Practice Network: Class-Action Lawsuits in the Employment Discrimination
Context: What I-O Psychologists Should Know
Michael M. Harris
University of Missouri-St. Louis
Peopleclick Research Institute
Class-action lawsuits generate a great deal of attention in the media. If you go to the Internet and type in
www.bigclassaction.com, you will not only find a plethora of employment-related class-action lawsuits listed, but you will also be able to submit information about your own possible class-action lawsuit, which will be evaluated at no charge by attorneys! Recent class-action lawsuits in the employment discrimination area (e.g., Texaco; Coca-Cola) have resulted in settlements of nearly $200 million. A group of females claiming sex discrimination by Wal-Mart could end up being part of the largest class-action lawsuit in history with approximately 1.5 million plaintiffs and potential liabilities in the hundreds of millions of dollars. The purpose of this column is to provide a basic overview of class-action employment discrimination lawsuits for I-O psychologists so that they better understand the underlying requirements in such cases (readers interested in a basic overview written from a lawyers perspective should examine Arbery (2003). Towards that end, we first summarize the basic criteria for class certification and identify some basic tactics that may be used by I-O psychologists to establish or defend against these criteria. Next, we summarize some typical class-certification cases for the purpose of examining how the courts have analyzed certain key issues of relevance to I-O psychologists. Finally, we conclude with major implications for I-O psychologists.
Criteria for Certifying a Class
The pivotal decision in most employment discrimination class-action lawsuits involves the certification of the purported class. When a single person files a discrimination claim, a single-plaintiff case, the charges may be resolved fairly easily and under the public radar. A class-action lawsuit, in contrast, is rarely resolved quickly. Class-action lawsuits require substantial time and resources. Thus, both plaintiffs and defendants allocate substantial time and resources to the class certification phase of a lawsuit. I-O psychologists are well suited to assist either plaintiffs or defendants at this stage.
Federal Rule of Civil Procedure 23 provides the criteria for class certification. In particular, Rule 23(a) stipulates the following four conditions for certifying a class:
1. Numerosity: The class must be so numerous that trying each individual case would be impractical;
2. Commonality: Common questions and facts of law must apply to all putative class members;
3. Typicality: The claims of the class members must be typical of those of the putative class; and
4. Adequacy of representation: The individuals filing the claim must be able to adequately represent the class.
Our focus in reviewing these four criteria is on areas where I-O psychologists may be of help. There are, of course, certain legal aspects that are more the purview of attorneys, and we do not comment on those aspects.
An employers Human Resource Information System (HRIS) may contain information needed to address the conditions for class certification. For criteria 1 and 4, for example, descriptive statistics may help to either support or refute numerosity and adequacy of representation. For criteria 2 and 3, plaintiffs and defendants often rely on statistical analyses to go beyond descriptive methods and conduct statistical studies to address typicality and commonality.
Numerosity may be addressed by identifying the number of individuals in an HRIS system with characteristics similar to those in the putative class. For example, if 10 females file a claim of employment discrimination against their employer and seek to represent a class of all females in clerical positions between 1995 and 2000, data in the HRIS system may provide counts of females in clerical positions at any time between 1995 and 2000. Other relevant information that may bolster a claim that the joinder of the individual cases would be impractical includes the geographic work location of putative class members.
In regard to adequacy, members of the purported class should share common characteristics with the class that they claim to represent. For example, for a lawsuit that includes claims of hiring discrimination, plaintiffs should include applicants denied employment by the defendant. For a promotion discrimination case, plaintiffs should include employees eligible for promotion. For a gender discrimination case, plaintiffs should be of the same gender. Again, HRIS data may be used to support claims regarding similarity of jobs, decisions, race, or gender and other aspects of the claim. In the above example, if putative class members are females employed in clerical positions between 1995 and 2000 and eligible for promotion during that time period, the adequacy of representation criteria may be met.
Numerosity and adequacy are relatively easy to show depending on the integrity and comprehensiveness of the HRIS database and the assistance of someone with knowledge of the system. However, the real challenge for plaintiffs pertains to evidence of typicality and commonality. Both plaintiffs and defendants may rely on statistical evidence to show or refute evidence of a pervasive pattern of discrimination throughout the defendants company.
To refute typicality and commonality, a defendant often claims that employment decisions are made on an individual, case-by-case basis. That is, the defendant may argue that since there is not a common decision-making practice that applies to all putative class members, there is no class. With backgrounds in both human resource practices and statistics, I-O psychologists are well suited to assist legal counsel for plaintiffs or defendants in establishing or refuting commonality and typicality. Plaintiffs want to present evidence of a pervasive pattern of discriminatory decisions throughout the company. Defendants will want to present evidence that contradicts such claims.
