731 F.2d 143 (2d Cir. 1984); Thompkins v. Stuttgart School Dist. No. 22, 787 F.2d 439 (8th Cir. 1986). b. Punitive damages are now recoverable under Title VII (subject to caps), the Rehabilitation Act and the ADA (subject to caps) against private employers where the plaintiff shows the employer engaged in a discriminatory practice or discriminatory practices with malice or with reckless indifference to the federally protected rights of an aggrieved individual. 42 U.S.C. 1981a(b)(1). c. In Kolstad v. American Dental Ass n, No (U.S. June 22, 1999), the Supreme Court held that an employer s conduct need not be independently egregious to satisfy 1981a s requirements for a punitive damages award, although evidence of egregious behavior may provide a valuable means by which an employee can show the malice or reckless indifference needed to qualify for such an award. The court stated that the terms malice and reckless indifference pertain not to the employer s awareness that it is engaging in discrimination, but to its knowledge that it may be acting in violation of federal law. The court recognized that there will be circumstances where intentional discrimination does not give rise to punitive damages liability under this standard, as where the employer is unaware of the relevant federal prohibition or discriminates with the distinct belief that its discrimination is lawful, where the underlying theory of discrimination is novel or otherwise 6

10 poorly recognized, or where the employer reasonably believes that its discrimination satisfies a bona fide occupational qualification defense or other statutory exception to liability. It then held that, in the punitive damages context, an employer may not be vicariously liable for the discriminatory employment decisions of managerial agents where these decisions are contrary to the employer s good faith efforts to comply with Title VII. 3. Back pay awards still comprise a significant portion of Title VII damage awards, however, in the routine discrimination class action. For instance, in Berger v. Iron Workers, Reinforced Rodmen, Local 201, 170 F.3d 1111 (D.C. Cir. 1999), the court addressed the calculation of damages for individual class members in a class action discrimination action against an union alleging discriminatory exclusion of black workers from the union. In that case, the court held that a special master appointed by the court did not commit clear error in rejecting a specific cohort methodology urged by the union s expert in determining the entitlement of black construction workers to back pay, where the special master concluded that the group of 38 individuals was a statistically insignificant sample and that the union could have found sufficient work for its members, so that fixed-pie analysis used by union s expert to determine the average number of hours that each union member would have worked in the absence of discrimination was not accurate. However, the court found that the special master, in calculating back pay for plaintiffs excluded from union membership and limited in referrals, erred in his assumption that the union could have found extra work for its members if black workers had been admitted. There was no evidence that there actually was such work available in the metropolitan area; if black workers wanted damages awarded on basis of hours not referred through the union hall, they should have borne the burden of at least making prima facie showing that additional hours were available to the union. III. PROCEDURAL ISSUES A Amendment to Rule 23 Allowing Discretionary Appeal 1. In 1998, Federal Rule of Civil Procedure 23 was amended to add a provision allowing a discretionary, interlocutory appeal of a class certification order. Fed. R. Civ. P. 23(f) now provides that: (f) Appeals. A court of appeals may in its discretion permit an appeal from an order of a district court granting or denying class action certification under this rule if application is made to it within ten days after entry of the order. An appeal does not stay proceedings in the district court unless the district judge or the court of appeals so orders. 2. Rule 23(f) thus provides an alternative means for review of a class certification order besides the certification of an order for interlocutory appeal pursuant to 28 7

11 U.S.C. 1292(b). Rule 23(f) has the advantage of not requiring approval of the district court, though it does require the exercise of discretion on the part of the Court of Appeals. B. Proposed Congressional Amendments to Rule 23 and Diversity Jurisdiction Statute 1. A proposed bill now pending in Congress would significantly alter both jurisdiction and procedure under Rule 23. The bill, entitled the Class Action Fairness Act of 1999, is intended to end perceived abuses (mainly by plaintiffs class counsel) in class action litigation. 