You are on page 1of 124

Legal Studies Research Paper Series

Title VII Disparate Impact Liability Makes Almost

Everything Presumptively Illegal

Research Paper No. 19-421


Gail L. Heriot

This paper can be downloaded without charge from the

Social Science Research Network Electronic Paper Collection: = 3482015

Electronic copy available at:

Working Draft
November 2019

Title VII Disparate Impact

Liability Makes Almost Everything
Presumptively Illegal …
It Gives the Federal Bureaucracy Extraordinary
Discretionary Power. But What Does It Do to the Rule
of Law? And Who Benefits?


Gail L. Heriot*

*Professor of Law, University of San Diego

& Member, U.S. Commission on Civil Rights.


Electronic copy available at:

***Table of Contents***

I. Introduction………………….………………………………………………………3

II. The 88th Congress Intended Title VII to Require Equal Treatment,
Not Equal Outcomes…………………………………………………………...…..6

III. Disparate Impact Liability Was the Brain Child of the EEOC,
Which Turned Out Not to Be a Toothless Tiger After All............................19

IV. Disparate Impact Liability’s Incoherence: Everything Has a Disparate

Impact on Some Group that Is Protected by Title VII..................................24

V. Griggs v. Duke Power Co.: The Supreme Court Deferred to the EEOC,
Not Congress, in Adopting Disparate Impact Liability..…………….....…..33

VI. The Pre-Wards Cove Business Necessity Defense Was So Difficult to

Establish that Many Employers Likely Opted
for Proportional Hiring Instead.…………………………………………...…..38

VII. The Effect of Civil Rights Act of 1991: Despite the Efforts of the Bush
Administration and its Congressional Allies, the Business Necessity
Defense Remains Very Difficult to Establish ………………………………...60

VIII. Criminal Background Checks: An Example of Disparate Impact

Liability in Action.……………………………………………………….……….80

IX. Perverse Effects: Disparate Impact Liability Can Harm its Intended
Beneficiaries (and in the Case of Criminal Background Checks
Probably Does)……………………………………………………….…………….89

X. Disparate Impact Liability Is Itself a Form of Discrimination

that Must Be Subjected to Strict Scrutiny ………………………...….…..….97

XI. Disparate Impact Liability Would Likely Fail to Survive Strict Scrutiny..105

XII. The Problem of Ubiquity: Are There Due Process and Separation of
Powers Issues Associated with the EEOC’s Discretion?..............................111

XIII. Conclusion……………………………………………………………………...…..121


Electronic copy available at:

I. Introduction.

The passage of Title VII of the Civil Rights Act of 19641 was historic. It was not,
however, intended to assert federal control over every aspect of the workplace. Its
carefully limited purpose was to prohibit employment discrimination based on race, color,
religion, sex and national origin. As William M. McCulloch, then-Ranking Member of the
House Committee on the Judiciary, put it:
[M]anagement prerogatives, and union freedoms are to be left
undisturbed to the greatest extent possible. Internal affairs of employers
and labor organizations must not be interfered with except to the limited
extent that correction is required in discrimination practices.2
Things haven’t quite worked out as originally intended. Today, the Equal
Employment Opportunity Commission wields considerable influence over every facet of
the employment relationship.3 It doesn’t just enforce prohibitions against discrimination.
Through its various “guidances,” this powerhouse agency with well over 2000 employees
manages employment practices generally. And it does so with surprisingly little
supervision by democratically elected officials or by the courts.
That isn’t good news. When a federal agency has enormous discretionary authority
over employment practices, it places even the most conscientious and law-abiding
employers at risk. One wrong move—or even one right move that EEOC staff members
happen to disapprove of—can lead to a costly investigation and a lawsuit. And the more
employees an employer has, the more risk the employer may be undertaking. Only the
most obstinate of bureaucrats could fail to see how this can discourage hiring. Jobs that
might have been created in the United States may end up elsewhere.

A significant chunk of the EEOC’s Title VII authority over employment practices
comes under the rubric of “disparate impact liability.” This form of liability was the
1 Civil Rights Act of 1964, Pub. L. No. 88–352, 78 Stat. 241, 253–66 (1964) (codified at 42 U.S.C. §

2 Statement of William M. McCulloch, et al., H.R. REP. NO. 914, 88th Cong., 2d Sess. at 29 (1964).
McCulloch was the House Judiciary Committee’s ranking member and was considered by many to
have been indispensable in passing the Act. W ILLIAM N. E SKRIDGE , J R ., P HILIP F RICKEY &
(4th ed. 2007) (discussing McCulloch’s critical role in passing the Civil Rights Act of 1964). See also
J.Y. Smith, Former Rep. William McCulloch Dies, W ASH . P OST , Feb. 23, 1980, at C8 (stating that
McCulloch has been called “the patron saint of civil rights legislation”).



Electronic copy available at:

brainchild of EEOC staff members, not of Congress. To put it bluntly, it is virtually
certain that in 1964 Title VII’s Congressional supporters would have strongly
disapproved.4 Yet, it gained the Supreme Court’s acceptance in Griggs v. Duke Power Co.
(1971)5 and the apparent acquiescence of Congress in 1991.6 Consequently, this internally
incoherent doctrine is now well-entrenched in the law.

The wrong turn happened rather quickly. Within just a few years of Title VII’s
passage, the EEOC, concerned over the recent tumult of the 1960s, began interpreting the
law to prohibit not just discrimination on the ground of race, color, religion, sex, or
national origin, but also job qualifications that make it likely that proportionally fewer
minority or female applicants will be hired than whites or males.7 Put differently, but not
unfairly, the EEOC interprets Title VII to require not just equal treatment, but near equal
results (at least insofar as it deems unequal results to have failed the standard of
“business necessity”).
Perhaps the most well-known recent example is the EEOC’s policy on job applicants
with felony records, which essentially holds that employers are required to hire at least
some of them.8 The rationale behind the policy is that despite the fact that the
overwhelming majority of Americans of all races are not felons, since proportionally more
African Americans than Americans of other races have felony convictions, they are
disproportionately affected when employers prefer to hire applicants with clean criminal

Note the obvious: A well-considered policy easing the transition of prisoners back
into the mainstream may indeed be a good idea.9 But Title VII is not such a policy. It
doesn’t say a word about prohibiting discrimination on the basis of criminal record. Yet
somehow the EEOC has come to interpret Title VII to cover conviction status.
Consequently, employers who wouldn’t dream of discriminating on the basis of race, but
who are uncomfortable employing ex-convicts of any race, are presumptively in violation.

4 See infra Part II.

5 Griggs v. Duke Power Co, 401 U.S. 424 (1971). See infra Part V.
6See Civil Rights Act of 1991, Pub. L. No. 102-166, 105 Stat. 1071-1100 (1991) (codified at various
sections of 29 U.S.C. and 42 U.S.C.). See infra Part VII.

7 See infra Part III.

OF THE CIVIL RIGHTS ACT OF 1964 (2012) (hereinafter the “2012 Guidance”). See also U.S. COMM’N ON

9 See infra Part VIII.


Electronic copy available at:

Unless they can prove that they had no choice—in the EEOC’s terms, that they were
compelled by “business necessity”—they can be sued on the improbable ground of race
discrimination. Some already have been.10

There is irony to all this. The Rev. Martin Luther King, Jr. famously looked
forward to the day when his children would be judged by the content of their character
rather than the color of their skin. Title VII was supposed to help get us there. But
current law turns his hope on its head. Under the Supreme Court’s decision in United
Steelworkers v. Weber (1979),11 Title VII is (implausibly) interpreted to permit explicit
preferential treatment based on race in affirmative action programs. Meanwhile, the
EEOC, applying its doctrine of liability for disparate impact, heavily limits the extent to
which the content of one’s character, at least as revealed by one’s criminal record, may be
taken into account.

The EEOC’s policy on criminal convictions is just one example of disparate impact
liability in action. There are many more. Indeed, the supply of potential bases for
disparate impact liability is virtually limitless, since, as this article will discuss,
everything or nearly everything has a disparate impact on some race, color, religion, sex,
or national origin group.12 That brings the law into disrepute.
How did we get to the point where everything or nearly everything an employer can
do is presumptively a violation of Title VII? What incentives does it create for employers?

10 See EEOC v. Peoplemark, 2011 U.S. Dist. LEXIS 38696 at *1-4 (W.D. Mich. 2011), aff’d U.S. Dist.
LEXIS 154429 (W.D. Mich. 2011), aff’d 732 F.3d 584 (6th Cir. 2013) (lawsuit brought against
company for its criminal background policy where EEOC wrongly accused employer of having more
stringent policy than it in fact had); EEOC v. Freeman, 961 F.Supp.2d 783, 803 (D. Md. 2013) (court
rejected testimony of EEOC experts, concluding that: “The story of the present action has been that
of a theory in search of facts to support it. But there are simply no facts here to support a theory of
disparate impact resulting from any identified, specific practice of the Defendant”). In Peoplemark,
the trial court ordered the EEOC to pay Peoplemark $751,942.48 in attorneys’ fees, expert witness
fees, and other costs. U.S. Dist. LEXIS 154429 at *12-13. Freeman was also dismissed. Win or lose,
however, employers who get sued are worse off—even when they are compensated for their
attorneys’ fees and costs. Many, like PepsiCo, would prefer to settle prior to litigation rather than
wind up in costly and time-consuming litigation. See, e.g. Press Release, Equal Emp’t Opportunity
Comm’n, Pepsi to Pay $3.13 Million and Made Major Policy Changes to Resolve EEOC Finding of
Nationwide Hiring Discrimination Against African Americans: Company’s Former Use of Criminal
Background Checks Discriminated Based on Race, Agency Found (Jan. 11, 2012), available at; Evan Pontz, Labor: Dealing with the
“New” EEOC, I NSIDE C OUNSEL (Nov. 14, 2011) (“Smart employers… will seek opportunities to
resolve EEOC charges quickly, efficiently, through other means (such as through private settlements
or private mediations) and to try to encourage the EEOC to expend its limited resources elsewhere”).

11 United Steelworkers v. Weber, 443 U.S. 193 (1979).

12 See infra Part IV.


Electronic copy available at:

What does it do to Americans’ respect for the law? Who, if anyone, has ultimately
benefitted? And why are even liberal lawyers concerned that it may be unconstitutional?13
These are questions that need to be addressed. But instead of addressing them, the
federal bureaucracy has spent decades trying to extend disparate impact into additional
areas of the law. And it has had some success in this.14

II. The 88th Congress Intended Title VII to Require Equal Treatment, Not
Equal Outcomes.

Title VII was supposed to be about discriminatory treatment, not disparate impact.
One among many ways that this is made clear is through Title VII’s text. Its central
prohibitions are as follows:
Section 703. Unlawful employment practices
(a) Employer practices
It shall be an unlawful employment practice for an employer—
(1) to fail or refuse to hire or to discharge any individual, or otherwise to
discriminate against any individual with respect to his compensation,
terms, conditions, or privileges of employment, because of such individual’s
race, color, religion, sex, or national origin; or
(2) to limit, segregate, or classify his employees or applicants for employment
in any way which would deprive … any individual of employment
opportunities or adversely affect his status … , because of such individual’s
race, color, religion, sex, or national origin.15

13 See infra Parts X & XI.

14  See
Smith v. City of Jackson, 544 U.S. 228 (2005); Texas Dept. of Hous. & Cmty. Affairs v.
Inclusive Cmtys, Project, Inc., 135 S. Ct. 2507 (2015).
15 § 703(a), 78 Stat, at 255-6 (codified at 42 U.S.C. § 2000e-2). The second provision was necessary in
order the take care of the “separate but equal” argument. Up until the Supreme Court’s decision in
Brown v. Board of Education, 347 U.S. 483 (1954), it was common to argue that state action that
provides “separate but equal” service to different races was not discriminatory. In Brown, the Court
determined that “separate but equal” schools violated the 14th Amendment’s Equal Protection
Clause, but it did not find that “separate but equal” arguments would fail in every context. 347 U.S.
483 at 495. Congress therefore was careful to proscribe all such separations in §703(a)(2).
Separations for the purpose of sexual privacy were authorized in a separate provision. See §703(e),
78 Stat. at 256 (codified at 42 U.S.C. § 2000e-2(e)). See also Browder v. Gayle, 352 U.S. 903 (1956)
(per curiam) (a decision that comes much closer to indicating that separate but equal in the context
of race is broadly unconstitutional).


Electronic copy available at:

The prohibition was drafted in two subsections in order to head off the argument
that “separate but equal” treatment is permissible. Note that it was written less than a
decade after Brown v. Board of Education. Brown itself had only held that “separate but
equal” is unlawfully discriminatory in the context of public education. The possibility that
it would be held to be acceptable elsewhere had to be kept in mind. It was thus important
to be clear in Title VII.
In both sections, using parallel language, Congress used the phrase “because of such
individual’s race, color religion, sex or national origin.” By doing so, it made explicit that
the reason for an employer’s action must be “race, color, religion, sex or national origin.”
Richard K. Berg was a Department of Justice attorney who advised Senator
Hubert Humphrey on the pending civil rights bill and who went on to be Deputy
General Counsel of the EEOC. He opposed the EEOC’s claim that Title VII
imposed liability for disparate impact and put his opposition this way:
“Discrimination is by its nature intentional. It involves both an action and a reason
for the action.”16 “To discriminate ‘unintentionally’ on grounds of race [],” he wrote,
“appears a contradiction in terms.”17  
An important clarification is useful here. While it is true that Title VII wasn’t
intended to outlaw job qualifications simply because they have a disparate impact on some
race, color, religion, sex, or national origin group, the fact that an employer chooses a job
qualification with a disparate impact on a group may be evidence that the employer is
motived by an intent to discriminate against that group. For example, if a bank were to
require loan officers to be at least six feet tall, one might infer from that requirement that

16See Richard K. Berg, Equal Employment Opportunity Under the Civil Rights Act of 1964, 31
B ROOK . L. R EV . 62, 71 (1964). His adversary in that debate was Alfred Blumrosen, the EEOC’s first
“Chief of Conciliations” and a strong advocate of disparate impact liability. See infra Part III. In
Black Employment and the Law, Blumrosen described Berg this way:

Richard Berg, who would become Deputy General Counsel, had been in the Justice
Department and acted as advisor to Senator Humphrey during the debates in the
Senate. He saw the statutory language in the context of its original informal setting
(or, more precisely, what he understood to be the original informal setting). He had a
clear and personal understanding of the legislative intent. He knew of compromises
which were actually made in the process of putting together the Civil Rights Act of
1964. The proposals he made, and his reaction to the proposal[s] of others, were
influenced by his concern with keeping the bargain which he understood had been
struck in the legislature. The statute was a compromise, and Mr. Berg intended to
be reasonably faithful to this compromise. As he saw it, legislative intent was, not a
standard for review of administrative action, but a reality to be observed.

A LFRED W. B LUMROSEN , B LACK E MPLOYMENT AND THE L AW 57 (1971). Blumrosen himself was less
concerned with the niceties about when the courts would sustain the policies he advocated. See infra
Part III.

17 Berg, supra note 16.


Electronic copy available at:

the bank is trying to exclude women, since women are rarely over six feet tall and it is
hard to fathom why the bank would care about the height of its loan officers. If this
evidence were combined, for example, with evidence of misogynistic remarks by the bank’s
manager and a history of sex discrimination, the inference might be quite strong.

On the other hand, if a bank were to require loan officers to have a clean criminal
record, few would jump immediately to the conclusion that the bank was attempting to
discriminate against men, even though such a requirement would have a strong disparate
impact on them. It would likely take a lot of additional evidence to persuade most people
that such discrimination was the true motivation for the requirement rather than caution
(excessive caution in the view of some) over the possibility of employee misconduct.

Hence no one disputes that Title VII plaintiffs should be able to use the fact of
disparate impact as evidence that the employer was motivated by an intent to discriminate
on an impermissible basis. But context matters. The evidence on either side may or may
not be persuasive, depending on the facts of the case. If the point is that disparate impact
might signal an intent to discriminate, an employer must always have the opportunity to
show that he or she was not so motivated.18

18To be sure, we humans are complex creatures whose motives can be difficult to divine and easy to
misinterpret. But that problem is common to many areas of the law, most notably the criminal law.
It is sometimes difficult to differentiate between an individual who takes the victim’s personal
computer erroneously believing it to be his own and the one who takes it knowing it to be the
victim’s. But no one argues that the defendant should not be accorded the opportunity to explain
that he was simply in error.

Of course, there can be statutory grey areas, such as the case where the actor is not fully conscious of
the reasons for his action. Suppose, for example, an employer rejects an African American job
applicant, when he would have (and indeed has) hired white applicants who are otherwise identical
in every relevant respect. But the employer doesn’t realize this about himself. What then? Does
that qualify as discrimination “because of such individual’s race” within the meaning of Title VII?

One can imagine Members of Congress disagreeing about the answer to that question in 1964. At
the time, a lot of race discrimination was about as subtle as a freight train. In some parts of the
country is was routine for employers to advertise, “Help Wanted—White” or “Help Wanted—Colored”
or for a “Man Friday” or a “Girl Friday.” See Marsha H. Nathanson, 33 Years Ago, a Newspaper’s
Discriminating Classifieds, B ALTIMORE S UN (June 9, 1993) (expressing surprise at seeing the
discriminatory advertisements in the Help Wanted section of the Baltimore Sun in 1960). Some of
Title VII’s Congressional supporters may well have expected it to apply only to blatant kinds of
discrimination and not to the employer who isn’t conscious of his bias.

But courts have interpreted Title VII (correctly in my view) to cover employers who treat job
applicants differently “because of [their] race, color, religion, sex or national origin” regardless of
whether the employer is conscious of it. Somewhere in the back of his mind, he has drawn a
distinction between African Americans and whites. Race is the “reason for [his] action.”


Electronic copy available at:

Disparate impact liability (as it is currently established in law) is an entirely
different concept. Under it, it doesn’t matter what the employer’s intent was.19 Even if the
employer can prove beyond a doubt that he was not motivated by race, color, religion, sex
or national origin, he is still presumptively in violation of Title VII. All that matters is the
fact of disparate impact and the employer’s failure to persuade the EEOC (or the court)
that he was compelled by “business necessity” to act as he did. (More on that later.)20
Lack of business necessity for one’s actions and lack of discriminatory intent are very
different concepts.

Even if the text of Title VII had failed to make it clear that it bans only employer
action taken “because of [the complainant’s] race, color, religion, sex or national origin,”
the legislative history does so.21 Congressional leaders assured their colleagues that Title
VII would not interfere with employer discretion to set job qualifications—so long as race,
color, religion, sex and national origin were not among them. Senators Clifford Case (R-
N.J.) and Joseph Clark (D-Pa.), the bill’s co-managers on the Senate floor, emphasized this
in their important, bipartisan interpretative memorandum:
19See Griggs, 401 U.S. at 432 (“good intent or absence of discriminatory intent does not redeem
employment procedures or testing mechanisms that operate as ‘built-in headwinds’ for minority
groups and are unrelated to measuring job capability”); id. (“Congress directed the thrust of the Act
to the consequences of employment practices, not simply the motivation”). See also infra Parts III-V.

20 See infra Parts VI-VII.

21The following exchange on the floor of the Senate shows the parties believed that the bill before
them required proof of intent to discriminate:

Mr. ERVIN: The language is “on account of race, color, creed, or national origin.”

Mr. CASE: That is correct.

Mr. ERVIN: That is a matter of intent. The intent in the man’s mind is going to be
judged, not by him, but by somebody else.

Mr. CASE: Yes, but in the time-honored custom of Anglo-Saxon jurisprudence, under
the terms of the bill it would be determined by a court of law.

Mr. ERVIN: It would be determined not by the external acts in the case, but by what
some Federal employee believes was in the mind at the time period.

Mr. CASE: The Senator from North Carolina is trying to make it appear that it is
unusual to have a determination of what is right or wrong depend on a mental state.
Practically speaking, this has been universally true.

110 C ONG . R EC . 7254 (1964). For other examples of statements in the Congressional Record that
make it clear that Members of Congress were assuming intent to be necessary, see Michael Evan
Gold, Griggs’ Folly: An Essay on the Theory, Problems, and Origin of the Adverse Impact Definition
of Employment Discrimination and a Recommendation for Reform, 7 I NDUS . R EL . L.J. 429, 494-500


Electronic copy available at:

There is no requirement in [T]itle VII that employers abandon bona fide
qualification tests where, because of differences in background and
education, members of some groups are able to perform better on these
tests than members of other groups. An employer may set his
qualifications as high as he likes, he may test to determine which
applicants have these qualifications, and he may hire, assign, and promote
on the basis of test performance.22
Note that Case and Clark used the term “bona fide qualification tests,” meaning
qualification tests adopted in good faith, and not “scientifically valid qualification tests” or
“qualification tests compelled by business necessity.” To Case and Clark, the issue was
whether the employer chose a particular job qualification because he believed it would
bring him better employees or because he believed it would help him exclude applicants
based on their race, color, religion, sex or national origin. Note also that they stated an
employer may set his qualifications “as high as he likes,” not “as high as he likes provided
he can prove that he has a business necessity for doing so.”
Case and Clark went on to state that Title VII “expressly protects the employer’s right
to insist that any prospective applicant, Negro or white, must meet the applicable job
qualifications.” “Indeed,” they wrote, “the very purpose of [T]itle VII is to promote hiring
on the basis of job qualifications, rather than on the basis of race or color.”23
All that can be said here is, “Yes, of course.” Title VII was intended to deny
employers the authority to decline to hire a job applicant who meets its hiring criteria, but
who is, in the employer’s view, the wrong race, color, religion, sex, or national origin. It
was not intended to create a bureaucracy that would regulate the fairness of the
employer’s choice of job qualifications. 24 So long as those job qualifications were not
adopted for the purpose of discriminating on the basis of race, color, religion, sex, or
national origin, they were intended to be at the employer’s discretion, not the EEOC’s or
the courts’. An employer that adopts a silly qualification, not well aimed at getting the
best employees for the job, can expect to be punished by the market.
There is an interesting backstory here. The two senators were inspired to make this
point in response to what would likely have been an obscure case—that of Leon Myart—
had it not been for the fact that historic civil rights legislation was under consideration by
Congress at the time. Instead this decision by a little-known, part-time hearing examiner
for the tiny Illinois Fair Employment Practices case turned out to be anything but obscure.

22Case & Clark Memorandum, 110 C ONG . R EC . 7213 (1964). The Memorandum also stated:
“Requirements for the keeping of records are a customary and necessary part of a regulatory statute.
They are particularly essential in [T]itle VII because whether or not a certain action is
discriminatory will turn on the motives of the respondent, which will usually be best evidenced by his
pattern of conduct on similar occasions.” Id. at 7214 (italics added).

23 Id. at 7247.
24 See supra note 2 and accompanying text.


Electronic copy available at:

Leon Myart, an African American, had applied to Motorola, Inc. for a production line
position involving the testing and adjusting of television sets. He was given a short
standardized aptitude test, which, according to Motorola, he had done poorly on and was
therefore not hired. Myart, however, complained to the Illinois Fair Employment
Practices Commission, where he argued that he had passed the test and that instead he
had been rejected on account of his race. Even before the passage of Title VII, if Myart’s
allegation was true, Motorola’s conduct was an obvious violation of Illinois law.
After he filed his complaint, a similar test was administered to Myart at the
Commission offices, which he passed. This lent some credibility to his allegation. Since
Motorola was unable to produce the actual exam, the hearing officer found in March of
1964 that Myart had indeed passed Motorola’s test and ordered the company to hire him.
Had the hearing officer stopped there, the case would have passed without much
notice from anyone in Congress or the media. But he went further, concluding that
regardless of whether Myart had passed, the exam did not "lend itself to equal opportunity
to qualify for the hitherto culturally deprived and the disadvantaged groups." He
therefore ordered Motorola to stop using it.25

As the Chicago Daily News put it, “[t]he uproar over that order reverberated around
the country.”26 The media used it as a sign that Congress had better be careful in how they
draft Title VII or it could prove disastrous. New York Times pundit Arthur Krock called
the decision “autocratic” and “a graphic illustration that when a political arm of
government assumes jurisdiction over the hiring and firing policies of private business,
the tendency is to expand this authority.”27

Krock reported that Motorola’s attorney planned to appeal “all the way to the
Supreme Court if necessary” and that the Employers Association of Chicago, which
represented 1400 companies, would be supporting Motorola.28 It turned out to be
25Myart, Charge No. 63C-127 (Ill. F.E.P.C. 1964) (reprinted in 110 CONG. REC. 5662, 5664 (1964)).
See also Motorola, Inc. v. Illinois Fair Employment Practices Commission, 34 Ill. 2d 266, 269-70, 215
N.E.2d 286, 288 (1966) (quoting the hearing examiner); Arthur Krock, In the Nation: A Pilot Ruling
on Equal Employment Opportunity, N.Y. T IMES , Mar. 13, 1964, at 32.

26 FEPC Fails Its Test, C HIC . D AILY N EWS , Nov. 23, 1964. See also Discrediting FEPC, C HIC . T RIB .,
May 23, 1964 (“The Illinois Fair Employment Practices Commission continues to dillydally with the
Motorola case, which has caused so much alarm throughout the country that it has affected the civil
rights bill now before the Senate”) (reprinted in 110 C ONG . R EC . A2909 (June 1, 1964) (extension of
remarks by the Hon. Robert H. Michel). See also The State Will Do Your Hiring, C HIC . T RIB ., March
7, 1964 (“A foretaste of what employers may expect if the ‘equal employment opportunity’ provisions
of the Federal civil rights bill become law has been provided here in Illinois under the State fair
employment practices act. An agent for the commission enforcing the act has just told Motorola,
Inc., that the State hereafter assumes to direct its hiring practices.”)

27 Krock, supra note 25.

28 Id.


Electronic copy available at:

unnecessary. By 1966—two years after the Civil Rights Act was passed by Congress—the
case made it to the Illinois Supreme Court. There, Motorola’s position was completely

But in early 1964, the Myart case was a real threat to Title VII’s passage. On the
floor of Congress, Rep. John Ashbrook (R-Ohio) cautioned that if Myart’s logic were to
prevail, “Merit will not count, only the likes and dislikes, the whims of those in authority .
. . .”30 Title VII’s most vociferous critics had a field day. Senator Strom Thurmond (D-SC)
praised an editorial in the Charleston Evening Post ridiculing the Myart case. The
editorial argued that it “could be” that the Myart arguments would be advanced in
connection with Title VII, if passed.31 Senator Sam Ervin (D-NC) declared that he “did not
understand” a provision of Title VII until he “read in the newspapers” about the Myart
case, implying that now he understood it all too well.32

Significantly, not a single member of Congress rose to defend Myart. No one

suggested that job qualifications that have a disparate impact could violate Title VII
regardless of whether the employer intends the racial effect. Everyone appeared to agree
that Myart was unsound public policy.

Case and Clark fully understood the importance of disavowing the hearing
examiner’s decision (and hence of the need for the statement in the Case & Clark
memorandum). Failure to convince their more hesitant colleagues that Title VII would
not allow for the hearing examiner’s decision in Myart could easily have torpedoed Title
VII. They therefore needed to accomplish two things. First, they had to convince their
colleagues that Title VII did not allow for Myart-style liability. Equally important, they
had to convince them that the EEOC would not have the power to twist the meaning on
Title VII after the law is passed.

On April 8, 1964, Senator Case discussed the Myart case specifically on the floor of
the Senate:

29 Motorola, Inc. v. Ill. Fair Employment Comm’n, 34 Ill.2d 266, 215 N.E.2d 286 (1966).

30 110 C ONG . R EC . 5311 (1964) (remarks of Rep. Ashbrook).

31110 C ONG . R EC . A1578-79 (1964) (extension of remarks of Sen. Thurmond). Later that year,
Thurmond switched his political affiliation to Republican. See also The Advantaged Need Not Apply,
C HARLESTON E VENING P OST , March 26, 1964, reprinted in 110 C ONG . R EC . A1578-79 (1964) (citing
the Myart case and stating “It if [sic] could be shown in a given case that an aptitude test, though
nondiscriminatory on its face, had, in fact, acted to deprive Negroes of employment, could it not then
be said that such a test ‘tended to deprive’ Negroes of equal employment? We think it could be said,
and we submit that, as surely as the sun will rise tomorrow, it would be said.”).

32 110 C ONG . R EC . 5614 (1964) (remarks of Sen. Ervin).


Electronic copy available at:

MR. CASE. Much has . . . been made of the case of Myart against
Motorola, Inc. It has been repeatedly and erroneously cited as an
example of the lengths to which the [proposed EEOC] might be
expected to go.

The decision is merely an initial or preliminary decision by a part-

time hearing examiner. The Illinois commission, according to its
chairman, “has not taken any stand of any kind . . . .”


Whatever the final action on the case, the citation of the examiner’s
finding has no application to title VII. First, as the subsections of
section 707 . . . make clear, the [EEOC], . . . unlike the Illinois
commission, would have no adjudicative functions. Only a Federal
court would have the authority to determine whether there had been
a violation of the act, and only the court could enforce compliance.
Second, under title VII, even a Federal court could not order an
employer to lower or change job qualifications simply because
proportionately fewer Negroes than white[s] are able to meet them.33

Case went on to say:

Title VII says only that covered employers cannot refuse to hire
someone simply because of his color. . . . [I]t expressly protects the
employer’s right to insist that any applicant meet the applicable job

He couldn’t have been clearer. But just in case, he also introduced a legal
memorandum on the Myart case that reiterated these points.35 And it is
again noteworthy that Case was not just any Senator. He was the
Republican floor leader for Title VII. He thus spoke authoritatively.

33 110 C ONG . R EC . 7246 (1964) (remarks of Sen. Case).

34 Id. (emphasis added).

35Id. at 7246. The memorandum, captioned “Myart v. Motorola, Inc.,” stated: “Of course, it is not
appropriate to comment here on whether the Motorola decision is correct as a matter of Illinois law.
This is for the State commission and the State courts to determine. It is enough to note that the
result seems questionable. There is no doubt, however, that such a result would be unmistakably
improper under the proposed Federal law. The Illinois case is based on the apparent premise that
the State law is designed to provide equal opportunity to Negroes, whether or not as well qualified as
white job applicants.”


Electronic copy available at:

But even his express disavowal of Myart was not enough for some of his
worried colleagues. If Title VII ended up being interpreted consistently with the
hearing examiner’s decision in Myart, the effect would be huge. It would, as Krock
suggested, put the federal government in control of every aspect of the employment
relationship.36 In order to ensure that Title VII would not wander into Myart
territory, key Senate Republicans—the swing voters who could make the difference
between successfully securing the bill’s passage and failure—demanded more than
just reassurances from Clark and Case. They wanted modifications to the bill’s

Senate Minority Leader Everett Dirksen (R-Ill.) took the lead in negotiating
those changes.38 Among the changes he required were two related directly to the
argument for disparate impact liability. The first was the addition of the word
“intentionally” to Section 706(g)(1), governing when relief may be granted. As
amended, the section would read: “If the court finds that the respondent has
intentionally engaged in or is intentionally engaging in an unlawful employment
practice charged in the complaint, the court may enjoin the respondent [or invoke
certain other limited remedies] ….”39

Inserting the word “intentionally” into Section 706(g)(1)--might seem to be

unnecessary, in the sense that Title VII’s text already made this clear. Indeed,
that is what then-Majority Whip Hubert Humphrey (D-Minn.) said about it in
explaining the effect of Dirksen’s amendment to his Senate colleagues:

Section 706(g) is amended to require a showing of

intentional violation of the title in order to obtain relief. This is a
clarifying change. Since the title bars only discrimination because of
race, color, religion, sex, or natural [sic] origin, it would seem to
already require intent, and, thus, the proposed change does not involve
any substantive change in the title. The express requirement of intent
is designed to make it wholly clear that inadvertent or accidential [sic]
discriminations will not violate the title or result in entry of court

36 See supra note 25 and accompanying text.

37See Daniel Rodriguez & Barry R. Weingast, The Positive Political Theory of Legislative History:
New Perspectives on the 1964 Civil Rights Act and Its Interpretation, 151 U. P ENN . L. R EV . 1417,
1470-73, 1474-98 (2003).

38For a discussion of the conflicting views on whether Dirksen’s role was important in the passage of
the 1964 Act, see Rodriguez & Weingast, supra note 37, at 1474-96 (calling the competing views the
“the innocuous Dirksen thesis” and the indispensable Dirksen thesis”).

39 Civil Rights Act of 1964, Pub. L. 88-352, §706(g), 78 Stat. 261 (codified at 42 U.S.C. § 2000e-5(g)(1))
(italics added).


Electronic copy available at:

orders. It means simply that the respondent must have intended to

To Humphrey, this might have been overkill, but he was quite willing to go along
with it.

But making that “wholly clear” was obviously important to securing the bill’s
passage. The stakes were high and Myart had frayed a lot of nerves. Dirksen
needed to ensure that the Myart hearing examiner’s views on the Illinois Fair
Employment Practices Act would not become the interpretative model for Title VII.
Humphrey thought Dirksen had accomplished that goal. Dirksen must have
thought so too.

Dirksen’s second amendment pertaining to the Myart fiasco was the deletion
of a major portion of the grandiloquent statement of Title VII’s purpose. The
deletion would have read: “The Congress hereby declares that the opportunity for
employment without discrimination of the types described in sections 704 and 705
is a right of all persons within the jurisdiction of the United States, and that it is
the national policy to protect the right of the individual to be free from such

This amendment may have seemed unimportant. Soaring prose at the

introduction of a statute is not ordinarily supposed to have operative value of its
own (which is why legislators have historically felt free to use it in what is
otherwise supposed to be a sober and precise legal document). But viewing
Dirksen’s amendment as unimportant may be a mistake. As Dean Daniel B.
Rodriguez and Professor Barry R. Weingast have written:

In light of what would become of civil rights in the first twenty years
after the enactment of Title VII, it is important not to understate the
significance of this deletion. Much of the thrust of courts’ expansive
readings of the civil rights laws in the 1960s and 1970s rested on
views concerning the proper interpretive scope to be given to vague
language in the legislation . . . . [I]n a series of important decisions
courts used just such language to rationalize expansionary readings
and to ignore specific limitations contained in the Act. But Dirksen
and his colleagues had no hindsight; they had the bill qua bill before
them. To the extent that Dirksen rationally feared that the broad
phrasing of section 701(a) would authorize courts to expand the scope

40 110 C ONG . R EC . 12,723-24 (1964) (remarks of Sen. Humphrey).

41 110 C ONG . R EC . 12,811 (1964).


Electronic copy available at:

of the Act, his intent in deleting the provision seems rather

Rodriguez & Weingast were suggesting that Dirksen may have anticipated the rise
of the “purposive” school of thought in statutory interpretation, the heyday of which
was to come in the next two decades.

Those weren’t the only amendments Dirksen sought. They are simply the
ones that relate most directly to the argument for disparate impact liability.43 Some
of the others relate to the issue indirectly, because they were aimed at ensuring the
EEOC would not have the power to push the application of Title VII beyond what
its Congressional supporters had intended. The House version of the bill had
already denied the EEOC adjudicatory powers akin to those enjoyed by the National
Labor Relations Board.44 To that Dirksen sought to add:
42  Rodriguez & Weingast, supra note 37, at 1492-93.

43For those who feared Title VII would lead to mandates for proportional hiring, the following
section was added as Section 703(j):

[N]othing contained in this title shall be interpreted to require any employer . . . to

grant preferential treatment to any individual or to any group because of the race,
color, religion, sex, or national origin of such individual or group on account of an
imbalance which may exist with respect to the total number or percentage of
persons of any race, color, religion, sex, or national origin employed by any employer
. . . .”

Civil Rights Act of 1964, § 703(j) 78 Stat. at (codified at 42 U.S.C. § 2000e-2(j)).

With regard to this amendment, Senator Humphrey explained:

A new subsection 703(j) is added to deal with the problem of racial balance among
employees. The proponents of this bill have carefully stated on numerous occasions
that title VII does not require an employer to achieve any sort of racial balance in his
work force by giving preferential treatment to any individual or group. Since doubts
have persisted, subsection (j) is added to state this point expressly. This subsection
does not represent any change in the substance of the title. It does state clearly and
accurately what we have maintained all along about the bill’s intent and meaning.”

110 C ONG . R EC . 12,723 (1964). United Steelworkers v. Weber, 443 U.S. 193 (1979), turned this
section upside down by holding that it did not apply to Kaiser Aluminum’s set aside of 50% of all
craft trainee positions for African Americans, a program adopted by Kaiser under pressure from the
Department of Labor’s Office of Federal Contract Compliance Programs.

44An early House version of the bill had already been modified to avoid giving the EEOC the power
to issue “cease and desist” orders along the lines of the National Labor Relations Board. For many
on Capitol Hill, the NLRB was a precedent they wanted to avoid. Twenty-five years earlier, the
American Federation of Labor had requested Congress to investigate the NLRB on the ground that it
heavily favored the more radical Congress of Industrial Organizations over the trade unions of the
American Federation of Labor. Rightly or wrongly, during the course of investigation the Special


Electronic copy available at:

(1) The denial of any substantive rule-making authority;45 and

(2) The denial of the authority to bring lawsuits against employers who have
engaged in a “pattern or practice” of discrimination.

