The But-For Theory of Anti-Discrimination Law
The But-For Theory of Anti-Discrimination Law
The But-For Theory of Anti-Discrimination Law
Citations:
Please note: citations are provided as a general guideline. Users should consult their preferred
citation format's style manual for proper citation formatting.
-- Your use of this HeinOnline PDF indicates your acceptance of HeinOnline's Terms and
Conditions of the license agreement available at
https://heinonline.org/HOL/License
-- The search text of this PDF is generated from uncorrected OCR text.
-- To obtain permission to use this article beyond the scope of your license, please use:
Copyright Information
THE BUT-FOR THEORY OF ANTI-DISCRIMINATION LAW
Katie Eyer*
This Article contends that the time has come to resolve the theoretical
crisis in anti-discrimination law. In a series of recent cases, the
Supreme Court has situated the question of whether an individual or
group would have fared differently "but for" their protected class
status as the central defining question of anti-discrimination law.
Moreover, the Court has suggested that this inquiry flows from anti-
discrimination law's plain text. As such, there are compelling
arguments to be made that a true disparate treatment principle-the
but-for principle-is the textually mandated inquiry in anti-
discrimination law, and that judicial deviations from this standard are
illegitimate.
This idea-that our anti-discrimination laws must reach all contexts
where the outcome would be different "butfor" the sex, race, or other
protected class status of those affected-is simultaneously conservative
in its aspirations and potentially radical in its legal effects. Such an
* Many thanks to Michael Carrier, Jessica Clarke, Tristin Green, Guha Krishnamurthi,
Alexandra Lahav, Shannon Minter, Zalman Rothschild, D'Andra Shu, Joseph Singer, Sandra
Sperino, Michael Selmi, Brian Soucek, Charles Sullivan, and Deb Widiss for helpful
conversations and feedback regarding this project, and to the editors of the Virginia Law
Review for excellent editorial suggestions. Special thanks are owed to Jessica Clarke, Sandra
Sperino, and Deb Widiss for extensive feedback. This Article was presented at the 15th
Annual Colloquium on Labor and Employment Law ("COSELL") and at the Association of
American Law Schools 2021 Meeting and received excellent feedback from participants.
1621
1622 Virginia Law Review [Vol. 107:1621
a. PrecedentialObstacles to Centering
"But For"........................................................ 1692
b. PrecedentialObstacles to Applying
"But For"........................................................ 1694
2. Statutory Language ................................................... 1695
C. JudicialHeadwinds.........................................................1700
1. Not Enough Summary Judgment ............................... 1701
2. Rule Creep.................................................................1704
3. JudicialDisfavorfor Anti-DiscriminationLaw
Claims ..................................................................... 1707
CONCLUSION ............................................................................... 1709
INTRODUCTION
Discrimination law has long been in theoretical crisis. Because its core
theory-disparate treatment-was recognized at a time when "disparate
treatment" and "intentional discrimination" were believed to be one and
1
the same, anti-discrimination law's foundational cases conflate the two.
This has created a fundamental question as to disparate treatment law's
central theoretical principles. Does disparate treatment law in fact prohibit
all "disparate treatment," (i.e., all decisions in which the outcome would
have been different "but for" race, sex, or other protected class status)?
Or does it prohibit only the narrower category of "intentional
discrimination" (i.e., decisions in which protected class status played a
conscious role)?
These questions about anti-discrimination law's core principles remain
unanswered even today, with important adverse consequences for anti-
discrimination law. In the absence of a clear commitment to barring
"disparate treatment," judicial law-making has run amok. Few judges
even ask the question of whether a policy decision-or an employment
action-would have turned out differently had the individual or group
affected been white, male, or of a majority religion. 2 Instead, across both
See, e.g., Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 335 & n.15, 382 (1977);
see also infra notes 39-60 and accompanying text (discussing this issue in depth).
2 See, e.g., Katie Eyer, The Return of the Technical McDonnell Douglas Paradigm, 94
Wash. L. Rev. 967, 1017 (2019). In the interest of brevity, I do not always list every protected
group when giving examples of the "but for" principle. This is not intended to suggest the
exclusion of other groups from the but-for principle, and, indeed, the application of the but-
for principle would be the same across all of the various contexts in which groups have
protections under statutory or constitutional anti-discrimination law.
1624 VirginiaLaw Review [Vol. 107:1621
3 See infra notes 61-71 and accompanying text; Sandra F. Sperino & Suja A. Thomas,
Unequal: How America's Courts Undermine Discrimination Law 13-14, 40-44, 158 (2017).
4 Gross v. FBL Fin. Servs., Inc., 557 U.S. 167, 180 (2009) (Age Discrimination in
Employment Act claim); see, e.g., Univ. of Tex. Sw. Med. Ctr. v. Nassar, 570 U.S. 338, 362
(2013) (Title VII retaliation claim); Comcast Corp. v. Nat'l Ass'n of Afr. Am.-Owned Media,
140 S. Ct. 1009, 1018-19 (2020) (42 U.S.C. § 1981 claim); Bostock v. Clayton Cnty., 140 S.
Ct. 1731, 1739-40 (2020) (Title VII sex discrimination claim); see also City of L.A. Dep't of
Water & Power v. Manhart, 435 U.S. 702, 711 (1978) (same); McDonald v. Santa Fe Trail
Transp. Co., 427 U.S. 273, 282 n.10 (1976) (Title VII race discrimination claim).
s See sources cited supra note 4.
2021] The But-For Theory of Anti-Discrimination Law 1625
But this Article suggests that the increased risks of embracing the but-
for principle are slight-and that the opportunity costs of not doing so are
considerable. The opportunity to recenter disparate treatment law around
what should be its core theoretical commitment is not one we ought to
take lightly. Without such a core theoretical commitment, we can expect
to continue to see an anti-discrimination law without any central rudder,
overrun by judge-made doctrines, and highly susceptible to individual
judicial biases. 9 Punitive or harmful government policies that would not
have been adopted "but for" the (minority) race of those affected will
continue to proliferate and go unremedied.' Employment decisions that
treat women, minorities and members of the LGBTQ community more
harshly than those who are men, white, cisgender and straight will
continue to be evaluated-and often dismissed-under a network of
doctrines that bear little relationship to whether differential treatment
occurred."
In contrast, an embrace of the but-for principle-and centering it as
anti-discrimination law's core commitment-offers myriad concrete
opportunities to argue for a more sensible and elegant approach to anti-
discrimination law. Under the but-for principle, our foundational inquiry
ought to be a simple and factual one: would the outcome have been
different "but for" the race, sex, or other protected class status of those
adversely affected? While in many cases answering this factual question
may be difficult-just as it is in, for example, tort claims-the procedure
for doing so is straightforward. The fact finder (jury or judge) ought to
consider all of the relevant evidence and consider whether it appears, by
a preponderance of the evidence, that a different outcome would have
resulted had the protected class status of those affected been different. For
example, would the Voter ID law have been passed, if those it had been
likely to disenfranchise were overwhelmingly white? Or would a man
9 See generally Sperino & Thomas, supra note 3, at 58-83 (describing in detail the judge-
made doctrines that the courts use to routinely dismiss discrimination claims); Katie Eyer,
That's Not Discrimination: American Beliefs and the Limits of Anti-Discrimination Law, 96
Minn. L. Rev. 1275, 1276 (2012) (noting that dismissals on summary judgment and motions
to dismiss are "extremely common in discrimination litigation, accounting for a full 86% of
litigated outcomes").
10 See, e.g., Petula Dvorak, We Scorned Addicts When They Were Black. It Is Different
Now That They Are White, Wash. Post (Apr. 12, 2018),
https://www.washingtonpost.com/local/we-hated-addicts-when-they-were-black-it-is-
different-now-that-they-are-white/2018/04/12/cd845f20-3e5b-11 e8-974f-
aacd97698cefstory.html [https://perma.cc/E49E-ZSGX].
" See Sperino & Thomas, supra note 3, at 1-4.
2021 ] The But-For Theory of Anti-DiscriminationLaw 1627
12 Cf. McCleskey v. Kemp, 481 U.S. 279, 293-95 (1987) (rejecting Equal Protection claim
in part on the grounds that statistical study could not identify the actors who engaged in
discrimination).
13 Cf. Linda Hamilton Krieger & Susan T. Fiske, Behavioral Realism in Employment
Discrimination Law: Implicit Bias and Disparate Treatment, 94 Calif. L. Rev. 997, 1029-38
(2006) (describing aspects of disparate treatment doctrine that assume the existence of a self-
aware discriminatory actor).
" See generally Sperino & Thomas, supra note 3 passim (detailing such doctrines and their
impact on anti-discrimination litigants); Eyer, supra note 9, at 1276 (noting that dismissals of
plaintiffs' claims at motions to dismiss and summary judgment account for 86% of litigated
outcomes in discrimination cases); Eyer, supra note 2, 969-72 (describing the ways in which
technical rules attached to the McDonnell Douglas paradigm are used to dismiss anti-
discrimination claims).
" See, e.g., Hamm v. Weyauwega Milk Prods., 332 F.3d 1058, 1066-67 (7th Cir. 2003)
(Posner, J., concurring).
1628 Virginia Law Review [Vol. 107:1621
16 See, e.g., Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1741, 1748-49 (2020).
17 See infra notes 226-29 and accompanying text.
'8 See infra Section IV.B.
19 See infra Part IV.
20 See infra Part V.
21 See infra Subsection V.A.2.
22 Id.
2021] The But-For Theory ofAnti-Discrimination Law 1629
differential treatment based on race, sex, national origin, etc.) rests at the
core of what most everyday people conceive of as invidious
discrimination. 26 And yet what should be a basic first-order question-
what is the central defining principle of disparate treatment law-remains
unsettled. This has left anti-discrimination law's core theory rudderless,
with predictably problematic results for the development of anti-
discrimination law doctrine.
This problematic failure to define anti-discrimination law's core theory
has arisen from a confluence of factors attributable at least in part to the
time and place in which modern disparate treatment doctrine originated.
Though both constitutional and statutory anti-discrimination
proscriptions predated Brown v. Board of Education, modern anti-
discrimination law, and with it, modern disparate treatment doctrine, can
trace its origins to the 1960s and 1970s. 27 As the Supreme Court grappled
with the sequelae of Brown and of the Civil Rights Act of 1964, it
developed the set of doctrines, including the modern day categories of
"disparate treatment" (i.e., differential treatment based on protected class
status) and "disparate impact" (i.e., differential impact on a protected
class not resulting from disparate treatment), which continue to largely
define anti-discrimination law today. 28 But neither the Court nor legal
advocates and observers were equipped at the time with a full
understanding of how discrimination operates in the disparate treatment
context.
Rather, as scholars such as Linda Hamilton Krieger have shown,
prevailing theories of discrimination in the 1960s and 1970s understood
29 See Linda Hamilton Krieger, The Content of Our Categories: A Cognitive Bias Approach
to Discrimination and Equal Employment Opportunity, 47 Stan. L. Rev. 1161, 1176 (1995).
30 Id.
31 Of course, even during the 1950s and 1960s, many racist actors believed their actions
were justified by race-neutral factors. See, e.g., Nancy Maclean, Freedom Is Not Enough: The
Opening of the American Workplace 49-50, 54-55, 67, 252 (2006); Katie Eyer, The New Jim
Crow Is the Old Jim Crow, 128 Yale L.J. 1002, 1025-31 (2019).
32 See Krieger, supra note 29, at 1187-88; Charles R. Lawrence 111, The Id, the Ego, and
Equal Protection: Reckoning with Unconscious Racism, 39 Stan. L. Rev. 317, 329-31 (1987);
Rachel D. Godsil & L. Song Richardson, Racial Anxiety, 102 Iowa L. Rev. 2235, 2247-56
(2017) (arguing that anxiety about being seen as racist or being profiled can result in disparate
treatment).
3 See, e.g., Eyer, supra note 31, at 1047-50.
14 See sources cited supra note 32.
1632 VirginiaLaw Review [Vol. 107:1621
can and does occur, even where "intent"-at least in the strong sense of a
self-aware discriminatory actor-does not exist.
It is important to emphasize that this form of disparate treatment (i.e.,
disparate treatment in the absence of strong self-aware intent) remains
disparate treatment, rather than disparate impact, as that doctrine has
been defined. For example, in a situation where a police officer shoots a
Black man because he perceives him as more dangerous based on his race,
that is disparate treatment, even if the officer does not consciously
perceive himself to be motivated by race. Similarly, a large employer that
systematically taps men, rather than women, for promotion, because men
are perceived of (because of their sex) as "management material" has
engaged in disparate treatment, even if the decision-makers involved are
unaware of the role that sex played in their decision-making. The key
question is would the outcome have been different "but for" the sex, race
or other protected class status of those affected? If so, disparate treatment
has indeed occurred.
In contrast, a true disparate impact model asks a different question.
Disparate impact doctrine imposes no requirement that the policy or
practice at issue was the result of disparate treatment, but instead simply
asks if it has a differential impact on a protected group (by race, sex,
etc.). 35 If the policy or practice produces such a differential impact,
disparate impact doctrine then asks the question of whether there is some
substantial justification for the policy or practice. 36 Finally, the plaintiff
has the opportunity to show that even if the policy is justified by neutral
factors, there is some alternative practice that would have a lesser impact
but would equally meet the defendants' objectives. 37 Thus, although
disparate impact doctrine shares with a true disparate treatment paradigm
a lack of a requirement of self-aware intent, it does not require a showing
that the outcome would have been different "but for" the protected class
status of those affected.