A variety of statistical techniques may be used to either support or refute either sides claims. One less widely known technique by I-O psychologists is cohort analysis, which might be used to establish commonality and typicality in the class-certification stage of an employment discrimination lawsuit. In a cohort analysis, one groups employees by similar job characteristics and examines decisions along racial or gender lines. A cohort analysis for class certification includes the following steps:
1. Identify the cohort variables required to examine commonality and
typicality.1 This may include, at a minimum, a variable representing time, such as a year if you have data available over a few years, a variable representing decision lines, such as manager or department or division, and a variable representing job duties which may be job title, job family, or job group.
2. Identify the cohort date, such as December 31.
3. Identify the appropriate statistical test such as a hypergeometric, binomial, or rank sum.
4. Examine decisions along racial or gender lines.
5. Identify whether any disparities reach the level accepted by courts as evidence of discrimination, usually 2 standard
1 As discussed below, there are mixed opinions about whether other job factors, such as education or experience related to the merits of the case need to be included at this stage.
2 In addition, industrial psychologists can calculate the exact p-values in order to make statements regarding the statistical significance of the disparities.
Using the example from above, lets say that we have HRIS data available for all employees in our company being sued for discriminating against females in promotion. We determine that we need to group employees by gender, job title, and department at the end of the year for each year 1995 to 2000. We identify that 1,000 out of 2,000 (i.e., 50%) of employees in Department A with the job title clerk are female on December 31, 1995. Next, we identify that the company promoted 200 employees from this cohort in the next year. If the company made these decisions without regard to gender, the company would have promoted approximately 100 or 50% females. If the company only promoted 40, the difference or disparity between the actual number of females promoted and the expected number of females promoted is 60. This shortfall can be expressed as a number of standard deviations. Courts generally accept 2 standard deviations as the threshold for evidence of discrimination.
In order to examine patterns throughout the company, we can aggregate the results across cohorts to examine disparities at the level of year, job title, department, or other cohort variable. In the above example, the first question is whether the shortfall of 60 equates to a number of standard deviations greater than 2. Lets say that it does. The next question is whether the shortfall can be attributed to a single year, department, or job. Shortfalls (i.e., fewer females promoted than expected given their representation in the cohort) across many departments, years, and jobs provides evidence of a pervasive pattern of discrimination throughout the company. If, however, the shortfall can be attributed to a single year, department, or job, the defense may argue against a pervasive pattern and against class certification.
Sample Class-Certification Cases
In this section, we review some recent cases in which class-certification issues arose starting with a brief comment on early key class-certification issues and following up with a description of two cases focusing on selection issues and two cases focusing on compensation issues. In these examples, you will see the importance of presenting both evidence that the employment practices under consideration are centralized or decentralized and that the application of these practices generally discriminates or generally doesnt discriminate against the putative class.
General Telephone v. Falcon (No. 81-574) represents an early class-certification employment discrimination case, which ended up in the Supreme Court and was ruled on in 1982. At issue in that case was whether a Mexican American employee claiming race discrimination in promotions could represent a class of Mexican American job applicants regarding hiring discrimination. In overturning a lower court decision to certify the class, the Supreme Court noted that [s]ignificant proof that an employer operated under a general policy of discrimination conceivably could justify a class of both applicants and employees if the discrimination manifested itself in hiring and promotion practices in the same general fashion, such as through
entirely subjective decisionmaking processes (italics added for emphasis). As will be mentioned in cases that follow next, this quote has led to much subsequent discussion regarding subjective versus objective decision making for employment discrimination.
Class-Certification Cases: Selection and Placement Issues. In the two cases that follow, we focus on selection and placement issues (the reader should note that both lawsuits involved issues besides selection and placement, such as compensation). We begin first with a case in which the judge sided with the plaintiffs, followed by a case where the judge sided with the defendant.
In Melodee Shores et al. v. Publix Super Markets (95-1162-CIV-T-25E), plaintiffs sued the grocery store chain for sex discrimination and sought class certification, arguing that gender stereotypes were pervasive throughout the organization. The judge characterized Publix as using a centralized policy of decentralized decision making in which certain practices, such as manuals and handbooks governing promotional opportunities, formed the basis for its centralized practices. In support of their argument that these centralized practices were the cause of sex discrimination, the plaintiffs argued that Publixs requirement that anyone wishing to be promoted to store manager had to first work as a stocker created a barrier because women were either discouraged from working as stockers or were refused such positions. Furthermore, because only employees who served as store managers or department managers were able to move into managerial positions, plaintiffs claimed that there was a barrier that prevented them from reaching higher-level positions. The plaintiffs also argued that store managers exercised considerable subjectivity in their HR decisions (e.g., in initial hiring). Plaintiffs argued that the absence of written guidelines or training in making those decisions made them susceptible to bias against women.