Dirksen got his way on both these amendments, too. The Attorney General
and hence ultimately the Department of Justice’s Civil Rights Division, which were
thought to be more politically accountable to the President than the “independent”
EEOC, were instead assigned the responsibility for bringing “pattern or practice”
lawsuits and those lawsuits were limited to those in which it could be established
that the employer had engaged in a pattern or practice of discrimination.46

One additional amendment that made it into the final bill was the so-called
“Tower Amendment”47 proposed by a vocal opponent of Title VII, John Tower (D-
Tex.). Tower’s amendment was aimed narrowly at the Myart case’s use of tests.
His original draft of the amendment could have been interpreted to mean that the
use of professionally-developed workplace tests are always legal. But critics pointed
out that this language would authorize even tests that are being given precisely
because of their anticipated discriminatory effect rather than in spite of it.48 Tower
therefore amended the language to read:
Committee to Investigate the National Labor Relations Board (popularly known as the “Smith
Committee”) accusations of communist influence were leveled against the Board. Resignations
followed. Nevertheless, the notion that the NLRB was staffed, not just by those with a pro-
labor/anti-management bias, but by those with a bias toward the more confrontational tactics of
industrial unions, took a long time to fade. The passage of Title VII as part of the Civil Rights Act of
1964 was not a foregone conclusion, especially in the Senate. Modeling the EEOC after the NLRB
would have been a non-starter. See H.R. 405, 88th Cong., 1st Sess. §§ 10-11 (1963); H.R. Rep. No. 570,
88th Cong., 1st Sess. 10-11. (1963). See also Civil Rights: Private Cause of Action Exists Under Title
VII Notwithstanding EEOC Determination of No Reasonable Cause, 1971 D UKE L.J. 467, 468-69
(1971) (describing process by which cease-and-desist power was stripped from bill).

45See §713(a), 78 Stat. at 265 (codified at 42 U.S.C. § 2000e-12 (granting rulemaking authority on
procedural issues).

46§707, 78 Stat. at 261-2). See also Rodriguez & Weingast, supra note 37, at 1489. This section of
the 1964 Act and other sections were amended by Equal Employment Opportunity Act of 1972, Pub.
L. No. 92-261, §5, 86 Stat. 103, 107 (1972) (codified at 42 U.S.C. § 2000e et seq.).
47 See Rodriguez & Weingast, supra note 37, at 1504-17.

48Both Senator Case and Senator Humphrey commented that any amendment of this type is
unnecessary. See 110 CONG. REC. 13,503-04 (1964) (remarks of Sen. Case). Senator Humphrey
stated: “Every concern of which this amendment seeks to take cognizance has already been taken
care in title VII . . . . These tests are legal. They do not need to be legalized a second time. They
are legal unless used for the purpose of discrimination. The amendment is unnecessary.” Id. at
13,504 (remarks of Sen. Humphrey).


Electronic copy available at:

Notwithstanding any other provision of this subchapter, it shall not be an
unlawful employment practice for an employer . . . to give and to act upon the
results of any professionally developed ability test provided that such test, its
administration or action upon the results is not designed, intended or used to
discriminate because of race, color, religion, sex or national origin.49

As modified, the amendment passed (although Tower nevertheless voted

against the final bill).50

All seemed to be going well for the passage of the bill. With these and other
amendments, the Civil Rights Act of 1964, including Title VII, passed the Senate on
June 19, 1964 by a vote of 73 to 27, more than enough to get past the Senate
filibuster. The Senate version of the bill was then adopted in the House of
Representatives on July 2, 1964 by a vote of 289-126. In both cases, the “yea” votes
were disproportionately Republican, further highlighting the fact that Dirksen’s
interventions were crucial. President Lyndon Baines Johnson signed the bill into
law that same day.51

By then, Myart had become a story of local interest only. The Chicago
Tribune reported that a state senator was calling for a legislative inquiry into the
case on the ground that it has “cast a cloud” over the business climate in the state.52
The implication was that Myart-style thinking held no sway outside of Illinois and
thus put Illinois at a competitive disadvantage. Given the text of Title VII and its
very explicit legislative history, this would have seemed obviously true at the
federal level. Business Week reported matter-of-factly that the recently-enacted

Because a vote against Tower’s first draft could be construed as an endorsement of the hearing
examiner’s position in Myart, some members felt obligated to dispel any such notion. Senator Case
stated, “I feel certain that no Member of the Senate disagrees with the views of [Senator Tower]
concerning the Motorola case.” 110 C ONG . R EC . 13,503 (remarks of Sen. Case). He added, “I want it
to be clearly understood, so far as I am concerned—and I believe I speak for all members of the
committee, the captains, and the leadership—that our position against [the first draft of the Tower
Amendment] and the vote we shall cast against it do not mean approval of the Motorola case or that
the bill embodies anything like the action taken by the examiner in that case.” Id. at 13,504.

49 42 U.S.C. § 2000e-2(h).

50 Rodriguez & Weingast, supra note37, at 1508.

51See Civil Rights Act of 1964, WIKIPEDIA.COM,, []. On the
Senate side, 82% of Republicans and 69% of Democrats voted in favor. One the House side, 80% of
Republicans and 63% of Democrats voted in favor.

52 Asks Illinois Legislative Probe of FEPC, C HIC . T RIB ., November 21, 1964 at 3.


Electronic copy available at:

Title VII would not permit the controversial result before the hearing examiner in

Dirksen’s mission had been accomplished. Or so everyone seemed to think.54

III. Disparate Impact Liability Was the Brain Child of the EEOC,
which Turned Out Not to Be a Toothless Tiger After All.

No one publicly objected to Dirksen’s insertion of the word “intentional” into

Section 706(g)(1). It was uncontroversial at the time. But some advocates were
infuriated by Dirksen’s reconfiguration of the EEOC to deprive it of substantive
rulemaking and litigation authority. One scholar famously called it “a poor
enfeebled thing.”55 It was commonplace to refer to the EEOC as a “toothless tiger”
until it received the power to pursue pattern or practice lawsuits in the name of the
United States against private employers in 1972.56 Indeed, the EEOC’s own web
site acknowledges exactly that: “Because of its lack of enforcement powers, most
civil rights groups viewed the Commission as a ‘toothless tiger.’”57

But the plain truth was that without these and other modifications to the bill
required by Dirksen, the Civil Rights Act of 1964 would not have passed.58 Like
most major legislation, it was the product of compromise.

53See Hiring Tests Wait for the Score: Case Involving Motorola’s Employment Test Raises Issue of
Whether Employers Can Use Screening Devices That Might Discriminate Against “Deprived” Persons,
B US . W EEK , Feb. 13, 1965 at 45 (reporting that Title VII would not permit that result).

54For a further look at the legislative history of Title VII, see Michael Evan Gold, Griggs’ Folly: An
Essay on the Theory, Problems, and Origin of the Adverse Impact Definition of Employment
Discrimination and a Recommendation for Reform, 7 I NDUS . R EL . L.J. 429 (1985) (Gold
“demonstrates that Congress did not intend to mandate adoption of adverse impact theory when it
established Title VII.”).


06 (1966).
56Equal Employment Opportunity Act of 1972, Pub. L. No. 92-261, 86 Stat. 103 (1972) (codified at 42
U.S.C. § 2000e et seq.).

57EEOC 35th Anniversary: 1965-1971: A “Toothless Tiger” Helps Shape the Law and Educate the
Public, EEOC.GOV,

58Among the other amendments pressed by Senator Dirksen and acquiesced in by Congress were (1)
a restriction requiring complaints to attempt to exhaust their remedies before state fair employment
practices commissions before they file a complaint with the EEOC; (2) an exemption for employers
whose employees work less than twenty weeks per year (largely designed for agricultural


Electronic copy available at:

What authority did the EEOC initially have? Not a lot. As originally
constituted by Title VII, the EEOC’s job was to act as investigator and mediator
when complaints are filed. It wasn’t glamorous work. Its first chairman, Franklin
Delano Roosevelt, Jr., a hard-drinking former Congressman and son of the former
President, appeared to be bored with the job. He spent a great deal of his time on
his yacht, causing one EEOC staff member to compose a ditty, “Franklin’s Away” (to
the tune of “Anchors Aweigh”), which could be sung during his absences.59
Roosevelt left the EEOC after only a year.

Nevertheless, there was plenty of this unglamorous work to do. While almost
half of the states already had fair employment practices laws in place when Title
VII was passed, the rest did not, including several where employers were expected
to be especially recalcitrant. So, just as Congress intended, the EEOC’s then-small
staff proceeded case by case—at least at first.

One thing that is inevitably true with major legislation, however, is that it
will raise questions in the minds of those who are attempting to comply with it.
Some of those questions will be susceptible to easy answers, some will not be. One
can imagine the kinds of questions that EEOC employees might have been asked by
worried employers:

Employer’s Question: What if it’s not me who wants to discriminate?

What if it’s my customers who are insisting that I send only white
electricians into their homes? Does that take me outside the statute?
EEOC Answer: No, it doesn’t. You still have to follow the law.

Employer’s Question: I own three, separately-incorporated franchises

of a national chain of hardware stores. None of them employs more
than 40 employees, but altogether they employ 107. The national and
its combined franchises employ well over 12,000. From time to time, I
assign one or more employees from one franchise to another when
necessary to make things run smoothly. I read in the newspaper that
the new law only covers businesses that employ at least a certain
number of employees. It’s not like I want to discriminate, but I need to
employers); and (3) the deletion of a provision that would have allowed outside organizations like the
NAACP to sue on behalf of an employee or applicant for employment. Most of Dirksen’s changes can
fairly be described as weakening the bill. But one amendment—the expansion of the definition of
“employment agency” to include union hiring halls —significantly strengthened the bill. See Daniel
Rodriguez & Barry R. Weingast, The Positive Political Theory of Legislative History: New
Perspectives on the 1964 Civil Rights Act and Its Interpretation, 151 U. P ENN . L. R EV . 1417, 1490

59 B LUMROSEN , supra note 16, at 55.  



Electronic copy available at:

know: Are my franchises covered? EEOC Answer: Let me check with
our General Counsel’s office and get back to you. We may need some
more information from you about how your franchises operate before we
can answer that.

It is hardly surprising that EEOC officials would undertake to publish

answers to the questions they were hearing repeatedly.60 Nor is it surprising that
employers as well as employees and job applicants would appreciate those
interpretative “guidances.” This was a niche that one way or another was going to
be filled. Employers have every incentive to want to stay on the right side of the
law. Free advice from the EEOC on how to do so is a good thing from their

But publishing such “guidances” also had the potential to spin out of control.
The temptation would always be to use them to establish what the EEOC staff
wanted the law to be rather than what it was. Instead of interpreting Title VII in
good faith, guidances would soon become quasi-legislation—disguised as
interpretation, but in reality imposing new duties on employers not found in Title
VII itself.61

None of this should be surprising. It is in the nature of bureaucracy. It

naturally seeks to expand its powers, often beginning by occupying niches that are
otherwise unoccupied. Over time, a little power often becomes a lot of power. What
is surprising is how upfront EEOC officials were about their tactics in accumulating
that power.
Alfred Blumrosen, the EEOC’s first “Chief of Conciliations” and disparate
impact liability’s primary architect, was unabashed in describing the extent to
which the EEOC was (and in his view should be) aggressive in its interpretation of
Title VII: “Creative administration converted a powerless agency operating under
an apparently weak statute into a major force for the elimination of employment
discrimination,” he reflected with obvious pride in 1971. 62 A rough translation
might be this: Congress intentionally limited our agency’s powers, but we managed
to come up with creative ways around the limitations.
Blumrosen was a Progressive’s Progressive with little affinity for the
concept of limited government. He complained that law schools spend too much time
teaching their students about the importance of limited government, the rule of law,
and the consequent need to keep a sharp eye on government officials. As a
60 See B LUMROSEN , supra note 16, at 52.

61 See infra 87-88 and note 248.

62 BLUMROSEN, supra note 16, at 53.


Electronic copy available at:

government official himself, he wanted more emphasis on how government officials
can be made more powerful:
[Legal education] rarely deals with the affirmative aspects of
administration. Rather, the law schools provide elaborate intellectual
equipment to restrict the efforts of administrators. Constitutional law and
administrative law courses are still largely concerned with what
government may not do, rather than with how it should decide what it
may do.63
Blumrosen characterized this aspect of legal education—one that is fundamental to
the rule of law—as “negativism.” Alas, his words betray an unfortunate eagerness
for EEOC officials like himself to define the limits of their own authority.
Students impatient with the negativism of present legal education would
be better equipped as lawyers if they would focus sharply on the question
of “how we can best fulfill the purposes which brought our agency into
being” rather than on the question of “whether the courts will sustain this
course of action.”64
I have a measure of sympathy for Blumrosen. He was part of the generation
of civil rights policymakers profoundly influenced by the turbulence of the late
1960s65—something that has been easy to lose sight since then. Perhaps he believed
the nation’s immediate survival was at stake. He repeatedly pushed the EEOC to
interpret Title VII with an eye toward effectuating what he perceived as its broad
purpose—improving African-American employment as quickly as possible—rather
than with an eye towards what the statute actually said or what the courts would
be likely to uphold. Perhaps his immediate fears put him beyond caring about the
rule of law. While I believe this to be short-sighted, history is full of that sort of
short-sightedness. Every generation has had to try to dig itself out of the things the
previous generation did in a panic.
Blumrosen took the position that Title VII authorizes the EEOC to issue “fair
qualification standards.”66 He argued that there should be three relevant inquiries:
(1) Does the standard … exclude a higher proportion of minority group
members … ? (2) If so, is this exclusion justified by business necessity … ?

63 Id. at 53 (emphasis in original).
64  Id.

65  See id. at 3.
66  Id. at 255-69.


Electronic copy available at:

(3) Is the employer entitled to require this capability in his applicants
without providing training programs … ?67
Thus was born disparate impact liability. Blumrosen’s sense of urgency led
him to an approach that was fundamentally both anti-democratic and lawless—
something he implicitly conceded when he wrote that whether courts would sustain
a particular course of action was of lesser importance to him.
Writing in 1968 for the NAACP’s The Crisis magazine, EEOC Commissioner
Samuel Jackson also implicitly conceded that disparate impact liability was the
product of EEOC creativity, rather than a sober read of Title VII’s actual text or
legislative history:

“[The] EEOC has taken its interpretation of Title VII a step further
than other agencies have taken their statutes. It has reasoned that in
addition to discrimination in employment, it is also an unlawful
practice to fail or refuse to hire, to discharge or to compensate
unevenly … on criteria which prove to have a demonstrable racial
effect without a clear and convincing business motive.”68

Note Jackson’s admission that the new approach was “in addition to
discrimination in employment.” Yet discrimination is what Title VII bans, not job
qualifications that “prove to have a demonstrable racial effect.”

Jackson attempted to justify the EEOC’s departure from Title VII’s intent
requirement by asserting that “Congress, with its elaborate exploration of the
economic plight of the minority worker, sought to establish a comprehensive
instrument with which to adjust the needless employment hardships resulting from
the arbitrary operation of personnel practices, as well as purposeful
discrimination.”69 It’s not clear where he got that impression. The 1964 statute
67  Id. at 255.

68  Samuel C. Jackson, EEOC vs. Discrimination, Inc, T HE C RISIS 16-17 (January 1968). Jackson is

described by Alfred Blumrosen this way:

Samuel Jackson … was a leader of the left wing of the NAACP and a
Republican. He … understood how the administrative process could be shaped
to achieve social results, and was busy from the first day attempting to influence
and control the direction of the Commission. His efforts brought him into
conflict with every chairman and staff director. He did good work in shaping the
law, in insisting on procedures wh[i]ch protected complainants, and in
conducting hearings on discrimination in the textile industry in the Carolinas.

Blumrosen, supra note 16, at 56.  

69 Samuel C. Jackson, EEOC vs. Discrimination, Inc., T HE C RISIS 16-17 (January 1968).


Electronic copy available at:

was no more “a comprehensive instrument with which to adjust … needless
employment hardships” than the No Child Left Behind Act70 is a comprehensive
instrument by which the federal bureaucracy is given authority to “adjust
circumstances to make education better” or the Patriot Act71 is a comprehensive
instrument by which the military is authorized to “adjust circumstances to ensure
national security.” Nor should anyone want them to be.

Some members of the 88th Congress may indeed have been concerned about
the “economic plight of the minority worker.” But what they did about it was pass
the Civil Rights Act of 1964, including Title VII, an anti-discrimination law. It did
not authorize the EEOC to do whatever it thought best to relieve that plight.
Suggesting otherwise runs the risk of teaching legislators not to inform themselves
on broad problems out of fear that future agency officials will view that as broad
authority to do what they think necessary to remedy the problem, not what the
legislators actually agreed by majority vote to do.

Moreover, when an administrative agency appeals to what it views as

Congress’s higher purpose rather than to the actual texts that a majority of
Members of Congress actually agreed on, it isn’t simply pushing aside what those
legislators, through the hard work of compromise, actually agreed to. It is making
future legislation less likely. If legislators can’t trust executive branch officials to
enforce the law as actually written, they would be fools to keep passing legislation
arrived at by compromise.72

The EEOC’s decision to ignore Title VII was intentional. As historian Hugh
Davis Graham wrote concerning this period in its history:

The EEOC legal staff was aware from the beginning that a normal,
traditional, and literal interpretation of Title VII could blunt their efforts
[based on disparate impact theory] against employers who used either
professionally developed tests or bona fide seniority systems. The EEOC’s
own official history of these early years records with unusual candor the
commission’s fundamental disagreement with its founding charter, especially
Title VII’s literal requirement that the discrimination be intentional.73
70No Child Left Behind Act of 2001, Pub. L. No. 107-110, 115 Stat. 1425 (2001) (codified in scattered
sections of Titles 15, 20, 42 & 47 of U.S.C.).

71 USA PATRIOT Act of 2001, Pub. L. No. 107-56, 115 Stat. 272 (2001) (codified in scattered sections
of Titles 8, 12, 15, 18, 20, 31, 42, 47, 49, 50 & 51 of U.S.C.).
72 See Rodriguez & Weingast, supra note 37, at 1417.

H UGH D AVIS G RAHAM , T HE C IVIL R IGHTS E RA 248-49 (Oxford Univ. Press 1990) (citing EEOC

Administrative History).


Electronic copy available at:

IV. Disparate Impact Liability’s Incoherence: Everything Has a
Disparate Impact on Some Group That Is Protected by Title VII.

The EEOC’s decision to ignore Congress’s intent is troubling. But at least as

troubling is that fact that disparate impact liability is incoherent. All job
qualifications have a disparate impact. It is no exaggeration to state that there is
always some group covered by Title VII that will do comparatively poorly with any
particular job qualification. It won’t always be the same group, but Title VII
protects a vast array of racial, national origin, religious and color groups as well as
men and women, so it will always be some group.74
Consider: On average, men are stronger than women, while women are
generally more capable of fine handiwork.75 Chinese Americans and Korean
Americans score higher on standardized math tests and other measures of
mathematical ability than most other national origin groups. They are also more
likely to hold a B.S., B.Eng., M.S. or Ph.D. in one of the hard sciences or
There is a lot more. South Asian Indian Americans are disproportionately
likely to have experience in motel management than the rest of the population.77
African Americans are statistically over-represented in many professional athletics
as well as in many areas of the entertainment industry.78 African American college
74  Brief of Gail Heriot & Peter Kirsanow as Amici Curiae Supporting Petitioners, Texas Dep’t of
Hous. & Cmty. Affairs v. Inclusive Cmtys. Project, Inc., 135 S. Ct. 2507 (2015) (No 13-1371), at 19–
21, 2014 WL 6660913.
75  A.E.J.
Miller et al., Gender Differences in Strength and Muscle Fiber Characteristics, 66 E UR . J.
/pubmed/8477683 [ that men are generally stronger than
women); Michael Peters et al., Marked Sex Differences on a Fine Motor Skill Task Disappear When
Finger Size Is Used as Covariate, 75 J. A PPLIED P SYCHOL . 87 (1990), [] (finding that women
performed significantly better than men on a fine motor skill test).
76  ChuanshengChen & Harold Stevenson, Motivation and Mathematics Achievement: A Comparative
Study of Asian-American, Caucasian-American, and East Asian High School Students, 66 C HILD
D EV . 1215 (1995), 2.x
[] (finding that Asian Americans outperformed whites on a standard
mathematics exam).
A MERICAN D REAM 1 (2012) (observing that subcontinent Indian Americans own about half of all
motels in the United States).
78  Richard
Lapchick, The 2018 Racial and Gender Report Card: National Football League,


Electronic copy available at:

students earn a disproportionate share of college degrees in public administration
and social services. Asian Americans are less likely to have majored in
Psychology.79 Unitarians are more likely to have college degrees than Baptists.80
And it doesn’t stop. Native Americans are less likely to have access to high-
speed internet service. They are hence less likely to learn of jobs that are posted
only on the internet and less likely to be able to comply with requirements that job
applications be submitted through the employer’s web site.81 Such disparities can
be unstable. In the past, African Americans were also less likely to have internet
access, but, for many African-American sub-groups, this does not seem to be the
case now.82
Some of the disparities are surprising. Cambodian Americans are
disproportionately likely to own or work for doughnut shops and hence are more
likely to have experience in that industry when it is called for by an employer.83 The
reasons behind other disparities may be more obvious: Non-Muslims are more

[](noting that over 69% of National Football League athletes are
African American).
See Peter L. Hinrichs, Racial and Ethnic Differences in College Major Choice, FED. RES. BANK OF
CLEVELAND, (March 31, 2015),
differences-in-college-major-choice.aspx [].

80 Compare Unitarians, P EW R ESEARCH C TR .,
landscape-study/religious-denomination/unitarian/ [] (finding that 67%
of Unitarians have completed a college degree) with Baptists in the Mainline Tradition, P EW
mainline-trad/ [] (finding that 13% of Baptists in the mainline tradition
have completed college).

81 Hansi Lo Wang, Native Americans On Tribal Land Are 'The Least Connected' To High-Speed
Internet, NPR.COM (Dec. 6, 2018)
tribal-land-are-the-least-connected-to-high-speed-internet [].

82See Aaron Smith, African Americans and Technology Use, P EW R ESEARCH C TR . (Jan. 6, 2014),
[] (“[Y]oung, college-educated, and higher-income African Americans
are just as likely as their white counterparts to use the internet and to have broadband service at
home. Some 86% of African Americans ages 18-29 are home broadband adopters, as are 88% of black
college graduates and 91% of African Americans with an annual household income of $75,000 or
more per year. These figures are all well above the national average for broadband adoption, and are
identical to whites of similar ages, incomes, and education levels.”).

83See Seth Mydans, Long Beach Journal: From Cambodia to Doughnut Shops, N.Y. T IMES , May 26,
1995, at 10.


Electronic copy available at:

likely than Muslims to have an interest in wine and hence develop qualifications
necessary to get a job in the winemaking industry.84
Residency requirements for municipal jobs in Washington, D.C. may
disadvantage Asian Americans, since they disproportionately live in nearby
Arlington County, Virginia. Residency requirements for county jobs in Arlington
County may disproportionately disadvantage African Americans, who are more
likely to live in Washington or in nearby Prince George’s County, Maryland.85
One might think that somewhere, somehow there would be a real-life job
qualification without disparate impact on any particular group. But if there is—
apart from sheer random selection—I have not found it. For a while, right-
handedness (or left-handedness) struck me as a possibility. But it turns out that
some cultures (e.g. Chinese) around the world tend to regard left-handedness as an
undesirable trait, and many parents try to train their children out of it.86
Consequently, those cultures have proportionally more right-handed adults than
others. As a group, they will therefore be advantaged when employers require a
right-handed worker and disadvantaged when employers want a left-handed
I have in the past promised to pay $10,000 to the favorite charity of anyone
who can bring to my attention a job qualification that has made a difference in a
real case and has no disparate impact on any race, color, religion, sex, or national
origin group. So far I have not had to pay.
Note also that disparate impact liability applies to standards for promotions,
terminations and every other kind of employment decision.87 Moreover, it applies to
subjective as well as objective job qualifications.88 So if an employer only hires
84Laurence Michalik et al., Religion and Alcohol in the U.S. National Alcohol Survey: How
Important Is Religion for Abstention and Drinking?, 87 D RUG & A LCOHOL D EPENDENCE 268, 275
[] (finding relatively high levels of abstention from alcohol among

[]. See also NAACP v. North Hudson Regional Fire & Rescue, 665 F.3d
464, 477 (3d Cir. 2011) (disparate impact liability for residency requirements for fire and rescue
86 Howard I. Kushner, Why Are There (Almost) No Left-Handers in China?,37 E NDEAVOR 71 (2013).

87 See, e.g., George v. Farmers Elec. Coop., Inc., 715 F.2d 175 (5th Cir. 1983) (termination); Wilmore
v. Wilmington, 699 F.2d 667 (3d Cir. 1983) (promotion).

88 Watson v. Fort Worth Bank & Tr., 487 U.S. 977, 990 (1988).


Electronic copy available at:

employees with three strong letters of recommendation from a previous employer or
teacher, that, too, likely will have a disparate impact on national origin groups
whose members are disproportionately recent immigrants. Contacting previous
employers or teachers from their nation of origin may not be feasible. Alternatively,
an employer who likes to hire employees with the same taste in television and
movies as she has will certainly find this has a disparate impact.89
What is the upshot of the ubiquity of disparate impact? Since all job
qualifications have a disparate impact on some group, all employer decisions are
subject to second-guessing by the EEOC (or by the courts in the case of a private
lawsuit).90 The same goes for standards for promotions, terminations, and task
assignment. It goes for everything. Employers must be prepared to show that their
actions are driven by “business necessity,” which, as this article will discuss infra at
Parts VI-VII, is not an easy task.
This is a profound change in the American workplace—and indeed in American
culture. Americans like to believe we are living in a comparatively free country.
We like to believe the rule of law prevails. That should mean that private citizens
are not required to justify every action (or be prepared to justify every action) to a
government official. It is not clear, however, how that belief can be squared with an
independent government agency that has made everything an employer might do
presumptively illegal. Disparate impact liability (in conjunction with other changes
in employment law and policy) has led to more bureaucratized employment
practices. The “personnel offices” of the mid-20th century have given way to the
much larger “human resources offices” of the present.
Supporters of disparate impact liability downplay the level of discretion
enjoyed by the EEOC as a result of disparate impact liability. Sometimes their
argument is that disparate impact’s ubiquity is not a problem, because the EEOC
has agreed to abide by a “four-fifths rule.” 91 That rule was first announced in the
Trends.pdf [].
See Watson, 487 U.S. at 990 (recognizing that disparate impact liability applies to subjective as

well as objective job qualifications).

91The Uniform Guidelines on Employee Selection Procedures section on the “four-fifths rule” reads
as follows:

D. Adverse impact and the "four-fifths rule."

A selection rate for any race, sex, or ethnic group which is less than four-fifths (4/5)
(or eighty percent) of the rate for the group with the highest rate will generally be
regarded by the Federal enforcement agencies as evidence of adverse impact, while
a greater than four-fifths rate will generally not be regarded by Federal enforcement
agencies as evidence of adverse impact. Smaller differences in selection rate may


Electronic copy available at:

Uniform Guidelines on Employee Selection Procedures (which were published by
the Civil Service Commission, the Department of Labor, the Department of Justice
and the EEOC in 1978). Under it, if a particular job qualification leads to a
“selection rate for any race, sex, or ethnic group” that is “greater than four-fifths” of
the “rate for the group with the highest rate” it will not ordinarily be challenged by
federal enforcement agencies for having a disparate impact (although it may still
form the basis of a private lawsuit).
This number was, of course, pulled out of a hat. There is nothing in Title VII
about a “four-fifths rule.” Private litigants and courts are not bound by it. But more
important, it doesn’t solve the problem even at the EEOC. Selection rates of less
than four-fifths relative to the group with the highest rate are extremely common.
Just as everything or nearly everything has a disparate impact, everything or
nearly everything has a selection rate that fails the “four fifths rule” for some race,
color, religion, sex, or national origin group.
Consider, for example, the horse racing industry. Of the five top-grossing
North American jockeys of 2015, all were Hispanic males.92 Height and weight
restrictions make it less likely that an African-American male will qualify.93
Moreover, while the “four-fifths” rule purports to be practical, it is of very limited
use in practice, even if disparate impacts that flunk the “four-fifths rule” were more
nevertheless constitute adverse impact, where they are significant in both statistical
and practical terms or where a user's actions have discouraged applicants
disproportionately on grounds of race, sex, or ethnic group. Greater differences in
selection rate may not constitute adverse impact where the differences are based on
small numbers and are not statistically significant, or where special recruiting or
other programs cause the pool of minority or female candidates to be atypical of the
normal pool of applicants from that group. Where the user's evidence concerning the
impact of a selection procedure indicates adverse impact but is based upon numbers
which are too small to be reliable, evidence concerning the impact of the procedure
over a longer period of time and/or evidence concerning the impact which the
selection procedure had when used in the same manner in similar circumstances
elsewhere may be considered in determining adverse impact. Where the user has
not maintained data on adverse impact as required by the documentation section of
applicable guidelines, the Federal enforcement agencies may draw an inference of
adverse impact of the selection process from the failure of the user to maintain such
data, if the user has an underutilization of a group in the job category, as compared
to the group's representation in the relevant labor market or, in the case of jobs
filled from within, the applicable work force.

29 C.F.R. § 1607.4(D).

92 Maggie McGrath, Top-Earning Jockeys, F ORBES, June 29, 2016, at 28.

ADULTS: UNITED STATES, 2007–2010 (2012),
[]. See Jockey, W IKIPEDIA .COM ,


Electronic copy available at:

rare. Prior to adopting a particular job qualification, employers usually have no
way of knowing what the selection rates will be. All they can be sure of is that the
results won’t be equal across the board, since nothing ever is. Indeed, they will not
know who will apply or which groups they will be members of until after the process
is underway. Often they won’t have information about applicants’ race, religion and
national origin even after the process has been concluded. Sometimes they still
won’t know the applicant’s sex.
Disparate impact liability’s supporters therefore have another argument. Even
though everything or nearly everything may have a disparate impact that fails the
“four-fifths rule,” the EEOC has not attempted to push disparate impact to its
logical limits. For example, while the EEOC has frequently attempted to prevent
employers from requiring a high school diploma, to my knowledge it has never
prioritized preventing employers from requiring letters of recommendation.
Instead, it has focused on written tests,94 physical strength and endurance tests,95
criminal background tests,96 high school diploma requirements,97 personal credit
histories,98 residency requirements,99 and a few others. The EEOC picks and
chooses the battles it will fight (though it has no control over which disparate
impact lawsuits a private litigant will bring).
But rather than being comforted by the EEOC’s enormous discretion, we should
be especially troubled. It essentially makes the EEOC a super-legislature. It can
choose which employment practices it wishes to outlaw and when it wishes to
outlaw them. Such discretion even gives the EEOC the ability to choose a
particular employer or industry it wishes to penalize, knowing that it almost always
can identify some employment practice engaged in by its target that has a disparate

94 29 C.F.R. § 1607.

95See, e.g., CSX Transportation to Pay $3.2 Million to Settle EEOC Disparate Impact Sex
Discrimination Case (June 13, 2018),

96See, e.g., EEOC Files Suit Against Two Employers for Use of Criminal Background Checks (June
11, 2013), [].

97 Griggs v. Duke Power Co., 401 U.S. 424 (1971)..



HAVE A NEGATIVE EFFECT ON CERTAIN APPLICANTS (specifying that requiring that employees live
within a certain geographic region may violate the


Electronic copy available at:

impact.100 (And since a single Commissioner can authorize a lawsuit, a single
Commissioner has that power.) Yet unlike legislators, EEOC officials are not
elected, and Congress went out of its way in 1964 to deny them the authority to
issue substantive rules under Title VII.
The EEOC’s influence would have been more limited if Congress hadn’t acted
in 1972 to make civil rights enforcement more efficient.101 In that year, it
transferred the power to bring pattern or practice litigation against private
employers from the Department of Justice’s Civil Rights Division to the EEOC
(though the authority to bring pattern or practice lawsuits against public employers
remains with the Civil Rights Division). It also gained the ability to bring lawsuits
under § 706 of Title VII.
This newly-gained power combined with the power to issue “disparate impact
liability” guidances has proven to be extremely potent. Indeed, this has turned out
to be one of those peculiar occasions on which what appears to be a weaker power is
really a stronger one. If it had been given substantive rulemaking power, that
power would have been subject to notice and comment procedures, including the
requirement that it explain the reasons for its action. In addition, once a rule was
promulgated, it would have been considered a final agency action within the
meaning of the Administrative Procedure Act and hence appealable to an Article III
court.102 Guidances, on the other hand, are not subject to notice and comment and
cannot be as easily brought before a court of law to determine whether they are in
fact in compliance with Title VII.103

100See, e.g., Chick-Fil-A Franchisee at Concord Commons to Pay $10,000 to Settle EEOC
Pregnancy Discrimination Suit (June 5, 2014),
14.cfm [] (not a disparate impact case).

101Equal Employment Opportunity Act of 1972, Pub. L. No. 92-261, 86 Stat. 103 (1972) (codified at
42 U.S.C. § 2000e et seq.).

102 5 U.S.C. § 704.

103The State of Texas sought and obtained partial injunctive relief in connection with the 2012
Guidance (criminal background checks). Texas v. EEOC, 933 F.3d 433 (5th Cir. 2019). The main
issues in the case were procedural—whether the 2012 Guidance is a final agency action, whether
Texas has standing to sue, whether the issue is ripe for adjudication. They bear directly on the
question of how EEOC guidances can be challenged in court.

Initially, the District Court dismissed the action, but this was reversed by the U.S. Court of Appeals
for the Fifth Circuit. See Texas v. EEOC, 827 F.3d 372 (5th Cir. 2016). Less than two months later,
the Court of Appeals, on motion for rehearing, vacated its opinion and remanded the case to the
District Court for the Northern District of Texas for reconsideration in light of the recently-decided
U.S. Army Corps of Engineers v. Hawkes Co, 136 S. Ct. 1807 (2016). After consideration, that court
issued an order granting judgment to Texas in part and to the EEOC in part and granting partial
injunctive relief. See Texas v. EEOC, 2018 U.S. Dist. LEXIS 30558 (N.D. Tex. 2018) (enjoining the


Electronic copy available at:

When guidances are combined with the ability to conduct long and costly
investigations followed by litigation against an employer, smart employers quickly
figure out how best to protect themselves: Keep your head down and follow the
guidance. Even if the employer is pretty confident the guidance is an overreach by
the agency, unless following it would be a very serious burden, it usually makes
sense to obey rather than disobey and wait to be sued. It may be years before the
employer would be challenged by the EEOC or a private litigant. In the meantime,
the potential damages that can be assessed against it would be building up. If the
employer turned out to be wrong about whether the courts would ultimately uphold
the EEOC’s guidance, it would have to pay those damages. Even if it turned out to
be right, it would still be hammered on investigation and litigation expenses.104
This is essentially the definition of arbitrary power: Everything is
presumptively illegal and the EEOC gets to decide which illegal acts it would most
like to pursue. Multiply the cost of compliance across hundreds of thousands of
employers and it is enormous.
Note that included with the discretion to pick which employment practices
will be challenged on disparate impact grounds is the discretion to pick which
groups will get protected and which will not. Blumrosen took it for granted that
African Americans were the ones that Title VII was intended to benefit. But in fact
Title VII protects a very large number of groups. Will disparate impacts that affect
African Americans be taken more seriously than those that protect Mexican
EEOC from enforcing the 2012 Guidance against Texas). Both parties appealed, resulting in the
2019 decision cited above.

See also Steven J. Lindsay, Timing Judicial Review of Agency Interpretation in Chevron’s Shadow,
127 Y ALE L.J. 2448 (2018).

104George Mason University law professor Robert Anthony has discussed how agencies, can
effectively coerce private actors using their power to issue mere “guidances” rather than
rules issued according to notice-and-comment procedures (which the EEOC does not have the
authority to issue). Anthony viewed this as faithless to the spirit of the Administrative
Procedure Act. He wrote:

[D]espite any professed reservation of discretion, a nonlegislative document as a

practical matter can quite readily impose binding standards or obligations upon
private parties. Their discretion is constrained even if the agency’s is not. A
test more consistent with the spirit of the APA than one looking to the
constraints on the agency’s discretion would be one that considered whether the
intended or actual constraints on the private persons’ discretion (that is, upon
their freedom of action) amount to binding them in a practical sense. If so, the
recitation that discretion is reserved should be of no moment, and the agency’s
circumvention of legislative rulemaking procedures should be redressed.

Robert A. Anthony, Interpretive Rules, Policy Statements, Guidances, Manuals, and the Like—
Should Federal Agencies Use Them to Bind the Public?, 41 D UKE L. J. 1311, 1360-61 (1992). See also
infra note 247.


