In 2014, a nonprofit called Students for Fair Admissions (SFFA) sued Harvard University, alleging that its race-conscious admissions program discriminated against Asian American applicants.1 Over the next few years, the lawsuit drew both attention and controversy.2 Some criticized Harvard for imposing de facto Asian American quotas.3 Others suggested that SFFA’s challenge to Harvard’s admissions practices was a wolf in sheep’s clothing — an attempt to invalidate affirmative action in the guise of preventing discrimination.4 Recently, in Students for Fair Admissions, Inc. v. President & Fellows of Harvard College,5 the First Circuit ruled that Harvard’s admissions program was legal.6 But the court avoided addressing the question that could pose the greatest threat to affirmative action in the future: Would proving that the college discriminated against Asian American applicants have been sufficient to bring down Harvard’s entire affirmative-action admissions program? Although the First Circuit was right to let Harvard admissions stand, its failure to grapple explicitly with this question left open the possibility of a new form of strict scrutiny fatal to affirmative action.
An application winding its way through Harvard’s undergraduate process first stops at the desk of a regional admissions officer.7 That officer, familiar with the applicant’s school and geographic region, gives the applicant an overall rating and a rating in four subcategories: academic, extracurricular, athletic, and personal.8 The officers then meet in subcommittees to advance the applications they believe merit further review.9 Finally, the full admissions committee votes on recommended admits, makes a tentative offer list, and makes cuts to that list until it reaches its target number of admissions offers.10
Harvard can explicitly consider the race of an applicant at multiple points in the review process. Individual admissions officers can do so when they assign an applicant their overall rating,11 subcommittees can do so when they make recommendations,12 and the full committee can do so when it cuts students from its tentative admit list.13 In addition, the Dean of Admissions shares the projected racial makeup of the admitted class with the committee throughout a year’s admissions cycle.14 These projections influence who the committee selects for its tentative list and who it cuts.15
SFFA challenged Harvard’s admissions practices under Title VI of the Civil Rights Act.16 SFFA argued that Harvard violated Title VI for three reasons that fit neatly within affirmative action doctrine: engaging in impermissible racial balancing, using race as a mechanical plus factor, and not adequately considering race-neutral alternatives.17 But SFFA also argued that Harvard violated Title VI for a reason not conventionally associated with challenges to affirmative action programs: intentionally discriminating against Asian American applicants.18
The district court agreed with SFFA that strict scrutiny, or the requirement that a race-conscious program be “narrowly tailored” to a “compelling . . . interest,”19 was the right standard under which to review its claims.20 However, comparing Harvard’s admissions process to the admissions processes challenged in relevant Supreme Court precedent, the court rejected SFFA’s traditional affirmative action claims.21 In addition, the court rejected SFFA’s intentional discrimination claim after examining the statistical regression models created by SFFA’s and Harvard’s experts.22 Both experts modeled the impact of race on an applicant’s admissions chances, but they disagreed over methodology and arrived at different conclusions.23 Primarily, they disagreed over whether to include in their models a variable for the “personal” ratings assigned to applicants.24 SFFA’s expert argued that personal ratings were themselves inflected with racial bias and that including them in the model would therefore skew its results by diminishing the predictive value of the race variable.25 Harvard’s expert argued the opposite: “[P]ersonal rating[s] capture . . . relevant characteristics unrelated to race” for which no other variable accounts.26 Omitting the personal ratings, SFFA’s expert found that being Asian American had a statistically significant negative effect on an applicant’s chance of admission.27 But Harvard’s expert found that, including a personal rating variable, being Asian American had no statistically significant effect on admissions chances.28 While the court was more persuaded by Harvard’s expert’s methodology, it concluded that “models with and without the personal rating are econometrically reasonable and provide evidence that is probative of the effect of race on the admissions process.”29 Therefore, the statistical evidence was “inconclusive” and did not prove intentional discrimination.30
