On the Legal Front: Understanding Grant v. Metro: Wards Cove Reloaded?
Eric M. Dunleavy
DCI Consulting Group
Florida Institute of Technology
One of the more complicated legal issues related to personnel selection concerns the difference between pattern or practice and disparate impact theories. Both theories involve class-wide allegations of discrimination but differ with regard to employer intent and the legal scenarios that follow. A recent 6th Circuit ruling considered this issue, and we think that this case has the potential to escalate to the Supreme Court. The case is Grant v. Metro, and in its most recent form it was decided on August 26, 2011 by a divided three-judge panel of the 6th Circuit Court [2011 U.S. App. LEXIS 18054]. It is a class action suit alleging both pattern or practice and disparate impact based on stock statistics that compared workforce representation at various levels.
Eric continually (half) jokes that Art is the only person in the country who understood the Supreme Court ruling in Wards Cove v. Antonio (1989) other than the justices involved. Art has long maintained that Wards Cove should have been a pattern or practice case, not an adverse impact case (e.g., Gutman, 2005; Gutman, Koppes & Vodanovich, 2010 Ch.2). What the 5–4 Supreme Court ruling did in Wards Cove was apply pattern or practice rules to adverse impact, and most agree that this was a key factor for CRA-91. Sure enough, Art has been waiting for this type of case (with facts similar to Wards Cove) since the Civil Rights Act of 1991 (CRA-91) was enacted.
The story begins with the Supreme Court’s ruling in Watson v. Fort Worth Bank (1988) [487 US 977] and Wards Cove v. Antonio (1989) [490 US 642]. These two cases are not connected so much by their facts but, rather, by the need for the Supreme Court to amplify a plurality opinion in Watson, a truly good adverse impact case, in Wards Cove, which should have been Int. Teamsters v. United States (1977) [431 US 324] revisited as a pattern or practice case. Then there was the failed attempt to modify Wards Cove and five other 1989 Supreme Court rulings in the Civil Rights Restoration Act of 1990 (CRRA-90) and the successful modification of these cases in the Civil Rights Act of 1991 (CRA-91). We start by reviewing the chronology of rulings in Grant v. Metro. We then revisit the Supreme Court rulings in Watson and Wards Cove and point out some disconnects. We then conclude with some speculation about where Grant v. Metro may be going and why it matters.
We chose to end the column on somewhat of a tangent, but one of which we think the SIOP community should be aware. OFCCP is taking a cognitive ability testing case before an administrative law judge (ALJ) at the Department of Labor (DOL). OFCCP testing cases are not often litigated, and this case is notable for some other reasons. As such we wanted to keep the SIOP community posted.
Grant v. Metro
The facts of this case focus on employment practices at Metro Water Services (MWS) in Nashville and Davidson County, Tennessee (Metro). Plaintiffs alleged that Metro discriminated against Black employees under both pattern or practice and disparate impact theories. A set of “preselection” procedures were challenged, including tailored job qualifications, selective interviewing, and subjective decision making. There was no attempt to isolate (or disaggregate) these practices so as to link them with specific employment outcomes. The plaintiffs generally charged that the “preselection” procedures hurt Black employees with respect to posthiring opportunities, most notably promotion, but also with job assignments, pay, accommodations, discipline, and other terms and conditions of their employment.
At the district court level, Judge William J. Haynes, Jr. granted class certification for “all former, current, and future African-American employees of …Metro…from the period January 1, 2000 to the present.” However, a jury ruled for Metro on the pattern or practice charge, and Judge Haynes reserved judgment on adverse impact (Grant v. Metro  [727 F.Supp 2nd 677]). The plaintiffs appealed on pattern or practice, requesting a new trial. Judge Haynes agreed, and his motion was upheld by a three-judge panel of the 6th Circuit (See In re Metro  [ 606 F.3d 855]). A new trial on pattern or practice therefore awaits. Of primary interest here is that the 6th Circuit also ordered Judge Haynes to rule on adverse impact, and his ruling was that the plaintiffs presented sufficient prima facie evidence of adverse impact. The plaintiffs were awarded back pay (to be determined by a special master) and immediate injunctive relief prohibiting Metro from conducting oral interviews for MWS promotions or imposing an interview requirement for lateral transfers.
