Home Home | About Us | Sitemap | Contact  
  • Info For
  • Professionals
  • Students
  • Educators
  • Media
  • Search
    Powered By Google

On the Legal Front: 
OFCCP Settlement Review: What Was the Burden on Bank of America? 



Arthur Gutman
Florida Institute of Technology

Eric Dunleavy
DCI Consulting

On January 21, 2010, Linda S. Chapman, a Department of Labor administrative law judge (ALJ), ruled that Bank of America (BoA) was guilty of a pattern or practice of discrimination against Black applicants for entry-level teller and clerical/administrative jobs in 1993, and again from 2002 to 2005.1 ALJ Chapman ruled that the OFCCP established a prima facie case of discrimination that BoA could not rebut. There are several reasons why we think this ruling is important, so bear with us as we develop the story. We start with an overview of the traditional role of the OFCCP in enforcing Executive Order 11246 (EO 11246) followed by an overview of the OFCCP’s recent emphasis on “systemic discrimination.” We then present the facts of the BoA case and conclude with the reasons we believe this ruling is important.

1 The text of the ruling is available at http://op.bna.com/dlrcases.nsf/r?Open=kmgn-82dlq7.

Traditional Role of the OFCCP

EO 11246 has three major parts corresponding to (I) federal agencies, (II) procurement contracts, and (III) construction contracts. Part I is enforced by the EEOC, and Parts II and III are enforced by the OFCCP, a branch of the DOL. Contractors doing business with the federal government under Parts II and III must comply with the EEO Clause, which contains the nine provisions depicted in Table 1. Among these, Provisions 6, 7, and 8 empower the OFCCP to regulate, investigate, and impose sanctions and penalties for failure to comply with Provisions 1 or 2.


Table 1
The Equal Opportunity Clause

Provision 1 Nondiscrimination based on race, color, religion, sex and national origin
Provision 2 Affirmative action based on race, color, religion, sex, and national origin
Provision 3 Posting notices for employees and applicants for employment
Provision 4 EEO statement for all advertisements and solicitations
Provision 5 Notification of unions of obligations under EO 11246
Provision 6 Agreement to comply with EO 11246 and DOL rules and    regulations
Provision 7 Agreement to furnish information, books, and records by    request of the DOL
Provision 8 Agreement to DOL sanctions and penalties for noncompliance
Provision 9 Inclusion of above provisions in subcontracts and purchase orders


Historically, the OFCCP’s main focus has been on Provision 2. In accordance with Revised Order #4 of the OFCCP Regulations (1971), contractors with 50 or more employees must submit annual EEO-1 reports, and contracts in excess of $50,000 must develop affirmative actions plans (or AAPs) to correct underutilization of minorities and women. According to Revised Order #4, underutilization is assessed by comparing the percentage of minorities and women in the workforce to the percentage of qualified and available minorities and women in the relevant labor pool.3  However, recent descriptive reviews of OFCCP enforcement (Cohen & Dunleavy, 2009; Cohen & Dunleavy, 2010) suggest that, toward the end of the Bush administration, OFCCP focused primarily on nondiscrimination under Provision 1.

2 In addition, the OFCCP requires preapproved AAPs for contracts in excess of $1 million. In the past, the OFCCP also used other methods, including imposed plans, hometown plans, and the so-called “standard clause.“

3 In most cases, the relevant labor pool is in the immediate vicinity of the company. In other cases, it is much wider and may include the entire country (e.g., for university professors).

The OFCCP’s primary tool is the compliance review via desk audits on EEO-1 and AAP data, as well as onsite reviews. The OFCCP must attempt to gain voluntary compliance if a contractor is deemed to be noncompliant. However, if an agreement cannot be reached, the OFCCP may impose its sanctions and penalties. Once imposed, the contractor has to appeal the ruling to the OFCCP and the case is heard by an ALJ (as in the BoA case). The contractor must then appeal to the Secretary of Labor (and lose) in order to gain access to federal district court. Once in court, the burden of proof that a violation was not committed is on the contractor.

An additional point to note is that federal courts have permitted the OFCCP to render affected class rulings. For example, in US v. Duquesne Light Co. (1976), the DOJ sued to enforce an OFCCP order for back pay to an affected class of minority and female employees and applicants. Critical for present purposes, the OFCCP has used affected class ruling to gain compliance with Provision 1 in Table 1 in recent years.