2. Most significantly, the bill would federalize many class action cases by altering the diversity of citizenship requirement in class actions, allowing a case to be brought in or removed to federal court where there is diversity between any plaintiff and any defendant. Specifically, the proposed bill would impose the following requirements: The Principle of Plain English : The bill requires that all class notices and notices of proposed class action settlements be written in easily understood terms and that settlement notices include the amount and source of attorneys fees, along with other material information. Notice to State Attorneys General: The bill requires notification of all state attorneys general of any proposed class settlement that would affect residents of their states. This notice gives state attorneys general an independent voice and an opportunity to object if the settlement terms are unfair. The stated intent is to provide more consumer protection for their citizens against abusive practices. Attorneys Fees Based on Actual Damages: The bill provides that class action attorneys fees are to be based on a reasonable percentage of damages actually paid to class members, the actual costs of complying with the terms of a settlement agreement, as well as any future financial benefits. In the alternative, the bill provides that, to the extent the law permits, fees may be based on a reasonable hourly (lodestar) rate. This provision is intended to discourage settlements that give attorneys exorbitant fees based on hypothetical overvaluation of coupon settlements, yet allow for reasonable fees in all cases, including those involving injunctive relief. Removal of Multistate Class Actions to Federal Court: The bill allows more class action lawsuits to be removed from state court to federal court, either by a defendant or an unnamed class member. A class action would qualify for federal jurisdiction if the total damages exceed $75,000 and parties include 8

12 citizens from multiple states. However, the bill provides that cases would remain in state court where the substantial majority of class and primary defendants are from the same state and that state s law would govern, or the primary defendants are states and a federal court would be unable to order the relief requested. Mandatory Sanctions for Frivolous Suits: The bill would require a court to impose sanctions for violations of Rule 11 of the Federal Rules of Civil Procedure in class action pleadings. The nature and extent of sanctions would remain discretionary. C. Communications with Class Members 1. Abdallah v. Coca-Cola Co., 79 Fair Empl. Prac. Cas. (BNA) 1490 (N.D. Ga. 1999), addressed communications by counsel to unnamed class members. The court held that lawyers for the plaintiffs in a race discrimination lawsuit filed against Coca-Cola Co. in Atlanta would talk freely to potential class members who contact them, but company communications to its employees regarding the lawsuit was subject to certain restrictions. Specifically, the court held that Coke, in any internal communications about the case to all employees except managers, would be required to include statements that it represents the company s position and that it will not retaliate against employees who participate in the lawsuit. The suit involved allegations of racial discrimination by black employees of Coke. The court order was prompted by an sent to Coke employees by Coca-Cola Chairman and Chief Executive Officer M. Douglas Ivester. Following widespread publicity about the lawsuit, Investor sent two messages about the litigation to all Coke employees in the United States defending the company s personnel and employment policies and claiming that the plaintiffs lawsuit contained numerous factual errors. D. Arbitration Issues 1. EEOC v. Frank s Nursery & Crafts, Inc., 79 Fair Empl. Prac. Cas. (BNA) 936 (6th Cir. 1999), held that the EEOC may pursue a Title VII claim for injunctive relief and damages against an employer even though the plaintiff waived her right to sue as part of an arbitration agreement. The court reasoned that Title VII does not permit an individual to take away the EEOC s enforcement authority even if she wishes to withdraw her charge; the legislative history of both Title VII and its 1972 amendments illustrate that Congress shifted much of responsibility of ensuring compliance with Title VII from private individuals to the EEOC; the EEOC is not an ordinary plaintiff and is not bound by procedural requirements applicable to private litigants because of the EEOC s role in representing the public interest; and empowering a private individual to take away the congressional mandate that EEOC represent an interest broader than that of a particular individual would grant 9