The August 26, 2011 ruling by the 6th Circuit addressed Metro’s appeal of Judge Haynes’ adverse impact ruling. As noted above, it was a split ruling in which two judges favored Metro and one judge dissented and favored the plaintiffs. Both the majority ruling and the dissent make sense but for different reasons.
The majority (Judges Batchelder and Sutton) focused on the first two prongs in CRA-91 for making an adverse impact prima facie case: (1) identifying “specific employment practices…responsible for any observed statistical disparities” and (2) proof that these practices cause adverse impact. Prong 1 has an exception such that if the decision-making process cannot be disaggregated for statistical analysis, then the entire process may be analyzed as a single employment practice.
Judge Batchelder, who wrote the majority opinion, ruled that the plaintiffs “never attempted to demonstrate that the elements of that process are incapable of separation for analysis.” This notion is similar to the requirement endorsed by the majority of Supreme Court justices in the Wal-Mart v. Dukes ruling that we reviewed in the last column. Explaining further, Judge Batchelder ruled:
The district court appears to have assumed that merely challenging the promotions process as a whole is sufficient to take advantage of the statutory exception, but that is simply not the law.…[CRA-91] clearly requires plaintiffs to identify and isolate specific employment practices.…A plaintiff may challenge the process as a whole only if he first demonstrates that its elements are incapable of separation.…The district court erred by allowing Plaintiffs to reap the advantages of the statutory exception without first meeting its requirements.
For Prong 2, the plaintiffs presented stock statistical comparisons showing that Blacks were disproportionately represented in lower paying jobs. Or in Batchelder’s words, the plaintiffs’ statistical expert:
[F]ocused specifically on the representation of “blacks in higher level positions compared to the overall black to white ratio at MWS.” He did not look at actual promotion rates, nor did he compare the ratios of black and white employees eligible for promotions with those who actually received promotions. He explained that, in light of MWS’s alleged practice of altering job qualifications and criteria, it was impossible to determine who was actually eligible for promotions.
In short, the majority ruling reduces the requirement to show actual applicant flow disparities (i.e., significant differences in specific selection rates) in order to prove causation from procedure to employment outcome. Broader comparisons of representation at different organizational levels were not appropriate for adverse impact inferences.
The dissenting judge (Clay) made three major points. First, he disagreed that the plaintiffs failed to identify specific employment practices. Second, he pointed to several pieces of anecdotal evidence of discrimination. For example, he cited the following example: “In one instance, Metro eliminated a bachelor’s degree requirement for a director position after a qualified Black employee applied, and awarded the position to a White applicant without a degree, even though the previous director had both bachelor’s and master’s degrees.”
Several such probative examples were cited. However, third, and most important for present purposes, Judge Clay opined that the statistical evidence provided by the plaintiffs was sufficient for a prima facie case of adverse impact because it showed that “the rate of promotions of Black employees, across nearly every job category, was three to four standard deviations lower than would be expected in the absence of discrimination.”
Both Batchelder and Clay raise important issues. However, Batchelder’s requirement for applicant flow disparities stemming from a facially neutral process is, in our opinion, the correct one for a prima facie case of adverse impact, whereas the anecdotal evidence of discriminatory decision making coupled with disparities between minority and nonminority employees in different job categories is the traditional requirement for a prima facie case of class-wide disparate treatment. Thus, we think that Judge Batchelder’s ruling was correct on disparate impact, and Judge Clay’s opinion is relevant to and should be reserved for the new trial on the pattern or practice charges. Now we turn to how this case is related to Watson and Wards Cove.