As such, it is clear today that there are two potential alternatives to a
disparate impact regime-"disparate treatment" and "intentional
discrimination"-alternatives that are not coextensive. But this set of
35 See 42 U.S.C. § 2000e-2(k); Ricci v. DeStefano, 557 U.S. 557, 577-78 (2009).
36 See sources cited supra note 35; see also Tex. Dep't of Hous. & Cmty. Affs. v. Inclusive
Cmty. Project, Inc., 576 U.S. 519, 533, 541 (2015) (describing the disparate impact standards
in the context of a Fair Housing Act case).
37 See sources cited supra note 35.
2021 ] The But-For Theory ofAnti-Discrimination Law 1633
understandings was not yet widespread in the 1960s and 1970s. 38 Thus,
our foundational disparate treatment cases reflect the prevailing
assumptions of the time-that disparate treatment and "intentional
discrimination" were one and the same. For example, in the constitutional
context, cases like Washington v. Davis-the seminal case which has
been understood to impose an intent standard-identify only one possible
alternative to a pure disparate impact standard, "discriminatory
39
purpose." But as the context of cases like Davis and its progeny make
clear, the Court did not consider itself to be opting for a purpose or intent
standard instead of a "disparate treatment" standard.40 Rather, the Court's
only aim in Davis was to reject a pure disparate impact standard in the
Equal Protection context-something it (erroneously) assumed was the
equivalent of a purpose inquiry.4 1
Similarly, early disparate treatment cases from the Title VII context
make clear that the Court understood itself to be defining the parameters
for a "disparate treatment" (as distinguished from a disparate impact)
paradigm but did not grapple with the possibility that disparate treatment
and "intentional discrimination" might not be coextensive. For example,
early cases in the individual disparate treatment context under Title VII
explicitly endorse a true disparate treatment standard, opining that "[t]he
central focus of the inquiry in a case such as this is always whether the
employer is treating 'some people less favorably than others because of
their race, color, religion, sex, or national origin,"'-while also
42
interchangeably referring to requirements of "motive" or "intent." So
too, such cases adopt an approach to the evidentiary structure and the
relevance of particular categories of proof that is consistent with a true
"disparate treatment" regime, without questioning whether such an
approach could reliably detect intent. 43 Even when requested by
defendants to address the question of whether "disparate treatment" could
exist independent of "discriminatory intent" in early statutory cases, the
Court simply ignored the question, apparently not perceiving any possible
distinction between the two.44
Thus, early individual disparate treatment cases in both the statutory
and constitutional context display a conceptual confusion of disparate
treatment and intentional discrimination. But the conflation of "disparate
treatment" with "intentional" discrimination is perhaps most apparent in
the Court's systemic disparate treatment cases of the same time. As the
Court developed the modern framework of anti-discrimination law
around the dual categories of "disparate treatment" and "disparate
impact," it further developed different doctrinal approaches for
"systemic" and "individual" disparate treatment. 4 5 As to the former
category, in which the allegation is that the defendant regularly and
systematically engaged in disparate treatment, the Court adopted a
method of proof, i.e., expected value statistics, that-while wholly
consistent with a true "disparate treatment" paradigm-can say little
about discriminatory intent. 46
As such, the central paradigm that continues to control the adjudication
of systemic disparate treatment cases today relies on properly constructed
immediately thereafter stating that the relevant inquiry is whether "the employer [is] treating
'some people less favorably than others because of their race, color, religion, sex, or national
origin."').
43 See McDonnell Douglas Corp. v. Green, 411 U.S. 792, 802-07 (1973); see also Banks
&
Ford, supra note 40, at 1073-80 (explaining that nothing in McDonnellDouglas turns on how
disparate treatment came about, whether from conscious or unconscious bias).
44 Furnco, 438 U.S. at 574 n.6; see also sources cited supra note 42.
4 See, e.g., Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 357-61 (1977).
46 See, e.g., id. at 337-40; Hazelwood Sch. Dist. v. United States, 433 U.S. 299, 307-09,
(1977); Castaneda v. Partida, 430 U.S. 482, 495-96, 496 n.17; see also Green, supra note 27
at 401-03, 411-17 (detailing the reasons why expected value statistics approach embraced in
Teamsters and Hazelwood, which has formed the crux of systemic disparate treatment
liability, does not necessarily allow a showing of purposeful discrimination). In some contexts
today, more sophisticated statistical methods, like regression analysis, take the place of
expected value in systemic disparate treatment contexts. See, e.g., Bazemore v. Friday, 478
U.S. 385, 398-401 (1986) (Brennan, J., concurring in part). However, for similar reasons,
regression analysis also would be incapable of detecting intent as opposed to disparate
treatment.
2021 ] The But-For Theory ofAnti-Discrimination Law 163 5
statistical design to ask the question of whether the "expected value," i.e.,
the expected numbers of a particular group receiving better or worse
outcomes in a particular context, is one that is likely to have resulted
absent disparate treatment. 47 Imagine, for example, that 50% of qualified
applicants for 100 teaching positions are African American, but that only
10% (10) of those hired are. 48 Assuming that all relevant non-racial
considerations have been accounted for in defining the relevant applicant
pool, we would have an "expected value" of 50 African American
teachers hired. 49 Statistical tests can then identify, based on sample size
and other considerations, whether this gap of 40 is statistically significant,
meaning that it is highly unlikely to have occurred absent disparate
treatment based on race. 0
Importantly this expected value methodology does not allow one to
identify intent, as opposed to disparate treatment.5 The biased decision-
maker who perceives African American candidates as less qualified, less
friendly, or less competent may well be unaware of the role that race is
playing in their decision, even as it produces disparate treatment on a
systematic scale, which the expected value test would detect. 2 Indeed,
history has shown that even among those who express explicit biases
openly, many may genuinely believe their actual decisions-which treat
minority groups differently-are based in the reality of non-invidious
factors.5 3 It is for this reason that early in the Civil Rights era, many
members of the white establishment could argue, and apparently
genuinely believe, that little actual disparate treatment existed, even as
they systematically perpetrated such disparate treatment.54
Nevertheless, the Court's seminal systemic disparate treatment cases-
like its seminal individual disparate treatment cases-do not address this
47 See generally Dianne Avery, Maria L. Ontiveros, Roberto L. Corrada, Michael Selmi
&
Melissa Hart, Employment Discrimination Law: Cases and Materials on Equality in the
Workplace 205-07 (8th ed. 2010) (describing "expected value" statistics but referring to it as
the "standard deviation" method).
'8 Id.
49 Id.
50 Id.
5I See generally Green, supra note 27, at 411-17 (detailing the ways that imposing a purpose
or intent requirement in systemic disparate treatment "would substantially alter the . . . use of
statistics [i.e., expected value statistics] endorsed by the Court and relied on by lower courts
and litigants for decades").
52 See sources cited supra note 32.
5 See sources cited supra note 31 and accompanying text.
54 Id.
1636 Virginia Law Review [Vol. 107:1621
tension, apparently because the Court was unaware it might exist. Thus,
seminal cases from both the statutory and constitutional context-such as
Teamsters, Hazelwood, and Castaneda-wholeheartedly endorse the
expected value methodology, while simultaneously opining that "[p]roof
of discriminatory motive is critical" for disparate treatment cases.55 As
the context of such cases make clear, the Court's objective in stating that
motive was important was, as in the individual disparate treatment
context, simply to distinguish disparate treatment from disparate impact.56
But the result was to conflate disparate treatment with "intentional
discrimination," leaving a profound theoretical conflict at the heart of
discrimination law.
This theoretical conflict has continued into the present day. Even in
contemporary cases, the Supreme Court has continued to conflate
disparate treatment with discriminatory intent and has failed to recognize
that more than one alternative to a pure disparate impact regime exists.5 7
Systemic disparate treatment cases continue to substantively rely on
statistical methodologies incapable of proving intent-while class
certification requirements for such cases assume intent is required. 58
Model jury instructions across the federal system employ different
standards ("disparate treatment" or "discriminatory intent") in charging
juries on the ultimate question in discrimination claims.5 9 Casebooks and
scholarly articles express a wide variety of mutually inconsistent
perspectives on what the core principles of disparate treatment law
ss See Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 324, 335-40, 335 n.15 (1977)
(accepting expected value statistical methodology as the central proof in a systemic disparate
treatment case, despite suggesting that discriminatory intent was the central requirement, in a
seminal Title VII systemic disparate treatment case, in which the court ruled on the systematic
non-hiring of Black and Spanish-surnamed drivers for "over-the-road" driver positions); see
also Hazelwood Sch. Dist. v. United States, 433 U.S. 299, 307-09, 308 n.13 (1977) (applying
the same methodology in the context of allegations of teacher non-hiring on the basis of race);
Castaneda v. Partida, 430 U.S. 482,495-97, 496 n.17 (1977) (applying the same methodology
in the context of constitutional allegations of systemic disparate treatment against Mexican-
Americans in grand jury selection).
56 See Teamsters, 431 U.S. at 336 n.15.
1 See, e.g., Young v. United Parcel Serv., Inc., 135 S. Ct. 1338, 1345, 1354-55 (2015);
Raytheon Co. v. Hernandez, 540 U.S. 44, 52-55 (2003).
58 See Green, supra note 27, at 405-17.
59 Compare Third Circuit Model Jury Instructions, 42 U.S.C § 1981, at 12-13 (Mar. 2018),
http://www.ca3.uscourts.gov/sites/ca3/files/6_Chap_6_2018_March.pdf (intentional
discrimination jury instruction), with Seventh Circuit Model Jury Instructions, § 3.01, at 59
(2017), http://www.ca7.uscourts.gov/pattern-jury-instructions/7th_cir_civilinstructions.pdf
[https://perma.cc/8TKA-QV29] (disparate treatment jury instruction).
2021 ] The But-For Theory ofAnti-Discrimination Law 1637
actually are, and whether the Supreme Court has embraced an "intentional
discrimination" or a true "disparate treatment" standard. 60
This continuing theoretical conflict has had important consequences.
In the statutory context, detethered from any core principle, judicial
lawmaking has run amok. Especially in the lower courts (where the vast
majority of discrimination cases are heard), the process of adjudicating
anti-discrimination claims bears virtually no relationship to any ultimate
question of discrimination-perhaps because it is not clear what that
ultimate question is. 6 1 Instead, judges deploy a wide array of technical
doctrines to, as Professors Sandra Sperino and Suja Thomas put it, "slice
and dice" claims, frequently leading to dismissal before trial. 62 In the
absence of a core central question to discipline the disparate treatment
inquiry, judges have felt free to develop legal doctrines that turn on
technical formalities, rather than the factual question of whether
discrimination took place. 6 3
Thus, for example, doctrines like the "same actor" doctrine are applied
by judges to dismiss cases where the same manager hired and fired the
plaintiff, even where there is compelling evidence that that manager was
bigoted. 4 Courts routinely dismiss cases for failure to show that an
employer's articulated reason for the decision is wholly false-even
65
where there is other persuasive evidence of disparate treatment. The
60 Compare Avery et al., supra note 47, at 89 (describing "discriminatory intent" as "the key
element in a disparate treatment" case), and David Benjamin Oppenheimer, Negligent
Discrimination, 141 U. Pa. L. Rev. 899, 923 (1993) (characterizing "conscious intent to
discriminate" as "the touchstone" of the Supreme Court's disparate treatment case law), with
Michael Selmi, Proving Intentional Discrimination: The Reality of Supreme Court Rhetoric,
86 Geo. L.J. 279, 287-89, 291-92, 294 (1997) (arguing that the Supreme Court's cases are
best understood as instantiating a causation-focused disparate treatment standard), and Noah
D. Zatz, Managing the Macaw: Third-Party Harassers, Accommodation, and the
Disaggregation of Discriminatory Intent, 109 Colum. L. Rev. 1357, 1374-75 (2009) (same).
61 See, e.g., Eyer, supra note 2, at 977-84.
62 See, e.g., Sperino & Thomas, supra note 3, at 152-55 (quoting Michael J. Zimmer, Slicing
& Dicing of Individual Disparate Treatment Law, 61 La. L. Rev. 577, 591 (2001)); Berrey,
Nelson & Nielsen, supra note 25, at 63.
63 See Eyer, supra note 2, at 977-84; Sperino & Thomas, supra note 3, at 152-55.
64 See, e.g., Maybin v. Hilton Grand Vacations Co., 343 F. Supp. 3d 988, 993-99 (D. Haw.
2018) (awarding summary judgment based on the "same actor" inference in an age
discrimination case, despite the fact that the manager regularly made ageist statements about
the plaintiff and other sales managers during plaintiff's period of employment); see also
Sperino & Thomas, supra note 3, at 69-71 (discussing the way the "same-actor" inference is
used against plaintiffs in statutory anti-discrimination cases).