On appeal, the Department of Justice (DOJ) submitted an amicus curiae brief clarifying why, in its view, SFFA’s evidence of discrimination against Asian American applicants should cause Harvard’s admissions program to fail strict scrutiny.31 According to DOJ, for an affirmative action program to satisfy the narrow tailoring requirement of strict scrutiny, it must not penalize members of any racial group.32 Narrow tailoring analysis, DOJ argued, requires “a comparison between members of racial groups that are favored and those that are not.”33 And since Harvard exhibited bias against Asian American applicants, the university was using race beyond what was necessary to achieve diversity.34
The First Circuit affirmed the district court.35 Writing for the panel, Judge Lynch36 analyzed Harvard’s affirmative action policy separately from SFFA’s claim of intentional discrimination against Asian American applicants.37 In the opinion’s admissions program section, the court ruled that Harvard’s explicit policies survived strict scrutiny, as they were narrowly tailored to the compelling interest of diversity.38 Finding that Harvard did not engage in racial balancing or use race as a mechanical plus factor, the panel emphasized the multiple types of diversity Harvard pursues in its student body and the holistic nature of the school’s review.39 And finding that Harvard adequately considered race-neutral alternative admissions schemes, the panel emphasized the hits that Harvard would take to its fundraising capabilities40 and the “increase[d] . . . isolation and alienation” that Black students might feel were Harvard to admit fewer Black students as a necessary consequence of adopting race-blind admissions.41
In the opinion’s intentional discrimination section, the First Circuit explained that the statistical models of Harvard’s admissions practices did not show bias against Asian American applicants on Harvard’s part.42 While being Asian American was negatively correlated with personal rating, the panel wrote, it was more likely that “factors external to Harvard . . . like personal essays and recommendations” were the cause of the correlation, rather than any racial animus.43 Because “[p]rivileged students likely have better access to schools with low student-to-teacher ratios and teachers and guidance counselors with more time to write strong, individualized recommendations,” and white students are more likely to be privileged than Asian American students, the panel noted the possibility that privilege was the true cause of the correlation, not Harvard’s racial prejudice.44 As a result, the district court made no error when it credited a statistical model using personal rating as a variable; “[w]ithout the personal rating, the model would suffer from omitted variable bias.”45
The First Circuit in Students for Fair Admissions came to two convincing conclusions: Harvard’s affirmative action policies were justified, and Harvard did not intentionally discriminate. But its opinion did not explain the relationship between them. Indeed, Judge Lynch wrote that the connection between the alleged intentional discrimination and Harvard’s explicit use of race was “not entirely clear.”46 As a result, the First Circuit failed to address the argument, advanced by DOJ, that may pose the greatest threat to affirmative action — that an intentional discrimination claim should be part of a race-conscious program’s strict scrutiny analysis. Incorporating allegations of intentional discrimination into such analysis would, without precedent, invalidate affirmative action programs unless a defendant can prove that they aren’t biased against any racial minority. Such a shift is explained neither by the doctrine nor by the purpose of strict scrutiny.
Strict scrutiny exists to ensure that preferences set along lines of race are justified. While courts and scholars debate what makes for a valid justification,47 strict scrutiny’s requirement that a race-conscious policy be narrowly tailored to a compelling interest serves this end by ensuring a particularly important stated purpose and an exceptionally strong fit between a policy’s mechanics and that purpose.48 The questions traditionally associated with narrow tailoring speak to this idea of fit: Is the defendant engaging in quotas or racial balancing?49 (If so, how could they possibly be promoting any meaningful sense of diversity?) Could the defendant achieve the same goal without using race?50 (If so, what possible valid reason could they have for deciding to use race?) The defendant bears the burden of answering.51 Implicit, however, in these inquiries of purpose and fit is a threshold question: Is the defendant implementing a racial preference at all?52 If not, then it makes no sense to examine whether the purpose of the alleged racial preference is compelling or the means narrowly tailored. The inquiry stops.