The Watson and Wards Cove Rulings
Only eight justices heard the Watson case, and they unanimously agreed that subjective selection decisions are subject to adverse impact rules. However, a plurality of four argued for fundamental changes in the adverse impact scenario. The facts are that Clara Watson, a Black woman, was passed over for promotion four times, each time in favor of a White applicant, and each time based on subjective ratings by White supervisors, including ratings of (a) job performance, (b) interview performance, and (c) past experience. It was unclear how these ratings were combined, but there was clearly bottom-line adverse impact for the total selection process.
Speaking for three others (Rehnquist, Scalia and White), Justice O’Connor proposed a major change in the adverse impact scenario originally formed in Griggs v. Duke Power (1971) [401 US 424] and Albemarle v. Moody (1975) [422 US 405] such that (a) plaintiffs must identify a cause(s) of adverse impact (which Clara Watson did), (b) prove the causal relationship statistically, and (c) force the defendant to articulate a legitimate nondiscriminatory reason to support the identified causes (as in disparate treatment cases such as McDonnell Douglas v. Green (1975). It was the third proposal that led to controversy because previously cases involving cognitive tests (e.g., Griggs and Albemarle) satisfied the first two proposals, thereby forcing the defendant to prove (not simply articulate) that the identified cause(s) is job related. In other words, O’Connor’s proposal transplanted a traditional pattern and practice burden to the adverse impact scenario, which was a substantially lighter burden than the traditional adverse impact burden of demonstrating job relatedness.
A year later, Justice Kennedy joined the Court, forging a majority of five that turned O’Connor’s plurality opinion in Watson into case law. The facts in Wards Cove were that two salmon packing companies had a hiring-hall arrangement for unskilled jobs (salmon packing) but used different procedures to hire skilled workers (e.g., machinists), who were paid, fed, and housed better than the unskilled packers. Eskimo and Filipino workers were overrepresented in the unskilled jobs and underrepresented in the skilled jobs. That’s why it should have been Teamsters revisited as a pattern or practice case. As in Teamsters, minorities were congregated in a less desirable job and Whites in more desirable jobs. However, in Teamsters, it was clear that minority and White workers were equally qualified; this was the arguable issue in Wards Cove. More on that part later. For present purposes, a majority of five used this case to cement O’Connor’s prior proposals related to the transplanted pattern or practice burden.
Then, Congress tried to overturn Wards Cove (and five other cases) in CRA-90, but President Bush vetoed the bill. The House easily overrode the veto, but the Senate failed to concur by only a single vote. The main reason for the veto was political disagreement on Wards Cove. Feeling the pressure of a near override, Congress got serious in CRA-91 and overturned Wards Cove and the other cases. CRA-91 kept O’Connor’s first two proposals but not her third. Of interest here for Grant v. Metro is the provision that if the cause of adverse impact cannot be identified because components of it cannot be disaggregated, then the total selection procedure should be analyzed as a single practice. More importantly for the general adverse impact scenario, the defense for proving adverse impact was restored to “job relatedness and consistency with business necessity,” which, for all intents and purposes, is a restatement of the original defense from Griggs and Albemarle.
Putting the Pieces Together
What characterized cases like Griggs and Albemarle is that proof of adverse impact required significant differences in applicant flow rates stemming from facially neutral processes (or implied chilling factors such as educational requirements). What characterized traditional pattern or practice cases such as Teamsters is the need to explain (not necessarily prove) why gross stock statistical disparities (either comparing representation across levels or to census data) existed. What muddied the waters was a slew of lower court cases in which plaintiffs charged both adverse impact and pattern or practice based on only stock statistics. Historically, this was not an issue because the ruling invariably went in the direction of the pattern or practice claim. That is, no adverse impact claim in any such case was supported while, at the same time, the pattern or practice claim was turned down. Similarly, the adverse impact charges failed every time the pattern or practice charges failed. The lone exception to this rule, both pre- and post-Wards Cove, was Wards Cove.