65 See, e.g., Youry v. Exec. Transp. Co., No. 11-4103, 2013 WL 4774447, at *5-6 (E.D. Pa.
Sept. 6, 2013) (awarding summary judgment because the plaintiff did not show the employer's
1638 Virginia Law Review [Vol. 107:1621
reason was factually false-despite evidence that the decision-maker had repeatedly referred
to the plaintiff with racially derogatory remarks, including during the conversation in which
the plaintiff was terminated); see also Sandra Sperino, McDonnell Douglas: The Most
Important Case in Employment Discrimination Law 164-5 (2018) (describing this
phenomenon). Note that this lower court practice is directly contradictory to Supreme Court
precedent. See, e.g., McDonald v. Santa Fe Trail Transp. Co., 427 U.S. 273, 282 n.10 (1976)
(specifically noting that the plaintiff need not prove that the alleged reason was not a real
reason, just that protected class status was a "but for" cause); St. Mary's Honor Ctr. v. Hicks,
509 U.S. 502, 514, 524-25 (1993) (holding that it is the factual question of discrimination, not
pretext, which is the ultimate inquiry in a Title VII case, even one brought via the McDonnell
Douglas paradigm).
66 See, e.g., Wallace v. Methodist Hosp. Sys., 271 F.3d 212, 222-24 (5th Cir. 2001)
(awarding JMOL after a jury verdict in plaintiff's favor on pregnancy discrimination claims
and dismissing as a "stray remark" the plaintiffs supervisor's overheard statement that the
plaintiff was fired because "she's been pregnant three times in three years."); see also Sperino,
supra note 65, at 204-09 (discussing how the stray remarks doctrine is used against
discrimination plaintiffs); Sperino & Thomas, supra note 3, at 60-69 (same); Jessica A.
Clarke, Explicit Bias, 113 Nw. U.L. Rev. 505, 540-47 (2018) (same). This lower court
practice also has been at least implicitly rejected by the Supreme Court. See Reeves v.
Sanderson Plumbing Prod. Inc., 530 U.S. 133, 152-53 (2000) (reversing a lower court decision
that had applied the stray remarks doctrine to disregard damning age-related remarks and
stating that the lower court impermissibly substituted its judgment for the jury's).
67 See, e.g., Mitchell v. Mills, 895 F.3d 365, 371-72 (5th Cir. 2018) (denying plaintiffs
claim of wage discrimination because the contractors that were used as comparators were not
similar enough to the plaintiff to meet the imposed "nearly identical" standard); Suzanne B.
Goldberg, Discrimination by Comparison, 120 Yale L.J. 728, 745-55 (2011); Charles
Sullivan, The Phoenix from the Ash: Proving Discrimination by Comparators, 60 Ala. L. Rev.
191, 216-20 (2009); see also Lewis v. City of Union City, 918 F.3d 1213, 1221-29 (11th Cir.
2019) (en banc) (delineating all of the ways a comparator must be the same in order to satisfy
the circuit's standards and support a prima facie case). The Supreme Court's decision in
McDonald v. Santa Fe Trail seems to directly repudiate this approach. See McDonald, 427
U.S. at 283 n.11.
68 See, e.g., Eyer, supra note 2, at 977-85. Often, these doctrines are tied to the application
of the so-called McDonnell Douglas paradigm. For a compelling argument for why the
McDonnell Douglas paradigm should be abandoned in favor of a textualist approach to anti-
2021] The But-For Theory of Anti-Discrimination Law 1639
discrimination claims, see Deborah A. Widiss, Proving Discrimination by the Text, 106 Minn.
L. Rev. (forthcoming 2021) (manuscript at 2-5).
69 See, e.g., McCleskey v. Kemp, 481 U.S. 279, 292-93 (1987).
70 See, e.g., Michelle Alexander, The New Jim Crow: Mass Incarceration in the Age of
Colorblindness 125-29 (2010); David Cole, No Equal Justice: Race and Class in the American
Criminal System 161-68 (1999); see also Fred O. Smith, Jr., Abstention in the Time of
Ferguson, 131 Harv. L. Rev. 2283, 2322 (2018) (discussing the ways that Younger abstention
arguments have led some courts to reject structural arguments challenging the criminalization
of poverty).
71 See, e.g., United States v. Armstrong, 517 U.S. 456, 468-70 (1996); Clarke, supra note
66, at 524-40, 547-71.
72 See, e.g., sources cited supra note 60.
73 See, e.g., Oppenheimer, supra note 60, at 967-72; Eyer, supra note 23, at 3 n.2 (collecting
sources).
74 See, e.g., Michael Selmi, Was the Disparate Impact Theory a Mistake?, 53 UCLA L. Rev.
701, 702-07 (2006) (observing and critiquing the extensive scholarly focus on disparate
impact); Charles A. Sullivan, Making Too Much of Too Little?: Why "Motivating Factor"
Liability Did Not Revolutionize Title VII, 62 Ariz. L. Rev. 357, 358-59 (2020) (describing
1640 Virginia Law Review [Vol. 107:1621
the extensive attention paid to "motivating factor" liability in the scholarly literature and
observing that "[o]ne wonders what all the fuss is about" in view of the disappointingly small
impact that "motivating factor" has had on anti-discrimination law).
75 It is important to note that other scholars have made arguments in the past that a true
disparate treatment principle does or should reside at the heart of anti-discrimination law. See,
e.g., Selmi, supra note 60, at 287-94; Zatz, supra note 60, at 1374-75; Green, supra note 27,
at 414-17. But, in the face of the ongoing conceptual crisis in anti-discrimination law, such
arguments have yet to generate a wider movement for focusing anti-discrimination law on a
true disparate treatment standard. This Article argues that the Supreme Court's recent case
law centering the "but for" principle offers perhaps a critical opportunity to spark such a wider
movement for a true disparate treatment standard.
76 See, e.g., Sharon Dietrich, Maurice Emsellem & Catherine Ruckelshaus, Work Reform:
The Other Side of Welfare Reform, 9 Stan. L. & Pol'y Rev. 53, 67 n.51 (1998) (noting that
the Civil Rights Act of 1991 arose in significant part from the outcry in the civil rights
advocacy community over the retrenchment of disparate impact liability in Wards Cove
Packing Co. v. Atonio, 490 U.S. 642 (1989)).
77 See, e.g., Eyer, supra note 2, at 1008 & n.250, 1016.
2021] The But-For Theory ofAnti-Discrimination Law 1641
78 Id. at 1008; see also e.g., Plaintiff-Appellant's Reply Brief at 2-7, Kozuma v. Barnhart,
2002 WL 32102095 (9th Cir. 2002) (No. 01-16109) (arguing that the "same actor inference"
did not apply given the circumstances of the case, but not contesting the legitimacy of the
doctrine).
79 Eyer, supra note 2, at 1016.
1642 Virginia Law Review [Vol. 107:1621
80 435 U.S. 702, 711 (1978); see also McDonald v. Santa Fe Trail Transp. Co., 427 U.S.
273, 282 n.10 (1976) (observing, two years before Manhart, that "no more is required [under
Title VII] to be shown than that race was a 'but for' cause.").
$' Manhart, 435 U.S. at 704.
82 Id. at 708-11.
83 Id. at 711 (emphasis added) (quoting Developments in the Law, Employment
Discrimination and Title VII of the Civil Rights Act of 1964, 84 Harv. L. Rev. 1109, 1170
(1971)).
2021 ] The But-For Theory ofAnti-DiscriminationLaw 1643
9" Comcast, 140 S. Ct. at 1014. Scholars have criticized both the historical accuracy of the
Court's characterization of the pedigree of the but-for causation test in tort law and the
appropriateness of its use of a tort frame in anti-discrimination law. See, e.g., Alexandra D.
Lahav, Why Justice Gorsuch Was Wrong About Causation in Comcast, 23 Green Bag 2D 205,
205 (2020) (questioning the historical accuracy of the assertion in Comcast that but-for
causation is an "ancient and simple" test); Sandra Sperino, The Tort Label, 66 Fla. L. Rev.
1051, 1052-54 (2014) (critiquing the Supreme Court's tendency to draw on tort law in
interpreting discrimination statutes). Nevertheless, this Article takes as its starting point the
assertions of the Supreme Court in recent cases such as Comcast and Nassar.
92 As Sandra Sperino has observed, the fact that Comcastwas a virtually unanimous opinion
signals that even the progressives on the Court have acceded to the controlling nature of the
but-for principle. See Sandra Sperino, Comcast and Bostock Offer Clarity on Causation
Standard, A.B.A. Human Rights Mag., Jan. 12, 2021, at 24-25.
93 Gross and its progeny have been critiqued by legal scholars on a variety of grounds. In
addition to those critiques discussed supra note 91, scholars have argued that Gross and its
progeny are not the correct reading of the relevant statutory text, that they disregarded the
legislative history of the Civil Rights Act of 1991, that the precedent of Price Waterhouse v.
Hopkins should have compelled a different result, and that they will deprive civil rights
litigants of adequate opportunity to make out their claims. See, e.g., Deborah A. Widiss,
Undermining Congressional Overrides: The Hydra Problem in Statutory Interpretation, 90
Tex. L. Rev. 859, 926-41 (2012); Leora F. Eisenstadt, Causation in Context, 36 Berkeley J.
of Emp. & Lab. L. 1, 3 n.6 (2015); James A. Macleod, Ordinary Causation: A Study in
Experimental Statutory Interpretation, 94 Ind. L.J. 957, 959-63 (2019); sources cited infra
notes 271-72. This Article takes as its starting point Gross and its progeny's endorsement of
the "but for" principle, and their description of the principle as textualist in nature, and thus
does not re-litigate methodological disputes over the principle's correctness or origins.
Substantive arguments regarding the adequacy of the "but for" principle as a vehicle for civil
rights litigation are addressed infra Parts III-V.
2021] The But-For Theory ofAnti-Discrimination Law 1645
94 See, e.g., Comcast, 140 S. Ct. at 1015; Nassar, 570 U.S. at 350-52; Gross v. FBL Fin.
Servs., 557 U.S. 167, 175-78 (2009).
95 Comcast, 140 S. Ct. at 1015 (emphasis added) (quoting 42 U.S.C. § 1981(a)).
96 Id.
1646 VirginiaLaw Review [Vol. 107:1621
104 See, e.g., Brief for Petitioners Altitude Express, Inc., and Ray Maynard at 13-18, 38-39,
Bostock, 140 S. Ct. 1731(No. 17-1623).
05 See Brief for Respondent Aimee Stephens, supra note 101 at 28-29; Brief of Statutory
Interpretation and Equality Law Scholars as Amici Curiae in Support of the Employees, supra
note 101 at 4-12.
106 See sources cited supra note 100.
107 See Bostock, 140 S. Ct. at 1740.
08 Id. at 1741-43, 1745--46.
1648 VirginiaLaw Review [Vol. 107:1621
109 Id.
"0 See Zatz, supra note 60, at 1376-78.
"' Mitchell N. Berman & Guha Krishnamurthi, Bostock Was Bogus: Textualism, Pluralism,
and Title VII, 97 Notre Dame L. Rev. (forthcoming 2021) (manuscript at 28).
112 Id.
113 Id. Berman and Krishnamurthi go on to reach what I think are erroneous conclusions
about the "but for" principle's capability of resolving the question at issue in Bostock. Id. at
28-29. But I agree with them that this scenario would not be actionable, albeit for reasons that
are distinctive from their own.
2021] The But-For Theory ofAnti-Discrimination Law 1649
114 See Bostock, 140 S. Ct. at 1741-43, 1745-46; see also Zatz, supra note 60, at 1377
(noting that internal membership causation is essential to disparate treatment claims).
115 As Zatz points out, there are some areas of anti-discrimination law that do not require
internal membership causation, but disparate treatment doctrine is not among them. See Zatz,
supra note 60, at 1373-82 (demonstrating that while disparate treatment requires internal
membership causation, third-party harassment cases do not require internal membership
causation for the plaintiff to prevail).
116 In my conversations with other anti-discrimination law scholars, many have expressed
that they found this part of the Court's opinion confusing and unpersuasive.
117 To the extent that some principle is needed to avoid scenarios like the one that Berman
and Krishnamurthi postulate in the disparate treatment context, the concept of internal and
external membership causation, which Zatz originated, seems a more helpful and intuitive way
of addressing the issue than the formulation of intent at issue in Bostock.
1650 VirginiaLaw Review [Vol. 107:1621
120 Comcast Co. v. Nat'l Ass'n of Afr. Am.-Owned Media, 140 S. Ct. 1009, 1016-17 (2020).
121 See, e.g., Gen. Bldg. Contractors Ass'n v. Pennsylvania, 458 U.S. 375, 383-91 (1982).
122 Id. at 384-85.
123 Id. at 389-90.
124 Of course, there is also a potential argument that the Court got it wrong in Comcast and
that motivating factor liability should have been found to be available since the Court had
already held that motivating factor liability is available in the Equal Protection context. See
Vill. of Arlington Heights v. Metro. Hous. Dev. Corp., 429 U.S. 252,265-70,270 n.21 (1977).
't Comcast, 140 S. Ct. at 1016-17.
126 Id.
1652 VirginiaLaw Review [Vol. 107:1621
"because of' or "on the basis" of race, they too endorse the but-for
principle as the relevant standard.' 27
While the assumptions on which this approach rests may be
questionable, taken at face value, it offers considerable opportunities to
argue that the but-for principle extends to the Equal Protection context.