Electronic copy available at:

Americans or Native Americans? Suppose, for example, an employer is told he
should not reject job applicants with criminal records, because to do so
disproportionately disadvantages African Americans. He therefore tries to come up
with an alternative to ensure that he only hires highly responsible individuals. He
could request letters of recommendation from a non-relative who has known the
applicant for at least 15 years. But that would disproportionately disadvantage
recent immigrants, which in his town tend to be from El Salvador or Guatemala.
Alternatively, he could depend exclusively on his existing employees to vouch for job
applicants they know well. But it just so happens that his existing employees are
disproportionately Navajo, who tend to disproportionately know other Navajos.
Finally, supporters of disparate impact liability argue that its pervasiveness
is not a problem, because the presumption of illegality can be easily overcome by a
showing of “business necessity.” But as Parts VI and VII will show, particularly in
its discussion of Albemarle Paper v. Moody and subsequent case law both before and
after the promulgation of the Civil Rights Act of 1991, this simply isn’t true.
Meanwhile, one effect of disparate impact liability is that it encourages
employers to shroud their hiring practices in secrecy. Announcing clear job
qualifications is simply an invitation for a disparate impact complaint, investigation
and lawsuit. Why do it? Why not be vague about how decisions are made? Indeed,
why not avoid having a clear policy at all?
One result is that job seekers don’t know what qualifications they need for
certain jobs and therefore fail to acquire those qualifications. Another is that job
seekers must waste their time applying for jobs they have no chance of getting.
But there are two more significant consequences. First, when employers are
vague about job qualifications, it is much harder to detect and prove discrimination.
Leon Myart was inspired to complain to the Illinois Fair Employment Practices
Commission, because he believed that he had done well on Motorola’s standardized
aptitude test and yet was not hired. If he had been told, “Motorola considers many
factors and has no test to screen applicants,” his suspicions might never have been
raised. If they had been raised, it would have been much more difficult to convince
anyone else that race had been a factor in Motorola’s decision not to hire him.
Second, when employers fail to set clear hiring criteria prior to hiring they are
allowing their hiring officials a great deal of discretion in making hiring decisions.
That discretion can be abused. These officials—who can be high-ranking officers or
mid-level managers—are thus in a position to steer those decisions toward their
personal favorites. In such situations, the twin vices of nepotism and cronyism are
more likely to flourish. None of this benefits the individuals most likely to suffer

V. Griggs v. Duke Power Co.: The Supreme Court Deferred to the

EEOC, Not Congress, in Adopting Disparate Impact Liability:


Electronic copy available at:

The facts of Griggs v. Duke Power Co.—the case that adopted the EEOC’s
“interpretation” of Title VII—were complicated, but well worth considering closely:
Duke Power operated a power generation facility in Draper, North Carolina.105 The
company had blatantly discriminated on the basis of race prior to the passage of
Title VII. But as Title VII was about to go into effect in 1965, its officers knew that
all explicit racially discriminatory policies would have to stop. And they did stop.
But new policies went into effect on the very day that Title VII took effect, which
certainly could cause a disinterested observer to suspect that Duke Power might be
attempting to smuggle race considerations in through the back door.
Prior to 1965, the plant had been organized into five operating departments: (1)
Labor, (2) Coal Handling, (3) Operations, (4) Maintenance, and (5) Laboratory and
Test. African Americans were permitted only in the Labor Department, where the
highest-paid employees earned less than the lowest-paid employees in the other
four departments, where whites were employed. Most promotions were made
within each department on the basis of job seniority, but occasionally white
employees could transfer from one department to another, although they usually
began in the lowest position in their new department.106

In 1955, Duke Power had instituted serious changes to its hiring and
promotion policies, which could not have been motivated by race, since at the time,
the company was openly discriminatory and had no need to hide it. Thenceforth a
high school diploma was required for initial assignment to any department except
Labor, and for transfer from the Coal Handling to one of the three higher-paying
departments. The effect of the latter requirement was that whites who had been
brought into the Coal Handling Department before the high school diploma
requirement would not be able to transfer into the higher-paying departments
unless they had a high school diploma. Whites without a high school diploma who
were already in one of the three higher-paying departments were unaffected.
African Americans were also unaffected, since they couldn’t transfer out of the
Labor Department anyway.107

When Title VII went into effect in 1965, the Company abandoned its policy of
restricting Negroes to the Labor Department (as it was required to). But it made a
few other changes as well. From that point on, African Americans could transfer
out of the Labor Department, but like whites in Coal Handling, they would need a
high school diploma. Moreover, to be newly hired into any department other than

105 Griggs v. Duke Power Co., 401 U.S. 424, 425-27 (1971).

106 Id. at 427.

107 Id. at 427-28.


Electronic copy available at:

Labor, one would have to have not only a high school diploma, but also satisfactory
scores on two aptitude tests that the company would administer.108

Was Duke Power motivated by a desire to exclude as many African

Americans as possible? Given the timing, a plausible case, maybe even a more-
than-plausible case can be made that it was. But the lower courts found
otherwise.109 Instead they at least purported to credit the story that Duke Power
was motivated purely by a desire to put a greater emphasis on the “aptitude” of its
new workers.

Was Duke Power lying through its corporate teeth? And were the lower
courts its willing accomplice? Maybe. But, if so, the Supreme Court didn’t think
the evidence for such a conclusion was strong enough to call them out on it. An
appellate court is supposed to accept the factual findings of a trial court unless
those findings are clearly erroneous.110

But the Supreme Court, speaking through Chief Justice Burger, wasn’t
willing just to let the matter go either. And therein lies the tragedy: As every law
student is told, hard cases make bad law. The Supreme Court may have wanted to
find a way to hold Duke Power liable despite the trial court’s finding of no intent to
discriminate on the basis of race, because it nevertheless suspected such an intent.

If so, it accomplished that goal by declaring that intent didn’t matter: “We do
not suggest that either the District Court or the Court of Appeals erred in
examining the employer’s intent; but good intent or absence of discriminatory intent
does not redeem” the Duke Power Company’s employment practices.111 According
to the Court, “Congress directed the thrust of the Act to the consequences of
employment practices, not simply the motivation.”112
108 Id.

Griggs v. Duke Power Co., 292 F. Supp. 243 (M.D.N.C. 1968), rev’d on other grounds, 420 F.2d

1225, 1232 (4th Cir. 1970), rev’d in part on other grounds, 401 U.S. 424 (1971).

110 FED. R. CIV. P. 52(a)(6).

111 Griggs, 401 U.S. at 432.

112Id. Since Griggs, the Court has applied disparate impact liability to the Age Discrimination in
Employment Act (“ADEA”), Pub. L. 90-202, 81 Stat. 602 (1967)(codified at 29 U.S.C. § 621 et seq.),
and the Fair Housing Act, Pub. L. 90-284, 82 Stat. 73 (1968)(codified at 42 U.S.C. § 3601 et seq.).
See Smith v. City of Jackson, 544 U.S. 228 (2005); Texas Dept. of Hous. & Cmty. Affairs v. Inclusive
Cmtys, Project, Inc., 135 S. Ct. 2507 (2015). But it has declined to interpret Title VI of the Civil
Rights Act of 1964 as a disparate impact statute. Alexander v. Sandoval, 532 U.S. 275 (2001).

The way that the Supreme Court arrived at its conclusion that Title VI did not enact disparate
impact liability is a little unusual. In Regents of the University of California v. Bakke, 438 U.S. 265
(1978), plaintiff Bakke argued that Title VI was clear: It banned discrimination against white as


Electronic copy available at:

The Court’s evidence for its assertion about where “Congress directed the
thrust of the Act” is unclear. Astonishingly, the Court never mentioned Senator
Dirksen’s amendment inserting the word “intentionally” into the text of Title VII.
Nor did it mention Senator Humphrey’s statement that the insertion was not a
substantive change, but only a clarification “mak[ing] it wholly clear that
inadvertent or accidental discrimination will not violate the title or result in entry
of court orders.”113 Furthermore, it made no effort to explain Congress’s use of the
phrase “because of such individual’s race, color, religion, sex, or national origin.” A
well as African Americans. Five members of the Court disagreed and held instead that Title VI
simply applies the same standard to federally-funded private entities that would ordinarily be
applied to state entities under the Equal Protection Clause of the Fourteenth Amendment. Put
differently, the five justices took the position that Title VI does not prohibit race discrimination, but
rather bans race discrimination only when it fails strict scrutiny.

The reasoning behind this decision was that several Members of Congress had stated on the
record that Title VI simply applies the Constitution to federally funded private entities.
(Oddly, this was likely only because those Members of Congress understood the Constitution
to forbid race discrimination, not because they did not intend to reject race discrimination
categorically.) And while the Court’s reasoning flies in the face of the statute’s actual text, it
allowed the Court to leave open the possibility that race-preferential admissions policies that
favor minorities could be legal (although the particular policy at issue in Bakke was held to
be a violation of Title VI).

In the long run, the Bakke case was a victory for advocates of race-preferential admissions, since
Justice Powell’s controlling opinion left open the possibility of a race-preferential admissions policy
based on a diversity theory. But it meant Title VI would be interpreted not to include disparate
impact. See Village of Arlington Heights v. Metropolitan Housing Development Corp., 429 U.S. 252,
265 (1977) (holding that the Fourteenth Amendment’s equal protection clause forbids only
intentional discrimination); Washington v. Davis, 426 U.S. 229, 242 (1976) (holding that the Fifth
Amendment’s equal protection component forbids only intentional discrimination).

“Disparate impact liability” does not necessarily mean the same thing from statute to statute. In
Smith, the Supreme Court held that if Title VII was intended to be a disparate impact statute, then
the ADEA must also have been, since the main prohibition in that Act is worded almost identically
(and the EEOC was developing theory of disparate impact liability right around the time the ADEA
was passed in December of 1967). But the ADEA also contains a provision permitting otherwise
prohibited action “where the differentiation is based on reasonable factors other than age.” 29 U.S.C.

Consequently, while the Court found that the ADEA is a disparate impact statute, it also found that
an employer will prevail if it can specify its criterion and show that its criterion was “reasonable.”
While no doubt employers would have preferred the Court to find that the ADEA did not allow for
disparate impact liability, is quite different from proving “business necessity.”

It remains to be seen what “disparate impact liability” will mean in the context of the Fair Housing
113 See supra note 39 and accompanying text.


Electronic copy available at:

wealth of legislative history—like the Clark and Case Memorandum’s statement
that “[a]n employer may set his qualifications as high as he likes, he may test to
determine which applicants have these qualifications”—also went unmentioned.114

Hugh Davis Graham said it best in his sweeping history, The Civil Rights
Era: Origins and Development of National Policy: “Burger’s interpretation of the
legislative intent of Congress in the Civil Rights Act would have been greeted with
disbelief in 1964.”115

The Court attempted to justify its decision as an act of deference to the

EEOC, arguing that “the administrative interpretation of the Act by the enforcing
agency is entitled to great deference.”116 This was also a curious argument in view
of the extent to which Congress went to ensure that the EEOC would not be an
enforcement agency, but only a “toothless” agency charged with investigating and
mediating Title VII complaints.117 Members of Congress had been explicit in
wanting the federal courts, not the EEOC, to be the governmental institution that
would decide issues of fact and law.118

114 See supra Part II.

P OLICY 1960-1972 387 (1990). See also Daniel Rodriguez & Barry R. Weingast, The Positive
Political Theory of Legislative History: New Perspectives on the 1964 Civil Rights Act and Its
Interpretation, 151 U. P ENN . L. R EV . 1417 (2003) (also arguing that the 88th Congress would have
been astonished at the application of disparate impact liability); W ILLIAM E SKRIDGE , P HILIP
C REATION OF P UBLIC P OLICY (3rd ed. 2001) (calling Griggs a “decision poorly reasoned and
vulnerable to the charge that it was a significant leap form the expectations of the enacting
D ISCRIMINATION L AWS 197 (1992) (“If in 1964 any sponsor of the Civil Rights Act had [suggested it
should be interpreted to allow for disparate impact liability], Title VII would have gone down to
thundering defeat . . . .”); D ONALD L. H OROWITZ, T HE C OURTS AND S OCIAL P OLICY 14-15 (1977)
(“There is convincing legislative history to show that Congress intended the opposite of the result
reached in Griggs”); Gary Bryner, Congress, Courts, and Agencies: Equal Employment and the
Limits of Policy Implementation, 96 P OL . S CI. Q. 411, 423 (1981) (“[T]he Court’s Griggs ruling . . .
conflicts with the wording and legislative history of Title VII”); Hugh Steven Wilson, A Second Look
at Griggs v. Duke Power Company: Ruminations on Job Testing, Discrimination, and the Role of the
Federal Courts, 58 V A . L. R EV . 844, 852-58 (1972)(“The overwhelming legislative response in
opposition to the result [in Myart] and the extensive commentary regarding the possibility of a
similar result under Title VII reveal the weakness of the Supreme Court’s [holding in Griggs]).
116 Griggs v. Duke Power Co., 401 U.S. 424, 433-34 (1971).

117 See supra Part II.

118See, e.g., 110 C ONG . R EC . 7246 (1964) (remarks of Sen. Case) (“Only a Federal court would have
the authority to determine whether there had been a violation of the act . . . .”).


Electronic copy available at:

Two additional points are worth making about the Court’s use of “great
deference.” First, in the decades since Griggs, the Court has devoted a good deal of
attention to the issue of deference to agency expertise.119 The issue, of course,
remains controversial with some commentators taking the position that the Court
has been too deferential, especially given the natural tendency on full display with
Title VII, for agency officials to interpret the law in ways that expand their own
power. Nevertheless, it is clear that, under the law as it stands today, an agency
like the EEOC, with no quasi-adjudicatory or substantive quasi-legislative
authority would not be entitled to “great deference.”120 At best its views would be
entitled to consideration.121 Second and more important, even when a standard of
“great deference” is applied, an interpretation that, in Graham’s words, “would have
been greeted with disbelief” by the authors of the statute should have been rejected
by the Court.

Griggs mentioned Myart only in a footnote. In it, the Court asserted that
Myart was distinguishable on the ground that in Myart, even job-related tests with
disparate impact were being outlawed while the standardized test in Griggs was
excluded only because it was found not to be job-related.122 But, if anything, there
was more evidence in Griggs that the test was job-related. Duke Power’s vice
president testified that the new requirements “‘were instituted on the Company’s
judgment that they generally would improve the overall quality of the workforce.”
In Myart, the issue wasn’t discussed. Moreover, the whole point of Myart’s
Congressional and media critics was that the government should not presume to tell
private businesses what is or is not job-related or appropriate as a job
qualification.123 As the Clark & Case Memorandum stated, an employer can set job
qualifications “as high as he likes”—a sentiment echoed by Senator Case
individually when he said, “Title VII says only that covered employers cannot refuse
to hire someone because of his color.”124
119See, e.g., United States v. Mead Corp., 533 U.S. 218 (2001); MCI Telecomm. Corp. v. Am.
Telephone & Telegraph Co., 512 U.S. 218 (1994); Edward J. DeBartolo Corp. v. Florida Gulf Coast
Bldg. & Constr. Trades Council, 485 U.S. 568 (1988); Chevron, U.S.A., Inc. v. Nat. Res. Def. Council,
Inc., 467 U.S. 837 (1984).

120See United States v. Mead Corp., 533 U.S. 218 (2001) (holding that tariff classification ruling by
the United States Customs Service was not entitled to judicial deference).

121 Compare Skidmore v. Swift & Co., 323 U.S. 134, 139-40 (1944).

122 Griggs v. Duke Power Co., 401 U.S. 432, 434 n.10 (1971).

123See, e.g., Arthur Krock, In the Nation: A Pilot Ruling on Equal Employment Opportunity, N.Y.
T IMES , Mar. 13, 1964, at 32.

124Ordinarily, of course, one would expect certain qualities—intelligence, punctuality, integrity,

collegiality, creativity, persistence, and civility—to be always valuable in an employee, all other
things being equal. We may argue about the best way to measure those qualities, but the truth is
that there is no way to measure any of them perfectly, especially given that employers need to make


Electronic copy available at:

VI. The Pre-Wards Cove Business Necessity Defense Was So Difficult
to Establish that Many Employers Likely Opted for Proportional Hiring

Without some sort of an affirmative defense, disparate impact liability would

be just a call for proportional hiring by race and sex. If the law imposes liability for
using job qualifications that have a disparate impact, employers will learn to avoid
that liability by hiring and promoting according to strict quotas. Even with an
affirmative defense, disparate impact liability will encourage proportional hiring if
the affirmative defense is difficult for employers to establish.125 Employers are

such determinations quickly and efficiently. The Griggs Court, however, was unwilling to assume
the intelligence (as measured by the test) was job-related in a manner sufficient to cause it to find
that Duke Power was acting out of business necessity.

The Tower Amendment suffered the same fate as Myart. It was held to apply only to tests when the
employer can prove they are job-related. See G EORGE A. R UTHERGLEN & J OHN J. D ONOHUE III,
E MPLOYMENT D ISCRIMINATION : L AW AND T HEORY 158 (3d ed. 2012) (critically questioning the
Griggs interpretation of the Tower Amendment).

125  Michael Carvin has elaborated on this point at length in a presentation given at Notre Dame

[I]n its purest sense—without some kind of defense for the employer, [disparate
impact liability] is a government mandate for proportional quotas. It is a command by
the government requiring employers to hire all groups in proportion to their
availability in the area. If an employer’s hiring practices result in a disparate impact
for a particular group (i.e., a group is not hired in proportion to its availability) and
Congress makes that illegal, Congress obviously is requiring employers to hire all
groups in proportion. . . .
I think that everyone agrees that, therefore, we must create an escape hatch for
employers. . . . [T]he key question became how stringent should this business
justification be? If it is impossible or unbelievably costly for an employer to justify the
disparate impact, the government essentially is mandating quotas. . . .

The “essential to job performance” standard, which was proposed by the traditional civil rights
groups in 1989, see infra note 188, was, according to Carvin, impossibly high:

[V]irtually nothing is essential to job performance, anybody who goes to law school
knows that a law degree is not essential to good performance as a lawyer. Neither
Abraham Lincoln nor Clarence Darrow had a law degree. More important, when you
apply for a job, no employer says, “Do you have the essential qualifications for this
job?;” “Do you have the minimum qualifications?;” or “I am for hiring for Cravath,
Swaine & Moore, oh, you have a Brooklyn law degree? Fine, here is your office.”
Employers do not ask questions about essential qualifications; instead, they engage in a
comparison of relative qualifications. There are 8 million people with the essential
qualifications to be a lawyer, let alone a civil service employee or a fire fighter, the


Electronic copy available at:

rarely stupid about how to protect themselves. And while a workforce that reflects
the demographics of the labor pool does not guarantee that an employer will not be
liable for the disparate impact of its job qualifications, as a practical matter it comes
pretty close.126

Griggs itself was vague about how an employer should go about rebutting a
prima facie case for liability. The key sentences (or at least the most often-quoted)
are these: “The touchstone is business necessity. If an employment practice which
operates to exclude Negroes cannot be shown to be related to job performance, the
practice is prohibited.”127

But what was meant by “business necessity”? How necessary must “business
necessity” be? We were told that business necessity has something to do with job-
relatedness. But how does one prove job-relatedness? How job-related must “job-
related” be? Does business necessity imply anything more that job-relatedness? On
those questions the Court gave only a few hints:

On the record before us, neither the high school completion

requirement nor the general intelligence test is shown to bear a
demonstrable relationship to successful performance of the jobs for
which it was used. Both were adopted, as the Court of Appeals noted,
without meaningful study of their relationship to job performance
ability. Rather, a vice president of the Company testified the
requirements were instituted on the Company’s judgment that they
generally would improve the overall quality of the workforce.

The evidence, however, shows that employees who have not completed
high school or taken the tests have continued to perform satisfactorily,

essential requirement for which is being vertical. But that is not how people make
decisions because it means that an employer can consider only the minimum
qualifications for the job. Employers can exclude only those applicants who do not
have the minimal qualifications. After excluding that group—those that can not [sic]
be vertical, the employer must hire on the basis of race because he is precluded from
engaging in a relative determination of the merits of the individual applicants.

Michael Carvin, Disparate Impact Claims Under the New Title VII, 68 N OTRE D AME L. R EV . 1153,
1153-54, 1156 (1993).

126In Connecticut v. Teal, 457 U.S. 440 (1982), the Court held that a “bottom line” hiring/promotion
pattern that favors minorities does not insulate an employer from a lawsuit by individual employees
to whom the employer would otherwise be liable in a disparate impact lawsuit. Put differently, an
employer is not absolved from liability just because it has hired (through an affirmative action plan
or otherwise) its proportionate share of minorities. Title VII protects individuals, not groups.

127 Griggs v. Duke Power Co., 401 U.S. 424, 431 (1971).


Electronic copy available at:

and make progress in departments for which the high school and test
criteria are now used. The promotion record of present employees …
suggests the possibility that the requirements may not be needed ….128

The only thing that Griggs makes clear is that business necessity is an affirmative
defense. If lack of business necessity were part of the plaintiff’s case, then the
decision would almost certainly have come out for Duke Power.

Perhaps the most troubling part of the Court’s discussion of business

necessity is this: While noting that some employees without high school diplomas
have nevertheless “performed satisfactorily,” it acknowledged in a footnote that
those with high school diplomas got promoted between July 2, 1965 to November 14,
1966 at slightly higher rates.129 That, of course, suggests that, on average, the high
school graduates performed a bit better. If so, why should it matter that some non-
graduates performed satisfactorily? Most employers are not seeking adequate
employees; they want the best employees they can get. If requiring high school
diplomas gave Duke Power even a slightly better pool of employees, shouldn’t it be
permitted to get that advantage? Aren’t slight advantages often the difference
between a successful business and an unsuccessful business? When the Court
stated that no “meaningful study” took place, does it mean such a study must take
place? What constitutes a “meaningful study”?

A. The Uniform Guidelines on Employee Selection Procedures: If the

Griggs decision provided little guidance, the EEOC was providing plenty. In 1966,
it had issued its Guidelines on Employee Testing Procedures, which was then
amended, expanded, and re-named the Guidelines on Employee Selection
Procedures in 1970. In turn, those guidelines were superseded by the Uniform
Employee Selection Procedures, which were a joint project of the Civil Service
Commission, the EEOC, the Department of Justice, and the Department of
Labor.130 They remain in effect.

The Uniform Guidelines for Employee Selection Procedures (attached to this

Article as Appendix A) are both stringent and likely incomprehensible to the typical
employer. They apply to all “tests and other selection procedures which are used as
a basis for any employment decision.” This is not limited to written tests or to

128 Griggs, 401 U.S. at 431-32.

129 Id. at 432 n. 7.

130These guidelines were an effort to bring uniformity of approach to the various government
agencies charged with enforcing Executive Order 11246 (Department of Labor) and Title VII (EEOC
for private employers, the Department of Justice for public employers) and with supervising federal
hiring (Civil Service Commission).


Electronic copy available at:

physical tests. Rather it includes any job, promotion or work assignment
qualification or criterion upon which decision makers might rely.

Employers must demonstrate “both the validity and utility of a selection

procedure.” That is, they need to prove that the test or other criterion is positively
correlated with successful job performance and also that the correlation is strong
enough to justify the test or criterion’s disparate impact. “Under no circumstances
will the general reputation of a test or other selection procedures, its author or its
publisher, or casual reports of its validity be accepted in lieu of evidence of validity.”

In examining the validity of a particular selection procedure—whether it

predicts that an applicant can do the job successfully—employers are expected to
use “professionally acceptable statistical procedures.” For practical purposes, that
means experts must be consulted. Indeed, if one is curious about why human
resources offices have become so bureaucratized in the last half century, one
significant contributing factor is certainly disparate impact liability and the
Uniform Guidelines for Employee Selection Procedures.

Adverse Impact and Test Validation: A Practitioner’s Guide to Valid and

Defensible Employment Testing is a 192-page, highly-technical tome written by Dan
Biddle. It is eye-opening for anyone who believes that compliance with Title VII’s
disparate impact liability requirements is easy. What is easy to grasp is how
innovative employment practices are less likely to be tried when employers face
such hurdles. 131

B. Albemarle Paper Co. v. Moody: Surprisingly or not, the Supreme Court

was as eager to defer to the EEOC’s various iterations of these guidelines as it had
been to the doctrine of disparate impact liability generally. Four years after Griggs,

V ALID AND D EFENSIBLE E MPLOYMENT T ESTING (2d ed. 2006). Biddle is described in the book
this way:

Dan A. Biddle, Ph.D. is the CEO of Biddle Consulting Group, Inc., (BCG) a
consulting firm specializing in the areas of test development/validation, Equal
Employment Opportunity compliance, and Human Resources software development.
Dan is also CEO of sister companies, Firefighters Selection, Inc. and Public Safety
Testing Solutions, Inc.—firms dedicated to providing fair and defensible testing
solutions to the protective services industry. BCG has consulted with numerous
Fortune 500 companies and hundreds of public sector agencies in matters pertaining
to these areas, and maintains over 1000 software or service clients worldwide. BCG
also provides expert witness/consulting services in state and federal litigation
matters, or in response to government audits.

Id. at xv.


Electronic copy available at:

in Albemarle Paper Co. v. Moody (1975), the Supreme Court followed the EEOC’s
1970 Guidelines on Employee Selection Procedures.132 It thus established
impossibly high standards for showing business necessity. Together, Albemarle
Paper and the various guidelines have effected a major change to employment
practices. Among other things, they have made it difficult in any but rare
circumstances to administer what are sometimes referred to as “paper-and-pencil”
tests. Tests that were once common (including the federal civil service
examinations) are now uncommon and require extraordinary preparation and study
before they are used.133

Like Duke Power, Albemarle Paper had been using two professionally-
developed general aptitude tests in its hiring process. But unlike Duke Power, it
had been using the general aptitude tests even before Title VII was made law, so it
is highly unlikely that its motivation had anything to do with race. Instead, it is
overwhelmingly likely that the policy was aimed at getting Albemarle Paper more
intelligent employees on the not unreasonable theory that more intelligent
employees would do a better job. Just after the decision in Griggs (and probably as
a result of the 1970 Guidelines on Employee Testing Procedures as well as Griggs
itself) it hired an industrial psychologist to attempt to determine scientifically
whether the tests really did help the company identify better employees.134
132 Albemarle Paper Co. v. Moody, 422 U.S. 405, 412 (1975).

133 See, e.g., Ricci v. DeStefano, 557 U.S. 557 (2009) (promotional examination for city firefighters).

134The 1970 Guidelines on Employee Testing Procedures provided the following with regard to
evidence of validity:

§ 1607.4 Evidence of validity.

(a) Each person using tests to select from among candidates for a position or for membership
shall have available for inspection evidence that the tests are being used in a manner which
does not violate § 1607.3. Such evidence shall be examined for indications of possible
discrimination, such as instances of higher rejection rates for minority candidates than
nonminority candidates. Furthermore, where technically feasible, a test should be validated
for each minority group with which it is used; that is, any differential rejection rates that
may exist, based on a test, must be relevant to performance on the jobs in question.
(b) The term “technically feasible” as used in these guidelines means having or obtaining a
sufficient number of minority individuals to achieve findings of statistical and practical
significance, the opportunity to obtain unbiased job performance criteria, etc. It is the
responsibility of the person claiming absence of technical feasibility to positively
demonstrate evidence of this absence.
(c) Evidence of a test’s validity should consist of empirical data demonstrating that the test is
predictive of or significantly correlated with important elements of work behavior which
comprise or are relevant to the job or jobs for which candidates are being evaluated.
(1) If job progressions structures and seniority provisions are so established that new
employees will probably, within a reasonable period of time and in a great majority of
cases, progress to a higher level, it may be considered that candidates are being
evaluated for jobs at that higher level. However, where job progression is not so nearly


Electronic copy available at:

The idea of having employers provide proof of the kind that would be
accepted by social scientists was never a sound one—but it is the approach taken by
the EEOC.135 First of all, if the EEOC is really wants it to be done according to
automatic, or the time span is such that higher level jobs or employees’ potential may be
expected to change in significant ways, it shall be considered that candidates are being
evaluated for a job at or near the entry level. This point is made to underscore the
principle that attainment of or performance at a higher level job is a relevant criterion
in validating employment tests only when there is a high probability that persons
employed will in fact attain that higher level job within a reasonable period of time.
(2) Where a test is to be used in different units of a multiunit organization and no
significant differences exist between units, jobs, and applicant populations, evidence
obtained in one unit may suffice for the others. Similarly, where the validation process
requires the collection of data throughout a multiunit organization, evidence of validity
specific to each unit may not be required. There may also be instances where evidence
of validity is appropriately obtained from more than one company in the same industry.
Both in this instance and in the use of data collected throughout a multiunit
organization, evidence of validity specific to each unit may not be required: Provided,
That no significant differences exist between units, jobs, and applicant populations.

35 Fed. Reg. 12334 (August 1, 1970).

135The EEOC guidelines require someone with both legal and statistical training to understand

§1607.5 Minimum standards for validation.

(a) For the purpose of satisfying the requirements of this part, empirical evidence in support of
a test’s validity must be based on studies employing generally accepted procedures for
determining criterion-related validity, such as those described in “Standards for Educational
and Psychological Tests and Manuals” published by the American Psychological Association,
1200 17th Street NW., Washington, D.C. 20036. Evidence of content or construct validity, as
defined in that publication, may also be appropriate where criterion-related validity is not
feasible. However, evidence for content or construct validity should be accompanied by
sufficient information from job analyses to demonstrate the relevance of the content (in the
case of job knowledge or proficiency tests) or the construct (in the case of trait measures).
Evidence of content validity alone may be acceptable for well-developed tests that consist of
suitable samples of the essential knowledge, skills or behaviors composing the job in
question. The types of knowledge, skills or behaviors contemplated here do not include those
which can be acquired in a brief orientation to the job.
(b) Although any appropriate validation strategy may be used to develop such empirical
evidence, the following minimum standards, as applicable, must be met in the research
approach and in the presentation of results which constitute evidence of validity:
(1) Where a validity study is conducted in which tests are administered to applicants, with
criterion data collected later, the sample of subjects must be representative of the
normal and typical candidate group for the job or jobs in question. This further assumes
that the applicant sample is representative of the minority population available for the
job or jobs in questions in the local labor market. Where a validity study is conducted in
which tests are administered to present employees, the sample must be representative of
the minority groups currently included in the applicant population. If it is not
technically feasible to include minority employees in validation studies conducted on the


Electronic copy available at:

present work force, the conduct of a validation study without minority candidates does
not relieve any person of his subsequent obligation for validation when inclusion of
minority candidates becomes technically feasible.
(2) Tests must be administered and scored under controlled and standardized conditions,
with proper safeguards to protect the security of test scores and to insure that scores do
not enter into any judgments of employee adequacy that are to be used as criterion
measures. Copies of tests and test manuals, including instructions for administration,
scoring, and interpretation of test  results,  that  are  privately  developed  and/or  are  not  
available  through  normal  commercial  channels  must  be  included  as  a  part  of  the  validation  
(3) The  work  behaviors  or  other  criteria  of  employee  adequacy  which  the  test  is  intended  to  predict  
or  identify  must  be  fully  described;  and  additionally,  in  the  case  of  rating  techniques,  the  
appraisal  form(s)  and  instructions  to  the  rater(s)  must  be  included  as  a  part  of  the  validation  
evidence.      Such  criteria  may  include  measures  other  than  actual  work  proficiency,  such  as  
training  time,  supervisory  ratings,  regularity  of  attendance  and  tenure.    Whatever  criteria  are  
used  they  must  represent  major  or  critical  work  behaviors  as  revealed  by  careful  job  analyses.  
(4) In  view  of  the  possibility  of  bias  inherent  in  subjective  evaluations,  supervisory  rating  
techniques  should  be  carefully  developed,  and  the  ratings  should  be  closely  examined  for  
evidence  of  bias.    In  addition,  minorities  might  obtain  unfairly  low  performance  criterion  scores  
for  reasons  other  than  supervisors’  prejudice,  as,  when,  as  new  employees,  they  have  had  less  
opportunity  to  learn  job  skills.    The  general  point  is  that  all  criteria  need  to  be  examined  to  
insure  freedom  from  factors  which  would  unfairly  depress  the  scores  of  minority  groups.  
(5) Differential  validity.    Data  must  be  generated  and  results  separately  reported  for  minority  and  
nonminority  groups  wherever  technically  feasible.    Where  a  minority  group  is  sufficiently  large  
to  constitute  an  identifiable  factor  in  the  local  labor  market,  but  validation  data  have  not  been  
developed  and  presented  separately  for  that  group,  evidence  of  satisfactory  validity  based  on  
other  groups  will  be  regarded  as  only  provisional  compliance  with  these  guidelines  pending  
separate  validation  of  the  test  for  the  minority  group  in  question.    (See  §  1607.9).    A  test  which  is  
differentially  valid  may  be  used  in  groups  for  which  it  is  valid  but  not  for  those  in  which  it  is  not  
valid.  In  this  regard,  where  a  test  is  valid  for  two  groups  but  one  group  characteristically  obtains  
higher  test  scores  than  the  other  without  a  corresponding  difference  in  job  performance,  cutoff  
scores  must  be  set  so  as  to  predict  the  same  probability  of  job  success  in  both  groups.  
(c) In  assessing  the  utility  of  a  test  the  following  considerations  will  be  applicable:  
(1) The  relationship  between  the  test  and  at  least  one  relevant  criterion  must  be  statistically  
significant.    This  ordinarily  means  that  the  relationship  should  be  sufficiently  high  as  to  have  a  
probability  of  no  more  than  1  in  20  to  have  occurred  by  chance.    However,  the  use  of  a  single  
test  as  the  sole  selection  device  will  be  scrutinized  closely  when  that  test  is  valid  against  only  
one  component  of  job  performance.  
(2) In  addition  to  statistical  significance,  the  relationship  between  the  test  and  criterion  should  
have  practical  significance.    The  magnitude  of  the  relationship  needed  for  practical  significance  
or  usefulness  is  affected  by  several  factors,  including:  
(i) The  larger  the  proportion  of  applicants  who  are  hired  for  or  placed  on  the  job,  the  
higher  the  relationship  needs  to  be  in  order  to  be  practically  useful.    Conversely,  a  
relatively  low  relationship  may  prove  useful  when  proportionately  few  job  vacancies  
are  available;  
(ii) The  larger  the  proportion  of  applicants  who  become  satisfactory  employees  when  not  
selected  on  the  basis  of  the  test,  the  higher  the  relationship  needs  to  be  between  the  
test  and  a  criterion  of  job  success  for  the  test  to  be  practically  useful.    Conversely,  a  
relatively  low  relationship  may  prove  useful  when  proportionately  few  applicants  turn  
out  to  be  satisfactory;  
(iii) The  smaller  the  economic  and  human  risks  involved  in  hiring  an  unqualified  applicant  
relative  to  the  risks  entailed  in  rejecting  a  qualified  applicant,  the  greater  the  


Electronic copy available at:

professionally acceptable standards (and it’s not clear that it does), it will
presumably require the employer to have a control group ready at hand. In the
context of hiring felons, for example, the employer would have to hire a sufficient
number of felons to show that they do indeed behave more irresponsibly than other
employees. Or in the context of an aptitude test, the employer would have to hire
some employees who did poorly on the test.136 That is asking a lot of employers.137
You can go over Title VII and its legislative history with a microscope looking for
evidence that Congress had in mind anything approaching such an approach and
you won’t find it. Instead, such a requirement directly contradicts Rep. McCulloch’s
statement concerning the need to keep interference to a minimum.138

The industrial psychologist hired by Albemarle Paper actually did quite well
considering the difficulty of his assignment. He found that test scores were indeed
correlated with better performance. And his findings were statistically significant
for all Albemarle Paper jobs collectively. But that wasn’t good enough for the Court.
It complained that, among other things, the psychologist was unable to generate
statistically significant results for each job category and each test. The Court
seemed unaware that in dealing with groups that small, the chance of generating
statistically significant results is vanishingly small.

relationship  needs  to  be  in  order  to  be  practically  useful.    Conversely,  a  relatively  low  
relationship  may  prove  useful  when  the  former  risks  are  relatively  high.  
136The alternative is to hire those who did well or very well, exclude those who did poorly or very
poorly, and then test the predictive power of the test only among those who were hired. This
produces error. A finding that “A” test-takers do not perform significantly better than “B” test-takers
does not means that the “A”s and “B”s do not produce results well above the “C”s, “D”s, and “F”s.

137 The 1970 Guidelines did provide an out for businesses too small to conduct a full-blown validity
study, but it was not an especially useful one, since it require small businesses to expend resources
building a substantial record, something they are probably not in a position to do:

§ 1607.7 Use of other validity studies.

In cases where the validity of a test cannot be determined pursuant to §1607.4 and § 1607.5 (e.g. the
number of subjects is less than that required for a technically adequate validation study, or an
appropriate criterion measure cannot be developed), evidence from validity studies conducted in
other organizations, such as that reported in test manuals and professional literature, may be
considered acceptable when: (a) The studies pertain to jobs which are comparable (i.e., have
basically the same task elements), and (b) there are no major differences in contextual variables or
sample composition which are likely to significantly affect validity. Any person citing evidence from
other validity studies as evidence of test validity for his own jobs must substantiate in detail job
comparability and must demonstrate the absence of contextual or sample differences cited in
paragraphs (a) and (b) of this section.