In Wards Cove, the district court favored the defendants on both pattern or practice and adverse impact. The 9th Circuit, while upholding the pattern or practice ruling, reversed the adverse impact ruling, thereby forcing the defendants to prove job relatedness of their selection procedures. Sound familiar? In Grant v. Metro, a jury ruled for the defense on pattern or practice, and the district court judge, after prompting from the 6th Circuit, upheld the adverse impact ruling. The district court judge also favored a motion for a new trial on pattern or practice, which the 6th Circuit endorsed. Like Wards Cove, there are many moving parts to the chronology of rulings in Grant v. Metro.
Allow us to speculate a bit. Regardless of what happens in the new trial, we think that there will likely be an appeal for en banc ruling by the 6th Circuit. Regardless of what this ruling ends up being, there will ultimately be a Supreme Court ruling. In our opinion, Wards Cove would have been business as usual had the Supreme Court correctly (a) identified the charges as valid for pattern or practice (thereby forcing defendants to articulate a legitimate explanation and plaintiffs to prove that the explanation is a pretext for class-wide disparate treatment) and (b) concluded that the adverse impact charges were invalid because there were no applicant flow statistics stemming from facially neutral processes (or implied chilling factors). The Supreme Court did not have to change any adverse impact rules. There has been no opportunity for such a ruling since Wards Cove.
Grant v. Metro also offers potential resolution for the “identification of a specific procedure” provision in CRA-91. Two out of three 6th Circuit judges ruled that Grant failed to carry its burden to show why the entire selection procedure should be treated as one practice, and a dissenting judge thought otherwise. However, this is the lesser of two concerns. Had there been failure to disaggregate in the face of bottom-line applicant flow disparities, we would agree with the dissenting judge. The more important issue is, in our opinion, that adverse impact and pattern or practice cases should not be confused with each other, which is, after all, what the Wards Cove ruling “accomplished.”1 Grant v Metro may ensure that they don’t. Stay tuned.
OFCCP Takes Disparate Impact Allegation Before an Administrative Law Judge
In early September OFCCP publically released2 a systemic discrimination complaint against Leprino Foods, which is a large producer of mozzarella cheese and whey products. Leprino has federal contracts totaling 5 million dollars, and as such must abide by Executive Order 11246. The allegation is that the company discriminated against qualified African-American, Asian, and Hispanic applicants for on-call laborer positions at the company’s Lemoore, CA facility by using an assessment that is not job related. OFCCP is seeking back pay and interest for at least 270 class members and job offers for at least 17 of the victims. The agency is also threatening cancellation of all existing federal contracts and debarment from future contracts.
The complaint is notable for a few reasons. First, it is an adverse impact allegation related to an assessment called WorkKeys, which was developed by ACT. Based on the complaint, Leprino uses the applied math, workplace observation, and locating information dimensions as part of their hiring process and argues that these are essential for on-call laborers that perform a variety of entry-level tasks such as inspecting products, monitoring equipment, and maintaining sanitation at the facility. According to the OFCCP, the selection rate based on this assessment over a 22-month period was 49% for minority applicants as compared to 72% for nonminority applicants. In the words of OFCCP Director Patricia Shiu, “the hiring process simply doesn’t pass the sniff test.” If recent OFCCP enforcement is any indication, then we assume that the sniff test is research meeting the requirements of the Uniform Guidelines.
Second, it is not often that OFCCP adverse impact cases make their way to an ALJ. Only a handful of OFCCP systemic discrimination cases are escalated to an ALJ in any given year, and over the last decade the vast majority of these cases have been pattern or practice allegations. Just about every OFCCP testing case that has settled over the last decade has done so voluntarily via a conciliation agreement and without the involvement of an ALJ.