Indeed, dozens of the Supreme Court's Equal Protection cases, ranging
from the immediate aftermath of the Fourteenth Amendment to today, use
precisely this type of language ("because of," "by reason of') to describe
the Equal Protection standard.1 28 Thus, for example, cases such as
Adarandv. Penarepeatedly state the principle that "government may treat
people differently because of their race only for the most compelling
reasons."1 29 So too, as the Supreme Court observed in Bolling v. Sharp,
"[a]s long ago as 1896, this Court declared the principle 'that the
Constitution of the United States ... forbids ... discrimination ...
against any citizen because of his race. ""30 Even decisions closely
associated with a strong requirement of intent or purpose, like Personnel
Administratorv. Feeney, ultimately state that such discriminatory purpose
exists where "the decisionmaker . .. selected or reaffirmed a particular
course of action at least in part 'because of,' not merely 'in spite of" its
effects on a particular group.'31
Finally, other well-established interpretive principles are likely to also
aid in making the argument that the but-for principle ought to reside at the
core of the constitutional anti-discrimination inquiry, not just the statutory
one. The Supreme Court has typically treated disparate treatment
discrimination in the constitutional and statutory contexts as following the
same principles, relying on cases from one context to inform the other.' 3 2
And there is nothing in the text of the Equal Protection clause itself that
would bar an interpretation that rested on a true disparate treatment
standard-indeed, arguably that is the most natural reading of the clause.
127 Id.
128 See, e.g., infra notes 129-31 and accompanying text.
129 Adarand Constructors, Inc. v. Pena, 515 U.S. 200, 227, 235 (1995) (emphasis added).
30 Bolling v. Sharpe, 347 U.S. 497, 499 (1954) (quoting Gibson v. Mississippi, 162
U.S.
565, 591 (1896)) (emphasis added).
131 Pers. Adm'r v. Feeney, 442 U.S. 256, 279 (1979); see also Selmi, supra note 60, at 291-
92 (interpreting Feeney as adopting what David Strauss has referred to as the "reversing the
groups" test, i.e., but-for causation).
132 See Primus, supra note 7, at 1354-55.
2021 ] The But-For Theory ofAnti-Discrimination Law 1653
("No State shall .. . deny to any person within its jurisdiction the equal
protection of the laws.") 133
equivalent of the "simple" and "traditional" tort law inquiry into "cause
in fact." 136 Thus, just as in the tort law context, the question that the but-
for principle asks is a factual, not a legal, one: would the outcome have
been different "but for" protected class status? 137 Assessing that factual
question at the trial stage simply requires the fact fmder to consider all of
the evidence offered by the plaintiff and the defendant in order to assess
whether the fact finder is persuaded-by a preponderance of the
evidence-that the outcome would have been different "but for" the race,
sex or other protected class status of the plaintiff.
Because the vast majority of discrimination cases are civil cases, it is
important to note that metaphysical certainty is not required in order for
a plaintiff to prevail at trial in making this showing.' 38 Rather, the plaintiff
must only persuade the fact finder (typically a jury) that it is "more likely
than not" that the outcome would have been different "but for" the
protected class status of the plaintiff.139 Thus, while there may be
circumstances in which it would be difficult or impossible to prove with
absolute certainty that "but for" causation exists, those circumstances tend
to be of little practical significance in reality.'4 0 Because of the standards
136 See, e.g., Univ. of Tex. Sw. Med. Ctr. v. Nassar, 570 U.S. 338, 346-47 (2013); Bostock
v. Clayton Cnty., 140 S. Ct. 1731, 1747 (2020).
137 See, e.g., Restatement (Third) of Torts: Liability for Physical and Emotional Harm § 26
(Am. L. Inst. 2010); see also U.S. Postal Serv. Bd. of Governors v. Aikens, 460 U.S., 711,
715-17 (1983) (emphasizing that the fundamental question in a Title VII disparate treatment
case is a factual question); St. Mary's Honor Ctr. v. Hicks, 509 U.S. 502, 511, 524 (1993)
(same).
138 For an overview of the criminal civil rights statutes that are subject to enforcement by
the Department of Justice, see, e.g., Alison M. Smith, Overview of Selected Federal Criminal
Civil Rights Statutes, Cong. Rsch. Serv. (Dec. 16, 2014),
https://sgp.fas.org/crs/misc/R43830.pdf [https://perma.cc/FB9F-L84G]. Criminal civil rights
enforcement may raise distinctive issues and is not the primary subject of this Article.
139 See, e.g., Ruth Maurice, Legal Standards of Proof, NOLO https://www.nolo.comlegal-
encyclopedia/legal-standards-proof.html [https://perma.cc/4KXJ-N5GR] (last visited Oct. 26,
2021) (describing the civil "preponderance of the evidence" standard of proof).
14 For this reason, I do not view Alexandra Lahav's theory of "chancy causation" to be
particularly relevant for the discrimination context. See Alexandra D. Lahav, Chancy
Causation in Tort, 14 J. Tort L. (forthcoming 2021) (manuscript at 2-3) (on file with author).
While there may not be epistemic certainty in applying the but-for principle given the
complications of fully reconstructing the counterfactual of a situation in which the plaintiff
was outside the protected class, there is no need for such a perfect certainty. Rather, the
plaintiff need only persuade the jury that it is "more likely than not" that the outcome would
have been different "but for" protected class status. Thus, while Lahav argues that the standard
of proof does not alleviate the problems of uncertain causation, id. at 10-11,1 disagree, at least
in those circumstances (such as discrimination) where causation is not exclusively proved via
probabilistic statistical studies.
2021] The But-For Theory ofAnti-Discrimination Law 1655
of proof in a civil case, the plaintiff need not achieve absolute certainty
via their proof.141 Rather, the plaintiff need only persuade the fact finder
that it is more likely than not, i.e., slightly more probable than not, that
the outcome would have been different but-for their protected class
status.14 2 In answering that question in a discrimination case, fact fmders
can (and must) ultimately rely on their own judgment to decide whether
it is more likely than not that the outcome would have been different for
someone outside the protected class-regardless of whether epistemic
certainty is possible. 143
Moreover, this is the most demanding standard that plaintiffs should
face under a but-for approach at any stage of the litigation. As the
Supreme Court held in its Comcast v. National Association of African-
American Owned Media decision, "the essential elements of a claim
remain constant through the life of a lawsuit."1 44 Thus, "while the
materials the plaintiff can rely on to show causation may change as a
lawsuit progresses from filing to judgment, the burden itself remains.
constant." 14 5 As such, in the context of a "but for" case, the burden at all
procedural stages is to show, to the standard demanded by the particular
procedural stage, that the outcome would have been different "but for"
the protected class status of those affected. Thus, while courts have often
applied distinctive and more difficult substantive standards to summary
judgment or even to motions to dismiss in anti-discrimination cases, this
is erroneous. Rather, the only permissible judicial inquiry under Comcast
is whether the plaintiff has plausibly alleged that they were treated
differently "but for" their protected class status (at the motions to dismiss
phase), or that a reasonable jury could so conclude (at summary judgment
and JMOL).
Importantly, the Supreme Court has made clear that this but-for
showing does not require a plaintiff to demonstrate "sole" causation-or
to disprove that legitimate considerations may have played some role in
the defendant's decision. Rather, protected class status need only be the
"straw that broke the camel's back"-a factor that made a difference.1 46
Thus, if an African American employee was late to work five times-and
was terminated on that basis-but a white employee who was similarly
tardy would not have been, the but-for principle would be satisfied. 147 So,
too, if a government entity would not have enacted a policy "but for" the
race or sex of those affected (for example, if a Republican legislature
would not have enacted a voter ID law if it disproportionally impacted
whites), that would suffice, even if there was some legitimate basis for
the law. 148 As the Court observed in the Bostock decision in 2020, the but-
for principle is thus a "sweeping standard," which finds liability if
protected class status was "one but-for cause of [the] decision," even
where other legitimate considerations also played a role.1 49
It is easy to see the potential of this standard on an individualized basis.
Currently, most anti-discrimination claims are dismissed before trial,
typically applying myriad technical legal doctrines.' 5 0 But in many such
cases, a reasonable jury could conclude as a factual matter that the
adverse action or policy that the plaintiff complains of would not have
happened "but for" the protected class status of those affected."5 Indeed,
taking "the straw that broke the camel's back" approach seriously, there
14 Burrage v. United States, 571 U.S. 204,211(2014); see also Bostock, 140 S. Ct. at 1739-
40, 1744 (making clear that protected class status under the "but for" standard need not be the
sole, or even primary cause, so long as without it the defendant's action would not have
occurred).
141 See Bostock, 140 S. Ct. at 1739 (observing that an employer cannot avoid liability under
Title VII just by "citing some other factor that contributed to its challenged employment
decision" and that if protected class status was one "but for" cause, the employer has violated
the law).
141 Cf. Restatement (Third) of Torts: Liability for Physical & Emotional Harm, § 26 cmt. e.
(2010) (observing that the cause in fact inquiry demands a "counterfactual inquiry").
149 Bostock, 140 S. Ct. at 1739-40.
50 See Sperino & Thomas, supra note 3, at 151-55 (quoting Michael J. Zimmer, Slicing
&
Dicing of Individual Disparate Treatment Law, 61 La. L. Rev. 577, 591 (2001)); Berrey,
Nelson & Nielsen, supra note 25, at 63. There are also a substantial number of discrimination
claims that settle before trial (most commonly before summary judgment), but the amount of
those settlements tends to be very low. See Sperino & Thomas, supra note 3, at 16; Eyer, supra
note 9, at 1290-91.
151 For a few examples of cases where it seems highly likely that a reasonable jury could
have found for the plaintiffs as a factual matter, see Eyer, supra note 2, at 982-83.
2021] The But-For Theory ofAnti-Discrimination Law 1657
52 See, e.g., Crawford v. Runyon, 37 F.3d 1338, 1341 (8th Cir. 1994), rev'd in part by
Torgerson v. Rochester, 643 F.3d 1031, 1060 (8th Cir. 2011).
"I Cf. Reeves v. Sanderson Plumbing, 530 U.S. 133, 147-49, 151-54 (2000) (making clear
that a variety of forms of evidence can be considered in determining the ultimate factual
question of discrimination, including evidence that the employer has offered a false reason for
their employment decision).
1 The lower courts have a dizzying array of technical rules that they have crafted out of
common factual situations that can arise in the context of discrimination law, such as, for
example, the so-called "same actor" inference (that we presume the same manager who hired
a minority worker will not turn around and fire them shortly thereafter on a discriminatory
basis). These technical doctrines-which seek to enforce as a matter of law what inferences
must be drawn from particular pieces of the factual record-are inconsistent with a true factual
inquiry, and would not be properly applied under a factual "but for" approach. See, e.g., Eyer,
supra note 2, at 1011 n.270 and accompanying text. However, that does not mean that the
types of situations these doctrines represent may not be useful evidence for an employer to try
to include in their arguments.
165 8 Virginia Law Review [Vol. 107:1621
"I See, e.g., Sandra Sperino, Evidentiary Inequality, 101 B.U. L. Rev. (forthcoming Dec.
2021) (manuscript at 56-58, 66-67) (on file with author) (calling for the courts to apply the
normal rules of civil procedure and evidence in discrimination cases, instead of engrafting
specialized evidentiary rules onto discrimination cases).
156 Wallace v. Methodist Hosp. Sys., 271 F.3d 212 (5th Cir. 2001).
5 Id. at 215-16.
157
1Id. at 216.
1 Id. at 216-17.
160 Id. at 217-18.
161 Id. at 221.
162 Id. at 218.
163 See supra notes 138-43 and accompanying text.
2021] The But-For Theory ofAnti-Discrimination Law 1659
would not have been fired for her (unquestionably significant) offense had
she not also been pregnant three times in the last three years. This and
many other cases will ultimately be a judgment call for the fact-finder,
based on all of the facts and circumstances it was presented with, coupled
with the perceived credibility of the witnesses.1 64
And indeed, a jury did conclude that Wallace was terminated because
of her pregnancies.1 65 Yet the U.S. Court of Appeals for the Fifth Circuit
affirmed a grant of judgment as a matter of law, applying a host of
technical doctrines to "slice and dice" Wallace's evidence.1 66 The Court
first opined that Wallace's non-pregnant comparators were not close
enough because they had committed infractions that were not identical to
Wallace's own. 167 It then dismissed Wallace's explicit evidence of
pregnancy discrimination (such as the remark that she was fired because
"she's been pregnant three times in the last three years") under the stray
remarks doctrine. 168 Ultimately, having dismantled the plaintiffs
evidence via these technical doctrines, the court concluded that the jury's
verdict could not stand. 169
The Fifth Circuit's approach should be clearly impermissible if the but-
for principle controls. 17 0 The technical doctrines applied by the Fifth
171
Circuit-such as rejecting comparators as insufficiently close and
ignoring damning testimony under the stray remarks doctrine-simply
have no place in a straightforward factual inquiry. Rather, the question
for the court is simply whether a reasonable jury could have concluded
17 2
that Wallace would not have been terminated "but for" her pregnancies.