138 See supra note 1 and accompanying text.


Electronic copy available at:

Strangely, the Court also complained that the industrial psychologist’s efforts
to validate the test involved efforts to correlate individual test scores with a
supervisor’s subjective ratings of individuals’ performance. The Court rejected
those supervisor ratings as inadequate measures of actual job performance. In
doing so, it was rejecting both objective tests and subjective all-things-considered
ratings of actual job performance. It is difficult to imagine what would satisfy the
Court.139 (And it is difficult to imagine the Court’s members select their own clerks
with such care.)

It is worth noting here that not everything in business (or indeed in life) is
susceptible to scientific proof. Indeed, every day we all make decisions—sometimes
even momentous ones—based on less-than-scientific evidence. It is a way of life for
all of us. Even if small and medium-sized businesses somehow managed to get
controls, they would still have little hope of having enough data to prove the value
139The following year, the Court decided Washington v. Davis, 426 U.S. 229 (1976). That case
concerned an entrance examination to the District of Columbia’s intensive 17-week police training
program, which leads to a job with the police department. The entrance exam was a standard civil
service examination developed by the U.S. Civil Service Commission, not by the police department.

The parties and the courts below had assumed that Title VII standards applied to the case and
briefed the case accordingly. The Court held, however, that in a Due Process case brought pursuant
to the Fifth Amendment that a federal law or policy (or, as in this case, a policy of the District of
Columbia), a violation occurs only if the law or policy involves an intentional effort to discriminate on
the basis of race. Disparate impact is insufficient.

The Court nevertheless stated that the defendants had effectively rebutted the possibility of Griggs-
style disparate impact liability. Washington, 426 U.S. at 250-52. It described the lengths that the
defendants had gone to prove that this standard civil service examination was in keeping with
business necessity:
The record includes a validation study of Test 21's relationship to performance in
the recruit training program. The study was made by D. L. Futransky of the
Standards Division, Bureau of Policies and Standards, United States Civil Service
Commission. App., at 99-109. Findings of the study included data "support[ing] the
conclusion that T[est] 21 is effective in selecting trainees who can learn the material
that is taught at the Recruit School." Id., at 103. Opinion evidence, submitted by
qualified experts examining the Futransky study and/or conducting their own
research, affirmed the correlation between scores on Test 21 and success in the
training program. E.g., Affidavit of Dr. Donald J. Schwartz (personnel research
psychologist, United States Civil Service Commission), App. 178, 183 ("It is my
opinion . . . that Test 21 has a significant positive correlation with success in the
MPD Recruit School for both Blacks and whites and is therefore shown to be
jobrelated . . ."); affidavit of Diane E. Wilson (personnel research psychologist,
United States Civil Service Commission), App. 185, 186 ("It is my opinion that there
is a direct and rational relationship between the content and difficulty of Test 21
and successful completion of recruit school training").

Id. at 251 n.17



Electronic copy available at:

of their hiring methods. That’s is why most employers must, to some degree, rely on
common sense and experience rather than science.

Imagine, for example, an employer who operates a shoe store. She cannot
scientifically prove that the job applicants she regards, after an interview, as more
polite will make better shoe salesmen. But it’s a pretty good guess that they do—
and certainly one she should be entitled to make.140 To prove it, however, she would
need to hire a large number of rude salesmen in order to have adequate controls
and then buy a chain of shoe stores in order to have enough data to generate
statistically significant results. That can’t be right.

If Albemarle Paper continues to state the law—and as this Article discusses

in Part VII, in some ways it does—employers are in a very difficult situation.
Interestingly, it was Justice Blackmun who saw this most clearly in 1975. In his
opinion concurring in the judgment, he wrote: “I fear that a too-rigid application of
the EEOC Guidelines will leave the employer little choice, save an impossibly
expensive and complex validation study, but to engage in a subjective quota system
of employment selection. This, of course, is far from the intent of Title VII.”141

C. Dothard v. Rawlinson: Two years later the Court again applied

disparate impact liability to Title VII claims. This time—in Dothard v. Rawlinson
(1977)142—it was concerned specifically with employment practices with a disparate
impact on women. It held that a requirement that prison personnel weigh at least
120 pounds and be at least 5 feet 2 inches tall was a violation of Title VII on account
of its disparate impact (despite the fact that most women would qualify under the

According to the Court, the state appellants had argued that the height and
weight requirements “have a relationship to strength, a sufficient but unspecified
140  An episode of the 1960s situation comedy The Dick van Dyke Show, entitled “Young Man with a
Shoehorn,” playfully challenged the conventional wisdom that polite salesmen are more successful
than impolite ones. In the episode, it turned out that customers wanted to buy shoes from an
impolite discount shoe salesman, because it indicated to them they were getting a bargain. But The
Dick van Dyke Show was a comedy. Don’t invest your savings in a shoe store that does not screen its
salespersons for politeness. See The Dick Van Dyke Show: Young Man with a Shoehorn (CBS
television broadcast Feb. 24, 1965), [
141Albemarle Paper Co. v. Moody, 422 U.S. 405, 448 (1975). (Blackmun, J., concurring in the

142 Dothard v. Rawlinson, 433 U.S. 321 (1977).

143 Id. at 323-24, 329, 332.


Electronic copy available at:

amount of which is essential to effective job performance as a correctional
counselor.”144 To this, the Court responded:

If the job-related quality that the appellants identify is bona fide, their
purpose could be achieved by adopting and validating a test for
applicants that measures strength directly. Such a test, fairly
administered, would fully satisfy the standards of Title VII because it
would be one that “measure[s] the person for the job and not the
person in the abstract.” Griggs v. Duke Power Co., 401 U.S. at 436.
But nothing in the present record even approaches such a

This appears to signal, perhaps in a confused manner, not just a continued

adherence to Griggs, but also to Albemarle Paper. Tests—including height and
weight requirements—must be validated.146
144 Id. at 331.

145 Id. at 332 (emphasis added).

146 But the Court is somewhat unclear about what must be validated. Albemarle Paper had tried to
follow the EEOC’s 1970 Guidelines on Employee Testing Procedures. In doing so, it had demanded
proof of the kind a social scientist would recognize that passing scores on the aptitude tests were
positively correlated with actual success on the job. Here the Court appears to want (1) proof that
Alabama’s desire for strong correctional counselor is “bona fide.” i.e. is being made in good faith and
is not a pretext for discrimination; (2) the adoption of some “direct” test of strength; and (3)

The first requirement is what Title VII was originally supposed to be about—actual discrimination.
The next two requirements are supposed to be related to Albemarle Paper. But it is not clear that
the Court is focused on what needs to be a valid measure of what. It complains that “[i]n the District
Court . . . the appellants produced no evidence correlating the height and weight requirements with
the requisite amount of strength thought essential to good job performance.” Id. at 331. But if it had
done so, would that have been enough? That seems unlikely. Surely, proof that Alabama’s height
and weight requirements can be correlated with some “direct” measure of strength “thought to be
essential to good performance” by Alabama would have been easy. It is overwhelmingly likely that
height at or above 5’ 2” and weight at or above 120 does indeed positively (if imperfectly) correlate
with “some” direct measure of strength (such as the ability to lift a certain number of pounds) that
Alabama might specify as necessary.

But if the Court wants a job requirement that “‘measure[s] the person for the job and not the person
in the abstract,’” that seems to mean that the direct measure of strength must also be shown to
correlate strongly with some measure of job performance. See id. at 332. But how should that “direct”
measure of strength be selected? For a warehouse worker one might be able to predict in advance
the largest item he or she may be called upon to lift. But for a correctional officer, the question may
be whether he or she has the level of strength necessary to break up a fight among inmates or to
discourage an inmate from attacking. What level of strength is that? And how often will it need to
be employed? That information is hard to come by. And with any luck, that skill will be held in


Electronic copy available at:

Two points seemed to elude the Court when it insisted that a “direct
measure” of strength should be employed rather than height and weight
requirements: First, while it is true that height and weight are not strictly
speaking measures of strength, it is not strength itself, but the appearance of
strength that is more important in keeping prisoners under control. For that,
height and weight may be a better measure than number of pounds one can lift or
any other more direct measure of strength.147 It may actually decrease the
likelihood that a correctional counselor will have to face violence.
Second, even if the state petitioners had established a direct measure of strength—
like ability to lift a certain weight—it would have been tricky business to validate it.
Strength is most necessary in emergencies—like prison riots—and emergencies are
(with any luck) few and far between. Yet the ability to deal with them is
undeniably crucial. The research necessary to validate such a job qualification could
take decades.

In any event, the upshot of Albemarle Paper and Dothard was that proof of
business necessity was not easy to come by.

D. New York City Transit Authority v. Beazer: The next occasion for the
Supreme Court to consider disparate impact liability was New York City Transit
Authority v. Beazer (1979).148 That case appears to be applying a more relaxed
standard and hence tends to be cited by those who maintain that the business
necessity standard should be and is relatively easy to establish.149 Beazer
concerned New York City Transit Authority’s policy of not employing persons who
use narcotic drugs, including those on a methadone maintenance program as part of
an intended cure for heroin addiction. Methadone users objected.150

The Court was unimpressed with the plaintiff’s evidence that declining to
hire those on a methadone maintenance program had a disparate impact on African
Americans or Hispanics. But it ultimately based its decision rejecting plaintiffs’
disparate impact liability cause of action on its conclusion that, in any event, the
147Interestingly, arguments inviting the Court to apply disparate impact liability to the Equal
Protection Clause of the Fourteenth Amendment or the equal protection component of the Due
Process Clause of the Fifth Amendment were rejected. See Village of Arlington Heights v. Metro.
Hous. Dev. Corp., 429 U.S. 252 (1977) (holding that the Fourteenth Amendment’s equal protection
clause forbids only intentional discrimination); Washington v. Davis, 426 U.S. 229 (1976) (holding
that the Fifth Amendment’s equal protection component forbids only intentional discrimination).

148 N.Y.C. Transit Auth. v. Beazer, 440 U.S. 568 (1979).

149 See infra Part VII for a discussion how Beazer was viewed as affecting the proper interpretation
of the Civil Rights Act of 1991.

150 Id. at 571-73.


Electronic copy available at:

New York Transit Authority had effectively rebutted any presumption against it by
demonstrating that the policy had a "manifest relationship to the employment in
question”—language taken directly from Griggs and quoted in Albemarle Paper.151

The Court’s analysis gave employers hope that they could sometimes produce
enough evidence to satisfy the business necessity standard. But the standard was
nevertheless set very high. The New York City Transit Authority produced large
amounts of evidence and expert testimony to justify its rule. But at least the
Supreme Court had agreed that it had done so successfully.

The Court noted that while “‘the strong majority’ of patients who have been
methadone maintenance for at least a year are free from illicit drug use,” “a
significant number are not.”152 “On this critical point,” the Court wrote, “the
evidence relied upon by the District Court reveals that even among participants
with more than 12 months’ tenure in methadone maintenance programs, the
incidence of drug and alcohol abuse may often approach and even exceed 25%.”153

The District Court had held that it was permissible for the Transit Authority
to exclude all methadone users from the 25% of jobs that are “safety sensitive.” As
for the rest of the jobs, the Supreme Court stated that the parties to the case also
agreed that the Transit Authority’s “legitimate employment goals of safety and
efficiency require the exclusion of all users of illegal narcotics, barbiturates, and
amphetamines, and of a majority of all methadone users.” (Italics added.) At issue,
therefore, was whether Title VII requires the Transit Authority to employ some
methadone users in non-safety sensitive jobs.

On that question, the Court was impressed by the fact that “the District
Court [had] noted that [safety and efficiency] goals are significantly served by--even
if they do not require –[the Transit Authority]'s rule as it applies to all methadone
users including those who are seeking employment in non-safety-sensitive
positions.” [Italics supplied.] According to the Court, “The record thus
demonstrates that [the Transit Authority]'s rule bears a "manifest relationship to
the employment in question" and hence the plaintiffs’ showing “clearly failed to
carry [their] ultimate burden of proving a violation of Title VII.154 The Court did

151 Id. at 587 n.31.

152 Id. at 575-76.

153 Id. at 576.

Id. at 587 n. 31 (quoting Griggs v. Duke Power Co.); Albemarle Paper Co. v. Moody, 422 U.S. 405,

425 (1975).


Electronic copy available at:

not elaborate beyond this; it did not purport to overrule Albemarle Paper, or

Contrast Beazer with Griggs. Griggs was unimpressed by evidence that on

average high school graduates were promoted very slightly more often than non-
graduates (and hence can be said to have performed slightly better). The Court
pointed instead to evidence that non-graduates hired before the requirements
tended to perform satisfactorily. What mattered in Beazer was that the group as a
whole would contain some bad apples who could have a negative effect on the
Authority’s safety and efficiency even if those hired could be kept out of safety-
sensitive positions.

As this Article will discuss in Part VII, during the negotiations over the Civil
Rights Act of 1991, the Bush Administration hoped that post 1991 Act courts would
view the Beazer decision as a turning point in the evolution of the business
necessity doctrine. But if Beazer was intended to be a significant modification of
Griggs and Albemarle Paper, it was not taken as such by U.S. Courts of Appeals. In
Williams v. Colorado School District No. 11 (1981), the court stated, “[I]n a
disparate impact case, unlike a disparate treatment case, a rational or legitimate,
nondiscriminatory reason is insufficient. The practice must be essential, the
purpose compelling.”156 Similarly, according to the court in EEOC v. Rath Packing
Co. (1986), “[T[he proper standard . . . is not whether [the employment practice in
question] is justified by routine business considerations but whether there is a
compelling need for … that practice . . . .”157

E. Watson v. Fort Worth Bank & Trust: The next Title VII disparate
impact case was Watson v. Fort Worth Bank & Trust (1988).158 This was the first
Supreme Court disparate impact case to deal specifically with a subjective job
qualification, such as a general evaluation of an employee’s suitability by a
155According to Senator Orrin Hatch shortly before the vote in the Senate on the Civil Rights Act of
1991, the Beazer decision “was well known to all sides in these negotiations and debates.” 137
C ONG . R EC . 28,842 (1991).

156 Williams v. Colo. Sch. Dist. No. 11, 641 F.2d 835, 842 (10th Cir. 1981).

EEOC v. Rath Packing Co., 787 F.2d 318, 331-32 (8th Cir. 1986) (emphasis in original) (citation

omitted) (quoting Kirby v. Colony Furniture Co., 613 F.2d 696, 705 n.6 (8th Cir. 1980)).
158 Watson v. Fort Worth Bank & Tr., 487 U.S. 977 (1988).

159Plaintiff Watson was an African-American woman who had applied for a promotion to a
supervisory position at the defendant bank on four occasions, but on each occasion she lost out to a
white applicant. The bank did not have any hard and fast job qualifications that Watson failed to
meet. Rather, it relied upon the subjective judgments of bank supervisors, all of whom were white at
the time. Id. at 982. The trial court had held in the bank’s favor on the issue of actual discrimination


Electronic copy available at:

The strongest argument for excluding subjective evaluations from disparate
impact analysis is that including them leads directly to proportional hiring. What
is the alternative to supervisors who disproportionately select white males for
promotions? Most likely it is supervisors who are careful to select a group that
“looks like America”—even if that means selecting them precisely because of their
race. From the employer’s standpoint, it’s so obvious as to be almost irresistible.
Disparate impact—especially when dealing with subjective assessments—leads to
proportional hiring or quotas.

The Supreme Court held that subjective job qualifications were covered by
disparate impact liability. In so holding, it was arguably being true to the logic of
Griggs, Albemarle Paper, Dothard, and Beazer (though not to the 1964 Act itself).
There is nothing in those cases to suggest a distinction between subjective and
objective job qualifications. If one assumes that Congress intended liability for
disparate impact, then it naturally applies to all job qualifications, not just objective
qualifications like test scores, college degrees and the like.160

But some members of the Court were apparently nervous. Watson thus
became the Court’s first Title VII case to articulate some discomfort with Griggs
and with the incentives created by wide-ranging disparate impact liability. But it
came in the part of Justice O’Connor opinion in which only a plurality joined--
O’Connor herself, Rehnquist, Scalia and White. It warned:

We agree that the inevitable focus on statistics in disparate impact

cases could put undue pressure on employers to adopt inappropriate
prophylactic measures. It is completely unrealistic to assume that
unlawful discrimination is the sole cause of people failing to gravitate
to jobs and employers in accord with the laws of chance. It would be
equally unrealistic to suppose that employers can eliminate, or
discover and explain, the myriad of innocent causes that may lead to
statistical imbalances in the composition of their workforces.161

(disparate treatment). Id. at 983-84. It also concluded that disparate impact liability did not apply to
the situation (though it apparently did not elaborate on the reasons). Id. On appeal, the U.S. Court
of Appeals for the Fifth Circuit held that the trial court was correct in declining to address plaintiff’s
disparate impact claim, because subjective job qualifications can be subjected only to liability for
actual discrimination, not for mere disparate impact. Id. at 984.

160 See id. at 990-91.

161Id. at 992 (citation omitted). Note that the Court seemed to be suggesting that if an employer
could explain the innocent causes that may lead to statistical imbalances, it would not be liable. But
“I have some tentative reason to believe my hiring practices will marginally increase the quality of
my workforce” is a reason innocent of discrimination. Yet it wouldn’t have passed muster under
Griggs, Albemarle Paper or Dothard.


Electronic copy available at:

O’Connor focused on how to deal with the incentives created by disparate impact
liability. She did not suggest in her opinion that Griggs should be overruled.
Instead, she seemed to indicate that to overrule Griggs would mean that
unconscious discrimination would not be covered by Title VII.

This was a mistake. There are three different possibilities here: (1)
consciously discriminatory treatment; (2) unconsciously discriminatory treatment;
and (3) disparate impact that is not motivated, consciously or unconsciously, by
race, color, religion, sex, or national origin. Griggs took the extreme position that
Congress intended the third category—actions that have a disparate impact (so long
as they are not compelled by business necessity)—to be prohibited by Title VII. It
said nothing about the second category.

Imagine three hypotheticals. The first is completely uncontroversial. If an

employer chooses not to hire Italian Americans because he thinks to himself that,
based on their national origin, they may be associated with organized crime, he is
engaging in consciously discriminatory treatment and thus violating Title VII.

The second is more arguable: If the employer declines to hire Italian

Americans whose qualifications are just as good or better than the other job
applicants that he does hire, because in the back of his mind he is associating them
with organized crime, he is unconsciously treating them differently from other job
applicants. To put it in other terms, if one runs the counterfactual—where the job
applicant had been a different national origin group but otherwise identical—he
would have gotten the job. This is a good candidate for treatment as a violation of
Title VII too.

It is the third hypothetical that tracks the disparate impact holding in

Griggs: The employer decides to require a Ph.D. in public administration for the job
and applies it evenhandedly. It wasn’t obvious to him that a non-Ph.D. would be
unable to perform the job. But all things considered he believed that he was more
likely to select the best possible person for the slot if he required such a degree. He
may not even have known it when he decided to require it, but a much smaller
proportion of Ph.D.s in public administration are Italian American than the
proportion of Italian Americans in the otherwise qualified population. When asked
whether it was “necessary” to have such a requirement, the employer replied, “No,
but I think it will help me get the best possible employee for the slot.”

If the Watson Court had overruled Griggs, that would have left a Title VII
concerned with discriminatory treatment rather than mere disparate impact. In
other words, overruling Griggs would not have required the court to hold that


Electronic copy available at:

Hypothetical #2 is not covered by Title VII. Instead, it would have ruled only
Hypothetical #3 out of bounds.162

Instead of calling for Griggs to be overruled, the Watson plurality first

warned that disparate impact liability could evolve to promote preferential
treatment and quotas:

Preferential treatment and the use of quotas by public employers

subject to Title VII can violate the Constitution, and it has long been
recognized that legal rules leaving any class of employers with “little
choice” but to adopt such measures would be “far from the intent of
Title VII.” Respondent and the United States are thus correct when
they argue that extending disparate impact analysis to subjective
employment practices has the potential to create a Hobson’s choice for
employers and thus to lead in practice to perverse results. If quotas
and preferential treatment become the only cost-effective means of
avoiding expensive litigation and potentially catastrophic liability,
such measures will be widely adopted. The prudent employer will be
careful to ensure that its programs are discussed in euphemistic
terms, but will be equally careful to ensure that the quotas are met.
Allowing the evolution of disparate impact analysis to lead to this

162A different argument could rule Hypothetical #2 out of bounds, but it would have nothing to do
Griggs. Recall that Dirksen amended Section 706(g)(1) to read: “If the court finds that the
respondent has intentionally engaged in or is intentionally engaging in an unlawful employment
practice charged in the complaint, the court may enjoin the respondent [or invoke certain other
limited remedies] ….” See supra note 39 and accompanying text. One could argue that Dirksen’s
insertion of the term “intentionally” should foreclose even the possibility that a court will hold
unconscious discriminatory treatment. But that is by no means a necessary interpretation of that
provision. As the Oxford English Dictionary acknowledges, one can be conscious or unconscious of
one’s intentions. C OMPACT O XFORD E NGLISH D ICTIONARY 317 (2d ed. 1991) (“conscious” is defined
as (among other things) “aware of what one is … intending to do”).

The fact that Dirksen was seeking to prevent the hearing examiner’s opinion in Myart from
becoming federal law makes this interpretation eminently plausible. Myart was based on disparate
impact. Since the distinction between a conscious act and an unconscious act is not that one is
intentional and the other is not, it is likely that Dirksen’s intent was simply to rule disparate impact
liability out of bounds. Another plausible interpretation of Dirksen’s amendment, which affects only
the remedial provisions of Title VII and not the prohibition itself, is that he intended the EEOC to
investigate and mediate both conscious and unconscious discrimination (i.e. unequal treatment,
regardless of whether the employer is unconscious of it), but that only conscious discrimination could
form the basis of a lawsuit before a federal court. Of course, whether an employer engages in
unconscious but actual discrimination (as opposed to mere disparate impact) can change over the
course of an EEOC investigation and mediation. Once the employer’s discriminatory treatment is
brought to the employer’s attention and the employer does not reverse course and begin to treat the
complainant on an equal basis with other job applicants or employees, it becomes more difficult for
the employer to maintain that its actions are unconscious.


Electronic copy available at:

result would be contrary to Congress’ clearly expressed intent, and it
should not be the effect of our decision today.163

It then took the position that two constraints should be observed to ensure
that disparate impact liability doesn’t get out of hand:

(1) Plaintiff must be responsible for identifying the practice that caused her to
be excluded and for showing that this practice has a disparate impact on job
applicants in a protected group of which she is a member.164

This was a curious requirement (and one that Congress would later re-
examine in connection with the Civil Rights Act of 1991).165 In Watson itself, how
should the plaintiff isolate what made the bank’s supervisors choose white
candidates over her? What does that mean? At what level of abstraction should it
be analyzed? Was her rejection because one or more of her supervisors value
experience more than plaintiff believes is necessary to pick a bank supervisor? Or
was the rejection because one or more of them values a degree in business or
accounting more than plaintiff believes is necessary to pick a bank supervisor? Or
was it because they generally overvalue experience no matter what the job category
is? Is it enough to show that those supervisors disproportionately choose white
candidates for no articulated reason? Or that the specific practice was letting those
particular individuals (and not certain other individuals) make the judgment?

(2) Defendant is responsible only for “producing evidence that its employment
practices are based on legitimate business reasons.” It is up to the plaintiff
to “‘show that other tests or selection devices, without a similarly
undesirable racial effect, would also serve the employer’s legitimate interest
in efficient and trustworthy workmanship.’”166

This appeared to be new. In Griggs, the defendant had the burden of proof on business
necessity. Now the plurality was explicitly stating that it was really a burden of
producing evidence, rather than a burden of proof.167 In addition, by phrasing it as
“legitimate business reason” the Watson plurality did indeed appear to be subtly softening
the concept of business necessity. But, again, this was only a plurality opinion.

163 Watson, 487 U.S. at 993 (citations omitted).

164 See id.. at 993-95.

165 See infra Part VII.

166 Watson, 487 U.S at 997-98 (citation omitted) (quoting Albemarle Paper, 422 U.S. at 425).

167 See Tex. Dept. of Cmty. Affairs v. Burdine, 450 U.S. 248 (1981).


Electronic copy available at:

F. Wards Cove Packing Co. v. Atonio: With Wards Cove Packing Co. v.
Atonio (1989),168 the plurality opinion in Watson or something very like it became
law—but only for a short period of time. Wards Cove concerned Alaska canneries
that operated only during the summer months. There were two broad categories of
jobs at these canneries: the unskilled jobs that involved actual canning and the
skilled jobs that did not. The unskilled jobs were filled primarily by Filipinos who
were hired through a union, and Native Alaskans, who lived year-round in close-by
communities. The skilled jobs—which included office workers, electricians, medical
personnel and others—were predominantly white. They were hired in the offseason
at the offices of the various cannery companies in Washington State and Oregon.169

The plaintiffs argued that these hiring practices had a disparate impact on
nonwhites. As proof, they offered the comparison between the racial compositions of
the unskilled cannery jobs and of the skilled non-cannery jobs.170 In a 5-4 decision,
the Court rejected that comparison, requiring instead that the plaintiff show
disparate impact in the relevant job markets for each kind of position.171

The Court then went on to endorse the plurality position in Watson in three
ways. First, it required that the plaintiff identify the specific employment practice
that had caused plaintiff’s exclusion and had a disparate impact on a race, color,
religion, sex or national origin group of which plaintiff is a member. Second, it
insisted that the “business necessity” defense does not require that the employment
168 Wards Cove Packing Co. v. Atonio, 490 U.S. 642 (1989).

169 Id. at 646-48.

170 Id. at 650.
171In doing so, it was applying the logic of Hazelwood School District v. United States, 433 U.S. 299
(1977), to disparate impact. Hazelwood School District had held that statistical evidence may be used
to help prove disparate treatment, but that apples must be compared to apples and oranges to
oranges. In determining whether a school district was discriminating against African American
applicants for teaching positions, litigants must present evidence of the proportion of African
Americans in the relevant geographical market who would be qualified teachers. Using the
proportion of all African Americans in the relevant geographic market regardless of whether they
have the credentials necessary to be teachers or the proportion of African American students in the
school system was insufficient.

Applying Hazelwood School District to disparate impact cases was certainly appropriate.
But its usefulness is limited. In disparate impact cases, there is often a fundamental disagreement
about how to define the qualified labor pool. Nevertheless, in Wards Cove it was clear the plaintiffs
had come nowhere near to getting the concept right. They were arguing that the racial composition
of defendants’ unskilled work force was very different from their skilled work force. That was not
helpful. The only evidence that would be helpful is a comparison of the qualified labor pool for each
of the skilled job categories and the actual hires. Although there may have been room for argument
about how to define the qualified labor pool, it was difficult to deny that plaintiffs had failed to define
it plausibly.


Electronic copy available at:

practice at issue be “essential” or “indispensible;” it is enough if it serves some
“legitimate employment goal.” Third, it placed the ultimate burden of proof on the
plaintiff to prove the employment practice does not serve some legitimate
employment goal (once the employer has produced evidence that it does).172

All these points were controversial. But the second point was particularly
controversial. Concerning it, the Court stated:

Though we have phrased the query differently in different cases, it is

generally well established that at the justification stage of such a
disparate-impact case, the dispositive issue is whether a challenged
practice serves, in a significant way, the legitimate employment goals
of the employer . . . A mere insubstantial justification in this regard
will not suffice, because such a low standard of review would permit
discrimination to be practiced through the use of spurious, seemingly
neutral employment practices. At the same time, though, there is no
requirement that the challenged practice be “essential” or
“indispensable” to the employer’s business for it to pass muster: this
degree of scrutiny would be almost impossible for most employers to
meet, and would result in a host of evils . . . .173

Note that the Court seemed to be hinting that discriminatory intent really is
Title VII’s point. It reasoned that too low a business necessity standard “would
permit discrimination to be practiced through the use of spurious . . . employment

The tone of the main dissent was one of outrage. Authored by Justice Stevens (who
had also authored the Beazer opinion) and joined by Justices Brennan, Marshall
and Blackmun, it opposed the majority on all three points. On the issue of
“business necessity,” the dissent argued that “[o]ur opinions always have
emphasized that, in a disparate impact case, the employer’s burden is weighty.”174

172  Wards Cove, 490 U.S. at 659,  
173 Id. (citations omitted).

174Id. at 671 (Stevens, J., dissenting). In order to support its thesis that the standard is very high,
the dissent quotes Dothard in this manner: “Later, we held that prison administrators had failed to
‘rebu[t] the prima facie case of discrimination by showing that the height and weight requirements
are … essential to effective job performance.’” Id.

The latter quote may have been a bit disingenuous. Dothard actually stated that defendants failed to
offer any evidence to “rebu[t] the prima facie case of discrimination by showing that the height and
weight requirements are job-related. These requirements, they say, have a relationship to strength,
a sufficient but unspecified amount of which is essential to effective job performance.” Dothard v.
Rawlinson, 433 U.S. 321,331 (1989). Dothard did not state that the specific height and weight
requirements must be “essential to effective job performance.” On the other hand, Dothard did state


Electronic copy available at:

Stevens asserted that he was “astonished” to read in the majority opinion
that there “‘is no requirement that the challenged practice be … “essential”.’” “This
casual—almost summary rejection of the statutory construction that developed in
the wake of Griggs is most disturbing.”

Coming from the author of the Beazer opinion, which suggested that an
employer is in the clear if there are “legitimate employment goals” that “are
significantly served by—even if they do not require—[the challenged employment
practice], this may seem a bit odd. But it is consistent with the interpretation of
Beazer in appellate decisions Williams v. Colorado School Dist. 11 (1981) and EEOC
v. Rath Packing Co. (1986).175

Stevens’ outrage, however, was selective. He was incensed that the Court
had decided Wards Cove in a way he considered inconsistent with Griggs and its
progeny. But he didn’t mind a bit how obviously and fundamentally Griggs was at
odds with Congress’s intent.

Wards Cove could simply have overruled Griggs by holding lack of

discriminatory intent is an element of plaintiff’s case (or lack of discriminatory
intent is an affirmative defense).176 As the Supreme Court stated in Payne v.

in a footnote that an employment practice with disparate impact “must be shown to be necessary to
safe and efficient job performance to survive a Title VII challenge.” Id. at 331 n. 14.
And while every lawyer knows from McCulloch v. Maryland, 17 U.S. 316 (1819) that “necessary”
does not necessarily mean “essential,” something that is “legitimate” is not necessarily “necessary.”
175 See supra notes 155 and 156 and accompanying text. Stevens had declined to join the plurality
opinion in Watson. Had he done so it would have made a majority. Instead he wrote that he would
prefer to “postpone any further discussion of the evidentiary standards set forth in our prior cases
until after the District Court has made appropriate findings concerning this plaintiff’s prima facie
evidence of disparate impact and this defendant’s explanation for its practice of giving supervisors
discretion in making certain promotions.” Watson v. Fort Worth Bank & Tr., 487 U.S. 977, 1011
(Stevens, J., concurring in the judgment). He also declined to join the opinion of Justice Blackman
(joined by Justices Brennan and Marshall), which criticized the plurality opinion for its treatment of
the business necessity defense. See id. at 1000-11 (Blackman, J., concurring in part and concurring
in the judgment),

176The doctrine of Constitutional avoidance might have been appropriate here. See infra Parts X-XII
(making constitutional argument). As the Supreme Court recently stated in Jennings v. Rodriguez,
“When ‘a serious doubt’ is raised about the constitutionality of an act of Congress, ‘it is a cardinal
principle that this Court will first ascertain whether a construction of the statute is fairly possible by
which the question may be avoided.’” Jennings v. Rodriguez, 138 S. Ct. 830, 842 (2018) (quoting
Crowell v. Benson, 285 U.S. 22, 62 (1932)); see also Edward J. DeBartolo Corp. v. Fla. Coast Bldg. &
Construction Traces Council, 485 U.S. 568, 575 (1988) (“where an otherwise acceptable construction
of a statute would raise serious constitutional problems, the Court will construe the statute to avoid
such problems unless such construction is plainly contrary to the intent of Congress”).


Electronic copy available at:

Tennessee (1991) “when governing decisions are unworkable or are badly reasoned,
‘this Court has never felt constrained to follow precedent.’”177 Many reasons cut in
favor of doing so then.178 Instead, it took a much milder approach. That approach
nevertheless infuriated many civil rights activists.179

VII. The Effect of Civil Rights Act of 1991: Despite the Efforts of the Bush
Administration and its Congressional Allies, the Business Necessity Defense
Remains Very Difficult to Establish:

In 1989, Members of Congress set about to overrule Wards Cove and three
other Supreme Court cases dealing with race and sex issues180--Patterson v. McLean
Credit Union (1989),181 Price Waterhouse v. Hopkins (1989),182 and Martin v. Wilks
177 Payne v. Tenn., 501 U.S. 808, 827 (1991) (quoting Smith v. Allwright, 321 U.S. 649, 665 (1944)).

178It is not just Griggs was inconsistent with both Title VII’s text and its legislative history. It is not
even just that it is internally incoherent given that all or nearly all job qualifications have a
disparate impact on some group; applying disparate impact will only result in the prohibition of

One cannot even justify Griggs by arguing that it advantages women and minorities relative to
others. While at first blush it might seem to, as Part IX of this Article, infra, will demonstrate, there
is empirical evidence that, in the long run, disparate impact liability may make women and
minorities worse off. When employers are prohibited from inquiring whether job applicants have
qualities that they think are important, they may fall back on race and sex stereotypes instead.

In addition, unlike many other Supreme Court precedents, no one had relied to his or her detriment
on the continued viability of Griggs. In that way it is unlike, for example, a tax deduction for
mortgage interest, which induces some individuals to buy a house and incur mortgage interest
payments who otherwise would not have.
179Two years later, Senator Howard Metzenbaum (D.-Ohio) described Wards Cove (along with other
race and sex decisions from the Spring of 1989) by stating that they “shook the foundations of
Federal civil rights law.” See 137 C ONG . R EC . 28,719-20 (1991) (remarks of Senator Howard

180The 101st Congress had strong Democratic majorities in both houses. In the Senate, there were 55
Democrats and to 45 Republicans; the House opened in 1989 with 259 Democrats and 174
Republicans. See 101st United States Congress, WIKIPEDIA.COM, [].

181Patterson v. McLean, 491 U.S. 164 (1989). Patterson was an African American woman who
worked for a small credit union. She brought a lawsuit alleging that she had been subjected to on-
the-job harassment, denied a promotion, and ultimately laid off on account of her race in violation of
the Civil Rights Act of 1866, 42 U.S.C. § 1981. She did not sue under Title VII, likely in part because
Title VII did not provide for damages for emotional suffering and loss of dignity, which in theory
would have been available under § 1981 if it were held to cover racial harassment claims.


Electronic copy available at:

(1989).183 Of these, the damages issue in Patterson was by far the most
consequential to advocates of the proposed legislation.184 Quickly getting around

This claim put the Supreme Court in an awkward position. The problem was that § 1981 (and its
companion provision § 1982) had never been intended by Congress to cover private race
discrimination. Those provisions were intended to guarantee all persons, regardless of race, the
same legal capacity to enter into and enforce contracts and to own and convey property. Such a right
was as against states (which set the rules as to who has the legal capacity to contract and to own and
convey property). These provisions had been badly misread in Jones v. Alfred H. Mayer Co., 392 U.S.
409 (1968) and later by Runyon v. McCrary, 427 U.S. 160 (1976) to be something close to 19th century
versions of the Civil Rights Act of 1964 (except broader and without the 1964 Act’s limitations on

In Patterson, the defendant credit union invited the Court to overrule Jones and Runyon, but the
Court, on the ground of stare decisis, declined to take that opportunity. But it did hold that racial
harassment relating to the conditions of employment in not actionable under § 1981. Rather, it
covered only the discriminatory refusal to form a contract (or the refusal to make a contract on
anything but discriminatory terms) and the right to enforce contracts in court or elsewhere.

This was an unwelcome result from the standpoint of potential plaintiffs, since Title VII’s remedial
provisions were, for reasons discussed infra 186, very limited, especially for plaintiffs whose primary
complaint was harassment. This was particularly so for plaintiffs whose primary complaint was for
sexual harassment, since they would not have been able to sue under § 1981 even if Patterson had
gone the other way. Patterson was nevertheless an important precedent for them, because it
highlighted the issue of Title VII’s limitations.

Providing plaintiffs whose primary complaint was racial or sexual harassment with what they
considered more effective Title VII remedies was almost certainly a more important issue at the time
than the disparate impact issues raised by Wards Cove.