Affect Class Rulings on “Systemic Discrimination”

The term “systemic discrimination” is essentially synonymous with the term “pattern or practice” of discrimination. Indeed, in the BoA ruling, ALJ Chapman used the two terms interchangeably. However, as we will witness below, the facts in the BoA case differ dramatically from those found in landmark Title VII pattern or practice rulings. In general, pattern or practice cases brought under Title VII feature gross statistical disparities between composition of the labor force as compared to a relevant labor pool (see Hazelwood School District v. US, 1977) or gross disparities between the compositions of two jobs within the same workforce (see Int. Teamsters v. US, 1977). For example, in Teamsters, Justice Steward defined pattern or practice as follows:

As the plaintiff, the Government bore the initial burden of making out a prima facie case of discrimination...and because it alleged a system-wide pattern or practice of resistance to the full enjoyment of Title VII rights, the Government ultimately had to prove more than the mere occurrence of isolated or “accidental” or sporadic discriminatory acts. It had to establish by a preponderance of the evidence that racial discrimination was the company’s standard operating procedure—the regular rather than the unusual practice [emphasis added].

The teamsters lost because there was an “inexorable zero” number of minorities in the lower paying of two bus driving jobs, and they could not offer a legitimate nondiscriminatory reason why this disparity existed. In comparison, in Hazelwood, where the plaintiffs won, the government could not prove a gross disparity between minorities in the workforce versus the relevant labor pool, and the Supreme Court ruled that whatever disparity existed was a function of the school district’s inability to compete with a larger surrounding city in recruiting minority teachers, a legitimate, nondiscriminatory reason.

It is useful to consider a few other recent OFCCP settlements to understand the implications of the BoA ruling. Toward this end, we focus on the three rulings depicted in Table 2.4 The statistics used in Teamsters and Hazelwood are commonly referred to as stock statistics. The statistics used in the three cases in Table 2 are commonly referred to as flow statistics. Thus, far from being in the image of Teamsters or Hazelwood, these are adverse impact cases more in the image of Griggs v. Duke Power (1971) and Albemarle v. Moody (1975). 

4 These and other settlements may be obtained directly on www.dol.gov or by Google search under the combination of “OFCCP” & “name of employer.”


Table 2
Sample Settlements Reported by DOL Office of Public Affairs (OPA) 

Alleged adverse impact on Blacks based on reading and math skills tests.
Total award = $850,000
Goodyear Tire & Rubber
Consent decree for alleged hiring discrimination against women.
Total award = $925,000
Vought Aircraft
Alleged adverse impact by screening procedure and test against women and Black and Asian men.
Total award = $1,5000,000

In Goodyear (2007), the OFCCP targeted hiring and selection procedures, and Goodyear agreed to pay $975,000 to approximately 800 women rejected for entry-level operative and laborer jobs and to hire 60 of them. The specific causes of adverse impact in this settlement are unclear. However, in Vought (2008), allegations of adverse impact featured an applicant screening procedure and a test, and Vought agreed to a $1,377,500 back pay award for 1,045 applicants. In Whirlpool (2005), allegations of adverse impact featured a reading and math skills test and a structured interview, and Whirlpool agreed to pay $850,000 to 800 rejected Black applicants. Interestingly, Whirlpool’s AAP statistics showed that the percentage of Blacks in the at-issue positions (30%) was greater than the percentage available in the workforce (25%). In addition, the OFCCP acknowledged this fact and agreed that the negative consequences were “inadvertent” yet emphasized disparities in applicant flow data.

The distinction between adverse impact and pattern or practice can be complex (see, for example Gutman, 2005). Nevertheless, they are clearly different in one critical dimension: The pattern or practice of discrimination implies a motive (or intention) to discriminate, whereas adverse impact does not. Our position below is that the BoA case should have probably been an adverse impact case not a pattern or practice case.

Facts of the BoA Case

The BoA case originated in 1993 against NationsBank, which was later acquired by BoA. After a 5-day onsite compliance review at a North Carolina branch, the OFCCP requested additional information from that branch and from branches in Tampa and South Carolina that had not been audited. NationsBank objected to discovery at the Tampa and South Carolina branches based on the 4th Amendment and, later, objected to discovery at the North Carolina branch as well. Early 4th Amendment rulings favored the bank, both within the DOL and the federal courts. However, these rulings were later overturned. The delay afforded the OFCCP time to add the charges for the later years. BoA argued that the OFCCP was afforded 15 years of discovery, which ALJ Chapman rejected on grounds that the delay was due to the 4th Amendment claim and subsequent appeals.