13 that individual the ability to govern whether and when EEOC may further the national initiative against employment discrimination. The court also held that the Federal Arbitration Act (FAA) did not bar the EEOC from seeking monetary damages under Title VII even though the employee had waived her right to sue as part of arbitration agreement, since an individual could not by waiving her statutory right to vindicate her own interest waive the statutory right of the EEOC to vindicate the public interest unless the government had agreed to such a waiver. Note that the Second Circuit has reached an opposite result on the issue of the right of the EEOC to recover money damages for a plaintiff who signed an arbitration agreement. See EEOC v. Kidder, Peabody & Co., 156 F.3d 298 (2d Cir.1998). E. Definition of Class 1. In Vizcaino v. U.S. District Court For The Western District Of Washington, 173 F.3d 713 (9th Cir. 1999), the Ninth Circuit s third decision involving Microsoft Corp. s use of contingent workers, the court ruled that the district court erred upon remand of the previous appeal in narrowing the class of contingent workers who may be entitled to participate in the company s employee stock purchase plan. In that case, the court granted the workers petition for a writ of mandamus to compel the district court to comply with the appeals court s July 1997 en banc decision holding that workers classified as independent contractors were improperly denied employee benefits. The district court had certified a class consisting of [a]ll persons employed by Microsoft... who are denied employee benefits because they are considered independent contractors or employees of third-party employment agencies, but who meet the definition of employees of Microsoft... under the common law. Id. at 716. However, on remand from the Ninth Circuit, the district court in July 1998 revised the class definition to exclude temporary employees hired after 1990, when the Internal Revenue Service reclassified a number of independent contractor positions and Microsoft converted many independent contractors to temporary agency employees. The Ninth Circuit found that the class definition was a central premise of both the panel and en banc decisions. It also found that Microsoft conceded that the class members generally were common law employees and that the company only reserved the right to object to the employee status of certain persons during certain periods. The en banc decision specifically stated that the stock purchase plan must, essentially, be made available to all employees. Id. at Accordingly, the district court on remand could not alter the class definition. 10

14 IV. USE OF STATISTICAL EVIDENCE IN EMPLOYMENT DISCRIMINATION CLASS ACTIONS A. Requirements for Use of Statistics 1. Statistics are routinely used in class action employment discrimination cases, both by plaintiffs and defendants. The starting point for a statistical analysis in an employment case is a comparison of the composition of the qualified applicant pool to the candidates selected for the positions at issue. See, e.g., Watson v. Fort Worth Bank & Trust, 487 U.S. 977, 997 (1988); Wards Cove Packing Co. v. Atonio, 490 U.S. 642, 649 (1989); Hazelwood School Dist. v. United States, 433 U.S. 299 (1977). See also Maddox v. Claytor, 764 F.2d 1539, (11th Cir. 1985) (Court holds that comparative study based on total Army base population, not on the group of applicants who were interviewed, or even on the group of applicants found qualified or the group of all applicants did not shed any light on plaintiffs discrimination claims). 2. In Hazelwood School District v. United States, 433 U.S. at 308, the Supreme Court established the proposition that for statistics to have any value in a discrimination case, the statistics must compare the composition of the qualified and available labor pool to the composition of the applicant(s) selected. In Hazelwood, African- American applicants for positions as teachers in a public school system alleged that they had been denied employment on the basis of their race. Plaintiffs sought to show they had been discriminated against by comparing the racial composition of the students to that of the teachers employed in the school system. Rejecting this analysis, the Supreme Court held that the relevant statistical analysis should have used as its base the qualified public school teacher population in the relevant labor market, not the racial composition of the students -- or, by implication, the racial composition of the community (emphasis added). 3. Once the appropriate labor pool comparison is identified, plaintiff then must show that there is a sufficient statistical disparity on the basis of race to allow a reasonable presumption of discrimination. See, e.g., Wards Cove, supra. The fact there may be some statistical disparities in selection based upon race is in itself insufficient to show racial discrimination. Watson, supra; Wards Cove, supra. 4. In practice, where statistical evidence establishes a disparity in the number of men and women, for example, in the employment context, this disparity can be attributed to one of three causes: the effect of legitimate, non-discriminatory selection criteria; unlawful discriminatory animus; or chance. Stender v. Lucky Stores, Inc., 803 F. Supp. 259, (N.D. Cal. 1992). Thus, one of the crucial questions for any pattern and practice case is determining at what point the disparity becomes sufficiently large, and the probability of chance sufficiently low, to allow an inference of discrimination to be drawn from the disparity. Id. 11