182Price Waterhouse v. Hopkins, 490 U.S. 228 (1989). Price Waterhouse had held that an employer
that had been motivated in part by race, color, religion, sex, or national origin would nevertheless
not be liable if it could establish that it would have made the same decision even if it had not been so

183 Martin v. Wilks, 490 U.S. 755 (1989). Martin was about challenging consent decrees. The white
fire fighters who brought the case had not been parties to the litigation that resulted in a consent
decree between the City of Birmingham, Alabama and African-American fire fighters establishing
goals for the fire department in hiring and promoting African-American fire fighters. The Supreme
Court held that the white fire fighters were not precluded from challenging employment decisions
taken pursuant to the consent decree, which the white fire fighters alleged were racially

For reasons only lawyers and historians tend to recognize easily, the original Title VII limited the

remedies available to plaintiffs to injunctions and lost wages.

The Constitution’s Seventh Amendment guarantees federal litigants the right to a jury trial “[i]n
Suits at common law, where the value in controversy shall exceed twenty dollars.” U.S. CONST.
amend. VII. But not all lawsuits are “at common law.” Eighteenth century English courts were of
several types. In “common law” courts, juries were employed to decide facts, but in courts of equity,
civil juries were not used. The Seventh Amendment merely codified this distinction.


Electronic copy available at:

Patterson in some manner would make a huge difference in the viability of
harassment lawsuits, which up to then were only a tiny portion of the employment
discrimination caseload. Once the Civil Rights Act of 1991 was passed, the number
of harassment cases exploded. During the period the Act was under consideration,
however, it was the disparate impact issues in Wards Cove that seemed to attract
the most controversy.

American courts merged common law and equity courts long ago. The same judges who preside over
suits “at common law” sit in equity when the occasion calls for it. But it is still true that a litigant’s
right to a civil jury depends on whether the lawsuit would traditionally have been viewed as one at
common law or in equity.

While it is an imperfect distinction a good rule of thumb is that equity courts issued injunctions,
including orders for the specific performance of contracts. Common law courts, on the other hand,
could order the defendant to pay damages in the form of money to the plaintiff.

Since Southern juries were not expected to be sympathetic to Title VII, the civil rights
attorneys active in the early 1960s would never have been inclined to ask for damages in the
form of money for things like emotional distress, loss of dignity or for pecuniary losses as a
consequence of discrimination. Such a request would trigger the right to a civil jury on the
part of the defendant. This is the last thing they wanted for their clients. As a result, a
limitation on remedies that allowed only for equitable remedies is something they could
easily agree to. Title VII was thus drafted to allow only for injunctions and lost wages
(which, although involving money, was considered a form of specific performance of a

By the 1980s, things had changed in two ways. Jury trials were no longer considered a
categorical problem and might even have been considered a good thing. Compensatory and
punitive damages were therefore desirable in the view of plaintiffs and their attorneys.
Second, the Supreme Court had held that on-the-job harassment can be a species of
discrimination cognizable under Title VII. Meritor Sav. Bank v. Vinson, 477 U.S. 57 (1986).

It is rare for the victim of sexual harassment to seek an injunction. It is a ham-fisted way to
deal with the problem and will in most cases either fail to make things better or indeed
affirmatively make them worse. It is also rare for the victim of sexual harassment to quit
her job before securing another, so lost wages are not at issue. Only for those victims who
found the harassment so intolerable that they quit without securing a new job or who were
fired because of their refusal to tolerate the harassment were Title VII’s original remedial
provisions useful.

Much more common were the potential litigants who would like to be compensated for the
emotional distress and loss of dignity caused by the harassment or for punitive damages (or
occasionally for consequential damages, such as when a victim quits her job on account of
the stress and therefore loses her home when she is unable to make the payments on the
mortgage). Title VII’s original remedial limitations stood in the way of this.

Patterson was a failed effort to get around those limitations in the context of race (Section
1981 does not apply to sex). After Patterson, it was going to take Congress to change the


Electronic copy available at:

The earliest versions of what eventually evolved into the Civil Rights Act of
1991 were more one-sided than later versions.185 The very first was sponsored by
Senator Edward Kennedy (D-Mass.) and apparently drafted shortly after the
Supreme Court decisions to which it responded. On the issue of disparate impact
liability, it sought to overrule Wards Cove in three ways—(1) the decision’s
requirement that plaintiffs identify the employment practice they wish to challenge
with particularity; (2) its placement of the ultimate burden of proof on plaintiffs to
prove lack of business necessity; and (3) its identification of “business necessity”
with “a legitimate employment goal.”

On the last of these, the bill would have instead defined business necessity
as “essential to effective job performance.”186 This standard appears to have been
intended as a direct response to language in Wards Cove. That opinion stated,
“there is no requirement that the challenged practice be ‘essential’ or ‘indispensible’
to the employer’s business for it to pass muster: this degree of scrutiny would be
almost impossible for most employers to meet, and would result in a host of evils we
have identified above.”

What were the “evils” that the Court had “identified above”? Earlier in the
opinion, the Court had state that “disparate impact liability” could lead to
employers being “haled into court and forced to engage in the expensive and time-
consuming task of defending the ‘business necessity’ of the methods used to select
the other members of his workforce.” It warned: “The only practicable option for
many employers would be to adopt racial quotas, insuring that no portion of their
workforces deviated in racial composition from the other portions thereof; this is a
result Congress expressly rejected in drafting Title VII.” 187

Given this, it is easy to see why some might refer to the bill as a “quota
bill.”188 Either S2104’s drafters disagreed with the Supreme Court’s majority that

185At least at the beginning of this process, supporters of the efforts believed they could muster veto-
proof majorities and hence did not need the cooperation of President George H.W. Bush. David
Lauter, Civil Rights Bill Vetoed by Bush: Job Bias: Congress is Not Likely to Muster the Votes
Needed for an Override. The Measures Supporters Accuse the President of Playing to White
Conservatives, L.A. T IMES (Oct. 23, 1990)
2961-story.html [] (stating that Gray and Thornburgh “push[ed] for a
hard-line stand” and “argued that the six Supreme Court decisions the bill sought to overturn were
correct and should be left alone” and that civil rights leaders “believ[ed] that they could muster the
votes to override a veto”).

186 S. REP. NO. 101-315, at 41 (1990) (italics added).

Wards Cove, 490 U.S. at 452-53. See 42 U.S.C. § 2000e-2(j); see also Watson v. Fort Worth Bank

& Tr., 487 U.S. 977, 992-994, and n. 2 (opinion of O’Connor, J.).

188 See infra note 197.


Electronic copy available at:

an “essential” standard would lead to quotas or they agreed it would lead to quotas
and considered that a good thing.189 It is not easy to believe they were unaware of
the language in the Wards Cove decision.190

A. The Civil Rights Act of 1990 Is Vetoed: Over the next year or so, the
bill was modified in a number of minor ways as critics attempted to weigh in.191 By
October of 1990, it appeared to be headed toward swift passage in both houses of
Congress.192 But Harvard law professor Charles Fried’s penetrating criticism, in a

189See generally 137 C ONG . R EC . 28,643 (“If you will look back at the language used by the Supreme
Court in Wards Cove, you will see that it was the obvious intent of Senator Kennedy’s original bill to
force employers to impose quotas upon themselves, as it used precisely the language that the
Supreme Court said would inevitably result in quotas!”) (remarks of Senator Slade Gorton).

190Either way, suggesting that those who argued that the bill would promote quotas had made the
issue up for political gain seems unfair. 137 C ONG . R EC . 28,868 (1991) (“I do not believe the
President or his advisers ever truly believed these bills were quota bills. They used the quota issue
for political advantage.”) (remarks of Sen. Dennis DeConcini (D.-Ariz.)).

191 Early versions of the bill were criticized by many, including distinguished legal experts and
representatives of employers. See, e.g., Hearings Before the Labor and Human Resources Committee
of the U.S. Senate on S. 2104, 136 C ONG . R EC . D142 (daily ed. Feb. 23, 1990); 136 C ONG . R EC . D156
(daily ed. Feb. 27, 1990); 136 C ONG . R EC . D184 (daily ed. Mar. 1, 1990); 136 C ONG . R EC . D222 (daily
ed. Mar. 7, 1990). Professor Nelson Lund has opined that without the support of the Bush
Administration, their criticisms would “almost certainly” have had little or no effect on the course of
events. See Nelson Lund, The Law of Affirmative Action in and After the Civil Rights Act of 1991:
Congress Invites Judicial Reform, 6 G EO . M ASON L. R EV . 87, 114 n. 141 (1997).

192Id. By the time the 1990 bill was presented to the President, the section dealing with business
necessity was in some ways narrower and in other ways broader than the original formulation. It
read as follows:


(o)(1) The term “required by business necessity’ means—
(A) in the case of employment practices involving selection such as tests, recruitment, evaluations, or
requirements of education, experience, knowledge, skill, ability or physical characteristics, or
practices primarily related to a measure of job performance, the practice or group of practices must
bear a significant relationship to successful performance of the job; or
(B) in the case of other employment decisions, not involving employment selection practices as
covered by subparagraph (A) (such as, but not limited to, a plant closing or bankruptcy), or that
involve rules relating methadone, alcohol or tobacco use, the practice or group of practices must bear
a significant relationship to a manifest business objective of the employer.
(2) In deciding whether the standards described in paragraph (1) for business necessity have been
met, unsubstantiated opinion and hearsay are not sufficient; demonstrable evidence is required. The
court may receive such evidence as statistical reports, validation studies, expert testimony,
performance evaluations, written records or notes related to the practice or decision, testimony of
individuals with knowledge or the practice or decision involved, other evidence as permitted by the
Federal Rules of Evidence, and the court shall give such weight, if any, to such evidence as is


Electronic copy available at:

New York Times opinion essay entitled The Civil Rights Sham of 1990, helped to
slow it down.193 Fried wasn’t timid about expressing his view. He called
characterizations of bill as “the most important civil rights legislation in a quarter-
century” as “public relations flimflam perpetrated by a cabal of overzealous civil
rights plaintiffs’ lawyers.” His own characterization was quite different:

The usual statement of the bill’s high purpose is that it would make it
easier for plaintiffs to win civil rights suits. Though this is hardly
stirring, at least it has the virtue of accuracy. It does not, however,
answer the embarrassing question of whether this greater ease is in
and of itself a good thing.

For instance, what if (as is the case with some provisions of [the bill]) it
makes it so much easier for plaintiffs to win that they can prevail even
over defendants who have done nothing wrong, who have not
(3) This subsection is meant to codify the meaning of “business necessity” as used in Griggs v. Duke
Power Co. … and to overrule the treatment of business necessity as a defense in Wards Cove Packing
Co. v. Atonio . . . .

S. 2104, The Civil Rights Act of 1990 (citations omitted) (vetoed).

Note that while it replaces “essential to effective job performance” with “bear[s] a significant
relationship to successful job performance,” it also restricts the kind of evidence that can be used to
prove that relationship. It also expands the range of activities that can be subjected to disparate
impact liability to things that are not traditionally thought of as employment practices, such as the
decision to close a particular plant or office or to file for bankruptcy. Such an approach would have
allowed courts to second guess employers on any number of business decisions that have a
significant (or perhaps a not-so-significant) effect on employment, such as a decision to go into a
particular line of business or to invest heavily in the advertising of a particular product.

193Charles Fried, The Civil Rights Sham of 1990, N.Y. T IMES , Oct. 4, 1990, at A29. Fried had also
served as Solicitor General under President Reagan.

194 Id. He explained his position more fully this way:

Its supporters explain that the bill reverses a number of Supreme
Court decisions in the 1988 term unfavorable to civil rights
plaintiffs. For too long, civil rights groups have been able to count
on judges—mainly lower Federal court judges, a majority of whom by
1981 had been appointed by President Jimmy Carter—to bend, stretch
and distort the law in favor of more and more ''creative'' civil
rights theories.

Ever since the Bakke case in 1978, which limited the use of racial
preferences, and Washington v. Davis in 1976, which ruled that only
intentional discrimination violated the Constitution, the Supreme
Court has been insisting on a more measured and lawful approach to
civil rights.


Electronic copy available at:

Fried recommended that Bush “veto this bill in its present form.” But it may
not have taken a lot of persuading. At least according to news reports, President
George H.W. Bush’s top legal advisors—White House Counsel Boyden Gray and
Attorney General Richard Thornburgh—were both strongly against the bill. In any
event, Bush followed Fried’s advice, and an effort to override that veto failed.195 In
his veto message, he condemned the bill as “a maze of highly legalistic language,”
which will “introduce the destructive force of quotas into our Nation's employment
system.”196 He went on:

Primarily through provisions governing cases in which em-

ployment practices are alleged to have unintentionally caused the
disproportionate exclusion of members of certain groups, [the bill]
creates powerful incentives for employers to adopt hiring and pro-
motion quotas. These incentives are created by the bill's new and
very technical rules of litigation, which will make it difficult for
employers to defend legitimate employment practices. In many
cases, a defense against unfounded allegations will be


In the series of 1988 cases, the Supreme Court made plainer than ever
that equal justice under law applies to civil rights defendants as
well as plaintiffs. I suggest it is that phenomenon that explains a
bill that indiscriminately seeks to reverse every ruling, no matter
how trivial (one of them written by Justice Brennan!), that might
possibly be seen as unfavorable to civil rights plaintiffs.

The great principle of [the bill], then, is just a

counter-message from Congress to the Supreme Court: Keep your hands off!
But why is this a message that President Bush would want to send?


195Bush referred to the bill as a “quota bill.” See Ann DeVoy, Bush Vetoes Civil Rights Bill, W ASH .
P OST (Oct. 23, 1990),
rights-bill/cd68a6c4-8529-471a-b4f7-08c26cf65ac0/ [].

196Message from the President Returning Without My Approval S. 2104, The Civil Rights Act of
1990, S. DOC. NO. 35,

197 Id. at 2. Among the bill’s other defects, Bush pointed out that “the plaintiff often need not even
show that any of the employer's practices caused a significant statistical disparity . . . . [T]he
employer's defense is confined to an unduly narrow definition of ‘business necessity’ that is
significantly more restrictive than that established by the Supreme Court in Griggs and in two
decades of subsequent decisions,” Bush wrote. “Thus unable to defend legitimate practices in court,


Electronic copy available at:

Some thought that with midterm elections only a couple of weeks ahead,
Bush and Congressional Republicans would pay a political price for that veto.
Others thought his veto was cynically designed to and would win them votes. Some
probably entertained both irreconcilable beliefs at the same time.198

Perhaps Bush did pay a price. Or perhaps his stance helped Congressional
Republicans. It is difficult to tell. Republicans did lose one seat in the Senate and
nine in the House of Representatives. That was fewer seats lost than what was
usual for an incumbent President during a midterm elections. On the other hand,
that may be because Bush started out with fewer seats held by Republicans in
Congress. He didn’t have as many to lose.199

B. The Civil Rights Act of 1991:

For whatever reason, the Bush White House and both parties in
Congress, made real efforts to compromise in 1991.200 But negotiations
employers will be driven to adopt quotas in order to avoid liability.” Id. See also Carvin, supra note

198 Compare DeVoy, supra note 195 (Sen. Edward Kennedy (D-Mass.) said Bush’s veto shows “that
he is more interested in appeasing extremists in his party than in providing simple justice”) with
DeVoy, supra note 195 (“The rhetoric may be gentler and kinder, but the policies of George Bush are
no less dangerous and regressive than those of Ronald Reagan and Ed Meese,” said Ralph Neas,
executive director of the Leadership Conference on Civil Rights.”).

199Mark P. Petracca, Letter: Midterm Congressional Elections, L.A. T IMES (Nov. 17, 1990), [

200According to Roger Clegg, the breakthrough is thought by some to have been related to the
confirmation of Supreme Court Justice Clarence Thomas, who had been strongly opposed by the
traditional hearings held shortly before the passage of the Act and who, late in the confirmation
process had been accused of the sexual harassment of one of his subordinates:

The temptation is strongest to see the connection as explaining some final concession
by the administration. President Bush was being the magnanimous patrician in the
wake of his victory, the explanation goes, and the WASP President was never one with
an instinct for the jugular on this issue anyway, and only too eager to wash his hands
of the groups’ blood after the unseemliness of the Thomas-Hill fight. For good
measure, this theory also points to the ascendancy of David Duke’s candidacy for
governor of Louisiana at the same time: President Bush could simply not abide
another race-charged national battle. Besides, he owed Senator Danforth after the
Thomas confirmation. So he told his lawyers to cut the best deal they could and be
done with it.

On the other side of the ledger were the problems being faced by the traditional civil rights
advocates, who had arguably burned their bridges with Justice Thomas’s former employer and
advocate Senator Danforth. Danforth had been trying to broker a deal between civil rights groups


Electronic copy available at:

nevertheless wore on for the better part of the year until mutually
acceptable ways to deal with Patterson,201 Price Waterhouse,202 and Martin203
and the Administration. After the Thomas debacle, his patience with them had likely worn thin.
Moreover, as Clegg stated:

The “quota bill” was turning into a public relations debacle [for traditional civil rights
groups]. An internal survey by the Leadership Conference on Civil Rights showed that
Americans largely viewed the groups as special pleaders for preferential treatment for
minorities, rather than principled defenders of fairness; the results of [the] survey, to
the groups’ embarrassment, were leaked. And what if, as they had hoped to do, the
administration and its allies at some point succeeded in attaching a dramatically
scaled-back version of the bill as a rider to some piece of popular legislation, forcing an
up-or-down vote? The passage of such a bill might end decisively any hopes for the
more sweeping changes the groups’ bill contemplated. So the groups faced considerable
pressure to strike a deal, too.

Roger Clegg, Introduction: A Brief Legislative History of the Civil Rights Act of 1991, 54 L A . L. R EV .
1459, 1469-70 (1994).

201 See supra notes 181 and 184. The crown jewel of the bill was its expansion of remedies available
to successful plaintiffs in cases involving intentional discrimination (not cases involving disparate
impact). The Civil Rights Act of 1991 amended Title VII to allow for compensatory damages
representing “future pecuniary losses, emotional pain, suffering, inconvenience, mental anguish, loss
of enjoyment of life, and other nonpecuniary damages” and also for punitive damages. The combined
award for compensatory and punitive damages available for each plaintiff was subject to caps. A
business with 15 to 100 employees was subject to a cap of $50,000. For a business of 101 to 200
employees, the cap was $100,000; for 201 to 500 employees, $200,000; and for more than 500
employees, $300,000. 42 U.S.C. §1981a.

See supra note 182. Price Waterhouse was dealt with by inserting the following to 42 U.S.C. §


(B) On a claim in which an individual proves a violation under section

2000e-2(m) and a respondent demonstrates that the respondent would
have taken the same action in the absence of the impermissible
motivating factor, the court—

(i) may grant declaratory relief, injunctive relief (except as provided

in clause (ii)), and attorney’s fees and costs demonstrated to be
directly attributable only to the pursuit of a claim under section
2000e-2(m) of this title; and
(ii) shall not award damages or issue an order requiring any
admission, reinstatement, hiring, promotion, or payment,
described in subparagraph (A).
See supra note 183. Martin was dealt with by inserting the following subsection into 42 U.S.C. §


(n)(1)(A) Notwithstanding any other provision of law, and except as provided in paragraph (2), an
employment practice that implements and is within the scope of a litigated or consent judgment or
order that resolves a claim of employment discrimination under the Constitution or Federal civil
rights laws may not be challenged under the circumstances described in subparagraph (B).


Electronic copy available at:

could be arrived at and the controversy over Wards Cove and disparate
impact liability could be papered over.204

The disagreement over Wards Cove and disparate impact proved in

some ways intractable. Bush Administration lawyers and their
Congressional allies believed Griggs to be an extraordinary deviation from
the intent of Congress in 1964. And, as I believe I have demonstrated in
Part II, they were right about that. They also believed that a business
necessity defense that is in any way difficult to establish creates an
incentive to engage in proportional hiring or quotas. And they were right
about that too. Indeed, it is not clear that their Congressional opponents
would have disagreed. Nevertheless, those opponents were equally firm in


(B) A practice described in subparagraph (A) may not be challenged in a claim under the
Constitution or Federal civil rights laws—

(i) by a person who, prior to the entry of the judgment or order described in subparagraph
(A), had—

(I) actual notice of the proposed judgment or order sufficient to apprise such person
that such judgment or order might adversely affect the interests and legal rights
of such person and that an opportunity was available to present objections to
such judgment or order by a future date certain; and
(II) a reasonable opportunity to present objections to such judgment or order; or

(ii) by a person whose interests were adequately represented by another person who had
previously challenged the judgment or order on the same legal grounds and with a
similar factual situation, unless there has been an intervening change in law or fact.
(2) Nothing in this subsection shall be construed to—

(A) alter the standard for intervention under rule 24 of the Federal Rules of Civil
Procedure or apply to the rights of parties who have successfully intervened
pursuant to such rule in the proceeding in which the parties intervened;
(B) apply to the rights of parties to the action in which a litigated or consent judgment
or order was entered, or of members of a class represented or sought to be
represented in such action, or of members of a group on whose behalf relief was
sought in such action by the Federal Government;
(C) prevent challenges to a litigated or consent judgment or order on the ground that
such judgment or order was obtained through collusion or fraud, or is transparently
invalid or was entered by a court lacking subject matter jurisdiction; or
(D) authorize or permit the denial to any person of the due process of law required by
the Constitution.
(3) Any action not precluded under the subsection that challenges an employment consent
judgment or order described in paragraph (1) shall be brought in the court, and if
possible before the judge, that entered such judgment or order. Nothing in this
subsection shall preclude a transfer of such action pursuant to section 1404 of title 28.

204 See Clegg, supra note 200.


Electronic copy available at:

believing that the business necessity defense needed to be difficult to
establish in order to achieve what they viewed as justice for plaintiffs.

Ultimately, they were able to agree only because the language they
adopted was being interpreted differently by each side. But unlike many of
the failures to achieve “a meeting or the minds” that have given rise to
contract claims for centuries, the players in this drama were aware that
their interpretations differed.205 Each side evidently hoped that courts
would eventually adopt its interpretation rather than the interpretation of
the other side of the controversy. They rolled the dice.

The Act added the following section to Section 703 of Title VII:

(k)(1)(A) An unlawful employment practice based on disparate impact

is established under this title only if—

(i) a complaining party demonstrates that a respondent uses a

particular employment practice that causes a disparate impact on the
basis of race, color, religion, sex, or national origin and the respondent
fails to demonstrate that the challenged practice is job related for the
position in questions and consistent with business necessity; or

(ii) the complaining party makes the demonstration described in

subparagraph (C) with respect to an alternative employment practice
and the respondent refuses to adopt such alternative employment

(B)(i) With respect to demonstrating that a particular employment

practice causes a disparate impact as described in subparagraph (A)(i),
the complaining party shall demonstrate that each particular
challenged employment practice causes a disparate impact, except that
if the complaining party can demonstrate to the court that the elements
of a respondent’s decisionmaking process are not capable of separation
for analysis, the decisionmaking process may be analyzed as one
employment practice.

(ii) If the respondent demonstrates that a specific employment practice

does not cause the disparate impact, the respondent shall not be
required to demonstrate that such practice is required by business

See, e.g., Raffles v. Wichelhaus, 159 Eng. Rep. 375 (Ex. 1864) (neither party bound to other’s

meaning when they disagree as to which Peerless ship was meant).


Electronic copy available at:

(C) The demonstration referred to by subparagraph (A)(ii) shall be in
accordance with the law as it existed on June 4, 1989, with respect to
the concept of “alternative employment practice.”

(2) A demonstration that an employment practice is required by

business necessity may not be used as a defense against a claim of
intentional discrimination under this title.

(3) Notwithstanding any other provision of this title, a rule barring

the employment of an individual who currently and knowingly uses or
possesses a controlled substance, as defined in schedules I and II of
section 102(6) of the Controlled Substances Act (21 U.S.C. 806(6)),
other than the use or possession of a drug taken under the supervision
of a licensed health care professional, or any other use or possession
authorized by the Controlled Substances Act or any other provision of
Federal Law, shall be considered an unlawful employment practice
under this title only if such rule is adopted or applied with an intent to
discriminate because of race, color, religion, sex, or national origin.

(b) No statements other than the interpretive memorandum appearing

at Vol. 137 Congressional Record S 15276 (daily ed. Oct. 25, 1991)
shall be considered legislative history of, or relied upon in any way as
legislative history in construing or applying, any provision of this Act
that relates to Wards Cove—Business
necessity/cumulation/alternative business practice.

A few things are clear from this. For example, the Bush
Administration and its allies in Congress clearly had agreed to place the
burden of proving business necessity on the defendant. The Wards Cove
innovation of making the defendant’s burden simply one of providing
evidence of business necessity and thus shifting the ultimate burden on the
plaintiff to prove lack of business necessity had disappeared.206

Other things are not so clear. The provision dealing with Wards
Cove’s requirement that plaintiffs specify the particular employment
practice that caused their exclusion and demonstrate its disparate impact
may be viewed as a win for the Bush Administration. The Wards Cove
requirement remains in effect except when it is impossible to sort things out
with precision. What that means is up for grabs. Moreover, all the
ambiguities initially inherent in that issue in 1989 remain ambiguous.207

206 See supra Part VI(F).

207 See supra Part VI(F).


Electronic copy available at:

One of the most interesting differences between the vetoed Civil
Rights Act of 1990 and the Civil Rights Act of 1991 as signed into the law
was the actual law’s statement that an “unlawful employment practice based
on disparate impact is established under this title only if ….” (Italics
supplied.) The earlier proposal had stated that an “unlawful employment
practice based on disparate impact is established under this title when ….”
(Italics supplied).208

Congress thus made it clear that it was not committing itself to a

particular conception of a disparate impact cause of action. The continued
evolution of that approach to Title VII appears to have been expected.209
Further refinements of the prima facie case and further affirmative defense
could be added. The statue did not purport to resolve all issues as to what
constitutes a prima facie case for disparate impact liability and what
constitutes a permissible defense.

Interestingly, if there was any effort to amend the other sections of

Title VII bearing on disparate impact—notably the prohibitions and the
section requiring intent before judicial remedies could be had—they were
successfully resisted. Reconciling that language with the new section will
not be easy. This raises the possibility, for example, that disparate impact
liability can still be interpreted in ways to avoid the constitutional issues
raised in Parts X, XI and XII.

Meanwhile, the most controversial issue—the meaning of “business

necessity”—is the one that was not resolved. The agreed upon language of
the statute was opaque.

The interpretative memorandum referred to in the statute was very

short. It helped a little perhaps, but not a lot. It directs our attention back
to Griggs and Dothard. Although it does not name them specifically, it also
directs us to Albemarle Paper, Beazer and (to the extent it states the law)
Watson. Perhaps we are to figure out what the law was just before Wards
Cove was decided and apply it. In full, the memorandum states:

Interpretive Memorandum

The final compromise on S. 1745 agreed to by several Senate sponsors,

including Senators Danforth, Kennedy, and Dole, and the
208 See S.2104, 101st Cong. §4(k)(vetoed).

209 Indeed, one could argue that Congress didn’t commit itself to a disparate impact cause of action at
all. Rather, if one exists, it must comply with Section 703(k).


Electronic copy available at:

Administration states that with respect to Wards Cove—Business
necessity/cumulation/alternative business practice—the exclusive
legislative history is as follows:

The terms `business necessity' and `job related' are intended to reflect
the concepts enunciated by the Supreme Court in Griggs v. Duke
Power Co., 401 U.S. 424 (1971), and in the other Supreme Court
decisions prior to Wards Cove Packing Co. v. Atonio, 490 U.S. 642

When a decision-making process includes particular, functionally-

integrated practices which are components of the same criterion,
standard, method of administration, or test, such as the height and
weight requirements designed to measure strength
in Dothard v. Rawlinson, 433 U.S. 321 (1977), the particular,
functionally-integrated practices may be analyzed as one employment

If Section 703(k) of the 1991 Act was intended to cause Members of

Congress to remain silent, it failed. But it is doubtful its drafters had that
thought; trying to prevent politicians from speaking seldom works. The
provision seems to be drafted instead to discourage courts from considering
such statements.211 In any event, shortly after it became clear the bill would
go through and be signed by the President, Republicans started
congratulating President Bush on standing firm while Democrats took the
position that he had caved. This means precisely nothing. It is exactly what
one would expect if the speakers disagreed on how to interpret the provision,
knew that they disagreed, and each hoped that courts would resolve the
210 137 C ONG . R EC . 28680 (1991).

211Senator John Danforth (R-Mo.) points out the wisdom of not taking what individual Senators put on
the record about the proper interpretation of an Act:

Now, with respect to the question of legislative history, I tried to address this,
probably not very skillfully, this morning. It is not unusual in the Senate for
various people to try to create legislative history and affect the interpretation
that a court might have of a statute. It is not unusual. Every time we have a
tax bill, we try to do that.

I can remember one night literally following one of my colleagues around the
floor of the Senate for fear that he would slip something into the
Congressional Record, and I would have to slip something else into the
Congressional Record. It is not unusual.

137 C ONG . R EC . 28877 (1991) (remarks of Sen. Danforth).


Electronic copy available at:

ambiguity to favor their side. Furthermore, it is also what one would expect
if they didn’t realize that they disagreed. Still the statements are a good
indicator of how each side hoped the language would be interpreted.

Senator Slade Gorton (R.-Wash.) was one of the most engaged

members of the Senate on the issues. He saw the provision as simply
eliminating Wards Cove as a precedent entitled to weight as such under the
doctrine of stare decisis. As he put it, “At most, the Senate has turned the
clock back to June 4, 1989, the day before the Wards Cove decision was
rendered.” While he was aware that his Democratic colleagues viewed the
Wards Cove formulation of the business necessity standard as having been
overturned, he stated that in his view, “the Supreme Court is free to render
the exact same substantive standard for disparate impact as it did in Wards
Cove.” He stated further:

[I am] of the firm belief that this compromise does not preclude the
Supreme Court from adopting a standard for disparate impact
cases as Justice White wrote in the Wards Cove decision. All we
have done is command the Supreme Court to reexamine that
standard de novo.212

Gorton’s statement was broadly consistent with a section-by-section analysis

that was inserted into the Congressional Record a few days later and in which he
joined.213 Representing the views of the Administration and a number of Republican
Senators, including Gorton, it emphasizes Beazer as the last majority decision before
Wards Cove and argues that the position taken in Wards Cove simply repeated the
Beazer approach. It noted:

[T]he present bill has codified the “business necessity” test

employed in Beazer and reiterated in Wards Cove. The language of
the bill is thus plainly not intended to make that test more onerous
for employers to satisfy than it had been under current law.214

Several Republican Senators further emphasized that the 1991 Act

was not an attempt to freeze the law as to disparate impact liability.
Rather it was intended to allow disparate impact liability law to
continue to evolve. Senator Pete Domenici (R.-N.M.) stated that “the
212 137 C ONG . R EC . 28644 (1991) (emphasis added).

213  Section-by-Section
Analysis Representing the Views of the Administration and Senators Burns,
Cochran, Dole, Garn, Gorton, Grassley, Hatch, Mack, McCain, McConnell, Murkowski, Simpson,
Seymour, and Thurmond, 137 C ONG . R EC . 29,035-40 (1991).
214 Id. at 29,038.


Electronic copy available at:

original version of the bill as well as the House version … attempted,
vainly I believe, to codify a rapidly evolving field of court-developed law
and to freeze it into a statutory straight jacket.”215 Senator Gorton was
in agreement with Domenici’s sentiment:

I also believe, as does the President and my distinguished

colleague from Utah, that it is no longer a quota bill because it
has abandoned at long last the intent to redefine the definition
of business necessity, as that definition has been elaborated by
the Supreme Court over the course of more than the past 20

Far from attempting a statutory definition of the term business

necessity, we have now left that definition in the hands of the
courts, where it belongs. That is a profound, significant change .
. . . .216

This is broadly consistent with the shift in the introductory phrase of Section 703(k)
from an unlawful employment practice based on disparate impact is established “when”
to it is established “only if.”217 Maneuvering room for the courts to add further
refinements and limitations appears to be preserved. On the other hand, the
“purposes” section of the text states that Congress was codifying the concepts of
“business necessity” and “job relatedness.”218

Democratic Senators appear to view Wards Cove as having been definitively

overruled by the 1991 Act. Senator Kennedy, for example, stated flatly that for nearly
two years, he and many of his colleagues had been attempting to put together a bill
that would overrule Wards Cove. “This amendment,” he said, “would do that.”219

In support of Kennedy’s view is the fact that it is clear from the text of the
Act that Congress disapproved of Wards Cove. Its findings specifically include “the
decision of the Supreme Court in Wards Cove Packing v. Atonio … has weakened

215 137 C ONG . R EC . 28715 (1991) (remarks of Sen. Domenici).

216 137 C ONG . R EC . 28642 (1991) (remarks of Sen. Gorton).

217 See supra note 208 and accompanying text.

218The purposes articulated in the enactment include, “to codify the concepts of ‘business necessity’
and ‘job related enunciated by the Supreme Court in Griggs v. Duke Power Co. … and in other
Supreme Court decisions prior to Wards Cove Packing v. Atonio ….” Civil Rights Act of 1991, Pub.
L. No. 102-166, §3(2), 105 Stat. 1071 (1991). citations omitted).

219 137 C ONG . R EC . 28878 (1991) (remarks of Sen. Kennedy).


Electronic copy available at:

the scope and effectiveness of Federal civil rights protections.”220 But that finding
(as well as Kennedy’s statement) can be explained by the Act’s treatment of the
burden of proof issue, which clearly overrules Wards Cove.

In the end, those who saw Wards Cove as the logical extension of previous
decisions would tend to side with the Bush Administration and Congressional
Republicans; those who saw it as a departure would tend to side with Congressional

C. The Fallout from the Civil Rights Act of 1991: As Oklahoma City
University law professor Andrew Spiropoulus has stated, “[T]hose who contend that
the Act establishes a strict business necessity defense and those who argue that the
Act enacted the more lenient business necessity defense both have plausible
arguments for their interpretations founded in two different lines of Supreme Court
precedent . . . . [N]either side can conclusively show that their interpretation was
embodied in the Act.”221

Since the passage of the 1991 Act, however, some commentators have continued
to argue for an extraordinarily high standard. A good example is William & Mary
law professor Susan Grover, who has stated:

The overarching issue continues to be whether the term “necessity” in

the business necessity defense literally requires that the
discriminatory practice be essential to the continued viability of the
business, or whether it requires something less. This Article argues for
the former interpretation.222

The standard that Grover argues for is presumably even higher than the
“essential to effective job performance” standard pushed for in the early
drafts of what became the ill-fated 1990 Act. She states:

That defense [the business necessity defense] should require an

employer to prove that its discriminatory practice is essential to its
continued operation. Under the structure created by the 1991 Act, an
employer must prove that the goal it seeks to achieve through the
220 Civil Rights Act of 1991 §2(2)(citations omitted).

221Andrew Spiropoulos, Defining the Business Necessity Defense to the Disparate Impact Cause of
Action: Finding the Golden Mean, 74 N.C. L. R EV . 1479, 1483-85 (1995).

222Susan Grover, The Business Necessity Defense in Disparate Impact Employment

Discrimination Cases, 30 G A . L. R EV . 387 (1996).


Electronic copy available at:

practice is crucial to its continued viability and, in turn, that the
practice selected is crucial to the achievement of that goal . . . .
Alternatives to this standard thwart the objectives of Title VII by
disproportionately favoring defendants.223

Professor Grover clarifies her use of the term “continued viability” by stating that it
means that “relinquishing the discriminatory practice will compel the employer to
cut back on its business, resulting in employee layoffs.”224

That is quite a standard. It is not enough that the job at issue will be badly
performed. It has to be bad enough to compel layoffs or bankruptcy. Imagine
applying it to the area of criminal background checks.225 Hiring an individual with
an extensive felony conviction record into a sensitive job increases the risk to the
employer, her employees, her customers and her property. But it will seldom be the
case that she can prove ex ante that it “will compel” her “to cut back” on her
business, “resulting in employee layoffs.”

The decisions of the federal courts have not been as radical as Glover.
Nevertheless, they have tended to come closer to Senator Kennedy’s view of
things than to Senator Gorton’s. In Bradley v. Pizzaco of Nebraska, Inc.
(1993), the Eighth Circuit Court of Appeals held that under the 1991 Act it
should apply pre-Wards Cove precedent. While the Interpretive
Memorandum directed courts only to Supreme Court precedent, the Court in
Bradley, not surprisingly, included Eighth Circuit interpretations of
Supreme Court precedent in its analysis. It held that the defendant can only
carry its burden to prove “business necessity” by showing it has a
“compelling need” for the employment practice at issue.226 Similarly, in
Nash v. Jacksonville (1995), the District Court for the Middle District of
Florida held that after the 1991 Act, “’business necessity’ is an affirmative
defense on which the defendant bears the burden of persuasion and
‘business necessity’ really means ‘necessity.’”227

223 Id. at 429-30.

224 Id. at n. 5.

225 See infra Part X.

226 7 F.3d 795, 797-99 (8th Cir. 1993).

227895 F. Supp. 1536, 1545 (M.D. Fla. 1995), aff’d without opinion, 85 F.3d 643 (11th Cir. 1996)
(unpublished table decision).