164 Id.
165 Wallace, 271 F.3d at 218.
166 Id. at 220-26.
167 Id. at 221-22.
161 Id. at 222-23.
169 Id. at 224.
170 See supra notes 135-43 and accompanying text.
171 It is important to emphasize that a but-for approach would provide a basis for pushing
back on-and would not reify or exacerbate-the current tendency of the courts to demand
close comparators in disparate treatment cases. See generally Goldberg, supra note 67, passim
(extensively describing the problematic comparator requirements that the lower courts have
imposed in the discrimination context). As the Supreme Court has made clear, the but-for
inquiry is a hypothetical counterfactual one. See, e.g., Comcast v. Nat'l Ass'n of Afr. Am.-
Owned Media, 140 S. Ct. 1009, 1014-15 (2020). As Wallace illustrates, any evidence bearing
on that counterfactual inquiry should be relevant, including many types of evidence that are
not comparator evidence of any kind. Thus, while identifying a close comparator may certainly
be useful for proving but-for causation, it should not be necessary.
172 See supra notes 135-43 and accompanying text.
1660 Virginia Law Review [Vol. 107:1621
And shorn of the Court's technical justifications, it seems clear that the
answer to that question is "yes." 173 Wallace thus provides a clear example
of how the but for principle could transform the approach-and
outcome-in many contemporary anti-discrimination cases.
And, while it may be hard for some to imagine anti-discrimination law
without all of the technical twists and turns, especially at summary
judgment, where most cases are currently dismissed, as then-Judge
Gorsuch observed, a simple focus on the factual question of causation is
actually far easier to apply. 7 4 Under the current approach, parties in an
anti-discrimination case can produce dozens of pages of briefmg at
summary judgment, addressing myriad nuances of specific technical
doctrines, all of which the court must address in order to resolve the
motion.' 7 5 But under a simple but-for approach, a court must simply
consider all of the facts and evidence before it, taken as a whole in the
light most favorable to the non-movant, and determine whether a
reasonable jury could find that the outcome would have been different
"but for" race, sex, or other protected class status. No technical tests or
specialized elements are required to answer this common-sense question.
And while no doubt judicial biases against discrimination claims would
continue to play a role in some adjudications, overall, such a system
would be far more likely to produce fair outcomes, eliminating the maze
of technical requirements that anti-discrimination law currently entails.
a complex legal puzzle, in which any piece out of place must result in the
dismissal of the plaintiffs' claims. 177 Prior precedent is treated as legally
dispositive of the unique factual circumstances of this case-despite the
fact that no prior case has adjudicated a case with these exact facts and
circumstances. 7 8 Evidence is "sliced and diced" without ever asking the
factual question of whether disparate treatment took place. 179 Cases are
dismissed based on collateral doctrines that have no relationship to the
question of whether discrimination took place.18 0
All of this is arguably illegitimate if the but-for principle controls. The
complex technical doctrines that the courts have engrafted onto the
disparate treatment inquiry-such as rigid comparator requirements, the
"same-actor" inference, the "stray remarks doctrine," and, indeed, the
McDonnell Douglas paradigm itself-all are impermissible distractions
from the factual question of discrimination if the but-for principle
controls.'81 Demands that the plaintiff shoehorn her proof into particular
categories-and the courts' propensity to address each of those categories
as if they were separate elements, rather than a unified whole-is also
inconsistent with a factual but-for approach.' 8 2 Searches for the particular
role of specific actors in producing disparate treatment-or for concrete
evidence of self-aware intent-provide no justification for dismissing
claims.' 83 Even the operation of independent doctrines-like standing or
the immunity of government officials-to bar discrimination claims can
at least be challenged more effectively when set against a backdrop of the
84
presumption that all "but for" discrimination ought to be proscribed.1
177 Id.
178 See Eyer, supra note 2, at 980-81, 1011 n. 270.
179 See generally Sperino & Thomas, supra note 3, at 152 (describing this phenomenon
extensively and using the phrase "sliced and diced").
180 See supra notes 69-71 and accompanying text.
181 See, e.g., Eyer, supra note 2, at 978-84; see also Widiss, supra note 68, at 29-34
(describing the ways that the McDonnellDouglas paradigm is often applied in conflict with
the principles behind but-for causation).
182 See, e.g., Eyer, supra note 2, at 1008-09.
183 See generally McCleskey v. Kemp, 481 U.S. 279, 293-99 (1987) (rejecting Equal
Protection claim based on a statistical showing that Black defendants were more likely to
receive the death penalty when they killed white victims, based in part on the difficulties of
identifying individual discriminatory actors, and a lack of specific evidence of discriminatory
intent).
184 See generally City of Los Angeles v. Lyons, 461 U.S. 95, 105-10 (1983) (holding that
plaintiff lacked standing to seek injunctive relief from illegal police chokehold); Imbler v.
Pachtman, 424 U.S. 409,427-28 (1976) (holding that prosecutors are absolutely immune from
civil liability under § 1983 in the conduct of their duties).
1662 VirginiaLaw Review [Vol. 107:1621
187 Comcast v. Nat'l Ass'n of Afr. Am.-Owned Media, 140 S. Ct. 1009, 1014, 1018 (2020);
there are reasons to believe that each of them might be receptive to cases
raising such arguments.
There are thus reasons to believe that but-for-centered arguments for
the reform of anti-discrimination law could succeed in future cases at the
Supreme Court, just as they did in Bostock. But it is important to note that
innumerable opportunities exist to center the but-for principle right now
in the lower courts, regardless of its further endorsement by the Supreme
Court. As described in Part II, supra, the Court has already endorsed the
but-for principle as the central textual mandate of anti-discrimination
law.1 94 And some circuits too have already independently adopted an
approach to anti-discrimination law that follows an analogous
approach. 195 Thus, advocacy within the lower courts-where the lion's
share of discrimination cases will be resolved-is both possible and
important. Indeed, pushing for the broad embrace of the but-for principle
in the lower courts-and its simple, factually focused inquiry-is an
important part of laying the groundwork for more substantial challenges,
at whatever level those challenges might occur.
A. Stereotyping Jurisprudence
As anti-discrimination scholars and advocates have long recognized,
1 96
stereotypes lie at the heart of many, if not most, discriminatory acts. As
such, scholars and advocates have long contended that anti-stereotyping
principles ought to play a central role in anti-discrimination law.1 97 And
in a seemingly substantial victory, a majority of the Justices in 1989
endorsed the perspective that gender stereotyping is impermissible under
Title VII in the case of Price Waterhouse v. Hopkins, albeit in a splintered
set of opinions.1 98
But while Price Waterhouse led to certain significant successes for
plaintiffs (including most notably, a wave of lower court cases
recognizing anti-transgender discrimination as sex discrimination),
stereotyping doctrine in general has not been as effective as advocates had
hoped.1 99 Cases addressing racial stereotyping, or other forms of non-sex-
200
based stereotyping have continued to see little success. And even
gender stereotyping has long been dogged by critiques that it lacks any
20! See, e.g., Hamm v. Weyauwega Milk Prods., 332 F.3d 1058, 1066-67 (7th Cir. 2003)
(Posner, J., concurring).
202 See, e.g., Brief for Petitioner Altitude Express at 40-43, Bostock v. Clayton Cnty., 140
S. Ct. 1731 (2020) (No. 17-1623); Brief of Institute for Faith & Family and Christian Family
Coalition as Amici Curiae at 11-14, Bostock, 140 S. Ct. 1731 (No. 17-1623) (criticizing lower
courts that "have morphed stereotyping into a separate species of sex discrimination and a
rationale for judicially amending Title VII.").
203 Bostock, 140 S. Ct. at 1748-49.
204 Id. Importantly, as Suzanne Goldberg has observed, courts do not require comparators
for this inquiry in the stereotyping context. See Goldberg, supra note 67, at 784; see also
Bostock, 140 S. Ct. at 1741-43, 1748-49 (recognizing that the but-for principle can apply
based on background knowledge of the differential stereotypes applied to men and women,
without requiring real-world comparators).
205 See infra notes 217-23 and accompanying text.
206 See infra notes 211-23 and accompanying text.
207 See Bostock, 140 S. Ct. at 1741-43, 1748.
2021] The But-For Theory ofAnti-Discrimination Law 1667
would not have been applied, "but for" protected class status. 208 This is
true both of adverse actions resulting from prescriptive stereotypes
(stereotypes about how a particular group should act or be), and
descriptive stereotypes (stereotypes about how a particular group does act
or is). 209 Thus, the operation of racial, religious, national origin, disability,
age, sex, and other stereotypes to effectuate disparate treatment against
individuals will, as a matter of course, violate the but-for principle. 2 10
While this does not mean that the mere existence of stereotyping
evidence is sufficient to show but-for discrimination in all instances, in
many circumstances, the existence of stereotype-based disparate
treatment may be easy to infer. For example, in cases involving family
responsibilities discrimination against women or men, it is often easy to
identify the ways that gender-specific stereotypes lead to "but for"
discrimination. 2 11 In the case of a woman who is denied a promotion
because she is a parent of small children, and thus stereotyped as being
uncommitted to the job, it takes no special evidence or expertise to
recognize that the same stereotype would not have been applied to a man
(indeed, a man might be assumed to need the extra income for his growing
family). 21 2 Conversely, in the context of a man who is penalized
especially harshly at work for taking time off to care for a newborn
(because of the stereotype that men are not supposed to engage in
caregiving), a simple "but for" analysis will also often lead to the
208 See infra note 210; see also Bornstein, supra note 197, at 962-74 (describing the ways
that stereotyping theory can be used to craft disparate treatment claims across the protected
classes).
209 See Bornstein, supra note 197, at 962-63 (describing the distinction between prescriptive
and descriptive stereotypes).
210 See, e.g., Bostock, 140 S. Ct. at 1741-43, 1748-49 (characterizing stereotyping as an
application of the "but for" principle, and linking the "but for" principle to Title VII's "because
of' language). As the courts have recognized, the but-for principle also applies to most other
major federal anti-discrimination laws, since they tend to have identical or cognate language
as a part of their core proscriptions. See, e.g., Gross v. FBL Fin. Servs., 557 U.S. 167, 175-78
(2009) (private sector age claims); Univ. of Tex. Sw. Med. Ctr. v. Nassar, 570 U.S. 338, 352
(2013) (Title VII retaliation provisions); Serwatka v. Rockwell Automation, Inc., 591 F.3d
957, 962 (7th Cir. 2010) (ADA).
211 See, e.g., Joan C. Williams & Stephanie Bornstein, Caregivers in the Courtroom: The
Growing Trend of Family Responsibilities Discrimination, 41 Univ. S.F. L. Rev. 171, 174-81
(2006).
212 See, e.g., Chadwick v. Wellpoint, 561 F.3d 38, 46-47, 48 n.12 (1st Cir. 2009).
1668 Virginia Law Review [Vol. 107:1621
213 See, e.g., Catherine Albiston & Lindsey Trimble O'Connor, Just Leave, 39 Harv. J.L.
&
determination that a facially neutral policy would not have been adopted "but for" the
protected class status of those affected.
218 See, e.g., Ricci v. DeStefano, 557 U.S. 557, 579-80 (2009) (holding that it was disparate
treatment for City to refuse to certify test results because of how the test would affect minority
candidates as a group).
219 Id.
220 See, e.g., United States v. Clary, 846 F. Supp. 768, 783-87 (E.D. Mo. 1994), rev'd 34
F.3d 709 (8th Cir. 1994).
221 Id. See also Dvorak, supra note 10 (making the point that we have treated addiction crises
affecting the Black and white communities very differently); Harden v. Hillman, 993 F.3d
465, 482-85 (6th Cir. 2021) (describing the pernicious racial stereotypes that have long
surrounded cocaine use, and specifically crack-cocaine).
222 EEOC v. Catastrophe Mgmt. Sols., 852 F.3d 1018, 1021-22 (11th Cir. 2016) (employer
policy requiring "professional" hairstyles, interpreted by the employer as prohibiting
dreadlocks).
223 Id.
1670 VirginiaLaw Review [Vol. 107:1621
situated within the but-for principle. This is important, since the most
stubborn obstacles that stereotyping jurisprudence has faced-the
unwillingness to extend it to other groups (beyond sex), and critiques of
its lack of textual grounding-are evidently erroneous when viewed
through the lens of "but for." With only minor exceptions, the but-for
principle extends to all major protected groups under federal anti-
discrimination law.224 Thus, to the extent that stereotyping jurisprudence
is founded in the but-for principle, it must extend to all such groups as
well. And because the but-for principle arises from anti-discrimination
law's central text, it upends traditional critiques of stereotyping
jurisprudence as textually ungrounded and instead suggests that anti-
stereotyping principles arise from the core of anti-discrimination law's
statutory text. 2 5
B. Negligent Discrimination
So-called "negligent discrimination" has also been a long-standing
project of anti-discrimination law scholars. Building on the insights of
Charles Lawrence III, Professor David Oppenheimer in 1993 developed
for the first time a theory of "negligent discrimination." 2 2 6 As Professor
Oppenheimer noted, much discrimination does not in fact take place as a
result of conscious intent to discriminate. 22 7 Assuming that the Court's
disparate treatment opinions generally required such self-aware intent, he
further argued that the Court ought to recognize so-called "negligent
discrimination"-discrimination where the "employer fails to take all
reasonable steps to prevent discrimination that it knows or should know
is occurring, or that it expects or should expect to occur." 22 8 Over the last
three decades, scholars have continued to build on Professor
Oppenheimer's work in calling for anti-discrimination law to recognize a
variety of forms of "negligent" or "reckless" discrimination. 229
As Professor Oppenheimer and those who have built on his work have
properly recognized, "intentional discrimination" and "disparate
treatment" are not co-extensive. 230 History, experience, and the social
sciences all tell us that people can and do engage in "disparate treatment"
(in the sense that their actions would have been different "but for" the
race, sex or other protected class status of those affected), without a self-
aware intent to treat others differently based on protected class status.23 1
Indeed, as described in Part I, supra, it is precisely the conflation of these
two non-commensurate concepts-"intentional discrimination" and
"disparate treatment"-that has led to the conceptual crisis in anti-
discrimination law. 232 Professor Oppenheimer (and those who have built
on his work) were no doubt correct that excluding all cases in which the
actor does not possess a self-aware intent to discriminate would lead to
much disparate treatment going unremedied, and would be a crisis for
anti-discrimination law.