15 5. To answer this question, courts typically look to the Z-value of a particular disparity. Webb v. Missouri Pac. R.R. Co., 826 F. Supp. 1192, 1202 (E.D. Ark. 1993); Stender, 803 F. Supp. at 323. A Z-value measures the number of standard deviations that exists between an observed value and the expected value. Thus, in situations in which a negative Z-value exists means that the observed value was less than what would be expected by chance from instances in which the particular group s protected status was not a factor. While disparities of two or three deviations are normally sufficient to raise an inference of discrimination, courts are extremely cautious about any inferences drawn from deviations between 1 and 3. Stender, 803 F. Supp. at 323 (quoting Gay v. Waiters and Dairy Lunchmen s Union, Local No. 30, 694 F.2d 531, 551 (9th Cir. 1982)). 6. The sample size also must be sufficiently large to draw valid statistical conclusions and the analysis must account for potential causes of any disparity other than discriminatory reasons. In Fisher v. Wayne Dalton Corp., 139 F.3d 1137 (7th Cir. 1998), for instance, the court held that statistical evidence did not establish that protected-age employee s loss of job in reduction in force was due to age discrimination, even though five of the six oldest office workers lost their jobs, since the sample size was too small to be meaningful, Plaintiff presented no evidence as to the employees qualifications and did not address potentially explanatory variables such as the inverse relationship between age and computer familiarity, two of the five were not actually discharged at all, since one refused a remaining position in department and the other bumped and replaced a less senior employee. B. Examples of the Use of Statistics in Employment Discrimination Cases 1. Ibarra v. Martin, 143 F.3d 286 (7th Cir. 1998), held that statistics concerning terminations by the county probation department chief did not support an inference of intentional discrimination against Hispanics, even though five of 33 discharged employees over five-year period were Hispanic and this amounted to 15 percent of terminations or 153 percent more than would be expected because only six percent of department is Hispanic, where the pool was too small for percentage descriptions to mean anything. Moreover, even if two Hispanics who impliedly resigned and non-hispanics whose termination were also recorded as implied resignation were excluded from the statistics, the termination rate for Hispanics was 11 percent, which was not statistically different from the hypothetical expected six-percent termination rate given the size of the pool. 2. Lalla v. City of New Orleans, No , 1999 WL (E.D. La. Mar. 12, 1999), held that the New Orleans fire department s 50/50 racial hiring quota violates the equal protection rights of white firefighter recruits. In reaching this result, the court found that statistical evidence of a gross manifest imbalance 12

16 between the representation of African Americans in the city s workforce and in the city s fire department was not conclusive evidence of discrimination that could justify racial hiring preferences, the district court held. According to the city s uncontested statistics, the fire department workforce was 17 percent black in 1991 while the city labor force was 59.7 percent black. The fire department s sole reliance on statistics -- without additional specific evidence of recent discrimination, including anecdotal evidence, prior judicial findings, or poor testing methods -- is insufficient to show the city has a compelling state interest that would justify using race as a factor in hiring decisions. 3. Lutheran Church-Missouri Synod v. FCC, 154 F.3d 487 (D.C. Cir. 1998), noted that the fact that statistical evidence can be relevant in determining whether the employer s past practice is discriminatory was not equivalent to concluding that absence of proportionality makes out discrimination. Comparing the proportionality of minorities in the employer s workforce to the proportionality of minorities in the overall population (the Metropolitan Statistical Area or MSA) is never the relevant comparison under such cases; rather, the racial composition of those holding at-issue jobs is compared with the racial composition of qualified applicants or qualified persons in the labor market. That the relevant statistical gauge was not the proportionality of minorities in the overall population is clear from the antidiscrimination rationale of Title VII -- the purpose of statistical evidence is to expose possible discriminatory intent, not to establish a workforce that mirrors the racial breakdown of the MSA. If discrimination under Title VII were defined as non-proportionality, much of the Supreme Court s recent equal protection cases would make little sense. That is, [a]t the heart of the Constitution s guarantee of equal protection lies the simple command that the Government must treat citizens as individuals, not as simply components of a racial, religious, sexual or national class. Id. at 494. C. Using Statistical Evidence To Infer Class-Wide Intentional Discrimination 1. In order to establish an inference of class-wide pattern and practice discrimination, plaintiffs must present statistical evidence demonstrating that there is a longlasting and gross disparity in the representation of members of the class from what would be expected from employment decisions made without regard to the class members protected status. See, e.g., International Bhd. of Teamsters v. United States, 431 U.S. 324, 340 (1977); Hazelwood School Dist. v. United States, 433 U.S. 299, 307 (1977); Anderson v. Douglas & Lomason Co., 26 F.3d 1277, 1285 (5th Cir. 1994), (citations omitted); Webb v. Missouri Pac. R.R.. Co., 826 F. Supp. 1192, 1202 (E.D. Ark. 1993); Stender, 803 F. Supp. at Statistics that show a gross disparity in the treatment of workers based on their protected status may by themselves constitute prima facie proof of a pattern or practice of discrimination. See Hazelwood, 433 U.S. at 307-8; Teamsters,