Electronic copy available at:

If anything, the standard for proving “business necessity” got tougher as time
passed. In Lanning v. Southeastern Pennsylvania Transportation Authority (1999),
the Third Circuit Court of Appeals was called upon to decide what constitutes
“business necessity” in the context of a physical fitness test for applicants for job as
police officers. The court held that a cut-off score with a disparate impact “is
impermissible unless shown to measure the minimum qualifications necessary for
successful performance of the job in question.”228

Such a standard, if taken seriously, is quite extraordinary. Employers seek

the best-qualified employees available to them, not the minimally qualified (i.e. the
one an employer would be willing to accept if it has no better options). 229 If the
“business necessity” defense is set too high, then, as the Watson plurality and the
Wards Cove Court warned, disparate impact liability is simply a call for quotas (or
for random hiring). Savvy employers will naturally opt to hire by the numbers
rather than face liability. Alternatively, rather than accept a minimally qualified
workforce, some may seek greater control over their affairs by moving their
facilities elsewhere.

The “minimum qualifications” test has nevertheless shown up repeatedly in

the case law in recent decades. The Third Circuit Court of Appeals has adhered to
it on two occasions,230 and other courts have applied it as well.231

The opinion in Easterling v. Connecticut Department of Correction (2011), is

particularly interesting.232 In it, the District Court for the District of Connecticut
stated that Lanning’s minimum qualifications rule is one of two judicial efforts to
flesh out the business necessity defense in the wake of the 1991 Act. The other is
what the Easterling court referred to as the “Significantly Correlated Standard,”
which it identifies as the approach taken by the Second Circuit Court of Appeals in
228181 F.3d 478, 481, 489 (3d Cir. 1999) (italics added), cert. denied, 538 U.S. 1131. Judge Weis filed
a strong and detailed dissent.

229 See Carvin, supra note 125.

230NAACP v. North Hudson Reg’l Fire & Rescue, 665 F.3d 464, 477 (3d Cir. 2011) (residency
requirements for fire and rescue employees); El v. Southeast Penn. Transp. Auth., 479 F.3d 232, 242
(3d Cir. 2007) (citing Lanning and noting that the minimum qualifications test is consistent with the
fact that Congress continues to call the test “business necessity” not “business convenience” or some
weaker term).

231See, e.g., Easterling v. Conn. Dep’t of Corr., 783 F. Supp. 2d 323 (D. Conn. 2011) (citing and
applying Lanning as well as an alternative standard); Green v. Town of Hamden, 73 F. Supp. 2d 192
(D. Conn. 1999) (citing Lanning and stating “[n]either the CPS validation study nor the Town’s
decisional process resulting in the 60% cutoff figure can be seen as reasonably delineating between
applicants who were competent to perform the entry level job and those that were not”).

232 Easterling, 783 F. Supp. 2d 323.


Electronic copy available at:

Gulino v. New York State Education Department (2006).233 Easterling applied both
standards and produced the same result each time. It also considered the
possibility that both standards apply.

The case concerned a four-part physical fitness test that applied to applicants
for a job as a correctional officer at the Connecticut Department of Correction. The
cut-offs for each part were adjusted for both sex and age. Thus, for example, a male
applicant age 50 or over was required to do better than a female applicant age 30-39
on the 1.5 mile run in order to get a job, but was given slightly more leeway than a
female applicant age 21-29. The plaintiff in the case, a female applicant, had passed
all tests except the 1.5 mile run. She brought a Title VII action under a disparate
impact theory.

In applying the minimum qualifications test, the court stated:

The parties agree that the 1.5 mile is a test that measures an
individual’s aerobic capacity. … The [Department of Correction] cannot
plausibly  argue  that  a  time  of  12:25  for  21-­‐29  year-­‐old  men  is  a  valid
predictor of the aerobic capacity minimally necessary for successful
completion of the tasks of an Correction Officer, if [it] also permitted 21-
29 year old women to complete the 1.5 mile run in 14:49, and 50 year-
old women to complete the 1.5 mile run in 17:14. By definition, cutoff
times that vary by gender and age cannot represent a measure of the
minimum aerobic capacity necessary for successful performance as a
[Correction Officer]. Only a single cutoff time could meet this

The Easterling court’s application of the “Significantly Correlated

Standard” was different, but not less demanding. In fleshing out that
standard, the court traced its lineage to Albemarle Paper Co. v. Moody. In
doing so, it implicitly took the position that Beazer did not overrule Albemarle
Paper and that Albemarle Paper and its extraordinarily rigorous
requirements for establishing the validity of a particular employment practice
are still essentially good law. According to Easterling, the Court in Albemarle
Paper “embraced the EEOC Guidelines which require that tests be ‘shown, by
professionally acceptable methods, to be ‘predictive of or significantly
correlated with important elements of work behavior which comprise or are
relevant to the job or jobs for which candidates are being evaluated.”’”235

Easterling, 783 F. Supp. 2d at 335 (citing Gulino v N.Y. State Educ. Dep’t., 460 F.3d 361 (2d Cir.

234 783 F. Supp. 2d at 335.

235 Id. at 337(quoting Albemarle Paper, 422 U.S. at 431, quoting 29 C.F.R. § 1607.4(C)).


Electronic copy available at:

Easterling initially takes the position that the Significantly
Correlated Standard is the one that it must apply in the Second Circuit.
Toward the end of the opinion, however, it discusses the possibility that both
standards apply—that the Significantly Correlated Standard applies to the
job-relatedness requirement for disparate impact and the Minimum
Qualifications Standard is a separate requirement relating to a separate
requirement of business necessity. Such an issue, however, did not need to be
resolved, since the job qualification at issue failed in any event.

Meanwhile, the EEOC itself does not seem to have even

considered the possibility that the 1991 Act (or Beazer) is inconsistent with
Albemarle Paper or Uniform Guidelines on Employee Selection Procedures.
The latter remain firmly in place. Its 2012 Guidance cites Griggs, Albemarle
Paper and Dothard as authority, but not Beazer.

Neither the post-1991 Act judicial precedent nor the EEOC’s

interpretation of the 1991 Act have been held by the Supreme Court to state
correctly. But they give a sense of how the law is being applied today.

VIII. Criminal Background Checks: An Example of Disparate Impact

Liability in Action.

When employment discrimination attorneys think of disparate impact

liability, they most often think of written tests (as in Albemarle Paper) or physical
tests (as in Dothard). These have been the most common applications of the
Uniform Guidelines on Employee Selection Procedures.236 In the last decade or so,
however, it is criminal background reports that has received the greatest attention
from the EEOC.

It is understandable why EEOC officials might want to do something to help

re-integrate more ex-offenders into the outside world. This is important work that
needs to be done. But it is a good example of how the EEOC has used Title VII to
pursue goals that have little to do with the subject matter entrusted to it by the 88th

Moreover, there are good ways and bad ways to accomplish that goal. Quite
apart from the rule of law issues—whether Griggs properly interpreted Title VII,
how the 1991 Act should be interpreted, and even whether disparate impact
236The Uniform Guidelines for Employee Selection Procedures themselves apply to much more than


Electronic copy available at:

liability is constitutionally infirm—applying disparate impact liability to this
context is bad policy.

One better way to encourage the re-integration of ex-offenders is the Work

Opportunity Tax Credit Program.237 Under it, employers who choose to hire a
qualified ex-offender get a small tax credit. No one is forced or threatened with
litigation to participate. Those businesses that perceive themselves as benefiting
from the arrangement will be the ones that take advantage of it. Eligibility and
other ground rules are clearly defined, so no one need be confused about what the
law permits.

Employers have many things to worry about when they hire. All of them are
vulnerable. If they make a wrong choice, they can wind up with someone who is
undependable, difficult to work with or incompetent. Recidivism rates make it clear
that hiring an ex-offender carries with it serious risks.238

A bad employee can steal from the employer, harass fellow employees, drive
away the customers, and cause devastating harm.239 Moreover, employers can end
237See Small Business Job Protection Act of 1996, Pub. L. No. 104-188, 110 Stat. 1755 (1996). The
Work Opportunity Tax Credit Program was originally set to expire on September 30, 1997. §1201,
110 Stat. at 1772. It has been revised and extended on several occasions, most recently in the
Protecting Americans From Tax Hikes Act of 2015, enacted as Division Q of the Consolidated
Appropriations Act 2016, Pub. L. No. 114-113, 129 Stat. 2242, 3056 (2015).

238According to Bureau of Justice statistics, “Five in 6 (83%) of state prisoners released in 2005
across 30 states were arrested at least once during the 9 years following their release.” The average
number of re-arrests is five. See Mariel Alper, Matthew R. Durose & Joshua Markman, SPECIAL
Recidivism rates were higher for males (84.2%) than for females (76.8%), and higher for those 24
years old or younger (90.1%) than for those 40 and older (76.5%). But they were high in all
categories. Id. at 6 tbl.3.

It is sometimes said that an ex-offender who has not yet been re-arrested after eight years is unlikely
to be re-arrested for the first time in the ninth year. And in some sense this is true. But this is
sometimes over-stated as a result of a misreading of the statistics. The Bureau of Justice found that
only 1% of those released were re-arrested for the first time in their ninth year. But that is 1% of the
total group, not 1% of those who had not been re-arrested. Since the vast majority of the group had
already been re-arrested one or more times, the likelihood that an ex-offender who had not already
been re-arrested would be re-arrested in the ninth year is between 5% and 6%. That is still low, but
it is nevertheless higher than the population without a felony record.

239See. e.g., Roberto Acosta, Accountant Embezzled Over $1 Million from Port Huron Charity, Police
Say, MLIVE.COM (Flint) (April 1, 2019),
embezzled-over-1-million-from-port-huron-charity-police-say.html [];
Mark Guarino, Kristine Phillips, and Frances Stead Sellers, Man Accused in Aurora Mass Shooting
Had Been Convicted for Beating Girlfriend with a Baseball Bat, WASH. POST (Feb. 17, 2019),
after-being-fired-illinois-police-say/ []; Pat Reavy, Charges: Woman


Electronic copy available at:

up legally responsible for the actions of their employees under doctrines of negligent
hire or supervision, respondeat superior and actual or apparent authority. The need
to fire an employee often brings lawsuits and thus must be avoided where possible.
No wonder employers are sometimes hesitant to hire. Policymakers need to avoid
making them more hesitant.

But that doesn’t mean that no employer will find hiring ex-offenders an
attractive option. Jobs vary immensely. Some provide the employee with very little
opportunity for wrongdoing; others can be made that way by adding a little extra
supervision. Individuals with criminal records vary immensely too. There are some
whose integrity is not open to serious doubt; there are others who will likely do well
when working with colleagues who are aware of their weaknesses and sensitive to
the need to avoid creating problems. A modest tax credit can be a useful tool to
persuade an employer who is considering hiring an ex-offender but has not yet
taken the plunge. In the long run, if administered properly, this program can reduce
crime and save the taxpayer money.

The Work Opportunity Tax Credit Program allows the employers who are in
the best position to offer employment to ex-offenders (or to a particular ex-offender)
to self-select. Some employers may find that they are in a good position to hire a
large number of ex-offenders; others may prefer to hire none. The latter group won’t
have to worry about their ability to prove to the satisfaction of any government
bureaucrat that they had good reason for their decision; instead, they simply won’t
be able to enjoy the tax credit that employers who make the opposite decision will
enjoy. The important thing is that the decision will be made by individuals who are
intimately familiar with the actual job and job applicant at issue and have an
incentive to make the right decision instead of by far away bureaucrats and judges,
who have no such familiarity with the situation.

The most recent iteration of the EEOC’s policy on criminal convictions—

issued in 2012—has none of the virtues of the Work Opportunity Tax Credit

Convicted of Embezzlement Charged with Embezzling to Pay Restitution, KSL.COM (Nov. 5, 2018),
embezzling-to-pay-restitution; Courtney Astolfi, Cleveland McDonald’s Manager Accused of Shooting
at Customer in Drive-Thru Line, CLEVELAND.COM (Feb. 23, 2018),
[]; Justin P. Hicks, New Embezzlement Charge Comes 5 Months After
Woman Released from Prison, MLIVE.COM (Grand Rapids)(Jan. 15, 2018),
[]; Richard Webner, Accountant Tied to Embezzlement from Centro Has
History of Bank Fraud, Bankruptcy, SAN ANTONIO EXPRESS-NEWS (Dec.18, 2017),
Centro-12434856.php [; Nick Leonard, Casper Woman Gets 3-5 Years for Embezzling $53,000 From
12-24 Club, K2RADIO.COM (June 1, 2017),
embezzling-52000-from-12-24-club/ []. All of these cases note the
employee’s previous convictions.  


Electronic copy available at:

Program. Not only does it make a race issue out of what should not be a race issue,
it ham-fistedly discourages all employers from even checking into the criminal
backgrounds of its job applicants. If they do check, it discourages them from acting
on the information that they obtain.

Part of the 2012 Guidance purports to be simply an ordinary application of

Griggs. Because acting on criminal background checks is thought to have a
disparate impact on African-American and Hispanic employees, employers are
instructed that they must be able to demonstrate a “business necessity” for doing so.

The 2012 Guidance was not the first time disparate impact analysis had been
applied to criminal records. Not long after Griggs, in Green v. Missouri Pacific
Railroad,240 the U.S. Court of Appeals for the Eighth Circuit attempted to apply
disparate impact’s logic to Missouri Pacific’s hiring policies.241 It took the position
endorsed by the EEOC that Griggs requires an employer who refuses to hire a job
applicant on account of his criminal record to demonstrate business necessity for its
action. Neither conscious nor unconscious intent to discriminate on the basis of
race was a necessary element of the cause of action.242

240 523 F.2d 1290 (8th Cir. 1975) (“Green I”).

241 Gregory v. Litton Sys., Inc., 316 F. Supp. 401 (C.D. Cal. 1970), modified on other grounds and
aff’d 472 F.2d 631 (9th Cir. 1972), is also worth noting. In it, the court held that an employer could
not, consistently with Title VII, decline to hire job applicants with 14 arrests for offenses other than
traffic violations but no convictions. The court stated:

There is no evidence to support a claim that persons who have suffered no criminal
convictions but have been arrested on a number of occasions can be expected, when
employed, to perform less efficiently or less honestly than other employees. In fact,
the evidence in the case was overwhelmingly to the contrary. Thus, information
concerning a prospective employee’s record of arrests without convictions, is
irrelevant to his suitability or qualification for employment. In recognition of this
irrelevance, the County of Los Angeles, a large-scale employer, has ceased to ask for
arrest information in application for employment.

Id. at 402-03. The court did not specify the evidence that it was relying on. Whatever it was, it was
unlikely to apply to an employee with fourteen arrests. I am very doubtful that Judge Hill would
have hired a law clerk, a housekeeper, a baby sitter or any other job applicant with a record of
fourteen arrests himself without overwhelming evidence that the particular job applicant’s character
had changed radically.

242  In
Green I, the Missouri Pacific Railroad had a policy of not hiring job applicants with criminal
records. As a result, 5.35% of African-American and 2.23% of white job applicants were rejected. To
explain its policy, Missouri Pacific advanced the following concerns (in the court’s words): “1) fear of
cargo theft, 2) handling company funds, 3) bonding qualifications, 4) possible impeachment of an
employee as witness, 5) possible liability for hiring persons with known violent tendencies, 6)
employment disruption caused by recidivism, and 7) alleged lack of moral character of persons with
convictions.” Green I, 523 F.2d at 1298.


Electronic copy available at:

Despite its name, the 2012 Guidance provides little guidance to employers on
what constitutes “business necessity,” apart from making it clear that the need to
hire only employees who were likely to be law abiding was not a clear case of
business necessity. Conscientious employers are at sea. Even employers who
decline to hire job applicants with felony convictions because they are required to do
so by state law are warned in the 2012 Guidance that they may be opening
themselves up to a lawsuit.243

Note that if the fact that state law commands an employer to act in a
This was not good enough. The Green I panel chided Missouri Pacific for failing to “validat[e] its
policy with respect to conviction records.” Id. at 1298. The EEOC had argued as amicus curiae and
the court apparently agreed that the law requires an employer to prove with the precision of a social
scientist that the job qualifications it uses produce better employees.

Green I went on to state that “blacks who have been summarily denied employment by MoPac on the
basis of conviction records have been discriminated against on the basis of race in violation of Title
VII and that the district court should enjoin MoPac's practice of using convictions as an absolute bar
to employment.” Id. at 1298-99.

On remand, the trial court issued an order enjoining Missouri Pacific from disqualifying job
applicants on account of any past criminal conviction. But it specifically allowed Missouri Pacific to
consider an applicants' prior criminal record as a factor in making individual hiring decisions so long
as defendant takes into account (1) the nature and gravity of the offense or offenses, (2) the time that
has passed since the conviction and/or completion of sentence, and (3) the nature of the job for which
the applicant has applied. These three items have come to be known as the “Green factors.”

Neither side of the litigation was happy with this result. But it was the plaintiffs, not the defendant,
who were most dissatisfied and appealed. They requested that the court enjoin Missouri Pacific from
considering a job applicant’s criminal record at all unless and until it could produce a valid study
showing that ex-offenders do indeed perform poorly at Missouri Pacific as employees. A different
panel of the Eighth Circuit (with only Judge Heaney in common) held that the Green I panel’s
discussion of the need for a validated study was dictum and affirmed the trial court. Green v. Mo.
Pac. R.R., 549 F.2d 1158 (8th Cir. 1977) (“Green II”).

243 The proposed Certainty in Enforcement Act of 2015 (H.R. 548) would have overruled the EEOC
on this narrow point. A hearing was held on March 24, 2015, but ultimately no action was taken.
See Testimony of Gail Heriot Before the House Subcommittee on Workforce Protections
(Mar. 24, 2015),

Note that licensing requirements, too, may go further than the EEOC’s policy. Hence, under this
approach, if the EEOC disagrees with a state legislature about whether a plumbing license should be
a requirement to engage in plumbing for hire, the EEOC’s policy should trump the state legislature’s
judgment. The 2012 Guidance denies that its policies could conflict with licensing requirements, but
it is difficult to see how this view can be squared with its view that state laws on hiring ex-offenders
can be overridden by EEOC policy via the Constitution’s Supremacy Clause. U.S. CONST. art. vi, cl. 2.


Electronic copy available at:

particular way is not enough to establish “business necessity,” then it is doubtful
that an employer could ever know for sure that it can decline to hire an employee on
the basis of his criminal record or indeed on the basis of anything else that has a
disparate impact (as all things do). The 2012 Guidance appears to be designed for
this purpose—intimidating employers into hiring as many ex-offenders as possible.

Insofar as the EEOC’s pattern of enforcement provides insight into what the
EEOC would consider sufficient to prove business necessity, it shows that the
EEOC is playing hardball. It has targeted employers whose line of work is sensitive
enough that the need for relatively clean criminal records would have been viewed
by many as an obvious business necessity. That the EEOC would pursue employers
like G4S Secure Solutions USA—a provider of security personnel—is strong
evidence of the agency’s overreach on this issue.

Julie Payne, Senior Vice President and General Counsel of G4S Secure
Solutions USA, testified before the U.S. Commission on Civil Rights on December 7,

Our clients expect, and most require, that G4S supplies

security officers that have passed an extensive criminal background
check. I am here today to cast light on the EEOC’s targeting of
companies, including my own, over legitimate and necessary business

In April of 2010, David Coleman, an individual with two prior

convictions for theft, applied for a position with G4S in Pennsylvania.
G4S did not hire Mr. Coleman, as a result of his convictions. He filed
a charge of discrimination, claiming that our refusal to hire him was
based on his race and that use of criminal background checks
adversely impacts African Americans.

The EEOC Philadelphia Office expanded the Agency

investigation beyond the charging party, to include G4S applicants
and employees across the United States. The EEOC sent a series of
requests for information that were incredibly extensive and
burdensome and sought vast amount of information related to our
hiring policies.

They requested information about every employee and

applicant of G4S, its parent companies, its subsidiaries, its affiliates,
its successors, its predecessors, agents and assigns. It sought the
information for a period of time dating back to the date when G4S
first implemented its policy regarding criminal convictions to the


Electronic copy available at:


G4S … was founded in 1958 and has relied on criminal

background screening from the beginning. As a result of this inquiry,
we have hired multiple lawyers, statisticians, experts, to assist us
with complying with the onerous request. We have spent hundreds
of thousands of dollars, to an uncertain result.244

As Ms. Payne testified, security guards must be trustworthy. They

frequently have access to the most sensitive precincts within an employer’s place of
business. Often they work unsupervised and at hours when their activities go
unobserved by others. Yet this was not enough for the EEOC.

The EEOC’s pattern of enforcement thus adds to the likelihood that some
employers will err on the side of not doing criminal background checks even when
courts might have been considered them a “business necessity.” Some of them will
later regret that decision as employees with criminal records will likely disappoint
their employers at higher rates than those without such records.

Other employers will undertake the background checks, but will be careful—
sometimes against their better judgment—to hire at least some of those with
criminal convictions in order to demonstrate their compliance with the 2012
Guidance. Some of these employers will also come to regret their decisions (though
unlike the employers who opt not to investigate the criminal records of their job
applicants, these employers will know where they went wrong). Instead of helping
to place ex-offenders in right jobs, this approach will put ex-offenders in the wrong

Instead of trying to help employers understand what constitutes “business

necessity” within the meaning of Griggs, the 2012 Guidance purports to outline a
set of steps for employers may wish to follow in hiring employees in order to avoid
liability. The EEOC is not always clear about those steps, but the guidance appears
to anticipate that employers will wait until they have selected the applicant or
applicants they intend to hire and only then (if ever) proceed to check their criminal


added), A more extensive written version of Ms.
Payne’s testimony is at pp. 239-49. Id.

The fact that Title VII makes EEOC investigations and mediations confidential, 42 U.S.C. § 2000e-
8(e), adds to the degree to which EEOC policymaking has tended to escape both public scrutiny and
government oversight. The EEOC was not cooperative with the U.S. Commission on Civil Rights
during the preparation of the above-cited report.


Electronic copy available at:

backgrounds. At that point, while the 2012 Guidance insists that it “does not
necessarily require individualized assessment in all circumstances,” it states that
“the use of a screen that does not include individualized assessment is more likely
to violate Title VII ... [and] the use of individualized assessments can help
employers avoid Title VII liability....”

One can be confident that most employers will read this as requiring
individualized assessments at least for members of the groups for whose benefit the
policy said to be intended--African Americans and Hispanics. (A by-product of the
EEOC’s efforts to shoehorn the problems of ex-offenders into the race discrimination
paradigm is that its policy will likely only apply to members of racial minorities.)
But it is unclear what an individual assessment is. It does, however, appear to
require, at minimum, the application of the following factors: (1) the nature and
gravity of the offense or offense; (2) the time that has passed since the conviction
and/or completion of the sentence; and (3) the nature of the job for which the
applicant has applied.245 The 2012 Guidance then followed this with a further
“suggestion” that employers use a second, more subjective assessment. What is
clear is that, at minimum, the subjective assessment ordinarily must include “notice
that [the job applicant] has been screened out because of a criminal conviction, an
opportunity to ... demonstrate that [the employer’s policy] should not be applied due
to his particular circumstances; and ... whether the additional information ...
warrants an exception ... show[ing] that the policy as applied is not job related and
consistent with business necessity.”246

The 2012 Guidance thus requires that employers telegraph to job applicants
that they have been screened out on account of their criminal records. In doing so, it
exponentially increases the odds of a lawsuit. The vague discussion of business
necessity in the guidance and the EEOC’s enforcement history mean that no
decision to reject a job applicant with a felony record can be regarded as completely
safe. Reasonable minds will frequently disagree as to whether the employer has
sufficient reason to reject a job applicant. The fact that ex-offenders are frequently
unreasonable should not be lost sight of.

Once an ex-offender brings a case to the EEOC’s attention (or the employer’s
exercise of discretion is brought to the EEOC’s attention by other means), it will be
the EEOC’s judgment concerning what constitutes “business necessity,” rather than
the employer’s, that counts. The only way to avoid this is for the employer to bend
over backwards to hire ex-offenders and to sometimes have to pay the price for
having done so.

Is the 2012 Guidance in compliance with the Administrative Procedure Act

245 See supra note 241.

246 2012 Guidance at 15.


Electronic copy available at:

(the “APA”)?247 That is less than clear, though, alas, any analysis of a particular
guidance’s status under the APA is beyond the scope of this article.248

247 5 U.S.C. § 500 et seq.

248 A few thoughts may be useful: The guidance certainly appears to be a “rule” within the meaning
of that Act. See 5 U.S.C. § 551(4) (defining “rule” as “an agency statement of general or particular
applicability and future effect designed to implement, interpret, or prescribe law or policy”). That
means it requires that full notice-and-comment procedures be observed unless the guidance falls
within one of the exceptions listed in the statute. Since the EEOC is not authorized to promulgate
substantive notice-and-comment rules under Title VII, if the 2012 Guidance doesn’t fall into one of
the exceptions, the problem cannot be remedied by subjecting it to notice-and-comment procedures in
the future.

The only two exceptions that seem possibly applicable are those for “interpretive rules” and “general
statements of policy.” See 5 U.S.C. § 553(b)(A) (specifying that notice-and-comment requirements do
not apply “to interpretive rules, general statements of policy, or rules of agency organization,
procedure, or practice”). According to Attorney General’s Manual on the Administrative Procedure
Act, interpretive rules are “rules or statements issued by an agency to advise the public of the
agency’s construction of the statutes and rules which it administers.” To be an interpretive rule, a
rule must indeed be an interpretation and not an extension of some statute, rule, or case. It cannot
add duties that not derived in good faith from the statute, rule or court opinion using the ordinary
methods employed by the courts and the legal profession. On the other hand, an agency’s good faith
interpretation of an ambiguous statute or rule is not outside the definition of an “interpretive rule”
simply because it happens to be more expansive than the regulated parities would like. Drawing the
line can be difficult. But it is no easy to see how the set of procedures envisioned by the 2012
Guidance are interpretations of Title VII or of Griggs. Instead they appear to be prophylactic in
nature. For example, by setting up a “subjective individualized assessment” procedure under which
employers tell job applicants that they would have been hired but for their prior felony conviction the
EEOC has provided employers with a huge incentive to construe the business necessity defense
narrowly. That is not the same thing as providing insight into the proper interpretation the
business necessity defense.

The other possible exception is a “statement of general policy.” But here, too, there are difficulties.
According to Attorney General’s Manual on the Administrative Procedure Act, “general statements
of policy” are “statements issued by an agency to advise the public prospectively of the manner in
which the agency proposes to exercise a discretionary power.” Most typically, a statement of general
policy will announce which kinds of violations that it intends to apply its resources to. But the
EEOC has no discretion to impose duties on regulated parties not found in the underlying statute,
rule or court opinion. Again, reasonable minds may differ in good faith over how expansively that
statute, rule or court decision should be read, so line drawing may be difficult. See Pac. Gas & Elec.
v. Fed. Power Comm’n, 506 F.2d 33, 37 (D.C. Cir. 1974) (quoting Professor Kenneth Culp Davis)
(calling the distinction between substantive rules and general statements of policy “fuzzy”). But even
with that cautionary note, the 2012 Guidance seems to extend the law by inventing a prophylactic
procedure designed to put employers in a position in which they will be extremely hesitant to use the
business necessity defense. That isn’t a forewarning of how it intends to exercise its discretion, since
it never had the discretion to impose such a procedure. Whatever “business necessity” might mean,
it does not mean “(a) decide which employee you wish to hire; (b) only then should you conduct a
criminal background test; (c) if you find a violation, be sure to consider it in light of (i) the nature and
gravity of the offense or offenses, (ii) the time that has passed since the conviction and/or completion
of sentence, and (iii) the nature of the job for which the applicant has applied; and (c) allow the
applicant a chance to explain why he should nevertheless by hired. Rather than a statement about


Electronic copy available at:

IX. Perverse Effects: Disparate Impact Liability Can Harm its Intended
Beneficiaries (and in the Case of Criminal Background Checks Probably

What happens when Title VII is interpreted to prevent employers from hiring
on the basis on job qualifications that the employer regards as significant? It’s
likely that many employers will rely on proxies for that qualification. Perversely,
sometimes those proxies will be racial or sexual stereotypes. For example, if an
employer wants to hire someone with better than average math skills and she is not
permitted to administer a math test, she may, consciously or unconsciously, choose
the Chinese American over the Irish American. If she wants to hire proofreaders
who excel at English grammar and punctuation and cannot test for that talent
without risking a lawsuit, she may quietly go for the Irish American over the
how the EEOC intends to exercise its discretionary power, this appears to be an effort to corral
employers into procedures that the EEOC believes will result in more hiring of ex-offenders.

To be sure, some cases have held that “[t]he critical distinction between a substantive rule and a
general statement of policy is the different practical effect that these two types of pronouncements
have in subsequent administrative proceedings . . . . A properly adopted substantive rule establishes
a standard of conduct which has the force of law . . . . The underlying policy embodied in the rule is
not generally subject to challenge before the agency. A general statement of policy, on the other
hand, does not establish a ‘binding norm.’” E.g., Pac. Gas & Elec., 506 F.2d at 38 (citations omitted)..
The Court’s focus on the practical effect in subsequent proceedings rather than the practical effect on
regulated parties may be explainable in part by the fact that plaintiff in the case was going to have
an opportunity to make its case before the Commission shortly. Its effort to bring the issue before a
court was thus decidedly premature. In cases in which such an opportunity is not available, courts
have been more willing to consider the practical effect on private parties. See Appalachian Power
Co. v. EPA, 208 F.3d 1015, 1021 (D.C. Cir. 2000)(“if [an agency’s document] leads private parties . . .
to believe that [the agency] will declare permits invalid unless they comply with the terms of the
document, then the agency’s document is for all practical purposes ‘binding’”). See also Columbia
Broad. Sys., Inc. v. U.S., 316 U.S. 407 (1942) (pre-APA case); Robert A. Anthony, Interpretive Rules,
Policy Statements, Guidances, Manuals, and the Like—Should Federal Agencies Use Them to Bind
the Public?, 41 D UKE L. J. 1311, 1360-61 (1992).

For information on the State of Texas’s challenge to the 2012 Guidance, Texas v. EEOC, see supra
note 103.


Electronic copy available at:

Chinese American on the assumption that an Irish American comes from a family
that has been speaking English for generations while the Chinese American does
Of course, prohibiting real discrimination is exactly what Title VII was
supposed to be about. Congress was well aware that some discrimination—call it
“statistical discrimination”—is rooted in stereotypes than may (or may not) have
some basis in fact. For example, on average women really are less physically able to
lift heavy weights than men. That is not a myth. But if an employer wants an
employee who is strong, it should look for evidence of that characteristic and not
depend on sexual stereotypes, even accurate stereotypes. Similarly, if an employer
wants an employee who is well-educated, polite, assertive, mathematically inclined,
gregarious, or trustworthy, it should look for evidence of those characteristics and
not depend on racial, color, religious, sexual, or national origin stereotypes.

But the success of that approach depends upon the ability of employers to
seek evidence of the actual desired traits. If the employer is looking for employees
who can lift 50 pounds with ease, it needs to be able to test for that. If the employer
is looking for employees who are good at math or at English grammar, it needs
some way to gauge that.

The requirement that employers not reject job applicants who have felony
records is an especially interesting example. If the employer is looking for
trustworthy employees who will not commit crimes, it needs some source of
information. The applicant’s criminal record (or lack of a criminal record) is often
the best available method for separating the cases that are most likely to be a
problem from those that are not. If employers are prohibited from using that
method, they may consciously or unconsciously, be tempted to use race, color,
religion, sex, or national origin as a proxy for criminal record.

This is an area where we have some evidence, albeit sketchy. In Perceived

Criminality, Criminal Background Checks and the Racial Hiring Practices of
Employers, three economists discussed the double effect of heavily regulating
criminal background checks. As they explain it, it must be kept in mind that
African-American and Hispanic men are not simply more likely to have a criminal
conviction on their record, they are more likely to be perceived as more likely to
have one. Consequently, if the EEOC discourages employers from checking the
criminal background of job applicants out of fear of liability, some will likely shy
away from hiring young African-American or Hispanic males in the (not necessarily
unfounded) belief that members of these groups are somewhat more likely to have
criminal records than, for example, elderly Asian-American females. The EEOC’s
attempt to prevent the “disparate impact effect” creates an incentive for a “real


Electronic copy available at:

discrimination effect.”249

Note also that some employers may make adjustments to their hiring policies
that are not motivated by race and hence do not violate Title VII’s prohibition on
intentional discrimination, but which nevertheless further disadvantage job
applicants from groups that are traditionally viewed as disadvantaged. Consider
the following hypothetical: An employer regularly hires young, unskilled, high
school dropouts as full-time packers for his moving van business. The company’s
business location is in a part of town that yields a labor pool that is
disproportionately, but not overwhelmingly, African American and Hispanic. Until
his lawyer instructed the owner that the requirement of “subjective individualized
assessments” made excluding applicants with criminal records too risky, he had
been doing background checks and declining to hire most of those with a criminal
record. But after he stopped conducting those checks, he hired a young, white 19-
year-old who ended up stealing from one of the employer’s customers. Another
recent hire turned out to have a serious drug problem. The employer does not know
it, but criminal background checks would have identified those employees as risky.
All the employer knows is that he is dissatisfied with the employees he has been
getting lately. He therefore decides to convert the full-time jobs that come open to
part-time jobs and to advertise in the campus newspaper at a nearby highly
competitive liberal arts college. He figures (rightly or wrongly) that the students
there will likely be more trustworthy than the pool he had been hiring from. Given
the demographics of the school’s student body, this yields an overall labor pool that
has proportionately fewer minorities than before. Under such a scenario, the
EEOC’s policy would have accomplished precisely the opposite of its intentions.

Or consider another scenario: The business is a local carwash company

operating within metropolitan Los Angeles. It employed 50 unskilled, young adults,
and compared to its competitors, it used low-tech, labor-intensive methods and paid
low wages. Prior the 2012 Guidance, its CEO used to screen all job applicants for
criminal convictions. He stopped doing it when his brother-in-law, a personal injury
lawyer, told him this could get him in trouble. Alas, one of his employees,
unbeknownst to him an ex-offender, turned out to be a troublemaker. He attacked
and killed one of his co-workers and seriously injured two others. After the police
investigation, the worker’s compensation claims, and the loss of customers, the CEO
decided to close the location and establish an automated car wash that employed
only 5 trusted employees. The rest of the employees were laid off.

From a policy standpoint, the most obvious question is which effect

dominates—the effect of employing more African Americans and Hispanics (because
fewer ex-offenders are eliminated from the pool) or the effect of employing fewer
(because the employer avoids hiring minorities). The evidence adduced by the

Harry J. Holzer, Steven Raphael & Michael A. Stoll, Perceived Criminality, Criminal Background

Checks, and the Racial Hiring Practices of Employers, 49 J.L. & E CON . 451 (2006).  


Electronic copy available at:

authors of Perceived Criminality, Criminal Background Checks and the Racial
Hiring Practices of Employers indicates that it may be the latter. That article
examined the answers to interview questions provided by slightly over 3000
employers that hired workers without college degrees from Atlanta, Boston, Detroit,
and Los Angeles in the early 1990s. Approximately half those employers either
always or sometimes conduct criminal background checks on job applicants.
Further data collected in 2001 in Los Angeles showed this number had climbed
from 48.2% to 63.2% for that city specifically.250
The article found that employers who conduct background checks were
more likely to have recently hired an African-American applicant than employers
who do not. Among those employers who were unwilling to hire ex-offenders, the
employers who checked were 10.7% more likely to have recently hired an African
American. This finding was highly significant.
It is often difficult to distinguish cause from effect. In conducting studies of
this kind, one could argue that the reason that employers who undertake
background checks are more likely to hire African Americans is that they face labor
pools that are heavily African American and are biased against African Americans.
But the authors used statistical methods to account for this possibility as best they
could and still found the evidence they found.251
More recent scholarship tends to confirm that laws making it difficult for
employers to find out whether applicants have a criminal record hurt rather than
help African American and Hispanic men. A fascinating example is an experiment
conducted by Amanda Agan and Sonja Starr. They sent approximately 15,000
fictitious online job applications to employers in New Jersey and New York City
before and after the adoption by those two jurisdictions of “Ban the Box” legislation.
Prior to the passage of that legislation, some of the employers routinely asked about
criminal records and some of them not.252

250  Id. at 457 (fig. 4). See also Michael A. Stoll, Ex-Offenders, Criminal Background Checks, and

Racial Consequences in the Labor Market, 2009 C HICAGO L EGAL F ORUM 381, 400-406.
251  Similarly,
research has been undertaken attempting to confirm or refute the hypothesis that easy
availability of criminal background information benefits black males as a group overall by comparing
the black-to-white wage ratio in states that make criminal records broadly available to that in states
that do not. Shawn D. Bushway, Labor Market Effects of Permitting Access to Criminal History
Records, 20 J. C ONTEMP . C RIM . J USTICE 276 (2004). Bushway’s data did indeed show that states
that make criminal records broadly available have higher black-to-white wage ratios, but those data
were too skimpy for this difference to be statistically significant. Bushway has called for more
research. Id. at 288-89.
252  Amanda
Agan & Sonja Starr, Ban the Box, Criminal Records, and Racial Discrimination: A
Field Experiment, 133 Q. J. E CON . 191 (2018).