Third, until recently it has been rare for the agency to threaten debarment. Recall that debarment from federal contracts is in theory the most damaging punishment in the OFCCP’s arsenal and is typically used with discretion in response to the most egregious violations of EO 11246. This is an adverse impact case, and as such the alleged discrimination is unintentional. For this reason the case seems different from other scenarios where debarment was threatened in response to the inability to conciliate an allegation of intentional discrimination under a pattern or practice theory.
One other consideration is worth noting. Given what the assessment measures, it is reasonable to expect adverse impact. The claim doesn’t shed light on much other than the fact that Leprino considers inferences made from the assessment to be job related. Again, given what the assessment measures, it is reasonable to expect some correlation with work outcomes, and we wonder whether this case may test the validity generalization argument with an ALJ. Some information available on the ACT Web site may offer a preview.3
In the past year, ACT commissioned two external experts to evaluate the WorkKeys system relative to the EEOC’s 1978 Uniform Guidelines on Employee Selection Procedures as well as more current professional standards. In a joint report issued in April 2010, two prominent industrial-organizational psychologists, Dr. Frank Schmidt of The University of Iowa and Dr. James C. Sharf of Sharf & Associates, Employment Risk Advisors, Inc., in Alexandria, VA, drew the following conclusions about the validity of the use of WorkKeys assessments as employee selection criteria.
A concluding comment written by Dr. Sharf states:
WorkKeys assessments are professionally defensible, content-valid measures of verbal, quantitative, and technical/problem-solving skills and abilities that are necessary prerequisites to successfully acquiring skills and performing tasks in the workplace. This conclusion is [also] supported by validity generalization principles, which are now well documented by research in the field and well accepted within contemporary industrial psychology. WorkKeys assessments measure specific cognitive skills/abilities/aptitudes, which, when used in combinations of three or more assessments, are a measure of general cognitive ability. Thus, it is my opinion that use of the WorkKeys assessments and the National Career Readiness Certificate as described [in the study] is legally defensible relative to the Uniform Guidelines, and that the WorkKeys assessments are “job related and consistent with business necessity,” as required under the Civil Rights Act of 1964.
Stay tuned. 2012 should be interesting on the legal front.
Albemarle v. Moody (1975) [422 US 405].
Grant v. Metro [2011 U.S. App. LEXIS 18054].
Grant v. Metro (2010) [727 F.Supp 2nd 677].
Grant v. Metro (2010)[606 F.3d 855].
Griggs v. Duke Power (1971) [401 US 424].
International Brotherhood of Teamsters v. United States (1977) [431 U.S. 324].
McDonnell Douglas v. Green (1975) [411 US 792].
Wal-Mart Stores, Inc. v. Dukes. (2011) [131 S.Ct. 2541].
Wards Cove v. Antonio (1989) [490 US 642].
Watson v. Fort Worth Bank (1988) [487 US 977].
Gutman, A. (2005). Adverse impact: Judicial, regulatory, and statutory authority. In F. J. Landy (Ed.), Employment discrimination litigation: Behavioral, quantitative, and legal perspectives (pp. 20–46). San Francisco, CA: Jossey Bass.
Gutman, A., Koppes, L. L., & Vodanovich, S. J. (2010). EEO law and personnel practices (3rd edition). New York, NY: Routledge, Taylor & Francis.
1 We note one caveat that this column has covered before. When comparing Title VII case law to OFCCP enforcement of Executive Order 11246, the difference between pattern or practice and adverse impact becomes even ambiguous. Based on OFCCP conciliation agreements and consent decrees, it appears that OFCCP consistently applies a pattern or practice theory of discrimination to applicant flow disparities when a selection system is ambiguous and not clearly organized into specific steps that could be challenged under an adverse impact theory. In most settlements the agency alleges that the selection process was (a) not applied in a standardized way for all applicants and (b) to the disadvantage of a protected group, and as such a pattern or practice argument may be reasonable because the process is not truly facially neutral.