17 U.S. at ; see also EEOC v. O & G Spring and Wire Forms Specialty Co., 38 F.3d 872, 876 (7th Cir. 1994); Anderson, 26 F.3d at 1285; EEOC v. Olson s Dairy Queens, Inc., 989 F.2d 165 (5th Cir. 1993); Cota v. Tucson Police Dept., 783 F. Supp. 458, 467 (D. Ariz. 1992); Cook v. Billington, 59 Fair Emp. Prac. Cas. (BNA) 1010 (D.D.C. 1992). In fact, some courts have inferred intentional discrimination solely from statistical evidence. O & G Spring and Wire, 38 F.3d at 879; Olson s Dairy Queens, 989 F.2d at ; Cook, 59 Fair Empl. Prac. Cas. at However, in practice, such cases are rare. Cota, 783 F. Supp. at 467; see also Forehand v. Florida State Hosp. at Chattahoochee, 89 F. 3d 1562, 1575 (11th Cir. 1996) (affirming dismissal of pattern and practice claim finding statistical evidence alone insufficient to raise inference of discrimination). Thus, given that statistics alone rarely provide the requisite inference of discrimination, most plaintiffs are forced to bolster their statistics with anecdotal evidence of individual discrimination. See Anderson, 26 F.3d at 1285; see also Cota, 783 F. Supp. at 467; Olson s Dairy Queens, 989 F.2d at 168. Often, such testimony can assist in bringing the cold numbers convincingly to life. Teamsters, 431 U.S. at See also Mullin v. Raytheon Co., 2 F. Supp. 2d 165 (D. Mass. 1998) (statistics purporting to show that the employees aged 50 and older were more likely to be demoted during restructuring than younger workers did not establish that the employer s justification for demoting protected-age employee, which was that it has policy of paying the employees according to level of work that they currently perform and not according to historic salary levels, was pretextual; no inference that older workers were more likely to be demoted can be drawn when calculations are made for protected-age class as a whole, statistics did not address differences in the employees experience, employment history, or value to the employer, and record does not indicate any connection between statistics and the employer s policies and practices toward the employee), aff d, 164 F.3d 696 (1st Cir.), petition for cert. filed (May 5, 1999) (No ). D. Rebutting Statistical Evidence 1. Where the weakness in a plaintiff s statistical evidence is apparent, courts will refuse to infer discriminatory intent. For instance, in Anderson, plaintiffs used statistics to compare the seniority of white and black employees prior to promotion to support their argument that the company was discriminating against blacks by promoting less senior whites. Anderson, 26 F.3d at In response, the defendant offered anecdotal evidence of white employees being promoted only after black employees rejected the position. Based on such evidence, the court found plaintiffs statistics fatally flawed since the evidence did not consider black employees who turned down promotions later offered to white employees. Id. Thus, the court refused to infer any intentional discrimination. 14

18 2. By contrast, courts give less weight to weaknesses in plaintiffs statistical evidence where other substantial evidence is present. In Stender v. Lucky Stores, Inc., 803 F. Supp. 259 (N.D. Cal. 1992), the defendant was unable to rebut the significant disparities between the promotions of men and women. Lucky attempted to prove that the statistical disparities between the promotions of men and women were explained by women s aversion to working night shifts and to differences in the job interests of men and women. The court rejected these arguments. Although noting that the statistical evidence was significant for some positions, and not for others, the court attributed the lack of significant disparities in those positions to the fact that women were blocked from upper management positions at the lower rungs of the promotional ladder. Id. at 333. In addition, the court relied on the direct and indirect evidence of discriminatory intent on the part of Lucky s managers, concluding that Lucky s promotion process was tainted by discretionary decisionmaking and the use of variable and subjective criteria. Accordingly, the court found that the evidence was sufficient to sustain plaintiffs burden of proving disparate treatment in Lucky s promotion practices as a whole. Id. at Even if evidence is presented establishing a statistical disparity, courts nonetheless that statistics are not irrefutable; they come infinite variety. Teamsters, 431 U.S. at 340. Thus, as one court explained, statistically significant results are by no means legally significant. Or, stated another way, statistical evidence which shows that chance is less likely to be the explanation for any observed disparity is not by itself sufficient to demonstrate that discrimination is the more likely explanation for an employer s conduct. Hence, courts rely on statistics with caution, ensuring that any inference of intentional discrimination does not arise from just a battle of statistics. Cota, 783 F. Supp. at 468 (quoting American Fed n of State, County and Mun. Employees, AFL-CIO v. Washington, 770 F.2d 1401, 1407 (9th Cir. 1985)). 