Electronic copy available at:

Some of their findings were hardly surprising. They found that employers that
ask about criminal records are 62% more likely to call back an applicant with no
record. Interestingly, New York City employers were especially sensitive to
criminal records; they were 78% more likely to call back an applicant with a clean
record as opposed to only 45% more likely for New Jersey employers.
But here’s the rub: Prior to “Ban the Box” legislation, white applicants
received 7% more call backs from employers who had asked about criminal records
than otherwise similar black applicants. After “Ban the Box” legislation went into
effect, the gap increased to 45%. Rather than helping black applicants, “Ban the
Box” legislation was hurting them.
Jennifer L. Doleac and Benjamin Hansen took a different approach. They
examined individual-level data from the 2004-2014 Current Population Survey,
taking advantage of the adoption and timing of various “Ban the Box” laws. They
focused on relatively young (ages 25-34), low-skilled (i.e. no college degree), black
and Hispanic men, seeking to find whether “Ban the Box” laws help or hurt. Their
findings were troubling for anyone who had hoped such laws would be beneficial to
such men:
We find net negative effects on employment for these groups: Young,
low-skilled black men are 3.4 percentage points (5.1%) less likely to be
employed after [Ban the Box] than before. This effect is statistically
significant (p < 0.05) and robust to a variety of alternative
specifications and sample definitions. We also find the [Ban the Box]
reduces employment by 2.3 percentage points (2.9%) for young, low-
skilled Hispanic men. This effect is only marginally significant (p <
0.10) but also fairly robust. Both effects are unexplained by pre-
existing trends in employment, and – for black men – persist long after
the policy change. The effects are larger for the least skilled in this
group (those with no high school diploma or GED), for whom a recent
incarceration is more likely.253

If these are the effects of “Ban the Box” legislation, it is likely that the 2012
Guidance also has counterproductive effects. Moreover, there is specific evidence
that other bans on employment practices with disparate impact have been counter-
productive too. Credit reports are a good example. Over the course of the last few
decades, some employers found that consulting the credit reports of job applicants
was a good way to ensure that an employee was reliable. An individual who
conscientiously paid her bills on time was thought to be more likely to perform the
tasks assigned to her conscientiously. And there is some evidence that, on the
253  Jennifer L. Doleac & Benjamin Hansen, The Unintended Consequences of “Ban the Box”:
Statistical Discrimination and Employment Outcomes When Criminal Histories are Hidden (August


Electronic copy available at:

average, this is true.254 Nevertheless, because this practice is thought to have a
disparate impact on some racial and national origin groups, employers can be
subject to disparate impact liability for using them. In Deleting a Signal: Evidence
from Pre-Employment Credit Checks, Alexander W. Bartik and Scott T. Nelson
found evidence that bans on credit checks decrease the chances that African
Americans will find a job and increases the chance that African Americans will be
terminated if they do find a job.255
Similarly, whether or not requirements that a job applicant pass a drug test
have a disparate impact on African American and/or Hispanic applicants, there is
evidence that prohibiting employers from drug testing their job applicants has a
negative effect on the employment of African Americans.256

The counterproductive effects of disparate impact law are not limited to such
direct backfires. One of the most important areas in which disparate impact
liability has left its mark is in drastically reducing the use of written tests.257 These
tests range from the standardized aptitude tests like those involved in Myart,
Griggs and Albemarle Paper to more specific tests of knowledge thought to be
necessary or useful for the job by the employer, such as that used in Ricci v.
DeStefano (2009). Tests of these sorts used to be common. These days they are
relatively uncommon. An employer must go to heroic lengths to justify their use.

The irony is that in an earlier era, written tests weren’t just permissible, they
were state-of-the-art in fair play. Their purpose was to help uncover talent in
unexpected places and to prevent the well-connected from walking away with all
the prizes.

Two examples stand out. More than a century ago, the Pendleton Civil
Service Reform Act of 1883 established competitive civil service exams as the
preferred method of choosing federal employees, replacing the patronage system,
254J.B. Bernerth, S. G. Taylor, H. J. Walker, and D. S. Whitman, An Empirical Investigation of
Dispositional Antecedents and Performance-Related Outcomes of Credit Scores, 97 J. A PP . P SYCH .
469–48 (2012).
255Alexander W. Bartik & Scott T. Nelson, Deleting a Signal: Evidence from Pre-Employment Credit
Checks (MIT Working Paper 16-01) (2019) (previously titled Credit Reports as Resumes: The
Incidence of Pre-Employment Credit Screening),
256  Abigail
K. Wozniak, Discrimination and the Effects of Drug Testing on Black Employment, 97
R EV . E CON . & S TATS . 548 (2015).
257See, e.g., Karen Matthews, Test Meant to Screen Teachers Instead Weeded Out Minorities,
Associated Press (Mar. 11, 2017) (New York poised to get rid of a teacher certification examination
because “[j]ust 46 percent of Hispanic and 41 percent of black test takers passed it on the first try,
compared with 64 percent of white candidates”),


Electronic copy available at:

which was viewed then as the height of corruption. The concept of merit selection
was synonymous with the competitive exam. After the passage of the Pendleton Act,
having “connections” would no longer be the ticket to a government job. Today only
a small proportion of federal jobs are decided on the basis of an exam.

Similarly, the SAT was designed to replace a system that gave graduates of
elite prep schools an advantage. Prior to the SAT, Ivy League universities required
knowledge of Latin and Ancient Greek, both of which were always taught at prep
schools, but which were less available at public schools, particularly those in rural
areas. The whole purpose of the SAT was to measure “aptitude for learning” rather
than the mastery of some specific area of knowledge. The point was to put students
of different backgrounds on a more level playing field.

Some might argue that decreased reliance on tests of this sort is a good thing.
Tests that haven’t been validated result in error. (Indeed, even tests that have been
validated result in error.) People who might well have been good at a job get
screened out; people who will be ultimately unsuccessful in the job fail to get
screened out.

But it is important to keep in mind that the substitutes for these sorts of
tests—“holistic evaluations”—have serious problems too. Holistic evaluations based
on a job applicant’s interview and his resume also exclude some applicants who
would have performed just fine and they fail to exclude some applicants who
perform poorly.

Indeed, there is plenty of evidence that, despite all the obvious shortcomings
of mechanical assessments, they are generally more accurate than more holistic
approaches.258 For example, in 1971, University of Oregon psychology professor

258 William M. Grove, David H. Zald, Boyd S. Lebow, Beth E. Snitz & Chad Nelson, Clinical Versus
Mechanical Prediction: A Meta-Analysis, 12 P SYCH . A SSESS . 19 (2000) (in a meta-analysis of over
130 empirical studies, “[o]n average mechanical prediction techniques were about 10% more accurate
than clinical predictions”). The study found that mechanical methods out-performed clinical
predictions in 33% to 47% of the studies examined. While the clinical predictions were often as good
of the mechanical predictions, they were only substantially better in 6% to 16% of those studies.
This does not mean that poorly-conceived tests will always yield better results than a more holistic
evaluation by a discerning hiring supervisor. The mechanical tests employed examined in this meta-
analysis had sometimes been studied well at an earlier point in time and sometimes had not been.
Some were diagnostic tests, some tests were for academic performance, and some were related to job
performance. But the overall pattern was that “mechanical predictions of human behavior are equal
to or superior to clinical prediction methods for a wide range of circumstances.” Id. at 19.

The Supreme Court’s has declined to allow race-preferential admissions policies that award a certain
number points to minority students. See Gratz v. Bollinger, 539 U.S. 244 (2003). But it has
encouraged holistic approaches. Grutter v. Bollinger, 539 U.S. 306 (2003). This has led some
commentators to point out the plentiful data suggesting problems with holistic methods. See
Margaret E. Brooks et al., Distinction Bias in Applicant Reactions to Using Diversity Information in


Electronic copy available at:

Robyn Dawes famously conducted one of the first experiments comparing the
results of the holistic approach with stricter measures of student academic
potential. Dawes took the faculty “holistic ratings” of applicants for slots in the
psychology graduate program and compared them to the actual performance of the
admitted students in their first year. He found the correlation was only .19, thus
explaining only 3.61% of the variance. By contrast, taking each student’s Graduate
Record Examination scores and combining them with that student’s undergraduate
GPA and a measure of the quality of that undergraduate institution yielded a
correlation of .38 with that student’s first-year graduate performance. It thus
explained 14.44% of the variance (i.e. quadrupling the amount of variance
accounted for).259

What else has replaced tests? For good or ill, in the last several decades,
employers have been much more likely to require or quietly give very heavy weight
to a college degree. Of course, requiring a college degree has a disparate impact on
African American and Hispanics too. But, in the broad exercise of its discretion, the
EEOC has taken less interest in employers that require job applicants to have a
college degree than it has in employers who administer written tests, like those in
Griggs, Albemarle Paper and Ricci. It’s not clear why (though some have suggested
that it is because the higher education industry is usually viewed as a left-of-center

It is not necessarily a good thing that employers must rely on college degrees
rather than on simple, easily-administered written tests. Young people make huge
investments in time and money in order to acquire a college degree. Many of them
ultimately fail in their efforts, but are left with heavy debt. Of those who succeed,
some would have done well even if they had not attended college, but had gone
straight into the workforce, demonstrating their skills through a written test.

Of course, a liberal arts education is a good thing in and of itself. I have a

deep regard for it when it is done right. But it isn’t everyone’s cup of tea. Some
(maybe many more than a few) would have preferred to enter the workforce sooner

Selection, 17 INT’L J. SELECT. & ASSESS. 377, 378 (2009): “In the most Supreme Court case dealing
directly with this issue, the Court mandated a more holistic approach to affirmative action—despite
evidence that mechanical methods of combining data consistently outperform holistic approaches.”
See also Scott Highhouse & John A. Kostech, Holistic Assessment for Selection and Placement, in K.

Robyn M. Dawes, A Case Study of Graduate Admissions: Application of Three Principles of


Human Decision Making, 26 AM. PSYCH. 180 (1971).

260Of course, if the employer only gives heavy weight to a college degree (or conceals the fact it is
requiring one), it is unlikely to come to the EEOC’s attention.


Electronic copy available at:

if that option had been available to them. And it is likely that some employers that
currently look for college degrees would have preferred, if given the option, to select
employees by testing for specific knowledge.261

Americans ages 25-64 are somewhat more likely to hold a college degree than
individuals in the same age group from the United Kingdom, Norway, Australia,
Sweden or Denmark. They are much more likely to hold a college degree than their
counterparts in France, Germany or Italy.262 Is it possible that Americans over-
invest in college? And is it possible that Griggs-based legal problems with written
tests are a significant part of the reason?

Note that colleges and universities are not governed by Title VII in choosing
their students. Education is not employment. Colleges and universities have
largely stuck to the SAT, the LSAT, the MCAT, the GMAT and the GRE, because
they know that for all their shortcomings, these tests are useful. Indeed, colleges
and universities administer their own tests, written by individual faculty members,
which are rarely validated in any way, in nearly all their classes. These tests
matter to a student’s GPA, which in turn matter in terms of whether they will
graduate, the professional and graduate schools programs to which they will be
admitted and the jobs they will be offered. In some respects, employers who insist
on or weigh heavily a college diploma (or especially those that insist on a college
diploma from an elite or selective institution) are using that as a proxy for
successful performance on written tests.

The irony is that disparate impact liability has ended up giving those lucky
enough to have had the opportunity to go to college an advantage over those who,
261Another effect of the legal risks involved in employer-administered paper-and-pencil tests is that
those risks are probably adding to the pressure for occupational licensing. There is now bipartisan
support for the view that occupational licensing has gotten out of hand. The number of jobs for
which there is some sort of licensing requirement has mushroomed in the last few decades. Very
often the licensing requirement involves a paper-and-pencil test of exactly the sort the EEOC has
been making war on for the last half century. But because the state that employs these tests is not
acting as an employer, it is not subject to disparate impact liability under Title VII. A paper-and-
pencil test associated with an occupational license is likely worse for those groups that are
disadvantaged by paper-and-pencil tests more generally. An occupational licensing test will be the
same for everyone who seeks to work in that occupation. On the other hand, if there were no license
required, but employers were permitted to test of knowledge they thought necessary or advisable for
the job without fear of being sued, employers would likely differ from one another in how they tested.
For one thing, they might differ in exactly what knowledge they thought they should test for. They
might also differ in how they tested for it. Even on a standardized test, employers might differ on
how high a score to demand—whereas there is a uniform pass-threshold on a licensing exam.


degrees-or-higher-age. For individuals age 25-34, the U.K., Norway, and Australia surpass the rates
for the United States.


Electronic copy available at:

for whatever reason, enter the workforce earlier. Not everyone is in a position to
undertake massive debt and spend four years of his or her life earning a college

X. Disparate Impact Liability Is Itself a Form of Discrimination that Must

Be Subjected to Strict Scrutiny.

Ricci v. DeStefano (2009),263 is an unusual illustration of how fear of liability

can lead to discrimination. In it, the defendant City of New Haven wanted to
promote the best qualified firefighters to higher ranking positions. It went to
enormous trouble and expense to produce a written examination that would fairly
measure applicants’ knowledge of fire science, hoping that a court would agree that
it was “job-related” and hence that it was acting from “business necessity.” It took
lots of time and input from many experts. In turn, the firefighters themselves went
to enormous trouble to prepare for the exam.

But local activists didn’t like the results, since no African American
candidates scored high enough for a promotion.264 They took to staging loud and
insistent protests. City officials therefore cancelled the results, arguing that they
were compelled to do so, because they could otherwise be held liable under a
disparate impact theory. The white and Hispanic candidates who had passed the
test sued.

Plaintiffs argued persuasively that cancelling the results of the examination

was a stark example of actual discrimination. If the racial identities of the winners
had been different, they’d have been promoted. To put it more broadly, traditional
liability for discrimination (which focuses on equal treatment) and disparate impact
liability (which focuses on equal results) were at loggerheads in that case. Indeed,
263 557 U.S. 557 (2009).

264 According to the Court:

Seventy-seven candidates completed the lieutenant examination—43 whites, 19

blacks, and 15 Hispanics. Of those, 34 candidates passed—25 whites, 6 blacks, and 3
Hispanics. Eight lieutenant positions were vacant at the time of the examination.
As the rule of three operated, this meant that the top 10 candidates were eligible for
an immediate promotion to lieutenant. All 10 were white. Subsequent vacancies
would have allowed at least 3 black candidates to be considered for promotion to

Forty-one candidates completed the captain examination—25 whites, 8 blacks, and 8

Hispanics. Of those, 22 candidates passed—16 whites, 3 blacks, and 3 Hispanics.
Seven captain positions were vacant at the time of the examination. Under the rule
of three, 9 candidates were eligible for an immediate promotion to captain—7 whites
and 2 Hispanics.

Ricci, 557 U.S. at 566 (citations omitted).


Electronic copy available at:

they are always at loggerheads. The Ricci case brought this point into sharp relief,
because New Haven retroactively set its promotion criteria to favor African
Americans. At that point, the victims of the action were obvious, identifiable, and
displeased enough to bring a lawsuit.

But even in more ordinary cases, disparate impact necessarily involves

discrimination. One possibility is that it pushes employers to quotas. Alternatively,
it has its intended effect: It pushes employers to choose job qualifications that they
otherwise would not, precisely because those qualifications were better for
particular groups based on their race, color, religion, sex, or national origin than the
qualifications the employers would have chosen.

The Ricci case gave the Court the opportunity to wade into the legal swamp
that it helped create with its decision in Griggs. But it only put its toe in. It held
that an employer may intentionally discriminate in the manner of the City of New
Haven only if there is “a strong basis in evidence” that it would otherwise be liable
for disparate impact. A merely colorable disparate impact claim is insufficient.

The Court went on to hold that New Haven did not have such a “strong basis
in evidence” and hence granted plaintiffs’ motion for summary judgment in its
favor. But the opinion is not especially clear as to how it drew that conclusion.
New Haven certainly had the “legitimate employment goal” as required by Wards
Cove. And it presented evidence from several expert witnesses that the knowledge
tested for was strongly job-related and consistent with business necessity. But it is
not clear that it had the kind of scientific evidence that would be necessary to pass
muster under Albemarle Paper or the Uniform Guidelines on Employee Selection
Procedures. Was that the standard? If not, what was?265

265 The Court explained itself this way: “The is no genuine dispute that the examinations were job-
related and consistent with business necessity.” On the other hand:

[R]espondents note some candidates’ complaints that certain examination questions

were contradictory or did not specifically apply to firefighting practices in New
Haven. But [the vice president of the company that designed the examinations for
New Haven] told the [Civil Service Board] that [the company] had addressed those
concerns—that it entertained “a handful” of challenges to the validity of particular
examination questions, that it “reviewed those challenges and provided feedback [to
the City] as to what we thought the best course of action was,” and that he could
remember at least one question [the company] had thrown out (“offer[ing] credit to
everybody for that particular question”). For his part, [a competitor of the company
that designed the examinations] said he “suspect[ed] that some of the criticisms ….
[leveled] by candidates” were not valid.

Ricci at 588 (citations omitted).

But the burden of proof was on the defendant. Could a trier of fact had found that the vice president of a
company that produces examinations of this kind is not to be trusted? What if they also had discounted


Electronic copy available at:

The most interesting aspect of the case, however, came in Justice Scalia’s
concurrence. He began with these words:

I join in the Court’s opinion in full, but write separately to observe that its
resolution of this dispute merely postpones the evil day on which the Court
will have to confront the question: Whether, or to what extent, are the
disparate-impact provisions of the Civil Rights Act of 1964 consistent
with the Constitution’s guarantee of equal protection? The question is
not an easy one.266

Scalia was likely thinking of law review articles that were then fairly recent that
reluctantly questioned the constitutionality of disparate impact liability.267

The triggering event for that discussion was the Court’s decision in Grutter v.
Bollinger (2003)—one of two cases against the University of Michigan for its race-
preferential admissions policies. Prior to Grutter, Court decisions had often signaled
what many considered obvious: The Equal Protection Clause really is an equal protection
clause in that it protects whites against discrimination to the same degree it protects
African Americans; it protects men to the same degree it protects women and so on.268
But hope sprang eternal among supporters of race-preferential admissions policies that
the Court would nonetheless hold that strict scrutiny—the judicial test that applies to
statutes and state policies that discriminate on the basis of race—would only apply to
racial minorities and not to whites.

It was a vain hope. In Grutter and its twin, Gratz v. Bollinger,269 the Court applied
strict scrutiny to all laws and state policies that discriminate on the basis of race,
regardless of which race is getting the short end of the stick. A contrary ruling would
have been incoherent. If strict scrutiny can only benefit minority races, that would
superficially suggest that Asian Americans could benefit from it, but not whites. And
American Indians could benefit, presumably including the (very large) group who are
(admittedly lukewarm) supportive statements from his competitor that “some” of the criticisms of the
examinations were not valid? It is unclear where the line is being drawn here.

266 Id. at 594 (Scalia, J. concurring).

267Charles A. Sullivan, The World Turned Upside Down?: Disparate Impact Claims by White Males,
98 N W . L. R EV . 1505, 1505 (2004); Richard A. Primus, Equal Protection and Disparate Impact:
Round Three, 117 H ARV . L. R EV . 493, 528 (2003).

268See, e.g., City of Richmond v. J.A. Croson Co. 488 U.S. 469, 493 (1989) (plurality opinion). See
also Adarand Constructors, Inc. v. Peña, 515 U.S. 200 (1995) (Fifth Amendment due process case).

269 539 U.S. 244 (2003).



Electronic copy available at:

biracial. Further, especially given the prominence of identity politics today, it would be
impossible to exclude Hispanics from the protection afforded by strict scrutiny, even
though “Hispanic” is not a race but rather a broad ethnicity. And if one “minority”
ethnicity is included, it is difficult to see how other ethnicities—from Armenian
Americans to Zimbabwean Americans—could be left out. This quickly results in every
group being a minority.

Still for the purpose of the debate over disparate impact liability, what mattered in
Grutter was this: The Court held that strict scrutiny must be applied to a policy that
systematically disadvantages members of a minority race, it must also be applied it to a
policy that systematically disadvantages the majority race.270 That conclusion may seem
unsurprising. But apart from suggestions in the 1980s and 90s that disparate impact may
give rise to quotas, few, if anyone, had thought through the implications of that for
disparate impact liability.

Seton Hall law professor Charles A. Sullivan has argued that he used to “firmly
announce” to his students that disparate impact theory “was not available to whites and
males.”271 When the Court began taking the position that strict scrutiny must be employed
on behalf of members of majority as well as minority races, however, he began to realize
that applying disparate impact theory only on behalf of women and racial minorities
would raise serious constitutional difficulties.

Sullivan is right that disparate impact law has traditionally been applied only to
women and minorities. To my knowledge, the EEOC has never brought a disparate
impact investigation or lawsuit on behalf of white males. Nor has the Supreme Court
entertained a disparate impact case on behalf of anyone other than women and racial and
national origin minorities. Its past decisions indicate it did not expect to. In Griggs, the
court repeatedly noted that the purpose of disparate impact liability was to assist African
Americans or non-whites in particular. One of the “objective[s] of Congress in the
enactment of Title VII,” it wrote, “was to remove barriers that have operated in the past to
270 On the other hand, in applying the strict scrutiny standard in Grutter, the Court seriously
watered it down. Traditionally, for a racially discriminatory law to be upheld, a state must establish
that it is pursuing a “compelling interest” and that its law or policy is “narrowly tailored” to achieve
that purpose. Just as traditionally the standard is regarded as very high. Indeed, in 1972, leading
constitutional scholar Gerald Gunther famously called it “‘strict’ in theory and fatal in fact.” Gerald
Gunther, Foreword, In Search of Evolving Doctrine on a Changing Court: A Model for a Newer
Equal Protection, 86 H ARV . L. R EV . 1, 8 (1972). But Grutter told the unusual step of deferring to the
University of Michigan in determining whether its asserted purpose—diversity—was a compelling
interest. As many scholars have pointed out, “deference” and strict scrutiny are opposites. The
whole point is for the Court to refuse to defer to the state, but rather to conduct a “searching inquiry”
of its own.

271Charles A. Sullivan, The World Turned Upside Down?: Disparate Impact Claims by White Males,
98 N W . L. R EV . 1505, 1505 (2004); Richard A. Primus, Equal Protection and Disparate Impact:
Round Three, 117 H ARV . L. R EV . 493, 528 (2003).


Electronic copy available at:

favor an identifiable group of white employees over other employees.”272 It concluded that
if “an employment practice which operates to exclude Negroes cannot be shown to be
related to job performance, the practice is prohibited.”273

In 1981, the U.S. Commission on Civil Rights issued a report that flatly stated that
disparate impact liability “cannot be sensibly applied to white males” given that the
purpose of the liability is to uproot historical and contemporary sexism and racism.274
Contemporary commentators agreed. Professor Martha Chamallas stated, “In sum,
disparate impact has been inherently one-sided. Blacks and women may object to a test
that tends to reduce job opportunities for them. … It is probable that the courts, in an
effort to reduce the intrusion on employer discretion, will continue to limit disparate
impact challenges to those brought by minorities.”275

The only court to address the issue squarely—Livingston v. Roadway Express,

Inc.—also agreed that disparate impact theory is ordinarily unavailable to whites or
males.276 It held that white males are limited to lawsuits based on a theory of intentional
discrimination. It stated that “in impact cases . . . a member of a favored group must
show background circumstances supporting the inference that a facially neutral policy
with a disparate impact is in fact a vehicle for unlawful discrimination.” While a few
white, male private litigants have attempted to employ a disparate impact theory in Title
VII, to my knowledge, none has ever secured a judgment in his favor.

That was the zeitgeist when Congress undertook to amend Title VII with the 1991
Act. Members who discussed the various drafts of the disparate impact provisions to the
statute perceived it as applying to women and racial minorities in particular. Among the
many statements in the Congressional Record demonstrating this are:

Sen. John Glenn: “The Civil Rights Act of 1991 would reverse …
Wards Cove versus Atonio and restore . . . Griggs . . . . In Griggs, the
Supreme Court held that practices which disproportionately exclude qualified

272 Griggs v. Duke Power Co, 401 U.S. 424, 429-30 (1971) (italics added).

273 Id. at 431 (italics added).

OF DISCRIMINATION 21 n.20 (1981).

275Martha Chamallas, Evolving Conceptions of Equality Under Title VII: Disparate Impact Theory
and the Demise of the Bottom Line Principle, 31 U.C.L.A. L. R EV . 305, 366-68 (1983). See also David
A. Strauss, The Myth of Color Blindness, 1986 S UP . C T . R EV . 99 (arguing that affirmative action and
disparate impact theory are conceptually related).

276 802 F.2d 1250, 1252 (10th Cir. 1986).


Electronic copy available at:

women and minorities … are unlawful unless they serve a business

Sen. Howard Metzenbaum: The 1991 amendments provide “that

employment practices which disproportionately exclude women or minorities
are unlawful, unless employers prove both that these practices are ‘job
related …’ and that they are ‘consistent with business necessity.’”278

Rep. Charles Stenholm: “The substitute creates a new standard of

‘business necessity’ that a business must meet to defend an employment
practice whose result is a ‘disparate impact’—meaning the percentage of the
employer’s work force comprising women, minorities, or a given religious
group, does not almost identically match that group's percentage in the
available labor pool.”279

Rep. Harold Ford: “The Griggs standard worked well . . . . Under

Griggs, employers who chose to use selection practices with a significant
disparate impact on women or minorities had to defend the practices by
showing business necessity.”280

Those are just a few examples.281 Contemporaneous media reports also support
the understanding that the amendments’ disparate impact provisions apply only to
women and minorities. For example, the New York Times echoed what was said on
the floor of Congress with this statement: Under the amendments, “[i]f workers
show that a particular practice tends to exclude women or minority members, then

277 137 CONG. REC. 29,064 (1991).

278 137 CONG. REC. 33,483 (1991).

279 137 CONG. REC. 13,537 (1991).

280 137 C ONG . R EC . 13,530 (1991).

281 See also 137 C ONG . R EC . 29,026 (1991) (statement of Sen. Dodd) (“[I]n Wards Cove Packing Co.
versus Atonio, the Supreme Court overturned an 18-year precedent set by the Griggs . . . decision
regarding . . . discrimination based upon the disparate impact of business hiring of minorities.”); 137
C ONG . R EC . 29,048 (1991) (statement of Sen. Kohl) (“Under this proposal employers must justify
work rules if . . . the rules have a disparate impact on women and minorities.”); 137 C ONG . R EC .
13,539 (1991) (statement of Rep. Fish) (“The complaining party in a disparate impact case carriers
[sic] the heavy burden of linking adverse impact on women of [sic] members of minority groups to a
specific practice or practices unless the employer's own conduct essentially forecloses the possibility
of establishing such linkage.”). For additional examples, see Charles A. Sullivan, The World Turned
Upside Down?: Disparate Impact Claims by White Males, 98 N W . L. R EV . 1505, 1539-40 n.169


Electronic copy available at:

the employer must show that the practice is ‘job-related … and consistent with
business necessity.’”282

More recent scholars have agreed. University of Michigan law professor

Richard A. Primus has written that “[w]hat authority there is supports the view
that employment practices with disparate impacts on historically dominant classes
are, as a matter of law, not actionable under Title VII.”283 Similarly, Stanford law
professor John J. Donohue III has stated, “I conclude that disparate impact analysis
will not protect white males as a matter of theory. … The first prong of a disparate
impact case—finding a practice that adversely affects a member of a protected
class—will not be met since white males will not be deemed to be ‘protected’ under
this doctrine.”284
But if so, Grutter clearly demands that disparate impact liability as applied to
race be subjected to strict scrutiny. It is simple. The form of liability is designed to
individuals of a certain race or races and not of others. To be sure, that is not the
end of the story. Some laws and policies survive strict scrutiny. But the first step—
the application of strict scrutiny—is straightforward.
While one might be able to argue that disparate impact liability as applied to
sex need only passed “intermediate scrutiny,” the equal protection standard
ordinarily applied to cases of sex discrimination, that standard too is fairly high.
Moreover, it is doubtful that Congress (or the courts) would want to retain disparate
impact liability for women if it is prohibited as to race. It is hence doubtful that a
Court would separately retain disparate impact based on sex if it were to find
disparate impact liability based on race to violate the Constitution.
Professor Sullivan has recognized that if disparate impact liability applies only
to some races and not others, it must be subjected to strict scrutiny. He therefore
has urged a reinterpretation of disparate impact liability so that it would allow it to
benefit to whites and males as well as non-whites and females. But there are
several problems with such a reinterpretation, which would, in essence, extend the
reach of a statute whose reach is already extraordinary. To begin with, it isn’t what
the 1991 Congress—that is the only Congress that ever actually considered the
question of disparate impact liability in anything approaching a positive manner—
passed. Moreover, there is no evidence that Congress would have supported such an

282See, e.g., Robert Pear, With Rights Act Comes Fight to Clarify Congress’s Intent, N.Y. T IMES (Nov.
18, 1991).

Richard A. Primus, Equal Protection and Disparate Impact: Round Three, 117 H ARV . L. R EV .

493, 528 (2003).

John J. Donohue III, Understanding the Reasons for and Impact of Legislatively Mandated

Benefits for Selected Workers, 53 S TAN . L. R EV . 897 (2001).


Electronic copy available at:

Expanding the reach of a statute in order to save it from unconstitutionality is
tricky business. Suppose, for example, Congress had passed an unconstitutional tax
on Hispanics. It would be improper for a court to simply impose that tax on
everyone, since that court has no evidence that Congress would have been willing to
tax all Americans if it had known its original tax would be found unconstitutional.
The proper thing to do would be to nullify the original tax and let Congress decide
whether to promulgate legislation imposing a generally applicable tax. The same
logic applies here.
The second reason to reject Professor Sullivan’s attempt to “save” disparate
impact liability is that it fails to save it. Even if there were overwhelming evidence
that Congress would prefer a generally applicable disparate impact doctrine to no
disparate impact doctrine at all, it would make no difference. Generally applicable
disparate impact theory is racially discriminatory too. The Constitution protects
individuals from race discrimination, not groups.
Suppose, for example, disparate impact liability is applied to help African
Americans where they are under-represented and whites where they are under-
represented. If so, the result is more race discrimination, not color-blindness. It
doesn’t make a white applicant for a job as a New Haven firefighter feel better to
know that the playing field would have been tilted in his group’s favor if he were
applying for a position in the NBA. He isn’t qualified for the NBA even with a field
tilted sharply in his favor. But he is qualified to be a firefighter, and there his
ambitions are thwarted by policies that discriminate against him based on race.285

To put it another way, if half the job categories in the country discriminate
against one group and the other half discriminate against another group, that’s not
equity. That is a whole lot of discrimination. If the job that you wanted and are
most qualified for discriminates against you on the basis of race, you’ve been
victimized. Period. Few of us have the training and skills to move seamlessly from
one profession to another. Few of us would want to.

Consequently, regardless of whether disparate impact liability applies only to

women and minorities or to all groups, strict scrutiny must apply.

XI. Disparate Impact Liability Would Likely Fail to Survive Strict


285Another way in which disparate impact liability discriminates by race is in standing. Suppose a
white firefighter who applied for a promotion in New Haven failed the test. He does not benefit from
the fact that African Americans fail the test at higher rates than whites. He has no standing to sue.
Yet his individual predicament is no different from that of the African American applicants. As a
result of a test that he believes is not proven to be “necessary,” he didn’t get a promotion.


Electronic copy available at:

It is difficult to see how a doctrine as convoluted as disparate impact liability
can survive strict scrutiny. First of all, simply improving the job prospects of
women and minorities relative to men and whites has little chance of being held to
be a compelling governmental interest. A parallel argument was made in
University of California Regents v. Bakke (1978) and it was soundly rejected in
Justice Lewis Powell’s controlling opinion. 286 Powell was explicit: Admissions
preferences for racial minorities for the “purpose” of increasing the number of
minority doctors are unconstitutional. This is discrimination for discrimination’s
Even if improving the job prospects of women and minorities relative to others
were a plausible compelling governmental interest, disparate impact liability is not
narrowly tailored to achieve it. Evidence that disparate impact has in fact improved
the job prospects of women and minorities is decidedly lacking. Indeed, the
evidence tends toward the contrary. In the context of criminal record checks,
African Americans, the main group thought to benefit from the disparate liability,
may be worse off. Moreover, this is not the only area for which there is evidence of
unintended consequences. Credit checks, too, may be backfiring. If disparate
impact liability is causing employers to fall back on improper stereotypes (or even if
it is just causing them to position their businesses in race-neutral ways that have a
disparate impact), African Americans’ job prospects as a whole may be decreased
rather than increased.287
In addition, the tendency of disparate impact liability to cause employers to
adopt vague hiring and promotion standards or to be secretive about their
standards makes it easier to hide actual discrimination when it occurs. For all
these reasons it is hard to assert with anything close to the necessary confidence
that disparate impact liability is narrowly tailored to improve the job prospects of
women and minorities generally.
How much proof should be demanded in order to justify disparate impact
liability? Demands of proof imposed by the EEOC on employers in disparate impact
cases have been extraordinary: It wants scientific proof. Why should the courts
demand any less proof that disparate impact liability benefits women and
minorities (or that it has any sort of beneficial effect) than the EEOC has demanded
for employee selection procedures? If they do (or if they just demand a fraction of
that), it is unlikely they will get it.
A second candidate for a “compelling governmental interest” might be the
promotion of workplace diversity. In Grutter, the Court did indeed find that taking
advantage of the pedagogical advantages of a diverse class was a compelling
interest in the context of a law school education. But any direct analogy to Grutter

286 438 U.S. 265 (1978).

287 See supra Part IX.


Electronic copy available at:

v. Bollinger seems unlikely to carry the day.288 In Grutter, a 5-4 opinion, the Court
arrived at its conclusion by deferring to the judgment of the University of Michigan,
a creature of the State of Michigan. While the Court was criticized for this move at
least it had the fig leaf of academic freedom for its argument. Here there would be
no such leaf. The Court would have to defer to Congress (if it accepted the notion
that Congress really has acquiesced to disparate impact liability). That would mean
the end of strict scrutiny, the whole point of which is to hold lawmakers to an
extremely high standard. Deference is the opposite of strict scrutiny.
If the argument for a compelling interest is going to center on workplace
diversity, the Court will have to conduct its own “searching inquiry” into the issue.
Such an inquiry is unlikely to result in the conclusion that proponents of disparate
impact liability desire. “Diversity is Our Strength” is an often-expressed meme one
can find in workplace bulletin boards as well as all over the internet. But the
empirical data as reported by Elizabeth Mannix and Margaret A. Neale, however, is
much more mixed.289 Indeed, they report the weight of the evidence runs the other
As the workplace has become increasingly diverse, there has been a
tension between the promise and the reality of diversity in team
process and performance. The optimistic view holds that diversity
will lead to an increase in the variety of perspectives and
approaches brought to a problem and to opportunities for
knowledge sharing, and hence lead to greater creativity and quality
of team performance. However, the preponderance of the evidence
favors a more pessimistic view: that diversity creates social
divisions, which in turn create negative performance outcomes for
the group.290
A few good examples should suffice. Three professors of business, Barton H.
Hamilton, Jack A. Nickerson, and Hideo Owan, examined the productivity of
different teams at the Koret Corporation’s garment manufacturing facility in Napa,
California. This was a simple case for measuring productivity, since Koret teams
were paid by the number of pieces they produced and hence careful records could be
kept. Contrary to what one would expect from the optimistic view of diversity,
teams that were composed of only one ethnicity were the most productive, though

288 539 U.S. 306 (2003).

Elizabeth Mannix & Margaret A. Neale, What Differences Make a Difference?: The Promise and

Reality of Diverse Teams in Organizations, 6 P SYCH . S CI. P UB . I NTEREST 31 (2005).