The prima facie case was based on statistical analyses by Dr. David L. Crawford and the testimony of three excluded applicants. Crawford testified that there were significant applicant flow disparities in 1993 and from 2002 to 2005, well in excess of two standard deviations for two steps in the selection process: (a) applicants selected for job interviews and (b) interviewees selected for the jobs. Critical to Crawford’s analysis, applicants excluded based on drug tests (coded “RD”)5 and reference checks (coded “RR”) were not included in his analysis, whereas applicants rejected based on credit checks (coded “RC”) and hours not compatible with the position (coded RH) were included. In addition, Crawford aggregated the teller and clerical/administrative jobs in his analyses.

5 Drug tests were administered in 1993, but not in the later years.

BoA’s defense was based on the testimony of two bank recruiters and its own statistical expert (Dr. Joan G. Haworth). The recruiters testified that in the early 1990s, jobs were advertised twice weekly in the local newspaper and were posted with the EEOC. Applicants then came in to complete applications. An administrative assistant then tore off attachments with EEO information, and decisions were made regarding which applicants to interview. Accordingly, the recruiters did not know the race of any applicant until they came in for interviews. A similar process was described for the years between 2002 and 2005, with the addition of Internet recruitment.

Haworth, cited several flaws in Crawford’s analysis, most notably: (a) mistakes in the data that were not corrected, (b) aggregation of the clerical and administrative jobs, and (c) inclusion in the analyses of applicants excluded based on credit checks (coded RC) and time incompatibilities (coded RH) were inconsistent with the OFCCP Internet applicant rule (see Reynolds, 2006 for a review). Haworth testified that when the jobs were disaggregated, there were no shortfalls for the clerical/administrative jobs, and the shortfalls for the teller jobs were eliminated when the RC and RH codes were excluded from the analysis.

There were other data and statistical criticisms back and forth between Crawford and Haworth. This was clearly a complicated case. Nevertheless, the main issues in the eyes of ALJ Chapman were that Crawford was justified in aggregating the two classifications and in including the applicants coded RC and RH. Based on interpretations of major Supreme Court pattern or practice rulings including Teamsters and Hazelwood, Chapman ruled there were “gross” statistical disparities sufficient for a prima case of a pattern or practice of discrimination that BoA could not rebut.

Implications of the ALJ’s Ruling

There are three major reasons for questioning the ruling in this case. First, this was not a clear pattern or practice case based on traditional Title VII precedent. As noted earlier, pattern or practice cases are built on stock statistics as in Teamsters and Hazelwood. Here, the claim of pattern or practice was built on applicant flow disparities that should have been decided under adverse impact rules established in Griggs and Albemarle and codified by Congress in the Civil Rights Act of 1991.

Second, the traditional burdens used in disparate treatment and pattern or practice cases may have been “reversed.” Let’s imagine that this was a traditional pattern or practice case brought by the EEOC (or the plaintiffs themselves) to federal district court. Following the precedents established in Teamsters and Hazelwood, the plaintiffs would present their statistical evidence together with individual claims of disparate treatment. The defense could then rebut the prima case with its own statistical evidence, and the court would decide if there is a prima facie case. In the BoA case, there were two refutable statistical disparities, and it is arguable that a federal district court judge could rule there is no prima case based on either of two reasons. First, the statistical disparities were not significant when proper controls were used (i.e., the disposition codes in contention). Second, these were 2 disparities out of 33 job classifications, which may be insufficient evidence to make a prima facie case of “systemic” discrimination.

However, assuming that the prima facie case is made, the burden on the defense based on case law is to articulate (without having to prove) nondiscriminatory reasons why the disparities exist, thereby forcing the plaintiff to prove with direct or indirect evidence that the stated reasons are a pretext for racial discrimination. ALJ Chapman did not follow this route but, instead, placed the burden of proof on BoA. Citing Teamsters, Chapman ruled that “the bank cannot rebut this statistical evidence merely by arguing that its decisions were legitimate or nondiscriminatory.” This is a misinterpretation of Justice Stewart’s ruling in Teamsters, which was that an “inexorable zero” cannot be explained absent illegal discrimination. In contrast, Stewart accepted an articulation in Hazelwood. In our opinion (which of course is not binding), BoA had several plausible articulations: that the recruiters did not know the race of the applicants prior to interviews, that they were unbiased in their decision making, and/or there is no bias in exclusions based on the various codes used.