In DOJ’s telling, Harvard’s alleged bias against Asian Americans represented a failure of the university to narrowly tailor their race-conscious practices to their diversity interest.53 But it was only by asking the threshold question with regard to some racial minorities, and asking the narrow tailoring questions with regard to a different minority, that DOJ’s argument had the appearance of force. In reality, SFFA was challenging two different practices: 1) Harvard’s (admitted) preference for Black and Hispanic applicants over equally qualified applicants of any other race, and 2) Harvard’s (denied) penalization of Asian American applicants vis-à-vis white applicants.54 The first practice, viewed on its own, clearly meets the threshold criterion to apply strict scrutiny. But it likely also meets the narrow tailoring and compelling interest criteria to survive it. The second practice, also viewed on its own, would clearly fail the narrow tailoring and compelling interest requirements of strict scrutiny.55 But this would matter only if it passed the threshold criterion to apply such scrutiny. By describing Harvard as incorporating race into the admissions process generally, DOJ’s argument attempts to use the first practice to meet the threshold for strict scrutiny, and the second practice to invalidate the affirmative action program as a failure of narrow tailoring.
Merging the practices in this way would have had two critical consequences in Students for Fair Admissions. First, it would have meant that, like for other elements of narrow tailoring, Harvard held the burden to prove it did not intentionally discriminate against Asian American applicants.56 Second, it would have meant that establishing discrimination against Asian Americans would bring down Harvard’s entire affirmative action program.57
The First Circuit avoided these consequences only by avoiding the actual decision of whether to merge its analyses of Harvard’s admissions practices. Structurally, the practices were separated into “Admissions Program” (affirmative action) and “Intentional Discriminat[ion]” sections.58 But in the latter, Judge Lynch assumed arguendo what could only be justified by analyzing the practices together — that Harvard held the burden to disprove bias against Asian American applicants.59 And in order to rule in Harvard’s favor despite this convenient assumption, the First Circuit analyzed evidence as if it had barely moved the burden at all. The statistical evidence proffered by SFFA showed that, even accounting for all numerical factors including the scores admissions officers assigned for the strength of teacher and guidance counselor recommendations, those officers gave Asian American applicants lower personal ratings than they gave white applicants.60 Still the panel accepted Harvard’s contention that it was likely the applicants’ personal statements and the specific language used in the recommendations that caused this discrepancy, without requiring Harvard to undergo the sizeable task of presenting evidence disproving implicit bias on the part of their admissions officers.61 Instead, it was enough that Harvard “called . . . into question”62 the inferences SFFA drew from the statistical evidence. This analysis was far from the “stringent” narrow tailoring requirement advocated by DOJ.63
A more rigorous analysis would have revealed that DOJ’s practice-merging was unjustified. In arguing that the intentional discrimination analysis should be a part of narrow tailoring, DOJ relied on the proposition articulated in Grutter v. Bollinger64 that “[n]arrow tailoring . . . requires that a race-conscious admissions program not unduly harm members of any racial group.”65 But viewed in proper context, this was a statement about comparing all of those favored by a racial preference to all of those not favored, not about comparing those favored to a subgroup of those not favored, or about comparing two subgroups of those not favored.66 In other words, rather than articulating an additional requirement, the Grutter language relied on by DOJ is best interpreted as merely restating a defendant’s obligation to show a close fit between their practice and their purpose. Consistent with Grutter, the question of whether a defendant intentionally discriminates should be part of a narrow tailoring analysis only to the extent that the alleged biases are shown to be consequences of the policy that triggers strict scrutiny in the first place.
The facts of Harvard’s admissions process, and of affirmative action programs in general, support this interpretation. While the “tips” Harvard gives to Black and Hispanic applicants are a result of explicit decisionmaking, any penalties Harvard assigns to Asian American applicants are more likely to be a result of stereotyping or implicit bias.67 If admissions officers make decisions about Asian American applicants according to stereotypes or implicit biases, they could and presumably would do so even under a system with no affirmative action. Therefore, there is no reason to think eliminating explicit consideration of race from the admissions process would eliminate the alleged Asian American penalty. This dynamic is constant throughout affirmative action programs — explicit racial considerations are separate from, and are often meant to work against, stereotypes.68
Affirmative action survived the First Circuit’s undertheorized account of admissions practices and may be yet again on its way to the Supreme Court.69 But the premise of DOJ’s argument — that the gateway to strict scrutiny can be opened using one minority group and entered using another — would likely lead to a different outcome in front of a majority of Justices deeply skeptical of race-conscious practices.70 Advocates would be wise to avoid conflating narrow-tailoring arguments with intentional discrimination arguments. Doing so has no basis in the law or the reality of affirmative action.