4. With such caution in mind, before an inference of intentional discrimination can be drawn from statistical evidence, some courts have begun to require that the statistics eliminat[e] nondiscriminatory explanations for the disparate treatment by showing disparate treatment between comparable individuals. Rea v. Martin Marietta Corp., 29 F.3d 1450, 1456 (10th Cir. 1994) (finding plaintiffs attempts to establish pretext in ADEA action based on statistical evidence which did not account for performance evaluations or departmental rankings was not a comparison of similarly situated employees) (quoting Fallis v. Kerr-McGee Corp., 944 F.2d 743, 746 (10th Cir. 1991)). Stated another way, [w]hile statistical evidence may create an inference of discrimination, the evidence may be so flawed as to render it insufficient to raise a jury question. Furr v. Seagate Technology, Inc., 82 F.3d 980, 986 (10th Cir. 1996) (determining that plaintiffs statistics flawed since all employees were grouped together regardless of specialty or skill). Therefore, the usefulness of statistics in raising an inference of discrimination, like 15

19 all evidence, depends on all of the surrounding facts and circumstances. Teamsters, 431 U.S. at 340; Anderson, 26 F.3d at In order to rebut an inference of discrimination derived from statistical evidence, some courts simply allow the employer to introduce proof that the statistics are inaccurate or insignificant. Anderson, 26 F.3d at 1285 (quoting Bernard v. Gulf Oil Corp., 841 F.2d 547, 568 (5th Cir. 1988) (citations omitted)). However, other courts have stated that the defendant may not rebut an inference of discrimination by merely pointing to flaws in the plaintiffs statistics. Stender, 803 F. Supp. at 322 (quoting EEOC v. General Tel. Co. of Northwest, 885 F.2d 575, 581 (9th Cir. 1989)). Instead, these courts require one who claims the statistics are flawed due to a missing factor to establish that this factor can explain the disparities as a product of a legitimate, nondiscriminatory selection criterion. Stender, 803 F.Supp at 322 (quoting General Telephone, 885 F.2d at 580). 6. One potential means of attacking plaintiffs statistics is that plaintiffs expert has failed to properly define the relevant labor pool for comparison purposes: a. Webb v. Derwinski, 868 F. Supp (E.D. Mo 1994) (statistics showing that U.S. Department of Veterans Affairs hired relatively small number of blacks for accounting-related or fiscal jobs has little probative value inasmuch as underlying statistic, measured by the black general population, included persons who were not qualified for positions; relevant data would show number of qualified blacks hired for these jobs), aff d, 68 F.3d 479 (8th Cir. 1995). b. Because there is rarely any disagreement as to which jobs are at-issue, the dispute most always hinges on determining the relevant labor market. See Forehand, 89 F.3d at 1572 (in context of promotions, parties disagreed on the pool from which promotions were granted); O & G Spring and Wire, 38 F.3d at (finding that determination of relevant labor market is crucial to analyzing whether there is a statistically significant disparity); Anderson, 26 F.3d at 1286 (holding that to infer discrimination must compare relevant portion of employer s work force with the qualified population in the relevant market); Olson s Dairy Queens, 989 F.2d at 168 (stating that usefulness of statistics depends on the validity of the basic reference population in comparison to the employer s work force). c. Some courts state the proper measure of the relevant labor market should always be the pool of actual applicants. See Stender, 803 F. Supp. at 326 (citing Moore v. Hughes Helicopters, Inc., 708 F.2d 475, 482 (9th Cir. 1983)). Meanwhile, other courts only view the actual applicant pool as the basis for evaluating an employer s hiring practices. See Anderson, 26 F.3d at 1287; see also Forehand, 89 F.3d at 1574 (declining to adopt per se rule that applicant 16