Interestingly, there was testimony in the BoA case that the plaintiffs would not have succeeded proving pretext. Most notably, the OFCCP’s auditor testified that exclusion based on credit checks is not suspect for bank jobs, and in addition, Crawford, the OFCCP’s expert, testified that he had no evidence to believe there was bias on the part of the recruiters. This is particularly notable given the recent attention the use of credit checks has received in the EEO context.

Our third reason for questioning the ruling is perhaps the most important one—a federal judge did not try this case nor will one ever try it if BoA agrees to settle with the OFCCP. The history of OFCCP rulings is that very few contractors go through the two-part appeals process to get to federal court. Therefore, unlike an EEOC determination, which can go to federal court directly, the OFCCP has, in effect, the power to interpret federal law as they see fit, as long as contractors eschew the two-part appeals process. Many expect that BoA will appeal to the Secretary of Labor and, in the face of a likely negative ruling, may take the case to federal court. Only time will tell.

Implications for I-O Psychologists

There are a number of useful lessons to learn from the BoA ruling for I-O psychologists working in federal contractor organizations and/or who are responsible for selection systems used by contractors. We highlight three here. First, for better or worse, this case is representative of the vast majority of recent OFCCP “systemic” settlements. In these cases the allegation is usually a pattern or practice of discrimination in hiring, and an unstructured hiring process characterized by subjective judgment is identified as the discriminatory mechanism. In fact, the vast majority of recent OFCCP settlements have not been traditional adverse impact cases where a structured selection procedure (e.g., a test, work sample, interview, etc.) is challenged along the Uniform Guidelines on Employee Selection Procedures (UGESP) lines because it causes a disparity. For example, only 10 of 73 OFCCP settlements in fiscal year 2008 focused on the adequacy of a specific selection procedure. In the rest of those settlements, a discriminatory hiring process that could not be broken down into components was challenged, and as such, an applicant-to-hire analysis was considered the probative measure of disparity.

Second, statistical significance testing is being used as a “stand-alone” measure of disparity in the vast majority of OFCCP settlements. The specific statistics being used most often are the two-sample Z test of independent proportions and, when the applicant pool has less than 30 applicants, Fisher’s exact test. Note that these tests appear to be used regardless of which sampling model (e.g., binomial or hypergeometric) best mirrors the reality of the selection event. It is also worth noting that a violation of the 4/5th rule was seen in only 2 of 73 OFCCP settlements from FY 2008. Thus, although the OFCCP enforces UGESP and the OFCCP compliance manual requires a 4/5th rule analysis to demonstrate practical significance, statistical significance tests of selection rates stemming from applicant flow data are the most common “stand alone” disparity analyses seen in OFCCP settlements. We think that, for better or worse, the role of practical significance measures in the EEO context (OFCCP enforcement, EEOC enforcement, private litigation) has declined in recent years, and it will be interesting to monitor this issue over time as applicant behavior continues to mature with Internet application systems.

Third, the BoA case is a prototypical example of how disposition codes can make a huge difference in the results of a disparity analysis. The OFCCP Internet applicant definition allows federal contractors to differentiate job seekers from actual applicants, and only actual applicants are included in disparity analyses. Basic qualifications that are noncomparative, objective, and job related can be used to differentiate job seekers from actual applicants if they are implemented in a standardized fashion. In this case, the inferences from disparity analyses hinged on two disposition codes (credit checks and time incompatibilities) and whether persons disqualified for these reasons should be considered applicants. Although these disposition codes seemed reasonable as exclusionary criteria on their face, the ALJ disagreed, and disparities were significant when these persons were included in the analysis. 


     Cohen, D. B., & Dunleavy, E. M. (2009). The Center for Corporate Equality releases a review of OFCCP settlements from fiscal year 2007. The Industrial-Organizational Psychologist, 47(1), 145–149.
     Cohen, D. B., & Dunleavy, E. M. (2010). A review of OFCCP settlements from fiscal year 2008. Washington DC: Center for Corporate Equality.
     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: Jossey Bass.
     Reynolds, D. (2006). OFCCP guidance on defining a job applicant in the Internet age: The final word? The Industrial-Organizational Psychologist, 42(2), 107–113.

Cases Cited

     Albermarle Paper Co. v. Moody (1975) 422 US 405.
     Griggs v. Duke Power Co. (1971) 401 US 424.
     Hazelwood School Dist. v. United States (1977) 433 US 299.
     International Brotherhood of Teamsters v. United States (1977) 431 US 324.
     US v Duquesne Light Co. (W.D. Pa. 1976) 423 F.Supp 507.