290 Id.


Electronic copy available at:

this finding did not hold for marginal changes in team composition. Diversity in age
was also a negative for productivity.291
Another study entitled Homophily in Entrepreneurial Team Formation by Paul
A. Gompers, Kevin Huang and Sophie Wang is also interesting.292 “Homophily” is
defined as the tendency among individuals to associate and bond with others who
are similar. This was a study of first-year MBA students at the Harvard Business
School who were assigned to design and launch a real microbusiness. In the
Spring semester of 2013, students were assigned to teams of 5-7 individuals. These
teams were designed to be diverse in terms of sex, ethnicity, education and past
working experience. Later, in the Spring semesters 2014-16, students were
permitted to form their own teams. These teams tended to be somewhat less
diverse, thus providing an opportunity to study what effect team diversity has in
this context.
The authors found that “homophily” of ethnicity “increases team performance
by lifting teams in bottom quantiles to median performance quantiles, probably
because it reduces conflicts and enhance [sic] communication efficiency within the
group, but it does not induce superior performance in the right hand tail of the
distribution.”293 To state it in more succinct terms, “ethnic homogeneity reduces the
likelihood of bad outcomes, but does not increase the likelihood of extremely positive
outcome[s].”294 Significantly, however, homophily was not a negative for achieving
extremely positive outcomes.
Both these studies have obvious limitations. Harvard business are not a cross
section of America; indeed, many Harvard students have international
backgrounds. And their assigned task—to create a viable microbusiness—was not
representative of the various tasks American employers assign their employees to
do. While the Koret Corporation employees are a very different population from the
Harvard students and were assigned to do very different tasks, it is true of them too
that they are not a cross section of America and their task is not representative of
what American employees are asked to do. Still, in both cases diversity on the basis
of ethnicity was not found to increase performance. And the Mannix-Neale study
concluded that the preponderance of the evidence “favors a more pessimistic view”
of diversity.
291Barton H. Hamilton, Jackson A. Nickerson & Hideo Owan, Diversity and Productivity in

292Paul A. Gompers, Kevin Huang & Sophie Q. Wang, Homophily in Entrepreneurial Team
Formation, (Nat’l Bureau of Econ. Research, Working Paper No. 23459, 2017),

293 Id. at 23.

294 Id. at 24.


Electronic copy available at:

Note that analogizing the diversity issue under Title VII to the diversity issue
in Grutter is unlikely to be persuasive even putting aside the deference to academic
authority aspect of Grutter. In Grutter, the University of Michigan wanted to engage
in race-preferential admissions. There was no outside governmental entity
attempting to force the University of Michigan to accept more diversity than it
preferred. By contrast, disparate impact liability is much more aggressive. It is
imposed on employers. The argument for workplace diversity is that it is
advantageous to employers, allowing them to complete the tasks at hand better.
But how could it possibly be that the federal government has figured this out, but
the employers themselves (unlike the University of Michigan in Grutter) haven’t?295
Suppose, however, the Court is willing to hold just that: That even though
employers themselves are unaware of the advantages of diversity and even though
the preponderance of the empirical evidence shows that “diversity” in the workplace
is at best overrated, those advantages are great and Congress can force employers
to take advantage of them. That only brings us to the issue of whether disparate
impact liability is narrowly tailored to achieve that end. 296 Here, again, we are
faced with the problem that there is no evidence disparate impact liability even
helps increase workplace diversity.297 Indeed, there is some evidence that it does
just the opposite.
Even if there were such evidence, the notion that workplace diversity is a
compelling need cannot possibly be said to apply across the board. There are
thousands and thousands of different jobs out there, from airline pilot to zookeeper.
One might be able to make a case for diversity being an asset for some of them. But
it would be farcical to claim it for all of them.

295Nobel laureate economist Gary Becker has noted that race discrimination is apt to be at its worst
when it is imposed by law (as with Jim Crow laws) or when it is practiced by individuals or entities
with significant monopoly power (as with government, public utilities, or (under certain
circumstances) labor unions). Businesses with a taste for race discrimination that face significant
competition will often, over time, lose ground to their competitors and hence have an incentive to
reform themselves and go for the best employees it can get rather than the ones who are the “right”
race. If so, it is curious that the Supreme Court chose a state university to defer to in Grutter. While
private competition to state universities exist, the subsidies to state universities are so large that
their practical monopoly power is great. Disparate impact liability under Title VII, on the other
hand, applies to a vast number of wholly private businesses engaged in competitive, sometimes
highly competitive enterprises. This is where discrimination is least likely to thrive. Second-
guessing them on whether their particular business would benefit from diversity such that they
should be forced, against their will, to discriminate in favor of under-represented groups seems a

See Fisher v. Univ. of Tex., 570 U.S. 297 (2013) (holding that no deference is appropriate on

narrow tailoring even in the context of higher education).

And there is certainly no rigorous proof of the kind demanded in the Uniform Guidelines on

Employee Selection Procedures, Albemarle Paper, and post-1991 Act cases have required.


Electronic copy available at:

For example, few would claim that there is a “women’s way” as opposed to a
“man’s way” to work on an automobile manufacturing assembly line. Similarly, one
would be hard-pressed to come up with a reason that African American culture
makes a unique contribution to working as a dishwasher or a structural engineer.
Nor does any other culture. Arab Americans do not claim to bring a special
perspective to auto repair or to dry-cleaning. No group does. Whether it’s dental
surgery, taxi-driving or brick-laying, the diversity of one’s workforce is really not a
factor. At the very least then, disparate impact liability would be held not to be
narrowly tailored to serve the interest of diversity, since it applies to too many jobs.
But there are more problems still. In Parents Involved in Community Schools
v. Seattle School District No. 1 (2007), the Court was asked to decide whether
students can be assigned to schools based on their race in order to promote
classroom diversity. 298 Whether diversity was a compelling state interest in
Kindergarten – 12th grade turned out not to be a controlling issue, because the
Court held that if schools are interested in promoting diversity in the classroom
they may only do so through individualized assessments modeled after the kind of
college admissions process approved by the Court in Grutter.
Such a process would, at minimum, have required school districts to consider
the full range of ways that a particular student could bring diversity to a classroom.
For example, if schools were really concerned about capturing the pedagogical
benefits of diversity for their students, they will consider such things as socio-
economic class, experiences living in other countries, musical and artistic talent and
special knowledge. Adding an upper-middle class black student whose parents are
both brain surgeons to a class already full of white students whose parents are
physicians might promote diversity less than adding a poor white student who has
lived in a totalitarian country and writes poetry.
Disparate impact liability does not in any way promote these individualized
assessments. Consequently, even if a court were (1) to conduct its own inquiry into
the need for workplace diversity; and (2) conclude that that workplace diversity is a
compelling interest in all job categories and that disparate impact liability
effectively promotes workplace diversity, disparate impact liability would still not
survive strict scrutiny.
Under the circumstances, it makes sense to eliminate both “improving the job
prospects of women and minorities” and “workplace diversity” as compelling
interests that would allow disparate impact liability to survive strict scrutiny.
A third possibility that is likely to be advanced in favor of disparate impact
liability is that it helps root out actual discrimination based on race, color, religion,
sex and national origin. But disparate impact liability as it is currently practiced is
an awkward fit here. For the Court to find it is justified because of the need to root

298 551 U.S. 701 (2007).


Electronic copy available at:

out actual discrimination has the ring of “We had to destroy the village in order to
save it.”299 Disparate impact liability is itself actual discrimination.
On the other hand, the need to root out actual discrimination may indeed be
weighty enough to be a compelling governmental purpose. The problem is that if
that is the purpose of disparate impact liability, the doctrine is wildly overbroad.
The doctrine only makes sense as a weapon against actual discrimination if the fact
of disparate impact is simply evidence of intent to discriminate and can be
A good example is criminal background checks. Reasonable minds will differ
about the risk to the employer involved in a particular hiring of an ex-offender.
Reasonable minds will differ on whether employers have a moral duty to take some
risk in order to help re-integrate ex-offenders. But few will believe that employers
who have a policy against hiring ex-offenders are just using it as a means to engage
in race or national origin discrimination.
At most then, disparate impact should constitute a rebuttable presumption of
intent to discriminate. An employer would then have the opportunity to explain
why the trier of fact should not conclude that it was motivated by a desire to
discriminate on the basis of race, color, religion, sex, or national origin. In some
situations, this would be easy for the employer to do; in others it would take a lot to
overcome the presumption. Every case is different.
Griggs-style disparate impact liability, which is the subject of this article, does
not allow for disparate impact liability to be rebutted by evidence of lack of intent to
discriminate. The Court was explicit that the “absence of discriminatory intent
does not redeem” employment practices that have a disparate impact. Only
“business necessity” can do that, and “business necessity” is a concept that is
unrelated to discriminatory intent.
Consequently, while Griggs would not have to be overruled to fit disparate
impact liability into this this view of its purpose, it would need to be modified. And
while that modification would not solve all the problems associated with disparate
impact liability, it would solve some of them.

XII. The Problem of Disparate Impact’s Ubiquity: Are There Due

Process and Separation of Powers Issues Associated with the EEOC’s
299See Stephen L. Carter, Destroying a Quote’s History in Order to Save It: A Famous Vietnam War
Dispatch is Now 50 Years Old, But the Origins of the Phrase are Older than That, BLOOMBERG.COM,
(Feb. 9, 2018),

300See Larry Alexander, Disparate Impact: Fairness or Efficiency?, 50 S AN D IEGO L. R EV . 191



Electronic copy available at:

If the previous arguments (in Parts XI and XII) are meritorious, that means
the constitutional rights of job applicants and employees who are not members of
the particular group driving the disparate impact analysis in any particular case
are being violated.301 But what about employers? Are their constitutional rights
being violated too? Surely the ubiquity of disparate impact is reason for concern. It
is not easy to be a law-abiding employer in this environment.

An argument worth exploring might be this one: Suppose Congress passes a

statute that explicitly declares that an employer will be civilly liable for any pattern
of activity or practice in which it might engage in its capacity as an employer (e.g.
hiring, firing, promoting or assigning tasks). Suppose further that the statute
explicitly authorizes an executive agency to pick and choose which patterns of
activity or practices it wishes to pursue through guidances rather than through
notice-and-comment rules. That would surely raise troubling issues—perhaps most
obviously, due process issues. How different is that from Title VII as interpreted by

I will leave it to others to flesh out this argument more thoroughly. But I can
offer the following thoughts: Perhaps the hypothetical (as well as Title VII itself as
interpreted by Griggs) lies between two bodies of Supreme Court authority on the
constitutional implications of broad and vague delegations of power. One has held
legislative enactments to be unconstitutionally vague in the context of criminal
vagrancy laws and in some other settings too. The other—that involve the
delegation of rulemaking power to administrative agencies—has generally declined
to do so.

Consider first the cases holding many criminal vagrancy laws to be void-for-
vagueness. For example, in Papachristou v. City of Jacksonville (1972)302 an
ordinance read as follows:

Rogues and vagabonds, or dissolute persons who go about begging,

common gamblers, person who use juggling or unlawful games or

301At one time or another that likely includes all job applicants and employees or just about all of
them. In Griggs, it is non-African Americans who would have a high school diploma and would have
passed the two tests. In Dothard, it is men who are 5’2” or more and 120 pounds or more. In
NAACP v. North Hudson Reg’l Fire & Rescue, 665 F.3d 464, 477 (3d Cir. 2011), it was the Hispanic
and white non-Hispanic firefighters who would have qualified as residents. These are not the only
persons who are hurt. In Griggs, for example, African Americans who have a high school diploma
and would have passed the test are also put to a competitive disadvantage. But it is more difficult to
argue that Title VII (as interpreted) is discriminating against them on the basis of race.

302405 U.S. 156 (1972). See Coates v. Cincinnati, 402 U.S. 611 (1971). See also Johnson v. United
States, 135 S. Ct. 2551 (2015).


Electronic copy available at:

plays, common drunkards, common night walkers, thieves, pilferers or
pickpockets, traders in stolen property, lewd, wanton and lascivious
persons, keepers of gambling places, common railers and brawlers,
persons wandering or strolling around from place to place without any
lawful purpose or object, habitual loafers, disorderly persons, persons
neglecting all lawful business and habitually spending their time by
frequenting houses of ill fame, gaming houses, or places where
alcoholic beverages are sold or served, persons able to work but
habitually living upon the earnings of their wives or minor children
shall be deemed vagrants and, upon conviction in the Municipal Court
shall be punished as provided for Class D offenses.303

The facts of the case were these: Margaret Papachristou, another white
woman, and two African American men had been arrested under the ordinance for
driving in a car that stopped near a used car lot that had been broken into more
than once recently. Police denied that race had anything to do with it, but the
circumstances of the arrest suggest otherwise. Several other defendants in the case
had been arrested of violations of the act based on similarly dubious facts.

The Court held the ordinance to be void for vagueness “both in the sense that
it ‘fails to give a person of ordinary intelligence fair notice that his contemplated
conduct is forbidden by the statute’ . . . and because it encourages arbitrary and
erratic arrests and convictions.”304

The statute in Kolender v. Lawson (1983) was perhaps less sweeping. Among
other things, it required a person “who loiters or wanders upon the streets or from
place to place without apparent reason or business” “to identify himself” “ when
requested by any peace officer … if the surrounding circumstances are such as to
indicate to a reasonable man that the public safety demands such identification.”305
It was the requirement that such a person “identify himself” that was at issue in the
case. A California Court of Appeal had interpreted this provision to require “a
credible and reliable” identification, which in turn was defined as an identification
“carrying reasonable assurance that the identification is authentic and providing
means for later getting in touch with the person who has identified himself.”306

303 Jacksonville Ordinance Code § 26-57 (quoted in Papachristou, 405 U.S. 156, 156 n.1).

304 Papachristou, 405 U.S. at 162 (citations omitted).

305California Penal Code Ann. § 647(e) (West 1970) (quoted in Kolender v. Lawson, 461 U.S. 352, 352
n.1 (1983)).

306 People v. Solomon, 33 Cal. App. 3d 429, 438, 108 Cal. Rptr. 867, 873 (1973).


Electronic copy available at:

The defendant San Diego police chief admitted that acceptable identification
was highly discretionary on the part of the peace officer. For a jogger without his
wallet, reciting one’s name and address may be enough. For other cases, more
formal identification might be required. Failure to satisfy the officer was grounds
for arrest.

The Court found this to be unacceptably vague and hence unconstitutional:

“We conclude § 647(e) is unconstitutionally vague on its face because it encourages
arbitrary enforcement by failing to describe with sufficient particularity what a
suspect must do in order to satisfy the statute.”307

Both Papachristou and Kolender were criminal cases and hence

distinguishable from the case of disparate impact under Title VII.308 But no bright
line can be drawn between criminal and civil cases for the purposes of due process.
It cannot be (and it isn’t) that vagueness is never a due process issue in civil cases.
Indeed, the void-for vagueness doctrine has been held to nullify statutes in non-
criminal cases. In Federal Communications Commission v. Fox Television Stations,
Inc. (2012)(“Fox II”),309 the FCC assessed an administrative fine pursuant to 47
C.F.R. § 73.3999 against Fox Television Stations for allowing fleeting foul language
on the airwaves. The Court set aside that sanction on due process grounds (and not
on First Amendment grounds) holding the FCC’s policy to be impermissibly vague
to the point that Fox Television did not have fair notice or what sort of behavior was

Sessions v. Dimaya (2018) is another example.310 In that case, the Court dealt
was a statute that allowed for the deportation of aliens who are convicted of crimes
of violence, including “any other offense that is a felony and that, by its nature,
involves a substantial risk that physical force against the person or property of

307 Kolender, 461 U.S. at 361.

308 It is also true that both deal with defendants who are acting in their purely personal capacity and
not in a business or professional capacity. Ordinarily, we might expect businesses to extend a bit
more effort to understand the law in their particular area than we expect from individuals in a non-
business or professional setting. But that doesn’t mean that it is fine to expose businesses to vague
laws. In Connally v. Gen. Constr. Co., 269 U.S. 385 (1926), the respondent was acting in its capacity
as an employer. Oklahoma law required all state contractors to pay their employees current
prevailing wages for the locality and subjected them to criminal penalties for the failure to do so.
The Court held this to be impermissibly vague and a violation of due process. Disparate impact law
is so vague that no amount of effort will enable employers to avoid legal exposure. They just have to
hope that what they do will escape the notice of the EEOC.

309 567 U.S. 239 (2012).

310  138  S.  Ct.  1204  (2018).    Cf.  Caperton  v.  A.T.  Massey  Coal  Co.,  556  U.S.  868  (2009);  Mathews  v.  Eldridge,  424  

U.S.  319  (1976)  (due  process  issues  in  civil  cases).  



Electronic copy available at:

another may be used in the course of committing the offense.”311 Although the case
involved underlying criminal activity, the immigration statute was not criminal in
nature. Deportation in this context is a civil remedy. The Court nevertheless held
the immigration statute to be void for vagueness.

In his concurrence, Justice Gorsuch pointed out that historically American

courts have refused to apply vague laws “in criminal cases involving relatively
modest penalties” and that “they have applied the doctrine in civil cases too.”312 “As
one court put it,” he wrote, “‘all laws’ ‘ought to be expressed in such a manner as
that its meaning may be unambiguous, and in such language as may be readily
understood by those upon whom it is to operate.’”313

Fox II and Dimaya shouldn’t be taken to mean that identical standards of

vagueness can be applied to all areas of the law, criminal or non-criminal. There
are many non-criminal areas of the law; each is different. In Fox II, concerned an
administrative agency levying a fine. From the standpoint of the party paying fine,
if there is a distinction between a criminal fine and an administrative fine, it is a
subtle one. Both are essentially punishment; both are intended to deter behavior
deemed to be contrary to law. It would be awkward to hold that the void for
vagueness doctrine applies to statutes that allow judges to impose a fine, but not
statutes that allow agencies to impose a fine. Dimaya goes further. The
deportation remedy is not punishment. Punishing aliens more severely than
citizens is not what it is about and might well be unconstitutional. Instead, it is
simply a civil measure to prevent further crime within the country’s borders. It
applies to aliens, because does not own them the same responsibility to provide a
homeland and hence deportation is an option.

In both cases, it is a government official who made the determination to “go

after” a particular person and not others. In that respect, they are similar to
Papachristou and Kolender.

Contrast them with a “core” civil case (which to my mind would be a lawsuit
for unintentional tort resulting in physical injury). Even a first-year law student
knows that civil and criminal procedure are two very different things. The
safeguards that apply in the criminal context—like the requirement that all offense
be codified and the need to prove guilt beyond a reasonable doubt—do not always
apply in the civil context.
311  See  18  U.S.C.  §  16.  

312  Dimaya,  138  S.  Ct.  at  1226  (Gorsuch,  J.,  concurring).    For  the  latter  point,  Justice  Gorsuch  cited  Drake  v.  

Drake,  15  N.C.  110,  115  (1833)  and  Commonwealth  v.  Bank  of  Pennsylvania,  3  Watts  &  Serg.  173,  177  (Pa.  
313  Id.  (quoting  McConvill  v.  Mayor  and  Aldermen  of  Jersey  City,  39  N.J.L.  38,  42  (1876)).  


Electronic copy available at:

The reasons for this are complex. For one thing, in the typical core civil
lawsuit, the fact of an injury is obvious (or if it’s not the plaintiff bear the burden of
proving it). The issue is who should bear it. Since someone will have to bear the
loss, no matter what, the downside of an erroneous determination that the
defendant should bear it is thought to be less serious than the downside of
erroneously punishing the defendant.

But there are important structural differences between a tort lawsuit and
criminal case that make it more difficult to abuse the ability to bring a tort lawsuit
than it is to abuse the ability to bring a criminal prosecution. They are themselves
“procedural safeguards” (or at least substantive law that substitute for procedural
safeguards: A lawsuit can’t be brought by just anyone. It has to be an injured
person. And the injured person can’t choose sue just anyone. It has to be someone
who conduct proximately caused the injury. Plaintiffs bringing a lawsuit in
negligence often have discretion. They can sue the deep-pocket tortfeasor instead of
the impecunious one. But they cannot choose to sue someone who could have caused
their harm, but didn’t. Or someone who caused a similar harm to somebody else.
With precious few exceptions in extraordinary cases, the requirement that plaintiff
prove that the defendant proximately caused her injury is solid.314

Disparate impact cases are more like Papachristou, Kolender, Fox II and
Dimaya than they are like ordinary tort lawsuits. They place enormous discretion
in the hands of the EEOC or, in the case of a lawsuit against a state or local
governmental entity, in the hands of the Attorney General (and hence the
Department of Justice).315 The discretion placed in the hands of the EEOC is
especially troubling, because the EEOC, unlike the Department of Justice, is an
independent agency that does not report to the President, who has been elected to
exercise executive power. While President can remove the Attorney General or a
U.S. Attorney at his discretion, an EEOC Commissioner can only be removed for

314See, e.g., Sindell v. Abbott Laboratories, 26 Cal. 3d 588 (1980) (market share liability for
diethylstilbestrol manufacturers who marketed drug as miscarriage preventative). See also Gail
Heriot, An Essay on the Civil-Criminal Distinction With Special Reference to Punitive Damages, 7 J.
C ONTEMP . L EGAL I SSUES 43 (1996).

315Note that comparatively few disparate impact cases are brought by private plaintiffs (although
they are always a possibility). Employer must therefore be on guard. The reason for fewer private
cases is that Title VII allows for compensatory damages only for actual discrimination (“disparate
treatment”). That is not to say that a private plaintiff cannot get money for a disparate impact case.
But it must take the form of back pay. When disparate treatment is alleged and proven, the plaintiff
can get consequential damages too, which may include damages for emotional distress or for such
things as costs incurred in dealing with an eviction that resulted from not being able to pay the rent
on one’s apartment. And under the right circumstances punitive damages are available. Most
plaintiffs therefore strongly prefer to bring disparate treatment cases.


Electronic copy available at:

misconduct.316 Note that EEOC disparate impact investigations and lawsuits are
not just minor inconveniences. They routinely cost employers millions of dollars.317

The EEOC almost certainly would argue that it has endeavored to never use
its power unjustly. But even if courts were to make the heroic assumption that the
EEOC would always try hard to honor such a commitment, it wouldn’t matter. As
the Court stated in Fox II, “the due process protection against vague regulations
‘does not leave [regulated parties] … at the mercy of noblesse oblige.’”318

Moreover, EEOC procedures are rigged in precisely the opposite direction.

Under a procedure known as a “commissioner charge,” a single Commissioner can,
without any underlying complaint, initiate an investigation and eventually bring a
lawsuit case against any employer subject to Title VII. That is an incredible power
for a single commissioner. There is no single member veto.

There are additional elements that might drive disparate impact cases in the
direction of Papachristou, Kolender, Fox II and Dimaya. First, discrimination on
the basis of race, color, religion, sex and national origin is an activity that carries

316In Federal Communications Commission v. Fox Television Stations, Inc., 556 U.S. 502 (2009)
(“Fox I”) (Breyer, J., dissenting), Justice Breyer discusses why it is especially troubling for Congress
to vest excessive discretion in the hands of an independent agency, which cannot be controlled by a
President who has been elected according to democratic procedures. “[An independent] agency’s
comparative freedom from ballot-box control,” he wrote, “makes it all the more important that courts
review its decisionmaking to assure compliance with applicable provisions of the law—including law
requiring that major policy decisions be based upon articulable reasons.” Id. at 547. Although Justice
Scalia, who wrote the majority opinion, disagreed with Breyer on this point, the part of his opinion
that criticized Breyer’s point was not joined by a majority.

317Pepsi Beverages Co. agreed to pay $3.13 million and provide training and job offers to job
applicants who had been rejected on account of their criminal records as a result of an EEOC
investigation. In EEOC v. BMW Manufacturing Co., No. 7:13-cv-01583 (D.S.C. consent decree filed
Sept. 8, 2015), BMW Manufacturing Co. agreed to pay $1.6 million and other relief. In EEOC v. CSX
Transportation, Inc., Case No. 3:17-cv-03731 (S.D. W. Va.), CSX agreed to pay $3.2 million to settle a
lawsuit arising out of the disparate impact based on sex of the company’s strength tests.

When the EEOC brings an action under § 706, class members may turn out to be fewer than was
originally thought or their damages may turn out to be less than originally thought or for whatever
reason they may choose not to collect or may never learn of the funds. In those cases, the extra
funds are not returned to the employer. Rather, to ensure deterrence, they go designated charitable
causes under the doctrine of cy pres. For example, in EEOC v. Bass Pro, 4:11-CV-3425 (S.D. Tex.
2017), it was agreed that excess funds would be devoted to programs that engage inner city youth in
outdoor activity. Michelle Ma, Bass Pro Shops to Pay $10.5 Million to Settle Discrimination Suit,
WALL ST. J. (July 26, 2017),
settle-discrimination-suit-1501092896 [].
Fox II, 132 S. Ct. at 2318 (brackets and ellipsis in original)(quoting United States v. Stevens, 559

U.S. 460 (2010).


Electronic copy available at:

with it a great deal of social stigma. Indeed, in the modern world, the level of
stigma that attaches to accusations of such discrimination is unique. One can lose
jobs,319 customers,320 or contracts.321 Indeed, it happens with startling frequency.
Web sites are full of petitions calling for boycotts of businesses for some real or

319John Blake, How an Internet Mob Falsely Painted a Chipotle Employee as Racist, CNN.COM (May
27, 2019) (employee fired for appearing to be refusing service to African American males but offered
her job back when a stranger uncovered earlier tweets by her “victims” bragging about running away
with food without paying at Chipotle restaurants),
internet-mob-chipotle-video/index.html; Andy Ngo, At this Portland Bakery, White Guilt Poisons the
Batter, QUILLETTE.COM (June 5, 2018) (two bakery employees fired for declining to serve African
American who arrived after closing time; management acknowledged that employees’ action was not
racially motivated, but fired them anyway on the ground that the customer felt discriminated

320Civil Rights Leaders Call for Boycott of San Francisco Giants, ABC7NEWS.COM (Nov. 27, 2018)
(team owner donated to campaign of Republican candidate for Senate in Mississippi, who had
lightheartedly expressed her appreciation for a friend by stating, “If he invited me to a public
hanging, I’d be in the front row”),
sf-giants/4768421/; Janelle Griffith, Spike Lee, T.I. Boycott Gucci, Prada Over “Blackface” Fashion,
NBCNEWS.COM (Feb. 10, 2019) (sweater design that unintentionally resembles blackface condemned
as racist and companies boycotted),
gucci-prada-over-blackface-fashion-n969821; Allyson Chiu, “Haute Couture Blackface”: Gucci
Apologizes and Pulls “Racist” Sweater, W ASH . P OST , (Feb. 7, 2019, available at

321Eddie Scarry, Social Justice Mob Cancels Book, Because “Eating” Is a Racial Stereotype of
Something, W ASHINGTON E XAMINER (May 13, 2019) (novelist’s contract with book distributor
canceled after Twitter mob call her out as a racist for complaining about subway employee eating on


Electronic copy available at:

imagined racial transgression.322 One can get oneself besieged on Twitter.323 One
can get death threats.324

If Griggs was right, Title VII has taken conduct that is considered odious by
nearly all Americans and has defined that conduct to include things so far beyond
its core case that essentially any pattern of activity in which an employer might
engage as an employer is covered.

Second, unlike a prosecutor, the EEOC is in a position to issue guidances to

employers (as discussed in Parts III and IV) and follow them up with investigations
and lawsuits.325 Imagine if the prosecutors or the police in Papachristou and
Kolender had issued elaborate (but not exhaustive) lists of things that they believed
violated their broadly worded statutes. Most individuals would meekly avoid the
conduct on the list and hence increase their chances of avoiding prosecution. It may
well be that nobody would have enough at stake to qualify for a pre-enforcement
declaratory judgment. As a result, everyone’s freedom would be constricted
somewhat. Cumulatively this would be culturally transformative.


322E.g., Boycott Applebee’s Restaurants,

applebee-s-restaurants; Boycott Prada!,; Shut Down
Disneyland’s Racist Jungle Cruise Ride,
down-disneyland-s-racist-jungle-safari-ride; Boycott and Ban Bill Maher For Fueling Racism & Hate
In America: End “Real Time,”
racism-hate-in-america-end-real-time; Boycott United Airlines,; Starbucks: Commit to Fixing
Your Own Racism!,
See Jonah Goldberg, Twitter Mobs and the Constant Hunt for Fresh Outrage, AMERICAN


hunt-for-fresh-outrage/ .
324Dale Hurd, Covington Catholic Reopens Under Police Protection After Death
Threats, Like Tweet That Puts Boys in Wood Chipper, CBNNEWS.COM (Jan. 23,
325Note that unlike a notice-and-comment regulation, a guidance is in theory merely an
interpretation of what the law already is. That means that it states what the law was even before
the guidance was issued. An employer, therefore, cannot wait for a guidance to be issued to comply
with it. They apply retroactively. Indeed, the EEOC routinely brings pattern or practice lawsuits
against employers for their actions before a guidance is issued. Guidances are not just notice-and-
comment rules without going to all the fuss of notice and comment; in effect they function as ex post
facto laws.


Electronic copy available at:

That is a lot of power for an agency that Congress intended to have no
substantive rulemaking power. It is hard to believe the Congress intended this is in
1964, in 1972 or in 1991.

At some point, due process concerns over disparate impact liability under
Title VII must arise, must they not?

The second body of Supreme Court precedent concerns the non-delegation

doctrine. It differs from the previous body in two respects. First, while the previous
body is about notice to private parties and excessive executive discretion, the second
is about excessive legislative-like discretion in the hands of an administrative
agency. It is rare for the issue of notice to private parties come up in the non-
delegation case; the agency’s actions are usually quite clear. Second, the previous
body sometimes comes out in favor of the regulated private party; the modern non-
delegation cases don’t.

It has often been said that, for all practical purposes, the non-delegation
doctrine does not exist anymore—i.e. that the Supreme Court has permitted
Congress to delegate so much open-ended rulemaking authority to administrative
agencies that there may be no level of delegation that Congress will ever be inclined
to make that will not be approved.326 But disparate impact may be a special case.
Even the current, highly-deferential non-delegation doctrine does not go so far as to
permit Congress to make everything or nearly everything presumptively illegal and
then allow an agency without substantive rulemaking authority to pick and choose
which cases to pursue.

No one disputes that the Constitution permits Congress to endow executive

branch personnel with a certain level of quasi-legislative discretion. But there are
at least in theory limits to Congress’s authority to do so; too much discretion is
thought to be an exercise of true legislative power and hence unconstitutional.327
326See, e.g., Whitman v. Amer. Trucking Ass’n, Inc., 531 U.S. 457 (2001) (holding that Congress may
give the EPA rulemaking authority to set “ambient air quality standards the attainment and
maintenance of which in the judgment of the Administrator, based on . . . criteria . . . and allowing
an adequate margin of safety, are requisite to protect the public health”); Amer. Power & Light Co. v.
SEC, 329 U.S. 90 (1946) (holding that Congress may give the SEC the power to reject corporate
reorganizations that “unduly or unnecessarily complicate the structure" or "unfairly or inequitably
distribute voting power among security holders.”); NBC v. U.S., 319 U.S. 190 (1943) (holding that
Congress may grant the FCC the power to allocate broadcasting licenses in “the public interest,
convenience, and necessity”).

327The textual support for the doctrine that Congress may not delegate its “legislative” authority
arises from Section 1 of Article I of the U.S. Constitution states, “All legislative Powers herein
granted shall be vested in a Congress of the United States, which shall consist of a Senate and House
of Representatives.” There is no reason to believe this language was anything but deliberate—that
“[a]ll” was indeed meant to mean all.327 See Whitman, 531 U.S. at 472 (2001)(“In a delegation
challenge, the constitutional question is whether the statute has delegated legislative power to the


Electronic copy available at:

The classic formulation of the doctrine in this area was articulated in J.W.
Hampton, Jr. & Co. v. United States (1928), a tariff-setting case, and is generally
known as the intelligible principle standard. As the Supreme Court put it in that

“If Congress shall lay down by legislative act an intelligible principle to

which the person or body authorized to fix such rates is directed to
conform, such legislative action is not a forbidden delegation of
legislative power.”328

Whether the intelligible principle standard ever had any teeth in a case that
involves the authorization of notice-and-comment rulemaking to an executive
agency is a question that can be debated. What is clear is that the Supreme Court
has found no constitutional violation in cases in which Congress has delegated
extremely broad notice-and-comment rulemaking power to an administrative
agency created for a specific purpose. For example, in Whitman v. American
Trucking Association, Inc. (2001), the Supreme Court held that Congress may give
the EPA rulemaking authority to set “ambient air quality standards the attainment
and maintenance of which in the judgment of the Administrator, based on…

agency. Article I, § 1, of the Constitution vests ‘[a]ll legislative Powers herein granted … in a
Congress of the United States.’”)

As John Locke put it:

The power of the legislative, being derived from the people by a positive voluntary grant and
institution, can be no other than what that positive grant conveyed, which being only to
make laws, and not to make legislators, the legislative can have no power to transfer their
authority of making laws and place it in other hands.

J OHN L OCKE , S ECOND T REATISE OF G OVERNMENT (1690). Locke was not unrealistically rigid in his
thinking about the function of government. He well recognized the need for the executive to have
what he called “power to act according to discretion.” In his Commentaries on the Laws of England,
William Blackstone similarly noted that the crown could issue “binding” proclamations that are
grounded in the idea that while “the making of laws is entirely the work of … the legislative branch
…, yet the manner, time, and circumstances of putting those laws into execution must frequently be
left to the discretion of the executive magistrate.” W ILLIAM B LACKSTONE . C OMMENTARIES ON THE
L AWS OF E NGLAND (1765). Like Locke, of course, Blackstone would have been very familiar to the

328J.W. Hampton Jr. & Co. v. United States, 276 U.S. 394, 409 (1928). The intelligible principle
standard has been criticized. In Whitman v. American Trucking Association, Inc. (2001), Justice
Thomas stated in concurrence, “I am not convinced that the intelligible principle doctrine serves to
prevent all cessions of legislative power. I believe there are cases in which the principle is
intelligible and yet the significance of the delegated decision is simply too great for the decision to be
called anything other than ‘legislative.’” Whitman v. Amer. Trucking Ass’n, Inc., 531 U.S. 457,
487(2001) (Thomas, J., concurring).


Electronic copy available at:

criteria… and allowing an adequate margin of safety, are requisite to protect the
public health.”329

Nevertheless, such cases are distinguishable from Title VII (as interpreted by
Griggs). In the non-delegation line of cases, Congress did not begin by presumptively
prohibiting all or nearly all conduct in which an employer could engage and then
leave it to an executive agency to determine when to bring an enforcement action
and when not to. Instead, nothing was prohibited or mandated until the agency
promulgated a regulation prohibiting or mandating particular conduct.

Furthermore, in the non-delegation cases, the administrative agencies had to

go through notice-and-comment procedures, including all the safeguards associated
with those procedures, before regulations binding on regulated parties could be
issued. By contrast, the EEOC works through pattern or practice lawsuits
supplemented by guidances that end up having the force of notice-and-comment

In a sense, Title VII (as interpreted by Griggs) is the opposite of a delegation

problem. Congress didn’t delegate the authority to “legislate” to an administrative agency.
Instead it acted itself; indeed it over-acted, over-legislated by making everything or nearly
everything an employer might do in connection with employment presumptively illegal.
In a sense, the EEOC is doing what executive branch officials always do—making
decisions about how to allocate its resources to enforce the law as best it can. In that
sense, it is similar to Papachristou and Kolender. But in another sense for Congress to
prohibit everything or nearly everything that an employer might do is to fail to legislate at
all. In effect, it is the administrative agency that is legislating, but it is doing so through
its guidances and decisions to file lawsuits rather than through anything that seems in
form to be like legislation.

XIII. Conclusion.

Members of the 88th Congress tried much harder than they should have
needed to in order to ensure they would not be misunderstood. But in establishing

329 Whitman, 531 U.S. 457 (2001).

330In addition, in the non-delegation cases, the administrative agencies involved had somewhat
narrow jurisdiction relative to the EEOC. In Whitman, for example, the EPA’s authority applied only
to entities that discharged things into the air that the EPA considered harmful to public health. In
American Power & Light Co. v. SEC, 329 U.S. 90 (1946), the SEC’s authority was over corporate
reorganizations. In NBC v. United States, 319 U.S. 190 (1953), the FCC was charged with the
responsibility for allocating broadcast licenses, a hugely important task, but not one that gives it
authority over every business. The jurisdiction of the agency was fairly narrow. By contrast, any
administrative agency that controls employment essentially controls everything. Corporations act
through their employees. Without them, corporations cannot act at all.  


Electronic copy available at:

disparate impact liability, the EEOC and the Griggs Court saw to it those efforts
would be for naught.

Up until 1991, Griggs would have been an excellent candidate for an outright
and explicit overruling. But the Civil Rights Act of 1991’s backhanded recognition of
the disparate impact cause of action, while incomplete (it states a cause of action can
be maintained “only if” certain limitations are fulfilled and does not amend the
prohibitions of the statute or the provision that allows judicial remedies only for
“intentional” violations), makes that more difficult.

In his concurrence in Ricci, Justice Scalia make it clear that he that expected
that eventually the question of its constitutionality would come before the Court.
“The question is not an easy one,” he wrote.331

This article has discussed what I consider to be disparate impact’s ubiquity

(and hence the doctrine’s incoherence). It has discussed the astonishing discretion
that gives to the EEOC (and to a lesser extent private plaintiffs). And it has
questioned whether the doctrine has helped its intended beneficiaries, presenting
evidence that, at least in some of its more prominent contexts, it has not and that, by
creating incentives for employers to be vague about job qualifications, it can make it
harder to prove illegal discrimination or other kinds of favoritism. Finally, it has
sketched out arguments for disparate impact liability’s unconstitutionality.

The case that Scalia expected to come before the Court did not arrive in his
lifetime. It remains to be seen when or if it will arrive.

331 Ricci v. DeStefano, 557 U.S. 557, 594 (Scalia, J., concurring).


Electronic copy available at: