Introduction
In the summer of 2020, the Supreme Court decided the blockbuster case Bostock v. Clayton County, holding that Title VII prohibits employment discrimination on grounds of sexual orientation and gender identity. The opinion, authored by Justice Neil Gorsuch, claimed to base the result in textualism and the “simple” test of but-for causation. The three dissenters, in opinions by Justices Samuel Alito and Brett Kavanaugh, took an opposing view that the statute did not cover discrimination based on the employee’s sexual orientation—but also claimed to ground their opinions in textualism.
This collection of conflicting opinions ignited a battle over the meaning of textualism and its relationship to conservative and liberal movements. Justice Alito wrote:
The Court’s opinion is like a pirate ship. It sails under a textualist flag, but what it actually represents is a theory of statutory interpretation that Justice Scalia excoriated––the theory that courts should ‘update’ old statutes so that they better reflect the current values of society.
Several conservative scholars followed this in tow. Josh Blackman and Randy Barnett labeled Justice Gorsuch’s opinion as “halfway textualism,” and Nelson Lund described it as “analytically untenable,” an “outlandish judicial performance,” and a “fatally flawed” application of “textualist principles.” The opinion even kindled questions of whether conservatives should discard textualism altogether, in favor of a judicial methodology that is expressly guided by conservative moral values.
On the liberal side of the discourse, the decision was roundly praised, and rightly so with respect to the result. It was a momentous victory for lesbian, gay, and transgender people and, indeed, for greater social justice more generally. It even led some scholars to reexamine textualism and its potential for advancing liberal and progressive causes.
More granularly, it also has led scholars to extol the but-for causation standard—which the majority opinion used to justify its purportedly textualist result—as the best way forward for anti-discrimination law. Chief among them is Professor Katie Eyer. In her article The But-For Theory of Anti-Discrimination Law, Eyer explains that anti-discrimination law has “display[ed] a conceptual confusion of disparate treatment and intentional discrimination.” This, in turn, has led to judicial lawmaking that improperly limits the reach of anti-discrimination law. And, she notes, advocacy and scholarship have been jumbled and disorganized in opposing these efforts. A large part of that, in her view, is the absence of a central message and a central principle for anti-discrimination law to build upon. In light of these confusions, she contends that we should “reorient[]” the basic factual inquiries of anti-discrimination law to the but-for causation test, for it is both grounded in textualism and can best “ensur[e] that anti-discrimination law can achieve its basic promises.”
However, but-for causation, especially as the Court understands it, is unlikely to improve the situation. In a recent coauthored article, Bostock was Bogus: Textualism, Pluralism, and Title VII, Mitch Berman and I contend that the result in Bostock is not in fact justified by textualism. Specifically, we argue that the ordinary meaning of Title VII does not cover discrimination “because of” an individual’s sexual orientation. Furthermore, we explain how the proper understanding of but-for causation does not cover discrimination on the basis of one’s sexual orientation. To this end, we formulate a constraint on how one chooses the comparator in but-for reasoning—what we call the Principle of Conservation in Motivational Analysis (PCM):
[In performing counterfactual analysis,] [w]hen changing one fact requires changing other facts too, the analyst must not change facts that are known, confidently believed, or stipulated to have been among the actor’s motivating reasons in favor of facts that are not likely, or less likely, to have been among the actor’s motivating reasons.
Applied to Bostock, we contend that this would constrain the choice of comparator and thus show that the plaintiffs were not discriminated against but for their sex. Consequently, because we favor the result in Bostock, and because the proper textualist but-for analysis does not deliver it, we think that provides reason to disfavor textualism. That is our beef with textualism.
But life is short, and that Article is long, so I do not rehash it here. Instead, I contend here that the Court’s simple but-for causation test, by its own lights, does not advance anti-discrimination law. To be sure, Eyer and I agree on the goals of advancing anti-discrimination law: that Title VII be more capacious in recognizing discrimination on the basis of protected traits in its varied forms, that plaintiffs have greater ability to raise their claims, and ultimately that there be less invidious discrimination. Yet the but-for test simply fails to advance these goals. First, it does not cover under Title VII core cases of discrimination that it ought to recognize—including discrimination against bisexual, pansexual, and trans people. Second, the simple but-for test can be used as a sword to cut down policies that have made our workplaces safer and less discriminatory. This leads me to conclude that simple but-for causation is not the appropriate foundation for anti-discrimination law. Instead, I suggest that we should approach anti-discrimination statutes with a pluralist lens, and I find support for this in Eyer’s own analysis.
This Essay proceeds in three Parts. First, I briefly set forth the reasoning of the Bostock majority and explain the simple but-for causation test. Second, I observe that the simple but-for test surprisingly fails to cover cases of discrimination that it ought to cover, like discrimination against bisexual and pansexual individuals and people whose sexual presentation is ambiguous. Third, I show how the simple but-for test has broad over-coverage that may threaten important workplace anti-discrimination policies, including affirmative action. Finally, I conclude with a discussion of why a pluralist interpretation of Title VII better realizes the aims of anti-discrimination law.
I. The Bostock Majority and the “Simple” But-For Test
In early 2019, the Supreme Court granted certiorari in three cases—Zarda v. Altitude Express, Bostock v. Clayton County, and Equal Employment Opportunity Commission v. R.G. & G.R. Harris Funeral Homes—and consolidated them for hearing. In a 6–3 decision, the Court held that discrimination on the basis of sexual orientation or transgender status constitutes discrimination “because of [an] individual’s . . . sex” and therefore violates Title VII. The majority opinion, written by Justice Gorsuch, contended that the text of the statute demanded this result. Title VII states in relevant part:
It shall be an unlawful employment practice for an employer—
(1) to fail or refuse to hire or to discharge any individual, or otherwise to discriminate against any individual with respect to his compensation, terms, conditions, or privileges of employment, because of such individual’s race, color, religion, sex, or national origin; or
(2) to limit, segregate, or classify his employees or applicants for employment in any way which would deprive or tend to deprive any individual of employment opportunities or otherwise adversely affect his status as an employee, because of such individual’s race, color, religion, sex, or national origin.
In parsing this text, Justice Gorsuch first located the key phrase “because of such individual’s . . . sex.” He stated that “[i]n the language of law,” a person is fired “because of” their sex if their sex is a “but-for” cause of the discrimination. He then explained that the but-for test operates in the following way: We “change one thing at a time and see if the outcome changes. If it does, we have found a but-for cause.”
Consider then the majority’s application of the but-for test to Bostock’s and Zarda’s cases. Gerald Bostock was employed by Clayton County, Georgia as a child welfare advocate, and David Zarda was a skydiving instructor working for Altitude Express. Both men, Bostock and Zarda, alleged that they were terminated for being gay—that is, because of their sexual orientation. The majority applied the simple but-for test as follows: We would change one thing—Bostock’s and Zarda’s sex, while keeping everything else constant. Most relevantly, we would keep constant the fact that Bostock and Zarda were attracted to men. Thus, the comparators of Bostock and Zarda would be women who were attracted to men. And because the employers, Clayton County and Altitude Express, would presumably not have terminated women who were attracted to men, “changing the employee’s sex would have yielded a different choice by the employer.” Thus, on the majority’s account, Bostock and Zarda were fired “because of” their “sex.”
Next consider how the majority applied the but-for test to Aimee Stephens’s claims of discrimination based on her transgender status in Equal Employment Opportunity Commission v. R.G. & G.R. Harris Funeral Homes. Suppose “an employer . . . fires a transgender person who was identified as a male at birth but who now identifies as a female. If the employer retains an otherwise identical employee who was identified as female at birth,” then “the employer intentionally penalizes a person identified as male at birth for traits or actions that it tolerates in an employee identified as female at birth.” Consequently, the employee would not have been terminated but for the employee’s sex and was therefore discriminated against “because of” the employee’s sex.
In both cases, we see the simple but-for test at work. However, we observe that the test doesn’t actually work as Justice Gorsuch says: When applying the test in these cases, we aren’t changing just one thing. In the Bostock–Zarda example, the employee is (1) a man, (2) attracted to men, and (3) gay. The comparator is (1) a woman, (2) attracted to men, and (3) heterosexual. So, two things changed. And if we’re open to changing two things, then there is an alternative comparator: a person who is (1) a woman, (2) attracted to women, and (3) gay. The majority opinion doesn’t provide a principled reason why we should choose one comparator over the other.
In our forthcoming work, Berman and I suggest that PCM does limit the choice of comparator, and that in the Bostock and Zarda cases, the alternative comparator—the woman who is attracted to women—is the appropriate choice. Recall PCM’s demand in performing counterfactual analysis: “When changing one fact requires changing other facts too, the analyst must not change facts that are known, confidently believed, or stipulated to have been among the actor’s motivating reasons in favor of facts that are not likely, or less likely, to have been among the actor’s motivating reasons.” As a general rule, when engaging in but-for analysis, we want the counterfactual to be as close to the actual world as possible. PCM aims to capture this proper way to conduct counterfactual reasoning with respect to the actor’s motivations, imperfect as it may be.
Applying it to the Bostock and Zarda cases, we see that the majority’s comparator—the heterosexual woman—violates PCM because it changes what we have strong reason to believe is among the actor’s motivating reasons—that the targeted employee was gay. In contrast, the alternative comparator keeps that fact of sexual orientation constant. And using the alternative comparator, there is good reason to think that the comparators would have also been fired, given their sexual orientation. Consequently, under proper but-for analysis, Bostock does not vindicate the result that discrimination based on an individual’s sexual orientation is discrimination based on an individual’s sex, and thus actionable under Title VII.
Eyer disagrees with us, arguing that we have the wrong comparator. She contends that the alternate comparator illicitly uses a superfluous category of sexual orientation, which is simply built from the categories of sex/gender and object of attraction. In response, one could raise the charge that the object of attraction is the superfluous category, once sex/gender and orientation become fixed. That line of argument won’t resolve the question. Thus, we proffer a principled, non-ad hoc basis on which to choose between the comparators: PCM.
For Eyer, then, there are two avenues of further response. First, Eyer may have a competing explanation for why her preferred comparator is better. For example, the explanation may rest on which characteristics are more fundamental. That would require far more explanation, and in any event would likely undercut the posited simplicity of the but-for test. Or it could be that, if there are multiple comparators, the plaintiff may choose among them in framing the argument. Then, so long as the plaintiff, equipped with their choice of comparator, can show the but-for relationship with a protected ground, they have shown the requisite “because of” relationship in Title VII. This but-for test may be expansive in the types of discrimination it cognizes.
* * *
Eyer sees much promise in this simple but-for standard and she thinks it should serve as a new foundation for anti-discrimination law. She writes, “In an era where textualism is the ascendant method of statutory interpretation at the Supreme Court, this type of argument may have considerable promise in addressing the conceptual confusion at the core of disparate treatment doctrine, and in mandating a true disparate treatment standard.”
That conceptual confusion, as Eyer sees it, is in the tension of whether our anti-discrimination law is about “disparate treatment” or “intentional discrimination.” Eyer contends that disparate treatment and intentional discrimination are not coextensive, yet case law has often treated them as such. This results in a tension of what must be shown to prove discrimination: Must the plaintiff show that they were treated less favorably because of a protected trait, or that the employer engaged in intentional discrimination, or both? And there are related questions on how to treat cases of systemic discrimination as opposed to individual discrimination. Eyer observes that, in the muddled understanding of anti-discrimination law, “judicial lawmaking has run amok.” As she relates, courts have concocted a number of technical barriers to plaintiffs bringing claims to vindicate their rights against discrimination.
She claims this is where but-for causation can provide a solution. Eyer contends that the but-for test can resolve the conceptual confusion between whether anti-discrimination law is about disparate treatment or intentional discrimination. This, she claims, was shown by (among other examples) the Bostock case—which she thinks shows the potential for the simple but-for test. And from the advocacy lens, this could also provide greater potential for recourse for plaintiffs.
Eyer acknowledges that Bostock did not dispense with the intent requirement. Instead, she says that the Court has conceived of the intent requirement as the employer’s intent to subject the employee to differential treatment. She proffers that this simple but-for causation standard, with this more basic intent requirement, is what anti-discrimination scholars should build upon.
Then Eyer recognizes four alternative paradigms of discrimination: (1) stereotyping jurisprudence; (2) negligent discrimination; (3) disparate impact; and (4) the motivating factor paradigm. She argues that the simple but-for test either wholly or partially incorporates these alternatives, and where they conflict, the but-for test is more favorable for the promises of anti-discrimination law. Thus, the simple but-for test can serve as the proper foundation for rebuilding anti-discrimination law.
II. The Under-Coverage of the Simple But-For Test
Along with other advantages, Eyer celebrates the simple but-for test for its ability to recognize discrimination on the basis of sexual orientation and transgender status as unlawful discrimination under Title VII. Indeed, she suggests that this has potential beyond this important-but-singular question. To her, the simple but-for test has the potential to recognize other kinds of discrimination that, under the prior conceptual confusions of anti-discrimination law, may have otherwise escaped the reach of Title VII.
However, this seeming benefit is illusory. First, the simple but-for test cannot even recognize discrimination on the basis of bisexual and pansexual orientation as Title VII discrimination. Second, it cannot recognize pretextual policies targeted at discriminating against transgender individuals as Title VII discrimination. The supposed payoff of the simple but-for test—which comes at the steep expense of some absurd results—falls away.
A. Bisexual and Pansexual Discrimination
Consider individuals who identify as bisexual or pansexual. I understand an individual to be bisexual if they are romantically attracted to both men and women (or both males and females). I understand an individual to be pansexual if they are romantically attracted to individuals of any sex/gender. Now consider an employer who has a policy against hiring bisexual or pansexual employees. Under Bostock, such a policy would appear to violate Title VII, because it is discrimination based on the individual’s sexual orientation and therefore discrimination because of such individual’s sex.
But the simple but-for test fails to produce this conclusion. Let’s run the simple test on an example to see why. Imagine two employees Xander, a bisexual man, and Yasmine, a pansexual female. They both work at Zizi, Inc., which has the aforementioned policy against employing bisexual or pansexual people. When the owner and manager, Jack Zizi, learns of their sexual orientations, he fires both Xander and Yasmine. Was this discrimination because of—that is but for—Xander’s and Yasmine’s sex?
Xander is (1) a man, and (2) attracted to both men and women (bisexual). So for the comparator X, we change one thing, Xander’s sex, from man to woman. Comparator X is (1) a woman, and (2) attracted to both men and women (bisexual). But comparator X would still be fired under the policy. Thus, Xander’s sex—being a man—is not a but-for cause of his firing.
It’s a similar result for Yasmine. Yasmine is (1) a female, and (2) attracted to individuals of any sex/gender. Changing just Yasmine’s sex from female to male, we have that comparator Y is (1) a male, and (2) attracted to individuals of any sex/gender. But under the policy, comparator Y is still fired, because they are still pansexual. Thus, Yasmine’s sex—being a female—is not a but-for cause of her firing. In both cases, Xander and Yasmine are not discriminated against because of their sex, and this is not cognizable as Title VII discrimination under the simple but-for test.
One way of resisting this is to appeal to analogy: Surely, if discrimination for being gay or being lesbian is discrimination because of one’s sex, then so too would discrimination for being bisexual and pansexual.
That seems right, as a matter of sound legal reasoning. But that isn’t how the simple but-for test works. It’s a different test that adds analogical reasoning to but-for causation—call it the “but-for + analogies” test. And embracing the need for analogy shows that the simple but-for test is not enough.
This strikes me as a deeply concerning result for the Bostock majority’s simple but-for test. One of the main virtues of the simple but-for test is that it delivered the just outcome that discrimination because of an individual’s sexual orientation is discrimination because of such individual’s sex. But as this example shows, that isn’t always the case, at least as far as the simple version of the test goes.
B. Pretextual Trans Discrimination
Even as a purely textual matter, Title VII was rightly read to recognize that discrimination on the basis of an individual’s transgender status is discrimination because of such individual’s “sex.” That’s because the ordinary meaning of “sex” includes transgender status. The Bostock majority attempts to capture this through the but-for test by using the example of a transgender employee “who was identified as a male at birth but who now identifies as a female.” Justice Gorsuch reasoned that if the employer were to treat the employee differently if they were identified as female at birth, then that is discrimination “but-for” sex, which violates Title VII.
But this is tenuous reasoning. What if the employer formulates the policy differently? Suppose an employer has a policy against hiring or employing “anyone whose sex/gender cannot be determined by the employer by the employee’s appearance.” The employer then fires an employee Jamie on the basis that their manager cannot determine Jamie’s sex/gender by Jamie’s appearance.
The simple but-for test applies as follows: Let’s assume (without loss of generality) that (1) Jamie is a trans man, and (2) Jamie’s appearance is such that Jamie’s manager—the agent of the employer—cannot determine Jamie’s sex/gender. Changing one thing at a time, we can make comparator J a cis man (or a trans woman, cis woman, or nonbinary). But, by operation of the simple but-for test—which tells us to change as little as possible in the comparator—(2) remains the same: J’s appearance is such that their employer cannot determine their sex/gender. Comparator J is still fired by their employer. Consequently, under the simple but-for test, Jamie’s “sex” is not a but-for cause of Jamie’s termination, and thus their termination is not “because of such individual’s . . . sex.” Here again the but-for test has failed our intuition about the right result: that such a policy is unlawful discrimination under Title VII.
One initial response is to assert that (2), Jamie/J’s appearance, would not have remained the same, if (1) Jamie/J’s sex were to change. But there’s no reason why not. It is certainly possible for J to have essentially the same facial features, keep the same hairstyle, wear the same clothes, etc.—such that J’s appearance would not reveal J’s sex/gender. And given the possibility, keeping (2) the same is exactly what is required of us when implementing the simple but-for test. In Justice Gorsuch’s words, “a but-for test directs us to change one thing at a time and see if the outcome changes. If it does, we have found a but-for cause.” That’s what we’re doing.
Another counter is that, properly understood, the term “sex” encompasses the appearance of one’s sex as well. This may well resuscitate the simple but-for test in this case; if discriminating against an individual based on their appearance is just discrimination because of such individual’s sex, then this conduct is unlawful under Title VII. But this is in no way special to the simple but-for test. Proposing that “sex” encompasses appearance of sex would enable the other aforementioned accounts—anti-stereotyping; negligent discrimination (because it would be intentional); disparate impact; and motivating factor—to cognize this policy as discrimination based on one’s sex. Indeed, this move seems to be more in accord with the anti-stereotyping account, which observes that much sex discrimination is based on discrimination of individuals for failure to conform to sex stereotypes.
Ultimately, I contend this response reveals that the simple but-for test isn’t doing the work to show that this hypothetical anti-transgender policy is actionable sex discrimination. Rather, the failure of the simple but-for test requires that we broaden the meaning of the term “sex” to obtain the intuitively correct result. This should give us pause in thinking that the simple but-for test itself serves as a solid foundation for a progressive anti-discrimination law. It just doesn’t deliver the intuitively correct results.
Finally, Eyer suggests another move that abstracts from the individual to the group. She says, “where a policy or practice would not have been adopted ‘but for’ the group adversely affected (or the group advantaged), the but-for principle is violated.” Here, the idea might be that but for the existence of trans individuals—and trans identity—the employer would not have adopted this appearance rule. Thus, this counts as discrimination “because of” trans identity, which in turn is “because of” sex.
First, this would raise some difficult epistemic issues. It is difficult to know how one would prove an employer had a generalized group in mind when formulating a neutral policy. I envision that the employer would need to expressly state so or perhaps demonstrate an animus toward the impacted group. But this gets us to the kind of analysis of intentions that the but-for test was supposed to allow us to circumvent.
More importantly, it may have a drastic over-coverage problem. Consider a simple “no racism” policy. A “no racism” policy only exists in light of the fact that there are races. That is, but for the existence of the races, a “no racism” policy wouldn’t be promulgated. Does that mean every “no racism” policy violates Title VII? Eyer surely has in mind a more sophisticated, restrictive construction of this “group” but-for test. However, such a but-for test occurs to me as anything but simple, and indeed I think it will rely on various types of reasoning to define its contours.
* * *
This is not to say that the result in Bostock was incorrect or that Title VII does not cover discrimination based on one’s sexual orientation or transgender status. As suggested above, a broad interpretation of the term “sex” may allow for recognizing sexual orientation discrimination and pretextual discrimination targeting transgender individuals. Another way to recognize such discrimination under Title VII may be through what I have called “conceptual causation.” Under this proffered doctrine, A is “because of” B, when “A conceptually depends on B”—that is, “an analysis of A requires an analysis of B.” Indeed, the Bostock majority at times seems to rely on this doctrine. Professor Benjamin Eidelson has proffered an intriguing account that builds upon conceptual causation, which he calls the dimensional account. And finally, there is a strong pluralist argument in favor of Bostock’s result. The key point is that these other ways to recognize transgender and sexual orientation discrimination under Title VII are distinct and independent from the simple but-for test. And that’s for good reason—the simple but-for test doesn’t do the requisite work.
III. The Over-Coverage of the Simple But-For Test
I have shown that, despite the result in Bostock, the simple but-for test fails to cover even basic forms of sexual orientation discrimination and pretextual policies that target transgender status. But if that were not enough, the but-for test would also recognize as discrimination under Title VII many kinds of employment policies that have been key to ensuring workplaces are safe from discrimination.
As a theoretical matter, this is perhaps unsurprising. In no other area of the law does but-for causation constitute the extent of a legal test of liability. In tort, causation requires a showing of but-for causation, but one must show proximate causation as well. That’s in part because but-for causation can be expansive. Our biological parents giving birth to us are but-for causes of everything we do. But it would be a stretch to say that a misbehaving employee was fired because of their biological parents. So in embracing the simple but-for test, we are pellucidly signing up for over-coverage problems. But beyond absurd cases, below I explain how this may impede anti-discrimination policies.
A. Restricting Anti-Bigotry Policies
First, consider an employer who promulgates an anti-bigotry policy for employees. Under the policy, employees who behave in racist or sexist ways will be disciplined or terminated. Now, we can recognize that, as a matter of social fact, there are certain kinds of acts that are generally understood to be bigoted if committed by an out-group person, but not considered bigoted if committed by an in-group person. This is typically the case, for example, with uttering certain slurs or speaking with particular vocabulary or in vernacular.
Suppose a white employee calls a Black coworker a slur. This is the kind of behavior that may be seen as racist when committed by the white coworker, but would not be if uttered by a Black person. As a consequence of the racist behavior, the white employee is promptly fired. My intuition is that the employer’s firing of the white employee for calling a Black coworker a slur is not discriminatory as a matter of Title VII. And I think that is an important result for the operation of an anti-bigotry policy that is sensitive to social context and history.
However, the but-for test would decide otherwise. The white employee may argue that their race was a but-for cause of their firing: The white employee (1) is white, and (2) used a particular slur to refer to a Black person. We change just (1), such that the comparator is a Black person. By assumption, had the Black coworker used the particular slur, it would not have been seen as racist. Consequently, the comparator—who is Black—would not have been fired. The white employee’s race was therefore a but-for cause of his firing, thus violating Title VII.
One might be inclined to embrace this conclusion, that under Title VII employees should never receive differential treatment, even based on social facts about what constitutes racism or sexism in different contexts. I disagree; anti-discrimination law is not furthered by the principle that all policies must be blind to the history and social facts about bigotry.
B. Affirmative Action
Nothing makes clearer the importance of sensitivity to social context and history than affirmative action. “Affirmative action” refers to programs in which “minority groups may be given an advantage in admissions or employment (in order to account for historical and contemporary discrimination or to ensure diversity).” There are several justifications for affirmative action. Among them are that affirmative action is necessary to restore society to a nonracist (or nonmisogynistic) position; that it remedies discrimination, either latent or overt, in hiring and admissions processes; that it promotes the interests of diversity; and that it prevents discrimination by combating tokenism and stereotype threat.
Affirmative action programs can take many forms and I don’t attempt here to cover all programs that receive such a label. The hallmark of an affirmative action program in hiring is that, by its operation, it will result in some minority individual receiving a position that they otherwise would not have received. And, in hiring, when there are more applicants than positions—which is regularly the case—this will result in some minority individual receiving a position that a non-minority individual would have otherwise received. This is uncontroversial—that’s the whole reason for the program. Speaking generally then, affirmative action programs are important for furthering the promises of anti-discrimination law. And indeed, as demonstrated by United Steelworkers of America v. Weber, Johnson v. Transportation Agency, and United States v. Paradise, such programs are legal if employed for an appropriate purpose—such as to remedy historical discrimination.
Yet such programs would fail the simple but-for test. Consider an affirmative action hiring program that operates on race. As discussed above, by operation of the program, some individual who belongs to a racial minority will receive the position, which some individual who belongs to a racial nonminority would have otherwise received. Thus, but for the race of the individual who belongs to a racial nonminority, they would have received the job. And thus, under the simple but-for test, that would count as a Title VII violation.
A number of scholars worry that the natural extension of Bostock’s but-for causation analysis is the invalidation of affirmative action programs under Title VII and similar anti-discrimination statutory schemes. Eyer contends that the ship has sailed with respect to affirmative action programs that explicitly use racial classifications—they are already constitutionally disfavored in our textualist era regardless of the but-for principle, and therefore should not serve as an impediment to adopting the but-for test.
Eyer instead takes solace in the fact that, in the next constitutional battleground, “race-intentional remedial policies that do not explicitly classify [based on race]” may be bolstered by the but-for principle. Consider Eyer’s example of school integration policies that do not explicitly use racial classifications but are race intentional in that they seek a racially integrated school. Eyer contends that such a program would pass constitutional muster under the but-for test, because the measures would have been adopted regardless of the race of the individuals affected.
This strikes me as the wrong way to proceed. First, I do not think that we should concede that affirmative action programs that explicitly use racial classifications are constitutionally invalid or doomed. If it is true that today’s textualism is incompatible with such affirmative action programs, then all the worse for today’s textualism. Indeed, I might be so bold as to say that is a reductio ad absurdum of textualism—and certainly not a reason to adopt the purportedly textualist but-for test. And insofar as Eyer is suggesting we fall in line with the Court’s mistaken jurisprudence, I humbly suggest we should instead hold our own.
Moreover, I am not at all convinced that the but-for test will be a useful tool in crafting a sensible jurisprudence for race-intentional policies that do not explicitly use racial classifications. Consider again the example of school integration. Suppose there is a de facto segregated municipality. The municipality undertakes a race-intentional policy that does not use explicit classifications in order to create a racially integrated school. Eyer states that “it seems a much more difficult claim to suggest that their actions were disparate treatment, i.e., that they would not have been taken ‘but for’ the race of those affected.” This occurs to me as mistaken. By hypothesis, the policy is “race-intentional”; it thus takes into account the race of the affected individuals, even if not explicitly stated. If the race of the affected individuals were different, we would certainly expect a different policy. Eyer considers the policy action of geographically locating the school. If the racial makeup of the school district was substantially different in terms of percentages and geographical breakdown, wouldn’t we expect that the school would need to be located differently to maximize racial integration?
If it is the case that such race-intentional policies that do not use explicit classifications are constitutionally valid under the but-for test, what mischief is in store? Can a race-intentional policy aimed at segregating schools pass constitutional muster? We could craft a similar example simply substituting the nefarious purpose that would analogously pass the but-for test. That is a clear reductio, suggesting that the but-for test cannot proffer a sensible jurisprudence of race-intentional policies that are explicitly neutral. Consequently, the but-for test neither preserves the important results of affirmative action, nor promises advancement with respect to race-intentional, but explicitly neutral policies. That is sufficient reason to think that the simple but-for test is just not the way forward for those who want a robust anti-discrimination law that aims at and is capable of rectifying historical oppression.
C. Requiring Ignorance
Finally, to show how expansive the simple but-for test would be, consider a hypothetical concerning an employee lying on paperwork. Suppose an employer, in the course of hiring, has paperwork that includes a self-identification survey that asks about race, sex, and sexual orientation. The paperwork is optional, but the employer adds a notation that, in filling out the paperwork, prospective employees should not lie about this information. Seems fair enough.
A prospective employee lies about their race, thinking that it will better their chances of getting hired. At the conclusion of a long process, they are ultimately hired. After some time, the employee states to other coworkers that they lied on the paperwork. One of the coworkers reports this to the hiring manager. The hiring manager examines the documentation and approaches the employee to ask them about it. The employee then confesses to lying about their race and the hiring manager fires the employee for lying on the application.
I think it’s intuitively clear that if the employer fires the employee for lying on the hiring paperwork about their race, that isn’t and shouldn’t be actionable discrimination under Title VII as race discrimination. But that is precisely the conclusion we might draw from applying the simple but-for test.
Suppose for example a white employee claims to be Black. Here is what we know about the employee: (1) They are white; and (2) they wrote on the paperwork that they were Black. If we are to change only one thing at a time, namely the employee’s race from white to Black, then the employee would not have lied about their race on the paperwork and wouldn’t be fired by the employer. Thus, the employee’s race was a but-for cause of the firing. Embracing the simple but-for test, we come to the conclusion that the employer fired the employee “because of” the employee’s race, in violation of Title VII. That occurs to me as wrong—and devastatingly so for the but-for test.
It may seem like a far-fetched example, but it’s not. And the potential overreach of the but-for test can have substantial consequences. It may severely restrict the ability of companies to promote diversity and equity in their ranks. If companies cannot even ask about diversity, trusting that they may obtain genuine and accurate information, how can they even run a diversity or affirmative action program?
Conclusion
I have shown that the but-for test would serve as a poor foundation for a progressive anti-discrimination law. It would fail to cover basic cases of discrimination, including on the basis of sexual orientation and trans status. And it would also prohibit important anti-bigotry and affirmative action policies in workplaces. But the question remains: How should we rebuild our anti-discrimination law in the wake of lingering conceptual confusions?
I think that pluralism is the way forward in interpreting anti-discrimination statutes like Title VII. Specifically, in interpreting and applying Title VII, I contend that one should attend to a variety of factors: original textual meaning, current meaning, legislative intentions and broader purposes, historical practice, avoiding unforeseen absurdities, and society’s shared moral commitments, among others. In many cases, these different modalities will align. When they don’t, we are confronted with a hard case; weighing the factors is not mechanical and it can be genuinely difficult and controversial.
For many, at least at first impression, this lack of certainty makes pluralism a disfavored approach. Eyer herself thinks that the lack of a simplified approach, for scholars and activists to unify behind, has been an obstacle in fixing anti-discrimination law.
It shouldn’t be. There is good reason to favor pluralism in the context of constitutional interpretation, and many of those considerations translate to the statutory context. With respect to anti-discrimination law in particular, I contend that a simplified theory, like simple but-for causation, will be narrow, unworkable, or both: it will wrongfully exclude cases of discrimination from its aegis or cover so many as to be rendered useless.
Most telling, I think, is that Eyer’s own analysis itself exhibits the hallmarks of pluralist analysis. She frames her defense of the simple but-for test in the context of the “basic promises” of anti-discrimination law. In arguing that the but-for test fulfills these promises, Eyer appeals to the history and precedent of anti-discrimination law in grounding its aims. She contends that the but-for theory is most consistent and consilient with the statutory text and the Court’s precedent. And she argues that but-for causation best preserves our intuitions on what are valid discrimination claims. To this point, she claims among other things that the but-for theory would enhance the prospects of plaintiffs bringing discrimination claims, while not hindering affirmative action programs.
Now I have explained why I disagree about the purported benefits of the but-for test, finding it peppered with intuitive deficits and absurdities. But I agree that these are exactly the factors and modalities that we should consider in interpreting Title VII. Instead of searching for a simplified theory, like but-for causation, I contend we should embrace such a pluralist method. The pluralist method can be applied in each case, with careful attention to the statute’s legislative purposes, its text, current society’s shared moral commitments regarding the subject matter, and the feasible operation of the statute. That—and not simple but-for causation—is the method of interpretation that could rationally deliver the rightly celebrated result in Bostock. And that is the method that we should embrace in shaping our anti-discrimination law and continuing the march for equality.