In defense, Publix argued that self-selection, (i.e., women preferring traditionally female jobs) rather than store manager bias was responsible for women favoring stereotypically female jobs. Furthermore, Publix argued that mere subjectivity in decision making was not necessarily indicative of discrimination.
Citing the Supreme Courts decision from Falcon, the judge ruled that in the class-certification stage, subjectivity of the decision-making process could be considered a factor in showing commonality, one of the four conditions for class certification. The judge also pointed to the lack of formal job postings, and the use of a system where managers had considerable discretion in terms of whom to select, as indicative of discrimination.
In terms of statistical analyses, the plaintiffs offered descriptive information indicating that women tended to be assigned to lower-level jobs in disproportionate numbers. For example, they showed that the vast majority of newly hired men, but only 12% of the newly hired women, worked in front service positions. Almost no women worked as grocery or produce clerks. The company did not contest the numbers; instead, the company argued that had gender differences in vocational interests and qualifications been taken into account these job placement differences would be explained away and offered an experts report to support those arguments. The judge summarized these positions as a battle of experts, concluding that it was inappropriate for the Court to determine the ultimate correctness of either parties contentions in the
context of class certification (emphasis added), and that the plaintiffs statistics were sufficient for the class-certification claim. This case provides an example of the distinction often made between evidence related to the merits of a class-action discrimination case and evidence related to whether a group of plaintiffs meet the conditions for class certification.
The judge ruled in favor of the defendant in the class-certification part of Rhodes v. Cracker Barrel Old Country Store (4:99-CV-217-HLM; for those with patience, this decision could serve as a case study for a graduate class on selection or employment discrimination, given its detailed discussion). Although most of the details are well beyond the scope of this article, the judge reviewed the HR system used at Cracker Barrel in great detail, including the hiring system, the training procedures, and the promotion process used throughout the company. Cracker Barrel consists of over 450 stores located in 41 states. Overall, the company employs about 50,000 workers. Very briefly, Cracker Barrel provided supervisors with staffing guidebooks, which included interviewing rules, sample questions, information regarding question legality, and related information. Although the details changed somewhat from year to year, supervisors typically received a one- day training program to cover the staffing guidebooks. Of particular interest for the present article, however, the judge concluded that while the HR policies and practices were centrally created by the company, the company did not apply the policies and practices in a centralized fashion. Specifically, the judge asserted that these policies and practices are applied by hundreds or perhaps thousands of relatively autonomous decision makers scattered over 450 stores in 41 states. He added that the fact that these policies and practices explicitly prohibit discrimination indicates that it is only individualized practices of the many autonomous decision makers that could be at issue.
Both the plaintiffs and the defendants experts presented a plethora of statistical analyses. The defendants experts critiqued many of the analyses presented by the plaintiffs experts. One of the major criticisms offered by the defendants experts is that the plaintiffs expert frequently failed to control for store; indeed, stores often varied in their pass rates on many of these variables. Among the results was the finding that when the defendants controlled for store, race differences in the pass rates on various tests tended to disappear. Thus, a store-level examination of the data argued against typicality and commonality.
Class-Certification Cases: Compensation Issues. In the next two cases, we focus on compensation issues. As before, we begin first with a case in which the judge sided with the plaintiffs, followed by a case where the judge sided with the defendant.
In a recent case, Warren et al. v. Xerox (01-CV-2909 [JG]), decided in January, 2004, class certification was again raised. The major charge in this case was that Xerox systemically assigned African Americans to inferior sales territories and refused to promote or transfer them to better territories, despite their performance. The plaintiffs were employed in the U.S. Customer Operations of Xerox, which was divided into several dozen Customer Business Units or CBUs. The CBUs functioned as independent organizations, with the responsibility for meeting centrally determined objectives. In terms of subjective decision making, the plaintiffs argued that the process was entirely subjective and therefore susceptible to discrimination. Xerox argued in response that its compensation decisions were not completely subjective, since there are various objective factors that were used. The judge ruled that the existence of some objective factors does not eliminate the possibility that the process is completely subjective, such as when those so-called objective variables may have been inappropriately applied. Thus, as noted by the judge, the fact that company-wide practices may be implemented differently in local sales operations does not negate the finding of commonality where, as here, the policy or practice was applied to the entire class.
The statistical evidence offered by the plaintiffs consisted of analyses by an expert showing that on an aggregate, company-wide basis, even controlling for job grade, sales experience, and tenure, African Americans earned less than Whites in the years covered by the lawsuit. In defense, Xerox argued that compensation was determined by four major factors, including base salary, sales territory, quota, and revenue produced by the salesperson. Moreover, the specific base salary was, according to the defendant, set by the local sales managers within centrally determined parameters and with instructions and guidance from Xeroxs Human Resources Department. When analyzed separately for each CBU, few of them revealed race differences. Thus, Xerox argued that there was no evidence of widespread pay discrimination. In choosing between the rather different conclusions of the plaintiffs and defendants experts, the judge took a similar stance to the one taken in
Melodee Shores et al. v. Publix Super Markets and asserted that in the class-certification stage the plaintiffs are not required to prove that they would prevail on the merits. Based on these considerations, the judge ruled in favor of the plaintiffs.
By way of comparison, in a recent decision (Moore et al. v. Boeing, 2004 U.S. Dist. LEXIS 5959), the judge ruled against class certification for the plaintiffs. Very briefly, the plaintiffs sought to represent female employees in the St. Louis area and alleged class discrimination in terms of pay. The number of business units in the St. Louis area varied, depending on the year considered (e.g., there were four units until mid-2002). Each unit had its own management structure and HR department.
The plaintiffs claimed that Boeing headquarters provided guidelines on pay, but these guidelines provided only vague considerations for the managers. The managers had considerable latitude on how to make those decisions. As such, the plaintiffs argued that these guidelines applied to all employees, supporting commonality. Boeing asserted, in defense, that there was not a specific company-wide policy or practice that could be pointed to other than excessive subjectivity. Siding with the defendant, the judge noted that excessive subjectivity might be a criticism of a practice but is not a policy or practice per se.
In reviewing the statistical evidence for pay discrimination, the judge noted that when aggregated within a unit, while the data
sometimes supported the argument that women were paid less, the data frequently did not show women were disadvantaged when separated by job groups. For example, for the year 2000, the plaintiffs expert studied 14 job groups and found that women were paid less than men in 11 of them. However, the differences were statistically significant in only three of the groups, and in one case, women appeared to be paid significantly
more than men. The judge concluded that the data does [sic] not show that there is a company-wide policy of discrimination. The judge therefore denied the motion for class certification.
In sum, there appears to be some variation in court opinions regarding the key issues in a class-certification employment discrimination case. These differences may be a function of different jurisdictions and different judges, so caution is needed in drawing firm conclusions from these cases to other situations (see Roehling, 1993, for more information on the dangers of drawing conclusions from prior court outcomes). Nevertheless, the issues raised throughout these cases provide insights into some of the issues that I-O psychologists might be called upon to apply their expertise.
Implications for I-O Psychologists
Given the risk involved, we think that it is important for I-O psychologists to remain knowledgeable and current with developments in the area of class- certification employment discrimination cases. It is our belief that the increasing use of the Internet may lead to the creation of highly centralized HR practices and may therefore expose companies to greater risk of class certification. In this light, we offer the following recommendations:
1. Be aware that developing and implementing centralized hiring, promotion, and pay practices could be viewed by the courts as evidence of commonality and typicality.
2. From a legal standpoint, there may be a delicate balance between practices that are too subjective and those that are too objective. Straddling the fence between overly subjective and overly objective practices may,
from a legal standpoint, be optimal.
3. Use of statistics to monitor potential disparate impact appears important. In addition, where disparate impact is found, it is important to examine whether the impact can be isolated to a particular part of the company or whether there is evidence of a pervasive pattern of discrimination. Companies should also provide appropriate oversight to ensure that disparate treatment is not occurring.
4. Continuous education of decision makers is important to ensure that they understand legal and professional guidelines for making HR decisions, including choosing proper tests for selection and promotion.
5. Regular auditing of an organizations hiring, promotion, pay, and other I-O systems and processes is needed to ensure that these practices meet professional standards.
In conclusion, at the time of writing this column, there is a pending decision regarding class certification in a major sex discrimination case
(Dukes v. Wal-Mart Stores, Inc.; see www.walmartclass.com). We believe that the outcome of this case is likely to affect future class-certification cases for years to come. In any event, class-action lawsuits in the employment discrimination area are likely to remain on the scene for years to come. We urge I-O psychologists to familiarize themselves with the issues they present.
As always, we would like to hear your comments, reactions, and experiences. Please contact either Michael Harris
(firstname.lastname@example.org) and/or Lisa Harpe (Lisa.Harpe@PeopleClick.com).
Arbery, W. C. (January 13, 2003). The threat of employment class-actions hovers.
The National Law Journal. Available at: http://www.hunton.com/pdfs/article/threat_employment_class_actions.pdf.
Roehling, M. V. (1993). Extracting policy from judicial opinions: The dangers of policy capturing in a field setting.
Personnel Psychology, 46, 477502.
July 2004 Table
of Contents | TIP Home
| SIOP Home