20 flow data is the best measure); Olson s Dairy Queens, 989 F.2d at 168; Cook, 59 Fair Empl. Prac. Cas. at d. Courts have accepted other statistical benchmarks when the applicant flow figures are flawed, or otherwise unavailable. Anderson, 26 F.3d at 1287; Olson s Dairy Queen, 989 F. 2d at 169; see also Stender, 803 F. Supp. at 326 (stating that applicant flow to be used unless there is proof that the applicant pool is inaccurate). Situations in which applicant flow figures have been deemed inapplicable include where the employer failed to retain a complete applicant data set, or where it was determined that the employer s past treatment towards the relevant demographic group led those members to believe that submitting an application would be futile. See Anderson, 26 F.2d at 1288; Stender, 803 F. Supp. at 326; E.E.O.C. v. Joe s Stone Crab, Inc., 969 F. Supp. 727 (S.D. Fla. 1997); see also Winbush v. Iowa by Glenwood State Hosp., 66 F.3d 1471, 1481 (8th Cir. 1995). 7. Another common flaw in statistical analyses is the failure to consider nondiscriminatory factors that could account for any statistical discrepancy. For example: a. Holifield v. Reno, 115 F.3d 1555 (11th Cir. 1997) (statistical reports studying percentage of persons in selected management positions within Federal Bureau of Prisons who were minority-group members on nationwide basis were not relevant to discriminatory treatment claim of black doctor that he was discharged from federal correctional institution because of his race). b. Thus, if the employer considers, for example, an applicant s prior experience or interests and preferences as an important factor, those factors should be taken into statistical analysis. See, e.g., O & G Spring and Wire, 38 F.3d at (factoring into the statistical analysis the applicant s self-selection and prior experience); Stender, 803 F. Supp. at 332, 334 (taking into account the applicant s preferences). c. However, although a party s statistical evidence is most probative where it considers all relevant factors, the Supreme Court has made it clear that a party s failure to do so may not be fatal: [F]ailure to include variables will affect the analysis probativeness, not its admissibility. Importantly, it is clear that a regression analysis that includes less than all measurable variables may serve to prove a plaintiff s case. A plaintiff in a Title VII suit need not prove discrimination with scientific certainty; rather, his or her burden is to prove discrimination by preponderance of the evidence. 17

21 Bazemore v. Friday, 478 U.S. 385, 400 (1986) (citing Texas Dept. of Community Affairs v. Burdine, 450 U.S. 248, 252 (1981)). 8. In order for a defendant to rationalize statistical disparities, courts have required the employer s justifications to be consistent with their own recruiting and hiring practices. For example, in reversing the trial court in Olson s Dairy Queen, the Court of Appeals for the Fifth Circuit determined that the EEOC s statistical evidence was sufficient to establish a prima facie case of disparate treatment. Olson s Dairy Queens, 989 F.2d at 168. The statistics showed the extent to which the restaurant s actual hiring practices produced a different black to non-black employee mix than would be expected if the defendant s hiring practices had been entirely race-neutral. Id. at 166. The court rejected the employer s claim that any statistical disparities were the result of the proximity of the applicant s residence to the employer s restaurants and the racial makeup of the local school district. Although recognizing the restaurants were located in predominantly white neighborhoods, the court ruled that there was no proof that proximity was a factor in the defendant s hiring decisions. Id. at 169. Moreover, while acknowledging that defendant s employees were mostly from the local school area, the court found that the defendant s actual applicant pool primarily consisted of persons outside the area high school district. Id. 9. In O & G Spring and Wire, the EEOC alleged that for six years the defendant discriminated against blacks and persons over age 40 in recruitment and hiring. O & G Spring and Wire, 38 F.3d at 874. At trial, the EEOC presented evidence that: 1) blacks represented the largest group in the walk-in pool; 2) four black witnesses had been rejected for jobs at the company; and 3) statistical evidence showed the number of black applicants that should have been hired based on the relevant labor market. Id. at 874. Approximately two-thirds of the defendant s hires during the relevant period were walk-ins off the street. During that time, 99 people were hired for the low-skilled positions at-issue. Of these, only five were black and all of them were hired after the EEOC filed its discrimination claim. Moreover, available applications indicated that at least one-eighth of the applicants were black. To defend the low number of black applicants hired, the company gave one primary justification: only walk-in applications accepted during a two-week window period were hired because the type of workers from whom it received applications were very transient and it was a waste of time to work from stale applications. In addition, the company claimed that the blacks who did apply were interested in immediate work only, and when none was available, they did not want to fill out applications. Rejecting both of these arguments, the court found the EEOC s statistical evidence, corroborated by anecdotal evidence and the increased number of blacks hired, established a prima facie case of intentional discrimination. Id. at