But as should be evident from the discussion in Parts I-III, supra,
"negligent discrimination" is in many ways a pale substitute for the but-
for principle. It simply is not necessary as a theory of disparate treatment
if the but-for principle controls. 233 Under the but-for principle, an entity
will be liable wherever the harm would not have occurred "but for" the
protected class status of the individual harmed. 234 One need not layer on
a complicated inquiry into whether the entity "knew or should have
known" that such discrimination would occur-and indeed doing so
arguably engrafts additional contra-textual obstacles to plaintiffs'
230 See, e.g., Oppenheimer, supra note 60, at 900-04; Bornstein, supra note 229, at 1059-
60.
23 See supra Part I.
232 Id.
233 It is important to note that there may still be more narrow instances in which negligence
or recklessness principles are important in anti-discrimination law, especially in the context
of employer responses to harassment, and employer liability for certain types of damages. But
the but-for principle makes clear that the heartland of what many scholars have sought to
address with negligent discrimination arguments-instances of disparate treatment that may
be effectuated without conscious intent-does not require a negligence paradigm, and that
indeed such a paradigm is likely to impose additional unnecessary burdens on plaintiffs as
compared to a "but for" approach.
234 See supra Part II; see also 42 U.S.C. § 2000e(b) (defining "employer" to include agents).
Note that the constitutional law context is different insofar as there are circumstances in which
sovereign immunity or § 1983 doctrines bar holding the entity itself liable. See, e.g., Will v.
Mich. Dep't of State Police, 491 U.S. 58, 80-81 (1989) (discussing the holding in United
States v. Fox, 94 U.S. 315, 318 (1877) that neither the state nor state officials acting in their
official capacity are "person[s]" within the meaning of § 1983). However, those restrictions
exist regardless of what the underlying standard is for assessing discrimination in the first
instance, the subject of negligent discrimination and the but-for principle.
1672 Virginia Law Review [Vol. 107:1621
relief.235 Thus, the very set of cases that Oppenheimer and other scholars
have meant to target-cases where disparate treatment has occurred, but
where the defendant did not act with self-aware discriminatory intent-
fall squarely within the but-for principle without any additional inquiry. 2 36
The but-for principle is thus not only an adequate but a superior theory to
"negligent discrimination" for instituting a true disparate treatment
standard. It provides a much simpler and more straightforward means to
the same end.
But what about cases like Bostock, which continue to suggest that a
showing of some form of intent is still required in anti-discrimination
law? As an initial matter, while cases like Bostock, Gross, and Nassar
situate the but-for principle as arising from the "ordinary meaning" of
anti-discrimination law's core language, they have not so characterized
intent.237 Thus, textualist principles arguably lean in favor of an
exclusively but-for focused standard. And while many contemporary anti-
discrimination cases do situate intent or purpose as a requirement, it is
clear that such cases have simply conflated disparate treatment and
"intent," rather than affirmatively selecting one or the other standard. 238
Therefore, one ultimate goal of a but-for-centered project can and should
be to challenge existing language in Supreme Court case law situating
intent as a necessary, distinct requirement.
But even assuming that in the near-term some nominal intent
requirement may be attached to the but-for principle, cases like Bostock
make clear that that requirement does not entail a showing of self-aware
conscious discriminatory intent. 2 39 Indeed, as described in Part II, supra,
Bostock makes clear that all that is required is an intention to take
differential adverse action against an individual or group (discrimination
in the literal sense)-actions that would, in turn, not have been taken "but
for" the protected class status of the individual or groups affected. 24 As
235 Cf. Oppenheimer, supra note 60, at 900 (employers would be liable for negligent
discrimination when they "fail[] to take all reasonable steps to prevent discrimination that
[they] know[] or should know is occurring, or that [they] expect[] or should expect to occur").
236 See supra Parts I-1I.
237 See, e.g., Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1739-40 (2020) (emphasis added)
(describing the but-for principle as being mandated by Title VII's text, but as to intent stating
only that "[i]n so-called 'disparate treatment' cases . . . this Court has also held that the
difference in treatment . . . must be intentional.").
238 See supra Part I.
239 See Bostock, 140 S. Ct. at 1741-43, 1745-46.
240 Id.
2021 ] The But-For Theory of Anti-Discrimination Law 1673
C. DisparateImpact
Since the 1970s, the disparate impact paradigm-which asks whether
a neutral policy or practice has a disparate impact on a protected group-
has captured the imagination of many anti-discrimination law scholars
and advocates. 24 3 Situating the paradigm as a preferred alternative to
disparate treatment law, such scholars and advocates have contended that
a disparate impact approach best effectuates anti-subordination
objectives-objectives which ought to be at the core of our anti-
discrimination inquiry. 244 Thus, they have argued that a disparate impact
v. Feeney, 442 U.S. 256, 266, 278 (1979). Note that importantly, a "but for" argument is
different than the "foreseeability" argument rejected in Feeney. See id. at 278.
256 See, e.g., United States v. Morrison, 529 U.S. 598, 630-34 (2000) (Souter, J., dissenting);
Caitlin Dickerson, Seth Freed Wessler & Miriam Jordan, Immigrants Say They Were
Pressured into Unneeded Surgeries, N.Y. Times (Sept. 29, 2020),
https://www.nytimes.com/2020/09/29/us/ice-hysterectomies-surgeries-georgia.html
[https://perma.cc/MY2Y-VQGC].
257 Washington v. Davis, 426 U.S. 229, 237-48 (1976).
258 See, e.g., Alexander v. Sandoval, 532 U.S. 275, 280-81, 293 (2001) (noting that Title VI
only prohibits disparate treatment, and finding no implied private cause of action to enforce
Title VI regulations prohibiting disparate impact discrimination).
259 See, e.g., Selmi, supra note 74, at 704-06; see also Eyer, supra note 9, at 1300-01
(describing results from studies showing that many people do not perceive classic disparate
impact contexts, such as disparate impacts on a protected group arising from testing-based
disparities, as discrimination).
260 See Int'l Bhd. of Teamsters v. United States, 431 U.S. 324, 335-36 n.15 (1977).
261 See, e.g., Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1736-37 (2020); Comcast Corp.
v.
Nat'l Ass'n of Afr. Am.-Owned Media, 140 S. Ct. 1009, 1012 (2020); Univ. of Tex. Sw. Med.
Ctr. v. Nassar, 570 U.S. 338, 340 (2013); Gross v. FBL Fin., 557 U.S. 167, 169 (2009).
2021] The But-For Theory ofAnti-Discrimination Law 1677
262 See Bostock, 140 S. Ct. at 1739; Comcast, 140 S. Ct. at 1014-15; Nassar, 570 U.S. at
350-52; Gross, 557 U.S. at 176-77.
263 See Sullivan, supra note 74, at 358-59 (describing the extensive attention paid to
"motivating factor" liability in the scholarly literature and observing that "one wonders what
all the fuss is about" in view of the disappointingly small impact that "motivating factor" has
had).
2" See 42 U.S.C. §§ 2000e-2(m), 2000e-5(g)(2)(B); Price Waterhouse v. Hopkins, 490 U.S.
228, 258 (1989) (plurality opinion).
265 See sources cited supra note 264.
1678 Virginia Law Review [Vol. 107:1621
.
a majority group employee. 273
Importantly, however, the claims of anti-discrimination scholars and
advocates-that the but-for standard requires a showing that protected
274
class status was the exclusive consideration-have been wrong. The
but-for standard does not require a showing that protected class status was
the sole cause, or that no other factors played a role. 275 Thus, the employee
who committed an infraction-but would not have been fired were they
outside the protected class-prevails under the but-for standard. 276 So too,
government actions that may have some legitimate underlying
justification-but would not have been taken "but for" the protected class
status of those affected-will result in defendant liability under the but-
for approach. 277 As long-standing articulations of the but-for principle (in,
for example, tort law) make evident, it is simply erroneous to contend
otherwise. 278
Moreover, if there were any doubt on this front, the Supreme Court's
recent decision in Bostock v. Clayton County eliminates it. 279 As the Court
holds in Bostock, under the but-for standard "a defendant cannot avoid
liability just by citing some other factor that contributed to its challenged
employment decision. So long as the plaintiff's [protected class status]
was one but-for cause of that decision, [it] is enough to trigger the law."
280 As the Court properly observes, "[t]his can be a sweeping standard,"
since "[o]ften, events have multiple but-for causes." 2 8 1 As such, under the
but-for principle, protected class status need only be one factor among
many-as the Court put it in a prior decision, the "straw that broke the
camel's back." 2 82
At this point, it should be evident that, understood properly, the but-for
principle can amply accommodate cases in which there are "mixed
motives"-or more clearly stated, multiple causal factors. The mere fact
that an employer may have been partially spurred to action by legitimate
misconduct, or that a government entity may have responded in part to
legitimate public concerns, does not answer the question of whether the
employer or government entity has violated the but-for principle. Indeed,
there are likely to be many cases in which some legitimate considerations
which makes absolutely clear that the "but for" standard is not a sole cause standard-should
lead to a reversal of these precedents. Id. at 27-30.
275 See, e.g., Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1739 (2020).
276 Id.; see also McDonald v. Santa Fe Trail Transp. Co., 427 U.S. 273, 283-84 (1976).
277 See, e.g., Burrage v. United States, 571 U.S. 204, 211 (2014).
278 See, e.g., Restatement (Third) of Torts § 26 cmt. c (Am. L. Inst. 2010).
279 See Bostock, 140 S. Ct. at 1739.
280 Id.
281 Id.
2 82
Burrage, 571 U.S. at 211.
2021 ] The But-For Theory ofAnti-DiscriminationLaw 1681
The foregoing Parts have made the case that there is a long-standing
theoretical crisis in anti-discrimination law-arising from the conflation
of "disparate treatment" and "discriminatory intent"-and that now is the
time to address it. As laid out in Parts II-IV, the Supreme Court's recent
cases-centering the but-for principle as the central defining feature of
anti-discrimination law-provide considerable potential to do so. But it is
important to acknowledge that there remain significant potential
headwinds to such a project-including the entrenched views of many of
those most committed in theory to expanding anti-discrimination
protections. This Part takes up potential obstacles to resolving the
theoretical crisis in anti-discrimination law, addressing in turn, likely
progressive headwinds to a but-for project, potential legal headwinds, and
likely sites of judicial resistance to refocusing anti-discrimination law
around the but-for principle.
A. ProgressiveHeadwinds
Without the buy-in of those committed to strengthening the protections
of anti-discrimination law-anti-discrimination advocates and scholars-
it is clear that a but-for-centered approach to revitalizing anti-
discrimination law will not succeed. Indeed, while the growth of the but-
for principle in anti-discrimination law may be inexorable (as evidenced
by progressives' failure to stop its spread), its effects and application
within anti-discrimination law will not be. If anti-discrimination scholars
1682 Virginia Law Review [Vol. 107:1621
1. Opposition to Textualism
One of the most obvious potential obstacles to progressive embrace of
the but-for principle is the Supreme Court's grounding of the but-for
principle in textualism. 281 What is a strength of the but-for principle in its
ability to provide effective arguments in the courts, becomes a weakness
283 See, e.g., Pelcha v. MW Bancorp, Inc., 998 F.3d 318, 324 (6th Cir. 2021).
284 Others have also written about progressive concerns about the but-for principle and
possible responses to them, see Jessica Clarke, The Virtues of Formal Causation 54 (Feb. 13,
2021) (unpublished manuscript) (on file with author).
285 See generally supra Part H (describing the Supreme Court's treatment of the "but for"
principle as arising from the plain text of anti-discrimination law).
2021 ] The But-For Theory ofAnti-Discrimination Law 1683
286 See, e.g., Doug Kendall & Jim Ryan, The Case for New Textualism, Democracy (2011),
https://democracyjournal.org/magazine/2 1/the-case-for-new-textualism/
[https://perma.cc/LA7V-3U65] (making this observation and arguing against this tendency).
Note that there is a sizeable and growing group of progressives who disagree with this
perspective, including this author. See, e.g., Katie Eyer, Progressive Textualism (articulating
a normative theory of progressive textualism) (unpublished manuscript).
287 See id.; see also, e.g., Scott Fruehwald, Pragmatic Textualism and the Limits of Statutory
Interpretation: Dale v. Boy Scouts of America, 35 Wake Forest L. Rev. 973, 991-92 (2000)
(collecting critiques of textualism); Cary Franklin, Living Textualism, 2020 S. Ct. Rev.
(forthcoming 2021) (manuscript at 4-8) (on file with author) (critiquing textualism in the
context of a critique of Bostock).
