On the Legal Front:
Lewis v. City of Chicago: What Is a Timely Claim of Disparate Impact?
Eric Dunleavy
DCI Consulting
Art Gutman
Florida Institute of Technology
On September 30, 2009, the Supreme Court announced that they will hear Lewis v. City of Chicago, a disparate impact case that started in 1997. In this case a test measuring a set of cognitive abilities and job knowledge was used to rank applicants for entry-level firefighter positions into different categories (i.e., well-qualified, qualified, and not qualified). Job offers were made to the well-qualified group of applicants first. A disparate impact claim was made by African-American test takers, asserting that the test was discriminatory. A district court initially ruled in favor of the plaintiffs in 2005, concluding that the test was not sufficiently job related.
However, that decision was overturned by a 7th circuit appeals ruling in June of 2008 on the grounds that the initial claim of discrimination made back in 1997 was not timely because it was made more than 300 days after the test results were made known to applicants. In the district court ruling, Judge Joan B. Gottschall concluded that the timely filing period began after employment decisions (i.e., actual hires) were made based on test scores, presumably under the rationale that this was the point in time when applicants understood whether or not they may have been discriminated against by the employment decision process. The Supreme Court has framed the legal question of interest as:
Where an employer adopts an employment practice that discriminates against African Americans in violation of Title VII’s disparate impact provision, must a plaintiff file an EEOC charge within 300 days after the announcement of the practice, or may a plaintiff file a charge within 300 days after the employer’s use of the discriminatory practice?
The timing of this case is interesting for a number of reasons. First, readers of this column are likely aware that the legal question of interest in Lewis is similar to the legal question considered by the Supreme Court in Ledbetter v. Goodyear Tire Co. (2007; see this column in the January 2008 TIP for a review of that ruling). The issue of “timeliness” was later considered directly by Congress, and the Ledbetter Fair Pay Act reversed the Supreme Court ruling (see this column in the July 2009 TIP for a review of the Ledbetter Act).
The Ledbetter ruling focused on a claim of intentional discrimination in pay under Title VII. The Ledbetter Act, on the other hand, may have intentionally been written using broader language, and it is reasonable to infer that Congress wanted to ensure that the issue of timeliness was a nonfactor in scenarios when potential victims of discrimination may be unsure as to whether they were actually discriminated against. Thus, one important dimension of Lewis is whether the Ledbetter Act could be applied to an adverse impact claim. Even if the answer is no,1 could a similar rationale regarding the likelihood that potential victims were unaware of being discriminated against be used to support the timeliness of the claim when inferences from test scores are made again and again over time?
1 Note that the original district court ruling in Lewis (2005) treated the claim as timely 4 years before the Ledbetter Act was passed in 2009. The appeals court ruling occurred in June of 2008, which was also before the Ledbetter Act was passed.
Lewis is also timely for another reason. In the October 2009 TIP, this column summarized the intricate Supreme Court ruling in Ricci v. Destefano (2009), where the Court ruled that New Haven discriminated against White and Hispanic applicants by throwing promotion exam results out in part because of potential adverse impact against African-American applicants. Shortly before the current article was due, Michael Briscoe, an African-American firefighter who took that promotion exam and applied to a lieutenant position in the New Haven Firefighter force and was rejected, filed a claim of disparate impact under Title VII.2 Of course, based on the Supreme Court ruling, New Haven already discriminated against 17 White employees and 1 Hispanic employee by throwing those test results out to avoid an adverse impact claim. Now, the city may have a traditional adverse impact case on its hands in addition to the already decided reverse-discrimination case.3
2 See Briscoe v. New Haven, D. Conn., No. 3:09cv1642, complaint filed 10/15/09.
3 Note that it is unclear whether this claim could be thrown out given that the promotions likely to be given to the White and Hispanic applicants were essentially mandated by the Supreme Court.
According to various media reports, Michael Briscoe scored highly on the oral portion of the exam and is challenging the scheme that weighted the oral component 40% and the written component 60%.4 Had the test been weighted 70% oral and 30% written, Briscoe asserts that he would have been promoted. Importantly, the issue of timely filing may play an important role in the future of this case. If the appeals court decision in Lewis stands, it may be unlikely that test scores made known to test takers in 2003 would allow for a “fresh” claim of discrimination in 2009. However, if the ruling is reversed and the timely filing period resets whenever test results are used to make a new round of employment decisions, the claim may be timely as soon as promotions in New Haven are made.
4 As described in the last Legal Front column, and in more detail by the I-O psychologists who submitted a brief to the Supreme Court, that 60–40 weighting scheme was arbitrary and decided upon in a collective bargaining agreement.
Lewis v. City of Chicago: The Facts of the Case
In July of 1995, 26,000 applicants participated in a selection process for entry-level firefighter jobs. One part of this process was a written test that measures a narrow set of cognitive abilities and job knowledge via multiple choice questions and written responses to video simulations. Based on overall test scores, applicants were categorized into three groupings: well-qualified (greater than 89 on the test), qualified (between 65 and 88 on the test), and not qualified (less than 65 on the test). These groupings were created to allow for “administrative convenience” in making hiring decisions and treated as rank-order bands. Job offers would first be made randomly to applicants in the highest well-qualified group. If the well-qualified group was exhausted, job offers would begin to be made randomly from the qualified group.
Applicants were made aware of their test scores and grouping status in January of 1996, about 6 months after they took the test. Applicants were told that those in the qualified group were unlikely to be hired but that they would remain on the eligibility list. News reports estimated that only about 600–700 hires would be made from the list, and given that there were about 1,800 well-qualified applicants, it may have been reasonable for qualified applicants to be pessimistic about receiving a job offer. However, actual hiring from the well-qualified group did not begin until May of 1996 and would continue over time.
This scenario may present some ambiguity regarding whether applicants categorized into the qualified group were in a position to know whether they may have been discriminated against. No matter how the test results were framed, White applicants were substantially more likely to be categorized as well qualified (making up 76% of the well-qualified group) compared to African-American applicants (making up only 12% of the well-qualified list). In fact, about 12% of White test takers passed into the well-qualified group, whereas about 2% of African-American test takers passed into the well-qualified group.
However, at the point where test takers knew their test scores and grouping status, no hiring decisions had been made. Job offers would eventually be made using random selection from the well-qualified list. Thus, understanding the practical impact of the test in terms of actual hiring decisions may have been difficult until a later date when hiring decisions were known. In a comment that was noted in both the district and appeals court rulings, Chicago Mayor Richard Daley expressed dissatisfaction with the racial breakdown of the well-qualified list. Certainly at that point African-American applicants may have been aware that, on the average, they were less likely to be categorized into the well-qualified group. However, it may have be unclear whether African-American applicants were substantially less likely to be hired over time, although that would likely be the case given subgroup representation differences in the well-qualified group.
In April of 1996, members of the African-American Firefighters League met with attorneys to consider whether they had a disparate impact case. The attorneys wanted to get more information about the test and potential defense strategies available to the city before making a firm decision about whether to make a claim. Eventually a claim of discrimination was made in March of 1997, more than 400 days after the test results were initially reported. Later in litigation, plaintiff lawyers acknowledged that they assumed the timeliness clock would reset each time employment decisions (i.e., a hire) were made based on test score inferences. At the district court level, the case would hinge on the merits of the disparate impact claim in the tradition of Griggs (1971) and Albemarle (1975) rulings, and the timeliness issue would be a tangent.
The District Court Ruling
As expected, the district court ruling in 2005 focused on the adequacy of the test. In fact, the timeliness of the claim received little attention. One sentence in the decision mentioned that the timeliness issue was briefed before the court in 2004 and that the city’s motion for summary judgment based on an untimely claim was denied. Thus, in essence, Judge Gottschall concluded that a fresh violation of Title VII harmed qualified applicants each time the city hired applicants from the well-qualified list.
The district court ruling read like the prototypical adverse impact case. The city conceded adverse impact against African Americans in using the test to classify applicants. White test takers scored about one standard deviation higher than African-American test takers on the test composite, and White applicants were more than five times as likely to be categorized as well-qualified as compared to African-American applicants. The city acknowledged the adverse impact and argued that the test measured some relevant cognitive aspects of the job. In addition, the city asserted that the cut score differentiating qualified from well qualified was “administratively convenient.”
A traditional content-oriented validation strategy was used to develop the test, and a job analysis identified important work behaviors and the knowledge, skills, and abilities necessary to perform them. No criterion research had been formally conducted. However, after the fact it was noted that (a) eventually applicants from the qualified group were hired after the well-qualified group had been exhausted, and (b) applicants from the well-qualified group performed no better on the job than did applicants from the qualified group. This finding cast doubt on the appropriateness of the cut score that differentiated well-qualified from qualified. The city had assumed that higher scores equated to higher performance without empirically evaluating this notion. In fact, cut-score/banding recommendations by the test developer were made based on the standard error of measurement, and were ignored by the city.
In addition, Judge Gottschall concluded that it was “unclear what the video simulation portion of the test was assessing,” and that one ability measured by the test (note taking) was not rated as important based on the job analysis. Thus, Judge Gottschall concluded that the test was flawed and that the impact was not justified by job relatedness. One other point to note is that, according to Judge Gottschall, the plaintiffs likely would have met the burden of reasonable alternatives even if the city had met its job-relatedness burden. Judge Gottschall suggested that random selection from a combination of the well-qualified and qualified groups would have been equally valid and resulted in less adverse impact. Although this approach as a reasonable alternative is somewhat questionable (and inconsistent with testimony on the standard error of measurement of the test), the take-home message remains that the test was fatally flawed from a legal defensibility perspective, and Judge Gottschall ruled in favor of the plaintiffs.
The Appeals Court Ruling
The city appealed the ruling, but instead of challenging conclusions about the adequacy of the test, the appeal challenged the timeliness of the claim. In the 7th U.S. Circuit of Appeals, Judge Posner,5 writing on behalf of Judges Bauer and Easterbrook, concluded that the timely filing period started at the point where scores were made known to test takers and not when decisions were made based on inferences from those scores. Judge Posner called the assumption of a continuing violation reset each time hiring decisions were made a fatal mistake made by plaintiff attorneys. The appeals court estimated that the initial charge of discrimination was made to EEOC 420 days after the individual results were sent out by the city, and 417 to 419 days after test takers received their results.
5 This name may be familiar to many readers. Judge Posner was interviewed by Frank Landy for his book Employment Discrimination Litigation (2005).
The appeals court did not view the Ledbetter ruling as relevant to this case. However, there was no meaningful distinction made between disparate treatment and disparate impact in the ruling. Instead, the appeals court considered the Lewis scenario to be similar to Delaware State College v. Ricks (1980), which was a disparate treatment case where the Supreme Court ruled that timely filing period began when tenure was denied to a college professor not when the plaintiff was terminated at the end of a 1-year contract negotiated after the tenure rejection. In both cases, the later potential act of discrimination (in Ricks, the termination after the 1-year contract; in Lewis, not being hired each time the well-qualified list was used to make hires) was ruled an automatic consequence of earlier discrimination and not a fresh act of discrimination.
The appeals court noted that the mayor’s comments about the lack of diversity in test results could be treated almost as a proxy for the start of the timely filing period because at that point diligent applicants would have known that they may be victims of discrimination. The Lewis context was also compared similarly to Lorance v. AT&T Technologies (1989), where a contractual modification to a seniority system eventually produced demotions years later. A claim of intentional discrimination was made after the demotions were made and was ruled untimely because it was the contractual modification years earlier that was the discriminatory act.
The appeals court suggested that the plaintiffs mistakenly relied on Beavers v. American Cast Iron Pipe (1992) to support their timeliness claim. In this case the Supreme Court considered a claim because it related to an insurance policy that eventually led to the rejection of dependent coverage denial. In this scenario the discriminatory policy was the only cause of the denial, and there was no intervening neutral act, as the judges perceived in Lewis. Judge Posner concluded that this treatment versus impact issue had no relevance to the date of the timely filing period.
In addition, the appeals court suggested that if reasonable diligence does not allow plaintiffs to know if they have been injured by discrimination, then the timely filing period is functioning as Title VII intended. The appeals court also concluded that a continuing violation theory does not apply because automatic consequences are not the same as fresh acts of discrimination. Further, the court suggested that a continued aggregation theory prevalent in recent retaliation and sexual harassment claims does not apply to Lewis either because automatic consequences are not the same as multiple instances of trivial deterring behavior that later aggregate to actionable discrimination. Judge Posner also spent considerable space focusing on lawyer error. Specifically, the opinion reiterated that plaintiff lawyers should not have assumed continuing violation and that complaints should be functionally independent from potential employer defenses.
Other Relevant Context
It shouldn’t come as a surprise that various lower courts have considered similar timeliness issues before. However, lower courts have come to different conclusions. For example, the 3rd Circuit (Bronze Shields Inc. v. New Jersey, 1981) and 6th Circuit (Cox v. City of Memphis, 2000) have ruled similar to the 7th Circuit decision in Lewis. However, the 2nd Circuit (Guardians v. Civil Service, 1980), 5th Circuit (Gonzalez v. Firestone Tire & Rubber Co., 1980), and 9th Circuit (Bouman v. Block, 1991) have essentially agreed with a continuing violation (or new and overt violation) argument in this context; the charging period starts anew when new employment decisions are made based on inferences from discriminatory procedures. The 11th Circuit (Beavers, 1992) and the D.C. Circuit (Anderson v. Zubieta, 1999) have ruled similarly to the 2nd, 5th, and 9th in slightly different contexts.
The solicitor general was asked to submit a brief on Lewis and was joined by the EEOC. Given the circuit disagreement on this issue, it isn’t surprising that the solicitor general concluded that the timeliness of a disparate impact claim warrants Supreme Court review. The brief differentiates treatment from impact and suggests that the continuing violation theory is actually unnecessary because it assumes one discriminatory act that affects other employment decisions over time. That is to say, in the disparate impact context, victims can clearly pinpoint each time they were not hired as a discriminatory act, and victims could feel impact anytime they were not hired when someone else was hired. The brief also urges that the language of Title VII be given precedent over a set of disparate treatment cases (e.g., Delaware and Lorance). Specifically, the brief emphasizes language including discriminatory uses of a test and employer acts based upon results of a test to support the notion that the timely filing period restarts each time a round of decision making stems from inferences based on a discriminatory test.
It is interesting to take both sides of this issue to extreme ends. The solicitor general’s brief alludes to the notion that, in theory, an employer could discriminate for an unlimited period of time if the timely filing period starts at the announcement of test results. Taken to an extreme end, an employer could announce test results, wait 300 days, and select from a higher scoring group with potential insulation regardless of the adequacy of the test. On the other hand, the city pointed out that, if the timely filing period starts anew each time a selection decision is made, a long period of equal employment opportunity could be moot if a later round of hiring results in adverse impact for a shorter time period many years later.
Does the Ledbetter Act Apply to Lewis?
Recall that one major theme of the Ledbetter Act was the notion that compensation discrimination was inherently different from discrimination via other employment decisions like hiring and promotion. The difference is a function of the fact that, in many situations, employees may not know that they are being discriminated against in pay because coworker pay information is often not publically available. However, hiring and promotion decisions are generally discernible, and thus known to applicants and employees. For this reason, many in the EEO community have predicted that the Ledbetter Act will be narrowly applied to few cases where the employment practice of interest is not some form of pay. However, the scenario in Lewis may be an exception to the notion that employees are immediately aware of discrimination in employment decisions like hiring. In this case test takers were aware of their test score and their band but not aware of specific hiring decisions based on the test.
So will the Ledbetter Act play a role in Lewis? The Act differentiated pay discrimination from other more overt acts of discrimination and endorses, in effect, a “continuing violation” theory of discrimination that expands the timely filing period. According to the Ledbetter Act, discrimination may happen at three different times depending on specific context, and those times are when:
- A discriminatory compensation decision or other practice is adopted;
- An individual becomes subject to a discriminatory compensation decision or other practice;
- An individual is affected by application of a discriminatory compensation decision or other practice, including each time wages, benefits, or other compensation is paid, resulting in whole or in part from such a decision or other practice.
Additionally:
- The purpose of the act is to clarify that a discriminatory compensation decision or other practice that is unlawful under such Acts occurs each time compensation is paid pursuant to the discriminatory compensation decision or other practice, and for other purposes.
- The Supreme Court ruling in Ledbetter was flawed in that the limitation imposed by the Court on the filing of discriminatory compensation claims ignores the reality of wage discrimination and is at odds with the robust application of the civil rights laws that Congress intended.
Note that the phrase or other practice has been and continues to be issue of legal contention. Some in the EEO arena are worried that employment decisions other than compensation could be covered by this phrase that were not intended by Congress. Recent case law from district courts has exemplified what does and does not exemplify the “other practices” phrase from the Ledbetter Act. For example, in Gentry v. Jackson State University (2009), the denial of tenure to a university professor position, later resulting in failure to receive a salary increase, is a compensation-related decision under the Ledbetter Act. In addition, in Mikula v. Allegheny County of Pennsylvania (2009), an employer’s refusal to respond to an employee’s request for a pay raise was deemed a covered compensation practice. Likewise, in Gilmore v. Macy’s Retain Holdings (2009), placement into a particular sales department was deemed a covered practice under Ledbetter. However, in Leach v. Baylor College of Medicine (2009), the court refused to apply the Ledbetter Act and a reassignment of job duties was not a covered practice. In Rowland v. Certain Teed Corp. (2009), failure to promote was not considered a covered decision. As the above rulings show, there is substantial disagreement across district in what is considered an “other practice” under the Ledbetter Act.
We think that there could be three relevant questions considered by the Court in Lewis regarding the Ledbetter Act. Specifically,
- Does the Ledbetter Act generalize from a disparate treatment scenario to a disparate impact scenario?
- On a related note, does a continuing violation theory apply to impact cases? Or similarly, does the continued inference of a discriminatory test score over time imply a new and overt discriminatory act every time test results are used to make decisions?
- Does the Ledbetter Act generalize from compensation as the employment decision of interest to employment decisions like hiring via the phrase “other practices” in some situations?
Based on the majority ruling in Ledbetter and the argument put forth by the dissenting justices (i.e., that pay was different from other employment decisions), it may be unlikely that the Supreme Court treats both compensation and hiring decisions as obvious employment decisions. However, given the time lag between test administration and the long term hiring based on test results in Lewis, it may not be surprising to see the Supreme Court justices that dissented in the Ledbetter ruling to conclude that potential victims were not in a reasonable position to know that they had been discriminated against in Lewis. This may be an informal application of Ledbetter rationale to the disparate impact scenario, as opposed to an actual application of the Ledbetter Act. One last point to consider is that many of the precedential cases considered by the district and circuit courts in Lewis were also considered in Ledbetter, so it will be interesting to see whether and how the Court differentiates the two scenarios.
Implications for I-O Psychologists
Some of the features in Lewis may look familiar to many I-O psychologists that develop selection procedures. For this reason the Lewis ruling may have important implications for I-Os, albeit in a technical manner that shouldn’t change I-O practice. In situations where a large number of hires will be made over an extended period of time, it is common to create eligibility lists in both the public and private sector, particularly for entry-level positions with a large number of applicants. This allows for a continuous sample of applicants to be considered over a long period of time and for potential hires to be readily available as jobs become open.
This scenario is particularly attractive given economic context. For example, if economic conditions allow for substantial hiring, an eligibility list may allow the organization to hire as necessary. If economic conditions do not allow hiring at all (as was the case for many organizations recently), the eligibility list may exist in stasis until hiring can occur. In addition, the organization may be able to minimize the cost of data collection/selection procedure implementation by assessing candidates at one period of time instead of multiple times.
Note that this scenario begs the question about how long eligibility lists should exist. Recent research on the currency of job analysis data may be useful in answering this question. For example, if evidence suggests that the job has not changed in recent years, then perhaps job-related scores can be maintained in an eligibility list for long periods of time. Refer to Bobko, Roth, and Buster (2008) for a legally defensible process intended to assess the “up-to-dateness” of job analysis data. In addition, whether an applicant’s score on job-related constructs becomes stale or outdated over time (regardless of changes to the job) is another issue to consider.
Modeling the adverse impact of this type of eligibility system may be challenging. It is often valuable to consider the adverse impact potential (in the form of d statistics before pass/fail decisions are made) of specific construct measures. Given that test takers may challenge a specific step in the selection process, it is also often useful to assess the actual adverse impact of a step in the process, and classifying applicants into bands can function as a step in the process. In this context the cut-score/banding point is of critical importance, and the I-O psychologist may want to consider multiple cut scores using empirically based criteria (e.g., the standard error of measurement). However, the adverse impact of actual employment decisions over time based on the selection procedure that produced the eligibility list is more complex to model. For example, results may change over time such that adverse impact may be deemed meaningful in some slice of the data (e.g., after 20 selections) but not before then.
Finally, it is important to note that the timeliness of an impact claim may be a moot point if selection procedures are sufficiently job related and reasonable alternatives were considered. This cannot be overemphasized: When organizations ask how to minimize legal risk, one best practice is to spend the time and effort necessary to create reliable and job-related selection procedures up front that also attempt to minimize adverse impact. In addition, as we have seen in the vast majority of recent adverse impact enforcement, implementation characteristics such as subtest weighting, cut scores, and banding strategies should be empirically and not arbitrarily based. It will be interesting to see whether Lewis becomes the intersection of divergent trends in recent EEO litigation. The Supreme Court is expected to hear oral argument early in 2010.
References
Bobko, P., Roth, P. L., & Buster, M. A. (2008). A systematic approach for assessing the currency (“up-to-dateness”) of job analytic information. Public Personnel Management, 37, 261–277.
Landy, F. J. (2005). A judge’s view: Interviews with federal judges about expert witness testimony. In F. J. Landy (Ed.), Employment discrimination litigation: Behavioral, quantitative, and legal perspectives. San Francisco: Jossey-Bass.
Cases Cited
Albemarle Paper Co. v. Moody, 422 U.S. 405 (1975).
Anderson v. Zubieta, 180 F.3d 329, 335-36 (D.C. Cir. 1999).
Beavers v. American Cast Iron Pipe Co., 975 F.2d 792, 796-800 (11th Cir. 1992).
Bouman v. Block, 940 F.2d 1211, 1221 (9th Cir. 1991).
Briscoe v. New Haven, D. Conn., No. 3:09cv1642, complaint filed 10/15/09.
Bronze Shields, Inc. v. N.J. Dep’t of Civil Serv., 667 F.2d 1074 (3d Cir. 1981).
Cox v. City of Memphis, 230 F.3d 199, 204-05 (6th Cir. 2000).
Delaware State College v. Ricks, 449 U.S. 250 (1980).
Gentry v. Jackson State Univ., 2009 U. S. Dist. LEXIS 35271 (S. D. Miss. April 17, 2009).
Gilmore v. Macy’s Retail Holdings, 2009 U. S. Dist. LEXIS 7894 (D. N.J. February 4, 2009).
Gonzalez v. Firestone Tire & Rubber Co., 610 F.2d 241 (5th Cir. 1980).
Griggs v. Duke Power Co. 401 US 424 (1971).
Guardians Association v. Civil Service Commission, 633 F.2d 232 (2d Cir. 1980).
Leach v. Baylor College of Medicine, 2009 U. S. Dist. Lexis 11845 (S. D. Tex. February 17, 2009).
Ledbetter v. Goodyear Tire 50 U.S. 618 (2007).
Lewis v. City of Chicago, No. 98 C 5596 (N.D. Ill. Mar. 22, 2005).
Lewis v. City of Chicago, 528 F. 3d 488 (2008).
Lorance v. AT & T Technologies, Inc., 490 U.S. 900, 910-11 (1989).
Mikula v. Allegheny County of Pennsylvania, 2009 U. S. App. Lexis 20217 (3rd Cir. September 10, 2009).
Ricci v. Destefano 129 S. Ct. 2658 (2009).
Rowland v. Certain Teed Corp., 2009 U. S. Dist. LEXIS 43706 (E.D. Pa. May 21, 2009).