288 See, e.g., Brief for Employment Law Professors as Amici Curiae at 5-13, Univ. of Tex.
Sw. Med. Ctr. v. Nassar, 570 U.S. 338 (2013) (No. 12-484); Brief for Employment Law
Professors as Amici Curiae at 4-10, Comcast v. Nat'l Ass'n of Afr. Am.-Owned Media, 140
S. Ct. 2561 (2020) (No. 18-1171). In the interest of full disclosure, this author joined the
aforementioned brief in Nassar, although my views on the proper textual interpretation of the
statutory language under review there have shifted since that time.
289 See Brief for Employment Law Professors as Amici Curiae at 6, Nassar, 570 U.S. 338
(No. 12-484).
1684 VirginiaLaw Review [Vol. 107:1621
290 See, e.g., Franklin, supra note 287, at 4-8; Mitchell N. Berman
& Guha Krishnamurthi,
Bostock was Bogus: Textualism, Pluralism, and Title VII 97 Notre Dame L. Rev. (forthcoming
2022) (manuscript at 28-30) (on file with author); Anuj Desai, Text Is Not Enough, 93 U.
Colo. L. Rev. (forthcoming 2021) (manuscript at 2-4) (on file with author); In Lieu of Fun,
supra note 101, https://youtu.be/zbxMG9Y9eB4?t-2622 [https://perma.cc/RBF5-YQ9M].
But cf. Eyer, supra note 101, at 73-80 (arguing before Bostock that textualism compelled the
result the Court ultimately reached); Grove, supra note 100, at 267-69 (2020) (suggesting that
Bostock was a legitimate application of textualism, but helps demonstrate that there are two
forms of textualism, one of which, "formalistic textualism," is much more likely to lead to
determinate results); Andrew Koppelman, Bostock, LGBT Discrimination, and the Subtractive
Moves, 105 Minn. L. Rev. Headnotes 1, 8-38 (2020) (describing Bostock as clearly correct
on the text, and describing why the "subtractive moves" made by the dissenters and some
lower court judges in similar contexts are contra-textual). Of course, several conservatives
also critiqued Bostock as an application of textualism. In addition to the dissenters, see, e.g.,
Josh Blackman & Randy Barnett, Justice Gorsuch's Halfway Textualism Surprises and
Disappoints in the Title VII Cases, Nat'l Rev. (June 26, 2020),
https://www.nationalreview.com/2020/06/justice-gorsuch-title-vii-cases-half-way-
textualism-surprises-disappoints/ [https://perma.cc/P5K9-XSZ7].
291 See, e.g., Cass Sunstein, Gorsuch Paves Way for Attack on Affirmative Action,
Bloomberg (June 17, 2020), https://www.bloomberg.com/opinion/articles/2020-06-
17/gorsuch-gay-rights-opinion-targets-affirmative-action [https://perma.cc/PH4Z-N6WN].
292 As I am developing more fully in other forthcoming work, I believe there are important
reasons to embrace textualism as a normative matter. See Eyer, supra note 286. But one need
not embrace this normative perspective in order to see the strategic utility of textualism as a
legal methodology.
293 See supra Part III.
2021] The But-For Theory ofAnti-Discrimination Law 1685
300 See Richmond v. J.A. Croson, 488 U.S. 469, 505 (1989).
301 See United Steelworkers of Am. v. Weber, 443 U.S. 193, 201 (1979).
302 See, e.g., Stacy Hawkins, How Diversity Can Redeem the McDonnellDouglas Standard,
83 Fordham L. Rev. 2457, 2467-69 (2015).
303 Unlike programs like affirmative action in which an identifiable person may be treated
differently based on their race, in the context of race-intentional policies, no specifically
identifiable individuals are treated differently. Thus, for example, while the decision to site a
school in a geographic location that will produce the maximum amount of racial integration
may have a partial racial motivation, it does not treat individuals-or groups-differently
based on race. Cf. Parents Involved in Cmty. Schs. v. Seattle Sch. Dist. No. 1, 551 U.S. 701,
789 (2007) (Kennedy, J., concurring) (noting that "[s]chool boards may pursue the goal of
bringing together students of diverse backgrounds and races through other means, including
strategic site selection of new schools; drawing attendance zones with general recognition of
the demographics of neighborhoods; allocating resources for special programs; recruiting
students and faculty in a targeted fashion; and tracking enrollments, performance, and other
statistics by race. These mechanisms are race conscious but do not lead to different treatment
based on a classification that tells each student he or she is to be defined by race, so it is
unlikely any of them would demand strict scrutiny to be found permissible.").
2021 ] The But-For Theory of Anti-DiscriminationLaw 1687
304 Over the last several decades, impact litigation organizations have repeatedly asked the
Supreme Court to take up this issue and squarely address it, and it has declined to do so. See
Katie Eyer, Constitutional Colorblindness and the Family, 162 U. Pa. L. Rev. 537, 600 n.292
(2014) (collecting petitions for certiorari review on this issue). Even in cases where the
question arguably was squarely presented, a majority of the Court has shown itself to be
uninterested in taking it up. See, e.g., Ricci v. DeStefano, 557 U.S. 557, 594 (2009) (Scalia,
J., concurring) (in an opinion not joined by any other member of the Court, arguing that the
Court should take up the issue of whether disparate impact doctrine-a form of race-
intentional action-is unconstitutional under the Equal Protection Clause).
301 See Kim Forde-Mazrui, The Constitutional Implications of Race-Neutral Affirmative
Action, 88 Geo. L.J. 2331, 2333-34 (2000).
306 Id.
307It is important to note that this is only true for-and I am only focused herein on-
programs that also do not explicitly racially distinguish at the individual level. (So, in the
example given above, the assumption is that the program makes its services available to all,
even though its objective may be to remediate racial disparities). Where programs explicitly
racially distinguish among individuals in eligibility criteria, they would fall under the logic of
the Supreme Court's affirmative action opinions instead.
1688 VirginiaLaw Review [Vol. 107:1621
310 See supra Section IV.D; see also Hillel J. Bavli, Causation in Civil Rights Legislation,
73 Ala. L. Rev. (forthcoming 2021) (manuscript at 3-5) (on file with author) (arguing against
the but-for causation standard, and suggesting that the "NESS" approach to causation be
adopted instead in anti-discrimination law).
311 See sources cited supra notes 90, 96-105, 124 and accompanying text.
312Comcast Corp. v. Nat'l Ass'n of Afr. Am.-Owned Media, 140 S. Ct. 1009, 1011-12
(2020).
31 Id. at 1019 n.* (Ginsburg, J., concurring in part).
314 See infra notes 346-50 and accompanying text. But cf. Bostock v. Clayton Cnty., 140 S.
Ct. 1731, 1739-40 (2020) (recognizing that this part of the Civil Rights Act of 1991 was
simply intended to impose a lower standard of causation).
1690 Virginia Law Review [Vol. 107:1621
But even if this concern did not exist, it is far from clear that
"motivating factor" burden-shifting is indeed superior to "but for." As
discussed supra, Charles Sullivan has shown that the "motivating factor"
provision has not had the on-the-ground effects that advocates had
hoped. 3 15 "Motivating factor" cases do not succeed at materially higher
rates, and the lion's share of anti-discrimination claims continue to be
raised under other paradigms (most notably the McDonnell Douglas
paradigm, which the Supreme Court has made clear is associated with
"but for"). 316 As Professor Sullivan details, advocates have been reluctant
to bring motivating factor claims because of concerns that juries will
"spli[t] the baby" (finding for the plaintiff on "motivating factor" but for
the defendant on "but for")-something that would leave the plaintiff
effectively without relief. 31 And judges have struggled to understand or
embrace the concept of what it might mean to find liability based on
something short of but-for causation. 3 18
Moreover, it is also important to note that there is far less daylight
between the "motivating factor" approach and the but-for approach than
progressives have traditionally suggested. As described in Part IV, supra,
progressives have at times suggested that the but-for principle demands
something akin to "sole" causation-perhaps to make their case for why
"motivating factor" is so urgent. 3 19 But this is false. Just like under the
motivating factor approach, under "but for," protected class status need
only be one consideration of many. As the Supreme Court put it, "a
defendant cannot avoid liability just by citing some other factor that
contributed to its challenged employment decision. So long as the
plaintiff's [protected class status] was one but-for cause of that decision,
that is enough to trigger the law."3 20
In short, while progressives may once have had good reasons for
thinking it was important to fight the turn to "but for," those reasons have
passed. "Motivating factor" has proved to be both less powerful, and more
risky than its proponents initially envisioned. And the turn to "but for" is
here to stay. Thus, today, there is little to be gained-and much to be
lost-by declining to embrace the potential of the but-for approach.
B. Legal Headwinds
Even if anti-discrimination scholars and advocates wholly embrace the
but-for principle, there will of course remain legal obstacles to using the
principle to effectuate the project described herein. As described in Part
I, supra, there is ample case law that uses the language of "intent" or
"purpose" and thus could provide a basis for arguments that a strong form
of intent is indeed a legal requirement. And the process of using the but-
for principle to attempt to dismantle the network of technical doctrinal
rules that have been engrafted onto the anti-discrimination law inquiry
will surely not be easy. Moreover, some anti-discrimination statutes,
including Title VII, also include language that could complicate the
argument that "but for" is the central defining inquiry in anti-
discrimination law.
This Section takes up those various legal obstacles and suggests that-
while they are real-they are not different in kind from those that any
major project of anti-discrimination law reform would face. Moreover,
because the project of resolving the theoretical crisis in anti-
discrimination law will necessarily be an iterative and multi-sited one, it
should not be viewed as an all or nothing affair. Rather, there are likely
to be many small victories-and many small losses-as the civil rights
community attempts to resituate anti-discrimination law around a true
disparate treatment standard, i.e., the but-for principle. It is important to
recognize that just as the conceptual crisis in anti-discrimination law took
many years to create, so too it is likely to take years to resolve. Current
cases offer the opportunity to begin that movement-and to secure
incremental victories for anti-discrimination litigants-but they will
require consistent and strategic action to produce their most significant
results.
1. Case Law
There are two fronts on which case law may stand as an obstacle to the
project described herein: (1) case law may be used to argue that the but-
for principle is not in fact the central defining principle of anti-
discrimination law (and that, for example, a strong form of self-aware
intent requirement exists); and (2) case law may be used to object to the
application of the but-for principle, suggesting that precedent demands a
different, more technical approach. As set out below, this second category
of case law should not be viewed as a reason not to pursue the but-for
1692 VirginiaLaw Review [Vol. 107:1621
"
As described in Part I, supra, the conceptual crisis in anti-
discrimination law has arisen from the conflation of "disparate treatment"
and "intentional discrimination" in the Supreme Court's doctrine. 321 As
such, there are ample cases that use the language of "discriminatory
intent" or "purpose"-even as they simultaneously situate disparate
treatment as the standard. 322 These cases no doubt will provide an
opportunity for those who might oppose a but-for-centered disparate
treatment law to argue the resolution of anti-discrimination law's
conceptual crisis suggested herein is foreclosed.
But it is important to note that any effort to resolve the conceptual crisis
in anti-discrimination law would run up against this difficulty-even one
that sought to resolve the conceptual crisis in favor of an "intent" or
"purpose" standard. As a variety of scholars have observed, there are
significant parts of the Supreme Court's case law that are also
irreconcilable with a true "intent" or "purpose" standard. 32 3 Precisely
because the doctrine has been conceptually confused-conflating
"disparate treatment" with "discriminatory intent"-the full body of what
the Supreme Court has said and done cannot be fully reconciled with
either a pure "disparate treatment" standard or a pure "intent"
approach.32 4
But as described in Part II, supra, recent Supreme Court case law has
tilted toward situating a true "disparate treatment" standard-the but-for
victory. 326 But as the Supreme Court's decision in Bostock makes clear,
even shy of achieving its ultimate goals, a "but for" movement may
produce important reforms (big and small) for anti-discrimination
litigants.
326It is perhaps too easy to forget that the campaign for marriage equality spanned five
decades and involved innumerable losses before the 2015 decision in Obergefell v. Hodges,
576 U.S. 644 (2015). Ultimately, those losses were all a part of the important iterative
transformation of society that allowed Obergefellto occur. See, e.g., Anthony Michael Kreis,
Marriage Demosprudence, 2016 U. Ill. L. Rev. 1679, 1709-11 (2016).
327 See supra note 77 and accompanying text.
321 See supra notes 61-63 and accompanying text.
329 See supra note 118 and accompanying text.
2021 ] The But-For Theory ofAnti-DiscriminationLaw 1695
2. Statutory Language
As described above, one of the principal strengths of using the but-for
principle as the vehicle for resolving the conceptual crisis in anti-
discrimination law is its textual grounding. As the Supreme Court has
held, the plain meaning of the causation language that appears at the heart
of most anti-discrimination laws-"because of' or "by reason of' or "on
the basis of'-is the but-for principle. 334 But there is also other language
in some anti-discrimination law statutes which may complicate arguments
that the but-for principle rests at the center of anti-discrimination law, or
that its principles should extend to all anti-discrimination law claims. 33 5
As set out below, while such language ought to be addressed by advocates
and scholars with care, it need not and should not be read to interfere with
the idea that the but-for principle is anti-discrimination law's central
defining inquiry. While the language of the anti-discrimination laws
varies, this Subsection focuses on Title VII of the Civil Rights Act of
1964, as its wording forms the basis for many of the other contemporary
anti-discrimination laws. 336
Title VII's statutory text prohibits discrimination "because of. . . race,
color, religion, sex, or national origin." 337 Thus, as the Supreme Court
recognized in Bostock, it falls squarely within the Court's holdings that
"the ordinary meaning of 'because of is 'by reason of or 'on account
of'"-and that those terms further connote "the 'simple' and 'traditional'
standard of but-for causation." 338 As such, Title VII can be situated as a
key example of the central textual significance of the but-for principle to
anti-discrimination law. 339 And Title VII plaintiffs-including both those
who raise status-based claims and those who raise retaliation claims-
ought to be able to take advantage of the arguments that the but-for
principle makes available. 34 0
But there is also other language within Title VII that could be used to
argue for a contrary proposition-and specifically, that plaintiffs must
show that defendants possessed self-aware intent. The words "intent,"
"motive" or "discriminatory purpose" (or cognate terminology) appear
multiple times in Title VII-and several times more in the statutory
provision in which Title VII's damages provisions are codified, 42 U.S.C.
§ 1981a. 34 1 Most of the places in which such language appears do not pose
a significant challenge to arguing that the but-for principle is the central
342 See, e.g., 42 U.S.C. § 2000e-2(h) (granting a safe harbor to "bona fide seniority or merit
system[s]," except where those systems are the result of an "intention to discriminate").
343 See supra note 76.
344 42 U.S.C. § 2000e-2(m); Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1739-40 (2020).
341 Bostock, 140 S. Ct. at 1739-40.
346 See Staub v. Proctor Hosp., 562 U.S. 411, 418-19 (2011); EEOC v. Abercrombie
&
34 See Staub, 562 U.S. at 418-19. Though the Court went on to find for the plaintiff, it did
so on a rationale considerably more complex and difficult for a plaintiff to meet than a simple
"but for" standard. See id. at 422.
350 See Abercrombie, 135 S. Ct. at 2033.
3 See, e.g., Ponce v. Billington, 679 F.3d 840, 844-45 (D.C. Cir. 2012); see also Desert
Palace, Inc. v. Costa, 539 U.S. 90, 94 (2003) (characterizing § 2000e-2(m) as creating an
"alternative" means of proving an unlawful employment practice).
352 Bostock v. Clayton Cnty., 140 S. Ct. 1731, 1739-40 (2020).
3 See, e.g., Abercrombie, 135 S. Ct. at 2032.
2021] The But-For Theory ofAnti-Discrimination Law 1699
C. JudicialHeadwinds
As described above, while there are legal arguments that can be raised
against a but-for project of reforming anti-discrimination law, there are
also many potential legal avenues for success. But it is important to
acknowledge that "law" itself may not be the only obstacle that the project
of resolving anti-discrimination law's theoretical crisis is likely to face in
the courts. Rather, the existing attitudes and beliefs of judges are also
likely to serve as a potential barrier to reform. In particular, judges' beliefs
that summary judgment is important to preserve judicial resources,
reasonable jury could find for the plaintiff on the issue under
consideration. 368 The Supreme Court has made clear, moreover, that there
is no basis for varying the question or standards under consideration at
different procedural stages of the litigation-something that is clearly
violated in the case of the technical standards that the lower courts apply
to dismiss discrimination claims. 369 Moreover, to the extent that the view
of judges is that the summary judgment must be available in
discrimination cases in particular, this, of course, violates the
requirements of trans-substantivity that are a basic premise of the Federal
Rules of Civil Procedure. 370 Finally, there is no textual basis in Title VII
(or other anti-discrimination laws) for the summary-judgment-promoting
doctrines that the lower courts have embraced-and indeed such
doctrines are in conflict with the plain text insofar as they distract from
the core question of whether discrimination was "because
of. . . [protected class status]." 371
But even assuming that a desire to preserve the regular availability of
summary judgment was a legally legitimate basis for adopting (or
rejecting) standards of anti-discrimination law, it is far from clear that the
concerns that have animated judges in wishing to ensure the availability
of summary judgment are factually warranted. Leaving aside for the
moment distaste for anti-discrimination cases (an issue discussed at
length in Subsection V.C.3, infra), the most common justification offered
for the need for the availability of summary judgment is judicial
resources.3 72 Specifically, it is common to hear claims that judges need to
award summary judgment in a certain portion of cases because the courts
do not have the resources to try a much higher proportion of claims. 373
Thus, summary judgment might be thought of as an unfortunate but
necessary reality of the modern legal system in which the volume of cases
exceeds the judicial capacity. 374
36 See, e.g., Reeves v. Sanderson Plumbing Prods., Inc., 530 U.S. 133, 149-50 (2000).
369 See, e.g., Comcast v. Nat'l Ass'n of Afr. Am.-Owned Media, 140 S. Ct. 1009, 1014
(2020).
370 See generally Leatherman v. Tarrant Cnty. Narcotics Intel. & Coordination Unit, 507
U.S. 163, 168-69 (1993) (holding that the imposition of higher procedural standards on
particular classes of federal claims must come through amendment of the federal rules, not
judicial action).
37' 42 U.S.C. § 2000e-2(a)(1).
372 See, e.g., Thomas, supra note 175, at 177.
373 Id.
374 Id.
2021] The But-For Theory ofAnti-Discrimination Law 1703
"I See Berrey, Nelson & Nielsen, supra note 25, at 63.
376 There are risks to both parties of going to trial, even following the denial of a summary
judgment motion. For the plaintiff, they must face the risk of losing the case and walking away
with nothing, whereas for the defendant, they must face the risk of substantially more onerous
conditions (financial or otherwise) being imposed on them at trial (if they lose) than they might
be able to negotiate in a settlement.
377 See Thomas, supra note 175, at 177-79.
378 See Thomas, supra note 175, at 178 n.165.
379 See id. at 178.
380 Id.
381 See Nancy Gertner, Losers' Rules, 122 Yale L.J. Online 109, 113 (2012).
381 Adeyemi v. District of Columbia, 525 F.3d 1222, 1226 (D.C. Cir. 2008); see also Walton
v. Powell, 821 F.3d 1204, 1212 (10th Cir. 2016) (critiquing the tendency to overlay complex
legal frameworks on the discrimination inquiry at summary judgment, and noting that
"summary judgment [is] supposed to be that-summary").
1704 VirginiaLaw Review [Vol. 107:1621
Finally, it is important to note that not all judges share the perspective
that the inquiry at summary judgment in discrimination cases ought to be
technically oriented and promote the availability of summary judgment
awards. For example, as then-Judge Gorsuch put it in rejecting the
application of technical rules to First Amendment retaliation claims (in
favor of a straightforward factual approach):
[T]oday motions practice, and especially summary judgment
motions practice, seems to have assumed a place near the center of the
legal universe: almost no one makes it to trial anymore. With that
development surely comes a strong temptation to anoint summary
judgment with unique significance and adorn it with special rules and
procedures. But the truth is summary judgment was supposed to be
that-summary. Not a maddening maze. Not a paper blizzard. Not a
replacement for the trial as the preferred means for resolving
disputes.3 83
Ultimately, while judicial perception of the need for rules that would
permit summary judgment in discrimination cases is a real obstacle, it
would stand as an obstacle to any project of anti-discrimination reform.
But strong, indeed overwhelming, legal arguments exist for opposing
such a view-and even from a pure policy perspective, it is far from clear
that reducing awards of summary judgment in discrimination cases would
pose a challenge for judicial resources.
2. Rule Creep
A second likely judicial obstacle to a project of recentering anti-
discrimination law around the but-for principle is what I refer to as "rule
creep." As observed in Part III, supra, some circuits (including most
notably the U.S. Court of Appeals for the Seventh Circuit) have already
in theory adopted a simple, factually focused approach to anti-
discrimination claims. 384 But a review of recent cases reveals that in many
instances this has not stopped judges from reattaching technical rules to
the new, ostensibly straightforward, factual inquiry. 385 This "rule creep"
could pose a genuine risk to a project of but-for focused reform, insofar
386 See supra Subsection V.C.1 (addressing the perception among some judges that there is
a need for doctrine that will allow elevated levels of summary judgment) and infra Subsection
V.C.3 (discussing general judicial disfavor for anti-discrimination law claims).
387 1 have had adjudicators of discrimination cases express this view to me in informal
conversations about this project and prior work on the McDonnell Douglas paradigm.
388 See sources cited supra notes 385-87 and accompanying text.
389 See, e.g., Russell K. Robinson, Perceptual Segregation, 108 Colum. L. Rev. 1093, 1106-
39 (2008).
391 See id. at 1106-39, 1155-56, 1164-66.
1706 Virginia Law Review [Vol. 107:1621
justify awarding summary judgment, but not denying it. 391 Indeed, the
only arguably "technical" rule that meaningfully benefits anti-
discrimination plaintiffs today-the "pretext" rule-is a rule that the
Supreme Court has made clear is not "technical" at all, but rather simply
arises from the straightforward factual inferences that are permissible
where a litigant is not truthful. 392 (So, just as in other areas of the law, if
a defendant lies in the discrimination context, this can allow a jury to infer
guilt). 3 93 As such, the notion that anti-discrimination law's technical rules
allow for more neutral adjudication-as opposed to biasing the courts in
favor of discrimination defendants-is simply unsupportable in the
current legal landscape.
As such, for judges who are concerned about the potential for bias
when assessing anti-discrimination law cases untethered to technical
rules, it is important to emphasize that technical rules do not effectuate
this purpose-but that faithful adherence to the summary judgment
standards can. 394 At summary judgment, of course, the judge is supposed
to draw all inferences in favor of the non-moving party (typically the
plaintiff), and to ask only whether any reasonable jury could find in their
favor. 395 Thus, contrary to the current technical rules (which often as a
matter of law draw inferences againstplaintiffs), the judge should think
through what is the "best case" scenario for what this evidence could
mean-even if the judge does not agree. 3 96 Taking seriously these
summary judgment standards-something too few judges do today-
would better serve the principles of neutrality and anti-bias than the
current technical architecture of anti-discrimination law.
Finally, it is important to note that whatever the causes of "rule
creep"-disfavor of anti-discrimination claims, or legitimate concerns
about judicial biases-the strategic advocacy responses to "rule creep"
should generally be the same. Just as laid out in Part III and Section V.B,
supra, the project of resolving the theoretical crisis in anti-discrimination
law will be an iterative one, and preventing "rule creep" is just one
example of the need for this type of consistent advocacy. An opinion
stating that the but-for standard is the central inquiry-and that technical
rules ought not be applied-is simply the beginning, not the end point of
advocacy efforts to address the current crisis in anti-discrimination law.
Rather, consistent advocacy and messaging that the inquiry must be a
simple and factual one-and a willingness to eschew technical arguments
even where they may appear to benefit plaintiffs-will be critical to
ensuring that "rule creep" does not quietly reinstate problematic technical
legal standards.
397 See, e.g., Lee Reeves, Pragmatism Over Politics: Recent Trends in Lower Court
Employment Discrimination Jurisprudence, 73 Mo. L. Rev. 481, 519-21 (2008).
398 See supra notes 174-82 and accompanying text.
1708 Virginia Law Review [Vol. 107:1621
CONCLUSION
405 See, e.g., Berrey, Nelson & Nielsen, supra note 25, at 115-19, 121-25.
406 See, e.g., Spencer G. Park, Note, Providing Equal Access to Equal Justice: A Statistical
Study of Non-Prisoner Pro Se Litigation in the United States District Court for the Northern
District of California in San Francisco, 48 Hastings L.J. 821, 823 (1997) (finding that "civil
rights" claims, including discrimination claims, were the largest substantive category of non-
prisoner pro se filings); see also Berrey, Nelson & Nielsen, supra note 25, at 68 (describing
the dramatically lower success rates that pro se discrimination litigants experience).
1710 VirginiaLaw Review [Vol. 107:1621
treated differently based on their race, sex, or other protected class status.
The Supreme Court has held that a true disparate treatment principle-
the but-for principle-resides at the core of anti-discrimination law.
Moreover, it has made clear that this principle controls as a matter of anti-
discrimination law's plain text. This set of holdings affords myriad
opportunities to argue-in both the statutory and constitutional
contexts-that all disparate treatment must be proscribed.
While this objective may seem modest, in today's legal landscape it is
not. If all disparate treatment is proscribed, then it is illegitimate for courts
to dismiss cases based on technical legal barriers like the McDonnell
Douglas paradigm, the "same actor" rule, or rigid comparator
requirements. If all disparate treatment is proscribed, then the courts
ought not demand individualized evidence of self-aware intent to
discriminate (though such evidence may certainly still be useful to show
disparate treatment). If all disparate treatment is proscribed, then
doctrines that routinely immunize certain actors from being held
responsible for their discriminatory acts should be found illegitimate.
The but-for principle thus offers innumerable opportunities to
challenge the anti-plaintiff structures of anti-discrimination law today.
Simply by insisting that our commitment to prohibiting disparate
treatment must be honored, we can dismantle the biased architecture of
modem anti-discrimination law. But this important opportunity is not
self-executing: it will take a movement for "but for" to see its full effects.
Thus, whether anti-discrimination advocates and scholars embrace the
but-for principle-or fail to do so-may determine the future of anti-
discrimination law.